[llvm] r349520 - [X86] Use SADDSAT/SSUBSAT instead of ADDS/SUBS

Nikita Popov via llvm-commits llvm-commits at lists.llvm.org
Tue Dec 18 10:28:22 PST 2018


Author: nikic
Date: Tue Dec 18 10:28:22 2018
New Revision: 349520

URL: http://llvm.org/viewvc/llvm-project?rev=349520&view=rev
Log:
[X86] Use SADDSAT/SSUBSAT instead of ADDS/SUBS

Migrate the X86 backend from X86ISD opcodes ADDS and SUBS to generic
ISD opcodes SADDSAT and SSUBSAT. This also improves scodegen for
@llvm.sadd.sat() and @llvm.ssub.sat() intrinsics.

This is a followup to D55787 and part of PR40056.

Differential Revision: https://reviews.llvm.org/D55833

Modified:
    llvm/trunk/lib/Target/X86/X86ISelLowering.cpp
    llvm/trunk/lib/Target/X86/X86ISelLowering.h
    llvm/trunk/lib/Target/X86/X86InstrAVX512.td
    llvm/trunk/lib/Target/X86/X86InstrFragmentsSIMD.td
    llvm/trunk/lib/Target/X86/X86InstrSSE.td
    llvm/trunk/lib/Target/X86/X86IntrinsicsInfo.h
    llvm/trunk/test/CodeGen/X86/sadd_sat_vec.ll
    llvm/trunk/test/CodeGen/X86/ssub_sat_vec.ll

Modified: llvm/trunk/lib/Target/X86/X86ISelLowering.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/X86/X86ISelLowering.cpp?rev=349520&r1=349519&r2=349520&view=diff
==============================================================================
--- llvm/trunk/lib/Target/X86/X86ISelLowering.cpp (original)
+++ llvm/trunk/lib/Target/X86/X86ISelLowering.cpp Tue Dec 18 10:28:22 2018
@@ -830,14 +830,20 @@ X86TargetLowering::X86TargetLowering(con
     }
 
     setOperationAction(ISD::UADDSAT,            MVT::v16i8, Legal);
-    setOperationAction(ISD::UADDSAT,            MVT::v8i16, Legal);
+    setOperationAction(ISD::SADDSAT,            MVT::v16i8, Legal);
     setOperationAction(ISD::USUBSAT,            MVT::v16i8, Legal);
+    setOperationAction(ISD::SSUBSAT,            MVT::v16i8, Legal);
+    setOperationAction(ISD::UADDSAT,            MVT::v8i16, Legal);
+    setOperationAction(ISD::SADDSAT,            MVT::v8i16, Legal);
     setOperationAction(ISD::USUBSAT,            MVT::v8i16, Legal);
+    setOperationAction(ISD::SSUBSAT,            MVT::v8i16, Legal);
     // Use widening instead of promotion.
     for (auto VT : { MVT::v8i8, MVT::v4i8, MVT::v2i8,
                      MVT::v4i16, MVT::v2i16 }) {
-      setOperationAction(ISD::UADDSAT,          VT,  Custom);
-      setOperationAction(ISD::USUBSAT,          VT,  Custom);
+      setOperationAction(ISD::UADDSAT, VT, Custom);
+      setOperationAction(ISD::SADDSAT, VT, Custom);
+      setOperationAction(ISD::USUBSAT, VT, Custom);
+      setOperationAction(ISD::SSUBSAT, VT, Custom);
     }
 
     setOperationAction(ISD::INSERT_VECTOR_ELT,  MVT::v8i16, Custom);
@@ -1212,9 +1218,13 @@ X86TargetLowering::X86TargetLowering(con
     setOperationAction(ISD::UMIN,      MVT::v4i64,  Custom);
 
     setOperationAction(ISD::UADDSAT,   MVT::v32i8,  HasInt256 ? Legal : Custom);
-    setOperationAction(ISD::UADDSAT,   MVT::v16i16, HasInt256 ? Legal : Custom);
+    setOperationAction(ISD::SADDSAT,   MVT::v32i8,  HasInt256 ? Legal : Custom);
     setOperationAction(ISD::USUBSAT,   MVT::v32i8,  HasInt256 ? Legal : Custom);
+    setOperationAction(ISD::SSUBSAT,   MVT::v32i8,  HasInt256 ? Legal : Custom);
+    setOperationAction(ISD::UADDSAT,   MVT::v16i16, HasInt256 ? Legal : Custom);
+    setOperationAction(ISD::SADDSAT,   MVT::v16i16, HasInt256 ? Legal : Custom);
     setOperationAction(ISD::USUBSAT,   MVT::v16i16, HasInt256 ? Legal : Custom);
+    setOperationAction(ISD::SSUBSAT,   MVT::v16i16, HasInt256 ? Legal : Custom);
 
     for (auto VT : { MVT::v32i8, MVT::v16i16, MVT::v8i32 }) {
       setOperationAction(ISD::ABS,  VT, HasInt256 ? Legal : Custom);
@@ -1334,7 +1344,9 @@ X86TargetLowering::X86TargetLowering(con
       setOperationAction(ISD::SELECT,           VT, Custom);
       setOperationAction(ISD::TRUNCATE,         VT, Custom);
       setOperationAction(ISD::UADDSAT,          VT, Custom);
+      setOperationAction(ISD::SADDSAT,          VT, Custom);
       setOperationAction(ISD::USUBSAT,          VT, Custom);
+      setOperationAction(ISD::SSUBSAT,          VT, Custom);
 
       setOperationAction(ISD::BUILD_VECTOR,     VT, Custom);
       setOperationAction(ISD::EXTRACT_VECTOR_ELT, VT, Custom);
@@ -1596,7 +1608,9 @@ X86TargetLowering::X86TargetLowering(con
       setOperationAction(ISD::MUL,                VT, Custom);
       setOperationAction(ISD::VSELECT,            VT, Expand);
       setOperationAction(ISD::UADDSAT,            VT, Custom);
+      setOperationAction(ISD::SADDSAT,            VT, Custom);
       setOperationAction(ISD::USUBSAT,            VT, Custom);
+      setOperationAction(ISD::SSUBSAT,            VT, Custom);
 
       setOperationAction(ISD::TRUNCATE,           VT, Custom);
       setOperationAction(ISD::SETCC,              VT, Custom);
@@ -1678,7 +1692,9 @@ X86TargetLowering::X86TargetLowering(con
       setOperationAction(ISD::UMIN,         VT, Legal);
       setOperationAction(ISD::SETCC,        VT, Custom);
       setOperationAction(ISD::UADDSAT,      VT, Legal);
+      setOperationAction(ISD::SADDSAT,      VT, Legal);
       setOperationAction(ISD::USUBSAT,      VT, Legal);
+      setOperationAction(ISD::SSUBSAT,      VT, Legal);
 
       // The condition codes aren't legal in SSE/AVX and under AVX512 we use
       // setcc all the way to isel and prefer SETGT in some isel patterns.
@@ -23388,15 +23404,17 @@ static SDValue LowerADD_SUB(SDValue Op,
   return split256IntArith(Op, DAG);
 }
 
-static SDValue LowerUADDSAT_USUBSAT(SDValue Op, SelectionDAG &DAG) {
+static SDValue LowerADDSAT_SUBSAT(SDValue Op, SelectionDAG &DAG) {
   MVT VT = Op.getSimpleValueType();
   if (VT.getScalarType() == MVT::i1) {
     SDLoc dl(Op);
     switch (Op.getOpcode()) {
     default: llvm_unreachable("Expected saturated arithmetic opcode");
     case ISD::UADDSAT:
+    case ISD::SADDSAT:
       return DAG.getNode(ISD::OR, dl, VT, Op.getOperand(0), Op.getOperand(1));
     case ISD::USUBSAT:
+    case ISD::SSUBSAT:
       return DAG.getNode(ISD::AND, dl, VT, Op.getOperand(0),
                          DAG.getNOT(dl, Op.getOperand(1), VT));
     }
@@ -26194,7 +26212,9 @@ SDValue X86TargetLowering::LowerOperatio
   case ISD::ADD:
   case ISD::SUB:                return LowerADD_SUB(Op, DAG);
   case ISD::UADDSAT:
-  case ISD::USUBSAT:            return LowerUADDSAT_USUBSAT(Op, DAG);
+  case ISD::SADDSAT:
+  case ISD::USUBSAT:
+  case ISD::SSUBSAT:            return LowerADDSAT_SUBSAT(Op, DAG);
   case ISD::SMAX:
   case ISD::SMIN:
   case ISD::UMAX:
@@ -26277,11 +26297,13 @@ void X86TargetLowering::ReplaceNodeResul
     return;
   }
   case ISD::UADDSAT:
+  case ISD::SADDSAT:
   case ISD::USUBSAT:
+  case ISD::SSUBSAT:
   case X86ISD::VPMADDWD:
   case X86ISD::AVG: {
-    // Legalize types for ISD::UADDSAT/USUBSAT and X86ISD::AVG/VPMADDWD
-    // by widening.
+    // Legalize types for ISD::UADDSAT/SADDSAT/USUBSAT/SSUBSAT and
+    // X86ISD::AVG/VPMADDWD by widening.
     assert(Subtarget.hasSSE2() && "Requires at least SSE2!");
 
     EVT VT = N->getValueType(0);
@@ -27228,8 +27250,6 @@ const char *X86TargetLowering::getTarget
   case X86ISD::FGETEXPS_RND:       return "X86ISD::FGETEXPS_RND";
   case X86ISD::SCALEF:             return "X86ISD::SCALEF";
   case X86ISD::SCALEFS:            return "X86ISD::SCALEFS";
-  case X86ISD::ADDS:               return "X86ISD::ADDS";
-  case X86ISD::SUBS:               return "X86ISD::SUBS";
   case X86ISD::AVG:                return "X86ISD::AVG";
   case X86ISD::MULHRS:             return "X86ISD::MULHRS";
   case X86ISD::SINT_TO_FP_RND:     return "X86ISD::SINT_TO_FP_RND";

Modified: llvm/trunk/lib/Target/X86/X86ISelLowering.h
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/X86/X86ISelLowering.h?rev=349520&r1=349519&r2=349520&view=diff
==============================================================================
--- llvm/trunk/lib/Target/X86/X86ISelLowering.h (original)
+++ llvm/trunk/lib/Target/X86/X86ISelLowering.h Tue Dec 18 10:28:22 2018
@@ -226,10 +226,6 @@ namespace llvm {
       SCALEF,
       SCALEFS,
 
-      // Integer add/sub with signed saturation.
-      ADDS,
-      SUBS,
-
       // Unsigned Integer average.
       AVG,
 

Modified: llvm/trunk/lib/Target/X86/X86InstrAVX512.td
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/X86/X86InstrAVX512.td?rev=349520&r1=349519&r2=349520&view=diff
==============================================================================
--- llvm/trunk/lib/Target/X86/X86InstrAVX512.td (original)
+++ llvm/trunk/lib/Target/X86/X86InstrAVX512.td Tue Dec 18 10:28:22 2018
@@ -4830,9 +4830,9 @@ defm VPADD : avx512_binop_rm_vl_all<0xFC
                                     SchedWriteVecALU, 1>;
 defm VPSUB : avx512_binop_rm_vl_all<0xF8, 0xF9, 0xFA, 0xFB, "vpsub", sub,
                                     SchedWriteVecALU, 0>;
-defm VPADDS : avx512_binop_rm_vl_bw<0xEC, 0xED, "vpadds", X86adds,
+defm VPADDS : avx512_binop_rm_vl_bw<0xEC, 0xED, "vpadds", saddsat,
                                     SchedWriteVecALU, HasBWI, 1>;
-defm VPSUBS : avx512_binop_rm_vl_bw<0xE8, 0xE9, "vpsubs", X86subs,
+defm VPSUBS : avx512_binop_rm_vl_bw<0xE8, 0xE9, "vpsubs", ssubsat,
                                     SchedWriteVecALU, HasBWI, 0>;
 defm VPADDUS : avx512_binop_rm_vl_bw<0xDC, 0xDD, "vpaddus", uaddsat,
                                      SchedWriteVecALU, HasBWI, 1>;

Modified: llvm/trunk/lib/Target/X86/X86InstrFragmentsSIMD.td
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/X86/X86InstrFragmentsSIMD.td?rev=349520&r1=349519&r2=349520&view=diff
==============================================================================
--- llvm/trunk/lib/Target/X86/X86InstrFragmentsSIMD.td (original)
+++ llvm/trunk/lib/Target/X86/X86InstrFragmentsSIMD.td Tue Dec 18 10:28:22 2018
@@ -227,8 +227,6 @@ def SDTX86CmpPTest : SDTypeProfile<1, 2,
                                           SDTCisVec<1>,
                                           SDTCisSameAs<2, 1>]>;
 
-def X86adds    : SDNode<"X86ISD::ADDS", SDTIntBinOp, [SDNPCommutative]>;
-def X86subs    : SDNode<"X86ISD::SUBS", SDTIntBinOp>;
 def X86mulhrs  : SDNode<"X86ISD::MULHRS", SDTIntBinOp, [SDNPCommutative]>;
 def X86avg     : SDNode<"X86ISD::AVG" , SDTIntBinOp, [SDNPCommutative]>;
 def X86ptest   : SDNode<"X86ISD::PTEST", SDTX86CmpPTest>;

Modified: llvm/trunk/lib/Target/X86/X86InstrSSE.td
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/X86/X86InstrSSE.td?rev=349520&r1=349519&r2=349520&view=diff
==============================================================================
--- llvm/trunk/lib/Target/X86/X86InstrSSE.td (original)
+++ llvm/trunk/lib/Target/X86/X86InstrSSE.td Tue Dec 18 10:28:22 2018
@@ -3623,9 +3623,9 @@ defm PADDD   : PDI_binop_all<0xFE, "padd
                              SchedWriteVecALU, 1, NoVLX>;
 defm PADDQ   : PDI_binop_all<0xD4, "paddq", add, v2i64, v4i64,
                              SchedWriteVecALU, 1, NoVLX>;
-defm PADDSB  : PDI_binop_all<0xEC, "paddsb", X86adds, v16i8, v32i8,
+defm PADDSB  : PDI_binop_all<0xEC, "paddsb", saddsat, v16i8, v32i8,
                              SchedWriteVecALU, 1, NoVLX_Or_NoBWI>;
-defm PADDSW  : PDI_binop_all<0xED, "paddsw", X86adds, v8i16, v16i16,
+defm PADDSW  : PDI_binop_all<0xED, "paddsw", saddsat, v8i16, v16i16,
                              SchedWriteVecALU, 1, NoVLX_Or_NoBWI>;
 defm PADDUSB : PDI_binop_all<0xDC, "paddusb", uaddsat, v16i8, v32i8,
                              SchedWriteVecALU, 1, NoVLX_Or_NoBWI>;
@@ -3645,9 +3645,9 @@ defm PSUBD   : PDI_binop_all<0xFA, "psub
                              SchedWriteVecALU, 0, NoVLX>;
 defm PSUBQ   : PDI_binop_all<0xFB, "psubq", sub, v2i64, v4i64,
                              SchedWriteVecALU, 0, NoVLX>;
-defm PSUBSB  : PDI_binop_all<0xE8, "psubsb", X86subs, v16i8, v32i8,
+defm PSUBSB  : PDI_binop_all<0xE8, "psubsb", ssubsat, v16i8, v32i8,
                              SchedWriteVecALU, 0, NoVLX_Or_NoBWI>;
-defm PSUBSW  : PDI_binop_all<0xE9, "psubsw", X86subs, v8i16, v16i16,
+defm PSUBSW  : PDI_binop_all<0xE9, "psubsw", ssubsat, v8i16, v16i16,
                              SchedWriteVecALU, 0, NoVLX_Or_NoBWI>;
 defm PSUBUSB : PDI_binop_all<0xD8, "psubusb", usubsat, v16i8, v32i8,
                              SchedWriteVecALU, 0, NoVLX_Or_NoBWI>;

Modified: llvm/trunk/lib/Target/X86/X86IntrinsicsInfo.h
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/X86/X86IntrinsicsInfo.h?rev=349520&r1=349519&r2=349520&view=diff
==============================================================================
--- llvm/trunk/lib/Target/X86/X86IntrinsicsInfo.h (original)
+++ llvm/trunk/lib/Target/X86/X86IntrinsicsInfo.h Tue Dec 18 10:28:22 2018
@@ -319,8 +319,8 @@ static const IntrinsicData  IntrinsicsWi
   X86_INTRINSIC_DATA(avx2_packsswb, INTR_TYPE_2OP, X86ISD::PACKSS, 0),
   X86_INTRINSIC_DATA(avx2_packusdw, INTR_TYPE_2OP, X86ISD::PACKUS, 0),
   X86_INTRINSIC_DATA(avx2_packuswb, INTR_TYPE_2OP, X86ISD::PACKUS, 0),
-  X86_INTRINSIC_DATA(avx2_padds_b, INTR_TYPE_2OP, X86ISD::ADDS, 0),
-  X86_INTRINSIC_DATA(avx2_padds_w, INTR_TYPE_2OP, X86ISD::ADDS, 0),
+  X86_INTRINSIC_DATA(avx2_padds_b, INTR_TYPE_2OP, ISD::SADDSAT, 0),
+  X86_INTRINSIC_DATA(avx2_padds_w, INTR_TYPE_2OP, ISD::SADDSAT, 0),
   X86_INTRINSIC_DATA(avx2_permd, VPERM_2OP, X86ISD::VPERMV, 0),
   X86_INTRINSIC_DATA(avx2_permps, VPERM_2OP, X86ISD::VPERMV, 0),
   X86_INTRINSIC_DATA(avx2_phadd_d, INTR_TYPE_2OP, X86ISD::HADD, 0),
@@ -361,8 +361,8 @@ static const IntrinsicData  IntrinsicsWi
   X86_INTRINSIC_DATA(avx2_psrlv_d_256, INTR_TYPE_2OP, ISD::SRL, 0),
   X86_INTRINSIC_DATA(avx2_psrlv_q, INTR_TYPE_2OP, ISD::SRL, 0),
   X86_INTRINSIC_DATA(avx2_psrlv_q_256, INTR_TYPE_2OP, ISD::SRL, 0),
-  X86_INTRINSIC_DATA(avx2_psubs_b, INTR_TYPE_2OP, X86ISD::SUBS, 0),
-  X86_INTRINSIC_DATA(avx2_psubs_w, INTR_TYPE_2OP, X86ISD::SUBS, 0),
+  X86_INTRINSIC_DATA(avx2_psubs_b, INTR_TYPE_2OP, ISD::SSUBSAT, 0),
+  X86_INTRINSIC_DATA(avx2_psubs_w, INTR_TYPE_2OP, ISD::SSUBSAT, 0),
   X86_INTRINSIC_DATA(avx512_add_pd_512, INTR_TYPE_2OP, ISD::FADD, X86ISD::FADD_RND),
   X86_INTRINSIC_DATA(avx512_add_ps_512, INTR_TYPE_2OP, ISD::FADD, X86ISD::FADD_RND),
   X86_INTRINSIC_DATA(avx512_cmp_pd_128, CMP_MASK_CC, X86ISD::CMPM, 0),
@@ -920,8 +920,8 @@ static const IntrinsicData  IntrinsicsWi
   X86_INTRINSIC_DATA(avx512_packsswb_512, INTR_TYPE_2OP, X86ISD::PACKSS, 0),
   X86_INTRINSIC_DATA(avx512_packusdw_512, INTR_TYPE_2OP, X86ISD::PACKUS, 0),
   X86_INTRINSIC_DATA(avx512_packuswb_512, INTR_TYPE_2OP, X86ISD::PACKUS, 0),
-  X86_INTRINSIC_DATA(avx512_padds_b_512, INTR_TYPE_2OP, X86ISD::ADDS, 0),
-  X86_INTRINSIC_DATA(avx512_padds_w_512, INTR_TYPE_2OP, X86ISD::ADDS, 0),
+  X86_INTRINSIC_DATA(avx512_padds_b_512, INTR_TYPE_2OP, ISD::SADDSAT, 0),
+  X86_INTRINSIC_DATA(avx512_padds_w_512, INTR_TYPE_2OP, ISD::SADDSAT, 0),
   X86_INTRINSIC_DATA(avx512_permvar_df_256, VPERM_2OP, X86ISD::VPERMV, 0),
   X86_INTRINSIC_DATA(avx512_permvar_df_512, VPERM_2OP, X86ISD::VPERMV, 0),
   X86_INTRINSIC_DATA(avx512_permvar_di_256, VPERM_2OP, X86ISD::VPERMV, 0),
@@ -1004,8 +1004,8 @@ static const IntrinsicData  IntrinsicsWi
   X86_INTRINSIC_DATA(avx512_psrlv_w_128, INTR_TYPE_2OP, ISD::SRL, 0),
   X86_INTRINSIC_DATA(avx512_psrlv_w_256, INTR_TYPE_2OP, ISD::SRL, 0),
   X86_INTRINSIC_DATA(avx512_psrlv_w_512, INTR_TYPE_2OP, ISD::SRL, 0),
-  X86_INTRINSIC_DATA(avx512_psubs_b_512, INTR_TYPE_2OP, X86ISD::SUBS, 0),
-  X86_INTRINSIC_DATA(avx512_psubs_w_512, INTR_TYPE_2OP, X86ISD::SUBS, 0),
+  X86_INTRINSIC_DATA(avx512_psubs_b_512, INTR_TYPE_2OP, ISD::SSUBSAT, 0),
+  X86_INTRINSIC_DATA(avx512_psubs_w_512, INTR_TYPE_2OP, ISD::SSUBSAT, 0),
   X86_INTRINSIC_DATA(avx512_pternlog_d_128, INTR_TYPE_4OP, X86ISD::VPTERNLOG, 0),
   X86_INTRINSIC_DATA(avx512_pternlog_d_256, INTR_TYPE_4OP, X86ISD::VPTERNLOG, 0),
   X86_INTRINSIC_DATA(avx512_pternlog_d_512, INTR_TYPE_4OP, X86ISD::VPTERNLOG, 0),
@@ -1168,8 +1168,8 @@ static const IntrinsicData  IntrinsicsWi
   X86_INTRINSIC_DATA(sse2_packssdw_128, INTR_TYPE_2OP, X86ISD::PACKSS, 0),
   X86_INTRINSIC_DATA(sse2_packsswb_128, INTR_TYPE_2OP, X86ISD::PACKSS, 0),
   X86_INTRINSIC_DATA(sse2_packuswb_128, INTR_TYPE_2OP, X86ISD::PACKUS, 0),
-  X86_INTRINSIC_DATA(sse2_padds_b,      INTR_TYPE_2OP, X86ISD::ADDS, 0),
-  X86_INTRINSIC_DATA(sse2_padds_w,      INTR_TYPE_2OP, X86ISD::ADDS, 0),
+  X86_INTRINSIC_DATA(sse2_padds_b,      INTR_TYPE_2OP, ISD::SADDSAT, 0),
+  X86_INTRINSIC_DATA(sse2_padds_w,      INTR_TYPE_2OP, ISD::SADDSAT, 0),
   X86_INTRINSIC_DATA(sse2_pmadd_wd,     INTR_TYPE_2OP, X86ISD::VPMADDWD, 0),
   X86_INTRINSIC_DATA(sse2_pmovmskb_128, INTR_TYPE_1OP, X86ISD::MOVMSK, 0),
   X86_INTRINSIC_DATA(sse2_pmulh_w,      INTR_TYPE_2OP, ISD::MULHS, 0),
@@ -1191,8 +1191,8 @@ static const IntrinsicData  IntrinsicsWi
   X86_INTRINSIC_DATA(sse2_psrli_d,      VSHIFT, X86ISD::VSRLI, 0),
   X86_INTRINSIC_DATA(sse2_psrli_q,      VSHIFT, X86ISD::VSRLI, 0),
   X86_INTRINSIC_DATA(sse2_psrli_w,      VSHIFT, X86ISD::VSRLI, 0),
-  X86_INTRINSIC_DATA(sse2_psubs_b,      INTR_TYPE_2OP, X86ISD::SUBS, 0),
-  X86_INTRINSIC_DATA(sse2_psubs_w,      INTR_TYPE_2OP, X86ISD::SUBS, 0),
+  X86_INTRINSIC_DATA(sse2_psubs_b,      INTR_TYPE_2OP, ISD::SSUBSAT, 0),
+  X86_INTRINSIC_DATA(sse2_psubs_w,      INTR_TYPE_2OP, ISD::SSUBSAT, 0),
   X86_INTRINSIC_DATA(sse2_ucomieq_sd,   COMI, X86ISD::UCOMI, ISD::SETEQ),
   X86_INTRINSIC_DATA(sse2_ucomige_sd,   COMI, X86ISD::UCOMI, ISD::SETGE),
   X86_INTRINSIC_DATA(sse2_ucomigt_sd,   COMI, X86ISD::UCOMI, ISD::SETGT),

Modified: llvm/trunk/test/CodeGen/X86/sadd_sat_vec.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/X86/sadd_sat_vec.ll?rev=349520&r1=349519&r2=349520&view=diff
==============================================================================
--- llvm/trunk/test/CodeGen/X86/sadd_sat_vec.ll (original)
+++ llvm/trunk/test/CodeGen/X86/sadd_sat_vec.ll Tue Dec 18 10:28:22 2018
@@ -34,13453 +34,160 @@ declare <2 x i128> @llvm.sadd.sat.v2i128
 ; Legal types, depending on architecture.
 
 define <16 x i8> @v16i8(<16 x i8> %x, <16 x i8> %y) nounwind {
-; SSE2-LABEL: v16i8:
-; SSE2:       # %bb.0:
-; SSE2-NEXT:    pushq %rbp
-; SSE2-NEXT:    pushq %r15
-; SSE2-NEXT:    pushq %r14
-; SSE2-NEXT:    pushq %r13
-; SSE2-NEXT:    pushq %r12
-; SSE2-NEXT:    pushq %rbx
-; SSE2-NEXT:    movaps %xmm1, -{{[0-9]+}}(%rsp)
-; SSE2-NEXT:    movaps %xmm0, -{{[0-9]+}}(%rsp)
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %r9b
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %sil
-; SSE2-NEXT:    movl %r9d, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %r9b
-; SSE2-NEXT:    jno .LBB0_2
-; SSE2-NEXT:  # %bb.1:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %r9d
-; SSE2-NEXT:  .LBB0_2:
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movl %esi, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %sil
-; SSE2-NEXT:    jno .LBB0_4
-; SSE2-NEXT:  # %bb.3:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %esi
-; SSE2-NEXT:  .LBB0_4:
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %bl
-; SSE2-NEXT:    movl %ebx, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %bl
-; SSE2-NEXT:    jno .LBB0_6
-; SSE2-NEXT:  # %bb.5:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %ebx
-; SSE2-NEXT:  .LBB0_6:
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %dl
-; SSE2-NEXT:    movl %edx, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %dl
-; SSE2-NEXT:    jno .LBB0_8
-; SSE2-NEXT:  # %bb.7:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %edx
-; SSE2-NEXT:  .LBB0_8:
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %r10b
-; SSE2-NEXT:    movl %r10d, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %r10b
-; SSE2-NEXT:    jno .LBB0_10
-; SSE2-NEXT:  # %bb.9:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %r10d
-; SSE2-NEXT:  .LBB0_10:
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %r11b
-; SSE2-NEXT:    movl %r11d, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %r11b
-; SSE2-NEXT:    jno .LBB0_12
-; SSE2-NEXT:  # %bb.11:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %r11d
-; SSE2-NEXT:  .LBB0_12:
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %bpl
-; SSE2-NEXT:    movl %ebp, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %bpl
-; SSE2-NEXT:    jno .LBB0_14
-; SSE2-NEXT:  # %bb.13:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %ebp
-; SSE2-NEXT:  .LBB0_14:
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %r14b
-; SSE2-NEXT:    movl %r14d, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %r14b
-; SSE2-NEXT:    jno .LBB0_16
-; SSE2-NEXT:  # %bb.15:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %r14d
-; SSE2-NEXT:  .LBB0_16:
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %r15b
-; SSE2-NEXT:    movl %r15d, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %r15b
-; SSE2-NEXT:    jno .LBB0_18
-; SSE2-NEXT:  # %bb.17:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %r15d
-; SSE2-NEXT:  .LBB0_18:
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %r12b
-; SSE2-NEXT:    movl %r12d, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %r12b
-; SSE2-NEXT:    jno .LBB0_20
-; SSE2-NEXT:  # %bb.19:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %r12d
-; SSE2-NEXT:  .LBB0_20:
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %r13b
-; SSE2-NEXT:    movl %r13d, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %r13b
-; SSE2-NEXT:    movb %bl, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSE2-NEXT:    movb %dl, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSE2-NEXT:    jno .LBB0_22
-; SSE2-NEXT:  # %bb.21:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %r13d
-; SSE2-NEXT:  .LBB0_22:
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %dil
-; SSE2-NEXT:    movl %edi, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %dil
-; SSE2-NEXT:    jno .LBB0_24
-; SSE2-NEXT:  # %bb.23:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %edi
-; SSE2-NEXT:  .LBB0_24:
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %r8b
-; SSE2-NEXT:    movl %r8d, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %r8b
-; SSE2-NEXT:    jno .LBB0_26
-; SSE2-NEXT:  # %bb.25:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %r8d
-; SSE2-NEXT:  .LBB0_26:
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %bl
-; SSE2-NEXT:    movl %ebx, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %bl
-; SSE2-NEXT:    jno .LBB0_28
-; SSE2-NEXT:  # %bb.27:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %ebx
-; SSE2-NEXT:  .LBB0_28:
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %dl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
-; SSE2-NEXT:    movl %eax, %ecx
-; SSE2-NEXT:    addb %dl, %cl
-; SSE2-NEXT:    setns %cl
-; SSE2-NEXT:    addb %dl, %al
-; SSE2-NEXT:    movb %sil, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSE2-NEXT:    jno .LBB0_30
-; SSE2-NEXT:  # %bb.29:
-; SSE2-NEXT:    addb $127, %cl
-; SSE2-NEXT:    movl %ecx, %eax
-; SSE2-NEXT:  .LBB0_30:
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %dl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %sil
-; SSE2-NEXT:    movl %esi, %ecx
-; SSE2-NEXT:    addb %dl, %cl
-; SSE2-NEXT:    setns %cl
-; SSE2-NEXT:    addb %dl, %sil
-; SSE2-NEXT:    jno .LBB0_32
-; SSE2-NEXT:  # %bb.31:
-; SSE2-NEXT:    addb $127, %cl
-; SSE2-NEXT:    movl %ecx, %esi
-; SSE2-NEXT:  .LBB0_32:
-; SSE2-NEXT:    movzbl %sil, %ecx
-; SSE2-NEXT:    movd %ecx, %xmm0
-; SSE2-NEXT:    movzbl %al, %eax
-; SSE2-NEXT:    movd %eax, %xmm1
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3],xmm1[4],xmm0[4],xmm1[5],xmm0[5],xmm1[6],xmm0[6],xmm1[7],xmm0[7]
-; SSE2-NEXT:    movzbl %bl, %eax
-; SSE2-NEXT:    movd %eax, %xmm0
-; SSE2-NEXT:    movzbl %r8b, %eax
-; SSE2-NEXT:    movd %eax, %xmm2
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm2 = xmm2[0],xmm0[0],xmm2[1],xmm0[1],xmm2[2],xmm0[2],xmm2[3],xmm0[3],xmm2[4],xmm0[4],xmm2[5],xmm0[5],xmm2[6],xmm0[6],xmm2[7],xmm0[7]
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm2 = xmm2[0],xmm1[0],xmm2[1],xmm1[1],xmm2[2],xmm1[2],xmm2[3],xmm1[3]
-; SSE2-NEXT:    movzbl %dil, %eax
-; SSE2-NEXT:    movd %eax, %xmm0
-; SSE2-NEXT:    movzbl %r13b, %eax
-; SSE2-NEXT:    movd %eax, %xmm3
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm3 = xmm3[0],xmm0[0],xmm3[1],xmm0[1],xmm3[2],xmm0[2],xmm3[3],xmm0[3],xmm3[4],xmm0[4],xmm3[5],xmm0[5],xmm3[6],xmm0[6],xmm3[7],xmm0[7]
-; SSE2-NEXT:    movzbl %r12b, %eax
-; SSE2-NEXT:    movd %eax, %xmm0
-; SSE2-NEXT:    movzbl %r15b, %eax
-; SSE2-NEXT:    movd %eax, %xmm1
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3],xmm1[4],xmm0[4],xmm1[5],xmm0[5],xmm1[6],xmm0[6],xmm1[7],xmm0[7]
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm3[0],xmm1[1],xmm3[1],xmm1[2],xmm3[2],xmm1[3],xmm3[3]
-; SSE2-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm2[0],xmm1[1],xmm2[1]
-; SSE2-NEXT:    movzbl %r14b, %eax
-; SSE2-NEXT:    movd %eax, %xmm0
-; SSE2-NEXT:    movzbl %bpl, %eax
-; SSE2-NEXT:    movd %eax, %xmm2
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm2 = xmm2[0],xmm0[0],xmm2[1],xmm0[1],xmm2[2],xmm0[2],xmm2[3],xmm0[3],xmm2[4],xmm0[4],xmm2[5],xmm0[5],xmm2[6],xmm0[6],xmm2[7],xmm0[7]
-; SSE2-NEXT:    movzbl %r11b, %eax
-; SSE2-NEXT:    movd %eax, %xmm0
-; SSE2-NEXT:    movzbl %r10b, %eax
-; SSE2-NEXT:    movd %eax, %xmm3
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm3 = xmm3[0],xmm0[0],xmm3[1],xmm0[1],xmm3[2],xmm0[2],xmm3[3],xmm0[3],xmm3[4],xmm0[4],xmm3[5],xmm0[5],xmm3[6],xmm0[6],xmm3[7],xmm0[7]
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm3 = xmm3[0],xmm2[0],xmm3[1],xmm2[1],xmm3[2],xmm2[2],xmm3[3],xmm2[3]
-; SSE2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; SSE2-NEXT:    movd %eax, %xmm0
-; SSE2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; SSE2-NEXT:    movd %eax, %xmm2
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm2 = xmm2[0],xmm0[0],xmm2[1],xmm0[1],xmm2[2],xmm0[2],xmm2[3],xmm0[3],xmm2[4],xmm0[4],xmm2[5],xmm0[5],xmm2[6],xmm0[6],xmm2[7],xmm0[7]
-; SSE2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; SSE2-NEXT:    movd %eax, %xmm4
-; SSE2-NEXT:    movzbl %r9b, %eax
-; SSE2-NEXT:    movd %eax, %xmm0
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm4[0],xmm0[1],xmm4[1],xmm0[2],xmm4[2],xmm0[3],xmm4[3],xmm0[4],xmm4[4],xmm0[5],xmm4[5],xmm0[6],xmm4[6],xmm0[7],xmm4[7]
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1],xmm0[2],xmm2[2],xmm0[3],xmm2[3]
-; SSE2-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm3[0],xmm0[1],xmm3[1]
-; SSE2-NEXT:    punpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm1[0]
-; SSE2-NEXT:    popq %rbx
-; SSE2-NEXT:    popq %r12
-; SSE2-NEXT:    popq %r13
-; SSE2-NEXT:    popq %r14
-; SSE2-NEXT:    popq %r15
-; SSE2-NEXT:    popq %rbp
-; SSE2-NEXT:    retq
-;
-; SSSE3-LABEL: v16i8:
-; SSSE3:       # %bb.0:
-; SSSE3-NEXT:    pushq %rbp
-; SSSE3-NEXT:    pushq %r15
-; SSSE3-NEXT:    pushq %r14
-; SSSE3-NEXT:    pushq %r13
-; SSSE3-NEXT:    pushq %r12
-; SSSE3-NEXT:    pushq %rbx
-; SSSE3-NEXT:    movaps %xmm1, -{{[0-9]+}}(%rsp)
-; SSSE3-NEXT:    movaps %xmm0, -{{[0-9]+}}(%rsp)
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %r9b
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %sil
-; SSSE3-NEXT:    movl %r9d, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %r9b
-; SSSE3-NEXT:    jno .LBB0_2
-; SSSE3-NEXT:  # %bb.1:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %r9d
-; SSSE3-NEXT:  .LBB0_2:
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movl %esi, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %sil
-; SSSE3-NEXT:    jno .LBB0_4
-; SSSE3-NEXT:  # %bb.3:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %esi
-; SSSE3-NEXT:  .LBB0_4:
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %bl
-; SSSE3-NEXT:    movl %ebx, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %bl
-; SSSE3-NEXT:    jno .LBB0_6
-; SSSE3-NEXT:  # %bb.5:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %ebx
-; SSSE3-NEXT:  .LBB0_6:
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %dl
-; SSSE3-NEXT:    movl %edx, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %dl
-; SSSE3-NEXT:    jno .LBB0_8
-; SSSE3-NEXT:  # %bb.7:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %edx
-; SSSE3-NEXT:  .LBB0_8:
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %r10b
-; SSSE3-NEXT:    movl %r10d, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %r10b
-; SSSE3-NEXT:    jno .LBB0_10
-; SSSE3-NEXT:  # %bb.9:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %r10d
-; SSSE3-NEXT:  .LBB0_10:
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %r11b
-; SSSE3-NEXT:    movl %r11d, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %r11b
-; SSSE3-NEXT:    jno .LBB0_12
-; SSSE3-NEXT:  # %bb.11:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %r11d
-; SSSE3-NEXT:  .LBB0_12:
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %bpl
-; SSSE3-NEXT:    movl %ebp, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %bpl
-; SSSE3-NEXT:    jno .LBB0_14
-; SSSE3-NEXT:  # %bb.13:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %ebp
-; SSSE3-NEXT:  .LBB0_14:
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %r14b
-; SSSE3-NEXT:    movl %r14d, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %r14b
-; SSSE3-NEXT:    jno .LBB0_16
-; SSSE3-NEXT:  # %bb.15:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %r14d
-; SSSE3-NEXT:  .LBB0_16:
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %r15b
-; SSSE3-NEXT:    movl %r15d, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %r15b
-; SSSE3-NEXT:    jno .LBB0_18
-; SSSE3-NEXT:  # %bb.17:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %r15d
-; SSSE3-NEXT:  .LBB0_18:
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %r12b
-; SSSE3-NEXT:    movl %r12d, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %r12b
-; SSSE3-NEXT:    jno .LBB0_20
-; SSSE3-NEXT:  # %bb.19:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %r12d
-; SSSE3-NEXT:  .LBB0_20:
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %r13b
-; SSSE3-NEXT:    movl %r13d, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %r13b
-; SSSE3-NEXT:    movb %bl, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSSE3-NEXT:    movb %dl, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSSE3-NEXT:    jno .LBB0_22
-; SSSE3-NEXT:  # %bb.21:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %r13d
-; SSSE3-NEXT:  .LBB0_22:
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %dil
-; SSSE3-NEXT:    movl %edi, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %dil
-; SSSE3-NEXT:    jno .LBB0_24
-; SSSE3-NEXT:  # %bb.23:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %edi
-; SSSE3-NEXT:  .LBB0_24:
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %r8b
-; SSSE3-NEXT:    movl %r8d, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %r8b
-; SSSE3-NEXT:    jno .LBB0_26
-; SSSE3-NEXT:  # %bb.25:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %r8d
-; SSSE3-NEXT:  .LBB0_26:
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %bl
-; SSSE3-NEXT:    movl %ebx, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %bl
-; SSSE3-NEXT:    jno .LBB0_28
-; SSSE3-NEXT:  # %bb.27:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %ebx
-; SSSE3-NEXT:  .LBB0_28:
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %dl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %al
-; SSSE3-NEXT:    movl %eax, %ecx
-; SSSE3-NEXT:    addb %dl, %cl
-; SSSE3-NEXT:    setns %cl
-; SSSE3-NEXT:    addb %dl, %al
-; SSSE3-NEXT:    movb %sil, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSSE3-NEXT:    jno .LBB0_30
-; SSSE3-NEXT:  # %bb.29:
-; SSSE3-NEXT:    addb $127, %cl
-; SSSE3-NEXT:    movl %ecx, %eax
-; SSSE3-NEXT:  .LBB0_30:
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %dl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %sil
-; SSSE3-NEXT:    movl %esi, %ecx
-; SSSE3-NEXT:    addb %dl, %cl
-; SSSE3-NEXT:    setns %cl
-; SSSE3-NEXT:    addb %dl, %sil
-; SSSE3-NEXT:    jno .LBB0_32
-; SSSE3-NEXT:  # %bb.31:
-; SSSE3-NEXT:    addb $127, %cl
-; SSSE3-NEXT:    movl %ecx, %esi
-; SSSE3-NEXT:  .LBB0_32:
-; SSSE3-NEXT:    movzbl %sil, %ecx
-; SSSE3-NEXT:    movd %ecx, %xmm0
-; SSSE3-NEXT:    movzbl %al, %eax
-; SSSE3-NEXT:    movd %eax, %xmm1
-; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3],xmm1[4],xmm0[4],xmm1[5],xmm0[5],xmm1[6],xmm0[6],xmm1[7],xmm0[7]
-; SSSE3-NEXT:    movzbl %bl, %eax
-; SSSE3-NEXT:    movd %eax, %xmm0
-; SSSE3-NEXT:    movzbl %r8b, %eax
-; SSSE3-NEXT:    movd %eax, %xmm2
-; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm2 = xmm2[0],xmm0[0],xmm2[1],xmm0[1],xmm2[2],xmm0[2],xmm2[3],xmm0[3],xmm2[4],xmm0[4],xmm2[5],xmm0[5],xmm2[6],xmm0[6],xmm2[7],xmm0[7]
-; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm2 = xmm2[0],xmm1[0],xmm2[1],xmm1[1],xmm2[2],xmm1[2],xmm2[3],xmm1[3]
-; SSSE3-NEXT:    movzbl %dil, %eax
-; SSSE3-NEXT:    movd %eax, %xmm0
-; SSSE3-NEXT:    movzbl %r13b, %eax
-; SSSE3-NEXT:    movd %eax, %xmm3
-; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm3 = xmm3[0],xmm0[0],xmm3[1],xmm0[1],xmm3[2],xmm0[2],xmm3[3],xmm0[3],xmm3[4],xmm0[4],xmm3[5],xmm0[5],xmm3[6],xmm0[6],xmm3[7],xmm0[7]
-; SSSE3-NEXT:    movzbl %r12b, %eax
-; SSSE3-NEXT:    movd %eax, %xmm0
-; SSSE3-NEXT:    movzbl %r15b, %eax
-; SSSE3-NEXT:    movd %eax, %xmm1
-; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3],xmm1[4],xmm0[4],xmm1[5],xmm0[5],xmm1[6],xmm0[6],xmm1[7],xmm0[7]
-; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm3[0],xmm1[1],xmm3[1],xmm1[2],xmm3[2],xmm1[3],xmm3[3]
-; SSSE3-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm2[0],xmm1[1],xmm2[1]
-; SSSE3-NEXT:    movzbl %r14b, %eax
-; SSSE3-NEXT:    movd %eax, %xmm0
-; SSSE3-NEXT:    movzbl %bpl, %eax
-; SSSE3-NEXT:    movd %eax, %xmm2
-; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm2 = xmm2[0],xmm0[0],xmm2[1],xmm0[1],xmm2[2],xmm0[2],xmm2[3],xmm0[3],xmm2[4],xmm0[4],xmm2[5],xmm0[5],xmm2[6],xmm0[6],xmm2[7],xmm0[7]
-; SSSE3-NEXT:    movzbl %r11b, %eax
-; SSSE3-NEXT:    movd %eax, %xmm0
-; SSSE3-NEXT:    movzbl %r10b, %eax
-; SSSE3-NEXT:    movd %eax, %xmm3
-; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm3 = xmm3[0],xmm0[0],xmm3[1],xmm0[1],xmm3[2],xmm0[2],xmm3[3],xmm0[3],xmm3[4],xmm0[4],xmm3[5],xmm0[5],xmm3[6],xmm0[6],xmm3[7],xmm0[7]
-; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm3 = xmm3[0],xmm2[0],xmm3[1],xmm2[1],xmm3[2],xmm2[2],xmm3[3],xmm2[3]
-; SSSE3-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; SSSE3-NEXT:    movd %eax, %xmm0
-; SSSE3-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; SSSE3-NEXT:    movd %eax, %xmm2
-; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm2 = xmm2[0],xmm0[0],xmm2[1],xmm0[1],xmm2[2],xmm0[2],xmm2[3],xmm0[3],xmm2[4],xmm0[4],xmm2[5],xmm0[5],xmm2[6],xmm0[6],xmm2[7],xmm0[7]
-; SSSE3-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; SSSE3-NEXT:    movd %eax, %xmm4
-; SSSE3-NEXT:    movzbl %r9b, %eax
-; SSSE3-NEXT:    movd %eax, %xmm0
-; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm4[0],xmm0[1],xmm4[1],xmm0[2],xmm4[2],xmm0[3],xmm4[3],xmm0[4],xmm4[4],xmm0[5],xmm4[5],xmm0[6],xmm4[6],xmm0[7],xmm4[7]
-; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1],xmm0[2],xmm2[2],xmm0[3],xmm2[3]
-; SSSE3-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm3[0],xmm0[1],xmm3[1]
-; SSSE3-NEXT:    punpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm1[0]
-; SSSE3-NEXT:    popq %rbx
-; SSSE3-NEXT:    popq %r12
-; SSSE3-NEXT:    popq %r13
-; SSSE3-NEXT:    popq %r14
-; SSSE3-NEXT:    popq %r15
-; SSSE3-NEXT:    popq %rbp
-; SSSE3-NEXT:    retq
-;
-; SSE41-LABEL: v16i8:
-; SSE41:       # %bb.0:
-; SSE41-NEXT:    pextrb $15, %xmm1, %ecx
-; SSE41-NEXT:    pextrb $15, %xmm0, %edx
-; SSE41-NEXT:    movl %edx, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %dl
-; SSE41-NEXT:    jno .LBB0_2
-; SSE41-NEXT:  # %bb.1:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %edx
-; SSE41-NEXT:  .LBB0_2:
-; SSE41-NEXT:    pextrb $14, %xmm1, %ecx
-; SSE41-NEXT:    pextrb $14, %xmm0, %r11d
-; SSE41-NEXT:    movl %r11d, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %r11b
-; SSE41-NEXT:    jno .LBB0_4
-; SSE41-NEXT:  # %bb.3:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %r11d
-; SSE41-NEXT:  .LBB0_4:
-; SSE41-NEXT:    pextrb $13, %xmm1, %ecx
-; SSE41-NEXT:    pextrb $13, %xmm0, %edi
-; SSE41-NEXT:    movl %edi, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %dil
-; SSE41-NEXT:    jno .LBB0_6
-; SSE41-NEXT:  # %bb.5:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %edi
-; SSE41-NEXT:  .LBB0_6:
-; SSE41-NEXT:    pushq %rbp
-; SSE41-NEXT:    pushq %r15
-; SSE41-NEXT:    pushq %r14
-; SSE41-NEXT:    pushq %r13
-; SSE41-NEXT:    pushq %r12
-; SSE41-NEXT:    pushq %rbx
-; SSE41-NEXT:    pextrb $12, %xmm1, %ecx
-; SSE41-NEXT:    pextrb $12, %xmm0, %r14d
-; SSE41-NEXT:    movl %r14d, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %r14b
-; SSE41-NEXT:    jno .LBB0_8
-; SSE41-NEXT:  # %bb.7:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %r14d
-; SSE41-NEXT:  .LBB0_8:
-; SSE41-NEXT:    pextrb $11, %xmm1, %ecx
-; SSE41-NEXT:    pextrb $11, %xmm0, %ebp
-; SSE41-NEXT:    movl %ebp, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %bpl
-; SSE41-NEXT:    jno .LBB0_10
-; SSE41-NEXT:  # %bb.9:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %ebp
-; SSE41-NEXT:  .LBB0_10:
-; SSE41-NEXT:    pextrb $10, %xmm1, %ecx
-; SSE41-NEXT:    pextrb $10, %xmm0, %r15d
-; SSE41-NEXT:    movl %r15d, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %r15b
-; SSE41-NEXT:    jno .LBB0_12
-; SSE41-NEXT:  # %bb.11:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %r15d
-; SSE41-NEXT:  .LBB0_12:
-; SSE41-NEXT:    pextrb $9, %xmm1, %ecx
-; SSE41-NEXT:    pextrb $9, %xmm0, %r12d
-; SSE41-NEXT:    movl %r12d, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %r12b
-; SSE41-NEXT:    jno .LBB0_14
-; SSE41-NEXT:  # %bb.13:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %r12d
-; SSE41-NEXT:  .LBB0_14:
-; SSE41-NEXT:    pextrb $8, %xmm1, %ecx
-; SSE41-NEXT:    pextrb $8, %xmm0, %r13d
-; SSE41-NEXT:    movl %r13d, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %r13b
-; SSE41-NEXT:    jno .LBB0_16
-; SSE41-NEXT:  # %bb.15:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %r13d
-; SSE41-NEXT:  .LBB0_16:
-; SSE41-NEXT:    pextrb $7, %xmm1, %ecx
-; SSE41-NEXT:    pextrb $7, %xmm0, %r10d
-; SSE41-NEXT:    movl %r10d, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %r10b
-; SSE41-NEXT:    movl %ebp, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE41-NEXT:    jno .LBB0_18
-; SSE41-NEXT:  # %bb.17:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %r10d
-; SSE41-NEXT:  .LBB0_18:
-; SSE41-NEXT:    pextrb $6, %xmm1, %ecx
-; SSE41-NEXT:    pextrb $6, %xmm0, %r9d
-; SSE41-NEXT:    movl %r9d, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %r9b
-; SSE41-NEXT:    jno .LBB0_20
-; SSE41-NEXT:  # %bb.19:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %r9d
-; SSE41-NEXT:  .LBB0_20:
-; SSE41-NEXT:    pextrb $5, %xmm1, %ecx
-; SSE41-NEXT:    pextrb $5, %xmm0, %ebp
-; SSE41-NEXT:    movl %ebp, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %bpl
-; SSE41-NEXT:    movl %edi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE41-NEXT:    jno .LBB0_22
-; SSE41-NEXT:  # %bb.21:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %ebp
-; SSE41-NEXT:  .LBB0_22:
-; SSE41-NEXT:    pextrb $4, %xmm1, %ecx
-; SSE41-NEXT:    pextrb $4, %xmm0, %edi
-; SSE41-NEXT:    movl %edi, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %dil
-; SSE41-NEXT:    movl %edx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE41-NEXT:    jno .LBB0_24
-; SSE41-NEXT:  # %bb.23:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %edi
-; SSE41-NEXT:  .LBB0_24:
-; SSE41-NEXT:    pextrb $3, %xmm1, %edx
-; SSE41-NEXT:    pextrb $3, %xmm0, %eax
-; SSE41-NEXT:    movl %eax, %ecx
-; SSE41-NEXT:    addb %dl, %cl
-; SSE41-NEXT:    setns %cl
-; SSE41-NEXT:    addb %dl, %al
-; SSE41-NEXT:    jno .LBB0_26
-; SSE41-NEXT:  # %bb.25:
-; SSE41-NEXT:    addb $127, %cl
-; SSE41-NEXT:    movl %ecx, %eax
-; SSE41-NEXT:  .LBB0_26:
-; SSE41-NEXT:    pextrb $2, %xmm1, %ebx
-; SSE41-NEXT:    pextrb $2, %xmm0, %ecx
-; SSE41-NEXT:    movl %ecx, %edx
-; SSE41-NEXT:    addb %bl, %dl
-; SSE41-NEXT:    setns %dl
-; SSE41-NEXT:    addb %bl, %cl
-; SSE41-NEXT:    jno .LBB0_28
-; SSE41-NEXT:  # %bb.27:
-; SSE41-NEXT:    addb $127, %dl
-; SSE41-NEXT:    movl %edx, %ecx
-; SSE41-NEXT:  .LBB0_28:
-; SSE41-NEXT:    pextrb $0, %xmm1, %esi
-; SSE41-NEXT:    pextrb $0, %xmm0, %edx
-; SSE41-NEXT:    movl %edx, %ebx
-; SSE41-NEXT:    addb %sil, %bl
-; SSE41-NEXT:    setns %bl
-; SSE41-NEXT:    addb %sil, %dl
-; SSE41-NEXT:    jno .LBB0_30
-; SSE41-NEXT:  # %bb.29:
-; SSE41-NEXT:    addb $127, %bl
-; SSE41-NEXT:    movl %ebx, %edx
-; SSE41-NEXT:  .LBB0_30:
-; SSE41-NEXT:    pextrb $1, %xmm1, %esi
-; SSE41-NEXT:    pextrb $1, %xmm0, %r8d
-; SSE41-NEXT:    movl %r8d, %ebx
-; SSE41-NEXT:    addb %sil, %bl
-; SSE41-NEXT:    setns %bl
-; SSE41-NEXT:    addb %sil, %r8b
-; SSE41-NEXT:    jno .LBB0_32
-; SSE41-NEXT:  # %bb.31:
-; SSE41-NEXT:    addb $127, %bl
-; SSE41-NEXT:    movl %ebx, %r8d
-; SSE41-NEXT:  .LBB0_32:
-; SSE41-NEXT:    movzbl %dl, %edx
-; SSE41-NEXT:    movd %edx, %xmm0
-; SSE41-NEXT:    movzbl %r8b, %edx
-; SSE41-NEXT:    pinsrb $1, %edx, %xmm0
-; SSE41-NEXT:    movzbl %cl, %ecx
-; SSE41-NEXT:    pinsrb $2, %ecx, %xmm0
-; SSE41-NEXT:    movzbl %al, %eax
-; SSE41-NEXT:    pinsrb $3, %eax, %xmm0
-; SSE41-NEXT:    movzbl %dil, %eax
-; SSE41-NEXT:    pinsrb $4, %eax, %xmm0
-; SSE41-NEXT:    movzbl %bpl, %eax
-; SSE41-NEXT:    pinsrb $5, %eax, %xmm0
-; SSE41-NEXT:    movzbl %r9b, %eax
-; SSE41-NEXT:    pinsrb $6, %eax, %xmm0
-; SSE41-NEXT:    movzbl %r10b, %eax
-; SSE41-NEXT:    pinsrb $7, %eax, %xmm0
-; SSE41-NEXT:    movzbl %r13b, %eax
-; SSE41-NEXT:    pinsrb $8, %eax, %xmm0
-; SSE41-NEXT:    movzbl %r12b, %eax
-; SSE41-NEXT:    pinsrb $9, %eax, %xmm0
-; SSE41-NEXT:    movzbl %r15b, %eax
-; SSE41-NEXT:    pinsrb $10, %eax, %xmm0
-; SSE41-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; SSE41-NEXT:    pinsrb $11, %eax, %xmm0
-; SSE41-NEXT:    movzbl %r14b, %eax
-; SSE41-NEXT:    pinsrb $12, %eax, %xmm0
-; SSE41-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; SSE41-NEXT:    pinsrb $13, %eax, %xmm0
-; SSE41-NEXT:    movzbl %r11b, %eax
-; SSE41-NEXT:    pinsrb $14, %eax, %xmm0
-; SSE41-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; SSE41-NEXT:    pinsrb $15, %eax, %xmm0
-; SSE41-NEXT:    popq %rbx
-; SSE41-NEXT:    popq %r12
-; SSE41-NEXT:    popq %r13
-; SSE41-NEXT:    popq %r14
-; SSE41-NEXT:    popq %r15
-; SSE41-NEXT:    popq %rbp
-; SSE41-NEXT:    retq
+; SSE-LABEL: v16i8:
+; SSE:       # %bb.0:
+; SSE-NEXT:    paddsb %xmm1, %xmm0
+; SSE-NEXT:    retq
 ;
 ; AVX-LABEL: v16i8:
 ; AVX:       # %bb.0:
-; AVX-NEXT:    vpextrb $15, %xmm1, %ecx
-; AVX-NEXT:    vpextrb $15, %xmm0, %edx
-; AVX-NEXT:    movl %edx, %eax
-; AVX-NEXT:    addb %cl, %al
-; AVX-NEXT:    setns %al
-; AVX-NEXT:    addb %cl, %dl
-; AVX-NEXT:    jno .LBB0_2
-; AVX-NEXT:  # %bb.1:
-; AVX-NEXT:    addb $127, %al
-; AVX-NEXT:    movl %eax, %edx
-; AVX-NEXT:  .LBB0_2:
-; AVX-NEXT:    vpextrb $14, %xmm1, %ecx
-; AVX-NEXT:    vpextrb $14, %xmm0, %r11d
-; AVX-NEXT:    movl %r11d, %eax
-; AVX-NEXT:    addb %cl, %al
-; AVX-NEXT:    setns %al
-; AVX-NEXT:    addb %cl, %r11b
-; AVX-NEXT:    jno .LBB0_4
-; AVX-NEXT:  # %bb.3:
-; AVX-NEXT:    addb $127, %al
-; AVX-NEXT:    movl %eax, %r11d
-; AVX-NEXT:  .LBB0_4:
-; AVX-NEXT:    vpextrb $13, %xmm1, %ecx
-; AVX-NEXT:    vpextrb $13, %xmm0, %edi
-; AVX-NEXT:    movl %edi, %eax
-; AVX-NEXT:    addb %cl, %al
-; AVX-NEXT:    setns %al
-; AVX-NEXT:    addb %cl, %dil
-; AVX-NEXT:    jno .LBB0_6
-; AVX-NEXT:  # %bb.5:
-; AVX-NEXT:    addb $127, %al
-; AVX-NEXT:    movl %eax, %edi
-; AVX-NEXT:  .LBB0_6:
-; AVX-NEXT:    pushq %rbp
-; AVX-NEXT:    pushq %r15
-; AVX-NEXT:    pushq %r14
-; AVX-NEXT:    pushq %r13
-; AVX-NEXT:    pushq %r12
-; AVX-NEXT:    pushq %rbx
-; AVX-NEXT:    vpextrb $12, %xmm1, %ecx
-; AVX-NEXT:    vpextrb $12, %xmm0, %r14d
-; AVX-NEXT:    movl %r14d, %eax
-; AVX-NEXT:    addb %cl, %al
-; AVX-NEXT:    setns %al
-; AVX-NEXT:    addb %cl, %r14b
-; AVX-NEXT:    jno .LBB0_8
-; AVX-NEXT:  # %bb.7:
-; AVX-NEXT:    addb $127, %al
-; AVX-NEXT:    movl %eax, %r14d
-; AVX-NEXT:  .LBB0_8:
-; AVX-NEXT:    vpextrb $11, %xmm1, %ecx
-; AVX-NEXT:    vpextrb $11, %xmm0, %ebp
-; AVX-NEXT:    movl %ebp, %eax
-; AVX-NEXT:    addb %cl, %al
-; AVX-NEXT:    setns %al
-; AVX-NEXT:    addb %cl, %bpl
-; AVX-NEXT:    jno .LBB0_10
-; AVX-NEXT:  # %bb.9:
-; AVX-NEXT:    addb $127, %al
-; AVX-NEXT:    movl %eax, %ebp
-; AVX-NEXT:  .LBB0_10:
-; AVX-NEXT:    vpextrb $10, %xmm1, %ecx
-; AVX-NEXT:    vpextrb $10, %xmm0, %r15d
-; AVX-NEXT:    movl %r15d, %eax
-; AVX-NEXT:    addb %cl, %al
-; AVX-NEXT:    setns %al
-; AVX-NEXT:    addb %cl, %r15b
-; AVX-NEXT:    jno .LBB0_12
-; AVX-NEXT:  # %bb.11:
-; AVX-NEXT:    addb $127, %al
-; AVX-NEXT:    movl %eax, %r15d
-; AVX-NEXT:  .LBB0_12:
-; AVX-NEXT:    vpextrb $9, %xmm1, %ecx
-; AVX-NEXT:    vpextrb $9, %xmm0, %r12d
-; AVX-NEXT:    movl %r12d, %eax
-; AVX-NEXT:    addb %cl, %al
-; AVX-NEXT:    setns %al
-; AVX-NEXT:    addb %cl, %r12b
-; AVX-NEXT:    jno .LBB0_14
-; AVX-NEXT:  # %bb.13:
-; AVX-NEXT:    addb $127, %al
-; AVX-NEXT:    movl %eax, %r12d
-; AVX-NEXT:  .LBB0_14:
-; AVX-NEXT:    vpextrb $8, %xmm1, %ecx
-; AVX-NEXT:    vpextrb $8, %xmm0, %r13d
-; AVX-NEXT:    movl %r13d, %eax
-; AVX-NEXT:    addb %cl, %al
-; AVX-NEXT:    setns %al
-; AVX-NEXT:    addb %cl, %r13b
-; AVX-NEXT:    jno .LBB0_16
-; AVX-NEXT:  # %bb.15:
-; AVX-NEXT:    addb $127, %al
-; AVX-NEXT:    movl %eax, %r13d
-; AVX-NEXT:  .LBB0_16:
-; AVX-NEXT:    vpextrb $7, %xmm1, %ecx
-; AVX-NEXT:    vpextrb $7, %xmm0, %r10d
-; AVX-NEXT:    movl %r10d, %eax
-; AVX-NEXT:    addb %cl, %al
-; AVX-NEXT:    setns %al
-; AVX-NEXT:    addb %cl, %r10b
-; AVX-NEXT:    movl %ebp, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX-NEXT:    jno .LBB0_18
-; AVX-NEXT:  # %bb.17:
-; AVX-NEXT:    addb $127, %al
-; AVX-NEXT:    movl %eax, %r10d
-; AVX-NEXT:  .LBB0_18:
-; AVX-NEXT:    vpextrb $6, %xmm1, %ecx
-; AVX-NEXT:    vpextrb $6, %xmm0, %r9d
-; AVX-NEXT:    movl %r9d, %eax
-; AVX-NEXT:    addb %cl, %al
-; AVX-NEXT:    setns %al
-; AVX-NEXT:    addb %cl, %r9b
-; AVX-NEXT:    jno .LBB0_20
-; AVX-NEXT:  # %bb.19:
-; AVX-NEXT:    addb $127, %al
-; AVX-NEXT:    movl %eax, %r9d
-; AVX-NEXT:  .LBB0_20:
-; AVX-NEXT:    vpextrb $5, %xmm1, %ecx
-; AVX-NEXT:    vpextrb $5, %xmm0, %ebp
-; AVX-NEXT:    movl %ebp, %eax
-; AVX-NEXT:    addb %cl, %al
-; AVX-NEXT:    setns %al
-; AVX-NEXT:    addb %cl, %bpl
-; AVX-NEXT:    movl %edi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX-NEXT:    jno .LBB0_22
-; AVX-NEXT:  # %bb.21:
-; AVX-NEXT:    addb $127, %al
-; AVX-NEXT:    movl %eax, %ebp
-; AVX-NEXT:  .LBB0_22:
-; AVX-NEXT:    vpextrb $4, %xmm1, %ecx
-; AVX-NEXT:    vpextrb $4, %xmm0, %edi
-; AVX-NEXT:    movl %edi, %eax
-; AVX-NEXT:    addb %cl, %al
-; AVX-NEXT:    setns %al
-; AVX-NEXT:    addb %cl, %dil
-; AVX-NEXT:    movl %edx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX-NEXT:    jno .LBB0_24
-; AVX-NEXT:  # %bb.23:
-; AVX-NEXT:    addb $127, %al
-; AVX-NEXT:    movl %eax, %edi
-; AVX-NEXT:  .LBB0_24:
-; AVX-NEXT:    vpextrb $3, %xmm1, %edx
-; AVX-NEXT:    vpextrb $3, %xmm0, %eax
-; AVX-NEXT:    movl %eax, %ecx
-; AVX-NEXT:    addb %dl, %cl
-; AVX-NEXT:    setns %cl
-; AVX-NEXT:    addb %dl, %al
-; AVX-NEXT:    jno .LBB0_26
-; AVX-NEXT:  # %bb.25:
-; AVX-NEXT:    addb $127, %cl
-; AVX-NEXT:    movl %ecx, %eax
-; AVX-NEXT:  .LBB0_26:
-; AVX-NEXT:    vpextrb $2, %xmm1, %ebx
-; AVX-NEXT:    vpextrb $2, %xmm0, %ecx
-; AVX-NEXT:    movl %ecx, %edx
-; AVX-NEXT:    addb %bl, %dl
-; AVX-NEXT:    setns %dl
-; AVX-NEXT:    addb %bl, %cl
-; AVX-NEXT:    jno .LBB0_28
-; AVX-NEXT:  # %bb.27:
-; AVX-NEXT:    addb $127, %dl
-; AVX-NEXT:    movl %edx, %ecx
-; AVX-NEXT:  .LBB0_28:
-; AVX-NEXT:    vpextrb $0, %xmm1, %esi
-; AVX-NEXT:    vpextrb $0, %xmm0, %edx
-; AVX-NEXT:    movl %edx, %ebx
-; AVX-NEXT:    addb %sil, %bl
-; AVX-NEXT:    setns %bl
-; AVX-NEXT:    addb %sil, %dl
-; AVX-NEXT:    jno .LBB0_30
-; AVX-NEXT:  # %bb.29:
-; AVX-NEXT:    addb $127, %bl
-; AVX-NEXT:    movl %ebx, %edx
-; AVX-NEXT:  .LBB0_30:
-; AVX-NEXT:    vpextrb $1, %xmm1, %esi
-; AVX-NEXT:    vpextrb $1, %xmm0, %r8d
-; AVX-NEXT:    movl %r8d, %ebx
-; AVX-NEXT:    addb %sil, %bl
-; AVX-NEXT:    setns %bl
-; AVX-NEXT:    addb %sil, %r8b
-; AVX-NEXT:    jno .LBB0_32
-; AVX-NEXT:  # %bb.31:
-; AVX-NEXT:    addb $127, %bl
-; AVX-NEXT:    movl %ebx, %r8d
-; AVX-NEXT:  .LBB0_32:
-; AVX-NEXT:    movzbl %dl, %edx
-; AVX-NEXT:    vmovd %edx, %xmm0
-; AVX-NEXT:    movzbl %r8b, %edx
-; AVX-NEXT:    vpinsrb $1, %edx, %xmm0, %xmm0
-; AVX-NEXT:    movzbl %cl, %ecx
-; AVX-NEXT:    vpinsrb $2, %ecx, %xmm0, %xmm0
-; AVX-NEXT:    movzbl %al, %eax
-; AVX-NEXT:    vpinsrb $3, %eax, %xmm0, %xmm0
-; AVX-NEXT:    movzbl %dil, %eax
-; AVX-NEXT:    vpinsrb $4, %eax, %xmm0, %xmm0
-; AVX-NEXT:    movzbl %bpl, %eax
-; AVX-NEXT:    vpinsrb $5, %eax, %xmm0, %xmm0
-; AVX-NEXT:    movzbl %r9b, %eax
-; AVX-NEXT:    vpinsrb $6, %eax, %xmm0, %xmm0
-; AVX-NEXT:    movzbl %r10b, %eax
-; AVX-NEXT:    vpinsrb $7, %eax, %xmm0, %xmm0
-; AVX-NEXT:    movzbl %r13b, %eax
-; AVX-NEXT:    vpinsrb $8, %eax, %xmm0, %xmm0
-; AVX-NEXT:    movzbl %r12b, %eax
-; AVX-NEXT:    vpinsrb $9, %eax, %xmm0, %xmm0
-; AVX-NEXT:    movzbl %r15b, %eax
-; AVX-NEXT:    vpinsrb $10, %eax, %xmm0, %xmm0
-; AVX-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX-NEXT:    vpinsrb $11, %eax, %xmm0, %xmm0
-; AVX-NEXT:    movzbl %r14b, %eax
-; AVX-NEXT:    vpinsrb $12, %eax, %xmm0, %xmm0
-; AVX-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX-NEXT:    vpinsrb $13, %eax, %xmm0, %xmm0
-; AVX-NEXT:    movzbl %r11b, %eax
-; AVX-NEXT:    vpinsrb $14, %eax, %xmm0, %xmm0
-; AVX-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX-NEXT:    vpinsrb $15, %eax, %xmm0, %xmm0
-; AVX-NEXT:    popq %rbx
-; AVX-NEXT:    popq %r12
-; AVX-NEXT:    popq %r13
-; AVX-NEXT:    popq %r14
-; AVX-NEXT:    popq %r15
-; AVX-NEXT:    popq %rbp
+; AVX-NEXT:    vpaddsb %xmm1, %xmm0, %xmm0
 ; AVX-NEXT:    retq
   %z = call <16 x i8> @llvm.sadd.sat.v16i8(<16 x i8> %x, <16 x i8> %y)
   ret <16 x i8> %z
 }
 
 define <32 x i8> @v32i8(<32 x i8> %x, <32 x i8> %y) nounwind {
-; SSE2-LABEL: v32i8:
-; SSE2:       # %bb.0:
-; SSE2-NEXT:    pushq %rbp
-; SSE2-NEXT:    pushq %r15
-; SSE2-NEXT:    pushq %r14
-; SSE2-NEXT:    pushq %r13
-; SSE2-NEXT:    pushq %r12
-; SSE2-NEXT:    pushq %rbx
-; SSE2-NEXT:    pushq %rax
-; SSE2-NEXT:    movaps %xmm3, -{{[0-9]+}}(%rsp)
-; SSE2-NEXT:    movaps %xmm1, -{{[0-9]+}}(%rsp)
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %r8b
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %r11b
-; SSE2-NEXT:    movl %r8d, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %r8b
-; SSE2-NEXT:    jno .LBB1_2
-; SSE2-NEXT:  # %bb.1:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %r8d
-; SSE2-NEXT:  .LBB1_2:
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movl %r11d, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %r11b
-; SSE2-NEXT:    jno .LBB1_4
-; SSE2-NEXT:  # %bb.3:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %r11d
-; SSE2-NEXT:  .LBB1_4:
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %bl
-; SSE2-NEXT:    movl %ebx, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %bl
-; SSE2-NEXT:    jno .LBB1_6
-; SSE2-NEXT:  # %bb.5:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %ebx
-; SSE2-NEXT:  .LBB1_6:
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %sil
-; SSE2-NEXT:    movl %esi, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %sil
-; SSE2-NEXT:    jno .LBB1_8
-; SSE2-NEXT:  # %bb.7:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %esi
-; SSE2-NEXT:  .LBB1_8:
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %dl
-; SSE2-NEXT:    movl %edx, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %dl
-; SSE2-NEXT:    jo .LBB1_9
-; SSE2-NEXT:  # %bb.10:
-; SSE2-NEXT:    movb %dl, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSE2-NEXT:    jmp .LBB1_11
-; SSE2-NEXT:  .LBB1_9:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movb %al, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSE2-NEXT:  .LBB1_11:
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %dil
-; SSE2-NEXT:    movl %edi, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %dil
-; SSE2-NEXT:    movb %r8b, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSE2-NEXT:    movb %bl, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSE2-NEXT:    jno .LBB1_13
-; SSE2-NEXT:  # %bb.12:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %edi
-; SSE2-NEXT:  .LBB1_13:
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %r9b
-; SSE2-NEXT:    movl %r9d, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %r9b
-; SSE2-NEXT:    jno .LBB1_15
-; SSE2-NEXT:  # %bb.14:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %r9d
-; SSE2-NEXT:  .LBB1_15:
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %r10b
-; SSE2-NEXT:    movl %r10d, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %r10b
-; SSE2-NEXT:    movb %r11b, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSE2-NEXT:    jno .LBB1_17
-; SSE2-NEXT:  # %bb.16:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %r10d
-; SSE2-NEXT:  .LBB1_17:
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %bpl
-; SSE2-NEXT:    movl %ebp, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %bpl
-; SSE2-NEXT:    jno .LBB1_19
-; SSE2-NEXT:  # %bb.18:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %ebp
-; SSE2-NEXT:  .LBB1_19:
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %r14b
-; SSE2-NEXT:    movl %r14d, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %r14b
-; SSE2-NEXT:    jno .LBB1_21
-; SSE2-NEXT:  # %bb.20:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %r14d
-; SSE2-NEXT:  .LBB1_21:
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %r11b
-; SSE2-NEXT:    movl %r11d, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %r11b
-; SSE2-NEXT:    jno .LBB1_23
-; SSE2-NEXT:  # %bb.22:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %r11d
-; SSE2-NEXT:  .LBB1_23:
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %r8b
-; SSE2-NEXT:    movl %r8d, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %r8b
-; SSE2-NEXT:    jno .LBB1_25
-; SSE2-NEXT:  # %bb.24:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %r8d
-; SSE2-NEXT:  .LBB1_25:
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %r15b
-; SSE2-NEXT:    movl %r15d, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %r15b
-; SSE2-NEXT:    jno .LBB1_27
-; SSE2-NEXT:  # %bb.26:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %r15d
-; SSE2-NEXT:  .LBB1_27:
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %r12b
-; SSE2-NEXT:    movl %r12d, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %r12b
-; SSE2-NEXT:    jno .LBB1_29
-; SSE2-NEXT:  # %bb.28:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %r12d
-; SSE2-NEXT:  .LBB1_29:
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %r13b
-; SSE2-NEXT:    movl %r13d, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %r13b
-; SSE2-NEXT:    jno .LBB1_31
-; SSE2-NEXT:  # %bb.30:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %r13d
-; SSE2-NEXT:  .LBB1_31:
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %dl
-; SSE2-NEXT:    movl %edx, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %dl
-; SSE2-NEXT:    movb %dl, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSE2-NEXT:    movaps %xmm2, -{{[0-9]+}}(%rsp)
-; SSE2-NEXT:    movaps %xmm0, -{{[0-9]+}}(%rsp)
-; SSE2-NEXT:    movb %sil, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSE2-NEXT:    jno .LBB1_33
-; SSE2-NEXT:  # %bb.32:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movb %al, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSE2-NEXT:  .LBB1_33:
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %sil
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %dl
-; SSE2-NEXT:    movl %esi, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %sil
-; SSE2-NEXT:    jno .LBB1_35
-; SSE2-NEXT:  # %bb.34:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %esi
-; SSE2-NEXT:  .LBB1_35:
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movl %edx, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %dl
-; SSE2-NEXT:    movb %dl, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSE2-NEXT:    jno .LBB1_37
-; SSE2-NEXT:  # %bb.36:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movb %al, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSE2-NEXT:  .LBB1_37:
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %dl
-; SSE2-NEXT:    movl %edx, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %dl
-; SSE2-NEXT:    jno .LBB1_39
-; SSE2-NEXT:  # %bb.38:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %edx
-; SSE2-NEXT:  .LBB1_39:
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %bl
-; SSE2-NEXT:    movl %ebx, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %bl
-; SSE2-NEXT:    movb %bl, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSE2-NEXT:    jno .LBB1_41
-; SSE2-NEXT:  # %bb.40:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movb %al, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSE2-NEXT:  .LBB1_41:
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %bl
-; SSE2-NEXT:    movl %ebx, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %bl
-; SSE2-NEXT:    movb %bpl, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSE2-NEXT:    movb %r14b, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSE2-NEXT:    movb %r11b, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSE2-NEXT:    jno .LBB1_43
-; SSE2-NEXT:  # %bb.42:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %ebx
-; SSE2-NEXT:  .LBB1_43:
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %bpl
-; SSE2-NEXT:    movl %ebp, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %bpl
-; SSE2-NEXT:    movb %r8b, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSE2-NEXT:    jno .LBB1_45
-; SSE2-NEXT:  # %bb.44:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %ebp
-; SSE2-NEXT:  .LBB1_45:
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %r14b
-; SSE2-NEXT:    movl %r14d, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %r14b
-; SSE2-NEXT:    movb %r15b, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSE2-NEXT:    jno .LBB1_47
-; SSE2-NEXT:  # %bb.46:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %r14d
-; SSE2-NEXT:  .LBB1_47:
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %r15b
-; SSE2-NEXT:    movl %r15d, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %r15b
-; SSE2-NEXT:    movb %r12b, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSE2-NEXT:    jno .LBB1_49
-; SSE2-NEXT:  # %bb.48:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %r15d
-; SSE2-NEXT:  .LBB1_49:
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %r12b
-; SSE2-NEXT:    movl %r12d, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %r12b
-; SSE2-NEXT:    movb %r13b, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSE2-NEXT:    jno .LBB1_51
-; SSE2-NEXT:  # %bb.50:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %r12d
-; SSE2-NEXT:  .LBB1_51:
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %r13b
-; SSE2-NEXT:    movl %r13d, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %r13b
-; SSE2-NEXT:    movb %dil, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSE2-NEXT:    jno .LBB1_53
-; SSE2-NEXT:  # %bb.52:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %r13d
-; SSE2-NEXT:  .LBB1_53:
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %r8b
-; SSE2-NEXT:    movl %r8d, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %r8b
-; SSE2-NEXT:    movb %r9b, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSE2-NEXT:    jno .LBB1_55
-; SSE2-NEXT:  # %bb.54:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %r8d
-; SSE2-NEXT:  .LBB1_55:
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %r9b
-; SSE2-NEXT:    movl %r9d, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %r9b
-; SSE2-NEXT:    movb %r10b, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSE2-NEXT:    jno .LBB1_57
-; SSE2-NEXT:  # %bb.56:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %r9d
-; SSE2-NEXT:  .LBB1_57:
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %r10b
-; SSE2-NEXT:    movl %r10d, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %r10b
-; SSE2-NEXT:    movb %sil, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSE2-NEXT:    movb %dl, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSE2-NEXT:    jno .LBB1_59
-; SSE2-NEXT:  # %bb.58:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %r10d
-; SSE2-NEXT:  .LBB1_59:
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %dl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %r11b
-; SSE2-NEXT:    movl %r11d, %ecx
-; SSE2-NEXT:    addb %dl, %cl
-; SSE2-NEXT:    setns %cl
-; SSE2-NEXT:    addb %dl, %r11b
-; SSE2-NEXT:    movb %bl, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSE2-NEXT:    jno .LBB1_61
-; SSE2-NEXT:  # %bb.60:
-; SSE2-NEXT:    addb $127, %cl
-; SSE2-NEXT:    movl %ecx, %r11d
-; SSE2-NEXT:  .LBB1_61:
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %bl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movl %ecx, %edx
-; SSE2-NEXT:    addb %bl, %dl
-; SSE2-NEXT:    setns %dl
-; SSE2-NEXT:    addb %bl, %cl
-; SSE2-NEXT:    jno .LBB1_63
-; SSE2-NEXT:  # %bb.62:
-; SSE2-NEXT:    addb $127, %dl
-; SSE2-NEXT:    movl %edx, %ecx
-; SSE2-NEXT:  .LBB1_63:
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %bl
-; SSE2-NEXT:    movl %ebx, %edx
-; SSE2-NEXT:    addb %al, %dl
-; SSE2-NEXT:    setns %dl
-; SSE2-NEXT:    addb %al, %bl
-; SSE2-NEXT:    jno .LBB1_65
-; SSE2-NEXT:  # %bb.64:
-; SSE2-NEXT:    addb $127, %dl
-; SSE2-NEXT:    movl %edx, %ebx
-; SSE2-NEXT:  .LBB1_65:
-; SSE2-NEXT:    movzbl %bl, %esi
-; SSE2-NEXT:    movzbl %cl, %edi
-; SSE2-NEXT:    movzbl %r11b, %r11d
-; SSE2-NEXT:    movzbl %r10b, %r10d
-; SSE2-NEXT:    movzbl %r9b, %r9d
-; SSE2-NEXT:    movzbl %r8b, %r8d
-; SSE2-NEXT:    movzbl %r13b, %r13d
-; SSE2-NEXT:    movzbl %r12b, %eax
-; SSE2-NEXT:    movzbl %r15b, %ebx
-; SSE2-NEXT:    movzbl %r14b, %edx
-; SSE2-NEXT:    movzbl %bpl, %ebp
-; SSE2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %ecx # 1-byte Folded Reload
-; SSE2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %r12d # 1-byte Folded Reload
-; SSE2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %r15d # 1-byte Folded Reload
-; SSE2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %r14d # 1-byte Folded Reload
-; SSE2-NEXT:    movd %esi, %xmm12
-; SSE2-NEXT:    movd %edi, %xmm6
-; SSE2-NEXT:    movd %r11d, %xmm11
-; SSE2-NEXT:    movd %r10d, %xmm2
-; SSE2-NEXT:    movd %r9d, %xmm10
-; SSE2-NEXT:    movd %r8d, %xmm5
-; SSE2-NEXT:    movd %r13d, %xmm9
-; SSE2-NEXT:    movd %eax, %xmm1
-; SSE2-NEXT:    movd %ebx, %xmm8
-; SSE2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %r13d # 1-byte Folded Reload
-; SSE2-NEXT:    movd %edx, %xmm14
-; SSE2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %r11d # 1-byte Folded Reload
-; SSE2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %r8d # 1-byte Folded Reload
-; SSE2-NEXT:    movd %ebp, %xmm13
-; SSE2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %r9d # 1-byte Folded Reload
-; SSE2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %r10d # 1-byte Folded Reload
-; SSE2-NEXT:    movd %ecx, %xmm7
-; SSE2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %ebx # 1-byte Folded Reload
-; SSE2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; SSE2-NEXT:    movd %r12d, %xmm0
-; SSE2-NEXT:    movdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; SSE2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %r12d # 1-byte Folded Reload
-; SSE2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %esi # 1-byte Folded Reload
-; SSE2-NEXT:    movd %r15d, %xmm4
-; SSE2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %r15d # 1-byte Folded Reload
-; SSE2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %ebp # 1-byte Folded Reload
-; SSE2-NEXT:    movd %r14d, %xmm15
-; SSE2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %r14d # 1-byte Folded Reload
-; SSE2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %edx # 1-byte Folded Reload
-; SSE2-NEXT:    movd %r13d, %xmm0
-; SSE2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %r13d # 1-byte Folded Reload
-; SSE2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %edi # 1-byte Folded Reload
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm6 = xmm6[0],xmm12[0],xmm6[1],xmm12[1],xmm6[2],xmm12[2],xmm6[3],xmm12[3],xmm6[4],xmm12[4],xmm6[5],xmm12[5],xmm6[6],xmm12[6],xmm6[7],xmm12[7]
-; SSE2-NEXT:    movd %r11d, %xmm3
-; SSE2-NEXT:    movdqa %xmm3, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; SSE2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %r11d # 1-byte Folded Reload
-; SSE2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %ecx # 1-byte Folded Reload
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm2 = xmm2[0],xmm11[0],xmm2[1],xmm11[1],xmm2[2],xmm11[2],xmm2[3],xmm11[3],xmm2[4],xmm11[4],xmm2[5],xmm11[5],xmm2[6],xmm11[6],xmm2[7],xmm11[7]
-; SSE2-NEXT:    movd %r8d, %xmm11
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm2 = xmm2[0],xmm6[0],xmm2[1],xmm6[1],xmm2[2],xmm6[2],xmm2[3],xmm6[3]
-; SSE2-NEXT:    movd %r9d, %xmm12
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm5 = xmm5[0],xmm10[0],xmm5[1],xmm10[1],xmm5[2],xmm10[2],xmm5[3],xmm10[3],xmm5[4],xmm10[4],xmm5[5],xmm10[5],xmm5[6],xmm10[6],xmm5[7],xmm10[7]
-; SSE2-NEXT:    movd %r10d, %xmm10
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm9[0],xmm1[1],xmm9[1],xmm1[2],xmm9[2],xmm1[3],xmm9[3],xmm1[4],xmm9[4],xmm1[5],xmm9[5],xmm1[6],xmm9[6],xmm1[7],xmm9[7]
-; SSE2-NEXT:    movd %ebx, %xmm9
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm5[0],xmm1[1],xmm5[1],xmm1[2],xmm5[2],xmm1[3],xmm5[3]
-; SSE2-NEXT:    movd %eax, %xmm5
-; SSE2-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm2[0],xmm1[1],xmm2[1]
-; SSE2-NEXT:    movd %r12d, %xmm6
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm14 = xmm14[0],xmm8[0],xmm14[1],xmm8[1],xmm14[2],xmm8[2],xmm14[3],xmm8[3],xmm14[4],xmm8[4],xmm14[5],xmm8[5],xmm14[6],xmm8[6],xmm14[7],xmm8[7]
-; SSE2-NEXT:    movd %esi, %xmm8
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm7 = xmm7[0],xmm13[0],xmm7[1],xmm13[1],xmm7[2],xmm13[2],xmm7[3],xmm13[3],xmm7[4],xmm13[4],xmm7[5],xmm13[5],xmm7[6],xmm13[6],xmm7[7],xmm13[7]
-; SSE2-NEXT:    movd %r15d, %xmm13
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm7 = xmm7[0],xmm14[0],xmm7[1],xmm14[1],xmm7[2],xmm14[2],xmm7[3],xmm14[3]
-; SSE2-NEXT:    movd %ebp, %xmm3
-; SSE2-NEXT:    punpcklbw {{[-0-9]+}}(%r{{[sb]}}p), %xmm4 # 16-byte Folded Reload
-; SSE2-NEXT:    # xmm4 = xmm4[0],mem[0],xmm4[1],mem[1],xmm4[2],mem[2],xmm4[3],mem[3],xmm4[4],mem[4],xmm4[5],mem[5],xmm4[6],mem[6],xmm4[7],mem[7]
-; SSE2-NEXT:    movd %r14d, %xmm14
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm15[0],xmm0[1],xmm15[1],xmm0[2],xmm15[2],xmm0[3],xmm15[3],xmm0[4],xmm15[4],xmm0[5],xmm15[5],xmm0[6],xmm15[6],xmm0[7],xmm15[7]
-; SSE2-NEXT:    movd %edx, %xmm15
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm4[0],xmm0[1],xmm4[1],xmm0[2],xmm4[2],xmm0[3],xmm4[3]
-; SSE2-NEXT:    movd %r13d, %xmm4
-; SSE2-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm7[0],xmm0[1],xmm7[1]
-; SSE2-NEXT:    movd %edi, %xmm7
-; SSE2-NEXT:    punpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm1[0]
-; SSE2-NEXT:    movd %r11d, %xmm2
-; SSE2-NEXT:    punpcklbw {{[-0-9]+}}(%r{{[sb]}}p), %xmm11 # 16-byte Folded Reload
-; SSE2-NEXT:    # xmm11 = xmm11[0],mem[0],xmm11[1],mem[1],xmm11[2],mem[2],xmm11[3],mem[3],xmm11[4],mem[4],xmm11[5],mem[5],xmm11[6],mem[6],xmm11[7],mem[7]
-; SSE2-NEXT:    movd %ecx, %xmm1
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm10 = xmm10[0],xmm12[0],xmm10[1],xmm12[1],xmm10[2],xmm12[2],xmm10[3],xmm12[3],xmm10[4],xmm12[4],xmm10[5],xmm12[5],xmm10[6],xmm12[6],xmm10[7],xmm12[7]
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm10 = xmm10[0],xmm11[0],xmm10[1],xmm11[1],xmm10[2],xmm11[2],xmm10[3],xmm11[3]
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm5 = xmm5[0],xmm9[0],xmm5[1],xmm9[1],xmm5[2],xmm9[2],xmm5[3],xmm9[3],xmm5[4],xmm9[4],xmm5[5],xmm9[5],xmm5[6],xmm9[6],xmm5[7],xmm9[7]
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm8 = xmm8[0],xmm6[0],xmm8[1],xmm6[1],xmm8[2],xmm6[2],xmm8[3],xmm6[3],xmm8[4],xmm6[4],xmm8[5],xmm6[5],xmm8[6],xmm6[6],xmm8[7],xmm6[7]
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm8 = xmm8[0],xmm5[0],xmm8[1],xmm5[1],xmm8[2],xmm5[2],xmm8[3],xmm5[3]
-; SSE2-NEXT:    punpckldq {{.*#+}} xmm8 = xmm8[0],xmm10[0],xmm8[1],xmm10[1]
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm3 = xmm3[0],xmm13[0],xmm3[1],xmm13[1],xmm3[2],xmm13[2],xmm3[3],xmm13[3],xmm3[4],xmm13[4],xmm3[5],xmm13[5],xmm3[6],xmm13[6],xmm3[7],xmm13[7]
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm15 = xmm15[0],xmm14[0],xmm15[1],xmm14[1],xmm15[2],xmm14[2],xmm15[3],xmm14[3],xmm15[4],xmm14[4],xmm15[5],xmm14[5],xmm15[6],xmm14[6],xmm15[7],xmm14[7]
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm15 = xmm15[0],xmm3[0],xmm15[1],xmm3[1],xmm15[2],xmm3[2],xmm15[3],xmm3[3]
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm7 = xmm7[0],xmm4[0],xmm7[1],xmm4[1],xmm7[2],xmm4[2],xmm7[3],xmm4[3],xmm7[4],xmm4[4],xmm7[5],xmm4[5],xmm7[6],xmm4[6],xmm7[7],xmm4[7]
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm2[0],xmm1[1],xmm2[1],xmm1[2],xmm2[2],xmm1[3],xmm2[3],xmm1[4],xmm2[4],xmm1[5],xmm2[5],xmm1[6],xmm2[6],xmm1[7],xmm2[7]
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm7[0],xmm1[1],xmm7[1],xmm1[2],xmm7[2],xmm1[3],xmm7[3]
-; SSE2-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm15[0],xmm1[1],xmm15[1]
-; SSE2-NEXT:    punpcklqdq {{.*#+}} xmm1 = xmm1[0],xmm8[0]
-; SSE2-NEXT:    addq $8, %rsp
-; SSE2-NEXT:    popq %rbx
-; SSE2-NEXT:    popq %r12
-; SSE2-NEXT:    popq %r13
-; SSE2-NEXT:    popq %r14
-; SSE2-NEXT:    popq %r15
-; SSE2-NEXT:    popq %rbp
-; SSE2-NEXT:    retq
-;
-; SSSE3-LABEL: v32i8:
-; SSSE3:       # %bb.0:
-; SSSE3-NEXT:    pushq %rbp
-; SSSE3-NEXT:    pushq %r15
-; SSSE3-NEXT:    pushq %r14
-; SSSE3-NEXT:    pushq %r13
-; SSSE3-NEXT:    pushq %r12
-; SSSE3-NEXT:    pushq %rbx
-; SSSE3-NEXT:    pushq %rax
-; SSSE3-NEXT:    movaps %xmm3, -{{[0-9]+}}(%rsp)
-; SSSE3-NEXT:    movaps %xmm1, -{{[0-9]+}}(%rsp)
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %r8b
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %r11b
-; SSSE3-NEXT:    movl %r8d, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %r8b
-; SSSE3-NEXT:    jno .LBB1_2
-; SSSE3-NEXT:  # %bb.1:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %r8d
-; SSSE3-NEXT:  .LBB1_2:
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movl %r11d, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %r11b
-; SSSE3-NEXT:    jno .LBB1_4
-; SSSE3-NEXT:  # %bb.3:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %r11d
-; SSSE3-NEXT:  .LBB1_4:
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %bl
-; SSSE3-NEXT:    movl %ebx, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %bl
-; SSSE3-NEXT:    jno .LBB1_6
-; SSSE3-NEXT:  # %bb.5:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %ebx
-; SSSE3-NEXT:  .LBB1_6:
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %sil
-; SSSE3-NEXT:    movl %esi, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %sil
-; SSSE3-NEXT:    jno .LBB1_8
-; SSSE3-NEXT:  # %bb.7:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %esi
-; SSSE3-NEXT:  .LBB1_8:
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %dl
-; SSSE3-NEXT:    movl %edx, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %dl
-; SSSE3-NEXT:    jo .LBB1_9
-; SSSE3-NEXT:  # %bb.10:
-; SSSE3-NEXT:    movb %dl, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSSE3-NEXT:    jmp .LBB1_11
-; SSSE3-NEXT:  .LBB1_9:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movb %al, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSSE3-NEXT:  .LBB1_11:
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %dil
-; SSSE3-NEXT:    movl %edi, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %dil
-; SSSE3-NEXT:    movb %r8b, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSSE3-NEXT:    movb %bl, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSSE3-NEXT:    jno .LBB1_13
-; SSSE3-NEXT:  # %bb.12:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %edi
-; SSSE3-NEXT:  .LBB1_13:
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %r9b
-; SSSE3-NEXT:    movl %r9d, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %r9b
-; SSSE3-NEXT:    jno .LBB1_15
-; SSSE3-NEXT:  # %bb.14:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %r9d
-; SSSE3-NEXT:  .LBB1_15:
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %r10b
-; SSSE3-NEXT:    movl %r10d, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %r10b
-; SSSE3-NEXT:    movb %r11b, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSSE3-NEXT:    jno .LBB1_17
-; SSSE3-NEXT:  # %bb.16:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %r10d
-; SSSE3-NEXT:  .LBB1_17:
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %bpl
-; SSSE3-NEXT:    movl %ebp, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %bpl
-; SSSE3-NEXT:    jno .LBB1_19
-; SSSE3-NEXT:  # %bb.18:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %ebp
-; SSSE3-NEXT:  .LBB1_19:
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %r14b
-; SSSE3-NEXT:    movl %r14d, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %r14b
-; SSSE3-NEXT:    jno .LBB1_21
-; SSSE3-NEXT:  # %bb.20:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %r14d
-; SSSE3-NEXT:  .LBB1_21:
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %r11b
-; SSSE3-NEXT:    movl %r11d, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %r11b
-; SSSE3-NEXT:    jno .LBB1_23
-; SSSE3-NEXT:  # %bb.22:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %r11d
-; SSSE3-NEXT:  .LBB1_23:
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %r8b
-; SSSE3-NEXT:    movl %r8d, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %r8b
-; SSSE3-NEXT:    jno .LBB1_25
-; SSSE3-NEXT:  # %bb.24:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %r8d
-; SSSE3-NEXT:  .LBB1_25:
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %r15b
-; SSSE3-NEXT:    movl %r15d, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %r15b
-; SSSE3-NEXT:    jno .LBB1_27
-; SSSE3-NEXT:  # %bb.26:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %r15d
-; SSSE3-NEXT:  .LBB1_27:
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %r12b
-; SSSE3-NEXT:    movl %r12d, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %r12b
-; SSSE3-NEXT:    jno .LBB1_29
-; SSSE3-NEXT:  # %bb.28:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %r12d
-; SSSE3-NEXT:  .LBB1_29:
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %r13b
-; SSSE3-NEXT:    movl %r13d, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %r13b
-; SSSE3-NEXT:    jno .LBB1_31
-; SSSE3-NEXT:  # %bb.30:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %r13d
-; SSSE3-NEXT:  .LBB1_31:
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %dl
-; SSSE3-NEXT:    movl %edx, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %dl
-; SSSE3-NEXT:    movb %dl, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSSE3-NEXT:    movaps %xmm2, -{{[0-9]+}}(%rsp)
-; SSSE3-NEXT:    movaps %xmm0, -{{[0-9]+}}(%rsp)
-; SSSE3-NEXT:    movb %sil, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSSE3-NEXT:    jno .LBB1_33
-; SSSE3-NEXT:  # %bb.32:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movb %al, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSSE3-NEXT:  .LBB1_33:
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %sil
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %dl
-; SSSE3-NEXT:    movl %esi, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %sil
-; SSSE3-NEXT:    jno .LBB1_35
-; SSSE3-NEXT:  # %bb.34:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %esi
-; SSSE3-NEXT:  .LBB1_35:
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movl %edx, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %dl
-; SSSE3-NEXT:    movb %dl, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSSE3-NEXT:    jno .LBB1_37
-; SSSE3-NEXT:  # %bb.36:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movb %al, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSSE3-NEXT:  .LBB1_37:
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %dl
-; SSSE3-NEXT:    movl %edx, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %dl
-; SSSE3-NEXT:    jno .LBB1_39
-; SSSE3-NEXT:  # %bb.38:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %edx
-; SSSE3-NEXT:  .LBB1_39:
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %bl
-; SSSE3-NEXT:    movl %ebx, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %bl
-; SSSE3-NEXT:    movb %bl, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSSE3-NEXT:    jno .LBB1_41
-; SSSE3-NEXT:  # %bb.40:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movb %al, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSSE3-NEXT:  .LBB1_41:
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %bl
-; SSSE3-NEXT:    movl %ebx, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %bl
-; SSSE3-NEXT:    movb %bpl, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSSE3-NEXT:    movb %r14b, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSSE3-NEXT:    movb %r11b, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSSE3-NEXT:    jno .LBB1_43
-; SSSE3-NEXT:  # %bb.42:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %ebx
-; SSSE3-NEXT:  .LBB1_43:
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %bpl
-; SSSE3-NEXT:    movl %ebp, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %bpl
-; SSSE3-NEXT:    movb %r8b, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSSE3-NEXT:    jno .LBB1_45
-; SSSE3-NEXT:  # %bb.44:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %ebp
-; SSSE3-NEXT:  .LBB1_45:
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %r14b
-; SSSE3-NEXT:    movl %r14d, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %r14b
-; SSSE3-NEXT:    movb %r15b, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSSE3-NEXT:    jno .LBB1_47
-; SSSE3-NEXT:  # %bb.46:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %r14d
-; SSSE3-NEXT:  .LBB1_47:
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %r15b
-; SSSE3-NEXT:    movl %r15d, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %r15b
-; SSSE3-NEXT:    movb %r12b, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSSE3-NEXT:    jno .LBB1_49
-; SSSE3-NEXT:  # %bb.48:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %r15d
-; SSSE3-NEXT:  .LBB1_49:
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %r12b
-; SSSE3-NEXT:    movl %r12d, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %r12b
-; SSSE3-NEXT:    movb %r13b, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSSE3-NEXT:    jno .LBB1_51
-; SSSE3-NEXT:  # %bb.50:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %r12d
-; SSSE3-NEXT:  .LBB1_51:
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %r13b
-; SSSE3-NEXT:    movl %r13d, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %r13b
-; SSSE3-NEXT:    movb %dil, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSSE3-NEXT:    jno .LBB1_53
-; SSSE3-NEXT:  # %bb.52:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %r13d
-; SSSE3-NEXT:  .LBB1_53:
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %r8b
-; SSSE3-NEXT:    movl %r8d, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %r8b
-; SSSE3-NEXT:    movb %r9b, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSSE3-NEXT:    jno .LBB1_55
-; SSSE3-NEXT:  # %bb.54:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %r8d
-; SSSE3-NEXT:  .LBB1_55:
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %r9b
-; SSSE3-NEXT:    movl %r9d, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %r9b
-; SSSE3-NEXT:    movb %r10b, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSSE3-NEXT:    jno .LBB1_57
-; SSSE3-NEXT:  # %bb.56:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %r9d
-; SSSE3-NEXT:  .LBB1_57:
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %r10b
-; SSSE3-NEXT:    movl %r10d, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %r10b
-; SSSE3-NEXT:    movb %sil, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSSE3-NEXT:    movb %dl, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSSE3-NEXT:    jno .LBB1_59
-; SSSE3-NEXT:  # %bb.58:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %r10d
-; SSSE3-NEXT:  .LBB1_59:
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %dl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %r11b
-; SSSE3-NEXT:    movl %r11d, %ecx
-; SSSE3-NEXT:    addb %dl, %cl
-; SSSE3-NEXT:    setns %cl
-; SSSE3-NEXT:    addb %dl, %r11b
-; SSSE3-NEXT:    movb %bl, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSSE3-NEXT:    jno .LBB1_61
-; SSSE3-NEXT:  # %bb.60:
-; SSSE3-NEXT:    addb $127, %cl
-; SSSE3-NEXT:    movl %ecx, %r11d
-; SSSE3-NEXT:  .LBB1_61:
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %bl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movl %ecx, %edx
-; SSSE3-NEXT:    addb %bl, %dl
-; SSSE3-NEXT:    setns %dl
-; SSSE3-NEXT:    addb %bl, %cl
-; SSSE3-NEXT:    jno .LBB1_63
-; SSSE3-NEXT:  # %bb.62:
-; SSSE3-NEXT:    addb $127, %dl
-; SSSE3-NEXT:    movl %edx, %ecx
-; SSSE3-NEXT:  .LBB1_63:
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %al
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %bl
-; SSSE3-NEXT:    movl %ebx, %edx
-; SSSE3-NEXT:    addb %al, %dl
-; SSSE3-NEXT:    setns %dl
-; SSSE3-NEXT:    addb %al, %bl
-; SSSE3-NEXT:    jno .LBB1_65
-; SSSE3-NEXT:  # %bb.64:
-; SSSE3-NEXT:    addb $127, %dl
-; SSSE3-NEXT:    movl %edx, %ebx
-; SSSE3-NEXT:  .LBB1_65:
-; SSSE3-NEXT:    movzbl %bl, %esi
-; SSSE3-NEXT:    movzbl %cl, %edi
-; SSSE3-NEXT:    movzbl %r11b, %r11d
-; SSSE3-NEXT:    movzbl %r10b, %r10d
-; SSSE3-NEXT:    movzbl %r9b, %r9d
-; SSSE3-NEXT:    movzbl %r8b, %r8d
-; SSSE3-NEXT:    movzbl %r13b, %r13d
-; SSSE3-NEXT:    movzbl %r12b, %eax
-; SSSE3-NEXT:    movzbl %r15b, %ebx
-; SSSE3-NEXT:    movzbl %r14b, %edx
-; SSSE3-NEXT:    movzbl %bpl, %ebp
-; SSSE3-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %ecx # 1-byte Folded Reload
-; SSSE3-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %r12d # 1-byte Folded Reload
-; SSSE3-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %r15d # 1-byte Folded Reload
-; SSSE3-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %r14d # 1-byte Folded Reload
-; SSSE3-NEXT:    movd %esi, %xmm12
-; SSSE3-NEXT:    movd %edi, %xmm6
-; SSSE3-NEXT:    movd %r11d, %xmm11
-; SSSE3-NEXT:    movd %r10d, %xmm2
-; SSSE3-NEXT:    movd %r9d, %xmm10
-; SSSE3-NEXT:    movd %r8d, %xmm5
-; SSSE3-NEXT:    movd %r13d, %xmm9
-; SSSE3-NEXT:    movd %eax, %xmm1
-; SSSE3-NEXT:    movd %ebx, %xmm8
-; SSSE3-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %r13d # 1-byte Folded Reload
-; SSSE3-NEXT:    movd %edx, %xmm14
-; SSSE3-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %r11d # 1-byte Folded Reload
-; SSSE3-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %r8d # 1-byte Folded Reload
-; SSSE3-NEXT:    movd %ebp, %xmm13
-; SSSE3-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %r9d # 1-byte Folded Reload
-; SSSE3-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %r10d # 1-byte Folded Reload
-; SSSE3-NEXT:    movd %ecx, %xmm7
-; SSSE3-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %ebx # 1-byte Folded Reload
-; SSSE3-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; SSSE3-NEXT:    movd %r12d, %xmm0
-; SSSE3-NEXT:    movdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; SSSE3-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %r12d # 1-byte Folded Reload
-; SSSE3-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %esi # 1-byte Folded Reload
-; SSSE3-NEXT:    movd %r15d, %xmm4
-; SSSE3-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %r15d # 1-byte Folded Reload
-; SSSE3-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %ebp # 1-byte Folded Reload
-; SSSE3-NEXT:    movd %r14d, %xmm15
-; SSSE3-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %r14d # 1-byte Folded Reload
-; SSSE3-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %edx # 1-byte Folded Reload
-; SSSE3-NEXT:    movd %r13d, %xmm0
-; SSSE3-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %r13d # 1-byte Folded Reload
-; SSSE3-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %edi # 1-byte Folded Reload
-; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm6 = xmm6[0],xmm12[0],xmm6[1],xmm12[1],xmm6[2],xmm12[2],xmm6[3],xmm12[3],xmm6[4],xmm12[4],xmm6[5],xmm12[5],xmm6[6],xmm12[6],xmm6[7],xmm12[7]
-; SSSE3-NEXT:    movd %r11d, %xmm3
-; SSSE3-NEXT:    movdqa %xmm3, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; SSSE3-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %r11d # 1-byte Folded Reload
-; SSSE3-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %ecx # 1-byte Folded Reload
-; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm2 = xmm2[0],xmm11[0],xmm2[1],xmm11[1],xmm2[2],xmm11[2],xmm2[3],xmm11[3],xmm2[4],xmm11[4],xmm2[5],xmm11[5],xmm2[6],xmm11[6],xmm2[7],xmm11[7]
-; SSSE3-NEXT:    movd %r8d, %xmm11
-; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm2 = xmm2[0],xmm6[0],xmm2[1],xmm6[1],xmm2[2],xmm6[2],xmm2[3],xmm6[3]
-; SSSE3-NEXT:    movd %r9d, %xmm12
-; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm5 = xmm5[0],xmm10[0],xmm5[1],xmm10[1],xmm5[2],xmm10[2],xmm5[3],xmm10[3],xmm5[4],xmm10[4],xmm5[5],xmm10[5],xmm5[6],xmm10[6],xmm5[7],xmm10[7]
-; SSSE3-NEXT:    movd %r10d, %xmm10
-; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm9[0],xmm1[1],xmm9[1],xmm1[2],xmm9[2],xmm1[3],xmm9[3],xmm1[4],xmm9[4],xmm1[5],xmm9[5],xmm1[6],xmm9[6],xmm1[7],xmm9[7]
-; SSSE3-NEXT:    movd %ebx, %xmm9
-; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm5[0],xmm1[1],xmm5[1],xmm1[2],xmm5[2],xmm1[3],xmm5[3]
-; SSSE3-NEXT:    movd %eax, %xmm5
-; SSSE3-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm2[0],xmm1[1],xmm2[1]
-; SSSE3-NEXT:    movd %r12d, %xmm6
-; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm14 = xmm14[0],xmm8[0],xmm14[1],xmm8[1],xmm14[2],xmm8[2],xmm14[3],xmm8[3],xmm14[4],xmm8[4],xmm14[5],xmm8[5],xmm14[6],xmm8[6],xmm14[7],xmm8[7]
-; SSSE3-NEXT:    movd %esi, %xmm8
-; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm7 = xmm7[0],xmm13[0],xmm7[1],xmm13[1],xmm7[2],xmm13[2],xmm7[3],xmm13[3],xmm7[4],xmm13[4],xmm7[5],xmm13[5],xmm7[6],xmm13[6],xmm7[7],xmm13[7]
-; SSSE3-NEXT:    movd %r15d, %xmm13
-; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm7 = xmm7[0],xmm14[0],xmm7[1],xmm14[1],xmm7[2],xmm14[2],xmm7[3],xmm14[3]
-; SSSE3-NEXT:    movd %ebp, %xmm3
-; SSSE3-NEXT:    punpcklbw {{[-0-9]+}}(%r{{[sb]}}p), %xmm4 # 16-byte Folded Reload
-; SSSE3-NEXT:    # xmm4 = xmm4[0],mem[0],xmm4[1],mem[1],xmm4[2],mem[2],xmm4[3],mem[3],xmm4[4],mem[4],xmm4[5],mem[5],xmm4[6],mem[6],xmm4[7],mem[7]
-; SSSE3-NEXT:    movd %r14d, %xmm14
-; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm15[0],xmm0[1],xmm15[1],xmm0[2],xmm15[2],xmm0[3],xmm15[3],xmm0[4],xmm15[4],xmm0[5],xmm15[5],xmm0[6],xmm15[6],xmm0[7],xmm15[7]
-; SSSE3-NEXT:    movd %edx, %xmm15
-; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm4[0],xmm0[1],xmm4[1],xmm0[2],xmm4[2],xmm0[3],xmm4[3]
-; SSSE3-NEXT:    movd %r13d, %xmm4
-; SSSE3-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm7[0],xmm0[1],xmm7[1]
-; SSSE3-NEXT:    movd %edi, %xmm7
-; SSSE3-NEXT:    punpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm1[0]
-; SSSE3-NEXT:    movd %r11d, %xmm2
-; SSSE3-NEXT:    punpcklbw {{[-0-9]+}}(%r{{[sb]}}p), %xmm11 # 16-byte Folded Reload
-; SSSE3-NEXT:    # xmm11 = xmm11[0],mem[0],xmm11[1],mem[1],xmm11[2],mem[2],xmm11[3],mem[3],xmm11[4],mem[4],xmm11[5],mem[5],xmm11[6],mem[6],xmm11[7],mem[7]
-; SSSE3-NEXT:    movd %ecx, %xmm1
-; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm10 = xmm10[0],xmm12[0],xmm10[1],xmm12[1],xmm10[2],xmm12[2],xmm10[3],xmm12[3],xmm10[4],xmm12[4],xmm10[5],xmm12[5],xmm10[6],xmm12[6],xmm10[7],xmm12[7]
-; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm10 = xmm10[0],xmm11[0],xmm10[1],xmm11[1],xmm10[2],xmm11[2],xmm10[3],xmm11[3]
-; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm5 = xmm5[0],xmm9[0],xmm5[1],xmm9[1],xmm5[2],xmm9[2],xmm5[3],xmm9[3],xmm5[4],xmm9[4],xmm5[5],xmm9[5],xmm5[6],xmm9[6],xmm5[7],xmm9[7]
-; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm8 = xmm8[0],xmm6[0],xmm8[1],xmm6[1],xmm8[2],xmm6[2],xmm8[3],xmm6[3],xmm8[4],xmm6[4],xmm8[5],xmm6[5],xmm8[6],xmm6[6],xmm8[7],xmm6[7]
-; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm8 = xmm8[0],xmm5[0],xmm8[1],xmm5[1],xmm8[2],xmm5[2],xmm8[3],xmm5[3]
-; SSSE3-NEXT:    punpckldq {{.*#+}} xmm8 = xmm8[0],xmm10[0],xmm8[1],xmm10[1]
-; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm3 = xmm3[0],xmm13[0],xmm3[1],xmm13[1],xmm3[2],xmm13[2],xmm3[3],xmm13[3],xmm3[4],xmm13[4],xmm3[5],xmm13[5],xmm3[6],xmm13[6],xmm3[7],xmm13[7]
-; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm15 = xmm15[0],xmm14[0],xmm15[1],xmm14[1],xmm15[2],xmm14[2],xmm15[3],xmm14[3],xmm15[4],xmm14[4],xmm15[5],xmm14[5],xmm15[6],xmm14[6],xmm15[7],xmm14[7]
-; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm15 = xmm15[0],xmm3[0],xmm15[1],xmm3[1],xmm15[2],xmm3[2],xmm15[3],xmm3[3]
-; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm7 = xmm7[0],xmm4[0],xmm7[1],xmm4[1],xmm7[2],xmm4[2],xmm7[3],xmm4[3],xmm7[4],xmm4[4],xmm7[5],xmm4[5],xmm7[6],xmm4[6],xmm7[7],xmm4[7]
-; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm2[0],xmm1[1],xmm2[1],xmm1[2],xmm2[2],xmm1[3],xmm2[3],xmm1[4],xmm2[4],xmm1[5],xmm2[5],xmm1[6],xmm2[6],xmm1[7],xmm2[7]
-; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm7[0],xmm1[1],xmm7[1],xmm1[2],xmm7[2],xmm1[3],xmm7[3]
-; SSSE3-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm15[0],xmm1[1],xmm15[1]
-; SSSE3-NEXT:    punpcklqdq {{.*#+}} xmm1 = xmm1[0],xmm8[0]
-; SSSE3-NEXT:    addq $8, %rsp
-; SSSE3-NEXT:    popq %rbx
-; SSSE3-NEXT:    popq %r12
-; SSSE3-NEXT:    popq %r13
-; SSSE3-NEXT:    popq %r14
-; SSSE3-NEXT:    popq %r15
-; SSSE3-NEXT:    popq %rbp
-; SSSE3-NEXT:    retq
-;
-; SSE41-LABEL: v32i8:
-; SSE41:       # %bb.0:
-; SSE41-NEXT:    pextrb $15, %xmm3, %ecx
-; SSE41-NEXT:    pextrb $15, %xmm1, %edx
-; SSE41-NEXT:    movl %edx, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %dl
-; SSE41-NEXT:    jno .LBB1_2
-; SSE41-NEXT:  # %bb.1:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %edx
-; SSE41-NEXT:  .LBB1_2:
-; SSE41-NEXT:    pextrb $14, %xmm3, %ecx
-; SSE41-NEXT:    pextrb $14, %xmm1, %esi
-; SSE41-NEXT:    movl %esi, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %sil
-; SSE41-NEXT:    jno .LBB1_4
-; SSE41-NEXT:  # %bb.3:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %esi
-; SSE41-NEXT:  .LBB1_4:
-; SSE41-NEXT:    pushq %rbp
-; SSE41-NEXT:    pushq %r15
-; SSE41-NEXT:    pushq %r14
-; SSE41-NEXT:    pushq %r13
-; SSE41-NEXT:    pushq %r12
-; SSE41-NEXT:    pushq %rbx
-; SSE41-NEXT:    pextrb $13, %xmm3, %ecx
-; SSE41-NEXT:    pextrb $13, %xmm1, %edi
-; SSE41-NEXT:    movl %edi, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %dil
-; SSE41-NEXT:    jo .LBB1_5
-; SSE41-NEXT:  # %bb.6:
-; SSE41-NEXT:    movl %edi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE41-NEXT:    jmp .LBB1_7
-; SSE41-NEXT:  .LBB1_5:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    # kill: def $al killed $al def $eax
-; SSE41-NEXT:    movl %eax, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE41-NEXT:  .LBB1_7:
-; SSE41-NEXT:    pextrb $12, %xmm3, %ecx
-; SSE41-NEXT:    pextrb $12, %xmm1, %edi
-; SSE41-NEXT:    movl %edi, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %dil
-; SSE41-NEXT:    jno .LBB1_9
-; SSE41-NEXT:  # %bb.8:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %edi
-; SSE41-NEXT:  .LBB1_9:
-; SSE41-NEXT:    pextrb $11, %xmm3, %ecx
-; SSE41-NEXT:    pextrb $11, %xmm1, %ebp
-; SSE41-NEXT:    movl %ebp, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %bpl
-; SSE41-NEXT:    jno .LBB1_11
-; SSE41-NEXT:  # %bb.10:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %ebp
-; SSE41-NEXT:  .LBB1_11:
-; SSE41-NEXT:    pextrb $10, %xmm3, %ecx
-; SSE41-NEXT:    pextrb $10, %xmm1, %ebx
-; SSE41-NEXT:    movl %ebx, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %bl
-; SSE41-NEXT:    movl %ebp, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE41-NEXT:    jno .LBB1_13
-; SSE41-NEXT:  # %bb.12:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %ebx
-; SSE41-NEXT:  .LBB1_13:
-; SSE41-NEXT:    pextrb $9, %xmm3, %ecx
-; SSE41-NEXT:    pextrb $9, %xmm1, %ebp
-; SSE41-NEXT:    movl %ebp, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %bpl
-; SSE41-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE41-NEXT:    jno .LBB1_15
-; SSE41-NEXT:  # %bb.14:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %ebp
-; SSE41-NEXT:  .LBB1_15:
-; SSE41-NEXT:    pextrb $8, %xmm3, %ecx
-; SSE41-NEXT:    pextrb $8, %xmm1, %esi
-; SSE41-NEXT:    movl %esi, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %sil
-; SSE41-NEXT:    movl %edx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE41-NEXT:    jno .LBB1_17
-; SSE41-NEXT:  # %bb.16:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %esi
-; SSE41-NEXT:  .LBB1_17:
-; SSE41-NEXT:    pextrb $7, %xmm3, %ecx
-; SSE41-NEXT:    pextrb $7, %xmm1, %edx
-; SSE41-NEXT:    movl %edx, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %dl
-; SSE41-NEXT:    movl %edi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE41-NEXT:    jno .LBB1_19
-; SSE41-NEXT:  # %bb.18:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %edx
-; SSE41-NEXT:  .LBB1_19:
-; SSE41-NEXT:    pextrb $6, %xmm3, %ecx
-; SSE41-NEXT:    pextrb $6, %xmm1, %edi
-; SSE41-NEXT:    movl %edi, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %dil
-; SSE41-NEXT:    movl %ebx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE41-NEXT:    movl %ebp, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE41-NEXT:    jno .LBB1_21
-; SSE41-NEXT:  # %bb.20:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %edi
-; SSE41-NEXT:  .LBB1_21:
-; SSE41-NEXT:    pextrb $5, %xmm3, %ecx
-; SSE41-NEXT:    pextrb $5, %xmm1, %r8d
-; SSE41-NEXT:    movl %r8d, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %r8b
-; SSE41-NEXT:    jno .LBB1_23
-; SSE41-NEXT:  # %bb.22:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %r8d
-; SSE41-NEXT:  .LBB1_23:
-; SSE41-NEXT:    pextrb $4, %xmm3, %ecx
-; SSE41-NEXT:    pextrb $4, %xmm1, %r11d
-; SSE41-NEXT:    movl %r11d, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %r11b
-; SSE41-NEXT:    jno .LBB1_25
-; SSE41-NEXT:  # %bb.24:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %r11d
-; SSE41-NEXT:  .LBB1_25:
-; SSE41-NEXT:    pextrb $3, %xmm3, %ecx
-; SSE41-NEXT:    pextrb $3, %xmm1, %r10d
-; SSE41-NEXT:    movl %r10d, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %r10b
-; SSE41-NEXT:    jno .LBB1_27
-; SSE41-NEXT:  # %bb.26:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %r10d
-; SSE41-NEXT:  .LBB1_27:
-; SSE41-NEXT:    pextrb $2, %xmm3, %ecx
-; SSE41-NEXT:    pextrb $2, %xmm1, %r14d
-; SSE41-NEXT:    movl %r14d, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %r14b
-; SSE41-NEXT:    jno .LBB1_29
-; SSE41-NEXT:  # %bb.28:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %r14d
-; SSE41-NEXT:  .LBB1_29:
-; SSE41-NEXT:    pextrb $0, %xmm3, %ecx
-; SSE41-NEXT:    pextrb $0, %xmm1, %r9d
-; SSE41-NEXT:    movl %r9d, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %r9b
-; SSE41-NEXT:    jno .LBB1_31
-; SSE41-NEXT:  # %bb.30:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %r9d
-; SSE41-NEXT:  .LBB1_31:
-; SSE41-NEXT:    pextrb $1, %xmm3, %ecx
-; SSE41-NEXT:    pextrb $1, %xmm1, %ebp
-; SSE41-NEXT:    movl %ebp, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %bpl
-; SSE41-NEXT:    jno .LBB1_33
-; SSE41-NEXT:  # %bb.32:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %ebp
-; SSE41-NEXT:  .LBB1_33:
-; SSE41-NEXT:    pextrb $15, %xmm2, %ecx
-; SSE41-NEXT:    pextrb $15, %xmm0, %ebx
-; SSE41-NEXT:    movl %ebx, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    movl %esi, %r15d
-; SSE41-NEXT:    addb %cl, %bl
-; SSE41-NEXT:    jno .LBB1_35
-; SSE41-NEXT:  # %bb.34:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %ebx
-; SSE41-NEXT:  .LBB1_35:
-; SSE41-NEXT:    movl %edi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE41-NEXT:    pextrb $14, %xmm2, %ecx
-; SSE41-NEXT:    pextrb $14, %xmm0, %esi
-; SSE41-NEXT:    movl %esi, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    movl %edx, %edi
-; SSE41-NEXT:    addb %cl, %sil
-; SSE41-NEXT:    jno .LBB1_37
-; SSE41-NEXT:  # %bb.36:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %esi
-; SSE41-NEXT:  .LBB1_37:
-; SSE41-NEXT:    pextrb $13, %xmm2, %ecx
-; SSE41-NEXT:    pextrb $13, %xmm0, %edx
-; SSE41-NEXT:    movl %edx, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %dl
-; SSE41-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE41-NEXT:    jo .LBB1_38
-; SSE41-NEXT:  # %bb.39:
-; SSE41-NEXT:    movl %edx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE41-NEXT:    jmp .LBB1_40
-; SSE41-NEXT:  .LBB1_38:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    # kill: def $al killed $al def $eax
-; SSE41-NEXT:    movl %eax, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE41-NEXT:  .LBB1_40:
-; SSE41-NEXT:    movl %edi, %edx
-; SSE41-NEXT:    pextrb $12, %xmm2, %ecx
-; SSE41-NEXT:    pextrb $12, %xmm0, %edi
-; SSE41-NEXT:    movl %edi, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %dil
-; SSE41-NEXT:    movl %r15d, %esi
-; SSE41-NEXT:    jno .LBB1_42
-; SSE41-NEXT:  # %bb.41:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %edi
-; SSE41-NEXT:  .LBB1_42:
-; SSE41-NEXT:    pextrb $11, %xmm2, %ecx
-; SSE41-NEXT:    pextrb $11, %xmm0, %r15d
-; SSE41-NEXT:    movl %r15d, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %r15b
-; SSE41-NEXT:    jno .LBB1_44
-; SSE41-NEXT:  # %bb.43:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %r15d
-; SSE41-NEXT:  .LBB1_44:
-; SSE41-NEXT:    pextrb $10, %xmm2, %ecx
-; SSE41-NEXT:    pextrb $10, %xmm0, %r12d
-; SSE41-NEXT:    movl %r12d, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %r12b
-; SSE41-NEXT:    jno .LBB1_46
-; SSE41-NEXT:  # %bb.45:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %r12d
-; SSE41-NEXT:  .LBB1_46:
-; SSE41-NEXT:    pextrb $9, %xmm2, %ecx
-; SSE41-NEXT:    pextrb $9, %xmm0, %r13d
-; SSE41-NEXT:    movl %r13d, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %r13b
-; SSE41-NEXT:    movl %r11d, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE41-NEXT:    movl %r10d, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE41-NEXT:    movl %r14d, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE41-NEXT:    movl %r9d, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE41-NEXT:    movl %ebp, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE41-NEXT:    movl %ebx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE41-NEXT:    jno .LBB1_48
-; SSE41-NEXT:  # %bb.47:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %r13d
-; SSE41-NEXT:  .LBB1_48:
-; SSE41-NEXT:    pextrb $8, %xmm2, %ecx
-; SSE41-NEXT:    pextrb $8, %xmm0, %r11d
-; SSE41-NEXT:    movl %r11d, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %r11b
-; SSE41-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE41-NEXT:    movl %edx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE41-NEXT:    movl %r8d, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE41-NEXT:    jno .LBB1_50
-; SSE41-NEXT:  # %bb.49:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %r11d
-; SSE41-NEXT:  .LBB1_50:
-; SSE41-NEXT:    pextrb $7, %xmm2, %ecx
-; SSE41-NEXT:    pextrb $7, %xmm0, %r10d
-; SSE41-NEXT:    movl %r10d, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %r10b
-; SSE41-NEXT:    jno .LBB1_52
-; SSE41-NEXT:  # %bb.51:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %r10d
-; SSE41-NEXT:  .LBB1_52:
-; SSE41-NEXT:    pextrb $6, %xmm2, %ecx
-; SSE41-NEXT:    pextrb $6, %xmm0, %ebp
-; SSE41-NEXT:    movl %ebp, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    movl %edi, %r14d
-; SSE41-NEXT:    addb %cl, %bpl
-; SSE41-NEXT:    jno .LBB1_54
-; SSE41-NEXT:  # %bb.53:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %ebp
-; SSE41-NEXT:  .LBB1_54:
-; SSE41-NEXT:    pextrb $5, %xmm2, %ecx
-; SSE41-NEXT:    pextrb $5, %xmm0, %edi
-; SSE41-NEXT:    movl %edi, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %dil
-; SSE41-NEXT:    jno .LBB1_56
-; SSE41-NEXT:  # %bb.55:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %edi
-; SSE41-NEXT:  .LBB1_56:
-; SSE41-NEXT:    pextrb $4, %xmm2, %edx
-; SSE41-NEXT:    pextrb $4, %xmm0, %eax
-; SSE41-NEXT:    movl %eax, %ecx
-; SSE41-NEXT:    addb %dl, %cl
-; SSE41-NEXT:    setns %cl
-; SSE41-NEXT:    addb %dl, %al
-; SSE41-NEXT:    jno .LBB1_58
-; SSE41-NEXT:  # %bb.57:
-; SSE41-NEXT:    addb $127, %cl
-; SSE41-NEXT:    movl %ecx, %eax
-; SSE41-NEXT:  .LBB1_58:
-; SSE41-NEXT:    pextrb $3, %xmm2, %ebx
-; SSE41-NEXT:    pextrb $3, %xmm0, %ecx
-; SSE41-NEXT:    movl %ecx, %edx
-; SSE41-NEXT:    addb %bl, %dl
-; SSE41-NEXT:    setns %dl
-; SSE41-NEXT:    addb %bl, %cl
-; SSE41-NEXT:    jno .LBB1_60
-; SSE41-NEXT:  # %bb.59:
-; SSE41-NEXT:    addb $127, %dl
-; SSE41-NEXT:    movl %edx, %ecx
-; SSE41-NEXT:  .LBB1_60:
-; SSE41-NEXT:    pextrb $2, %xmm2, %esi
-; SSE41-NEXT:    pextrb $2, %xmm0, %edx
-; SSE41-NEXT:    movl %edx, %ebx
-; SSE41-NEXT:    addb %sil, %bl
-; SSE41-NEXT:    setns %bl
-; SSE41-NEXT:    addb %sil, %dl
-; SSE41-NEXT:    jno .LBB1_62
-; SSE41-NEXT:  # %bb.61:
-; SSE41-NEXT:    addb $127, %bl
-; SSE41-NEXT:    movl %ebx, %edx
-; SSE41-NEXT:  .LBB1_62:
-; SSE41-NEXT:    pextrb $0, %xmm2, %esi
-; SSE41-NEXT:    pextrb $0, %xmm0, %r8d
-; SSE41-NEXT:    movl %r8d, %ebx
-; SSE41-NEXT:    addb %sil, %bl
-; SSE41-NEXT:    setns %bl
-; SSE41-NEXT:    addb %sil, %r8b
-; SSE41-NEXT:    jno .LBB1_64
-; SSE41-NEXT:  # %bb.63:
-; SSE41-NEXT:    addb $127, %bl
-; SSE41-NEXT:    movl %ebx, %r8d
-; SSE41-NEXT:  .LBB1_64:
-; SSE41-NEXT:    pextrb $1, %xmm2, %esi
-; SSE41-NEXT:    pextrb $1, %xmm0, %r9d
-; SSE41-NEXT:    movl %r9d, %ebx
-; SSE41-NEXT:    addb %sil, %bl
-; SSE41-NEXT:    setns %bl
-; SSE41-NEXT:    addb %sil, %r9b
-; SSE41-NEXT:    jno .LBB1_66
-; SSE41-NEXT:  # %bb.65:
-; SSE41-NEXT:    addb $127, %bl
-; SSE41-NEXT:    movl %ebx, %r9d
-; SSE41-NEXT:  .LBB1_66:
-; SSE41-NEXT:    movzbl %r8b, %esi
-; SSE41-NEXT:    movd %esi, %xmm0
-; SSE41-NEXT:    movzbl %r9b, %esi
-; SSE41-NEXT:    pinsrb $1, %esi, %xmm0
-; SSE41-NEXT:    movzbl %dl, %edx
-; SSE41-NEXT:    pinsrb $2, %edx, %xmm0
-; SSE41-NEXT:    movzbl %cl, %ecx
-; SSE41-NEXT:    pinsrb $3, %ecx, %xmm0
-; SSE41-NEXT:    movzbl %al, %eax
-; SSE41-NEXT:    pinsrb $4, %eax, %xmm0
-; SSE41-NEXT:    movzbl %dil, %eax
-; SSE41-NEXT:    pinsrb $5, %eax, %xmm0
-; SSE41-NEXT:    movzbl %bpl, %eax
-; SSE41-NEXT:    pinsrb $6, %eax, %xmm0
-; SSE41-NEXT:    movzbl %r10b, %eax
-; SSE41-NEXT:    pinsrb $7, %eax, %xmm0
-; SSE41-NEXT:    movzbl %r11b, %eax
-; SSE41-NEXT:    pinsrb $8, %eax, %xmm0
-; SSE41-NEXT:    movzbl %r13b, %eax
-; SSE41-NEXT:    pinsrb $9, %eax, %xmm0
-; SSE41-NEXT:    movzbl %r12b, %eax
-; SSE41-NEXT:    pinsrb $10, %eax, %xmm0
-; SSE41-NEXT:    movzbl %r15b, %eax
-; SSE41-NEXT:    pinsrb $11, %eax, %xmm0
-; SSE41-NEXT:    movzbl %r14b, %eax
-; SSE41-NEXT:    pinsrb $12, %eax, %xmm0
-; SSE41-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; SSE41-NEXT:    pinsrb $13, %eax, %xmm0
-; SSE41-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; SSE41-NEXT:    pinsrb $14, %eax, %xmm0
-; SSE41-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; SSE41-NEXT:    pinsrb $15, %eax, %xmm0
-; SSE41-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; SSE41-NEXT:    movd %eax, %xmm1
-; SSE41-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; SSE41-NEXT:    pinsrb $1, %eax, %xmm1
-; SSE41-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; SSE41-NEXT:    pinsrb $2, %eax, %xmm1
-; SSE41-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; SSE41-NEXT:    pinsrb $3, %eax, %xmm1
-; SSE41-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; SSE41-NEXT:    pinsrb $4, %eax, %xmm1
-; SSE41-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; SSE41-NEXT:    pinsrb $5, %eax, %xmm1
-; SSE41-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; SSE41-NEXT:    pinsrb $6, %eax, %xmm1
-; SSE41-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; SSE41-NEXT:    pinsrb $7, %eax, %xmm1
-; SSE41-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; SSE41-NEXT:    pinsrb $8, %eax, %xmm1
-; SSE41-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; SSE41-NEXT:    pinsrb $9, %eax, %xmm1
-; SSE41-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; SSE41-NEXT:    pinsrb $10, %eax, %xmm1
-; SSE41-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; SSE41-NEXT:    pinsrb $11, %eax, %xmm1
-; SSE41-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; SSE41-NEXT:    pinsrb $12, %eax, %xmm1
-; SSE41-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; SSE41-NEXT:    pinsrb $13, %eax, %xmm1
-; SSE41-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; SSE41-NEXT:    pinsrb $14, %eax, %xmm1
-; SSE41-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; SSE41-NEXT:    pinsrb $15, %eax, %xmm1
-; SSE41-NEXT:    popq %rbx
-; SSE41-NEXT:    popq %r12
-; SSE41-NEXT:    popq %r13
-; SSE41-NEXT:    popq %r14
-; SSE41-NEXT:    popq %r15
-; SSE41-NEXT:    popq %rbp
-; SSE41-NEXT:    retq
+; SSE-LABEL: v32i8:
+; SSE:       # %bb.0:
+; SSE-NEXT:    paddsb %xmm2, %xmm0
+; SSE-NEXT:    paddsb %xmm3, %xmm1
+; SSE-NEXT:    retq
 ;
 ; AVX1-LABEL: v32i8:
 ; AVX1:       # %bb.0:
-; AVX1-NEXT:    pushq %rbp
-; AVX1-NEXT:    pushq %r15
-; AVX1-NEXT:    pushq %r14
-; AVX1-NEXT:    pushq %r13
-; AVX1-NEXT:    pushq %r12
-; AVX1-NEXT:    pushq %rbx
-; AVX1-NEXT:    vpextrb $15, %xmm1, %ecx
-; AVX1-NEXT:    vpextrb $15, %xmm0, %edx
-; AVX1-NEXT:    movl %edx, %eax
-; AVX1-NEXT:    addb %cl, %al
-; AVX1-NEXT:    setns %al
-; AVX1-NEXT:    addb %cl, %dl
-; AVX1-NEXT:    jo .LBB1_1
-; AVX1-NEXT:  # %bb.2:
-; AVX1-NEXT:    movl %edx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX1-NEXT:    jmp .LBB1_3
-; AVX1-NEXT:  .LBB1_1:
-; AVX1-NEXT:    addb $127, %al
-; AVX1-NEXT:    # kill: def $al killed $al def $eax
-; AVX1-NEXT:    movl %eax, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX1-NEXT:  .LBB1_3:
-; AVX1-NEXT:    vpextrb $14, %xmm1, %ecx
-; AVX1-NEXT:    vpextrb $14, %xmm0, %edx
-; AVX1-NEXT:    movl %edx, %eax
-; AVX1-NEXT:    addb %cl, %al
-; AVX1-NEXT:    setns %al
-; AVX1-NEXT:    addb %cl, %dl
-; AVX1-NEXT:    jno .LBB1_5
-; AVX1-NEXT:  # %bb.4:
-; AVX1-NEXT:    addb $127, %al
-; AVX1-NEXT:    movl %eax, %edx
-; AVX1-NEXT:  .LBB1_5:
-; AVX1-NEXT:    vpextrb $13, %xmm1, %ecx
-; AVX1-NEXT:    vpextrb $13, %xmm0, %esi
-; AVX1-NEXT:    movl %esi, %eax
-; AVX1-NEXT:    addb %cl, %al
-; AVX1-NEXT:    setns %al
-; AVX1-NEXT:    addb %cl, %sil
-; AVX1-NEXT:    jo .LBB1_6
-; AVX1-NEXT:  # %bb.7:
-; AVX1-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX1-NEXT:    jmp .LBB1_8
-; AVX1-NEXT:  .LBB1_6:
-; AVX1-NEXT:    addb $127, %al
-; AVX1-NEXT:    # kill: def $al killed $al def $eax
-; AVX1-NEXT:    movl %eax, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX1-NEXT:  .LBB1_8:
-; AVX1-NEXT:    vpextrb $12, %xmm1, %ecx
-; AVX1-NEXT:    vpextrb $12, %xmm0, %esi
-; AVX1-NEXT:    movl %esi, %eax
-; AVX1-NEXT:    addb %cl, %al
-; AVX1-NEXT:    setns %al
-; AVX1-NEXT:    addb %cl, %sil
-; AVX1-NEXT:    jno .LBB1_10
-; AVX1-NEXT:  # %bb.9:
-; AVX1-NEXT:    addb $127, %al
-; AVX1-NEXT:    movl %eax, %esi
-; AVX1-NEXT:  .LBB1_10:
-; AVX1-NEXT:    vpextrb $11, %xmm1, %ecx
-; AVX1-NEXT:    vpextrb $11, %xmm0, %edi
-; AVX1-NEXT:    movl %edi, %eax
-; AVX1-NEXT:    addb %cl, %al
-; AVX1-NEXT:    setns %al
-; AVX1-NEXT:    addb %cl, %dil
-; AVX1-NEXT:    jno .LBB1_12
-; AVX1-NEXT:  # %bb.11:
-; AVX1-NEXT:    addb $127, %al
-; AVX1-NEXT:    movl %eax, %edi
-; AVX1-NEXT:  .LBB1_12:
-; AVX1-NEXT:    vpextrb $10, %xmm1, %ecx
-; AVX1-NEXT:    vpextrb $10, %xmm0, %ebp
-; AVX1-NEXT:    movl %ebp, %eax
-; AVX1-NEXT:    addb %cl, %al
-; AVX1-NEXT:    setns %al
-; AVX1-NEXT:    addb %cl, %bpl
-; AVX1-NEXT:    jno .LBB1_14
-; AVX1-NEXT:  # %bb.13:
-; AVX1-NEXT:    addb $127, %al
-; AVX1-NEXT:    movl %eax, %ebp
-; AVX1-NEXT:  .LBB1_14:
-; AVX1-NEXT:    vpextrb $9, %xmm1, %ecx
-; AVX1-NEXT:    vpextrb $9, %xmm0, %ebx
-; AVX1-NEXT:    movl %ebx, %eax
-; AVX1-NEXT:    addb %cl, %al
-; AVX1-NEXT:    setns %al
-; AVX1-NEXT:    addb %cl, %bl
-; AVX1-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX1-NEXT:    jno .LBB1_16
-; AVX1-NEXT:  # %bb.15:
-; AVX1-NEXT:    addb $127, %al
-; AVX1-NEXT:    movl %eax, %ebx
-; AVX1-NEXT:  .LBB1_16:
-; AVX1-NEXT:    vpextrb $8, %xmm1, %ecx
-; AVX1-NEXT:    vpextrb $8, %xmm0, %esi
-; AVX1-NEXT:    movl %esi, %eax
-; AVX1-NEXT:    addb %cl, %al
-; AVX1-NEXT:    setns %al
-; AVX1-NEXT:    addb %cl, %sil
-; AVX1-NEXT:    movl %edx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX1-NEXT:    movl %ebp, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX1-NEXT:    movl %ebx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX1-NEXT:    jno .LBB1_18
-; AVX1-NEXT:  # %bb.17:
-; AVX1-NEXT:    addb $127, %al
-; AVX1-NEXT:    movl %eax, %esi
-; AVX1-NEXT:  .LBB1_18:
-; AVX1-NEXT:    vpextrb $7, %xmm1, %ecx
-; AVX1-NEXT:    vpextrb $7, %xmm0, %edx
-; AVX1-NEXT:    movl %edx, %eax
-; AVX1-NEXT:    addb %cl, %al
-; AVX1-NEXT:    setns %al
-; AVX1-NEXT:    addb %cl, %dl
-; AVX1-NEXT:    movl %edi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX1-NEXT:    jno .LBB1_20
-; AVX1-NEXT:  # %bb.19:
-; AVX1-NEXT:    addb $127, %al
-; AVX1-NEXT:    movl %eax, %edx
-; AVX1-NEXT:  .LBB1_20:
-; AVX1-NEXT:    vpextrb $6, %xmm1, %ecx
-; AVX1-NEXT:    vpextrb $6, %xmm0, %edi
-; AVX1-NEXT:    movl %edi, %eax
-; AVX1-NEXT:    addb %cl, %al
-; AVX1-NEXT:    setns %al
-; AVX1-NEXT:    addb %cl, %dil
-; AVX1-NEXT:    jno .LBB1_22
-; AVX1-NEXT:  # %bb.21:
-; AVX1-NEXT:    addb $127, %al
-; AVX1-NEXT:    movl %eax, %edi
-; AVX1-NEXT:  .LBB1_22:
-; AVX1-NEXT:    vpextrb $5, %xmm1, %ecx
-; AVX1-NEXT:    vpextrb $5, %xmm0, %ebp
-; AVX1-NEXT:    movl %ebp, %eax
-; AVX1-NEXT:    addb %cl, %al
-; AVX1-NEXT:    setns %al
-; AVX1-NEXT:    addb %cl, %bpl
-; AVX1-NEXT:    jno .LBB1_24
-; AVX1-NEXT:  # %bb.23:
-; AVX1-NEXT:    addb $127, %al
-; AVX1-NEXT:    movl %eax, %ebp
-; AVX1-NEXT:  .LBB1_24:
-; AVX1-NEXT:    vpextrb $4, %xmm1, %ecx
-; AVX1-NEXT:    vpextrb $4, %xmm0, %r11d
-; AVX1-NEXT:    movl %r11d, %eax
-; AVX1-NEXT:    addb %cl, %al
-; AVX1-NEXT:    setns %al
-; AVX1-NEXT:    addb %cl, %r11b
-; AVX1-NEXT:    jno .LBB1_26
-; AVX1-NEXT:  # %bb.25:
-; AVX1-NEXT:    addb $127, %al
-; AVX1-NEXT:    movl %eax, %r11d
-; AVX1-NEXT:  .LBB1_26:
-; AVX1-NEXT:    vpextrb $3, %xmm1, %ecx
-; AVX1-NEXT:    vpextrb $3, %xmm0, %r14d
-; AVX1-NEXT:    movl %r14d, %eax
-; AVX1-NEXT:    addb %cl, %al
-; AVX1-NEXT:    setns %al
-; AVX1-NEXT:    addb %cl, %r14b
-; AVX1-NEXT:    movl %ebp, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX1-NEXT:    jno .LBB1_28
-; AVX1-NEXT:  # %bb.27:
-; AVX1-NEXT:    addb $127, %al
-; AVX1-NEXT:    movl %eax, %r14d
-; AVX1-NEXT:  .LBB1_28:
-; AVX1-NEXT:    vpextrb $2, %xmm1, %ecx
-; AVX1-NEXT:    vpextrb $2, %xmm0, %r8d
-; AVX1-NEXT:    movl %r8d, %eax
-; AVX1-NEXT:    addb %cl, %al
-; AVX1-NEXT:    setns %al
-; AVX1-NEXT:    addb %cl, %r8b
-; AVX1-NEXT:    jno .LBB1_30
-; AVX1-NEXT:  # %bb.29:
-; AVX1-NEXT:    addb $127, %al
-; AVX1-NEXT:    movl %eax, %r8d
-; AVX1-NEXT:  .LBB1_30:
-; AVX1-NEXT:    vpextrb $0, %xmm1, %ecx
-; AVX1-NEXT:    vpextrb $0, %xmm0, %r10d
-; AVX1-NEXT:    movl %r10d, %eax
-; AVX1-NEXT:    addb %cl, %al
-; AVX1-NEXT:    setns %al
-; AVX1-NEXT:    addb %cl, %r10b
-; AVX1-NEXT:    jno .LBB1_32
-; AVX1-NEXT:  # %bb.31:
-; AVX1-NEXT:    addb $127, %al
-; AVX1-NEXT:    movl %eax, %r10d
-; AVX1-NEXT:  .LBB1_32:
-; AVX1-NEXT:    vpextrb $1, %xmm1, %ecx
-; AVX1-NEXT:    vpextrb $1, %xmm0, %r9d
-; AVX1-NEXT:    movl %r9d, %eax
-; AVX1-NEXT:    addb %cl, %al
-; AVX1-NEXT:    setns %al
-; AVX1-NEXT:    addb %cl, %r9b
-; AVX1-NEXT:    jno .LBB1_34
-; AVX1-NEXT:  # %bb.33:
-; AVX1-NEXT:    addb $127, %al
-; AVX1-NEXT:    movl %eax, %r9d
-; AVX1-NEXT:  .LBB1_34:
-; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm1
-; AVX1-NEXT:    vpextrb $15, %xmm1, %ecx
-; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm0
-; AVX1-NEXT:    vpextrb $15, %xmm0, %ebx
-; AVX1-NEXT:    movl %ebx, %eax
-; AVX1-NEXT:    addb %cl, %al
-; AVX1-NEXT:    setns %al
-; AVX1-NEXT:    movl %esi, %r12d
-; AVX1-NEXT:    addb %cl, %bl
-; AVX1-NEXT:    jno .LBB1_36
-; AVX1-NEXT:  # %bb.35:
-; AVX1-NEXT:    addb $127, %al
-; AVX1-NEXT:    movl %eax, %ebx
-; AVX1-NEXT:  .LBB1_36:
-; AVX1-NEXT:    vpextrb $14, %xmm1, %ecx
-; AVX1-NEXT:    vpextrb $14, %xmm0, %esi
-; AVX1-NEXT:    movl %esi, %eax
-; AVX1-NEXT:    addb %cl, %al
-; AVX1-NEXT:    setns %al
-; AVX1-NEXT:    movl %edx, %r13d
-; AVX1-NEXT:    addb %cl, %sil
-; AVX1-NEXT:    jno .LBB1_38
-; AVX1-NEXT:  # %bb.37:
-; AVX1-NEXT:    addb $127, %al
-; AVX1-NEXT:    movl %eax, %esi
-; AVX1-NEXT:  .LBB1_38:
-; AVX1-NEXT:    vpextrb $13, %xmm1, %ecx
-; AVX1-NEXT:    vpextrb $13, %xmm0, %edx
-; AVX1-NEXT:    movl %edx, %eax
-; AVX1-NEXT:    addb %cl, %al
-; AVX1-NEXT:    setns %al
-; AVX1-NEXT:    movl %edi, %ebp
-; AVX1-NEXT:    addb %cl, %dl
-; AVX1-NEXT:    jno .LBB1_40
-; AVX1-NEXT:  # %bb.39:
-; AVX1-NEXT:    addb $127, %al
-; AVX1-NEXT:    movl %eax, %edx
-; AVX1-NEXT:  .LBB1_40:
-; AVX1-NEXT:    vpextrb $12, %xmm1, %ecx
-; AVX1-NEXT:    vpextrb $12, %xmm0, %edi
-; AVX1-NEXT:    movl %edi, %eax
-; AVX1-NEXT:    addb %cl, %al
-; AVX1-NEXT:    setns %al
-; AVX1-NEXT:    addb %cl, %dil
-; AVX1-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX1-NEXT:    movl %edx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX1-NEXT:    jo .LBB1_41
-; AVX1-NEXT:  # %bb.42:
-; AVX1-NEXT:    movl %edi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX1-NEXT:    jmp .LBB1_43
-; AVX1-NEXT:  .LBB1_41:
-; AVX1-NEXT:    addb $127, %al
-; AVX1-NEXT:    # kill: def $al killed $al def $eax
-; AVX1-NEXT:    movl %eax, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX1-NEXT:  .LBB1_43:
-; AVX1-NEXT:    movl %ebp, %edi
-; AVX1-NEXT:    vpextrb $11, %xmm1, %ecx
-; AVX1-NEXT:    vpextrb $11, %xmm0, %r15d
-; AVX1-NEXT:    movl %r15d, %eax
-; AVX1-NEXT:    addb %cl, %al
-; AVX1-NEXT:    setns %al
-; AVX1-NEXT:    addb %cl, %r15b
-; AVX1-NEXT:    movl %r12d, %esi
-; AVX1-NEXT:    movl %r13d, %edx
-; AVX1-NEXT:    jno .LBB1_45
-; AVX1-NEXT:  # %bb.44:
-; AVX1-NEXT:    addb $127, %al
-; AVX1-NEXT:    movl %eax, %r15d
-; AVX1-NEXT:  .LBB1_45:
-; AVX1-NEXT:    vpextrb $10, %xmm1, %ecx
-; AVX1-NEXT:    vpextrb $10, %xmm0, %r12d
-; AVX1-NEXT:    movl %r12d, %eax
-; AVX1-NEXT:    addb %cl, %al
-; AVX1-NEXT:    setns %al
-; AVX1-NEXT:    addb %cl, %r12b
-; AVX1-NEXT:    jno .LBB1_47
-; AVX1-NEXT:  # %bb.46:
-; AVX1-NEXT:    addb $127, %al
-; AVX1-NEXT:    movl %eax, %r12d
-; AVX1-NEXT:  .LBB1_47:
-; AVX1-NEXT:    vpextrb $9, %xmm1, %ecx
-; AVX1-NEXT:    vpextrb $9, %xmm0, %r13d
-; AVX1-NEXT:    movl %r13d, %eax
-; AVX1-NEXT:    addb %cl, %al
-; AVX1-NEXT:    setns %al
-; AVX1-NEXT:    addb %cl, %r13b
-; AVX1-NEXT:    movl %r11d, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX1-NEXT:    movl %r14d, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX1-NEXT:    movl %r8d, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX1-NEXT:    movl %r10d, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX1-NEXT:    movl %r9d, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX1-NEXT:    movl %ebx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX1-NEXT:    jno .LBB1_49
-; AVX1-NEXT:  # %bb.48:
-; AVX1-NEXT:    addb $127, %al
-; AVX1-NEXT:    movl %eax, %r13d
-; AVX1-NEXT:  .LBB1_49:
-; AVX1-NEXT:    vpextrb $8, %xmm1, %ecx
-; AVX1-NEXT:    vpextrb $8, %xmm0, %r11d
-; AVX1-NEXT:    movl %r11d, %eax
-; AVX1-NEXT:    addb %cl, %al
-; AVX1-NEXT:    setns %al
-; AVX1-NEXT:    addb %cl, %r11b
-; AVX1-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX1-NEXT:    movl %edx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX1-NEXT:    movl %edi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX1-NEXT:    jno .LBB1_51
-; AVX1-NEXT:  # %bb.50:
-; AVX1-NEXT:    addb $127, %al
-; AVX1-NEXT:    movl %eax, %r11d
-; AVX1-NEXT:  .LBB1_51:
-; AVX1-NEXT:    vpextrb $7, %xmm1, %ecx
-; AVX1-NEXT:    vpextrb $7, %xmm0, %r10d
-; AVX1-NEXT:    movl %r10d, %eax
-; AVX1-NEXT:    addb %cl, %al
-; AVX1-NEXT:    setns %al
-; AVX1-NEXT:    addb %cl, %r10b
-; AVX1-NEXT:    jno .LBB1_53
-; AVX1-NEXT:  # %bb.52:
-; AVX1-NEXT:    addb $127, %al
-; AVX1-NEXT:    movl %eax, %r10d
-; AVX1-NEXT:  .LBB1_53:
-; AVX1-NEXT:    vpextrb $6, %xmm1, %ecx
-; AVX1-NEXT:    vpextrb $6, %xmm0, %ebp
-; AVX1-NEXT:    movl %ebp, %eax
-; AVX1-NEXT:    addb %cl, %al
-; AVX1-NEXT:    setns %al
-; AVX1-NEXT:    addb %cl, %bpl
-; AVX1-NEXT:    jno .LBB1_55
-; AVX1-NEXT:  # %bb.54:
-; AVX1-NEXT:    addb $127, %al
-; AVX1-NEXT:    movl %eax, %ebp
-; AVX1-NEXT:  .LBB1_55:
-; AVX1-NEXT:    vpextrb $5, %xmm1, %ecx
-; AVX1-NEXT:    vpextrb $5, %xmm0, %edi
-; AVX1-NEXT:    movl %edi, %eax
-; AVX1-NEXT:    addb %cl, %al
-; AVX1-NEXT:    setns %al
-; AVX1-NEXT:    addb %cl, %dil
-; AVX1-NEXT:    jno .LBB1_57
-; AVX1-NEXT:  # %bb.56:
-; AVX1-NEXT:    addb $127, %al
-; AVX1-NEXT:    movl %eax, %edi
-; AVX1-NEXT:  .LBB1_57:
-; AVX1-NEXT:    vpextrb $4, %xmm1, %edx
-; AVX1-NEXT:    vpextrb $4, %xmm0, %eax
-; AVX1-NEXT:    movl %eax, %ecx
-; AVX1-NEXT:    addb %dl, %cl
-; AVX1-NEXT:    setns %cl
-; AVX1-NEXT:    addb %dl, %al
-; AVX1-NEXT:    jno .LBB1_59
-; AVX1-NEXT:  # %bb.58:
-; AVX1-NEXT:    addb $127, %cl
-; AVX1-NEXT:    movl %ecx, %eax
-; AVX1-NEXT:  .LBB1_59:
-; AVX1-NEXT:    vpextrb $3, %xmm1, %ebx
-; AVX1-NEXT:    vpextrb $3, %xmm0, %ecx
-; AVX1-NEXT:    movl %ecx, %edx
-; AVX1-NEXT:    addb %bl, %dl
-; AVX1-NEXT:    setns %dl
-; AVX1-NEXT:    addb %bl, %cl
-; AVX1-NEXT:    jno .LBB1_61
-; AVX1-NEXT:  # %bb.60:
-; AVX1-NEXT:    addb $127, %dl
-; AVX1-NEXT:    movl %edx, %ecx
-; AVX1-NEXT:  .LBB1_61:
-; AVX1-NEXT:    vpextrb $2, %xmm1, %esi
-; AVX1-NEXT:    vpextrb $2, %xmm0, %edx
-; AVX1-NEXT:    movl %edx, %ebx
-; AVX1-NEXT:    addb %sil, %bl
-; AVX1-NEXT:    setns %bl
-; AVX1-NEXT:    addb %sil, %dl
-; AVX1-NEXT:    jno .LBB1_63
-; AVX1-NEXT:  # %bb.62:
-; AVX1-NEXT:    addb $127, %bl
-; AVX1-NEXT:    movl %ebx, %edx
-; AVX1-NEXT:  .LBB1_63:
-; AVX1-NEXT:    vpextrb $0, %xmm1, %esi
-; AVX1-NEXT:    vpextrb $0, %xmm0, %r8d
-; AVX1-NEXT:    movl %r8d, %ebx
-; AVX1-NEXT:    addb %sil, %bl
-; AVX1-NEXT:    setns %bl
-; AVX1-NEXT:    addb %sil, %r8b
-; AVX1-NEXT:    jo .LBB1_64
-; AVX1-NEXT:  # %bb.65:
-; AVX1-NEXT:    movl {{[-0-9]+}}(%r{{[sb]}}p), %r14d # 4-byte Reload
-; AVX1-NEXT:    jmp .LBB1_66
-; AVX1-NEXT:  .LBB1_64:
-; AVX1-NEXT:    movl {{[-0-9]+}}(%r{{[sb]}}p), %r14d # 4-byte Reload
-; AVX1-NEXT:    addb $127, %bl
-; AVX1-NEXT:    movl %ebx, %r8d
-; AVX1-NEXT:  .LBB1_66:
-; AVX1-NEXT:    vpextrb $1, %xmm1, %esi
-; AVX1-NEXT:    vpextrb $1, %xmm0, %r9d
-; AVX1-NEXT:    movl %r9d, %ebx
-; AVX1-NEXT:    addb %sil, %bl
-; AVX1-NEXT:    setns %bl
-; AVX1-NEXT:    addb %sil, %r9b
-; AVX1-NEXT:    jno .LBB1_68
-; AVX1-NEXT:  # %bb.67:
-; AVX1-NEXT:    addb $127, %bl
-; AVX1-NEXT:    movl %ebx, %r9d
-; AVX1-NEXT:  .LBB1_68:
-; AVX1-NEXT:    movzbl %r8b, %esi
-; AVX1-NEXT:    vmovd %esi, %xmm0
-; AVX1-NEXT:    movzbl %r9b, %esi
-; AVX1-NEXT:    vpinsrb $1, %esi, %xmm0, %xmm0
-; AVX1-NEXT:    movzbl %dl, %edx
-; AVX1-NEXT:    vpinsrb $2, %edx, %xmm0, %xmm0
-; AVX1-NEXT:    movzbl %cl, %ecx
-; AVX1-NEXT:    vpinsrb $3, %ecx, %xmm0, %xmm0
-; AVX1-NEXT:    movzbl %al, %eax
-; AVX1-NEXT:    vpinsrb $4, %eax, %xmm0, %xmm0
-; AVX1-NEXT:    movzbl %dil, %eax
-; AVX1-NEXT:    vpinsrb $5, %eax, %xmm0, %xmm0
-; AVX1-NEXT:    movzbl %bpl, %eax
-; AVX1-NEXT:    vpinsrb $6, %eax, %xmm0, %xmm0
-; AVX1-NEXT:    movzbl %r10b, %eax
-; AVX1-NEXT:    vpinsrb $7, %eax, %xmm0, %xmm0
-; AVX1-NEXT:    movzbl %r11b, %eax
-; AVX1-NEXT:    vpinsrb $8, %eax, %xmm0, %xmm0
-; AVX1-NEXT:    movzbl %r13b, %eax
-; AVX1-NEXT:    vpinsrb $9, %eax, %xmm0, %xmm0
-; AVX1-NEXT:    movzbl %r12b, %eax
-; AVX1-NEXT:    vpinsrb $10, %eax, %xmm0, %xmm0
-; AVX1-NEXT:    movzbl %r15b, %eax
-; AVX1-NEXT:    vpinsrb $11, %eax, %xmm0, %xmm0
-; AVX1-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX1-NEXT:    vpinsrb $12, %eax, %xmm0, %xmm0
-; AVX1-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX1-NEXT:    vpinsrb $13, %eax, %xmm0, %xmm0
-; AVX1-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX1-NEXT:    vpinsrb $14, %eax, %xmm0, %xmm0
-; AVX1-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX1-NEXT:    vpinsrb $15, %eax, %xmm0, %xmm0
-; AVX1-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX1-NEXT:    vmovd %eax, %xmm1
-; AVX1-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX1-NEXT:    vpinsrb $1, %eax, %xmm1, %xmm1
-; AVX1-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX1-NEXT:    vpinsrb $2, %eax, %xmm1, %xmm1
-; AVX1-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX1-NEXT:    vpinsrb $3, %eax, %xmm1, %xmm1
-; AVX1-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX1-NEXT:    vpinsrb $4, %eax, %xmm1, %xmm1
-; AVX1-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX1-NEXT:    vpinsrb $5, %eax, %xmm1, %xmm1
-; AVX1-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX1-NEXT:    vpinsrb $6, %eax, %xmm1, %xmm1
-; AVX1-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX1-NEXT:    vpinsrb $7, %eax, %xmm1, %xmm1
-; AVX1-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX1-NEXT:    vpinsrb $8, %eax, %xmm1, %xmm1
-; AVX1-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX1-NEXT:    vpinsrb $9, %eax, %xmm1, %xmm1
-; AVX1-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX1-NEXT:    vpinsrb $10, %eax, %xmm1, %xmm1
-; AVX1-NEXT:    movzbl %r14b, %eax
-; AVX1-NEXT:    vpinsrb $11, %eax, %xmm1, %xmm1
-; AVX1-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX1-NEXT:    vpinsrb $12, %eax, %xmm1, %xmm1
-; AVX1-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX1-NEXT:    vpinsrb $13, %eax, %xmm1, %xmm1
-; AVX1-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX1-NEXT:    vpinsrb $14, %eax, %xmm1, %xmm1
-; AVX1-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX1-NEXT:    vpinsrb $15, %eax, %xmm1, %xmm1
-; AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm1, %ymm0
-; AVX1-NEXT:    popq %rbx
-; AVX1-NEXT:    popq %r12
-; AVX1-NEXT:    popq %r13
-; AVX1-NEXT:    popq %r14
-; AVX1-NEXT:    popq %r15
-; AVX1-NEXT:    popq %rbp
+; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm2
+; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm3
+; AVX1-NEXT:    vpaddsb %xmm2, %xmm3, %xmm2
+; AVX1-NEXT:    vpaddsb %xmm1, %xmm0, %xmm0
+; AVX1-NEXT:    vinsertf128 $1, %xmm2, %ymm0, %ymm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: v32i8:
 ; AVX2:       # %bb.0:
-; AVX2-NEXT:    pushq %rbp
-; AVX2-NEXT:    pushq %r15
-; AVX2-NEXT:    pushq %r14
-; AVX2-NEXT:    pushq %r13
-; AVX2-NEXT:    pushq %r12
-; AVX2-NEXT:    pushq %rbx
-; AVX2-NEXT:    vpextrb $15, %xmm1, %ecx
-; AVX2-NEXT:    vpextrb $15, %xmm0, %edx
-; AVX2-NEXT:    movl %edx, %eax
-; AVX2-NEXT:    addb %cl, %al
-; AVX2-NEXT:    setns %al
-; AVX2-NEXT:    addb %cl, %dl
-; AVX2-NEXT:    jo .LBB1_1
-; AVX2-NEXT:  # %bb.2:
-; AVX2-NEXT:    movl %edx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX2-NEXT:    jmp .LBB1_3
-; AVX2-NEXT:  .LBB1_1:
-; AVX2-NEXT:    addb $127, %al
-; AVX2-NEXT:    # kill: def $al killed $al def $eax
-; AVX2-NEXT:    movl %eax, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX2-NEXT:  .LBB1_3:
-; AVX2-NEXT:    vpextrb $14, %xmm1, %ecx
-; AVX2-NEXT:    vpextrb $14, %xmm0, %edx
-; AVX2-NEXT:    movl %edx, %eax
-; AVX2-NEXT:    addb %cl, %al
-; AVX2-NEXT:    setns %al
-; AVX2-NEXT:    addb %cl, %dl
-; AVX2-NEXT:    jno .LBB1_5
-; AVX2-NEXT:  # %bb.4:
-; AVX2-NEXT:    addb $127, %al
-; AVX2-NEXT:    movl %eax, %edx
-; AVX2-NEXT:  .LBB1_5:
-; AVX2-NEXT:    vpextrb $13, %xmm1, %ecx
-; AVX2-NEXT:    vpextrb $13, %xmm0, %esi
-; AVX2-NEXT:    movl %esi, %eax
-; AVX2-NEXT:    addb %cl, %al
-; AVX2-NEXT:    setns %al
-; AVX2-NEXT:    addb %cl, %sil
-; AVX2-NEXT:    jo .LBB1_6
-; AVX2-NEXT:  # %bb.7:
-; AVX2-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX2-NEXT:    jmp .LBB1_8
-; AVX2-NEXT:  .LBB1_6:
-; AVX2-NEXT:    addb $127, %al
-; AVX2-NEXT:    # kill: def $al killed $al def $eax
-; AVX2-NEXT:    movl %eax, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX2-NEXT:  .LBB1_8:
-; AVX2-NEXT:    vpextrb $12, %xmm1, %ecx
-; AVX2-NEXT:    vpextrb $12, %xmm0, %esi
-; AVX2-NEXT:    movl %esi, %eax
-; AVX2-NEXT:    addb %cl, %al
-; AVX2-NEXT:    setns %al
-; AVX2-NEXT:    addb %cl, %sil
-; AVX2-NEXT:    jno .LBB1_10
-; AVX2-NEXT:  # %bb.9:
-; AVX2-NEXT:    addb $127, %al
-; AVX2-NEXT:    movl %eax, %esi
-; AVX2-NEXT:  .LBB1_10:
-; AVX2-NEXT:    vpextrb $11, %xmm1, %ecx
-; AVX2-NEXT:    vpextrb $11, %xmm0, %edi
-; AVX2-NEXT:    movl %edi, %eax
-; AVX2-NEXT:    addb %cl, %al
-; AVX2-NEXT:    setns %al
-; AVX2-NEXT:    addb %cl, %dil
-; AVX2-NEXT:    jno .LBB1_12
-; AVX2-NEXT:  # %bb.11:
-; AVX2-NEXT:    addb $127, %al
-; AVX2-NEXT:    movl %eax, %edi
-; AVX2-NEXT:  .LBB1_12:
-; AVX2-NEXT:    vpextrb $10, %xmm1, %ecx
-; AVX2-NEXT:    vpextrb $10, %xmm0, %ebp
-; AVX2-NEXT:    movl %ebp, %eax
-; AVX2-NEXT:    addb %cl, %al
-; AVX2-NEXT:    setns %al
-; AVX2-NEXT:    addb %cl, %bpl
-; AVX2-NEXT:    jno .LBB1_14
-; AVX2-NEXT:  # %bb.13:
-; AVX2-NEXT:    addb $127, %al
-; AVX2-NEXT:    movl %eax, %ebp
-; AVX2-NEXT:  .LBB1_14:
-; AVX2-NEXT:    vpextrb $9, %xmm1, %ecx
-; AVX2-NEXT:    vpextrb $9, %xmm0, %ebx
-; AVX2-NEXT:    movl %ebx, %eax
-; AVX2-NEXT:    addb %cl, %al
-; AVX2-NEXT:    setns %al
-; AVX2-NEXT:    addb %cl, %bl
-; AVX2-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX2-NEXT:    jno .LBB1_16
-; AVX2-NEXT:  # %bb.15:
-; AVX2-NEXT:    addb $127, %al
-; AVX2-NEXT:    movl %eax, %ebx
-; AVX2-NEXT:  .LBB1_16:
-; AVX2-NEXT:    vpextrb $8, %xmm1, %ecx
-; AVX2-NEXT:    vpextrb $8, %xmm0, %esi
-; AVX2-NEXT:    movl %esi, %eax
-; AVX2-NEXT:    addb %cl, %al
-; AVX2-NEXT:    setns %al
-; AVX2-NEXT:    addb %cl, %sil
-; AVX2-NEXT:    movl %edx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX2-NEXT:    movl %ebp, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX2-NEXT:    movl %ebx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX2-NEXT:    jno .LBB1_18
-; AVX2-NEXT:  # %bb.17:
-; AVX2-NEXT:    addb $127, %al
-; AVX2-NEXT:    movl %eax, %esi
-; AVX2-NEXT:  .LBB1_18:
-; AVX2-NEXT:    vpextrb $7, %xmm1, %ecx
-; AVX2-NEXT:    vpextrb $7, %xmm0, %edx
-; AVX2-NEXT:    movl %edx, %eax
-; AVX2-NEXT:    addb %cl, %al
-; AVX2-NEXT:    setns %al
-; AVX2-NEXT:    addb %cl, %dl
-; AVX2-NEXT:    movl %edi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX2-NEXT:    jno .LBB1_20
-; AVX2-NEXT:  # %bb.19:
-; AVX2-NEXT:    addb $127, %al
-; AVX2-NEXT:    movl %eax, %edx
-; AVX2-NEXT:  .LBB1_20:
-; AVX2-NEXT:    vpextrb $6, %xmm1, %ecx
-; AVX2-NEXT:    vpextrb $6, %xmm0, %edi
-; AVX2-NEXT:    movl %edi, %eax
-; AVX2-NEXT:    addb %cl, %al
-; AVX2-NEXT:    setns %al
-; AVX2-NEXT:    addb %cl, %dil
-; AVX2-NEXT:    jno .LBB1_22
-; AVX2-NEXT:  # %bb.21:
-; AVX2-NEXT:    addb $127, %al
-; AVX2-NEXT:    movl %eax, %edi
-; AVX2-NEXT:  .LBB1_22:
-; AVX2-NEXT:    vpextrb $5, %xmm1, %ecx
-; AVX2-NEXT:    vpextrb $5, %xmm0, %ebp
-; AVX2-NEXT:    movl %ebp, %eax
-; AVX2-NEXT:    addb %cl, %al
-; AVX2-NEXT:    setns %al
-; AVX2-NEXT:    addb %cl, %bpl
-; AVX2-NEXT:    jno .LBB1_24
-; AVX2-NEXT:  # %bb.23:
-; AVX2-NEXT:    addb $127, %al
-; AVX2-NEXT:    movl %eax, %ebp
-; AVX2-NEXT:  .LBB1_24:
-; AVX2-NEXT:    vpextrb $4, %xmm1, %ecx
-; AVX2-NEXT:    vpextrb $4, %xmm0, %r11d
-; AVX2-NEXT:    movl %r11d, %eax
-; AVX2-NEXT:    addb %cl, %al
-; AVX2-NEXT:    setns %al
-; AVX2-NEXT:    addb %cl, %r11b
-; AVX2-NEXT:    jno .LBB1_26
-; AVX2-NEXT:  # %bb.25:
-; AVX2-NEXT:    addb $127, %al
-; AVX2-NEXT:    movl %eax, %r11d
-; AVX2-NEXT:  .LBB1_26:
-; AVX2-NEXT:    vpextrb $3, %xmm1, %ecx
-; AVX2-NEXT:    vpextrb $3, %xmm0, %r14d
-; AVX2-NEXT:    movl %r14d, %eax
-; AVX2-NEXT:    addb %cl, %al
-; AVX2-NEXT:    setns %al
-; AVX2-NEXT:    addb %cl, %r14b
-; AVX2-NEXT:    movl %ebp, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX2-NEXT:    jno .LBB1_28
-; AVX2-NEXT:  # %bb.27:
-; AVX2-NEXT:    addb $127, %al
-; AVX2-NEXT:    movl %eax, %r14d
-; AVX2-NEXT:  .LBB1_28:
-; AVX2-NEXT:    vpextrb $2, %xmm1, %ecx
-; AVX2-NEXT:    vpextrb $2, %xmm0, %r8d
-; AVX2-NEXT:    movl %r8d, %eax
-; AVX2-NEXT:    addb %cl, %al
-; AVX2-NEXT:    setns %al
-; AVX2-NEXT:    addb %cl, %r8b
-; AVX2-NEXT:    jno .LBB1_30
-; AVX2-NEXT:  # %bb.29:
-; AVX2-NEXT:    addb $127, %al
-; AVX2-NEXT:    movl %eax, %r8d
-; AVX2-NEXT:  .LBB1_30:
-; AVX2-NEXT:    vpextrb $0, %xmm1, %ecx
-; AVX2-NEXT:    vpextrb $0, %xmm0, %r10d
-; AVX2-NEXT:    movl %r10d, %eax
-; AVX2-NEXT:    addb %cl, %al
-; AVX2-NEXT:    setns %al
-; AVX2-NEXT:    addb %cl, %r10b
-; AVX2-NEXT:    jno .LBB1_32
-; AVX2-NEXT:  # %bb.31:
-; AVX2-NEXT:    addb $127, %al
-; AVX2-NEXT:    movl %eax, %r10d
-; AVX2-NEXT:  .LBB1_32:
-; AVX2-NEXT:    vpextrb $1, %xmm1, %ecx
-; AVX2-NEXT:    vpextrb $1, %xmm0, %r9d
-; AVX2-NEXT:    movl %r9d, %eax
-; AVX2-NEXT:    addb %cl, %al
-; AVX2-NEXT:    setns %al
-; AVX2-NEXT:    addb %cl, %r9b
-; AVX2-NEXT:    jno .LBB1_34
-; AVX2-NEXT:  # %bb.33:
-; AVX2-NEXT:    addb $127, %al
-; AVX2-NEXT:    movl %eax, %r9d
-; AVX2-NEXT:  .LBB1_34:
-; AVX2-NEXT:    vextracti128 $1, %ymm1, %xmm1
-; AVX2-NEXT:    vpextrb $15, %xmm1, %ecx
-; AVX2-NEXT:    vextracti128 $1, %ymm0, %xmm0
-; AVX2-NEXT:    vpextrb $15, %xmm0, %ebx
-; AVX2-NEXT:    movl %ebx, %eax
-; AVX2-NEXT:    addb %cl, %al
-; AVX2-NEXT:    setns %al
-; AVX2-NEXT:    movl %esi, %r12d
-; AVX2-NEXT:    addb %cl, %bl
-; AVX2-NEXT:    jno .LBB1_36
-; AVX2-NEXT:  # %bb.35:
-; AVX2-NEXT:    addb $127, %al
-; AVX2-NEXT:    movl %eax, %ebx
-; AVX2-NEXT:  .LBB1_36:
-; AVX2-NEXT:    vpextrb $14, %xmm1, %ecx
-; AVX2-NEXT:    vpextrb $14, %xmm0, %esi
-; AVX2-NEXT:    movl %esi, %eax
-; AVX2-NEXT:    addb %cl, %al
-; AVX2-NEXT:    setns %al
-; AVX2-NEXT:    movl %edx, %r13d
-; AVX2-NEXT:    addb %cl, %sil
-; AVX2-NEXT:    jno .LBB1_38
-; AVX2-NEXT:  # %bb.37:
-; AVX2-NEXT:    addb $127, %al
-; AVX2-NEXT:    movl %eax, %esi
-; AVX2-NEXT:  .LBB1_38:
-; AVX2-NEXT:    vpextrb $13, %xmm1, %ecx
-; AVX2-NEXT:    vpextrb $13, %xmm0, %edx
-; AVX2-NEXT:    movl %edx, %eax
-; AVX2-NEXT:    addb %cl, %al
-; AVX2-NEXT:    setns %al
-; AVX2-NEXT:    movl %edi, %ebp
-; AVX2-NEXT:    addb %cl, %dl
-; AVX2-NEXT:    jno .LBB1_40
-; AVX2-NEXT:  # %bb.39:
-; AVX2-NEXT:    addb $127, %al
-; AVX2-NEXT:    movl %eax, %edx
-; AVX2-NEXT:  .LBB1_40:
-; AVX2-NEXT:    vpextrb $12, %xmm1, %ecx
-; AVX2-NEXT:    vpextrb $12, %xmm0, %edi
-; AVX2-NEXT:    movl %edi, %eax
-; AVX2-NEXT:    addb %cl, %al
-; AVX2-NEXT:    setns %al
-; AVX2-NEXT:    addb %cl, %dil
-; AVX2-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX2-NEXT:    movl %edx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX2-NEXT:    jo .LBB1_41
-; AVX2-NEXT:  # %bb.42:
-; AVX2-NEXT:    movl %edi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX2-NEXT:    jmp .LBB1_43
-; AVX2-NEXT:  .LBB1_41:
-; AVX2-NEXT:    addb $127, %al
-; AVX2-NEXT:    # kill: def $al killed $al def $eax
-; AVX2-NEXT:    movl %eax, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX2-NEXT:  .LBB1_43:
-; AVX2-NEXT:    movl %ebp, %edi
-; AVX2-NEXT:    vpextrb $11, %xmm1, %ecx
-; AVX2-NEXT:    vpextrb $11, %xmm0, %r15d
-; AVX2-NEXT:    movl %r15d, %eax
-; AVX2-NEXT:    addb %cl, %al
-; AVX2-NEXT:    setns %al
-; AVX2-NEXT:    addb %cl, %r15b
-; AVX2-NEXT:    movl %r12d, %esi
-; AVX2-NEXT:    movl %r13d, %edx
-; AVX2-NEXT:    jno .LBB1_45
-; AVX2-NEXT:  # %bb.44:
-; AVX2-NEXT:    addb $127, %al
-; AVX2-NEXT:    movl %eax, %r15d
-; AVX2-NEXT:  .LBB1_45:
-; AVX2-NEXT:    vpextrb $10, %xmm1, %ecx
-; AVX2-NEXT:    vpextrb $10, %xmm0, %r12d
-; AVX2-NEXT:    movl %r12d, %eax
-; AVX2-NEXT:    addb %cl, %al
-; AVX2-NEXT:    setns %al
-; AVX2-NEXT:    addb %cl, %r12b
-; AVX2-NEXT:    jno .LBB1_47
-; AVX2-NEXT:  # %bb.46:
-; AVX2-NEXT:    addb $127, %al
-; AVX2-NEXT:    movl %eax, %r12d
-; AVX2-NEXT:  .LBB1_47:
-; AVX2-NEXT:    vpextrb $9, %xmm1, %ecx
-; AVX2-NEXT:    vpextrb $9, %xmm0, %r13d
-; AVX2-NEXT:    movl %r13d, %eax
-; AVX2-NEXT:    addb %cl, %al
-; AVX2-NEXT:    setns %al
-; AVX2-NEXT:    addb %cl, %r13b
-; AVX2-NEXT:    movl %r11d, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX2-NEXT:    movl %r14d, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX2-NEXT:    movl %r8d, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX2-NEXT:    movl %r10d, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX2-NEXT:    movl %r9d, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX2-NEXT:    movl %ebx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX2-NEXT:    jno .LBB1_49
-; AVX2-NEXT:  # %bb.48:
-; AVX2-NEXT:    addb $127, %al
-; AVX2-NEXT:    movl %eax, %r13d
-; AVX2-NEXT:  .LBB1_49:
-; AVX2-NEXT:    vpextrb $8, %xmm1, %ecx
-; AVX2-NEXT:    vpextrb $8, %xmm0, %r11d
-; AVX2-NEXT:    movl %r11d, %eax
-; AVX2-NEXT:    addb %cl, %al
-; AVX2-NEXT:    setns %al
-; AVX2-NEXT:    addb %cl, %r11b
-; AVX2-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX2-NEXT:    movl %edx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX2-NEXT:    movl %edi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX2-NEXT:    jno .LBB1_51
-; AVX2-NEXT:  # %bb.50:
-; AVX2-NEXT:    addb $127, %al
-; AVX2-NEXT:    movl %eax, %r11d
-; AVX2-NEXT:  .LBB1_51:
-; AVX2-NEXT:    vpextrb $7, %xmm1, %ecx
-; AVX2-NEXT:    vpextrb $7, %xmm0, %r10d
-; AVX2-NEXT:    movl %r10d, %eax
-; AVX2-NEXT:    addb %cl, %al
-; AVX2-NEXT:    setns %al
-; AVX2-NEXT:    addb %cl, %r10b
-; AVX2-NEXT:    jno .LBB1_53
-; AVX2-NEXT:  # %bb.52:
-; AVX2-NEXT:    addb $127, %al
-; AVX2-NEXT:    movl %eax, %r10d
-; AVX2-NEXT:  .LBB1_53:
-; AVX2-NEXT:    vpextrb $6, %xmm1, %ecx
-; AVX2-NEXT:    vpextrb $6, %xmm0, %ebp
-; AVX2-NEXT:    movl %ebp, %eax
-; AVX2-NEXT:    addb %cl, %al
-; AVX2-NEXT:    setns %al
-; AVX2-NEXT:    addb %cl, %bpl
-; AVX2-NEXT:    jno .LBB1_55
-; AVX2-NEXT:  # %bb.54:
-; AVX2-NEXT:    addb $127, %al
-; AVX2-NEXT:    movl %eax, %ebp
-; AVX2-NEXT:  .LBB1_55:
-; AVX2-NEXT:    vpextrb $5, %xmm1, %ecx
-; AVX2-NEXT:    vpextrb $5, %xmm0, %edi
-; AVX2-NEXT:    movl %edi, %eax
-; AVX2-NEXT:    addb %cl, %al
-; AVX2-NEXT:    setns %al
-; AVX2-NEXT:    addb %cl, %dil
-; AVX2-NEXT:    jno .LBB1_57
-; AVX2-NEXT:  # %bb.56:
-; AVX2-NEXT:    addb $127, %al
-; AVX2-NEXT:    movl %eax, %edi
-; AVX2-NEXT:  .LBB1_57:
-; AVX2-NEXT:    vpextrb $4, %xmm1, %edx
-; AVX2-NEXT:    vpextrb $4, %xmm0, %eax
-; AVX2-NEXT:    movl %eax, %ecx
-; AVX2-NEXT:    addb %dl, %cl
-; AVX2-NEXT:    setns %cl
-; AVX2-NEXT:    addb %dl, %al
-; AVX2-NEXT:    jno .LBB1_59
-; AVX2-NEXT:  # %bb.58:
-; AVX2-NEXT:    addb $127, %cl
-; AVX2-NEXT:    movl %ecx, %eax
-; AVX2-NEXT:  .LBB1_59:
-; AVX2-NEXT:    vpextrb $3, %xmm1, %ebx
-; AVX2-NEXT:    vpextrb $3, %xmm0, %ecx
-; AVX2-NEXT:    movl %ecx, %edx
-; AVX2-NEXT:    addb %bl, %dl
-; AVX2-NEXT:    setns %dl
-; AVX2-NEXT:    addb %bl, %cl
-; AVX2-NEXT:    jno .LBB1_61
-; AVX2-NEXT:  # %bb.60:
-; AVX2-NEXT:    addb $127, %dl
-; AVX2-NEXT:    movl %edx, %ecx
-; AVX2-NEXT:  .LBB1_61:
-; AVX2-NEXT:    vpextrb $2, %xmm1, %esi
-; AVX2-NEXT:    vpextrb $2, %xmm0, %edx
-; AVX2-NEXT:    movl %edx, %ebx
-; AVX2-NEXT:    addb %sil, %bl
-; AVX2-NEXT:    setns %bl
-; AVX2-NEXT:    addb %sil, %dl
-; AVX2-NEXT:    jno .LBB1_63
-; AVX2-NEXT:  # %bb.62:
-; AVX2-NEXT:    addb $127, %bl
-; AVX2-NEXT:    movl %ebx, %edx
-; AVX2-NEXT:  .LBB1_63:
-; AVX2-NEXT:    vpextrb $0, %xmm1, %esi
-; AVX2-NEXT:    vpextrb $0, %xmm0, %r8d
-; AVX2-NEXT:    movl %r8d, %ebx
-; AVX2-NEXT:    addb %sil, %bl
-; AVX2-NEXT:    setns %bl
-; AVX2-NEXT:    addb %sil, %r8b
-; AVX2-NEXT:    jo .LBB1_64
-; AVX2-NEXT:  # %bb.65:
-; AVX2-NEXT:    movl {{[-0-9]+}}(%r{{[sb]}}p), %r14d # 4-byte Reload
-; AVX2-NEXT:    jmp .LBB1_66
-; AVX2-NEXT:  .LBB1_64:
-; AVX2-NEXT:    movl {{[-0-9]+}}(%r{{[sb]}}p), %r14d # 4-byte Reload
-; AVX2-NEXT:    addb $127, %bl
-; AVX2-NEXT:    movl %ebx, %r8d
-; AVX2-NEXT:  .LBB1_66:
-; AVX2-NEXT:    vpextrb $1, %xmm1, %esi
-; AVX2-NEXT:    vpextrb $1, %xmm0, %r9d
-; AVX2-NEXT:    movl %r9d, %ebx
-; AVX2-NEXT:    addb %sil, %bl
-; AVX2-NEXT:    setns %bl
-; AVX2-NEXT:    addb %sil, %r9b
-; AVX2-NEXT:    jno .LBB1_68
-; AVX2-NEXT:  # %bb.67:
-; AVX2-NEXT:    addb $127, %bl
-; AVX2-NEXT:    movl %ebx, %r9d
-; AVX2-NEXT:  .LBB1_68:
-; AVX2-NEXT:    movzbl %r8b, %esi
-; AVX2-NEXT:    vmovd %esi, %xmm0
-; AVX2-NEXT:    movzbl %r9b, %esi
-; AVX2-NEXT:    vpinsrb $1, %esi, %xmm0, %xmm0
-; AVX2-NEXT:    movzbl %dl, %edx
-; AVX2-NEXT:    vpinsrb $2, %edx, %xmm0, %xmm0
-; AVX2-NEXT:    movzbl %cl, %ecx
-; AVX2-NEXT:    vpinsrb $3, %ecx, %xmm0, %xmm0
-; AVX2-NEXT:    movzbl %al, %eax
-; AVX2-NEXT:    vpinsrb $4, %eax, %xmm0, %xmm0
-; AVX2-NEXT:    movzbl %dil, %eax
-; AVX2-NEXT:    vpinsrb $5, %eax, %xmm0, %xmm0
-; AVX2-NEXT:    movzbl %bpl, %eax
-; AVX2-NEXT:    vpinsrb $6, %eax, %xmm0, %xmm0
-; AVX2-NEXT:    movzbl %r10b, %eax
-; AVX2-NEXT:    vpinsrb $7, %eax, %xmm0, %xmm0
-; AVX2-NEXT:    movzbl %r11b, %eax
-; AVX2-NEXT:    vpinsrb $8, %eax, %xmm0, %xmm0
-; AVX2-NEXT:    movzbl %r13b, %eax
-; AVX2-NEXT:    vpinsrb $9, %eax, %xmm0, %xmm0
-; AVX2-NEXT:    movzbl %r12b, %eax
-; AVX2-NEXT:    vpinsrb $10, %eax, %xmm0, %xmm0
-; AVX2-NEXT:    movzbl %r15b, %eax
-; AVX2-NEXT:    vpinsrb $11, %eax, %xmm0, %xmm0
-; AVX2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX2-NEXT:    vpinsrb $12, %eax, %xmm0, %xmm0
-; AVX2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX2-NEXT:    vpinsrb $13, %eax, %xmm0, %xmm0
-; AVX2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX2-NEXT:    vpinsrb $14, %eax, %xmm0, %xmm0
-; AVX2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX2-NEXT:    vpinsrb $15, %eax, %xmm0, %xmm0
-; AVX2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX2-NEXT:    vmovd %eax, %xmm1
-; AVX2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX2-NEXT:    vpinsrb $1, %eax, %xmm1, %xmm1
-; AVX2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX2-NEXT:    vpinsrb $2, %eax, %xmm1, %xmm1
-; AVX2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX2-NEXT:    vpinsrb $3, %eax, %xmm1, %xmm1
-; AVX2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX2-NEXT:    vpinsrb $4, %eax, %xmm1, %xmm1
-; AVX2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX2-NEXT:    vpinsrb $5, %eax, %xmm1, %xmm1
-; AVX2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX2-NEXT:    vpinsrb $6, %eax, %xmm1, %xmm1
-; AVX2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX2-NEXT:    vpinsrb $7, %eax, %xmm1, %xmm1
-; AVX2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX2-NEXT:    vpinsrb $8, %eax, %xmm1, %xmm1
-; AVX2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX2-NEXT:    vpinsrb $9, %eax, %xmm1, %xmm1
-; AVX2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX2-NEXT:    vpinsrb $10, %eax, %xmm1, %xmm1
-; AVX2-NEXT:    movzbl %r14b, %eax
-; AVX2-NEXT:    vpinsrb $11, %eax, %xmm1, %xmm1
-; AVX2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX2-NEXT:    vpinsrb $12, %eax, %xmm1, %xmm1
-; AVX2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX2-NEXT:    vpinsrb $13, %eax, %xmm1, %xmm1
-; AVX2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX2-NEXT:    vpinsrb $14, %eax, %xmm1, %xmm1
-; AVX2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX2-NEXT:    vpinsrb $15, %eax, %xmm1, %xmm1
-; AVX2-NEXT:    vinserti128 $1, %xmm0, %ymm1, %ymm0
-; AVX2-NEXT:    popq %rbx
-; AVX2-NEXT:    popq %r12
-; AVX2-NEXT:    popq %r13
-; AVX2-NEXT:    popq %r14
-; AVX2-NEXT:    popq %r15
-; AVX2-NEXT:    popq %rbp
+; AVX2-NEXT:    vpaddsb %ymm1, %ymm0, %ymm0
 ; AVX2-NEXT:    retq
 ;
 ; AVX512-LABEL: v32i8:
 ; AVX512:       # %bb.0:
-; AVX512-NEXT:    pushq %rbp
-; AVX512-NEXT:    pushq %r15
-; AVX512-NEXT:    pushq %r14
-; AVX512-NEXT:    pushq %r13
-; AVX512-NEXT:    pushq %r12
-; AVX512-NEXT:    pushq %rbx
-; AVX512-NEXT:    vpextrb $15, %xmm1, %ecx
-; AVX512-NEXT:    vpextrb $15, %xmm0, %edx
-; AVX512-NEXT:    movl %edx, %eax
-; AVX512-NEXT:    addb %cl, %al
-; AVX512-NEXT:    setns %al
-; AVX512-NEXT:    addb %cl, %dl
-; AVX512-NEXT:    jo .LBB1_1
-; AVX512-NEXT:  # %bb.2:
-; AVX512-NEXT:    movl %edx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX512-NEXT:    jmp .LBB1_3
-; AVX512-NEXT:  .LBB1_1:
-; AVX512-NEXT:    addb $127, %al
-; AVX512-NEXT:    # kill: def $al killed $al def $eax
-; AVX512-NEXT:    movl %eax, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX512-NEXT:  .LBB1_3:
-; AVX512-NEXT:    vpextrb $14, %xmm1, %ecx
-; AVX512-NEXT:    vpextrb $14, %xmm0, %edx
-; AVX512-NEXT:    movl %edx, %eax
-; AVX512-NEXT:    addb %cl, %al
-; AVX512-NEXT:    setns %al
-; AVX512-NEXT:    addb %cl, %dl
-; AVX512-NEXT:    jno .LBB1_5
-; AVX512-NEXT:  # %bb.4:
-; AVX512-NEXT:    addb $127, %al
-; AVX512-NEXT:    movl %eax, %edx
-; AVX512-NEXT:  .LBB1_5:
-; AVX512-NEXT:    vpextrb $13, %xmm1, %ecx
-; AVX512-NEXT:    vpextrb $13, %xmm0, %esi
-; AVX512-NEXT:    movl %esi, %eax
-; AVX512-NEXT:    addb %cl, %al
-; AVX512-NEXT:    setns %al
-; AVX512-NEXT:    addb %cl, %sil
-; AVX512-NEXT:    jo .LBB1_6
-; AVX512-NEXT:  # %bb.7:
-; AVX512-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX512-NEXT:    jmp .LBB1_8
-; AVX512-NEXT:  .LBB1_6:
-; AVX512-NEXT:    addb $127, %al
-; AVX512-NEXT:    # kill: def $al killed $al def $eax
-; AVX512-NEXT:    movl %eax, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX512-NEXT:  .LBB1_8:
-; AVX512-NEXT:    vpextrb $12, %xmm1, %ecx
-; AVX512-NEXT:    vpextrb $12, %xmm0, %esi
-; AVX512-NEXT:    movl %esi, %eax
-; AVX512-NEXT:    addb %cl, %al
-; AVX512-NEXT:    setns %al
-; AVX512-NEXT:    addb %cl, %sil
-; AVX512-NEXT:    jno .LBB1_10
-; AVX512-NEXT:  # %bb.9:
-; AVX512-NEXT:    addb $127, %al
-; AVX512-NEXT:    movl %eax, %esi
-; AVX512-NEXT:  .LBB1_10:
-; AVX512-NEXT:    vpextrb $11, %xmm1, %ecx
-; AVX512-NEXT:    vpextrb $11, %xmm0, %edi
-; AVX512-NEXT:    movl %edi, %eax
-; AVX512-NEXT:    addb %cl, %al
-; AVX512-NEXT:    setns %al
-; AVX512-NEXT:    addb %cl, %dil
-; AVX512-NEXT:    jno .LBB1_12
-; AVX512-NEXT:  # %bb.11:
-; AVX512-NEXT:    addb $127, %al
-; AVX512-NEXT:    movl %eax, %edi
-; AVX512-NEXT:  .LBB1_12:
-; AVX512-NEXT:    vpextrb $10, %xmm1, %ecx
-; AVX512-NEXT:    vpextrb $10, %xmm0, %ebp
-; AVX512-NEXT:    movl %ebp, %eax
-; AVX512-NEXT:    addb %cl, %al
-; AVX512-NEXT:    setns %al
-; AVX512-NEXT:    addb %cl, %bpl
-; AVX512-NEXT:    jno .LBB1_14
-; AVX512-NEXT:  # %bb.13:
-; AVX512-NEXT:    addb $127, %al
-; AVX512-NEXT:    movl %eax, %ebp
-; AVX512-NEXT:  .LBB1_14:
-; AVX512-NEXT:    vpextrb $9, %xmm1, %ecx
-; AVX512-NEXT:    vpextrb $9, %xmm0, %ebx
-; AVX512-NEXT:    movl %ebx, %eax
-; AVX512-NEXT:    addb %cl, %al
-; AVX512-NEXT:    setns %al
-; AVX512-NEXT:    addb %cl, %bl
-; AVX512-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX512-NEXT:    jno .LBB1_16
-; AVX512-NEXT:  # %bb.15:
-; AVX512-NEXT:    addb $127, %al
-; AVX512-NEXT:    movl %eax, %ebx
-; AVX512-NEXT:  .LBB1_16:
-; AVX512-NEXT:    vpextrb $8, %xmm1, %ecx
-; AVX512-NEXT:    vpextrb $8, %xmm0, %esi
-; AVX512-NEXT:    movl %esi, %eax
-; AVX512-NEXT:    addb %cl, %al
-; AVX512-NEXT:    setns %al
-; AVX512-NEXT:    addb %cl, %sil
-; AVX512-NEXT:    movl %edx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX512-NEXT:    movl %ebp, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX512-NEXT:    movl %ebx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX512-NEXT:    jno .LBB1_18
-; AVX512-NEXT:  # %bb.17:
-; AVX512-NEXT:    addb $127, %al
-; AVX512-NEXT:    movl %eax, %esi
-; AVX512-NEXT:  .LBB1_18:
-; AVX512-NEXT:    vpextrb $7, %xmm1, %ecx
-; AVX512-NEXT:    vpextrb $7, %xmm0, %edx
-; AVX512-NEXT:    movl %edx, %eax
-; AVX512-NEXT:    addb %cl, %al
-; AVX512-NEXT:    setns %al
-; AVX512-NEXT:    addb %cl, %dl
-; AVX512-NEXT:    movl %edi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX512-NEXT:    jno .LBB1_20
-; AVX512-NEXT:  # %bb.19:
-; AVX512-NEXT:    addb $127, %al
-; AVX512-NEXT:    movl %eax, %edx
-; AVX512-NEXT:  .LBB1_20:
-; AVX512-NEXT:    vpextrb $6, %xmm1, %ecx
-; AVX512-NEXT:    vpextrb $6, %xmm0, %edi
-; AVX512-NEXT:    movl %edi, %eax
-; AVX512-NEXT:    addb %cl, %al
-; AVX512-NEXT:    setns %al
-; AVX512-NEXT:    addb %cl, %dil
-; AVX512-NEXT:    jno .LBB1_22
-; AVX512-NEXT:  # %bb.21:
-; AVX512-NEXT:    addb $127, %al
-; AVX512-NEXT:    movl %eax, %edi
-; AVX512-NEXT:  .LBB1_22:
-; AVX512-NEXT:    vpextrb $5, %xmm1, %ecx
-; AVX512-NEXT:    vpextrb $5, %xmm0, %ebp
-; AVX512-NEXT:    movl %ebp, %eax
-; AVX512-NEXT:    addb %cl, %al
-; AVX512-NEXT:    setns %al
-; AVX512-NEXT:    addb %cl, %bpl
-; AVX512-NEXT:    jno .LBB1_24
-; AVX512-NEXT:  # %bb.23:
-; AVX512-NEXT:    addb $127, %al
-; AVX512-NEXT:    movl %eax, %ebp
-; AVX512-NEXT:  .LBB1_24:
-; AVX512-NEXT:    vpextrb $4, %xmm1, %ecx
-; AVX512-NEXT:    vpextrb $4, %xmm0, %r11d
-; AVX512-NEXT:    movl %r11d, %eax
-; AVX512-NEXT:    addb %cl, %al
-; AVX512-NEXT:    setns %al
-; AVX512-NEXT:    addb %cl, %r11b
-; AVX512-NEXT:    jno .LBB1_26
-; AVX512-NEXT:  # %bb.25:
-; AVX512-NEXT:    addb $127, %al
-; AVX512-NEXT:    movl %eax, %r11d
-; AVX512-NEXT:  .LBB1_26:
-; AVX512-NEXT:    vpextrb $3, %xmm1, %ecx
-; AVX512-NEXT:    vpextrb $3, %xmm0, %r14d
-; AVX512-NEXT:    movl %r14d, %eax
-; AVX512-NEXT:    addb %cl, %al
-; AVX512-NEXT:    setns %al
-; AVX512-NEXT:    addb %cl, %r14b
-; AVX512-NEXT:    movl %ebp, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX512-NEXT:    jno .LBB1_28
-; AVX512-NEXT:  # %bb.27:
-; AVX512-NEXT:    addb $127, %al
-; AVX512-NEXT:    movl %eax, %r14d
-; AVX512-NEXT:  .LBB1_28:
-; AVX512-NEXT:    vpextrb $2, %xmm1, %ecx
-; AVX512-NEXT:    vpextrb $2, %xmm0, %r8d
-; AVX512-NEXT:    movl %r8d, %eax
-; AVX512-NEXT:    addb %cl, %al
-; AVX512-NEXT:    setns %al
-; AVX512-NEXT:    addb %cl, %r8b
-; AVX512-NEXT:    jno .LBB1_30
-; AVX512-NEXT:  # %bb.29:
-; AVX512-NEXT:    addb $127, %al
-; AVX512-NEXT:    movl %eax, %r8d
-; AVX512-NEXT:  .LBB1_30:
-; AVX512-NEXT:    vpextrb $0, %xmm1, %ecx
-; AVX512-NEXT:    vpextrb $0, %xmm0, %r10d
-; AVX512-NEXT:    movl %r10d, %eax
-; AVX512-NEXT:    addb %cl, %al
-; AVX512-NEXT:    setns %al
-; AVX512-NEXT:    addb %cl, %r10b
-; AVX512-NEXT:    jno .LBB1_32
-; AVX512-NEXT:  # %bb.31:
-; AVX512-NEXT:    addb $127, %al
-; AVX512-NEXT:    movl %eax, %r10d
-; AVX512-NEXT:  .LBB1_32:
-; AVX512-NEXT:    vpextrb $1, %xmm1, %ecx
-; AVX512-NEXT:    vpextrb $1, %xmm0, %r9d
-; AVX512-NEXT:    movl %r9d, %eax
-; AVX512-NEXT:    addb %cl, %al
-; AVX512-NEXT:    setns %al
-; AVX512-NEXT:    addb %cl, %r9b
-; AVX512-NEXT:    jno .LBB1_34
-; AVX512-NEXT:  # %bb.33:
-; AVX512-NEXT:    addb $127, %al
-; AVX512-NEXT:    movl %eax, %r9d
-; AVX512-NEXT:  .LBB1_34:
-; AVX512-NEXT:    vextracti128 $1, %ymm1, %xmm1
-; AVX512-NEXT:    vpextrb $15, %xmm1, %ecx
-; AVX512-NEXT:    vextracti128 $1, %ymm0, %xmm0
-; AVX512-NEXT:    vpextrb $15, %xmm0, %ebx
-; AVX512-NEXT:    movl %ebx, %eax
-; AVX512-NEXT:    addb %cl, %al
-; AVX512-NEXT:    setns %al
-; AVX512-NEXT:    movl %esi, %r12d
-; AVX512-NEXT:    addb %cl, %bl
-; AVX512-NEXT:    jno .LBB1_36
-; AVX512-NEXT:  # %bb.35:
-; AVX512-NEXT:    addb $127, %al
-; AVX512-NEXT:    movl %eax, %ebx
-; AVX512-NEXT:  .LBB1_36:
-; AVX512-NEXT:    vpextrb $14, %xmm1, %ecx
-; AVX512-NEXT:    vpextrb $14, %xmm0, %esi
-; AVX512-NEXT:    movl %esi, %eax
-; AVX512-NEXT:    addb %cl, %al
-; AVX512-NEXT:    setns %al
-; AVX512-NEXT:    movl %edx, %r13d
-; AVX512-NEXT:    addb %cl, %sil
-; AVX512-NEXT:    jno .LBB1_38
-; AVX512-NEXT:  # %bb.37:
-; AVX512-NEXT:    addb $127, %al
-; AVX512-NEXT:    movl %eax, %esi
-; AVX512-NEXT:  .LBB1_38:
-; AVX512-NEXT:    vpextrb $13, %xmm1, %ecx
-; AVX512-NEXT:    vpextrb $13, %xmm0, %edx
-; AVX512-NEXT:    movl %edx, %eax
-; AVX512-NEXT:    addb %cl, %al
-; AVX512-NEXT:    setns %al
-; AVX512-NEXT:    movl %edi, %ebp
-; AVX512-NEXT:    addb %cl, %dl
-; AVX512-NEXT:    jno .LBB1_40
-; AVX512-NEXT:  # %bb.39:
-; AVX512-NEXT:    addb $127, %al
-; AVX512-NEXT:    movl %eax, %edx
-; AVX512-NEXT:  .LBB1_40:
-; AVX512-NEXT:    vpextrb $12, %xmm1, %ecx
-; AVX512-NEXT:    vpextrb $12, %xmm0, %edi
-; AVX512-NEXT:    movl %edi, %eax
-; AVX512-NEXT:    addb %cl, %al
-; AVX512-NEXT:    setns %al
-; AVX512-NEXT:    addb %cl, %dil
-; AVX512-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX512-NEXT:    movl %edx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX512-NEXT:    jo .LBB1_41
-; AVX512-NEXT:  # %bb.42:
-; AVX512-NEXT:    movl %edi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX512-NEXT:    jmp .LBB1_43
-; AVX512-NEXT:  .LBB1_41:
-; AVX512-NEXT:    addb $127, %al
-; AVX512-NEXT:    # kill: def $al killed $al def $eax
-; AVX512-NEXT:    movl %eax, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX512-NEXT:  .LBB1_43:
-; AVX512-NEXT:    movl %ebp, %edi
-; AVX512-NEXT:    vpextrb $11, %xmm1, %ecx
-; AVX512-NEXT:    vpextrb $11, %xmm0, %r15d
-; AVX512-NEXT:    movl %r15d, %eax
-; AVX512-NEXT:    addb %cl, %al
-; AVX512-NEXT:    setns %al
-; AVX512-NEXT:    addb %cl, %r15b
-; AVX512-NEXT:    movl %r12d, %esi
-; AVX512-NEXT:    movl %r13d, %edx
-; AVX512-NEXT:    jno .LBB1_45
-; AVX512-NEXT:  # %bb.44:
-; AVX512-NEXT:    addb $127, %al
-; AVX512-NEXT:    movl %eax, %r15d
-; AVX512-NEXT:  .LBB1_45:
-; AVX512-NEXT:    vpextrb $10, %xmm1, %ecx
-; AVX512-NEXT:    vpextrb $10, %xmm0, %r12d
-; AVX512-NEXT:    movl %r12d, %eax
-; AVX512-NEXT:    addb %cl, %al
-; AVX512-NEXT:    setns %al
-; AVX512-NEXT:    addb %cl, %r12b
-; AVX512-NEXT:    jno .LBB1_47
-; AVX512-NEXT:  # %bb.46:
-; AVX512-NEXT:    addb $127, %al
-; AVX512-NEXT:    movl %eax, %r12d
-; AVX512-NEXT:  .LBB1_47:
-; AVX512-NEXT:    vpextrb $9, %xmm1, %ecx
-; AVX512-NEXT:    vpextrb $9, %xmm0, %r13d
-; AVX512-NEXT:    movl %r13d, %eax
-; AVX512-NEXT:    addb %cl, %al
-; AVX512-NEXT:    setns %al
-; AVX512-NEXT:    addb %cl, %r13b
-; AVX512-NEXT:    movl %r11d, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX512-NEXT:    movl %r14d, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX512-NEXT:    movl %r8d, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX512-NEXT:    movl %r10d, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX512-NEXT:    movl %r9d, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX512-NEXT:    movl %ebx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX512-NEXT:    jno .LBB1_49
-; AVX512-NEXT:  # %bb.48:
-; AVX512-NEXT:    addb $127, %al
-; AVX512-NEXT:    movl %eax, %r13d
-; AVX512-NEXT:  .LBB1_49:
-; AVX512-NEXT:    vpextrb $8, %xmm1, %ecx
-; AVX512-NEXT:    vpextrb $8, %xmm0, %r11d
-; AVX512-NEXT:    movl %r11d, %eax
-; AVX512-NEXT:    addb %cl, %al
-; AVX512-NEXT:    setns %al
-; AVX512-NEXT:    addb %cl, %r11b
-; AVX512-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX512-NEXT:    movl %edx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX512-NEXT:    movl %edi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX512-NEXT:    jno .LBB1_51
-; AVX512-NEXT:  # %bb.50:
-; AVX512-NEXT:    addb $127, %al
-; AVX512-NEXT:    movl %eax, %r11d
-; AVX512-NEXT:  .LBB1_51:
-; AVX512-NEXT:    vpextrb $7, %xmm1, %ecx
-; AVX512-NEXT:    vpextrb $7, %xmm0, %r10d
-; AVX512-NEXT:    movl %r10d, %eax
-; AVX512-NEXT:    addb %cl, %al
-; AVX512-NEXT:    setns %al
-; AVX512-NEXT:    addb %cl, %r10b
-; AVX512-NEXT:    jno .LBB1_53
-; AVX512-NEXT:  # %bb.52:
-; AVX512-NEXT:    addb $127, %al
-; AVX512-NEXT:    movl %eax, %r10d
-; AVX512-NEXT:  .LBB1_53:
-; AVX512-NEXT:    vpextrb $6, %xmm1, %ecx
-; AVX512-NEXT:    vpextrb $6, %xmm0, %ebp
-; AVX512-NEXT:    movl %ebp, %eax
-; AVX512-NEXT:    addb %cl, %al
-; AVX512-NEXT:    setns %al
-; AVX512-NEXT:    addb %cl, %bpl
-; AVX512-NEXT:    jno .LBB1_55
-; AVX512-NEXT:  # %bb.54:
-; AVX512-NEXT:    addb $127, %al
-; AVX512-NEXT:    movl %eax, %ebp
-; AVX512-NEXT:  .LBB1_55:
-; AVX512-NEXT:    vpextrb $5, %xmm1, %ecx
-; AVX512-NEXT:    vpextrb $5, %xmm0, %edi
-; AVX512-NEXT:    movl %edi, %eax
-; AVX512-NEXT:    addb %cl, %al
-; AVX512-NEXT:    setns %al
-; AVX512-NEXT:    addb %cl, %dil
-; AVX512-NEXT:    jno .LBB1_57
-; AVX512-NEXT:  # %bb.56:
-; AVX512-NEXT:    addb $127, %al
-; AVX512-NEXT:    movl %eax, %edi
-; AVX512-NEXT:  .LBB1_57:
-; AVX512-NEXT:    vpextrb $4, %xmm1, %edx
-; AVX512-NEXT:    vpextrb $4, %xmm0, %eax
-; AVX512-NEXT:    movl %eax, %ecx
-; AVX512-NEXT:    addb %dl, %cl
-; AVX512-NEXT:    setns %cl
-; AVX512-NEXT:    addb %dl, %al
-; AVX512-NEXT:    jno .LBB1_59
-; AVX512-NEXT:  # %bb.58:
-; AVX512-NEXT:    addb $127, %cl
-; AVX512-NEXT:    movl %ecx, %eax
-; AVX512-NEXT:  .LBB1_59:
-; AVX512-NEXT:    vpextrb $3, %xmm1, %ebx
-; AVX512-NEXT:    vpextrb $3, %xmm0, %ecx
-; AVX512-NEXT:    movl %ecx, %edx
-; AVX512-NEXT:    addb %bl, %dl
-; AVX512-NEXT:    setns %dl
-; AVX512-NEXT:    addb %bl, %cl
-; AVX512-NEXT:    jno .LBB1_61
-; AVX512-NEXT:  # %bb.60:
-; AVX512-NEXT:    addb $127, %dl
-; AVX512-NEXT:    movl %edx, %ecx
-; AVX512-NEXT:  .LBB1_61:
-; AVX512-NEXT:    vpextrb $2, %xmm1, %esi
-; AVX512-NEXT:    vpextrb $2, %xmm0, %edx
-; AVX512-NEXT:    movl %edx, %ebx
-; AVX512-NEXT:    addb %sil, %bl
-; AVX512-NEXT:    setns %bl
-; AVX512-NEXT:    addb %sil, %dl
-; AVX512-NEXT:    jno .LBB1_63
-; AVX512-NEXT:  # %bb.62:
-; AVX512-NEXT:    addb $127, %bl
-; AVX512-NEXT:    movl %ebx, %edx
-; AVX512-NEXT:  .LBB1_63:
-; AVX512-NEXT:    vpextrb $0, %xmm1, %esi
-; AVX512-NEXT:    vpextrb $0, %xmm0, %r8d
-; AVX512-NEXT:    movl %r8d, %ebx
-; AVX512-NEXT:    addb %sil, %bl
-; AVX512-NEXT:    setns %bl
-; AVX512-NEXT:    addb %sil, %r8b
-; AVX512-NEXT:    jo .LBB1_64
-; AVX512-NEXT:  # %bb.65:
-; AVX512-NEXT:    movl {{[-0-9]+}}(%r{{[sb]}}p), %r14d # 4-byte Reload
-; AVX512-NEXT:    jmp .LBB1_66
-; AVX512-NEXT:  .LBB1_64:
-; AVX512-NEXT:    movl {{[-0-9]+}}(%r{{[sb]}}p), %r14d # 4-byte Reload
-; AVX512-NEXT:    addb $127, %bl
-; AVX512-NEXT:    movl %ebx, %r8d
-; AVX512-NEXT:  .LBB1_66:
-; AVX512-NEXT:    vpextrb $1, %xmm1, %esi
-; AVX512-NEXT:    vpextrb $1, %xmm0, %r9d
-; AVX512-NEXT:    movl %r9d, %ebx
-; AVX512-NEXT:    addb %sil, %bl
-; AVX512-NEXT:    setns %bl
-; AVX512-NEXT:    addb %sil, %r9b
-; AVX512-NEXT:    jno .LBB1_68
-; AVX512-NEXT:  # %bb.67:
-; AVX512-NEXT:    addb $127, %bl
-; AVX512-NEXT:    movl %ebx, %r9d
-; AVX512-NEXT:  .LBB1_68:
-; AVX512-NEXT:    movzbl %r8b, %esi
-; AVX512-NEXT:    vmovd %esi, %xmm0
-; AVX512-NEXT:    movzbl %r9b, %esi
-; AVX512-NEXT:    vpinsrb $1, %esi, %xmm0, %xmm0
-; AVX512-NEXT:    movzbl %dl, %edx
-; AVX512-NEXT:    vpinsrb $2, %edx, %xmm0, %xmm0
-; AVX512-NEXT:    movzbl %cl, %ecx
-; AVX512-NEXT:    vpinsrb $3, %ecx, %xmm0, %xmm0
-; AVX512-NEXT:    movzbl %al, %eax
-; AVX512-NEXT:    vpinsrb $4, %eax, %xmm0, %xmm0
-; AVX512-NEXT:    movzbl %dil, %eax
-; AVX512-NEXT:    vpinsrb $5, %eax, %xmm0, %xmm0
-; AVX512-NEXT:    movzbl %bpl, %eax
-; AVX512-NEXT:    vpinsrb $6, %eax, %xmm0, %xmm0
-; AVX512-NEXT:    movzbl %r10b, %eax
-; AVX512-NEXT:    vpinsrb $7, %eax, %xmm0, %xmm0
-; AVX512-NEXT:    movzbl %r11b, %eax
-; AVX512-NEXT:    vpinsrb $8, %eax, %xmm0, %xmm0
-; AVX512-NEXT:    movzbl %r13b, %eax
-; AVX512-NEXT:    vpinsrb $9, %eax, %xmm0, %xmm0
-; AVX512-NEXT:    movzbl %r12b, %eax
-; AVX512-NEXT:    vpinsrb $10, %eax, %xmm0, %xmm0
-; AVX512-NEXT:    movzbl %r15b, %eax
-; AVX512-NEXT:    vpinsrb $11, %eax, %xmm0, %xmm0
-; AVX512-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX512-NEXT:    vpinsrb $12, %eax, %xmm0, %xmm0
-; AVX512-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX512-NEXT:    vpinsrb $13, %eax, %xmm0, %xmm0
-; AVX512-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX512-NEXT:    vpinsrb $14, %eax, %xmm0, %xmm0
-; AVX512-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX512-NEXT:    vpinsrb $15, %eax, %xmm0, %xmm0
-; AVX512-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX512-NEXT:    vmovd %eax, %xmm1
-; AVX512-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX512-NEXT:    vpinsrb $1, %eax, %xmm1, %xmm1
-; AVX512-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX512-NEXT:    vpinsrb $2, %eax, %xmm1, %xmm1
-; AVX512-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX512-NEXT:    vpinsrb $3, %eax, %xmm1, %xmm1
-; AVX512-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX512-NEXT:    vpinsrb $4, %eax, %xmm1, %xmm1
-; AVX512-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX512-NEXT:    vpinsrb $5, %eax, %xmm1, %xmm1
-; AVX512-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX512-NEXT:    vpinsrb $6, %eax, %xmm1, %xmm1
-; AVX512-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX512-NEXT:    vpinsrb $7, %eax, %xmm1, %xmm1
-; AVX512-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX512-NEXT:    vpinsrb $8, %eax, %xmm1, %xmm1
-; AVX512-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX512-NEXT:    vpinsrb $9, %eax, %xmm1, %xmm1
-; AVX512-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX512-NEXT:    vpinsrb $10, %eax, %xmm1, %xmm1
-; AVX512-NEXT:    movzbl %r14b, %eax
-; AVX512-NEXT:    vpinsrb $11, %eax, %xmm1, %xmm1
-; AVX512-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX512-NEXT:    vpinsrb $12, %eax, %xmm1, %xmm1
-; AVX512-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX512-NEXT:    vpinsrb $13, %eax, %xmm1, %xmm1
-; AVX512-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX512-NEXT:    vpinsrb $14, %eax, %xmm1, %xmm1
-; AVX512-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX512-NEXT:    vpinsrb $15, %eax, %xmm1, %xmm1
-; AVX512-NEXT:    vinserti128 $1, %xmm0, %ymm1, %ymm0
-; AVX512-NEXT:    popq %rbx
-; AVX512-NEXT:    popq %r12
-; AVX512-NEXT:    popq %r13
-; AVX512-NEXT:    popq %r14
-; AVX512-NEXT:    popq %r15
-; AVX512-NEXT:    popq %rbp
+; AVX512-NEXT:    vpaddsb %ymm1, %ymm0, %ymm0
 ; AVX512-NEXT:    retq
   %z = call <32 x i8> @llvm.sadd.sat.v32i8(<32 x i8> %x, <32 x i8> %y)
   ret <32 x i8> %z
 }
 
 define <64 x i8> @v64i8(<64 x i8> %x, <64 x i8> %y) nounwind {
-; SSE2-LABEL: v64i8:
-; SSE2:       # %bb.0:
-; SSE2-NEXT:    pushq %rbp
-; SSE2-NEXT:    pushq %r15
-; SSE2-NEXT:    pushq %r14
-; SSE2-NEXT:    pushq %r13
-; SSE2-NEXT:    pushq %r12
-; SSE2-NEXT:    pushq %rbx
-; SSE2-NEXT:    subq $232, %rsp
-; SSE2-NEXT:    movaps %xmm5, (%rsp)
-; SSE2-NEXT:    movaps %xmm1, -{{[0-9]+}}(%rsp)
-; SSE2-NEXT:    movb (%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %dl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %sil
-; SSE2-NEXT:    movl %edx, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %dl
-; SSE2-NEXT:    jno .LBB2_2
-; SSE2-NEXT:  # %bb.1:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %edx
-; SSE2-NEXT:  .LBB2_2:
-; SSE2-NEXT:    movb {{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movl %esi, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %sil
-; SSE2-NEXT:    jno .LBB2_4
-; SSE2-NEXT:  # %bb.3:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %esi
-; SSE2-NEXT:  .LBB2_4:
-; SSE2-NEXT:    movb {{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %dil
-; SSE2-NEXT:    movl %edi, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %dil
-; SSE2-NEXT:    jno .LBB2_6
-; SSE2-NEXT:  # %bb.5:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %edi
-; SSE2-NEXT:  .LBB2_6:
-; SSE2-NEXT:    movb {{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %r11b
-; SSE2-NEXT:    movl %r11d, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %r11b
-; SSE2-NEXT:    jno .LBB2_8
-; SSE2-NEXT:  # %bb.7:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %r11d
-; SSE2-NEXT:  .LBB2_8:
-; SSE2-NEXT:    movb {{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %r9b
-; SSE2-NEXT:    movl %r9d, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %r9b
-; SSE2-NEXT:    jno .LBB2_10
-; SSE2-NEXT:  # %bb.9:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %r9d
-; SSE2-NEXT:  .LBB2_10:
-; SSE2-NEXT:    movb {{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %r10b
-; SSE2-NEXT:    movl %r10d, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %r10b
-; SSE2-NEXT:    jno .LBB2_12
-; SSE2-NEXT:  # %bb.11:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %r10d
-; SSE2-NEXT:  .LBB2_12:
-; SSE2-NEXT:    movb {{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %bl
-; SSE2-NEXT:    movl %ebx, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %bl
-; SSE2-NEXT:    movb %dl, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSE2-NEXT:    jno .LBB2_14
-; SSE2-NEXT:  # %bb.13:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %ebx
-; SSE2-NEXT:  .LBB2_14:
-; SSE2-NEXT:    movb {{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %dl
-; SSE2-NEXT:    movl %edx, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %dl
-; SSE2-NEXT:    movb %bl, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSE2-NEXT:    jo .LBB2_15
-; SSE2-NEXT:  # %bb.16:
-; SSE2-NEXT:    movb %dl, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSE2-NEXT:    jmp .LBB2_17
-; SSE2-NEXT:  .LBB2_15:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movb %al, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSE2-NEXT:  .LBB2_17:
-; SSE2-NEXT:    movb {{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %dl
-; SSE2-NEXT:    movl %edx, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %dl
-; SSE2-NEXT:    jno .LBB2_19
-; SSE2-NEXT:  # %bb.18:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %edx
-; SSE2-NEXT:  .LBB2_19:
-; SSE2-NEXT:    movb {{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %r12b
-; SSE2-NEXT:    movl %r12d, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %r12b
-; SSE2-NEXT:    jno .LBB2_21
-; SSE2-NEXT:  # %bb.20:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %r12d
-; SSE2-NEXT:  .LBB2_21:
-; SSE2-NEXT:    movb {{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %bl
-; SSE2-NEXT:    movl %ebx, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %bl
-; SSE2-NEXT:    movb %r9b, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSE2-NEXT:    jno .LBB2_23
-; SSE2-NEXT:  # %bb.22:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %ebx
-; SSE2-NEXT:  .LBB2_23:
-; SSE2-NEXT:    movb %bl, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSE2-NEXT:    movb {{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %r9b
-; SSE2-NEXT:    movl %r9d, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %r9b
-; SSE2-NEXT:    jno .LBB2_25
-; SSE2-NEXT:  # %bb.24:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %r9d
-; SSE2-NEXT:  .LBB2_25:
-; SSE2-NEXT:    movl %edi, %r8d
-; SSE2-NEXT:    movb {{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %bl
-; SSE2-NEXT:    movl %ebx, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %bl
-; SSE2-NEXT:    jo .LBB2_26
-; SSE2-NEXT:  # %bb.27:
-; SSE2-NEXT:    movb %bl, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSE2-NEXT:    movl %esi, %edi
-; SSE2-NEXT:    jmp .LBB2_28
-; SSE2-NEXT:  .LBB2_26:
-; SSE2-NEXT:    movl %esi, %edi
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movb %al, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSE2-NEXT:  .LBB2_28:
-; SSE2-NEXT:    movb {{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %sil
-; SSE2-NEXT:    movl %esi, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %sil
-; SSE2-NEXT:    movb %dl, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSE2-NEXT:    jno .LBB2_30
-; SSE2-NEXT:  # %bb.29:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %esi
-; SSE2-NEXT:  .LBB2_30:
-; SSE2-NEXT:    movb {{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %dl
-; SSE2-NEXT:    movl %edx, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %dl
-; SSE2-NEXT:    jno .LBB2_32
-; SSE2-NEXT:  # %bb.31:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %edx
-; SSE2-NEXT:  .LBB2_32:
-; SSE2-NEXT:    movb {{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %bl
-; SSE2-NEXT:    movl %ebx, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %bl
-; SSE2-NEXT:    movaps %xmm6, {{[0-9]+}}(%rsp)
-; SSE2-NEXT:    movaps %xmm2, -{{[0-9]+}}(%rsp)
-; SSE2-NEXT:    movb %r10b, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSE2-NEXT:    movb %r12b, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSE2-NEXT:    movb %r9b, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSE2-NEXT:    movb %dl, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSE2-NEXT:    jno .LBB2_34
-; SSE2-NEXT:  # %bb.33:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %ebx
-; SSE2-NEXT:  .LBB2_34:
-; SSE2-NEXT:    movb {{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %dl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %r9b
-; SSE2-NEXT:    movl %edx, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %dl
-; SSE2-NEXT:    jno .LBB2_36
-; SSE2-NEXT:  # %bb.35:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %edx
-; SSE2-NEXT:  .LBB2_36:
-; SSE2-NEXT:    movb {{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movl %r9d, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %r9b
-; SSE2-NEXT:    movb %dl, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSE2-NEXT:    movb %sil, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSE2-NEXT:    jno .LBB2_38
-; SSE2-NEXT:  # %bb.37:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %r9d
-; SSE2-NEXT:  .LBB2_38:
-; SSE2-NEXT:    movb {{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %dl
-; SSE2-NEXT:    movl %edx, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %dl
-; SSE2-NEXT:    movb %dil, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSE2-NEXT:    movb %r8b, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSE2-NEXT:    movb %bl, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSE2-NEXT:    jno .LBB2_40
-; SSE2-NEXT:  # %bb.39:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %edx
-; SSE2-NEXT:  .LBB2_40:
-; SSE2-NEXT:    movb {{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %bpl
-; SSE2-NEXT:    movl %ebp, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %bpl
-; SSE2-NEXT:    movb %r11b, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSE2-NEXT:    jno .LBB2_42
-; SSE2-NEXT:  # %bb.41:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %ebp
-; SSE2-NEXT:  .LBB2_42:
-; SSE2-NEXT:    movb {{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %sil
-; SSE2-NEXT:    movl %esi, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %sil
-; SSE2-NEXT:    movb %dl, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSE2-NEXT:    jno .LBB2_44
-; SSE2-NEXT:  # %bb.43:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %esi
-; SSE2-NEXT:  .LBB2_44:
-; SSE2-NEXT:    movb {{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %dl
-; SSE2-NEXT:    movl %edx, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %dl
-; SSE2-NEXT:    jno .LBB2_46
-; SSE2-NEXT:  # %bb.45:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %edx
-; SSE2-NEXT:  .LBB2_46:
-; SSE2-NEXT:    movb {{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %dil
-; SSE2-NEXT:    movl %edi, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %dil
-; SSE2-NEXT:    jno .LBB2_48
-; SSE2-NEXT:  # %bb.47:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %edi
-; SSE2-NEXT:  .LBB2_48:
-; SSE2-NEXT:    movb {{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %bl
-; SSE2-NEXT:    movl %ebx, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %bl
-; SSE2-NEXT:    movb %bl, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSE2-NEXT:    movb %sil, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSE2-NEXT:    movb %dl, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSE2-NEXT:    jno .LBB2_50
-; SSE2-NEXT:  # %bb.49:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movb %al, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSE2-NEXT:  .LBB2_50:
-; SSE2-NEXT:    movb {{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %sil
-; SSE2-NEXT:    movl %esi, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %sil
-; SSE2-NEXT:    movb %dil, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSE2-NEXT:    jno .LBB2_52
-; SSE2-NEXT:  # %bb.51:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %esi
-; SSE2-NEXT:  .LBB2_52:
-; SSE2-NEXT:    movb {{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %dil
-; SSE2-NEXT:    movl %edi, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %dil
-; SSE2-NEXT:    jno .LBB2_54
-; SSE2-NEXT:  # %bb.53:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %edi
-; SSE2-NEXT:  .LBB2_54:
-; SSE2-NEXT:    movb {{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %r14b
-; SSE2-NEXT:    movl %r14d, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %r14b
-; SSE2-NEXT:    movb %r9b, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSE2-NEXT:    jno .LBB2_56
-; SSE2-NEXT:  # %bb.55:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %r14d
-; SSE2-NEXT:  .LBB2_56:
-; SSE2-NEXT:    movb {{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %r9b
-; SSE2-NEXT:    movl %r9d, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %r9b
-; SSE2-NEXT:    movb %bpl, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSE2-NEXT:    jno .LBB2_58
-; SSE2-NEXT:  # %bb.57:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %r9d
-; SSE2-NEXT:  .LBB2_58:
-; SSE2-NEXT:    movb {{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %dl
-; SSE2-NEXT:    movl %edx, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %dl
-; SSE2-NEXT:    movb %dl, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSE2-NEXT:    jno .LBB2_60
-; SSE2-NEXT:  # %bb.59:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movb %al, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSE2-NEXT:  .LBB2_60:
-; SSE2-NEXT:    movb {{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %dl
-; SSE2-NEXT:    movl %edx, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %dl
-; SSE2-NEXT:    jno .LBB2_62
-; SSE2-NEXT:  # %bb.61:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %edx
-; SSE2-NEXT:  .LBB2_62:
-; SSE2-NEXT:    movb {{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %r15b
-; SSE2-NEXT:    movl %r15d, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %r15b
-; SSE2-NEXT:    jno .LBB2_64
-; SSE2-NEXT:  # %bb.63:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %r15d
-; SSE2-NEXT:  .LBB2_64:
-; SSE2-NEXT:    movb {{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %bl
-; SSE2-NEXT:    movl %ebx, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %bl
-; SSE2-NEXT:    movaps %xmm7, {{[0-9]+}}(%rsp)
-; SSE2-NEXT:    movaps %xmm3, -{{[0-9]+}}(%rsp)
-; SSE2-NEXT:    movb %r14b, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSE2-NEXT:    movb %r9b, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSE2-NEXT:    movb %dl, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSE2-NEXT:    jo .LBB2_65
-; SSE2-NEXT:  # %bb.66:
-; SSE2-NEXT:    movb %bl, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSE2-NEXT:    jmp .LBB2_67
-; SSE2-NEXT:  .LBB2_65:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movb %al, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSE2-NEXT:  .LBB2_67:
-; SSE2-NEXT:    movb {{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %r11b
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %dl
-; SSE2-NEXT:    movl %r11d, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %r11b
-; SSE2-NEXT:    jno .LBB2_69
-; SSE2-NEXT:  # %bb.68:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %r11d
-; SSE2-NEXT:  .LBB2_69:
-; SSE2-NEXT:    movb {{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movl %edx, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %dl
-; SSE2-NEXT:    movb %dl, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSE2-NEXT:    movb %r15b, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSE2-NEXT:    jno .LBB2_71
-; SSE2-NEXT:  # %bb.70:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movb %al, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSE2-NEXT:  .LBB2_71:
-; SSE2-NEXT:    movb {{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %r12b
-; SSE2-NEXT:    movl %r12d, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %r12b
-; SSE2-NEXT:    jno .LBB2_73
-; SSE2-NEXT:  # %bb.72:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %r12d
-; SSE2-NEXT:  .LBB2_73:
-; SSE2-NEXT:    movb {{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %r14b
-; SSE2-NEXT:    movl %r14d, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %r14b
-; SSE2-NEXT:    jno .LBB2_75
-; SSE2-NEXT:  # %bb.74:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %r14d
-; SSE2-NEXT:  .LBB2_75:
-; SSE2-NEXT:    movb {{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %r15b
-; SSE2-NEXT:    movl %r15d, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %r15b
-; SSE2-NEXT:    jno .LBB2_77
-; SSE2-NEXT:  # %bb.76:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %r15d
-; SSE2-NEXT:  .LBB2_77:
-; SSE2-NEXT:    movb {{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %bpl
-; SSE2-NEXT:    movl %ebp, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %bpl
-; SSE2-NEXT:    jno .LBB2_79
-; SSE2-NEXT:  # %bb.78:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %ebp
-; SSE2-NEXT:  .LBB2_79:
-; SSE2-NEXT:    movb {{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %r13b
-; SSE2-NEXT:    movl %r13d, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %r13b
-; SSE2-NEXT:    jno .LBB2_81
-; SSE2-NEXT:  # %bb.80:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %r13d
-; SSE2-NEXT:  .LBB2_81:
-; SSE2-NEXT:    movb {{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %dl
-; SSE2-NEXT:    movl %edx, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %dl
-; SSE2-NEXT:    movb %dl, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSE2-NEXT:    jno .LBB2_83
-; SSE2-NEXT:  # %bb.82:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movb %al, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSE2-NEXT:  .LBB2_83:
-; SSE2-NEXT:    movb {{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %dl
-; SSE2-NEXT:    movl %edx, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %dl
-; SSE2-NEXT:    jo .LBB2_84
-; SSE2-NEXT:  # %bb.85:
-; SSE2-NEXT:    movb %dl, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSE2-NEXT:    jmp .LBB2_86
-; SSE2-NEXT:  .LBB2_84:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movb %al, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSE2-NEXT:  .LBB2_86:
-; SSE2-NEXT:    movb {{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %dl
-; SSE2-NEXT:    movl %edx, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %dl
-; SSE2-NEXT:    movb %dl, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSE2-NEXT:    jno .LBB2_88
-; SSE2-NEXT:  # %bb.87:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movb %al, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSE2-NEXT:  .LBB2_88:
-; SSE2-NEXT:    movb {{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %dl
-; SSE2-NEXT:    movl %edx, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %dl
-; SSE2-NEXT:    jo .LBB2_89
-; SSE2-NEXT:  # %bb.90:
-; SSE2-NEXT:    movb %dl, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSE2-NEXT:    jmp .LBB2_91
-; SSE2-NEXT:  .LBB2_89:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movb %al, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSE2-NEXT:  .LBB2_91:
-; SSE2-NEXT:    movb {{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %dl
-; SSE2-NEXT:    movl %edx, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %dl
-; SSE2-NEXT:    movb %dl, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSE2-NEXT:    jno .LBB2_93
-; SSE2-NEXT:  # %bb.92:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movb %al, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSE2-NEXT:  .LBB2_93:
-; SSE2-NEXT:    movb {{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %r8b
-; SSE2-NEXT:    movl %r8d, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %r8b
-; SSE2-NEXT:    jno .LBB2_95
-; SSE2-NEXT:  # %bb.94:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %r8d
-; SSE2-NEXT:  .LBB2_95:
-; SSE2-NEXT:    movb {{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %dl
-; SSE2-NEXT:    movl %edx, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %dl
-; SSE2-NEXT:    movb %dl, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSE2-NEXT:    jno .LBB2_97
-; SSE2-NEXT:  # %bb.96:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movb %al, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSE2-NEXT:  .LBB2_97:
-; SSE2-NEXT:    movb {{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %r9b
-; SSE2-NEXT:    movl %r9d, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %r9b
-; SSE2-NEXT:    jno .LBB2_99
-; SSE2-NEXT:  # %bb.98:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %r9d
-; SSE2-NEXT:  .LBB2_99:
-; SSE2-NEXT:    movb {{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %dl
-; SSE2-NEXT:    movl %edx, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %dl
-; SSE2-NEXT:    movb %dl, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSE2-NEXT:    movaps %xmm4, -{{[0-9]+}}(%rsp)
-; SSE2-NEXT:    movaps %xmm0, -{{[0-9]+}}(%rsp)
-; SSE2-NEXT:    movb %r12b, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSE2-NEXT:    jno .LBB2_101
-; SSE2-NEXT:  # %bb.100:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movb %al, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSE2-NEXT:  .LBB2_101:
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %r12b
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %r10b
-; SSE2-NEXT:    movl %r12d, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %r12b
-; SSE2-NEXT:    jno .LBB2_103
-; SSE2-NEXT:  # %bb.102:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %r12d
-; SSE2-NEXT:  .LBB2_103:
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movl %r10d, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %r10b
-; SSE2-NEXT:    jno .LBB2_105
-; SSE2-NEXT:  # %bb.104:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %r10d
-; SSE2-NEXT:  .LBB2_105:
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %dl
-; SSE2-NEXT:    movl %edx, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %dl
-; SSE2-NEXT:    jno .LBB2_107
-; SSE2-NEXT:  # %bb.106:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %edx
-; SSE2-NEXT:  .LBB2_107:
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %bl
-; SSE2-NEXT:    movl %ebx, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %bl
-; SSE2-NEXT:    movb %bl, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSE2-NEXT:    movb %r14b, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSE2-NEXT:    movb %r15b, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSE2-NEXT:    movb %bpl, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSE2-NEXT:    jno .LBB2_109
-; SSE2-NEXT:  # %bb.108:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movb %al, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSE2-NEXT:  .LBB2_109:
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %r14b
-; SSE2-NEXT:    movl %r14d, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %r14b
-; SSE2-NEXT:    movb %r13b, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSE2-NEXT:    jno .LBB2_111
-; SSE2-NEXT:  # %bb.110:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %r14d
-; SSE2-NEXT:  .LBB2_111:
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %r15b
-; SSE2-NEXT:    movl %r15d, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %r15b
-; SSE2-NEXT:    movb %r12b, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSE2-NEXT:    movb %r10b, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSE2-NEXT:    movb %dl, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSE2-NEXT:    jno .LBB2_113
-; SSE2-NEXT:  # %bb.112:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %r15d
-; SSE2-NEXT:  .LBB2_113:
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %r12b
-; SSE2-NEXT:    movl %r12d, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %r12b
-; SSE2-NEXT:    movl %r8d, %edx
-; SSE2-NEXT:    jno .LBB2_115
-; SSE2-NEXT:  # %bb.114:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %r12d
-; SSE2-NEXT:  .LBB2_115:
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %r13b
-; SSE2-NEXT:    movl %r13d, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %r13b
-; SSE2-NEXT:    movb %sil, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSE2-NEXT:    movb %dil, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSE2-NEXT:    movb %r11b, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSE2-NEXT:    jno .LBB2_117
-; SSE2-NEXT:  # %bb.116:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %r13d
-; SSE2-NEXT:  .LBB2_117:
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %sil
-; SSE2-NEXT:    movl %esi, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %sil
-; SSE2-NEXT:    jno .LBB2_119
-; SSE2-NEXT:  # %bb.118:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %esi
-; SSE2-NEXT:  .LBB2_119:
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %dil
-; SSE2-NEXT:    movl %edi, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %dil
-; SSE2-NEXT:    jno .LBB2_121
-; SSE2-NEXT:  # %bb.120:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %edi
-; SSE2-NEXT:  .LBB2_121:
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %r8b
-; SSE2-NEXT:    movl %r8d, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %r8b
-; SSE2-NEXT:    jno .LBB2_123
-; SSE2-NEXT:  # %bb.122:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %r8d
-; SSE2-NEXT:  .LBB2_123:
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %r10b
-; SSE2-NEXT:    movl %r10d, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %r10b
-; SSE2-NEXT:    movb {{[-0-9]+}}(%r{{[sb]}}p), %bpl # 1-byte Reload
-; SSE2-NEXT:    jno .LBB2_125
-; SSE2-NEXT:  # %bb.124:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %r10d
-; SSE2-NEXT:  .LBB2_125:
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %r11b
-; SSE2-NEXT:    movl %r11d, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %r11b
-; SSE2-NEXT:    jno .LBB2_127
-; SSE2-NEXT:  # %bb.126:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %r11d
-; SSE2-NEXT:  .LBB2_127:
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %bl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movl %ecx, %eax
-; SSE2-NEXT:    addb %bl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %bl, %cl
-; SSE2-NEXT:    movb %dl, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSE2-NEXT:    movb %r9b, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSE2-NEXT:    jno .LBB2_129
-; SSE2-NEXT:  # %bb.128:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %ecx
-; SSE2-NEXT:  .LBB2_129:
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %dl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %bl
-; SSE2-NEXT:    movl %ebx, %eax
-; SSE2-NEXT:    addb %dl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %dl, %bl
-; SSE2-NEXT:    jno .LBB2_131
-; SSE2-NEXT:  # %bb.130:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %ebx
-; SSE2-NEXT:  .LBB2_131:
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %dl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %r9b
-; SSE2-NEXT:    movl %r9d, %eax
-; SSE2-NEXT:    addb %dl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %dl, %r9b
-; SSE2-NEXT:    jno .LBB2_133
-; SSE2-NEXT:  # %bb.132:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %r9d
-; SSE2-NEXT:  .LBB2_133:
-; SSE2-NEXT:    movl %ebp, %eax
-; SSE2-NEXT:    movb {{[-0-9]+}}(%r{{[sb]}}p), %dl # 1-byte Reload
-; SSE2-NEXT:    movzbl %r9b, %ebp
-; SSE2-NEXT:    movl %ebp, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE2-NEXT:    movzbl %bl, %ebp
-; SSE2-NEXT:    movl %ebp, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE2-NEXT:    movzbl %cl, %ecx
-; SSE2-NEXT:    movl %ecx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE2-NEXT:    movzbl %r11b, %ecx
-; SSE2-NEXT:    movl %ecx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE2-NEXT:    movzbl %r10b, %ecx
-; SSE2-NEXT:    movl %ecx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE2-NEXT:    movzbl %r8b, %ecx
-; SSE2-NEXT:    movl %ecx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE2-NEXT:    movzbl %dil, %ecx
-; SSE2-NEXT:    movl %ecx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE2-NEXT:    movzbl %sil, %ecx
-; SSE2-NEXT:    movl %ecx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE2-NEXT:    movzbl %r13b, %ecx
-; SSE2-NEXT:    movl %ecx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE2-NEXT:    movzbl %r12b, %ecx
-; SSE2-NEXT:    movl %ecx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE2-NEXT:    movzbl %r15b, %ecx
-; SSE2-NEXT:    movl %ecx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE2-NEXT:    movzbl %r14b, %ecx
-; SSE2-NEXT:    movl %ecx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %ecx # 1-byte Folded Reload
-; SSE2-NEXT:    movl %ecx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %ecx # 1-byte Folded Reload
-; SSE2-NEXT:    movl %ecx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %ecx # 1-byte Folded Reload
-; SSE2-NEXT:    movl %ecx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %ecx # 1-byte Folded Reload
-; SSE2-NEXT:    movl %ecx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %ecx # 1-byte Folded Reload
-; SSE2-NEXT:    movl %ecx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %ecx # 1-byte Folded Reload
-; SSE2-NEXT:    movl %ecx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %ecx # 1-byte Folded Reload
-; SSE2-NEXT:    movl %ecx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %ecx # 1-byte Folded Reload
-; SSE2-NEXT:    movl %ecx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %ecx # 1-byte Folded Reload
-; SSE2-NEXT:    movl %ecx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %ecx # 1-byte Folded Reload
-; SSE2-NEXT:    movl %ecx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE2-NEXT:    movzbl %dl, %ecx
-; SSE2-NEXT:    movl %ecx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %ecx # 1-byte Folded Reload
-; SSE2-NEXT:    movl %ecx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %ecx # 1-byte Folded Reload
-; SSE2-NEXT:    movl %ecx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %ecx # 1-byte Folded Reload
-; SSE2-NEXT:    movl %ecx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %ecx # 1-byte Folded Reload
-; SSE2-NEXT:    movl %ecx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %ecx # 1-byte Folded Reload
-; SSE2-NEXT:    movl %ecx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %ecx # 1-byte Folded Reload
-; SSE2-NEXT:    movl %ecx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %ecx # 1-byte Folded Reload
-; SSE2-NEXT:    movl %ecx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %ecx # 1-byte Folded Reload
-; SSE2-NEXT:    movl %ecx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE2-NEXT:    movzbl %al, %eax
-; SSE2-NEXT:    movl %eax, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %r13d # 1-byte Folded Reload
-; SSE2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %r12d # 1-byte Folded Reload
-; SSE2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %r15d # 1-byte Folded Reload
-; SSE2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %r14d # 1-byte Folded Reload
-; SSE2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %ebp # 1-byte Folded Reload
-; SSE2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %ebx # 1-byte Folded Reload
-; SSE2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %r11d # 1-byte Folded Reload
-; SSE2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %r9d # 1-byte Folded Reload
-; SSE2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; SSE2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %ecx # 1-byte Folded Reload
-; SSE2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %edx # 1-byte Folded Reload
-; SSE2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %edi # 1-byte Folded Reload
-; SSE2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %esi # 1-byte Folded Reload
-; SSE2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %r8d # 1-byte Folded Reload
-; SSE2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %r10d # 1-byte Folded Reload
-; SSE2-NEXT:    movd {{[-0-9]+}}(%r{{[sb]}}p), %xmm14 # 4-byte Folded Reload
-; SSE2-NEXT:    # xmm14 = mem[0],zero,zero,zero
-; SSE2-NEXT:    movd {{[-0-9]+}}(%r{{[sb]}}p), %xmm2 # 4-byte Folded Reload
-; SSE2-NEXT:    # xmm2 = mem[0],zero,zero,zero
-; SSE2-NEXT:    movd {{[-0-9]+}}(%r{{[sb]}}p), %xmm12 # 4-byte Folded Reload
-; SSE2-NEXT:    # xmm12 = mem[0],zero,zero,zero
-; SSE2-NEXT:    movd {{[-0-9]+}}(%r{{[sb]}}p), %xmm4 # 4-byte Folded Reload
-; SSE2-NEXT:    # xmm4 = mem[0],zero,zero,zero
-; SSE2-NEXT:    movd {{[-0-9]+}}(%r{{[sb]}}p), %xmm8 # 4-byte Folded Reload
-; SSE2-NEXT:    # xmm8 = mem[0],zero,zero,zero
-; SSE2-NEXT:    movd {{[-0-9]+}}(%r{{[sb]}}p), %xmm6 # 4-byte Folded Reload
-; SSE2-NEXT:    # xmm6 = mem[0],zero,zero,zero
-; SSE2-NEXT:    movd {{[-0-9]+}}(%r{{[sb]}}p), %xmm10 # 4-byte Folded Reload
-; SSE2-NEXT:    # xmm10 = mem[0],zero,zero,zero
-; SSE2-NEXT:    movd {{[-0-9]+}}(%r{{[sb]}}p), %xmm5 # 4-byte Folded Reload
-; SSE2-NEXT:    # xmm5 = mem[0],zero,zero,zero
-; SSE2-NEXT:    movd {{[-0-9]+}}(%r{{[sb]}}p), %xmm13 # 4-byte Folded Reload
-; SSE2-NEXT:    # xmm13 = mem[0],zero,zero,zero
-; SSE2-NEXT:    movd {{[-0-9]+}}(%r{{[sb]}}p), %xmm7 # 4-byte Folded Reload
-; SSE2-NEXT:    # xmm7 = mem[0],zero,zero,zero
-; SSE2-NEXT:    movd {{[-0-9]+}}(%r{{[sb]}}p), %xmm11 # 4-byte Folded Reload
-; SSE2-NEXT:    # xmm11 = mem[0],zero,zero,zero
-; SSE2-NEXT:    movd {{[-0-9]+}}(%r{{[sb]}}p), %xmm3 # 4-byte Folded Reload
-; SSE2-NEXT:    # xmm3 = mem[0],zero,zero,zero
-; SSE2-NEXT:    movss {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 4-byte Reload
-; SSE2-NEXT:    # xmm0 = mem[0],zero,zero,zero
-; SSE2-NEXT:    movaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; SSE2-NEXT:    movd {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 4-byte Folded Reload
-; SSE2-NEXT:    # xmm1 = mem[0],zero,zero,zero
-; SSE2-NEXT:    movd {{[-0-9]+}}(%r{{[sb]}}p), %xmm15 # 4-byte Folded Reload
-; SSE2-NEXT:    # xmm15 = mem[0],zero,zero,zero
-; SSE2-NEXT:    movd {{[-0-9]+}}(%r{{[sb]}}p), %xmm9 # 4-byte Folded Reload
-; SSE2-NEXT:    # xmm9 = mem[0],zero,zero,zero
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm2 = xmm2[0],xmm14[0],xmm2[1],xmm14[1],xmm2[2],xmm14[2],xmm2[3],xmm14[3],xmm2[4],xmm14[4],xmm2[5],xmm14[5],xmm2[6],xmm14[6],xmm2[7],xmm14[7]
-; SSE2-NEXT:    movss {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 4-byte Reload
-; SSE2-NEXT:    # xmm0 = mem[0],zero,zero,zero
-; SSE2-NEXT:    movaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm4 = xmm4[0],xmm12[0],xmm4[1],xmm12[1],xmm4[2],xmm12[2],xmm4[3],xmm12[3],xmm4[4],xmm12[4],xmm4[5],xmm12[5],xmm4[6],xmm12[6],xmm4[7],xmm12[7]
-; SSE2-NEXT:    movd {{[-0-9]+}}(%r{{[sb]}}p), %xmm12 # 4-byte Folded Reload
-; SSE2-NEXT:    # xmm12 = mem[0],zero,zero,zero
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm4 = xmm4[0],xmm2[0],xmm4[1],xmm2[1],xmm4[2],xmm2[2],xmm4[3],xmm2[3]
-; SSE2-NEXT:    movss {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 4-byte Reload
-; SSE2-NEXT:    # xmm0 = mem[0],zero,zero,zero
-; SSE2-NEXT:    movaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm6 = xmm6[0],xmm8[0],xmm6[1],xmm8[1],xmm6[2],xmm8[2],xmm6[3],xmm8[3],xmm6[4],xmm8[4],xmm6[5],xmm8[5],xmm6[6],xmm8[6],xmm6[7],xmm8[7]
-; SSE2-NEXT:    movd {{[-0-9]+}}(%r{{[sb]}}p), %xmm8 # 4-byte Folded Reload
-; SSE2-NEXT:    # xmm8 = mem[0],zero,zero,zero
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm5 = xmm5[0],xmm10[0],xmm5[1],xmm10[1],xmm5[2],xmm10[2],xmm5[3],xmm10[3],xmm5[4],xmm10[4],xmm5[5],xmm10[5],xmm5[6],xmm10[6],xmm5[7],xmm10[7]
-; SSE2-NEXT:    movss {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 4-byte Reload
-; SSE2-NEXT:    # xmm0 = mem[0],zero,zero,zero
-; SSE2-NEXT:    movaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm5 = xmm5[0],xmm6[0],xmm5[1],xmm6[1],xmm5[2],xmm6[2],xmm5[3],xmm6[3]
-; SSE2-NEXT:    movd {{[-0-9]+}}(%r{{[sb]}}p), %xmm6 # 4-byte Folded Reload
-; SSE2-NEXT:    # xmm6 = mem[0],zero,zero,zero
-; SSE2-NEXT:    punpckldq {{.*#+}} xmm5 = xmm5[0],xmm4[0],xmm5[1],xmm4[1]
-; SSE2-NEXT:    movd {{[-0-9]+}}(%r{{[sb]}}p), %xmm14 # 4-byte Folded Reload
-; SSE2-NEXT:    # xmm14 = mem[0],zero,zero,zero
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm7 = xmm7[0],xmm13[0],xmm7[1],xmm13[1],xmm7[2],xmm13[2],xmm7[3],xmm13[3],xmm7[4],xmm13[4],xmm7[5],xmm13[5],xmm7[6],xmm13[6],xmm7[7],xmm13[7]
-; SSE2-NEXT:    movd {{[-0-9]+}}(%r{{[sb]}}p), %xmm2 # 4-byte Folded Reload
-; SSE2-NEXT:    # xmm2 = mem[0],zero,zero,zero
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm3 = xmm3[0],xmm11[0],xmm3[1],xmm11[1],xmm3[2],xmm11[2],xmm3[3],xmm11[3],xmm3[4],xmm11[4],xmm3[5],xmm11[5],xmm3[6],xmm11[6],xmm3[7],xmm11[7]
-; SSE2-NEXT:    movd {{[-0-9]+}}(%r{{[sb]}}p), %xmm11 # 4-byte Folded Reload
-; SSE2-NEXT:    # xmm11 = mem[0],zero,zero,zero
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm3 = xmm3[0],xmm7[0],xmm3[1],xmm7[1],xmm3[2],xmm7[2],xmm3[3],xmm7[3]
-; SSE2-NEXT:    movd {{[-0-9]+}}(%r{{[sb]}}p), %xmm10 # 4-byte Folded Reload
-; SSE2-NEXT:    # xmm10 = mem[0],zero,zero,zero
-; SSE2-NEXT:    punpcklbw {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Folded Reload
-; SSE2-NEXT:    # xmm1 = xmm1[0],mem[0],xmm1[1],mem[1],xmm1[2],mem[2],xmm1[3],mem[3],xmm1[4],mem[4],xmm1[5],mem[5],xmm1[6],mem[6],xmm1[7],mem[7]
-; SSE2-NEXT:    movd {{[-0-9]+}}(%r{{[sb]}}p), %xmm13 # 4-byte Folded Reload
-; SSE2-NEXT:    # xmm13 = mem[0],zero,zero,zero
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm9 = xmm9[0],xmm15[0],xmm9[1],xmm15[1],xmm9[2],xmm15[2],xmm9[3],xmm15[3],xmm9[4],xmm15[4],xmm9[5],xmm15[5],xmm9[6],xmm15[6],xmm9[7],xmm15[7]
-; SSE2-NEXT:    movd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 4-byte Folded Reload
-; SSE2-NEXT:    # xmm0 = mem[0],zero,zero,zero
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm9 = xmm9[0],xmm1[0],xmm9[1],xmm1[1],xmm9[2],xmm1[2],xmm9[3],xmm1[3]
-; SSE2-NEXT:    movd {{[-0-9]+}}(%r{{[sb]}}p), %xmm15 # 4-byte Folded Reload
-; SSE2-NEXT:    # xmm15 = mem[0],zero,zero,zero
-; SSE2-NEXT:    punpckldq {{.*#+}} xmm9 = xmm9[0],xmm3[0],xmm9[1],xmm3[1]
-; SSE2-NEXT:    movd {{[-0-9]+}}(%r{{[sb]}}p), %xmm7 # 4-byte Folded Reload
-; SSE2-NEXT:    # xmm7 = mem[0],zero,zero,zero
-; SSE2-NEXT:    punpcklqdq {{.*#+}} xmm9 = xmm9[0],xmm5[0]
-; SSE2-NEXT:    movd {{[-0-9]+}}(%r{{[sb]}}p), %xmm5 # 4-byte Folded Reload
-; SSE2-NEXT:    # xmm5 = mem[0],zero,zero,zero
-; SSE2-NEXT:    punpcklbw {{[-0-9]+}}(%r{{[sb]}}p), %xmm12 # 16-byte Folded Reload
-; SSE2-NEXT:    # xmm12 = xmm12[0],mem[0],xmm12[1],mem[1],xmm12[2],mem[2],xmm12[3],mem[3],xmm12[4],mem[4],xmm12[5],mem[5],xmm12[6],mem[6],xmm12[7],mem[7]
-; SSE2-NEXT:    movd {{[-0-9]+}}(%r{{[sb]}}p), %xmm4 # 4-byte Folded Reload
-; SSE2-NEXT:    # xmm4 = mem[0],zero,zero,zero
-; SSE2-NEXT:    punpcklbw {{[-0-9]+}}(%r{{[sb]}}p), %xmm8 # 16-byte Folded Reload
-; SSE2-NEXT:    # xmm8 = xmm8[0],mem[0],xmm8[1],mem[1],xmm8[2],mem[2],xmm8[3],mem[3],xmm8[4],mem[4],xmm8[5],mem[5],xmm8[6],mem[6],xmm8[7],mem[7]
-; SSE2-NEXT:    movd %r13d, %xmm1
-; SSE2-NEXT:    movdqa %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm8 = xmm8[0],xmm12[0],xmm8[1],xmm12[1],xmm8[2],xmm12[2],xmm8[3],xmm12[3]
-; SSE2-NEXT:    movd %r12d, %xmm1
-; SSE2-NEXT:    punpcklbw {{[-0-9]+}}(%r{{[sb]}}p), %xmm6 # 16-byte Folded Reload
-; SSE2-NEXT:    # xmm6 = xmm6[0],mem[0],xmm6[1],mem[1],xmm6[2],mem[2],xmm6[3],mem[3],xmm6[4],mem[4],xmm6[5],mem[5],xmm6[6],mem[6],xmm6[7],mem[7]
-; SSE2-NEXT:    movd %r15d, %xmm3
-; SSE2-NEXT:    movdqa %xmm3, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm2 = xmm2[0],xmm14[0],xmm2[1],xmm14[1],xmm2[2],xmm14[2],xmm2[3],xmm14[3],xmm2[4],xmm14[4],xmm2[5],xmm14[5],xmm2[6],xmm14[6],xmm2[7],xmm14[7]
-; SSE2-NEXT:    movd %r14d, %xmm14
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm2 = xmm2[0],xmm6[0],xmm2[1],xmm6[1],xmm2[2],xmm6[2],xmm2[3],xmm6[3]
-; SSE2-NEXT:    movd %ebp, %xmm3
-; SSE2-NEXT:    movdqa %xmm3, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; SSE2-NEXT:    punpckldq {{.*#+}} xmm2 = xmm2[0],xmm8[0],xmm2[1],xmm8[1]
-; SSE2-NEXT:    movd %ebx, %xmm12
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm10 = xmm10[0],xmm11[0],xmm10[1],xmm11[1],xmm10[2],xmm11[2],xmm10[3],xmm11[3],xmm10[4],xmm11[4],xmm10[5],xmm11[5],xmm10[6],xmm11[6],xmm10[7],xmm11[7]
-; SSE2-NEXT:    movd %r11d, %xmm8
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm13[0],xmm0[1],xmm13[1],xmm0[2],xmm13[2],xmm0[3],xmm13[3],xmm0[4],xmm13[4],xmm0[5],xmm13[5],xmm0[6],xmm13[6],xmm0[7],xmm13[7]
-; SSE2-NEXT:    movd %r9d, %xmm3
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm10[0],xmm0[1],xmm10[1],xmm0[2],xmm10[2],xmm0[3],xmm10[3]
-; SSE2-NEXT:    movd %eax, %xmm11
-; SSE2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %r15d # 1-byte Folded Reload
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm7 = xmm7[0],xmm15[0],xmm7[1],xmm15[1],xmm7[2],xmm15[2],xmm7[3],xmm15[3],xmm7[4],xmm15[4],xmm7[5],xmm15[5],xmm7[6],xmm15[6],xmm7[7],xmm15[7]
-; SSE2-NEXT:    movd %ecx, %xmm6
-; SSE2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %r12d # 1-byte Folded Reload
-; SSE2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %r9d # 1-byte Folded Reload
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm4 = xmm4[0],xmm5[0],xmm4[1],xmm5[1],xmm4[2],xmm5[2],xmm4[3],xmm5[3],xmm4[4],xmm5[4],xmm4[5],xmm5[5],xmm4[6],xmm5[6],xmm4[7],xmm5[7]
-; SSE2-NEXT:    movd %edx, %xmm13
-; SSE2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %r11d # 1-byte Folded Reload
-; SSE2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %r14d # 1-byte Folded Reload
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm4 = xmm4[0],xmm7[0],xmm4[1],xmm7[1],xmm4[2],xmm7[2],xmm4[3],xmm7[3]
-; SSE2-NEXT:    movd %edi, %xmm5
-; SSE2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %edi # 1-byte Folded Reload
-; SSE2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %r13d # 1-byte Folded Reload
-; SSE2-NEXT:    punpckldq {{.*#+}} xmm4 = xmm4[0],xmm0[0],xmm4[1],xmm0[1]
-; SSE2-NEXT:    movd %esi, %xmm15
-; SSE2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %esi # 1-byte Folded Reload
-; SSE2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %edx # 1-byte Folded Reload
-; SSE2-NEXT:    punpcklqdq {{.*#+}} xmm4 = xmm4[0],xmm2[0]
-; SSE2-NEXT:    movd %r8d, %xmm0
-; SSE2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %r8d # 1-byte Folded Reload
-; SSE2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %ebp # 1-byte Folded Reload
-; SSE2-NEXT:    punpcklbw {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Folded Reload
-; SSE2-NEXT:    # xmm1 = xmm1[0],mem[0],xmm1[1],mem[1],xmm1[2],mem[2],xmm1[3],mem[3],xmm1[4],mem[4],xmm1[5],mem[5],xmm1[6],mem[6],xmm1[7],mem[7]
-; SSE2-NEXT:    movd %r10d, %xmm10
-; SSE2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %ecx # 1-byte Folded Reload
-; SSE2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %r10d # 1-byte Folded Reload
-; SSE2-NEXT:    punpcklbw {{[-0-9]+}}(%r{{[sb]}}p), %xmm14 # 16-byte Folded Reload
-; SSE2-NEXT:    # xmm14 = xmm14[0],mem[0],xmm14[1],mem[1],xmm14[2],mem[2],xmm14[3],mem[3],xmm14[4],mem[4],xmm14[5],mem[5],xmm14[6],mem[6],xmm14[7],mem[7]
-; SSE2-NEXT:    movd %r15d, %xmm2
-; SSE2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %ebx # 1-byte Folded Reload
-; SSE2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %r15d # 1-byte Folded Reload
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm14 = xmm14[0],xmm1[0],xmm14[1],xmm1[1],xmm14[2],xmm1[2],xmm14[3],xmm1[3]
-; SSE2-NEXT:    movd %r12d, %xmm1
-; SSE2-NEXT:    movdqa %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; SSE2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; SSE2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %r12d # 1-byte Folded Reload
-; SSE2-NEXT:    punpcklbw {{[-0-9]+}}(%r{{[sb]}}p), %xmm12 # 16-byte Folded Reload
-; SSE2-NEXT:    # xmm12 = xmm12[0],mem[0],xmm12[1],mem[1],xmm12[2],mem[2],xmm12[3],mem[3],xmm12[4],mem[4],xmm12[5],mem[5],xmm12[6],mem[6],xmm12[7],mem[7]
-; SSE2-NEXT:    movd %r9d, %xmm1
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm3 = xmm3[0],xmm8[0],xmm3[1],xmm8[1],xmm3[2],xmm8[2],xmm3[3],xmm8[3],xmm3[4],xmm8[4],xmm3[5],xmm8[5],xmm3[6],xmm8[6],xmm3[7],xmm8[7]
-; SSE2-NEXT:    movd %r11d, %xmm8
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm3 = xmm3[0],xmm12[0],xmm3[1],xmm12[1],xmm3[2],xmm12[2],xmm3[3],xmm12[3]
-; SSE2-NEXT:    movd %r14d, %xmm12
-; SSE2-NEXT:    punpckldq {{.*#+}} xmm3 = xmm3[0],xmm14[0],xmm3[1],xmm14[1]
-; SSE2-NEXT:    movd %edi, %xmm7
-; SSE2-NEXT:    movdqa %xmm7, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm6 = xmm6[0],xmm11[0],xmm6[1],xmm11[1],xmm6[2],xmm11[2],xmm6[3],xmm11[3],xmm6[4],xmm11[4],xmm6[5],xmm11[5],xmm6[6],xmm11[6],xmm6[7],xmm11[7]
-; SSE2-NEXT:    movd %r13d, %xmm11
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm5 = xmm5[0],xmm13[0],xmm5[1],xmm13[1],xmm5[2],xmm13[2],xmm5[3],xmm13[3],xmm5[4],xmm13[4],xmm5[5],xmm13[5],xmm5[6],xmm13[6],xmm5[7],xmm13[7]
-; SSE2-NEXT:    movd %esi, %xmm14
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm5 = xmm5[0],xmm6[0],xmm5[1],xmm6[1],xmm5[2],xmm6[2],xmm5[3],xmm6[3]
-; SSE2-NEXT:    movd %edx, %xmm6
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm15[0],xmm0[1],xmm15[1],xmm0[2],xmm15[2],xmm0[3],xmm15[3],xmm0[4],xmm15[4],xmm0[5],xmm15[5],xmm0[6],xmm15[6],xmm0[7],xmm15[7]
-; SSE2-NEXT:    movd %r8d, %xmm15
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm2 = xmm2[0],xmm10[0],xmm2[1],xmm10[1],xmm2[2],xmm10[2],xmm2[3],xmm10[3],xmm2[4],xmm10[4],xmm2[5],xmm10[5],xmm2[6],xmm10[6],xmm2[7],xmm10[7]
-; SSE2-NEXT:    movd %ebp, %xmm7
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm2 = xmm2[0],xmm0[0],xmm2[1],xmm0[1],xmm2[2],xmm0[2],xmm2[3],xmm0[3]
-; SSE2-NEXT:    movd %ecx, %xmm13
-; SSE2-NEXT:    punpckldq {{.*#+}} xmm2 = xmm2[0],xmm5[0],xmm2[1],xmm5[1]
-; SSE2-NEXT:    movd %r10d, %xmm5
-; SSE2-NEXT:    punpcklqdq {{.*#+}} xmm2 = xmm2[0],xmm3[0]
-; SSE2-NEXT:    movd %ebx, %xmm10
-; SSE2-NEXT:    punpcklbw {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Folded Reload
-; SSE2-NEXT:    # xmm1 = xmm1[0],mem[0],xmm1[1],mem[1],xmm1[2],mem[2],xmm1[3],mem[3],xmm1[4],mem[4],xmm1[5],mem[5],xmm1[6],mem[6],xmm1[7],mem[7]
-; SSE2-NEXT:    movd %r15d, %xmm0
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm12 = xmm12[0],xmm8[0],xmm12[1],xmm8[1],xmm12[2],xmm8[2],xmm12[3],xmm8[3],xmm12[4],xmm8[4],xmm12[5],xmm8[5],xmm12[6],xmm8[6],xmm12[7],xmm8[7]
-; SSE2-NEXT:    movd %eax, %xmm8
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm12 = xmm12[0],xmm1[0],xmm12[1],xmm1[1],xmm12[2],xmm1[2],xmm12[3],xmm1[3]
-; SSE2-NEXT:    movd %r12d, %xmm3
-; SSE2-NEXT:    punpcklbw {{[-0-9]+}}(%r{{[sb]}}p), %xmm11 # 16-byte Folded Reload
-; SSE2-NEXT:    # xmm11 = xmm11[0],mem[0],xmm11[1],mem[1],xmm11[2],mem[2],xmm11[3],mem[3],xmm11[4],mem[4],xmm11[5],mem[5],xmm11[6],mem[6],xmm11[7],mem[7]
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm6 = xmm6[0],xmm14[0],xmm6[1],xmm14[1],xmm6[2],xmm14[2],xmm6[3],xmm14[3],xmm6[4],xmm14[4],xmm6[5],xmm14[5],xmm6[6],xmm14[6],xmm6[7],xmm14[7]
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm6 = xmm6[0],xmm11[0],xmm6[1],xmm11[1],xmm6[2],xmm11[2],xmm6[3],xmm11[3]
-; SSE2-NEXT:    punpckldq {{.*#+}} xmm6 = xmm6[0],xmm12[0],xmm6[1],xmm12[1]
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm7 = xmm7[0],xmm15[0],xmm7[1],xmm15[1],xmm7[2],xmm15[2],xmm7[3],xmm15[3],xmm7[4],xmm15[4],xmm7[5],xmm15[5],xmm7[6],xmm15[6],xmm7[7],xmm15[7]
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm5 = xmm5[0],xmm13[0],xmm5[1],xmm13[1],xmm5[2],xmm13[2],xmm5[3],xmm13[3],xmm5[4],xmm13[4],xmm5[5],xmm13[5],xmm5[6],xmm13[6],xmm5[7],xmm13[7]
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm5 = xmm5[0],xmm7[0],xmm5[1],xmm7[1],xmm5[2],xmm7[2],xmm5[3],xmm7[3]
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm10[0],xmm0[1],xmm10[1],xmm0[2],xmm10[2],xmm0[3],xmm10[3],xmm0[4],xmm10[4],xmm0[5],xmm10[5],xmm0[6],xmm10[6],xmm0[7],xmm10[7]
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm3 = xmm3[0],xmm8[0],xmm3[1],xmm8[1],xmm3[2],xmm8[2],xmm3[3],xmm8[3],xmm3[4],xmm8[4],xmm3[5],xmm8[5],xmm3[6],xmm8[6],xmm3[7],xmm8[7]
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm3 = xmm3[0],xmm0[0],xmm3[1],xmm0[1],xmm3[2],xmm0[2],xmm3[3],xmm0[3]
-; SSE2-NEXT:    punpckldq {{.*#+}} xmm3 = xmm3[0],xmm5[0],xmm3[1],xmm5[1]
-; SSE2-NEXT:    punpcklqdq {{.*#+}} xmm3 = xmm3[0],xmm6[0]
-; SSE2-NEXT:    movdqa %xmm9, %xmm0
-; SSE2-NEXT:    movdqa %xmm4, %xmm1
-; SSE2-NEXT:    addq $232, %rsp
-; SSE2-NEXT:    popq %rbx
-; SSE2-NEXT:    popq %r12
-; SSE2-NEXT:    popq %r13
-; SSE2-NEXT:    popq %r14
-; SSE2-NEXT:    popq %r15
-; SSE2-NEXT:    popq %rbp
-; SSE2-NEXT:    retq
-;
-; SSSE3-LABEL: v64i8:
-; SSSE3:       # %bb.0:
-; SSSE3-NEXT:    pushq %rbp
-; SSSE3-NEXT:    pushq %r15
-; SSSE3-NEXT:    pushq %r14
-; SSSE3-NEXT:    pushq %r13
-; SSSE3-NEXT:    pushq %r12
-; SSSE3-NEXT:    pushq %rbx
-; SSSE3-NEXT:    subq $232, %rsp
-; SSSE3-NEXT:    movaps %xmm5, (%rsp)
-; SSSE3-NEXT:    movaps %xmm1, -{{[0-9]+}}(%rsp)
-; SSSE3-NEXT:    movb (%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %dl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %sil
-; SSSE3-NEXT:    movl %edx, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %dl
-; SSSE3-NEXT:    jno .LBB2_2
-; SSSE3-NEXT:  # %bb.1:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %edx
-; SSSE3-NEXT:  .LBB2_2:
-; SSSE3-NEXT:    movb {{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movl %esi, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %sil
-; SSSE3-NEXT:    jno .LBB2_4
-; SSSE3-NEXT:  # %bb.3:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %esi
-; SSSE3-NEXT:  .LBB2_4:
-; SSSE3-NEXT:    movb {{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %dil
-; SSSE3-NEXT:    movl %edi, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %dil
-; SSSE3-NEXT:    jno .LBB2_6
-; SSSE3-NEXT:  # %bb.5:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %edi
-; SSSE3-NEXT:  .LBB2_6:
-; SSSE3-NEXT:    movb {{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %r11b
-; SSSE3-NEXT:    movl %r11d, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %r11b
-; SSSE3-NEXT:    jno .LBB2_8
-; SSSE3-NEXT:  # %bb.7:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %r11d
-; SSSE3-NEXT:  .LBB2_8:
-; SSSE3-NEXT:    movb {{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %r9b
-; SSSE3-NEXT:    movl %r9d, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %r9b
-; SSSE3-NEXT:    jno .LBB2_10
-; SSSE3-NEXT:  # %bb.9:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %r9d
-; SSSE3-NEXT:  .LBB2_10:
-; SSSE3-NEXT:    movb {{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %r10b
-; SSSE3-NEXT:    movl %r10d, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %r10b
-; SSSE3-NEXT:    jno .LBB2_12
-; SSSE3-NEXT:  # %bb.11:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %r10d
-; SSSE3-NEXT:  .LBB2_12:
-; SSSE3-NEXT:    movb {{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %bl
-; SSSE3-NEXT:    movl %ebx, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %bl
-; SSSE3-NEXT:    movb %dl, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSSE3-NEXT:    jno .LBB2_14
-; SSSE3-NEXT:  # %bb.13:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %ebx
-; SSSE3-NEXT:  .LBB2_14:
-; SSSE3-NEXT:    movb {{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %dl
-; SSSE3-NEXT:    movl %edx, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %dl
-; SSSE3-NEXT:    movb %bl, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSSE3-NEXT:    jo .LBB2_15
-; SSSE3-NEXT:  # %bb.16:
-; SSSE3-NEXT:    movb %dl, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSSE3-NEXT:    jmp .LBB2_17
-; SSSE3-NEXT:  .LBB2_15:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movb %al, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSSE3-NEXT:  .LBB2_17:
-; SSSE3-NEXT:    movb {{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %dl
-; SSSE3-NEXT:    movl %edx, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %dl
-; SSSE3-NEXT:    jno .LBB2_19
-; SSSE3-NEXT:  # %bb.18:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %edx
-; SSSE3-NEXT:  .LBB2_19:
-; SSSE3-NEXT:    movb {{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %r12b
-; SSSE3-NEXT:    movl %r12d, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %r12b
-; SSSE3-NEXT:    jno .LBB2_21
-; SSSE3-NEXT:  # %bb.20:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %r12d
-; SSSE3-NEXT:  .LBB2_21:
-; SSSE3-NEXT:    movb {{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %bl
-; SSSE3-NEXT:    movl %ebx, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %bl
-; SSSE3-NEXT:    movb %r9b, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSSE3-NEXT:    jno .LBB2_23
-; SSSE3-NEXT:  # %bb.22:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %ebx
-; SSSE3-NEXT:  .LBB2_23:
-; SSSE3-NEXT:    movb %bl, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSSE3-NEXT:    movb {{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %r9b
-; SSSE3-NEXT:    movl %r9d, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %r9b
-; SSSE3-NEXT:    jno .LBB2_25
-; SSSE3-NEXT:  # %bb.24:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %r9d
-; SSSE3-NEXT:  .LBB2_25:
-; SSSE3-NEXT:    movl %edi, %r8d
-; SSSE3-NEXT:    movb {{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %bl
-; SSSE3-NEXT:    movl %ebx, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %bl
-; SSSE3-NEXT:    jo .LBB2_26
-; SSSE3-NEXT:  # %bb.27:
-; SSSE3-NEXT:    movb %bl, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSSE3-NEXT:    movl %esi, %edi
-; SSSE3-NEXT:    jmp .LBB2_28
-; SSSE3-NEXT:  .LBB2_26:
-; SSSE3-NEXT:    movl %esi, %edi
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movb %al, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSSE3-NEXT:  .LBB2_28:
-; SSSE3-NEXT:    movb {{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %sil
-; SSSE3-NEXT:    movl %esi, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %sil
-; SSSE3-NEXT:    movb %dl, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSSE3-NEXT:    jno .LBB2_30
-; SSSE3-NEXT:  # %bb.29:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %esi
-; SSSE3-NEXT:  .LBB2_30:
-; SSSE3-NEXT:    movb {{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %dl
-; SSSE3-NEXT:    movl %edx, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %dl
-; SSSE3-NEXT:    jno .LBB2_32
-; SSSE3-NEXT:  # %bb.31:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %edx
-; SSSE3-NEXT:  .LBB2_32:
-; SSSE3-NEXT:    movb {{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %bl
-; SSSE3-NEXT:    movl %ebx, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %bl
-; SSSE3-NEXT:    movaps %xmm6, {{[0-9]+}}(%rsp)
-; SSSE3-NEXT:    movaps %xmm2, -{{[0-9]+}}(%rsp)
-; SSSE3-NEXT:    movb %r10b, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSSE3-NEXT:    movb %r12b, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSSE3-NEXT:    movb %r9b, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSSE3-NEXT:    movb %dl, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSSE3-NEXT:    jno .LBB2_34
-; SSSE3-NEXT:  # %bb.33:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %ebx
-; SSSE3-NEXT:  .LBB2_34:
-; SSSE3-NEXT:    movb {{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %dl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %r9b
-; SSSE3-NEXT:    movl %edx, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %dl
-; SSSE3-NEXT:    jno .LBB2_36
-; SSSE3-NEXT:  # %bb.35:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %edx
-; SSSE3-NEXT:  .LBB2_36:
-; SSSE3-NEXT:    movb {{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movl %r9d, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %r9b
-; SSSE3-NEXT:    movb %dl, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSSE3-NEXT:    movb %sil, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSSE3-NEXT:    jno .LBB2_38
-; SSSE3-NEXT:  # %bb.37:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %r9d
-; SSSE3-NEXT:  .LBB2_38:
-; SSSE3-NEXT:    movb {{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %dl
-; SSSE3-NEXT:    movl %edx, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %dl
-; SSSE3-NEXT:    movb %dil, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSSE3-NEXT:    movb %r8b, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSSE3-NEXT:    movb %bl, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSSE3-NEXT:    jno .LBB2_40
-; SSSE3-NEXT:  # %bb.39:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %edx
-; SSSE3-NEXT:  .LBB2_40:
-; SSSE3-NEXT:    movb {{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %bpl
-; SSSE3-NEXT:    movl %ebp, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %bpl
-; SSSE3-NEXT:    movb %r11b, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSSE3-NEXT:    jno .LBB2_42
-; SSSE3-NEXT:  # %bb.41:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %ebp
-; SSSE3-NEXT:  .LBB2_42:
-; SSSE3-NEXT:    movb {{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %sil
-; SSSE3-NEXT:    movl %esi, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %sil
-; SSSE3-NEXT:    movb %dl, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSSE3-NEXT:    jno .LBB2_44
-; SSSE3-NEXT:  # %bb.43:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %esi
-; SSSE3-NEXT:  .LBB2_44:
-; SSSE3-NEXT:    movb {{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %dl
-; SSSE3-NEXT:    movl %edx, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %dl
-; SSSE3-NEXT:    jno .LBB2_46
-; SSSE3-NEXT:  # %bb.45:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %edx
-; SSSE3-NEXT:  .LBB2_46:
-; SSSE3-NEXT:    movb {{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %dil
-; SSSE3-NEXT:    movl %edi, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %dil
-; SSSE3-NEXT:    jno .LBB2_48
-; SSSE3-NEXT:  # %bb.47:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %edi
-; SSSE3-NEXT:  .LBB2_48:
-; SSSE3-NEXT:    movb {{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %bl
-; SSSE3-NEXT:    movl %ebx, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %bl
-; SSSE3-NEXT:    movb %bl, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSSE3-NEXT:    movb %sil, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSSE3-NEXT:    movb %dl, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSSE3-NEXT:    jno .LBB2_50
-; SSSE3-NEXT:  # %bb.49:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movb %al, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSSE3-NEXT:  .LBB2_50:
-; SSSE3-NEXT:    movb {{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %sil
-; SSSE3-NEXT:    movl %esi, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %sil
-; SSSE3-NEXT:    movb %dil, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSSE3-NEXT:    jno .LBB2_52
-; SSSE3-NEXT:  # %bb.51:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %esi
-; SSSE3-NEXT:  .LBB2_52:
-; SSSE3-NEXT:    movb {{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %dil
-; SSSE3-NEXT:    movl %edi, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %dil
-; SSSE3-NEXT:    jno .LBB2_54
-; SSSE3-NEXT:  # %bb.53:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %edi
-; SSSE3-NEXT:  .LBB2_54:
-; SSSE3-NEXT:    movb {{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %r14b
-; SSSE3-NEXT:    movl %r14d, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %r14b
-; SSSE3-NEXT:    movb %r9b, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSSE3-NEXT:    jno .LBB2_56
-; SSSE3-NEXT:  # %bb.55:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %r14d
-; SSSE3-NEXT:  .LBB2_56:
-; SSSE3-NEXT:    movb {{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %r9b
-; SSSE3-NEXT:    movl %r9d, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %r9b
-; SSSE3-NEXT:    movb %bpl, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSSE3-NEXT:    jno .LBB2_58
-; SSSE3-NEXT:  # %bb.57:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %r9d
-; SSSE3-NEXT:  .LBB2_58:
-; SSSE3-NEXT:    movb {{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %dl
-; SSSE3-NEXT:    movl %edx, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %dl
-; SSSE3-NEXT:    movb %dl, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSSE3-NEXT:    jno .LBB2_60
-; SSSE3-NEXT:  # %bb.59:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movb %al, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSSE3-NEXT:  .LBB2_60:
-; SSSE3-NEXT:    movb {{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %dl
-; SSSE3-NEXT:    movl %edx, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %dl
-; SSSE3-NEXT:    jno .LBB2_62
-; SSSE3-NEXT:  # %bb.61:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %edx
-; SSSE3-NEXT:  .LBB2_62:
-; SSSE3-NEXT:    movb {{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %r15b
-; SSSE3-NEXT:    movl %r15d, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %r15b
-; SSSE3-NEXT:    jno .LBB2_64
-; SSSE3-NEXT:  # %bb.63:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %r15d
-; SSSE3-NEXT:  .LBB2_64:
-; SSSE3-NEXT:    movb {{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %bl
-; SSSE3-NEXT:    movl %ebx, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %bl
-; SSSE3-NEXT:    movaps %xmm7, {{[0-9]+}}(%rsp)
-; SSSE3-NEXT:    movaps %xmm3, -{{[0-9]+}}(%rsp)
-; SSSE3-NEXT:    movb %r14b, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSSE3-NEXT:    movb %r9b, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSSE3-NEXT:    movb %dl, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSSE3-NEXT:    jo .LBB2_65
-; SSSE3-NEXT:  # %bb.66:
-; SSSE3-NEXT:    movb %bl, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSSE3-NEXT:    jmp .LBB2_67
-; SSSE3-NEXT:  .LBB2_65:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movb %al, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSSE3-NEXT:  .LBB2_67:
-; SSSE3-NEXT:    movb {{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %r11b
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %dl
-; SSSE3-NEXT:    movl %r11d, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %r11b
-; SSSE3-NEXT:    jno .LBB2_69
-; SSSE3-NEXT:  # %bb.68:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %r11d
-; SSSE3-NEXT:  .LBB2_69:
-; SSSE3-NEXT:    movb {{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movl %edx, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %dl
-; SSSE3-NEXT:    movb %dl, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSSE3-NEXT:    movb %r15b, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSSE3-NEXT:    jno .LBB2_71
-; SSSE3-NEXT:  # %bb.70:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movb %al, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSSE3-NEXT:  .LBB2_71:
-; SSSE3-NEXT:    movb {{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %r12b
-; SSSE3-NEXT:    movl %r12d, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %r12b
-; SSSE3-NEXT:    jno .LBB2_73
-; SSSE3-NEXT:  # %bb.72:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %r12d
-; SSSE3-NEXT:  .LBB2_73:
-; SSSE3-NEXT:    movb {{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %r14b
-; SSSE3-NEXT:    movl %r14d, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %r14b
-; SSSE3-NEXT:    jno .LBB2_75
-; SSSE3-NEXT:  # %bb.74:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %r14d
-; SSSE3-NEXT:  .LBB2_75:
-; SSSE3-NEXT:    movb {{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %r15b
-; SSSE3-NEXT:    movl %r15d, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %r15b
-; SSSE3-NEXT:    jno .LBB2_77
-; SSSE3-NEXT:  # %bb.76:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %r15d
-; SSSE3-NEXT:  .LBB2_77:
-; SSSE3-NEXT:    movb {{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %bpl
-; SSSE3-NEXT:    movl %ebp, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %bpl
-; SSSE3-NEXT:    jno .LBB2_79
-; SSSE3-NEXT:  # %bb.78:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %ebp
-; SSSE3-NEXT:  .LBB2_79:
-; SSSE3-NEXT:    movb {{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %r13b
-; SSSE3-NEXT:    movl %r13d, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %r13b
-; SSSE3-NEXT:    jno .LBB2_81
-; SSSE3-NEXT:  # %bb.80:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %r13d
-; SSSE3-NEXT:  .LBB2_81:
-; SSSE3-NEXT:    movb {{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %dl
-; SSSE3-NEXT:    movl %edx, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %dl
-; SSSE3-NEXT:    movb %dl, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSSE3-NEXT:    jno .LBB2_83
-; SSSE3-NEXT:  # %bb.82:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movb %al, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSSE3-NEXT:  .LBB2_83:
-; SSSE3-NEXT:    movb {{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %dl
-; SSSE3-NEXT:    movl %edx, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %dl
-; SSSE3-NEXT:    jo .LBB2_84
-; SSSE3-NEXT:  # %bb.85:
-; SSSE3-NEXT:    movb %dl, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSSE3-NEXT:    jmp .LBB2_86
-; SSSE3-NEXT:  .LBB2_84:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movb %al, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSSE3-NEXT:  .LBB2_86:
-; SSSE3-NEXT:    movb {{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %dl
-; SSSE3-NEXT:    movl %edx, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %dl
-; SSSE3-NEXT:    movb %dl, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSSE3-NEXT:    jno .LBB2_88
-; SSSE3-NEXT:  # %bb.87:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movb %al, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSSE3-NEXT:  .LBB2_88:
-; SSSE3-NEXT:    movb {{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %dl
-; SSSE3-NEXT:    movl %edx, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %dl
-; SSSE3-NEXT:    jo .LBB2_89
-; SSSE3-NEXT:  # %bb.90:
-; SSSE3-NEXT:    movb %dl, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSSE3-NEXT:    jmp .LBB2_91
-; SSSE3-NEXT:  .LBB2_89:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movb %al, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSSE3-NEXT:  .LBB2_91:
-; SSSE3-NEXT:    movb {{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %dl
-; SSSE3-NEXT:    movl %edx, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %dl
-; SSSE3-NEXT:    movb %dl, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSSE3-NEXT:    jno .LBB2_93
-; SSSE3-NEXT:  # %bb.92:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movb %al, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSSE3-NEXT:  .LBB2_93:
-; SSSE3-NEXT:    movb {{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %r8b
-; SSSE3-NEXT:    movl %r8d, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %r8b
-; SSSE3-NEXT:    jno .LBB2_95
-; SSSE3-NEXT:  # %bb.94:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %r8d
-; SSSE3-NEXT:  .LBB2_95:
-; SSSE3-NEXT:    movb {{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %dl
-; SSSE3-NEXT:    movl %edx, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %dl
-; SSSE3-NEXT:    movb %dl, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSSE3-NEXT:    jno .LBB2_97
-; SSSE3-NEXT:  # %bb.96:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movb %al, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSSE3-NEXT:  .LBB2_97:
-; SSSE3-NEXT:    movb {{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %r9b
-; SSSE3-NEXT:    movl %r9d, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %r9b
-; SSSE3-NEXT:    jno .LBB2_99
-; SSSE3-NEXT:  # %bb.98:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %r9d
-; SSSE3-NEXT:  .LBB2_99:
-; SSSE3-NEXT:    movb {{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %dl
-; SSSE3-NEXT:    movl %edx, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %dl
-; SSSE3-NEXT:    movb %dl, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSSE3-NEXT:    movaps %xmm4, -{{[0-9]+}}(%rsp)
-; SSSE3-NEXT:    movaps %xmm0, -{{[0-9]+}}(%rsp)
-; SSSE3-NEXT:    movb %r12b, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSSE3-NEXT:    jno .LBB2_101
-; SSSE3-NEXT:  # %bb.100:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movb %al, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSSE3-NEXT:  .LBB2_101:
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %r12b
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %r10b
-; SSSE3-NEXT:    movl %r12d, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %r12b
-; SSSE3-NEXT:    jno .LBB2_103
-; SSSE3-NEXT:  # %bb.102:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %r12d
-; SSSE3-NEXT:  .LBB2_103:
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movl %r10d, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %r10b
-; SSSE3-NEXT:    jno .LBB2_105
-; SSSE3-NEXT:  # %bb.104:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %r10d
-; SSSE3-NEXT:  .LBB2_105:
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %dl
-; SSSE3-NEXT:    movl %edx, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %dl
-; SSSE3-NEXT:    jno .LBB2_107
-; SSSE3-NEXT:  # %bb.106:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %edx
-; SSSE3-NEXT:  .LBB2_107:
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %bl
-; SSSE3-NEXT:    movl %ebx, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %bl
-; SSSE3-NEXT:    movb %bl, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSSE3-NEXT:    movb %r14b, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSSE3-NEXT:    movb %r15b, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSSE3-NEXT:    movb %bpl, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSSE3-NEXT:    jno .LBB2_109
-; SSSE3-NEXT:  # %bb.108:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movb %al, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSSE3-NEXT:  .LBB2_109:
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %r14b
-; SSSE3-NEXT:    movl %r14d, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %r14b
-; SSSE3-NEXT:    movb %r13b, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSSE3-NEXT:    jno .LBB2_111
-; SSSE3-NEXT:  # %bb.110:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %r14d
-; SSSE3-NEXT:  .LBB2_111:
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %r15b
-; SSSE3-NEXT:    movl %r15d, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %r15b
-; SSSE3-NEXT:    movb %r12b, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSSE3-NEXT:    movb %r10b, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSSE3-NEXT:    movb %dl, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSSE3-NEXT:    jno .LBB2_113
-; SSSE3-NEXT:  # %bb.112:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %r15d
-; SSSE3-NEXT:  .LBB2_113:
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %r12b
-; SSSE3-NEXT:    movl %r12d, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %r12b
-; SSSE3-NEXT:    movl %r8d, %edx
-; SSSE3-NEXT:    jno .LBB2_115
-; SSSE3-NEXT:  # %bb.114:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %r12d
-; SSSE3-NEXT:  .LBB2_115:
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %r13b
-; SSSE3-NEXT:    movl %r13d, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %r13b
-; SSSE3-NEXT:    movb %sil, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSSE3-NEXT:    movb %dil, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSSE3-NEXT:    movb %r11b, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSSE3-NEXT:    jno .LBB2_117
-; SSSE3-NEXT:  # %bb.116:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %r13d
-; SSSE3-NEXT:  .LBB2_117:
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %sil
-; SSSE3-NEXT:    movl %esi, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %sil
-; SSSE3-NEXT:    jno .LBB2_119
-; SSSE3-NEXT:  # %bb.118:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %esi
-; SSSE3-NEXT:  .LBB2_119:
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %dil
-; SSSE3-NEXT:    movl %edi, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %dil
-; SSSE3-NEXT:    jno .LBB2_121
-; SSSE3-NEXT:  # %bb.120:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %edi
-; SSSE3-NEXT:  .LBB2_121:
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %r8b
-; SSSE3-NEXT:    movl %r8d, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %r8b
-; SSSE3-NEXT:    jno .LBB2_123
-; SSSE3-NEXT:  # %bb.122:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %r8d
-; SSSE3-NEXT:  .LBB2_123:
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %r10b
-; SSSE3-NEXT:    movl %r10d, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %r10b
-; SSSE3-NEXT:    movb {{[-0-9]+}}(%r{{[sb]}}p), %bpl # 1-byte Reload
-; SSSE3-NEXT:    jno .LBB2_125
-; SSSE3-NEXT:  # %bb.124:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %r10d
-; SSSE3-NEXT:  .LBB2_125:
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %r11b
-; SSSE3-NEXT:    movl %r11d, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %r11b
-; SSSE3-NEXT:    jno .LBB2_127
-; SSSE3-NEXT:  # %bb.126:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %r11d
-; SSSE3-NEXT:  .LBB2_127:
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %bl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movl %ecx, %eax
-; SSSE3-NEXT:    addb %bl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %bl, %cl
-; SSSE3-NEXT:    movb %dl, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSSE3-NEXT:    movb %r9b, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSSE3-NEXT:    jno .LBB2_129
-; SSSE3-NEXT:  # %bb.128:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %ecx
-; SSSE3-NEXT:  .LBB2_129:
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %dl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %bl
-; SSSE3-NEXT:    movl %ebx, %eax
-; SSSE3-NEXT:    addb %dl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %dl, %bl
-; SSSE3-NEXT:    jno .LBB2_131
-; SSSE3-NEXT:  # %bb.130:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %ebx
-; SSSE3-NEXT:  .LBB2_131:
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %dl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %r9b
-; SSSE3-NEXT:    movl %r9d, %eax
-; SSSE3-NEXT:    addb %dl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %dl, %r9b
-; SSSE3-NEXT:    jno .LBB2_133
-; SSSE3-NEXT:  # %bb.132:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %r9d
-; SSSE3-NEXT:  .LBB2_133:
-; SSSE3-NEXT:    movl %ebp, %eax
-; SSSE3-NEXT:    movb {{[-0-9]+}}(%r{{[sb]}}p), %dl # 1-byte Reload
-; SSSE3-NEXT:    movzbl %r9b, %ebp
-; SSSE3-NEXT:    movl %ebp, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSSE3-NEXT:    movzbl %bl, %ebp
-; SSSE3-NEXT:    movl %ebp, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSSE3-NEXT:    movzbl %cl, %ecx
-; SSSE3-NEXT:    movl %ecx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSSE3-NEXT:    movzbl %r11b, %ecx
-; SSSE3-NEXT:    movl %ecx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSSE3-NEXT:    movzbl %r10b, %ecx
-; SSSE3-NEXT:    movl %ecx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSSE3-NEXT:    movzbl %r8b, %ecx
-; SSSE3-NEXT:    movl %ecx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSSE3-NEXT:    movzbl %dil, %ecx
-; SSSE3-NEXT:    movl %ecx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSSE3-NEXT:    movzbl %sil, %ecx
-; SSSE3-NEXT:    movl %ecx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSSE3-NEXT:    movzbl %r13b, %ecx
-; SSSE3-NEXT:    movl %ecx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSSE3-NEXT:    movzbl %r12b, %ecx
-; SSSE3-NEXT:    movl %ecx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSSE3-NEXT:    movzbl %r15b, %ecx
-; SSSE3-NEXT:    movl %ecx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSSE3-NEXT:    movzbl %r14b, %ecx
-; SSSE3-NEXT:    movl %ecx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSSE3-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %ecx # 1-byte Folded Reload
-; SSSE3-NEXT:    movl %ecx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSSE3-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %ecx # 1-byte Folded Reload
-; SSSE3-NEXT:    movl %ecx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSSE3-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %ecx # 1-byte Folded Reload
-; SSSE3-NEXT:    movl %ecx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSSE3-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %ecx # 1-byte Folded Reload
-; SSSE3-NEXT:    movl %ecx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSSE3-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %ecx # 1-byte Folded Reload
-; SSSE3-NEXT:    movl %ecx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSSE3-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %ecx # 1-byte Folded Reload
-; SSSE3-NEXT:    movl %ecx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSSE3-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %ecx # 1-byte Folded Reload
-; SSSE3-NEXT:    movl %ecx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSSE3-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %ecx # 1-byte Folded Reload
-; SSSE3-NEXT:    movl %ecx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSSE3-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %ecx # 1-byte Folded Reload
-; SSSE3-NEXT:    movl %ecx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSSE3-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %ecx # 1-byte Folded Reload
-; SSSE3-NEXT:    movl %ecx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSSE3-NEXT:    movzbl %dl, %ecx
-; SSSE3-NEXT:    movl %ecx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSSE3-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %ecx # 1-byte Folded Reload
-; SSSE3-NEXT:    movl %ecx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSSE3-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %ecx # 1-byte Folded Reload
-; SSSE3-NEXT:    movl %ecx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSSE3-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %ecx # 1-byte Folded Reload
-; SSSE3-NEXT:    movl %ecx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSSE3-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %ecx # 1-byte Folded Reload
-; SSSE3-NEXT:    movl %ecx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSSE3-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %ecx # 1-byte Folded Reload
-; SSSE3-NEXT:    movl %ecx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSSE3-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %ecx # 1-byte Folded Reload
-; SSSE3-NEXT:    movl %ecx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSSE3-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %ecx # 1-byte Folded Reload
-; SSSE3-NEXT:    movl %ecx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSSE3-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %ecx # 1-byte Folded Reload
-; SSSE3-NEXT:    movl %ecx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSSE3-NEXT:    movzbl %al, %eax
-; SSSE3-NEXT:    movl %eax, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSSE3-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %r13d # 1-byte Folded Reload
-; SSSE3-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %r12d # 1-byte Folded Reload
-; SSSE3-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %r15d # 1-byte Folded Reload
-; SSSE3-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %r14d # 1-byte Folded Reload
-; SSSE3-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %ebp # 1-byte Folded Reload
-; SSSE3-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %ebx # 1-byte Folded Reload
-; SSSE3-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %r11d # 1-byte Folded Reload
-; SSSE3-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %r9d # 1-byte Folded Reload
-; SSSE3-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; SSSE3-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %ecx # 1-byte Folded Reload
-; SSSE3-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %edx # 1-byte Folded Reload
-; SSSE3-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %edi # 1-byte Folded Reload
-; SSSE3-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %esi # 1-byte Folded Reload
-; SSSE3-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %r8d # 1-byte Folded Reload
-; SSSE3-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %r10d # 1-byte Folded Reload
-; SSSE3-NEXT:    movd {{[-0-9]+}}(%r{{[sb]}}p), %xmm14 # 4-byte Folded Reload
-; SSSE3-NEXT:    # xmm14 = mem[0],zero,zero,zero
-; SSSE3-NEXT:    movd {{[-0-9]+}}(%r{{[sb]}}p), %xmm2 # 4-byte Folded Reload
-; SSSE3-NEXT:    # xmm2 = mem[0],zero,zero,zero
-; SSSE3-NEXT:    movd {{[-0-9]+}}(%r{{[sb]}}p), %xmm12 # 4-byte Folded Reload
-; SSSE3-NEXT:    # xmm12 = mem[0],zero,zero,zero
-; SSSE3-NEXT:    movd {{[-0-9]+}}(%r{{[sb]}}p), %xmm4 # 4-byte Folded Reload
-; SSSE3-NEXT:    # xmm4 = mem[0],zero,zero,zero
-; SSSE3-NEXT:    movd {{[-0-9]+}}(%r{{[sb]}}p), %xmm8 # 4-byte Folded Reload
-; SSSE3-NEXT:    # xmm8 = mem[0],zero,zero,zero
-; SSSE3-NEXT:    movd {{[-0-9]+}}(%r{{[sb]}}p), %xmm6 # 4-byte Folded Reload
-; SSSE3-NEXT:    # xmm6 = mem[0],zero,zero,zero
-; SSSE3-NEXT:    movd {{[-0-9]+}}(%r{{[sb]}}p), %xmm10 # 4-byte Folded Reload
-; SSSE3-NEXT:    # xmm10 = mem[0],zero,zero,zero
-; SSSE3-NEXT:    movd {{[-0-9]+}}(%r{{[sb]}}p), %xmm5 # 4-byte Folded Reload
-; SSSE3-NEXT:    # xmm5 = mem[0],zero,zero,zero
-; SSSE3-NEXT:    movd {{[-0-9]+}}(%r{{[sb]}}p), %xmm13 # 4-byte Folded Reload
-; SSSE3-NEXT:    # xmm13 = mem[0],zero,zero,zero
-; SSSE3-NEXT:    movd {{[-0-9]+}}(%r{{[sb]}}p), %xmm7 # 4-byte Folded Reload
-; SSSE3-NEXT:    # xmm7 = mem[0],zero,zero,zero
-; SSSE3-NEXT:    movd {{[-0-9]+}}(%r{{[sb]}}p), %xmm11 # 4-byte Folded Reload
-; SSSE3-NEXT:    # xmm11 = mem[0],zero,zero,zero
-; SSSE3-NEXT:    movd {{[-0-9]+}}(%r{{[sb]}}p), %xmm3 # 4-byte Folded Reload
-; SSSE3-NEXT:    # xmm3 = mem[0],zero,zero,zero
-; SSSE3-NEXT:    movss {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 4-byte Reload
-; SSSE3-NEXT:    # xmm0 = mem[0],zero,zero,zero
-; SSSE3-NEXT:    movaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; SSSE3-NEXT:    movd {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 4-byte Folded Reload
-; SSSE3-NEXT:    # xmm1 = mem[0],zero,zero,zero
-; SSSE3-NEXT:    movd {{[-0-9]+}}(%r{{[sb]}}p), %xmm15 # 4-byte Folded Reload
-; SSSE3-NEXT:    # xmm15 = mem[0],zero,zero,zero
-; SSSE3-NEXT:    movd {{[-0-9]+}}(%r{{[sb]}}p), %xmm9 # 4-byte Folded Reload
-; SSSE3-NEXT:    # xmm9 = mem[0],zero,zero,zero
-; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm2 = xmm2[0],xmm14[0],xmm2[1],xmm14[1],xmm2[2],xmm14[2],xmm2[3],xmm14[3],xmm2[4],xmm14[4],xmm2[5],xmm14[5],xmm2[6],xmm14[6],xmm2[7],xmm14[7]
-; SSSE3-NEXT:    movss {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 4-byte Reload
-; SSSE3-NEXT:    # xmm0 = mem[0],zero,zero,zero
-; SSSE3-NEXT:    movaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm4 = xmm4[0],xmm12[0],xmm4[1],xmm12[1],xmm4[2],xmm12[2],xmm4[3],xmm12[3],xmm4[4],xmm12[4],xmm4[5],xmm12[5],xmm4[6],xmm12[6],xmm4[7],xmm12[7]
-; SSSE3-NEXT:    movd {{[-0-9]+}}(%r{{[sb]}}p), %xmm12 # 4-byte Folded Reload
-; SSSE3-NEXT:    # xmm12 = mem[0],zero,zero,zero
-; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm4 = xmm4[0],xmm2[0],xmm4[1],xmm2[1],xmm4[2],xmm2[2],xmm4[3],xmm2[3]
-; SSSE3-NEXT:    movss {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 4-byte Reload
-; SSSE3-NEXT:    # xmm0 = mem[0],zero,zero,zero
-; SSSE3-NEXT:    movaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm6 = xmm6[0],xmm8[0],xmm6[1],xmm8[1],xmm6[2],xmm8[2],xmm6[3],xmm8[3],xmm6[4],xmm8[4],xmm6[5],xmm8[5],xmm6[6],xmm8[6],xmm6[7],xmm8[7]
-; SSSE3-NEXT:    movd {{[-0-9]+}}(%r{{[sb]}}p), %xmm8 # 4-byte Folded Reload
-; SSSE3-NEXT:    # xmm8 = mem[0],zero,zero,zero
-; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm5 = xmm5[0],xmm10[0],xmm5[1],xmm10[1],xmm5[2],xmm10[2],xmm5[3],xmm10[3],xmm5[4],xmm10[4],xmm5[5],xmm10[5],xmm5[6],xmm10[6],xmm5[7],xmm10[7]
-; SSSE3-NEXT:    movss {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 4-byte Reload
-; SSSE3-NEXT:    # xmm0 = mem[0],zero,zero,zero
-; SSSE3-NEXT:    movaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm5 = xmm5[0],xmm6[0],xmm5[1],xmm6[1],xmm5[2],xmm6[2],xmm5[3],xmm6[3]
-; SSSE3-NEXT:    movd {{[-0-9]+}}(%r{{[sb]}}p), %xmm6 # 4-byte Folded Reload
-; SSSE3-NEXT:    # xmm6 = mem[0],zero,zero,zero
-; SSSE3-NEXT:    punpckldq {{.*#+}} xmm5 = xmm5[0],xmm4[0],xmm5[1],xmm4[1]
-; SSSE3-NEXT:    movd {{[-0-9]+}}(%r{{[sb]}}p), %xmm14 # 4-byte Folded Reload
-; SSSE3-NEXT:    # xmm14 = mem[0],zero,zero,zero
-; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm7 = xmm7[0],xmm13[0],xmm7[1],xmm13[1],xmm7[2],xmm13[2],xmm7[3],xmm13[3],xmm7[4],xmm13[4],xmm7[5],xmm13[5],xmm7[6],xmm13[6],xmm7[7],xmm13[7]
-; SSSE3-NEXT:    movd {{[-0-9]+}}(%r{{[sb]}}p), %xmm2 # 4-byte Folded Reload
-; SSSE3-NEXT:    # xmm2 = mem[0],zero,zero,zero
-; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm3 = xmm3[0],xmm11[0],xmm3[1],xmm11[1],xmm3[2],xmm11[2],xmm3[3],xmm11[3],xmm3[4],xmm11[4],xmm3[5],xmm11[5],xmm3[6],xmm11[6],xmm3[7],xmm11[7]
-; SSSE3-NEXT:    movd {{[-0-9]+}}(%r{{[sb]}}p), %xmm11 # 4-byte Folded Reload
-; SSSE3-NEXT:    # xmm11 = mem[0],zero,zero,zero
-; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm3 = xmm3[0],xmm7[0],xmm3[1],xmm7[1],xmm3[2],xmm7[2],xmm3[3],xmm7[3]
-; SSSE3-NEXT:    movd {{[-0-9]+}}(%r{{[sb]}}p), %xmm10 # 4-byte Folded Reload
-; SSSE3-NEXT:    # xmm10 = mem[0],zero,zero,zero
-; SSSE3-NEXT:    punpcklbw {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Folded Reload
-; SSSE3-NEXT:    # xmm1 = xmm1[0],mem[0],xmm1[1],mem[1],xmm1[2],mem[2],xmm1[3],mem[3],xmm1[4],mem[4],xmm1[5],mem[5],xmm1[6],mem[6],xmm1[7],mem[7]
-; SSSE3-NEXT:    movd {{[-0-9]+}}(%r{{[sb]}}p), %xmm13 # 4-byte Folded Reload
-; SSSE3-NEXT:    # xmm13 = mem[0],zero,zero,zero
-; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm9 = xmm9[0],xmm15[0],xmm9[1],xmm15[1],xmm9[2],xmm15[2],xmm9[3],xmm15[3],xmm9[4],xmm15[4],xmm9[5],xmm15[5],xmm9[6],xmm15[6],xmm9[7],xmm15[7]
-; SSSE3-NEXT:    movd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 4-byte Folded Reload
-; SSSE3-NEXT:    # xmm0 = mem[0],zero,zero,zero
-; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm9 = xmm9[0],xmm1[0],xmm9[1],xmm1[1],xmm9[2],xmm1[2],xmm9[3],xmm1[3]
-; SSSE3-NEXT:    movd {{[-0-9]+}}(%r{{[sb]}}p), %xmm15 # 4-byte Folded Reload
-; SSSE3-NEXT:    # xmm15 = mem[0],zero,zero,zero
-; SSSE3-NEXT:    punpckldq {{.*#+}} xmm9 = xmm9[0],xmm3[0],xmm9[1],xmm3[1]
-; SSSE3-NEXT:    movd {{[-0-9]+}}(%r{{[sb]}}p), %xmm7 # 4-byte Folded Reload
-; SSSE3-NEXT:    # xmm7 = mem[0],zero,zero,zero
-; SSSE3-NEXT:    punpcklqdq {{.*#+}} xmm9 = xmm9[0],xmm5[0]
-; SSSE3-NEXT:    movd {{[-0-9]+}}(%r{{[sb]}}p), %xmm5 # 4-byte Folded Reload
-; SSSE3-NEXT:    # xmm5 = mem[0],zero,zero,zero
-; SSSE3-NEXT:    punpcklbw {{[-0-9]+}}(%r{{[sb]}}p), %xmm12 # 16-byte Folded Reload
-; SSSE3-NEXT:    # xmm12 = xmm12[0],mem[0],xmm12[1],mem[1],xmm12[2],mem[2],xmm12[3],mem[3],xmm12[4],mem[4],xmm12[5],mem[5],xmm12[6],mem[6],xmm12[7],mem[7]
-; SSSE3-NEXT:    movd {{[-0-9]+}}(%r{{[sb]}}p), %xmm4 # 4-byte Folded Reload
-; SSSE3-NEXT:    # xmm4 = mem[0],zero,zero,zero
-; SSSE3-NEXT:    punpcklbw {{[-0-9]+}}(%r{{[sb]}}p), %xmm8 # 16-byte Folded Reload
-; SSSE3-NEXT:    # xmm8 = xmm8[0],mem[0],xmm8[1],mem[1],xmm8[2],mem[2],xmm8[3],mem[3],xmm8[4],mem[4],xmm8[5],mem[5],xmm8[6],mem[6],xmm8[7],mem[7]
-; SSSE3-NEXT:    movd %r13d, %xmm1
-; SSSE3-NEXT:    movdqa %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm8 = xmm8[0],xmm12[0],xmm8[1],xmm12[1],xmm8[2],xmm12[2],xmm8[3],xmm12[3]
-; SSSE3-NEXT:    movd %r12d, %xmm1
-; SSSE3-NEXT:    punpcklbw {{[-0-9]+}}(%r{{[sb]}}p), %xmm6 # 16-byte Folded Reload
-; SSSE3-NEXT:    # xmm6 = xmm6[0],mem[0],xmm6[1],mem[1],xmm6[2],mem[2],xmm6[3],mem[3],xmm6[4],mem[4],xmm6[5],mem[5],xmm6[6],mem[6],xmm6[7],mem[7]
-; SSSE3-NEXT:    movd %r15d, %xmm3
-; SSSE3-NEXT:    movdqa %xmm3, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm2 = xmm2[0],xmm14[0],xmm2[1],xmm14[1],xmm2[2],xmm14[2],xmm2[3],xmm14[3],xmm2[4],xmm14[4],xmm2[5],xmm14[5],xmm2[6],xmm14[6],xmm2[7],xmm14[7]
-; SSSE3-NEXT:    movd %r14d, %xmm14
-; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm2 = xmm2[0],xmm6[0],xmm2[1],xmm6[1],xmm2[2],xmm6[2],xmm2[3],xmm6[3]
-; SSSE3-NEXT:    movd %ebp, %xmm3
-; SSSE3-NEXT:    movdqa %xmm3, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; SSSE3-NEXT:    punpckldq {{.*#+}} xmm2 = xmm2[0],xmm8[0],xmm2[1],xmm8[1]
-; SSSE3-NEXT:    movd %ebx, %xmm12
-; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm10 = xmm10[0],xmm11[0],xmm10[1],xmm11[1],xmm10[2],xmm11[2],xmm10[3],xmm11[3],xmm10[4],xmm11[4],xmm10[5],xmm11[5],xmm10[6],xmm11[6],xmm10[7],xmm11[7]
-; SSSE3-NEXT:    movd %r11d, %xmm8
-; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm13[0],xmm0[1],xmm13[1],xmm0[2],xmm13[2],xmm0[3],xmm13[3],xmm0[4],xmm13[4],xmm0[5],xmm13[5],xmm0[6],xmm13[6],xmm0[7],xmm13[7]
-; SSSE3-NEXT:    movd %r9d, %xmm3
-; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm10[0],xmm0[1],xmm10[1],xmm0[2],xmm10[2],xmm0[3],xmm10[3]
-; SSSE3-NEXT:    movd %eax, %xmm11
-; SSSE3-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %r15d # 1-byte Folded Reload
-; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm7 = xmm7[0],xmm15[0],xmm7[1],xmm15[1],xmm7[2],xmm15[2],xmm7[3],xmm15[3],xmm7[4],xmm15[4],xmm7[5],xmm15[5],xmm7[6],xmm15[6],xmm7[7],xmm15[7]
-; SSSE3-NEXT:    movd %ecx, %xmm6
-; SSSE3-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %r12d # 1-byte Folded Reload
-; SSSE3-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %r9d # 1-byte Folded Reload
-; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm4 = xmm4[0],xmm5[0],xmm4[1],xmm5[1],xmm4[2],xmm5[2],xmm4[3],xmm5[3],xmm4[4],xmm5[4],xmm4[5],xmm5[5],xmm4[6],xmm5[6],xmm4[7],xmm5[7]
-; SSSE3-NEXT:    movd %edx, %xmm13
-; SSSE3-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %r11d # 1-byte Folded Reload
-; SSSE3-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %r14d # 1-byte Folded Reload
-; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm4 = xmm4[0],xmm7[0],xmm4[1],xmm7[1],xmm4[2],xmm7[2],xmm4[3],xmm7[3]
-; SSSE3-NEXT:    movd %edi, %xmm5
-; SSSE3-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %edi # 1-byte Folded Reload
-; SSSE3-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %r13d # 1-byte Folded Reload
-; SSSE3-NEXT:    punpckldq {{.*#+}} xmm4 = xmm4[0],xmm0[0],xmm4[1],xmm0[1]
-; SSSE3-NEXT:    movd %esi, %xmm15
-; SSSE3-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %esi # 1-byte Folded Reload
-; SSSE3-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %edx # 1-byte Folded Reload
-; SSSE3-NEXT:    punpcklqdq {{.*#+}} xmm4 = xmm4[0],xmm2[0]
-; SSSE3-NEXT:    movd %r8d, %xmm0
-; SSSE3-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %r8d # 1-byte Folded Reload
-; SSSE3-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %ebp # 1-byte Folded Reload
-; SSSE3-NEXT:    punpcklbw {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Folded Reload
-; SSSE3-NEXT:    # xmm1 = xmm1[0],mem[0],xmm1[1],mem[1],xmm1[2],mem[2],xmm1[3],mem[3],xmm1[4],mem[4],xmm1[5],mem[5],xmm1[6],mem[6],xmm1[7],mem[7]
-; SSSE3-NEXT:    movd %r10d, %xmm10
-; SSSE3-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %ecx # 1-byte Folded Reload
-; SSSE3-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %r10d # 1-byte Folded Reload
-; SSSE3-NEXT:    punpcklbw {{[-0-9]+}}(%r{{[sb]}}p), %xmm14 # 16-byte Folded Reload
-; SSSE3-NEXT:    # xmm14 = xmm14[0],mem[0],xmm14[1],mem[1],xmm14[2],mem[2],xmm14[3],mem[3],xmm14[4],mem[4],xmm14[5],mem[5],xmm14[6],mem[6],xmm14[7],mem[7]
-; SSSE3-NEXT:    movd %r15d, %xmm2
-; SSSE3-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %ebx # 1-byte Folded Reload
-; SSSE3-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %r15d # 1-byte Folded Reload
-; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm14 = xmm14[0],xmm1[0],xmm14[1],xmm1[1],xmm14[2],xmm1[2],xmm14[3],xmm1[3]
-; SSSE3-NEXT:    movd %r12d, %xmm1
-; SSSE3-NEXT:    movdqa %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; SSSE3-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; SSSE3-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %r12d # 1-byte Folded Reload
-; SSSE3-NEXT:    punpcklbw {{[-0-9]+}}(%r{{[sb]}}p), %xmm12 # 16-byte Folded Reload
-; SSSE3-NEXT:    # xmm12 = xmm12[0],mem[0],xmm12[1],mem[1],xmm12[2],mem[2],xmm12[3],mem[3],xmm12[4],mem[4],xmm12[5],mem[5],xmm12[6],mem[6],xmm12[7],mem[7]
-; SSSE3-NEXT:    movd %r9d, %xmm1
-; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm3 = xmm3[0],xmm8[0],xmm3[1],xmm8[1],xmm3[2],xmm8[2],xmm3[3],xmm8[3],xmm3[4],xmm8[4],xmm3[5],xmm8[5],xmm3[6],xmm8[6],xmm3[7],xmm8[7]
-; SSSE3-NEXT:    movd %r11d, %xmm8
-; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm3 = xmm3[0],xmm12[0],xmm3[1],xmm12[1],xmm3[2],xmm12[2],xmm3[3],xmm12[3]
-; SSSE3-NEXT:    movd %r14d, %xmm12
-; SSSE3-NEXT:    punpckldq {{.*#+}} xmm3 = xmm3[0],xmm14[0],xmm3[1],xmm14[1]
-; SSSE3-NEXT:    movd %edi, %xmm7
-; SSSE3-NEXT:    movdqa %xmm7, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm6 = xmm6[0],xmm11[0],xmm6[1],xmm11[1],xmm6[2],xmm11[2],xmm6[3],xmm11[3],xmm6[4],xmm11[4],xmm6[5],xmm11[5],xmm6[6],xmm11[6],xmm6[7],xmm11[7]
-; SSSE3-NEXT:    movd %r13d, %xmm11
-; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm5 = xmm5[0],xmm13[0],xmm5[1],xmm13[1],xmm5[2],xmm13[2],xmm5[3],xmm13[3],xmm5[4],xmm13[4],xmm5[5],xmm13[5],xmm5[6],xmm13[6],xmm5[7],xmm13[7]
-; SSSE3-NEXT:    movd %esi, %xmm14
-; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm5 = xmm5[0],xmm6[0],xmm5[1],xmm6[1],xmm5[2],xmm6[2],xmm5[3],xmm6[3]
-; SSSE3-NEXT:    movd %edx, %xmm6
-; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm15[0],xmm0[1],xmm15[1],xmm0[2],xmm15[2],xmm0[3],xmm15[3],xmm0[4],xmm15[4],xmm0[5],xmm15[5],xmm0[6],xmm15[6],xmm0[7],xmm15[7]
-; SSSE3-NEXT:    movd %r8d, %xmm15
-; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm2 = xmm2[0],xmm10[0],xmm2[1],xmm10[1],xmm2[2],xmm10[2],xmm2[3],xmm10[3],xmm2[4],xmm10[4],xmm2[5],xmm10[5],xmm2[6],xmm10[6],xmm2[7],xmm10[7]
-; SSSE3-NEXT:    movd %ebp, %xmm7
-; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm2 = xmm2[0],xmm0[0],xmm2[1],xmm0[1],xmm2[2],xmm0[2],xmm2[3],xmm0[3]
-; SSSE3-NEXT:    movd %ecx, %xmm13
-; SSSE3-NEXT:    punpckldq {{.*#+}} xmm2 = xmm2[0],xmm5[0],xmm2[1],xmm5[1]
-; SSSE3-NEXT:    movd %r10d, %xmm5
-; SSSE3-NEXT:    punpcklqdq {{.*#+}} xmm2 = xmm2[0],xmm3[0]
-; SSSE3-NEXT:    movd %ebx, %xmm10
-; SSSE3-NEXT:    punpcklbw {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Folded Reload
-; SSSE3-NEXT:    # xmm1 = xmm1[0],mem[0],xmm1[1],mem[1],xmm1[2],mem[2],xmm1[3],mem[3],xmm1[4],mem[4],xmm1[5],mem[5],xmm1[6],mem[6],xmm1[7],mem[7]
-; SSSE3-NEXT:    movd %r15d, %xmm0
-; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm12 = xmm12[0],xmm8[0],xmm12[1],xmm8[1],xmm12[2],xmm8[2],xmm12[3],xmm8[3],xmm12[4],xmm8[4],xmm12[5],xmm8[5],xmm12[6],xmm8[6],xmm12[7],xmm8[7]
-; SSSE3-NEXT:    movd %eax, %xmm8
-; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm12 = xmm12[0],xmm1[0],xmm12[1],xmm1[1],xmm12[2],xmm1[2],xmm12[3],xmm1[3]
-; SSSE3-NEXT:    movd %r12d, %xmm3
-; SSSE3-NEXT:    punpcklbw {{[-0-9]+}}(%r{{[sb]}}p), %xmm11 # 16-byte Folded Reload
-; SSSE3-NEXT:    # xmm11 = xmm11[0],mem[0],xmm11[1],mem[1],xmm11[2],mem[2],xmm11[3],mem[3],xmm11[4],mem[4],xmm11[5],mem[5],xmm11[6],mem[6],xmm11[7],mem[7]
-; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm6 = xmm6[0],xmm14[0],xmm6[1],xmm14[1],xmm6[2],xmm14[2],xmm6[3],xmm14[3],xmm6[4],xmm14[4],xmm6[5],xmm14[5],xmm6[6],xmm14[6],xmm6[7],xmm14[7]
-; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm6 = xmm6[0],xmm11[0],xmm6[1],xmm11[1],xmm6[2],xmm11[2],xmm6[3],xmm11[3]
-; SSSE3-NEXT:    punpckldq {{.*#+}} xmm6 = xmm6[0],xmm12[0],xmm6[1],xmm12[1]
-; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm7 = xmm7[0],xmm15[0],xmm7[1],xmm15[1],xmm7[2],xmm15[2],xmm7[3],xmm15[3],xmm7[4],xmm15[4],xmm7[5],xmm15[5],xmm7[6],xmm15[6],xmm7[7],xmm15[7]
-; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm5 = xmm5[0],xmm13[0],xmm5[1],xmm13[1],xmm5[2],xmm13[2],xmm5[3],xmm13[3],xmm5[4],xmm13[4],xmm5[5],xmm13[5],xmm5[6],xmm13[6],xmm5[7],xmm13[7]
-; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm5 = xmm5[0],xmm7[0],xmm5[1],xmm7[1],xmm5[2],xmm7[2],xmm5[3],xmm7[3]
-; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm10[0],xmm0[1],xmm10[1],xmm0[2],xmm10[2],xmm0[3],xmm10[3],xmm0[4],xmm10[4],xmm0[5],xmm10[5],xmm0[6],xmm10[6],xmm0[7],xmm10[7]
-; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm3 = xmm3[0],xmm8[0],xmm3[1],xmm8[1],xmm3[2],xmm8[2],xmm3[3],xmm8[3],xmm3[4],xmm8[4],xmm3[5],xmm8[5],xmm3[6],xmm8[6],xmm3[7],xmm8[7]
-; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm3 = xmm3[0],xmm0[0],xmm3[1],xmm0[1],xmm3[2],xmm0[2],xmm3[3],xmm0[3]
-; SSSE3-NEXT:    punpckldq {{.*#+}} xmm3 = xmm3[0],xmm5[0],xmm3[1],xmm5[1]
-; SSSE3-NEXT:    punpcklqdq {{.*#+}} xmm3 = xmm3[0],xmm6[0]
-; SSSE3-NEXT:    movdqa %xmm9, %xmm0
-; SSSE3-NEXT:    movdqa %xmm4, %xmm1
-; SSSE3-NEXT:    addq $232, %rsp
-; SSSE3-NEXT:    popq %rbx
-; SSSE3-NEXT:    popq %r12
-; SSSE3-NEXT:    popq %r13
-; SSSE3-NEXT:    popq %r14
-; SSSE3-NEXT:    popq %r15
-; SSSE3-NEXT:    popq %rbp
-; SSSE3-NEXT:    retq
-;
-; SSE41-LABEL: v64i8:
-; SSE41:       # %bb.0:
-; SSE41-NEXT:    pextrb $15, %xmm5, %ecx
-; SSE41-NEXT:    pextrb $15, %xmm1, %r11d
-; SSE41-NEXT:    movl %r11d, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %r11b
-; SSE41-NEXT:    jno .LBB2_2
-; SSE41-NEXT:  # %bb.1:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %r11d
-; SSE41-NEXT:  .LBB2_2:
-; SSE41-NEXT:    pextrb $14, %xmm5, %ecx
-; SSE41-NEXT:    pextrb $14, %xmm1, %esi
-; SSE41-NEXT:    movl %esi, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %sil
-; SSE41-NEXT:    jno .LBB2_4
-; SSE41-NEXT:  # %bb.3:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %esi
-; SSE41-NEXT:  .LBB2_4:
-; SSE41-NEXT:    pextrb $13, %xmm5, %ecx
-; SSE41-NEXT:    pextrb $13, %xmm1, %edx
-; SSE41-NEXT:    movl %edx, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %dl
-; SSE41-NEXT:    jno .LBB2_6
-; SSE41-NEXT:  # %bb.5:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %edx
-; SSE41-NEXT:  .LBB2_6:
-; SSE41-NEXT:    pextrb $12, %xmm5, %ecx
-; SSE41-NEXT:    pextrb $12, %xmm1, %edi
-; SSE41-NEXT:    movl %edi, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %dil
-; SSE41-NEXT:    jno .LBB2_8
-; SSE41-NEXT:  # %bb.7:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %edi
-; SSE41-NEXT:  .LBB2_8:
-; SSE41-NEXT:    pushq %rbp
-; SSE41-NEXT:    pushq %r15
-; SSE41-NEXT:    pushq %r14
-; SSE41-NEXT:    pushq %r13
-; SSE41-NEXT:    pushq %r12
-; SSE41-NEXT:    pushq %rbx
-; SSE41-NEXT:    subq $76, %rsp
-; SSE41-NEXT:    pextrb $11, %xmm5, %ecx
-; SSE41-NEXT:    pextrb $11, %xmm1, %ebp
-; SSE41-NEXT:    movl %ebp, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %bpl
-; SSE41-NEXT:    jno .LBB2_10
-; SSE41-NEXT:  # %bb.9:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %ebp
-; SSE41-NEXT:  .LBB2_10:
-; SSE41-NEXT:    pextrb $10, %xmm5, %ecx
-; SSE41-NEXT:    pextrb $10, %xmm1, %ebx
-; SSE41-NEXT:    movl %ebx, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %bl
-; SSE41-NEXT:    movl %edx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE41-NEXT:    jno .LBB2_12
-; SSE41-NEXT:  # %bb.11:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %ebx
-; SSE41-NEXT:  .LBB2_12:
-; SSE41-NEXT:    pextrb $9, %xmm5, %ecx
-; SSE41-NEXT:    pextrb $9, %xmm1, %edx
-; SSE41-NEXT:    movl %edx, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %dl
-; SSE41-NEXT:    movl %ebp, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE41-NEXT:    movl %ebx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE41-NEXT:    jo .LBB2_13
-; SSE41-NEXT:  # %bb.14:
-; SSE41-NEXT:    movl %edx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE41-NEXT:    jmp .LBB2_15
-; SSE41-NEXT:  .LBB2_13:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    # kill: def $al killed $al def $eax
-; SSE41-NEXT:    movl %eax, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE41-NEXT:  .LBB2_15:
-; SSE41-NEXT:    pextrb $8, %xmm5, %ecx
-; SSE41-NEXT:    pextrb $8, %xmm1, %edx
-; SSE41-NEXT:    movl %edx, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %dl
-; SSE41-NEXT:    jno .LBB2_17
-; SSE41-NEXT:  # %bb.16:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %edx
-; SSE41-NEXT:  .LBB2_17:
-; SSE41-NEXT:    pextrb $7, %xmm5, %ecx
-; SSE41-NEXT:    pextrb $7, %xmm1, %ebp
-; SSE41-NEXT:    movl %ebp, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %bpl
-; SSE41-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE41-NEXT:    jno .LBB2_19
-; SSE41-NEXT:  # %bb.18:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %ebp
-; SSE41-NEXT:  .LBB2_19:
-; SSE41-NEXT:    pextrb $6, %xmm5, %ecx
-; SSE41-NEXT:    pextrb $6, %xmm1, %esi
-; SSE41-NEXT:    movl %esi, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %sil
-; SSE41-NEXT:    movl %r11d, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE41-NEXT:    jno .LBB2_21
-; SSE41-NEXT:  # %bb.20:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %esi
-; SSE41-NEXT:  .LBB2_21:
-; SSE41-NEXT:    pextrb $5, %xmm5, %ecx
-; SSE41-NEXT:    pextrb $5, %xmm1, %ebx
-; SSE41-NEXT:    movl %ebx, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %bl
-; SSE41-NEXT:    movl %edi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE41-NEXT:    movl %ebp, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE41-NEXT:    jo .LBB2_22
-; SSE41-NEXT:  # %bb.23:
-; SSE41-NEXT:    movl %ebx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE41-NEXT:    jmp .LBB2_24
-; SSE41-NEXT:  .LBB2_22:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    # kill: def $al killed $al def $eax
-; SSE41-NEXT:    movl %eax, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE41-NEXT:  .LBB2_24:
-; SSE41-NEXT:    pextrb $4, %xmm5, %ecx
-; SSE41-NEXT:    pextrb $4, %xmm1, %r13d
-; SSE41-NEXT:    movl %r13d, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %r13b
-; SSE41-NEXT:    jno .LBB2_26
-; SSE41-NEXT:  # %bb.25:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %r13d
-; SSE41-NEXT:  .LBB2_26:
-; SSE41-NEXT:    pextrb $3, %xmm5, %ecx
-; SSE41-NEXT:    pextrb $3, %xmm1, %ebx
-; SSE41-NEXT:    movl %ebx, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %bl
-; SSE41-NEXT:    jno .LBB2_28
-; SSE41-NEXT:  # %bb.27:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %ebx
-; SSE41-NEXT:  .LBB2_28:
-; SSE41-NEXT:    pextrb $2, %xmm5, %ecx
-; SSE41-NEXT:    pextrb $2, %xmm1, %ebp
-; SSE41-NEXT:    movl %ebp, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %bpl
-; SSE41-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE41-NEXT:    jno .LBB2_30
-; SSE41-NEXT:  # %bb.29:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %ebp
-; SSE41-NEXT:  .LBB2_30:
-; SSE41-NEXT:    pextrb $0, %xmm5, %ecx
-; SSE41-NEXT:    pextrb $0, %xmm1, %esi
-; SSE41-NEXT:    movl %esi, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %sil
-; SSE41-NEXT:    movl %edx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE41-NEXT:    jno .LBB2_32
-; SSE41-NEXT:  # %bb.31:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %esi
-; SSE41-NEXT:  .LBB2_32:
-; SSE41-NEXT:    pextrb $1, %xmm5, %ecx
-; SSE41-NEXT:    pextrb $1, %xmm1, %edx
-; SSE41-NEXT:    movl %edx, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %dl
-; SSE41-NEXT:    movl %ebx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE41-NEXT:    jno .LBB2_34
-; SSE41-NEXT:  # %bb.33:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %edx
-; SSE41-NEXT:  .LBB2_34:
-; SSE41-NEXT:    pextrb $15, %xmm6, %ecx
-; SSE41-NEXT:    pextrb $15, %xmm2, %r9d
-; SSE41-NEXT:    movl %r9d, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %r9b
-; SSE41-NEXT:    movl %ebp, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE41-NEXT:    jno .LBB2_36
-; SSE41-NEXT:  # %bb.35:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %r9d
-; SSE41-NEXT:  .LBB2_36:
-; SSE41-NEXT:    pextrb $14, %xmm6, %ecx
-; SSE41-NEXT:    pextrb $14, %xmm2, %ebp
-; SSE41-NEXT:    movl %ebp, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %bpl
-; SSE41-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE41-NEXT:    jno .LBB2_38
-; SSE41-NEXT:  # %bb.37:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %ebp
-; SSE41-NEXT:  .LBB2_38:
-; SSE41-NEXT:    pextrb $13, %xmm6, %ecx
-; SSE41-NEXT:    pextrb $13, %xmm2, %r14d
-; SSE41-NEXT:    movl %r14d, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %r14b
-; SSE41-NEXT:    movl %edx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE41-NEXT:    jno .LBB2_40
-; SSE41-NEXT:  # %bb.39:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %r14d
-; SSE41-NEXT:  .LBB2_40:
-; SSE41-NEXT:    pextrb $12, %xmm6, %ecx
-; SSE41-NEXT:    pextrb $12, %xmm2, %r10d
-; SSE41-NEXT:    movl %r10d, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %r10b
-; SSE41-NEXT:    jno .LBB2_42
-; SSE41-NEXT:  # %bb.41:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %r10d
-; SSE41-NEXT:  .LBB2_42:
-; SSE41-NEXT:    pextrb $11, %xmm6, %ecx
-; SSE41-NEXT:    pextrb $11, %xmm2, %ebx
-; SSE41-NEXT:    movl %ebx, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %bl
-; SSE41-NEXT:    movl %ebp, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE41-NEXT:    jno .LBB2_44
-; SSE41-NEXT:  # %bb.43:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %ebx
-; SSE41-NEXT:  .LBB2_44:
-; SSE41-NEXT:    pextrb $10, %xmm6, %ecx
-; SSE41-NEXT:    pextrb $10, %xmm2, %ebp
-; SSE41-NEXT:    movl %ebp, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %bpl
-; SSE41-NEXT:    jno .LBB2_46
-; SSE41-NEXT:  # %bb.45:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %ebp
-; SSE41-NEXT:  .LBB2_46:
-; SSE41-NEXT:    pextrb $9, %xmm6, %ecx
-; SSE41-NEXT:    pextrb $9, %xmm2, %edi
-; SSE41-NEXT:    movl %edi, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %dil
-; SSE41-NEXT:    jno .LBB2_48
-; SSE41-NEXT:  # %bb.47:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %edi
-; SSE41-NEXT:  .LBB2_48:
-; SSE41-NEXT:    pextrb $8, %xmm6, %ecx
-; SSE41-NEXT:    pextrb $8, %xmm2, %edx
-; SSE41-NEXT:    movl %edx, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %dl
-; SSE41-NEXT:    jno .LBB2_50
-; SSE41-NEXT:  # %bb.49:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %edx
-; SSE41-NEXT:  .LBB2_50:
-; SSE41-NEXT:    pextrb $7, %xmm6, %ecx
-; SSE41-NEXT:    pextrb $7, %xmm2, %esi
-; SSE41-NEXT:    movl %esi, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %sil
-; SSE41-NEXT:    movl %r13d, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE41-NEXT:    jno .LBB2_52
-; SSE41-NEXT:  # %bb.51:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %esi
-; SSE41-NEXT:  .LBB2_52:
-; SSE41-NEXT:    pextrb $6, %xmm6, %ecx
-; SSE41-NEXT:    pextrb $6, %xmm2, %r8d
-; SSE41-NEXT:    movl %r8d, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %r8b
-; SSE41-NEXT:    jno .LBB2_54
-; SSE41-NEXT:  # %bb.53:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %r8d
-; SSE41-NEXT:  .LBB2_54:
-; SSE41-NEXT:    pextrb $5, %xmm6, %ecx
-; SSE41-NEXT:    pextrb $5, %xmm2, %r11d
-; SSE41-NEXT:    movl %r11d, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %r11b
-; SSE41-NEXT:    movl %ebx, (%rsp) # 4-byte Spill
-; SSE41-NEXT:    movl %ebp, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE41-NEXT:    movl %edi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE41-NEXT:    movl %edx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE41-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE41-NEXT:    jno .LBB2_56
-; SSE41-NEXT:  # %bb.55:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %r11d
-; SSE41-NEXT:  .LBB2_56:
-; SSE41-NEXT:    pextrb $4, %xmm6, %ecx
-; SSE41-NEXT:    pextrb $4, %xmm2, %edx
-; SSE41-NEXT:    movl %edx, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %dl
-; SSE41-NEXT:    movl %r14d, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE41-NEXT:    jno .LBB2_58
-; SSE41-NEXT:  # %bb.57:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %edx
-; SSE41-NEXT:  .LBB2_58:
-; SSE41-NEXT:    pextrb $3, %xmm6, %ecx
-; SSE41-NEXT:    pextrb $3, %xmm2, %esi
-; SSE41-NEXT:    movl %esi, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %sil
-; SSE41-NEXT:    movl %r10d, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE41-NEXT:    movl %edx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE41-NEXT:    jno .LBB2_60
-; SSE41-NEXT:  # %bb.59:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %esi
-; SSE41-NEXT:  .LBB2_60:
-; SSE41-NEXT:    pextrb $2, %xmm6, %ecx
-; SSE41-NEXT:    pextrb $2, %xmm2, %edx
-; SSE41-NEXT:    movl %edx, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %dl
-; SSE41-NEXT:    movl %r9d, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE41-NEXT:    jno .LBB2_62
-; SSE41-NEXT:  # %bb.61:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %edx
-; SSE41-NEXT:  .LBB2_62:
-; SSE41-NEXT:    pextrb $0, %xmm6, %ecx
-; SSE41-NEXT:    pextrb $0, %xmm2, %edi
-; SSE41-NEXT:    movl %edi, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %dil
-; SSE41-NEXT:    jno .LBB2_64
-; SSE41-NEXT:  # %bb.63:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %edi
-; SSE41-NEXT:  .LBB2_64:
-; SSE41-NEXT:    pextrb $1, %xmm6, %ecx
-; SSE41-NEXT:    pextrb $1, %xmm2, %ebx
-; SSE41-NEXT:    movl %ebx, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %bl
-; SSE41-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE41-NEXT:    jno .LBB2_66
-; SSE41-NEXT:  # %bb.65:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %ebx
-; SSE41-NEXT:  .LBB2_66:
-; SSE41-NEXT:    pextrb $15, %xmm7, %ecx
-; SSE41-NEXT:    pextrb $15, %xmm3, %esi
-; SSE41-NEXT:    movl %esi, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %sil
-; SSE41-NEXT:    movl %edx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE41-NEXT:    jno .LBB2_68
-; SSE41-NEXT:  # %bb.67:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %esi
-; SSE41-NEXT:  .LBB2_68:
-; SSE41-NEXT:    pextrb $14, %xmm7, %ecx
-; SSE41-NEXT:    pextrb $14, %xmm3, %edx
-; SSE41-NEXT:    movl %edx, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %dl
-; SSE41-NEXT:    movl %edi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE41-NEXT:    jno .LBB2_70
-; SSE41-NEXT:  # %bb.69:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %edx
-; SSE41-NEXT:  .LBB2_70:
-; SSE41-NEXT:    pextrb $13, %xmm7, %ecx
-; SSE41-NEXT:    pextrb $13, %xmm3, %edi
-; SSE41-NEXT:    movl %edi, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %dil
-; SSE41-NEXT:    movl %ebx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE41-NEXT:    jno .LBB2_72
-; SSE41-NEXT:  # %bb.71:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %edi
-; SSE41-NEXT:  .LBB2_72:
-; SSE41-NEXT:    pextrb $12, %xmm7, %ecx
-; SSE41-NEXT:    pextrb $12, %xmm3, %r15d
-; SSE41-NEXT:    movl %r15d, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %r15b
-; SSE41-NEXT:    movl %r15d, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE41-NEXT:    movl %edx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE41-NEXT:    jno .LBB2_74
-; SSE41-NEXT:  # %bb.73:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    # kill: def $al killed $al def $eax
-; SSE41-NEXT:    movl %eax, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE41-NEXT:  .LBB2_74:
-; SSE41-NEXT:    pextrb $11, %xmm7, %ecx
-; SSE41-NEXT:    pextrb $11, %xmm3, %edx
-; SSE41-NEXT:    movl %edx, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %dl
-; SSE41-NEXT:    jno .LBB2_76
-; SSE41-NEXT:  # %bb.75:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %edx
-; SSE41-NEXT:  .LBB2_76:
-; SSE41-NEXT:    pextrb $10, %xmm7, %ecx
-; SSE41-NEXT:    pextrb $10, %xmm3, %ebp
-; SSE41-NEXT:    movl %ebp, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %bpl
-; SSE41-NEXT:    movl %r11d, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE41-NEXT:    jno .LBB2_78
-; SSE41-NEXT:  # %bb.77:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %ebp
-; SSE41-NEXT:  .LBB2_78:
-; SSE41-NEXT:    pextrb $9, %xmm7, %ecx
-; SSE41-NEXT:    pextrb $9, %xmm3, %r11d
-; SSE41-NEXT:    movl %r11d, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %r11b
-; SSE41-NEXT:    movl %r8d, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE41-NEXT:    jno .LBB2_80
-; SSE41-NEXT:  # %bb.79:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %r11d
-; SSE41-NEXT:  .LBB2_80:
-; SSE41-NEXT:    pextrb $8, %xmm7, %ecx
-; SSE41-NEXT:    pextrb $8, %xmm3, %r8d
-; SSE41-NEXT:    movl %r8d, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %r8b
-; SSE41-NEXT:    movl %ebp, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE41-NEXT:    jno .LBB2_82
-; SSE41-NEXT:  # %bb.81:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %r8d
-; SSE41-NEXT:  .LBB2_82:
-; SSE41-NEXT:    pextrb $7, %xmm7, %ecx
-; SSE41-NEXT:    pextrb $7, %xmm3, %r9d
-; SSE41-NEXT:    movl %r9d, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %r9b
-; SSE41-NEXT:    jno .LBB2_84
-; SSE41-NEXT:  # %bb.83:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %r9d
-; SSE41-NEXT:  .LBB2_84:
-; SSE41-NEXT:    pextrb $6, %xmm7, %ecx
-; SSE41-NEXT:    pextrb $6, %xmm3, %r10d
-; SSE41-NEXT:    movl %r10d, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %r10b
-; SSE41-NEXT:    jno .LBB2_86
-; SSE41-NEXT:  # %bb.85:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %r10d
-; SSE41-NEXT:  .LBB2_86:
-; SSE41-NEXT:    pextrb $5, %xmm7, %ecx
-; SSE41-NEXT:    pextrb $5, %xmm3, %r14d
-; SSE41-NEXT:    movl %r14d, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %r14b
-; SSE41-NEXT:    jno .LBB2_88
-; SSE41-NEXT:  # %bb.87:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %r14d
-; SSE41-NEXT:  .LBB2_88:
-; SSE41-NEXT:    pextrb $4, %xmm7, %ecx
-; SSE41-NEXT:    pextrb $4, %xmm3, %r12d
-; SSE41-NEXT:    movl %r12d, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %r12b
-; SSE41-NEXT:    movl %r12d, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE41-NEXT:    jno .LBB2_90
-; SSE41-NEXT:  # %bb.89:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    # kill: def $al killed $al def $eax
-; SSE41-NEXT:    movl %eax, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE41-NEXT:  .LBB2_90:
-; SSE41-NEXT:    pextrb $3, %xmm7, %ecx
-; SSE41-NEXT:    pextrb $3, %xmm3, %r13d
-; SSE41-NEXT:    movl %r13d, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %r13b
-; SSE41-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE41-NEXT:    jno .LBB2_92
-; SSE41-NEXT:  # %bb.91:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %r13d
-; SSE41-NEXT:  .LBB2_92:
-; SSE41-NEXT:    pextrb $2, %xmm7, %ecx
-; SSE41-NEXT:    pextrb $2, %xmm3, %esi
-; SSE41-NEXT:    movl %esi, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %sil
-; SSE41-NEXT:    movl %edx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE41-NEXT:    jno .LBB2_94
-; SSE41-NEXT:  # %bb.93:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %esi
-; SSE41-NEXT:  .LBB2_94:
-; SSE41-NEXT:    pextrb $0, %xmm7, %ecx
-; SSE41-NEXT:    pextrb $0, %xmm3, %edx
-; SSE41-NEXT:    movl %edx, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    movl %edi, %r15d
-; SSE41-NEXT:    addb %cl, %dl
-; SSE41-NEXT:    jno .LBB2_96
-; SSE41-NEXT:  # %bb.95:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %edx
-; SSE41-NEXT:  .LBB2_96:
-; SSE41-NEXT:    pextrb $1, %xmm7, %ecx
-; SSE41-NEXT:    pextrb $1, %xmm3, %edi
-; SSE41-NEXT:    movl %edi, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %dil
-; SSE41-NEXT:    jno .LBB2_98
-; SSE41-NEXT:  # %bb.97:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %edi
-; SSE41-NEXT:  .LBB2_98:
-; SSE41-NEXT:    pextrb $15, %xmm4, %ecx
-; SSE41-NEXT:    pextrb $15, %xmm0, %r12d
-; SSE41-NEXT:    movl %r12d, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %r12b
-; SSE41-NEXT:    movl %r12d, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE41-NEXT:    jno .LBB2_100
-; SSE41-NEXT:  # %bb.99:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    # kill: def $al killed $al def $eax
-; SSE41-NEXT:    movl %eax, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE41-NEXT:  .LBB2_100:
-; SSE41-NEXT:    pextrb $14, %xmm4, %ecx
-; SSE41-NEXT:    pextrb $14, %xmm0, %ebp
-; SSE41-NEXT:    movl %ebp, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %bpl
-; SSE41-NEXT:    jno .LBB2_102
-; SSE41-NEXT:  # %bb.101:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %ebp
-; SSE41-NEXT:  .LBB2_102:
-; SSE41-NEXT:    pextrb $13, %xmm4, %ecx
-; SSE41-NEXT:    pextrb $13, %xmm0, %ebx
-; SSE41-NEXT:    movl %ebx, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %bl
-; SSE41-NEXT:    movl %r15d, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE41-NEXT:    movl %r13d, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE41-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE41-NEXT:    movl %edx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE41-NEXT:    movl %edi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE41-NEXT:    movl %ebp, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE41-NEXT:    jo .LBB2_103
-; SSE41-NEXT:  # %bb.104:
-; SSE41-NEXT:    movl %ebx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE41-NEXT:    jmp .LBB2_105
-; SSE41-NEXT:  .LBB2_103:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    # kill: def $al killed $al def $eax
-; SSE41-NEXT:    movl %eax, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE41-NEXT:  .LBB2_105:
-; SSE41-NEXT:    pextrb $12, %xmm4, %ecx
-; SSE41-NEXT:    pextrb $12, %xmm0, %r12d
-; SSE41-NEXT:    movl %r12d, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %r12b
-; SSE41-NEXT:    jno .LBB2_107
-; SSE41-NEXT:  # %bb.106:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %r12d
-; SSE41-NEXT:  .LBB2_107:
-; SSE41-NEXT:    pextrb $11, %xmm4, %ecx
-; SSE41-NEXT:    pextrb $11, %xmm0, %r13d
-; SSE41-NEXT:    movl %r13d, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %r13b
-; SSE41-NEXT:    movl %r11d, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE41-NEXT:    movl %r8d, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE41-NEXT:    movl %r9d, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE41-NEXT:    movl %r10d, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE41-NEXT:    movl %r14d, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE41-NEXT:    jno .LBB2_109
-; SSE41-NEXT:  # %bb.108:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %r13d
-; SSE41-NEXT:  .LBB2_109:
-; SSE41-NEXT:    pextrb $10, %xmm4, %ecx
-; SSE41-NEXT:    pextrb $10, %xmm0, %r15d
-; SSE41-NEXT:    movl %r15d, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %r15b
-; SSE41-NEXT:    jno .LBB2_111
-; SSE41-NEXT:  # %bb.110:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %r15d
-; SSE41-NEXT:  .LBB2_111:
-; SSE41-NEXT:    pextrb $9, %xmm4, %ecx
-; SSE41-NEXT:    pextrb $9, %xmm0, %r14d
-; SSE41-NEXT:    movl %r14d, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %r14b
-; SSE41-NEXT:    jno .LBB2_113
-; SSE41-NEXT:  # %bb.112:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %r14d
-; SSE41-NEXT:  .LBB2_113:
-; SSE41-NEXT:    pextrb $8, %xmm4, %ecx
-; SSE41-NEXT:    pextrb $8, %xmm0, %ebp
-; SSE41-NEXT:    movl %ebp, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %bpl
-; SSE41-NEXT:    jno .LBB2_115
-; SSE41-NEXT:  # %bb.114:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %ebp
-; SSE41-NEXT:  .LBB2_115:
-; SSE41-NEXT:    pextrb $7, %xmm4, %ecx
-; SSE41-NEXT:    pextrb $7, %xmm0, %edi
-; SSE41-NEXT:    movl %edi, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %dil
-; SSE41-NEXT:    jno .LBB2_117
-; SSE41-NEXT:  # %bb.116:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %edi
-; SSE41-NEXT:  .LBB2_117:
-; SSE41-NEXT:    pextrb $6, %xmm4, %edx
-; SSE41-NEXT:    pextrb $6, %xmm0, %eax
-; SSE41-NEXT:    movl %eax, %ecx
-; SSE41-NEXT:    addb %dl, %cl
-; SSE41-NEXT:    setns %cl
-; SSE41-NEXT:    addb %dl, %al
-; SSE41-NEXT:    jno .LBB2_119
-; SSE41-NEXT:  # %bb.118:
-; SSE41-NEXT:    addb $127, %cl
-; SSE41-NEXT:    movl %ecx, %eax
-; SSE41-NEXT:  .LBB2_119:
-; SSE41-NEXT:    pextrb $5, %xmm4, %ebx
-; SSE41-NEXT:    pextrb $5, %xmm0, %ecx
-; SSE41-NEXT:    movl %ecx, %edx
-; SSE41-NEXT:    addb %bl, %dl
-; SSE41-NEXT:    setns %dl
-; SSE41-NEXT:    addb %bl, %cl
-; SSE41-NEXT:    jno .LBB2_121
-; SSE41-NEXT:  # %bb.120:
-; SSE41-NEXT:    addb $127, %dl
-; SSE41-NEXT:    movl %edx, %ecx
-; SSE41-NEXT:  .LBB2_121:
-; SSE41-NEXT:    pextrb $4, %xmm4, %esi
-; SSE41-NEXT:    pextrb $4, %xmm0, %edx
-; SSE41-NEXT:    movl %edx, %ebx
-; SSE41-NEXT:    addb %sil, %bl
-; SSE41-NEXT:    setns %bl
-; SSE41-NEXT:    addb %sil, %dl
-; SSE41-NEXT:    jno .LBB2_123
-; SSE41-NEXT:  # %bb.122:
-; SSE41-NEXT:    addb $127, %bl
-; SSE41-NEXT:    movl %ebx, %edx
-; SSE41-NEXT:  .LBB2_123:
-; SSE41-NEXT:    pextrb $3, %xmm4, %esi
-; SSE41-NEXT:    pextrb $3, %xmm0, %r8d
-; SSE41-NEXT:    movl %r8d, %ebx
-; SSE41-NEXT:    addb %sil, %bl
-; SSE41-NEXT:    setns %bl
-; SSE41-NEXT:    addb %sil, %r8b
-; SSE41-NEXT:    jno .LBB2_125
-; SSE41-NEXT:  # %bb.124:
-; SSE41-NEXT:    addb $127, %bl
-; SSE41-NEXT:    movl %ebx, %r8d
-; SSE41-NEXT:  .LBB2_125:
-; SSE41-NEXT:    pextrb $2, %xmm4, %esi
-; SSE41-NEXT:    pextrb $2, %xmm0, %r9d
-; SSE41-NEXT:    movl %r9d, %ebx
-; SSE41-NEXT:    addb %sil, %bl
-; SSE41-NEXT:    setns %bl
-; SSE41-NEXT:    addb %sil, %r9b
-; SSE41-NEXT:    jno .LBB2_127
-; SSE41-NEXT:  # %bb.126:
-; SSE41-NEXT:    addb $127, %bl
-; SSE41-NEXT:    movl %ebx, %r9d
-; SSE41-NEXT:  .LBB2_127:
-; SSE41-NEXT:    pextrb $0, %xmm4, %esi
-; SSE41-NEXT:    pextrb $0, %xmm0, %r10d
-; SSE41-NEXT:    movl %r10d, %ebx
-; SSE41-NEXT:    addb %sil, %bl
-; SSE41-NEXT:    setns %bl
-; SSE41-NEXT:    addb %sil, %r10b
-; SSE41-NEXT:    jno .LBB2_129
-; SSE41-NEXT:  # %bb.128:
-; SSE41-NEXT:    addb $127, %bl
-; SSE41-NEXT:    movl %ebx, %r10d
-; SSE41-NEXT:  .LBB2_129:
-; SSE41-NEXT:    pextrb $1, %xmm4, %esi
-; SSE41-NEXT:    pextrb $1, %xmm0, %r11d
-; SSE41-NEXT:    movl %r11d, %ebx
-; SSE41-NEXT:    addb %sil, %bl
-; SSE41-NEXT:    setns %bl
-; SSE41-NEXT:    addb %sil, %r11b
-; SSE41-NEXT:    jno .LBB2_131
-; SSE41-NEXT:  # %bb.130:
-; SSE41-NEXT:    addb $127, %bl
-; SSE41-NEXT:    movl %ebx, %r11d
-; SSE41-NEXT:  .LBB2_131:
-; SSE41-NEXT:    movzbl %r10b, %esi
-; SSE41-NEXT:    movd %esi, %xmm0
-; SSE41-NEXT:    movzbl %r11b, %esi
-; SSE41-NEXT:    pinsrb $1, %esi, %xmm0
-; SSE41-NEXT:    movzbl %r9b, %esi
-; SSE41-NEXT:    pinsrb $2, %esi, %xmm0
-; SSE41-NEXT:    movzbl %r8b, %esi
-; SSE41-NEXT:    pinsrb $3, %esi, %xmm0
-; SSE41-NEXT:    movzbl %dl, %edx
-; SSE41-NEXT:    pinsrb $4, %edx, %xmm0
-; SSE41-NEXT:    movzbl %cl, %ecx
-; SSE41-NEXT:    pinsrb $5, %ecx, %xmm0
-; SSE41-NEXT:    movzbl %al, %eax
-; SSE41-NEXT:    pinsrb $6, %eax, %xmm0
-; SSE41-NEXT:    movzbl %dil, %eax
-; SSE41-NEXT:    pinsrb $7, %eax, %xmm0
-; SSE41-NEXT:    movzbl %bpl, %eax
-; SSE41-NEXT:    pinsrb $8, %eax, %xmm0
-; SSE41-NEXT:    movzbl %r14b, %eax
-; SSE41-NEXT:    pinsrb $9, %eax, %xmm0
-; SSE41-NEXT:    movzbl %r15b, %eax
-; SSE41-NEXT:    pinsrb $10, %eax, %xmm0
-; SSE41-NEXT:    movzbl %r13b, %eax
-; SSE41-NEXT:    pinsrb $11, %eax, %xmm0
-; SSE41-NEXT:    movzbl %r12b, %eax
-; SSE41-NEXT:    pinsrb $12, %eax, %xmm0
-; SSE41-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; SSE41-NEXT:    pinsrb $13, %eax, %xmm0
-; SSE41-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; SSE41-NEXT:    pinsrb $14, %eax, %xmm0
-; SSE41-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; SSE41-NEXT:    pinsrb $15, %eax, %xmm0
-; SSE41-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; SSE41-NEXT:    movd %eax, %xmm1
-; SSE41-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; SSE41-NEXT:    pinsrb $1, %eax, %xmm1
-; SSE41-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; SSE41-NEXT:    pinsrb $2, %eax, %xmm1
-; SSE41-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; SSE41-NEXT:    pinsrb $3, %eax, %xmm1
-; SSE41-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; SSE41-NEXT:    pinsrb $4, %eax, %xmm1
-; SSE41-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; SSE41-NEXT:    pinsrb $5, %eax, %xmm1
-; SSE41-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; SSE41-NEXT:    pinsrb $6, %eax, %xmm1
-; SSE41-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; SSE41-NEXT:    pinsrb $7, %eax, %xmm1
-; SSE41-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; SSE41-NEXT:    pinsrb $8, %eax, %xmm1
-; SSE41-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; SSE41-NEXT:    pinsrb $9, %eax, %xmm1
-; SSE41-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; SSE41-NEXT:    pinsrb $10, %eax, %xmm1
-; SSE41-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; SSE41-NEXT:    pinsrb $11, %eax, %xmm1
-; SSE41-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; SSE41-NEXT:    pinsrb $12, %eax, %xmm1
-; SSE41-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; SSE41-NEXT:    pinsrb $13, %eax, %xmm1
-; SSE41-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; SSE41-NEXT:    pinsrb $14, %eax, %xmm1
-; SSE41-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; SSE41-NEXT:    pinsrb $15, %eax, %xmm1
-; SSE41-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; SSE41-NEXT:    movd %eax, %xmm2
-; SSE41-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; SSE41-NEXT:    pinsrb $1, %eax, %xmm2
-; SSE41-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; SSE41-NEXT:    pinsrb $2, %eax, %xmm2
-; SSE41-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; SSE41-NEXT:    pinsrb $3, %eax, %xmm2
-; SSE41-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; SSE41-NEXT:    pinsrb $4, %eax, %xmm2
-; SSE41-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; SSE41-NEXT:    pinsrb $5, %eax, %xmm2
-; SSE41-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; SSE41-NEXT:    pinsrb $6, %eax, %xmm2
-; SSE41-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; SSE41-NEXT:    pinsrb $7, %eax, %xmm2
-; SSE41-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; SSE41-NEXT:    pinsrb $8, %eax, %xmm2
-; SSE41-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; SSE41-NEXT:    pinsrb $9, %eax, %xmm2
-; SSE41-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; SSE41-NEXT:    pinsrb $10, %eax, %xmm2
-; SSE41-NEXT:    movzbl (%rsp), %eax # 1-byte Folded Reload
-; SSE41-NEXT:    pinsrb $11, %eax, %xmm2
-; SSE41-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; SSE41-NEXT:    pinsrb $12, %eax, %xmm2
-; SSE41-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; SSE41-NEXT:    pinsrb $13, %eax, %xmm2
-; SSE41-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; SSE41-NEXT:    pinsrb $14, %eax, %xmm2
-; SSE41-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; SSE41-NEXT:    pinsrb $15, %eax, %xmm2
-; SSE41-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; SSE41-NEXT:    movd %eax, %xmm3
-; SSE41-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; SSE41-NEXT:    pinsrb $1, %eax, %xmm3
-; SSE41-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; SSE41-NEXT:    pinsrb $2, %eax, %xmm3
-; SSE41-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; SSE41-NEXT:    pinsrb $3, %eax, %xmm3
-; SSE41-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; SSE41-NEXT:    pinsrb $4, %eax, %xmm3
-; SSE41-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; SSE41-NEXT:    pinsrb $5, %eax, %xmm3
-; SSE41-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; SSE41-NEXT:    pinsrb $6, %eax, %xmm3
-; SSE41-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; SSE41-NEXT:    pinsrb $7, %eax, %xmm3
-; SSE41-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; SSE41-NEXT:    pinsrb $8, %eax, %xmm3
-; SSE41-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; SSE41-NEXT:    pinsrb $9, %eax, %xmm3
-; SSE41-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; SSE41-NEXT:    pinsrb $10, %eax, %xmm3
-; SSE41-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; SSE41-NEXT:    pinsrb $11, %eax, %xmm3
-; SSE41-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; SSE41-NEXT:    pinsrb $12, %eax, %xmm3
-; SSE41-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; SSE41-NEXT:    pinsrb $13, %eax, %xmm3
-; SSE41-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; SSE41-NEXT:    pinsrb $14, %eax, %xmm3
-; SSE41-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; SSE41-NEXT:    pinsrb $15, %eax, %xmm3
-; SSE41-NEXT:    addq $76, %rsp
-; SSE41-NEXT:    popq %rbx
-; SSE41-NEXT:    popq %r12
-; SSE41-NEXT:    popq %r13
-; SSE41-NEXT:    popq %r14
-; SSE41-NEXT:    popq %r15
-; SSE41-NEXT:    popq %rbp
-; SSE41-NEXT:    retq
+; SSE-LABEL: v64i8:
+; SSE:       # %bb.0:
+; SSE-NEXT:    paddsb %xmm4, %xmm0
+; SSE-NEXT:    paddsb %xmm5, %xmm1
+; SSE-NEXT:    paddsb %xmm6, %xmm2
+; SSE-NEXT:    paddsb %xmm7, %xmm3
+; SSE-NEXT:    retq
 ;
 ; AVX1-LABEL: v64i8:
 ; AVX1:       # %bb.0:
-; AVX1-NEXT:    pushq %rbp
-; AVX1-NEXT:    pushq %r15
-; AVX1-NEXT:    pushq %r14
-; AVX1-NEXT:    pushq %r13
-; AVX1-NEXT:    pushq %r12
-; AVX1-NEXT:    pushq %rbx
-; AVX1-NEXT:    subq $76, %rsp
-; AVX1-NEXT:    vpextrb $15, %xmm3, %ecx
-; AVX1-NEXT:    vpextrb $15, %xmm1, %edx
-; AVX1-NEXT:    movl %edx, %eax
-; AVX1-NEXT:    addb %cl, %al
-; AVX1-NEXT:    setns %al
-; AVX1-NEXT:    addb %cl, %dl
-; AVX1-NEXT:    jo .LBB2_1
-; AVX1-NEXT:  # %bb.2:
-; AVX1-NEXT:    movl %edx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX1-NEXT:    jmp .LBB2_3
-; AVX1-NEXT:  .LBB2_1:
-; AVX1-NEXT:    addb $127, %al
-; AVX1-NEXT:    # kill: def $al killed $al def $eax
-; AVX1-NEXT:    movl %eax, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX1-NEXT:  .LBB2_3:
-; AVX1-NEXT:    vpextrb $14, %xmm3, %ecx
-; AVX1-NEXT:    vpextrb $14, %xmm1, %edx
-; AVX1-NEXT:    movl %edx, %eax
-; AVX1-NEXT:    addb %cl, %al
-; AVX1-NEXT:    setns %al
-; AVX1-NEXT:    addb %cl, %dl
-; AVX1-NEXT:    jno .LBB2_5
-; AVX1-NEXT:  # %bb.4:
-; AVX1-NEXT:    addb $127, %al
-; AVX1-NEXT:    movl %eax, %edx
-; AVX1-NEXT:  .LBB2_5:
-; AVX1-NEXT:    vpextrb $13, %xmm3, %ecx
-; AVX1-NEXT:    vpextrb $13, %xmm1, %esi
-; AVX1-NEXT:    movl %esi, %eax
-; AVX1-NEXT:    addb %cl, %al
-; AVX1-NEXT:    setns %al
-; AVX1-NEXT:    addb %cl, %sil
-; AVX1-NEXT:    jo .LBB2_6
-; AVX1-NEXT:  # %bb.7:
-; AVX1-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX1-NEXT:    jmp .LBB2_8
-; AVX1-NEXT:  .LBB2_6:
-; AVX1-NEXT:    addb $127, %al
-; AVX1-NEXT:    # kill: def $al killed $al def $eax
-; AVX1-NEXT:    movl %eax, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX1-NEXT:  .LBB2_8:
-; AVX1-NEXT:    vpextrb $12, %xmm3, %ecx
-; AVX1-NEXT:    vpextrb $12, %xmm1, %esi
-; AVX1-NEXT:    movl %esi, %eax
-; AVX1-NEXT:    addb %cl, %al
-; AVX1-NEXT:    setns %al
-; AVX1-NEXT:    addb %cl, %sil
-; AVX1-NEXT:    jno .LBB2_10
-; AVX1-NEXT:  # %bb.9:
-; AVX1-NEXT:    addb $127, %al
-; AVX1-NEXT:    movl %eax, %esi
-; AVX1-NEXT:  .LBB2_10:
-; AVX1-NEXT:    vpextrb $11, %xmm3, %ecx
-; AVX1-NEXT:    vpextrb $11, %xmm1, %edi
-; AVX1-NEXT:    movl %edi, %eax
-; AVX1-NEXT:    addb %cl, %al
-; AVX1-NEXT:    setns %al
-; AVX1-NEXT:    addb %cl, %dil
-; AVX1-NEXT:    jno .LBB2_12
-; AVX1-NEXT:  # %bb.11:
-; AVX1-NEXT:    addb $127, %al
-; AVX1-NEXT:    movl %eax, %edi
-; AVX1-NEXT:  .LBB2_12:
-; AVX1-NEXT:    vpextrb $10, %xmm3, %ecx
-; AVX1-NEXT:    vpextrb $10, %xmm1, %ebp
-; AVX1-NEXT:    movl %ebp, %eax
-; AVX1-NEXT:    addb %cl, %al
-; AVX1-NEXT:    setns %al
-; AVX1-NEXT:    addb %cl, %bpl
-; AVX1-NEXT:    jno .LBB2_14
-; AVX1-NEXT:  # %bb.13:
-; AVX1-NEXT:    addb $127, %al
-; AVX1-NEXT:    movl %eax, %ebp
-; AVX1-NEXT:  .LBB2_14:
-; AVX1-NEXT:    vpextrb $9, %xmm3, %ecx
-; AVX1-NEXT:    vpextrb $9, %xmm1, %ebx
-; AVX1-NEXT:    movl %ebx, %eax
-; AVX1-NEXT:    addb %cl, %al
-; AVX1-NEXT:    setns %al
-; AVX1-NEXT:    addb %cl, %bl
-; AVX1-NEXT:    movl %ebp, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX1-NEXT:    jo .LBB2_15
-; AVX1-NEXT:  # %bb.16:
-; AVX1-NEXT:    movl %ebx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX1-NEXT:    jmp .LBB2_17
-; AVX1-NEXT:  .LBB2_15:
-; AVX1-NEXT:    addb $127, %al
-; AVX1-NEXT:    # kill: def $al killed $al def $eax
-; AVX1-NEXT:    movl %eax, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX1-NEXT:  .LBB2_17:
-; AVX1-NEXT:    vpextrb $8, %xmm3, %ecx
-; AVX1-NEXT:    vpextrb $8, %xmm1, %ebp
-; AVX1-NEXT:    movl %ebp, %eax
-; AVX1-NEXT:    addb %cl, %al
-; AVX1-NEXT:    setns %al
-; AVX1-NEXT:    addb %cl, %bpl
-; AVX1-NEXT:    jno .LBB2_19
-; AVX1-NEXT:  # %bb.18:
-; AVX1-NEXT:    addb $127, %al
-; AVX1-NEXT:    movl %eax, %ebp
-; AVX1-NEXT:  .LBB2_19:
-; AVX1-NEXT:    vpextrb $7, %xmm3, %ecx
-; AVX1-NEXT:    vpextrb $7, %xmm1, %ebx
-; AVX1-NEXT:    movl %ebx, %eax
-; AVX1-NEXT:    addb %cl, %al
-; AVX1-NEXT:    setns %al
-; AVX1-NEXT:    addb %cl, %bl
-; AVX1-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX1-NEXT:    jno .LBB2_21
-; AVX1-NEXT:  # %bb.20:
-; AVX1-NEXT:    addb $127, %al
-; AVX1-NEXT:    movl %eax, %ebx
-; AVX1-NEXT:  .LBB2_21:
-; AVX1-NEXT:    vpextrb $6, %xmm3, %ecx
-; AVX1-NEXT:    vpextrb $6, %xmm1, %esi
-; AVX1-NEXT:    movl %esi, %eax
-; AVX1-NEXT:    addb %cl, %al
-; AVX1-NEXT:    setns %al
-; AVX1-NEXT:    addb %cl, %sil
-; AVX1-NEXT:    movl %edx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX1-NEXT:    jno .LBB2_23
-; AVX1-NEXT:  # %bb.22:
-; AVX1-NEXT:    addb $127, %al
-; AVX1-NEXT:    movl %eax, %esi
-; AVX1-NEXT:  .LBB2_23:
-; AVX1-NEXT:    vpextrb $5, %xmm3, %ecx
-; AVX1-NEXT:    vpextrb $5, %xmm1, %r11d
-; AVX1-NEXT:    movl %r11d, %eax
-; AVX1-NEXT:    addb %cl, %al
-; AVX1-NEXT:    setns %al
-; AVX1-NEXT:    addb %cl, %r11b
-; AVX1-NEXT:    movl %edi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX1-NEXT:    movl %ebp, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX1-NEXT:    movl %ebx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX1-NEXT:    jno .LBB2_25
-; AVX1-NEXT:  # %bb.24:
-; AVX1-NEXT:    addb $127, %al
-; AVX1-NEXT:    movl %eax, %r11d
-; AVX1-NEXT:  .LBB2_25:
-; AVX1-NEXT:    vpextrb $4, %xmm3, %ecx
-; AVX1-NEXT:    vpextrb $4, %xmm1, %r13d
-; AVX1-NEXT:    movl %r13d, %eax
-; AVX1-NEXT:    addb %cl, %al
-; AVX1-NEXT:    setns %al
-; AVX1-NEXT:    addb %cl, %r13b
-; AVX1-NEXT:    jno .LBB2_27
-; AVX1-NEXT:  # %bb.26:
-; AVX1-NEXT:    addb $127, %al
-; AVX1-NEXT:    movl %eax, %r13d
-; AVX1-NEXT:  .LBB2_27:
-; AVX1-NEXT:    vpextrb $3, %xmm3, %ecx
-; AVX1-NEXT:    vpextrb $3, %xmm1, %r8d
-; AVX1-NEXT:    movl %r8d, %eax
-; AVX1-NEXT:    addb %cl, %al
-; AVX1-NEXT:    setns %al
-; AVX1-NEXT:    addb %cl, %r8b
-; AVX1-NEXT:    jno .LBB2_29
-; AVX1-NEXT:  # %bb.28:
-; AVX1-NEXT:    addb $127, %al
-; AVX1-NEXT:    movl %eax, %r8d
-; AVX1-NEXT:  .LBB2_29:
-; AVX1-NEXT:    vpextrb $2, %xmm3, %ecx
-; AVX1-NEXT:    vpextrb $2, %xmm1, %ebp
-; AVX1-NEXT:    movl %ebp, %eax
-; AVX1-NEXT:    addb %cl, %al
-; AVX1-NEXT:    setns %al
-; AVX1-NEXT:    addb %cl, %bpl
-; AVX1-NEXT:    jno .LBB2_31
-; AVX1-NEXT:  # %bb.30:
-; AVX1-NEXT:    addb $127, %al
-; AVX1-NEXT:    movl %eax, %ebp
-; AVX1-NEXT:  .LBB2_31:
-; AVX1-NEXT:    vpextrb $0, %xmm3, %ecx
-; AVX1-NEXT:    vpextrb $0, %xmm1, %ebx
-; AVX1-NEXT:    movl %ebx, %eax
-; AVX1-NEXT:    addb %cl, %al
-; AVX1-NEXT:    setns %al
-; AVX1-NEXT:    addb %cl, %bl
-; AVX1-NEXT:    jno .LBB2_33
-; AVX1-NEXT:  # %bb.32:
-; AVX1-NEXT:    addb $127, %al
-; AVX1-NEXT:    movl %eax, %ebx
-; AVX1-NEXT:  .LBB2_33:
-; AVX1-NEXT:    vpextrb $1, %xmm3, %ecx
-; AVX1-NEXT:    vpextrb $1, %xmm1, %edi
-; AVX1-NEXT:    movl %edi, %eax
-; AVX1-NEXT:    addb %cl, %al
-; AVX1-NEXT:    setns %al
-; AVX1-NEXT:    addb %cl, %dil
-; AVX1-NEXT:    jno .LBB2_35
-; AVX1-NEXT:  # %bb.34:
-; AVX1-NEXT:    addb $127, %al
-; AVX1-NEXT:    movl %eax, %edi
-; AVX1-NEXT:  .LBB2_35:
-; AVX1-NEXT:    vextractf128 $1, %ymm3, %xmm3
-; AVX1-NEXT:    vpextrb $15, %xmm3, %ecx
-; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm1
-; AVX1-NEXT:    vpextrb $15, %xmm1, %edx
-; AVX1-NEXT:    movl %edx, %eax
-; AVX1-NEXT:    addb %cl, %al
-; AVX1-NEXT:    setns %al
-; AVX1-NEXT:    addb %cl, %dl
-; AVX1-NEXT:    movl %ebp, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX1-NEXT:    movl %ebx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX1-NEXT:    movl %edi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX1-NEXT:    jno .LBB2_37
-; AVX1-NEXT:  # %bb.36:
-; AVX1-NEXT:    addb $127, %al
-; AVX1-NEXT:    movl %eax, %edx
-; AVX1-NEXT:  .LBB2_37:
-; AVX1-NEXT:    vpextrb $14, %xmm3, %ecx
-; AVX1-NEXT:    vpextrb $14, %xmm1, %edi
-; AVX1-NEXT:    movl %edi, %eax
-; AVX1-NEXT:    addb %cl, %al
-; AVX1-NEXT:    setns %al
-; AVX1-NEXT:    addb %cl, %dil
-; AVX1-NEXT:    movl %edx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX1-NEXT:    jno .LBB2_39
-; AVX1-NEXT:  # %bb.38:
-; AVX1-NEXT:    addb $127, %al
-; AVX1-NEXT:    movl %eax, %edi
-; AVX1-NEXT:  .LBB2_39:
-; AVX1-NEXT:    vpextrb $13, %xmm3, %ecx
-; AVX1-NEXT:    vpextrb $13, %xmm1, %r12d
-; AVX1-NEXT:    movl %r12d, %eax
-; AVX1-NEXT:    addb %cl, %al
-; AVX1-NEXT:    setns %al
-; AVX1-NEXT:    addb %cl, %r12b
-; AVX1-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX1-NEXT:    jno .LBB2_41
-; AVX1-NEXT:  # %bb.40:
-; AVX1-NEXT:    addb $127, %al
-; AVX1-NEXT:    movl %eax, %r12d
-; AVX1-NEXT:  .LBB2_41:
-; AVX1-NEXT:    vpextrb $12, %xmm3, %ecx
-; AVX1-NEXT:    vpextrb $12, %xmm1, %r15d
-; AVX1-NEXT:    movl %r15d, %eax
-; AVX1-NEXT:    addb %cl, %al
-; AVX1-NEXT:    setns %al
-; AVX1-NEXT:    addb %cl, %r15b
-; AVX1-NEXT:    movl %r11d, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX1-NEXT:    jno .LBB2_43
-; AVX1-NEXT:  # %bb.42:
-; AVX1-NEXT:    addb $127, %al
-; AVX1-NEXT:    movl %eax, %r15d
-; AVX1-NEXT:  .LBB2_43:
-; AVX1-NEXT:    vpextrb $11, %xmm3, %ecx
-; AVX1-NEXT:    vpextrb $11, %xmm1, %ebp
-; AVX1-NEXT:    movl %ebp, %eax
-; AVX1-NEXT:    addb %cl, %al
-; AVX1-NEXT:    setns %al
-; AVX1-NEXT:    addb %cl, %bpl
-; AVX1-NEXT:    movl %r13d, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX1-NEXT:    jno .LBB2_45
-; AVX1-NEXT:  # %bb.44:
-; AVX1-NEXT:    addb $127, %al
-; AVX1-NEXT:    movl %eax, %ebp
-; AVX1-NEXT:  .LBB2_45:
-; AVX1-NEXT:    vpextrb $10, %xmm3, %ecx
-; AVX1-NEXT:    vpextrb $10, %xmm1, %ebx
-; AVX1-NEXT:    movl %ebx, %eax
-; AVX1-NEXT:    addb %cl, %al
-; AVX1-NEXT:    setns %al
-; AVX1-NEXT:    addb %cl, %bl
-; AVX1-NEXT:    jno .LBB2_47
-; AVX1-NEXT:  # %bb.46:
-; AVX1-NEXT:    addb $127, %al
-; AVX1-NEXT:    movl %eax, %ebx
-; AVX1-NEXT:  .LBB2_47:
-; AVX1-NEXT:    vpextrb $9, %xmm3, %ecx
-; AVX1-NEXT:    vpextrb $9, %xmm1, %edx
-; AVX1-NEXT:    movl %edx, %eax
-; AVX1-NEXT:    addb %cl, %al
-; AVX1-NEXT:    setns %al
-; AVX1-NEXT:    addb %cl, %dl
-; AVX1-NEXT:    jno .LBB2_49
-; AVX1-NEXT:  # %bb.48:
-; AVX1-NEXT:    addb $127, %al
-; AVX1-NEXT:    movl %eax, %edx
-; AVX1-NEXT:  .LBB2_49:
-; AVX1-NEXT:    vpextrb $8, %xmm3, %ecx
-; AVX1-NEXT:    vpextrb $8, %xmm1, %esi
-; AVX1-NEXT:    movl %esi, %eax
-; AVX1-NEXT:    addb %cl, %al
-; AVX1-NEXT:    setns %al
-; AVX1-NEXT:    addb %cl, %sil
-; AVX1-NEXT:    movl %ebp, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX1-NEXT:    movl %ebx, (%rsp) # 4-byte Spill
-; AVX1-NEXT:    movl %edx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX1-NEXT:    jo .LBB2_50
-; AVX1-NEXT:  # %bb.51:
-; AVX1-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX1-NEXT:    jmp .LBB2_52
-; AVX1-NEXT:  .LBB2_50:
-; AVX1-NEXT:    addb $127, %al
-; AVX1-NEXT:    # kill: def $al killed $al def $eax
-; AVX1-NEXT:    movl %eax, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX1-NEXT:  .LBB2_52:
-; AVX1-NEXT:    vpextrb $7, %xmm3, %ecx
-; AVX1-NEXT:    vpextrb $7, %xmm1, %r11d
-; AVX1-NEXT:    movl %r11d, %eax
-; AVX1-NEXT:    addb %cl, %al
-; AVX1-NEXT:    setns %al
-; AVX1-NEXT:    addb %cl, %r11b
-; AVX1-NEXT:    jno .LBB2_54
-; AVX1-NEXT:  # %bb.53:
-; AVX1-NEXT:    addb $127, %al
-; AVX1-NEXT:    movl %eax, %r11d
-; AVX1-NEXT:  .LBB2_54:
-; AVX1-NEXT:    vpextrb $6, %xmm3, %ecx
-; AVX1-NEXT:    vpextrb $6, %xmm1, %esi
-; AVX1-NEXT:    movl %esi, %eax
-; AVX1-NEXT:    addb %cl, %al
-; AVX1-NEXT:    setns %al
-; AVX1-NEXT:    addb %cl, %sil
-; AVX1-NEXT:    movl %r8d, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX1-NEXT:    movl %edi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX1-NEXT:    jno .LBB2_56
-; AVX1-NEXT:  # %bb.55:
-; AVX1-NEXT:    addb $127, %al
-; AVX1-NEXT:    movl %eax, %esi
-; AVX1-NEXT:  .LBB2_56:
-; AVX1-NEXT:    vpextrb $5, %xmm3, %ecx
-; AVX1-NEXT:    vpextrb $5, %xmm1, %edi
-; AVX1-NEXT:    movl %edi, %eax
-; AVX1-NEXT:    addb %cl, %al
-; AVX1-NEXT:    setns %al
-; AVX1-NEXT:    addb %cl, %dil
-; AVX1-NEXT:    jno .LBB2_58
-; AVX1-NEXT:  # %bb.57:
-; AVX1-NEXT:    addb $127, %al
-; AVX1-NEXT:    movl %eax, %edi
-; AVX1-NEXT:  .LBB2_58:
-; AVX1-NEXT:    vpextrb $4, %xmm3, %ecx
-; AVX1-NEXT:    vpextrb $4, %xmm1, %r13d
-; AVX1-NEXT:    movl %r13d, %eax
-; AVX1-NEXT:    addb %cl, %al
-; AVX1-NEXT:    setns %al
-; AVX1-NEXT:    addb %cl, %r13b
-; AVX1-NEXT:    movl %r13d, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX1-NEXT:    jno .LBB2_60
-; AVX1-NEXT:  # %bb.59:
-; AVX1-NEXT:    addb $127, %al
-; AVX1-NEXT:    # kill: def $al killed $al def $eax
-; AVX1-NEXT:    movl %eax, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX1-NEXT:  .LBB2_60:
-; AVX1-NEXT:    vpextrb $3, %xmm3, %ecx
-; AVX1-NEXT:    vpextrb $3, %xmm1, %ebp
-; AVX1-NEXT:    movl %ebp, %eax
-; AVX1-NEXT:    addb %cl, %al
-; AVX1-NEXT:    setns %al
-; AVX1-NEXT:    addb %cl, %bpl
-; AVX1-NEXT:    jo .LBB2_61
-; AVX1-NEXT:  # %bb.62:
-; AVX1-NEXT:    movl %ebp, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX1-NEXT:    jmp .LBB2_63
-; AVX1-NEXT:  .LBB2_61:
-; AVX1-NEXT:    addb $127, %al
-; AVX1-NEXT:    # kill: def $al killed $al def $eax
-; AVX1-NEXT:    movl %eax, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX1-NEXT:  .LBB2_63:
-; AVX1-NEXT:    vpextrb $2, %xmm3, %ecx
-; AVX1-NEXT:    vpextrb $2, %xmm1, %ebp
-; AVX1-NEXT:    movl %ebp, %eax
-; AVX1-NEXT:    addb %cl, %al
-; AVX1-NEXT:    setns %al
-; AVX1-NEXT:    addb %cl, %bpl
-; AVX1-NEXT:    movl %edi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX1-NEXT:    jno .LBB2_65
-; AVX1-NEXT:  # %bb.64:
-; AVX1-NEXT:    addb $127, %al
-; AVX1-NEXT:    movl %eax, %ebp
-; AVX1-NEXT:  .LBB2_65:
-; AVX1-NEXT:    vpextrb $0, %xmm3, %ecx
-; AVX1-NEXT:    vpextrb $0, %xmm1, %edi
-; AVX1-NEXT:    movl %edi, %eax
-; AVX1-NEXT:    addb %cl, %al
-; AVX1-NEXT:    setns %al
-; AVX1-NEXT:    addb %cl, %dil
-; AVX1-NEXT:    movl %r15d, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX1-NEXT:    jno .LBB2_67
-; AVX1-NEXT:  # %bb.66:
-; AVX1-NEXT:    addb $127, %al
-; AVX1-NEXT:    movl %eax, %edi
-; AVX1-NEXT:  .LBB2_67:
-; AVX1-NEXT:    vpextrb $1, %xmm3, %ecx
-; AVX1-NEXT:    vpextrb $1, %xmm1, %ebx
-; AVX1-NEXT:    movl %ebx, %eax
-; AVX1-NEXT:    addb %cl, %al
-; AVX1-NEXT:    setns %al
-; AVX1-NEXT:    addb %cl, %bl
-; AVX1-NEXT:    movl %r12d, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX1-NEXT:    jno .LBB2_69
-; AVX1-NEXT:  # %bb.68:
-; AVX1-NEXT:    addb $127, %al
-; AVX1-NEXT:    movl %eax, %ebx
-; AVX1-NEXT:  .LBB2_69:
-; AVX1-NEXT:    vpextrb $15, %xmm2, %ecx
-; AVX1-NEXT:    vpextrb $15, %xmm0, %edx
-; AVX1-NEXT:    movl %edx, %eax
-; AVX1-NEXT:    addb %cl, %al
-; AVX1-NEXT:    setns %al
-; AVX1-NEXT:    addb %cl, %dl
-; AVX1-NEXT:    movl %edi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX1-NEXT:    jno .LBB2_71
-; AVX1-NEXT:  # %bb.70:
-; AVX1-NEXT:    addb $127, %al
-; AVX1-NEXT:    movl %eax, %edx
-; AVX1-NEXT:  .LBB2_71:
-; AVX1-NEXT:    vpextrb $14, %xmm2, %ecx
-; AVX1-NEXT:    vpextrb $14, %xmm0, %edi
-; AVX1-NEXT:    movl %edi, %eax
-; AVX1-NEXT:    addb %cl, %al
-; AVX1-NEXT:    setns %al
-; AVX1-NEXT:    addb %cl, %dil
-; AVX1-NEXT:    jno .LBB2_73
-; AVX1-NEXT:  # %bb.72:
-; AVX1-NEXT:    addb $127, %al
-; AVX1-NEXT:    movl %eax, %edi
-; AVX1-NEXT:  .LBB2_73:
-; AVX1-NEXT:    vpextrb $13, %xmm2, %ecx
-; AVX1-NEXT:    vpextrb $13, %xmm0, %r10d
-; AVX1-NEXT:    movl %r10d, %eax
-; AVX1-NEXT:    addb %cl, %al
-; AVX1-NEXT:    setns %al
-; AVX1-NEXT:    addb %cl, %r10b
-; AVX1-NEXT:    jno .LBB2_75
-; AVX1-NEXT:  # %bb.74:
-; AVX1-NEXT:    addb $127, %al
-; AVX1-NEXT:    movl %eax, %r10d
-; AVX1-NEXT:  .LBB2_75:
-; AVX1-NEXT:    vpextrb $12, %xmm2, %ecx
-; AVX1-NEXT:    vpextrb $12, %xmm0, %r12d
-; AVX1-NEXT:    movl %r12d, %eax
-; AVX1-NEXT:    addb %cl, %al
-; AVX1-NEXT:    setns %al
-; AVX1-NEXT:    addb %cl, %r12b
-; AVX1-NEXT:    movl %r12d, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX1-NEXT:    jno .LBB2_77
-; AVX1-NEXT:  # %bb.76:
-; AVX1-NEXT:    addb $127, %al
-; AVX1-NEXT:    # kill: def $al killed $al def $eax
-; AVX1-NEXT:    movl %eax, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX1-NEXT:  .LBB2_77:
-; AVX1-NEXT:    vpextrb $11, %xmm2, %ecx
-; AVX1-NEXT:    vpextrb $11, %xmm0, %r14d
-; AVX1-NEXT:    movl %r14d, %eax
-; AVX1-NEXT:    addb %cl, %al
-; AVX1-NEXT:    setns %al
-; AVX1-NEXT:    addb %cl, %r14b
-; AVX1-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX1-NEXT:    jno .LBB2_79
-; AVX1-NEXT:  # %bb.78:
-; AVX1-NEXT:    addb $127, %al
-; AVX1-NEXT:    movl %eax, %r14d
-; AVX1-NEXT:  .LBB2_79:
-; AVX1-NEXT:    vpextrb $10, %xmm2, %ecx
-; AVX1-NEXT:    vpextrb $10, %xmm0, %r13d
-; AVX1-NEXT:    movl %r13d, %eax
-; AVX1-NEXT:    addb %cl, %al
-; AVX1-NEXT:    setns %al
-; AVX1-NEXT:    addb %cl, %r13b
-; AVX1-NEXT:    movl %r11d, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX1-NEXT:    movl %ebp, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX1-NEXT:    movl %ebx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX1-NEXT:    movl %edx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX1-NEXT:    jno .LBB2_81
-; AVX1-NEXT:  # %bb.80:
-; AVX1-NEXT:    addb $127, %al
-; AVX1-NEXT:    movl %eax, %r13d
-; AVX1-NEXT:  .LBB2_81:
-; AVX1-NEXT:    vpextrb $9, %xmm2, %ecx
-; AVX1-NEXT:    vpextrb $9, %xmm0, %r8d
-; AVX1-NEXT:    movl %r8d, %eax
-; AVX1-NEXT:    addb %cl, %al
-; AVX1-NEXT:    setns %al
-; AVX1-NEXT:    addb %cl, %r8b
-; AVX1-NEXT:    jno .LBB2_83
-; AVX1-NEXT:  # %bb.82:
-; AVX1-NEXT:    addb $127, %al
-; AVX1-NEXT:    movl %eax, %r8d
-; AVX1-NEXT:  .LBB2_83:
-; AVX1-NEXT:    vpextrb $8, %xmm2, %ecx
-; AVX1-NEXT:    vpextrb $8, %xmm0, %r15d
-; AVX1-NEXT:    movl %r15d, %eax
-; AVX1-NEXT:    addb %cl, %al
-; AVX1-NEXT:    setns %al
-; AVX1-NEXT:    addb %cl, %r15b
-; AVX1-NEXT:    jno .LBB2_85
-; AVX1-NEXT:  # %bb.84:
-; AVX1-NEXT:    addb $127, %al
-; AVX1-NEXT:    movl %eax, %r15d
-; AVX1-NEXT:  .LBB2_85:
-; AVX1-NEXT:    vpextrb $7, %xmm2, %ecx
-; AVX1-NEXT:    vpextrb $7, %xmm0, %r12d
-; AVX1-NEXT:    movl %r12d, %eax
-; AVX1-NEXT:    addb %cl, %al
-; AVX1-NEXT:    setns %al
-; AVX1-NEXT:    addb %cl, %r12b
-; AVX1-NEXT:    movl %r12d, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX1-NEXT:    jno .LBB2_87
-; AVX1-NEXT:  # %bb.86:
-; AVX1-NEXT:    addb $127, %al
-; AVX1-NEXT:    # kill: def $al killed $al def $eax
-; AVX1-NEXT:    movl %eax, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX1-NEXT:  .LBB2_87:
-; AVX1-NEXT:    vpextrb $6, %xmm2, %ecx
-; AVX1-NEXT:    vpextrb $6, %xmm0, %r12d
-; AVX1-NEXT:    movl %r12d, %eax
-; AVX1-NEXT:    addb %cl, %al
-; AVX1-NEXT:    setns %al
-; AVX1-NEXT:    addb %cl, %r12b
-; AVX1-NEXT:    jno .LBB2_89
-; AVX1-NEXT:  # %bb.88:
-; AVX1-NEXT:    addb $127, %al
-; AVX1-NEXT:    movl %eax, %r12d
-; AVX1-NEXT:  .LBB2_89:
-; AVX1-NEXT:    vpextrb $5, %xmm2, %ecx
-; AVX1-NEXT:    vpextrb $5, %xmm0, %ebp
-; AVX1-NEXT:    movl %ebp, %eax
-; AVX1-NEXT:    addb %cl, %al
-; AVX1-NEXT:    setns %al
-; AVX1-NEXT:    addb %cl, %bpl
-; AVX1-NEXT:    jno .LBB2_91
-; AVX1-NEXT:  # %bb.90:
-; AVX1-NEXT:    addb $127, %al
-; AVX1-NEXT:    movl %eax, %ebp
-; AVX1-NEXT:  .LBB2_91:
-; AVX1-NEXT:    vpextrb $4, %xmm2, %ecx
-; AVX1-NEXT:    vpextrb $4, %xmm0, %esi
-; AVX1-NEXT:    movl %esi, %eax
-; AVX1-NEXT:    addb %cl, %al
-; AVX1-NEXT:    setns %al
-; AVX1-NEXT:    addb %cl, %sil
-; AVX1-NEXT:    jno .LBB2_93
-; AVX1-NEXT:  # %bb.92:
-; AVX1-NEXT:    addb $127, %al
-; AVX1-NEXT:    movl %eax, %esi
-; AVX1-NEXT:  .LBB2_93:
-; AVX1-NEXT:    vpextrb $3, %xmm2, %ecx
-; AVX1-NEXT:    vpextrb $3, %xmm0, %edx
-; AVX1-NEXT:    movl %edx, %eax
-; AVX1-NEXT:    addb %cl, %al
-; AVX1-NEXT:    setns %al
-; AVX1-NEXT:    addb %cl, %dl
-; AVX1-NEXT:    movl %edi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX1-NEXT:    jno .LBB2_95
-; AVX1-NEXT:  # %bb.94:
-; AVX1-NEXT:    addb $127, %al
-; AVX1-NEXT:    movl %eax, %edx
-; AVX1-NEXT:  .LBB2_95:
-; AVX1-NEXT:    vpextrb $2, %xmm2, %ecx
-; AVX1-NEXT:    vpextrb $2, %xmm0, %edi
-; AVX1-NEXT:    movl %edi, %eax
-; AVX1-NEXT:    addb %cl, %al
-; AVX1-NEXT:    setns %al
-; AVX1-NEXT:    addb %cl, %dil
-; AVX1-NEXT:    jno .LBB2_97
-; AVX1-NEXT:  # %bb.96:
-; AVX1-NEXT:    addb $127, %al
-; AVX1-NEXT:    movl %eax, %edi
-; AVX1-NEXT:  .LBB2_97:
-; AVX1-NEXT:    vpextrb $0, %xmm2, %ecx
-; AVX1-NEXT:    vpextrb $0, %xmm0, %ebx
-; AVX1-NEXT:    movl %ebx, %eax
-; AVX1-NEXT:    addb %cl, %al
-; AVX1-NEXT:    setns %al
-; AVX1-NEXT:    addb %cl, %bl
-; AVX1-NEXT:    jno .LBB2_99
-; AVX1-NEXT:  # %bb.98:
-; AVX1-NEXT:    addb $127, %al
-; AVX1-NEXT:    movl %eax, %ebx
-; AVX1-NEXT:  .LBB2_99:
-; AVX1-NEXT:    vpextrb $1, %xmm2, %ecx
-; AVX1-NEXT:    vpextrb $1, %xmm0, %r11d
-; AVX1-NEXT:    movl %r11d, %eax
-; AVX1-NEXT:    addb %cl, %al
-; AVX1-NEXT:    setns %al
-; AVX1-NEXT:    addb %cl, %r11b
-; AVX1-NEXT:    movl %r11d, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX1-NEXT:    jno .LBB2_101
-; AVX1-NEXT:  # %bb.100:
-; AVX1-NEXT:    addb $127, %al
-; AVX1-NEXT:    # kill: def $al killed $al def $eax
-; AVX1-NEXT:    movl %eax, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX1-NEXT:  .LBB2_101:
-; AVX1-NEXT:    vextractf128 $1, %ymm2, %xmm1
-; AVX1-NEXT:    vpextrb $15, %xmm1, %ecx
-; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm0
-; AVX1-NEXT:    vpextrb $15, %xmm0, %r11d
-; AVX1-NEXT:    movl %r11d, %eax
-; AVX1-NEXT:    addb %cl, %al
-; AVX1-NEXT:    setns %al
-; AVX1-NEXT:    addb %cl, %r11b
-; AVX1-NEXT:    jno .LBB2_103
-; AVX1-NEXT:  # %bb.102:
-; AVX1-NEXT:    addb $127, %al
-; AVX1-NEXT:    movl %eax, %r11d
-; AVX1-NEXT:  .LBB2_103:
-; AVX1-NEXT:    vpextrb $14, %xmm1, %ecx
-; AVX1-NEXT:    vpextrb $14, %xmm0, %r9d
-; AVX1-NEXT:    movl %r9d, %eax
-; AVX1-NEXT:    addb %cl, %al
-; AVX1-NEXT:    setns %al
-; AVX1-NEXT:    addb %cl, %r9b
-; AVX1-NEXT:    movl %r9d, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX1-NEXT:    jno .LBB2_105
-; AVX1-NEXT:  # %bb.104:
-; AVX1-NEXT:    addb $127, %al
-; AVX1-NEXT:    # kill: def $al killed $al def $eax
-; AVX1-NEXT:    movl %eax, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX1-NEXT:  .LBB2_105:
-; AVX1-NEXT:    vpextrb $13, %xmm1, %ecx
-; AVX1-NEXT:    vpextrb $13, %xmm0, %r9d
-; AVX1-NEXT:    movl %r9d, %eax
-; AVX1-NEXT:    addb %cl, %al
-; AVX1-NEXT:    setns %al
-; AVX1-NEXT:    addb %cl, %r9b
-; AVX1-NEXT:    movl %r14d, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX1-NEXT:    movl %r13d, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX1-NEXT:    movl %r8d, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX1-NEXT:    movl %r15d, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX1-NEXT:    movl %r12d, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX1-NEXT:    movl %ebp, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX1-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX1-NEXT:    movl %edx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX1-NEXT:    movl %edi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX1-NEXT:    movl %ebx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX1-NEXT:    jno .LBB2_107
-; AVX1-NEXT:  # %bb.106:
-; AVX1-NEXT:    addb $127, %al
-; AVX1-NEXT:    movl %eax, %r9d
-; AVX1-NEXT:  .LBB2_107:
-; AVX1-NEXT:    vpextrb $12, %xmm1, %ecx
-; AVX1-NEXT:    vpextrb $12, %xmm0, %edx
-; AVX1-NEXT:    movl %edx, %eax
-; AVX1-NEXT:    addb %cl, %al
-; AVX1-NEXT:    setns %al
-; AVX1-NEXT:    addb %cl, %dl
-; AVX1-NEXT:    movl %r10d, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX1-NEXT:    jno .LBB2_109
-; AVX1-NEXT:  # %bb.108:
-; AVX1-NEXT:    addb $127, %al
-; AVX1-NEXT:    movl %eax, %edx
-; AVX1-NEXT:  .LBB2_109:
-; AVX1-NEXT:    vpextrb $11, %xmm1, %ecx
-; AVX1-NEXT:    vpextrb $11, %xmm0, %r13d
-; AVX1-NEXT:    movl %r13d, %eax
-; AVX1-NEXT:    addb %cl, %al
-; AVX1-NEXT:    setns %al
-; AVX1-NEXT:    addb %cl, %r13b
-; AVX1-NEXT:    jno .LBB2_111
-; AVX1-NEXT:  # %bb.110:
-; AVX1-NEXT:    addb $127, %al
-; AVX1-NEXT:    movl %eax, %r13d
-; AVX1-NEXT:  .LBB2_111:
-; AVX1-NEXT:    vpextrb $10, %xmm1, %ecx
-; AVX1-NEXT:    vpextrb $10, %xmm0, %r15d
-; AVX1-NEXT:    movl %r15d, %eax
-; AVX1-NEXT:    addb %cl, %al
-; AVX1-NEXT:    setns %al
-; AVX1-NEXT:    addb %cl, %r15b
-; AVX1-NEXT:    jno .LBB2_113
-; AVX1-NEXT:  # %bb.112:
-; AVX1-NEXT:    addb $127, %al
-; AVX1-NEXT:    movl %eax, %r15d
-; AVX1-NEXT:  .LBB2_113:
-; AVX1-NEXT:    vpextrb $9, %xmm1, %ecx
-; AVX1-NEXT:    vpextrb $9, %xmm0, %r14d
-; AVX1-NEXT:    movl %r14d, %eax
-; AVX1-NEXT:    addb %cl, %al
-; AVX1-NEXT:    setns %al
-; AVX1-NEXT:    addb %cl, %r14b
-; AVX1-NEXT:    jno .LBB2_115
-; AVX1-NEXT:  # %bb.114:
-; AVX1-NEXT:    addb $127, %al
-; AVX1-NEXT:    movl %eax, %r14d
-; AVX1-NEXT:  .LBB2_115:
-; AVX1-NEXT:    vpextrb $8, %xmm1, %ecx
-; AVX1-NEXT:    vpextrb $8, %xmm0, %ebp
-; AVX1-NEXT:    movl %ebp, %eax
-; AVX1-NEXT:    addb %cl, %al
-; AVX1-NEXT:    setns %al
-; AVX1-NEXT:    addb %cl, %bpl
-; AVX1-NEXT:    jno .LBB2_117
-; AVX1-NEXT:  # %bb.116:
-; AVX1-NEXT:    addb $127, %al
-; AVX1-NEXT:    movl %eax, %ebp
-; AVX1-NEXT:  .LBB2_117:
-; AVX1-NEXT:    vpextrb $7, %xmm1, %ecx
-; AVX1-NEXT:    vpextrb $7, %xmm0, %edi
-; AVX1-NEXT:    movl %edi, %eax
-; AVX1-NEXT:    addb %cl, %al
-; AVX1-NEXT:    setns %al
-; AVX1-NEXT:    addb %cl, %dil
-; AVX1-NEXT:    movl %edx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX1-NEXT:    jno .LBB2_119
-; AVX1-NEXT:  # %bb.118:
-; AVX1-NEXT:    addb $127, %al
-; AVX1-NEXT:    movl %eax, %edi
-; AVX1-NEXT:  .LBB2_119:
-; AVX1-NEXT:    vpextrb $6, %xmm1, %edx
-; AVX1-NEXT:    vpextrb $6, %xmm0, %eax
-; AVX1-NEXT:    movl %eax, %ecx
-; AVX1-NEXT:    addb %dl, %cl
-; AVX1-NEXT:    setns %cl
-; AVX1-NEXT:    addb %dl, %al
-; AVX1-NEXT:    jno .LBB2_121
-; AVX1-NEXT:  # %bb.120:
-; AVX1-NEXT:    addb $127, %cl
-; AVX1-NEXT:    movl %ecx, %eax
-; AVX1-NEXT:  .LBB2_121:
-; AVX1-NEXT:    vpextrb $5, %xmm1, %ebx
-; AVX1-NEXT:    vpextrb $5, %xmm0, %ecx
-; AVX1-NEXT:    movl %ecx, %edx
-; AVX1-NEXT:    addb %bl, %dl
-; AVX1-NEXT:    setns %dl
-; AVX1-NEXT:    addb %bl, %cl
-; AVX1-NEXT:    jno .LBB2_123
-; AVX1-NEXT:  # %bb.122:
-; AVX1-NEXT:    addb $127, %dl
-; AVX1-NEXT:    movl %edx, %ecx
-; AVX1-NEXT:  .LBB2_123:
-; AVX1-NEXT:    vpextrb $4, %xmm1, %esi
-; AVX1-NEXT:    vpextrb $4, %xmm0, %edx
-; AVX1-NEXT:    movl %edx, %ebx
-; AVX1-NEXT:    addb %sil, %bl
-; AVX1-NEXT:    setns %bl
-; AVX1-NEXT:    addb %sil, %dl
-; AVX1-NEXT:    jno .LBB2_125
-; AVX1-NEXT:  # %bb.124:
-; AVX1-NEXT:    addb $127, %bl
-; AVX1-NEXT:    movl %ebx, %edx
-; AVX1-NEXT:  .LBB2_125:
-; AVX1-NEXT:    vpextrb $3, %xmm1, %esi
-; AVX1-NEXT:    vpextrb $3, %xmm0, %r8d
-; AVX1-NEXT:    movl %r8d, %ebx
-; AVX1-NEXT:    addb %sil, %bl
-; AVX1-NEXT:    setns %bl
-; AVX1-NEXT:    addb %sil, %r8b
-; AVX1-NEXT:    movl %r9d, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX1-NEXT:    jno .LBB2_127
-; AVX1-NEXT:  # %bb.126:
-; AVX1-NEXT:    addb $127, %bl
-; AVX1-NEXT:    movl %ebx, %r8d
-; AVX1-NEXT:  .LBB2_127:
-; AVX1-NEXT:    vpextrb $2, %xmm1, %esi
-; AVX1-NEXT:    vpextrb $2, %xmm0, %r9d
-; AVX1-NEXT:    movl %r9d, %ebx
-; AVX1-NEXT:    addb %sil, %bl
-; AVX1-NEXT:    setns %bl
-; AVX1-NEXT:    addb %sil, %r9b
-; AVX1-NEXT:    jno .LBB2_129
-; AVX1-NEXT:  # %bb.128:
-; AVX1-NEXT:    addb $127, %bl
-; AVX1-NEXT:    movl %ebx, %r9d
-; AVX1-NEXT:  .LBB2_129:
-; AVX1-NEXT:    vpextrb $0, %xmm1, %esi
-; AVX1-NEXT:    vpextrb $0, %xmm0, %r10d
-; AVX1-NEXT:    movl %r10d, %ebx
-; AVX1-NEXT:    addb %sil, %bl
-; AVX1-NEXT:    setns %bl
-; AVX1-NEXT:    movl %r11d, %r12d
-; AVX1-NEXT:    addb %sil, %r10b
-; AVX1-NEXT:    jno .LBB2_131
-; AVX1-NEXT:  # %bb.130:
-; AVX1-NEXT:    addb $127, %bl
-; AVX1-NEXT:    movl %ebx, %r10d
-; AVX1-NEXT:  .LBB2_131:
-; AVX1-NEXT:    vpextrb $1, %xmm1, %esi
-; AVX1-NEXT:    vpextrb $1, %xmm0, %r11d
-; AVX1-NEXT:    movl %r11d, %ebx
-; AVX1-NEXT:    addb %sil, %bl
-; AVX1-NEXT:    setns %bl
-; AVX1-NEXT:    addb %sil, %r11b
-; AVX1-NEXT:    jno .LBB2_133
-; AVX1-NEXT:  # %bb.132:
-; AVX1-NEXT:    addb $127, %bl
-; AVX1-NEXT:    movl %ebx, %r11d
-; AVX1-NEXT:  .LBB2_133:
-; AVX1-NEXT:    movzbl %r10b, %esi
-; AVX1-NEXT:    vmovd %esi, %xmm0
-; AVX1-NEXT:    movzbl %r11b, %esi
-; AVX1-NEXT:    vpinsrb $1, %esi, %xmm0, %xmm0
-; AVX1-NEXT:    movzbl %r9b, %esi
-; AVX1-NEXT:    vpinsrb $2, %esi, %xmm0, %xmm0
-; AVX1-NEXT:    movzbl %r8b, %esi
-; AVX1-NEXT:    vpinsrb $3, %esi, %xmm0, %xmm0
-; AVX1-NEXT:    movzbl %dl, %edx
-; AVX1-NEXT:    vpinsrb $4, %edx, %xmm0, %xmm0
-; AVX1-NEXT:    movzbl %cl, %ecx
-; AVX1-NEXT:    vpinsrb $5, %ecx, %xmm0, %xmm0
-; AVX1-NEXT:    movzbl %al, %eax
-; AVX1-NEXT:    vpinsrb $6, %eax, %xmm0, %xmm0
-; AVX1-NEXT:    movzbl %dil, %eax
-; AVX1-NEXT:    vpinsrb $7, %eax, %xmm0, %xmm0
-; AVX1-NEXT:    movzbl %bpl, %eax
-; AVX1-NEXT:    vpinsrb $8, %eax, %xmm0, %xmm0
-; AVX1-NEXT:    movzbl %r14b, %eax
-; AVX1-NEXT:    vpinsrb $9, %eax, %xmm0, %xmm0
-; AVX1-NEXT:    movzbl %r15b, %eax
-; AVX1-NEXT:    vpinsrb $10, %eax, %xmm0, %xmm0
-; AVX1-NEXT:    movzbl %r13b, %eax
-; AVX1-NEXT:    vpinsrb $11, %eax, %xmm0, %xmm0
-; AVX1-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX1-NEXT:    vpinsrb $12, %eax, %xmm0, %xmm0
-; AVX1-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX1-NEXT:    vpinsrb $13, %eax, %xmm0, %xmm0
-; AVX1-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX1-NEXT:    vpinsrb $14, %eax, %xmm0, %xmm0
-; AVX1-NEXT:    movzbl %r12b, %eax
-; AVX1-NEXT:    vpinsrb $15, %eax, %xmm0, %xmm0
-; AVX1-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX1-NEXT:    vmovd %eax, %xmm1
-; AVX1-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX1-NEXT:    vpinsrb $1, %eax, %xmm1, %xmm1
-; AVX1-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX1-NEXT:    vpinsrb $2, %eax, %xmm1, %xmm1
-; AVX1-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX1-NEXT:    vpinsrb $3, %eax, %xmm1, %xmm1
-; AVX1-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX1-NEXT:    vpinsrb $4, %eax, %xmm1, %xmm1
-; AVX1-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX1-NEXT:    vpinsrb $5, %eax, %xmm1, %xmm1
-; AVX1-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX1-NEXT:    vpinsrb $6, %eax, %xmm1, %xmm1
-; AVX1-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX1-NEXT:    vpinsrb $7, %eax, %xmm1, %xmm1
-; AVX1-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX1-NEXT:    vpinsrb $8, %eax, %xmm1, %xmm1
-; AVX1-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX1-NEXT:    vpinsrb $9, %eax, %xmm1, %xmm1
-; AVX1-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX1-NEXT:    vpinsrb $10, %eax, %xmm1, %xmm1
-; AVX1-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX1-NEXT:    vpinsrb $11, %eax, %xmm1, %xmm1
-; AVX1-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX1-NEXT:    vpinsrb $12, %eax, %xmm1, %xmm1
-; AVX1-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX1-NEXT:    vpinsrb $13, %eax, %xmm1, %xmm1
-; AVX1-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX1-NEXT:    vpinsrb $14, %eax, %xmm1, %xmm1
-; AVX1-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX1-NEXT:    vpinsrb $15, %eax, %xmm1, %xmm1
-; AVX1-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX1-NEXT:    vmovd %eax, %xmm2
-; AVX1-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX1-NEXT:    vpinsrb $1, %eax, %xmm2, %xmm2
-; AVX1-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX1-NEXT:    vpinsrb $2, %eax, %xmm2, %xmm2
-; AVX1-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX1-NEXT:    vpinsrb $3, %eax, %xmm2, %xmm2
-; AVX1-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX1-NEXT:    vpinsrb $4, %eax, %xmm2, %xmm2
-; AVX1-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX1-NEXT:    vpinsrb $5, %eax, %xmm2, %xmm2
-; AVX1-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX1-NEXT:    vpinsrb $6, %eax, %xmm2, %xmm2
-; AVX1-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX1-NEXT:    vpinsrb $7, %eax, %xmm2, %xmm2
-; AVX1-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX1-NEXT:    vpinsrb $8, %eax, %xmm2, %xmm2
-; AVX1-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX1-NEXT:    vpinsrb $9, %eax, %xmm2, %xmm2
-; AVX1-NEXT:    movzbl (%rsp), %eax # 1-byte Folded Reload
-; AVX1-NEXT:    vpinsrb $10, %eax, %xmm2, %xmm2
-; AVX1-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX1-NEXT:    vpinsrb $11, %eax, %xmm2, %xmm2
-; AVX1-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX1-NEXT:    vpinsrb $12, %eax, %xmm2, %xmm2
-; AVX1-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX1-NEXT:    vpinsrb $13, %eax, %xmm2, %xmm2
-; AVX1-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX1-NEXT:    vpinsrb $14, %eax, %xmm2, %xmm2
-; AVX1-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX1-NEXT:    vpinsrb $15, %eax, %xmm2, %xmm2
-; AVX1-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX1-NEXT:    vmovd %eax, %xmm3
-; AVX1-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX1-NEXT:    vpinsrb $1, %eax, %xmm3, %xmm3
-; AVX1-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX1-NEXT:    vpinsrb $2, %eax, %xmm3, %xmm3
-; AVX1-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX1-NEXT:    vpinsrb $3, %eax, %xmm3, %xmm3
-; AVX1-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX1-NEXT:    vpinsrb $4, %eax, %xmm3, %xmm3
-; AVX1-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX1-NEXT:    vpinsrb $5, %eax, %xmm3, %xmm3
-; AVX1-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX1-NEXT:    vpinsrb $6, %eax, %xmm3, %xmm3
-; AVX1-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX1-NEXT:    vpinsrb $7, %eax, %xmm3, %xmm3
-; AVX1-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX1-NEXT:    vpinsrb $8, %eax, %xmm3, %xmm3
-; AVX1-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX1-NEXT:    vpinsrb $9, %eax, %xmm3, %xmm3
-; AVX1-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX1-NEXT:    vpinsrb $10, %eax, %xmm3, %xmm3
-; AVX1-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX1-NEXT:    vpinsrb $11, %eax, %xmm3, %xmm3
-; AVX1-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX1-NEXT:    vpinsrb $12, %eax, %xmm3, %xmm3
-; AVX1-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX1-NEXT:    vpinsrb $13, %eax, %xmm3, %xmm3
-; AVX1-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX1-NEXT:    vpinsrb $14, %eax, %xmm3, %xmm3
-; AVX1-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX1-NEXT:    vpinsrb $15, %eax, %xmm3, %xmm3
-; AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm1, %ymm0
-; AVX1-NEXT:    vinsertf128 $1, %xmm2, %ymm3, %ymm1
-; AVX1-NEXT:    addq $76, %rsp
-; AVX1-NEXT:    popq %rbx
-; AVX1-NEXT:    popq %r12
-; AVX1-NEXT:    popq %r13
-; AVX1-NEXT:    popq %r14
-; AVX1-NEXT:    popq %r15
-; AVX1-NEXT:    popq %rbp
+; AVX1-NEXT:    vextractf128 $1, %ymm2, %xmm4
+; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm5
+; AVX1-NEXT:    vpaddsb %xmm4, %xmm5, %xmm4
+; AVX1-NEXT:    vpaddsb %xmm2, %xmm0, %xmm0
+; AVX1-NEXT:    vinsertf128 $1, %xmm4, %ymm0, %ymm0
+; AVX1-NEXT:    vextractf128 $1, %ymm3, %xmm2
+; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm4
+; AVX1-NEXT:    vpaddsb %xmm2, %xmm4, %xmm2
+; AVX1-NEXT:    vpaddsb %xmm3, %xmm1, %xmm1
+; AVX1-NEXT:    vinsertf128 $1, %xmm2, %ymm1, %ymm1
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: v64i8:
 ; AVX2:       # %bb.0:
-; AVX2-NEXT:    pushq %rbp
-; AVX2-NEXT:    pushq %r15
-; AVX2-NEXT:    pushq %r14
-; AVX2-NEXT:    pushq %r13
-; AVX2-NEXT:    pushq %r12
-; AVX2-NEXT:    pushq %rbx
-; AVX2-NEXT:    subq $76, %rsp
-; AVX2-NEXT:    vpextrb $15, %xmm3, %ecx
-; AVX2-NEXT:    vpextrb $15, %xmm1, %edx
-; AVX2-NEXT:    movl %edx, %eax
-; AVX2-NEXT:    addb %cl, %al
-; AVX2-NEXT:    setns %al
-; AVX2-NEXT:    addb %cl, %dl
-; AVX2-NEXT:    jo .LBB2_1
-; AVX2-NEXT:  # %bb.2:
-; AVX2-NEXT:    movl %edx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX2-NEXT:    jmp .LBB2_3
-; AVX2-NEXT:  .LBB2_1:
-; AVX2-NEXT:    addb $127, %al
-; AVX2-NEXT:    # kill: def $al killed $al def $eax
-; AVX2-NEXT:    movl %eax, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX2-NEXT:  .LBB2_3:
-; AVX2-NEXT:    vpextrb $14, %xmm3, %ecx
-; AVX2-NEXT:    vpextrb $14, %xmm1, %edx
-; AVX2-NEXT:    movl %edx, %eax
-; AVX2-NEXT:    addb %cl, %al
-; AVX2-NEXT:    setns %al
-; AVX2-NEXT:    addb %cl, %dl
-; AVX2-NEXT:    jno .LBB2_5
-; AVX2-NEXT:  # %bb.4:
-; AVX2-NEXT:    addb $127, %al
-; AVX2-NEXT:    movl %eax, %edx
-; AVX2-NEXT:  .LBB2_5:
-; AVX2-NEXT:    vpextrb $13, %xmm3, %ecx
-; AVX2-NEXT:    vpextrb $13, %xmm1, %esi
-; AVX2-NEXT:    movl %esi, %eax
-; AVX2-NEXT:    addb %cl, %al
-; AVX2-NEXT:    setns %al
-; AVX2-NEXT:    addb %cl, %sil
-; AVX2-NEXT:    jo .LBB2_6
-; AVX2-NEXT:  # %bb.7:
-; AVX2-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX2-NEXT:    jmp .LBB2_8
-; AVX2-NEXT:  .LBB2_6:
-; AVX2-NEXT:    addb $127, %al
-; AVX2-NEXT:    # kill: def $al killed $al def $eax
-; AVX2-NEXT:    movl %eax, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX2-NEXT:  .LBB2_8:
-; AVX2-NEXT:    vpextrb $12, %xmm3, %ecx
-; AVX2-NEXT:    vpextrb $12, %xmm1, %esi
-; AVX2-NEXT:    movl %esi, %eax
-; AVX2-NEXT:    addb %cl, %al
-; AVX2-NEXT:    setns %al
-; AVX2-NEXT:    addb %cl, %sil
-; AVX2-NEXT:    jno .LBB2_10
-; AVX2-NEXT:  # %bb.9:
-; AVX2-NEXT:    addb $127, %al
-; AVX2-NEXT:    movl %eax, %esi
-; AVX2-NEXT:  .LBB2_10:
-; AVX2-NEXT:    vpextrb $11, %xmm3, %ecx
-; AVX2-NEXT:    vpextrb $11, %xmm1, %edi
-; AVX2-NEXT:    movl %edi, %eax
-; AVX2-NEXT:    addb %cl, %al
-; AVX2-NEXT:    setns %al
-; AVX2-NEXT:    addb %cl, %dil
-; AVX2-NEXT:    jno .LBB2_12
-; AVX2-NEXT:  # %bb.11:
-; AVX2-NEXT:    addb $127, %al
-; AVX2-NEXT:    movl %eax, %edi
-; AVX2-NEXT:  .LBB2_12:
-; AVX2-NEXT:    vpextrb $10, %xmm3, %ecx
-; AVX2-NEXT:    vpextrb $10, %xmm1, %ebp
-; AVX2-NEXT:    movl %ebp, %eax
-; AVX2-NEXT:    addb %cl, %al
-; AVX2-NEXT:    setns %al
-; AVX2-NEXT:    addb %cl, %bpl
-; AVX2-NEXT:    jno .LBB2_14
-; AVX2-NEXT:  # %bb.13:
-; AVX2-NEXT:    addb $127, %al
-; AVX2-NEXT:    movl %eax, %ebp
-; AVX2-NEXT:  .LBB2_14:
-; AVX2-NEXT:    vpextrb $9, %xmm3, %ecx
-; AVX2-NEXT:    vpextrb $9, %xmm1, %ebx
-; AVX2-NEXT:    movl %ebx, %eax
-; AVX2-NEXT:    addb %cl, %al
-; AVX2-NEXT:    setns %al
-; AVX2-NEXT:    addb %cl, %bl
-; AVX2-NEXT:    movl %ebp, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX2-NEXT:    jo .LBB2_15
-; AVX2-NEXT:  # %bb.16:
-; AVX2-NEXT:    movl %ebx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX2-NEXT:    jmp .LBB2_17
-; AVX2-NEXT:  .LBB2_15:
-; AVX2-NEXT:    addb $127, %al
-; AVX2-NEXT:    # kill: def $al killed $al def $eax
-; AVX2-NEXT:    movl %eax, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX2-NEXT:  .LBB2_17:
-; AVX2-NEXT:    vpextrb $8, %xmm3, %ecx
-; AVX2-NEXT:    vpextrb $8, %xmm1, %ebp
-; AVX2-NEXT:    movl %ebp, %eax
-; AVX2-NEXT:    addb %cl, %al
-; AVX2-NEXT:    setns %al
-; AVX2-NEXT:    addb %cl, %bpl
-; AVX2-NEXT:    jno .LBB2_19
-; AVX2-NEXT:  # %bb.18:
-; AVX2-NEXT:    addb $127, %al
-; AVX2-NEXT:    movl %eax, %ebp
-; AVX2-NEXT:  .LBB2_19:
-; AVX2-NEXT:    vpextrb $7, %xmm3, %ecx
-; AVX2-NEXT:    vpextrb $7, %xmm1, %ebx
-; AVX2-NEXT:    movl %ebx, %eax
-; AVX2-NEXT:    addb %cl, %al
-; AVX2-NEXT:    setns %al
-; AVX2-NEXT:    addb %cl, %bl
-; AVX2-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX2-NEXT:    jno .LBB2_21
-; AVX2-NEXT:  # %bb.20:
-; AVX2-NEXT:    addb $127, %al
-; AVX2-NEXT:    movl %eax, %ebx
-; AVX2-NEXT:  .LBB2_21:
-; AVX2-NEXT:    vpextrb $6, %xmm3, %ecx
-; AVX2-NEXT:    vpextrb $6, %xmm1, %esi
-; AVX2-NEXT:    movl %esi, %eax
-; AVX2-NEXT:    addb %cl, %al
-; AVX2-NEXT:    setns %al
-; AVX2-NEXT:    addb %cl, %sil
-; AVX2-NEXT:    movl %edx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX2-NEXT:    jno .LBB2_23
-; AVX2-NEXT:  # %bb.22:
-; AVX2-NEXT:    addb $127, %al
-; AVX2-NEXT:    movl %eax, %esi
-; AVX2-NEXT:  .LBB2_23:
-; AVX2-NEXT:    vpextrb $5, %xmm3, %ecx
-; AVX2-NEXT:    vpextrb $5, %xmm1, %r11d
-; AVX2-NEXT:    movl %r11d, %eax
-; AVX2-NEXT:    addb %cl, %al
-; AVX2-NEXT:    setns %al
-; AVX2-NEXT:    addb %cl, %r11b
-; AVX2-NEXT:    movl %edi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX2-NEXT:    movl %ebp, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX2-NEXT:    movl %ebx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX2-NEXT:    jno .LBB2_25
-; AVX2-NEXT:  # %bb.24:
-; AVX2-NEXT:    addb $127, %al
-; AVX2-NEXT:    movl %eax, %r11d
-; AVX2-NEXT:  .LBB2_25:
-; AVX2-NEXT:    vpextrb $4, %xmm3, %ecx
-; AVX2-NEXT:    vpextrb $4, %xmm1, %r13d
-; AVX2-NEXT:    movl %r13d, %eax
-; AVX2-NEXT:    addb %cl, %al
-; AVX2-NEXT:    setns %al
-; AVX2-NEXT:    addb %cl, %r13b
-; AVX2-NEXT:    jno .LBB2_27
-; AVX2-NEXT:  # %bb.26:
-; AVX2-NEXT:    addb $127, %al
-; AVX2-NEXT:    movl %eax, %r13d
-; AVX2-NEXT:  .LBB2_27:
-; AVX2-NEXT:    vpextrb $3, %xmm3, %ecx
-; AVX2-NEXT:    vpextrb $3, %xmm1, %r8d
-; AVX2-NEXT:    movl %r8d, %eax
-; AVX2-NEXT:    addb %cl, %al
-; AVX2-NEXT:    setns %al
-; AVX2-NEXT:    addb %cl, %r8b
-; AVX2-NEXT:    jno .LBB2_29
-; AVX2-NEXT:  # %bb.28:
-; AVX2-NEXT:    addb $127, %al
-; AVX2-NEXT:    movl %eax, %r8d
-; AVX2-NEXT:  .LBB2_29:
-; AVX2-NEXT:    vpextrb $2, %xmm3, %ecx
-; AVX2-NEXT:    vpextrb $2, %xmm1, %ebp
-; AVX2-NEXT:    movl %ebp, %eax
-; AVX2-NEXT:    addb %cl, %al
-; AVX2-NEXT:    setns %al
-; AVX2-NEXT:    addb %cl, %bpl
-; AVX2-NEXT:    jno .LBB2_31
-; AVX2-NEXT:  # %bb.30:
-; AVX2-NEXT:    addb $127, %al
-; AVX2-NEXT:    movl %eax, %ebp
-; AVX2-NEXT:  .LBB2_31:
-; AVX2-NEXT:    vpextrb $0, %xmm3, %ecx
-; AVX2-NEXT:    vpextrb $0, %xmm1, %ebx
-; AVX2-NEXT:    movl %ebx, %eax
-; AVX2-NEXT:    addb %cl, %al
-; AVX2-NEXT:    setns %al
-; AVX2-NEXT:    addb %cl, %bl
-; AVX2-NEXT:    jno .LBB2_33
-; AVX2-NEXT:  # %bb.32:
-; AVX2-NEXT:    addb $127, %al
-; AVX2-NEXT:    movl %eax, %ebx
-; AVX2-NEXT:  .LBB2_33:
-; AVX2-NEXT:    vpextrb $1, %xmm3, %ecx
-; AVX2-NEXT:    vpextrb $1, %xmm1, %edi
-; AVX2-NEXT:    movl %edi, %eax
-; AVX2-NEXT:    addb %cl, %al
-; AVX2-NEXT:    setns %al
-; AVX2-NEXT:    addb %cl, %dil
-; AVX2-NEXT:    jno .LBB2_35
-; AVX2-NEXT:  # %bb.34:
-; AVX2-NEXT:    addb $127, %al
-; AVX2-NEXT:    movl %eax, %edi
-; AVX2-NEXT:  .LBB2_35:
-; AVX2-NEXT:    vextracti128 $1, %ymm3, %xmm3
-; AVX2-NEXT:    vpextrb $15, %xmm3, %ecx
-; AVX2-NEXT:    vextracti128 $1, %ymm1, %xmm1
-; AVX2-NEXT:    vpextrb $15, %xmm1, %edx
-; AVX2-NEXT:    movl %edx, %eax
-; AVX2-NEXT:    addb %cl, %al
-; AVX2-NEXT:    setns %al
-; AVX2-NEXT:    addb %cl, %dl
-; AVX2-NEXT:    movl %ebp, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX2-NEXT:    movl %ebx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX2-NEXT:    movl %edi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX2-NEXT:    jno .LBB2_37
-; AVX2-NEXT:  # %bb.36:
-; AVX2-NEXT:    addb $127, %al
-; AVX2-NEXT:    movl %eax, %edx
-; AVX2-NEXT:  .LBB2_37:
-; AVX2-NEXT:    vpextrb $14, %xmm3, %ecx
-; AVX2-NEXT:    vpextrb $14, %xmm1, %edi
-; AVX2-NEXT:    movl %edi, %eax
-; AVX2-NEXT:    addb %cl, %al
-; AVX2-NEXT:    setns %al
-; AVX2-NEXT:    addb %cl, %dil
-; AVX2-NEXT:    movl %edx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX2-NEXT:    jno .LBB2_39
-; AVX2-NEXT:  # %bb.38:
-; AVX2-NEXT:    addb $127, %al
-; AVX2-NEXT:    movl %eax, %edi
-; AVX2-NEXT:  .LBB2_39:
-; AVX2-NEXT:    vpextrb $13, %xmm3, %ecx
-; AVX2-NEXT:    vpextrb $13, %xmm1, %r12d
-; AVX2-NEXT:    movl %r12d, %eax
-; AVX2-NEXT:    addb %cl, %al
-; AVX2-NEXT:    setns %al
-; AVX2-NEXT:    addb %cl, %r12b
-; AVX2-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX2-NEXT:    jno .LBB2_41
-; AVX2-NEXT:  # %bb.40:
-; AVX2-NEXT:    addb $127, %al
-; AVX2-NEXT:    movl %eax, %r12d
-; AVX2-NEXT:  .LBB2_41:
-; AVX2-NEXT:    vpextrb $12, %xmm3, %ecx
-; AVX2-NEXT:    vpextrb $12, %xmm1, %r15d
-; AVX2-NEXT:    movl %r15d, %eax
-; AVX2-NEXT:    addb %cl, %al
-; AVX2-NEXT:    setns %al
-; AVX2-NEXT:    addb %cl, %r15b
-; AVX2-NEXT:    movl %r11d, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX2-NEXT:    jno .LBB2_43
-; AVX2-NEXT:  # %bb.42:
-; AVX2-NEXT:    addb $127, %al
-; AVX2-NEXT:    movl %eax, %r15d
-; AVX2-NEXT:  .LBB2_43:
-; AVX2-NEXT:    vpextrb $11, %xmm3, %ecx
-; AVX2-NEXT:    vpextrb $11, %xmm1, %ebp
-; AVX2-NEXT:    movl %ebp, %eax
-; AVX2-NEXT:    addb %cl, %al
-; AVX2-NEXT:    setns %al
-; AVX2-NEXT:    addb %cl, %bpl
-; AVX2-NEXT:    movl %r13d, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX2-NEXT:    jno .LBB2_45
-; AVX2-NEXT:  # %bb.44:
-; AVX2-NEXT:    addb $127, %al
-; AVX2-NEXT:    movl %eax, %ebp
-; AVX2-NEXT:  .LBB2_45:
-; AVX2-NEXT:    vpextrb $10, %xmm3, %ecx
-; AVX2-NEXT:    vpextrb $10, %xmm1, %ebx
-; AVX2-NEXT:    movl %ebx, %eax
-; AVX2-NEXT:    addb %cl, %al
-; AVX2-NEXT:    setns %al
-; AVX2-NEXT:    addb %cl, %bl
-; AVX2-NEXT:    jno .LBB2_47
-; AVX2-NEXT:  # %bb.46:
-; AVX2-NEXT:    addb $127, %al
-; AVX2-NEXT:    movl %eax, %ebx
-; AVX2-NEXT:  .LBB2_47:
-; AVX2-NEXT:    vpextrb $9, %xmm3, %ecx
-; AVX2-NEXT:    vpextrb $9, %xmm1, %edx
-; AVX2-NEXT:    movl %edx, %eax
-; AVX2-NEXT:    addb %cl, %al
-; AVX2-NEXT:    setns %al
-; AVX2-NEXT:    addb %cl, %dl
-; AVX2-NEXT:    jno .LBB2_49
-; AVX2-NEXT:  # %bb.48:
-; AVX2-NEXT:    addb $127, %al
-; AVX2-NEXT:    movl %eax, %edx
-; AVX2-NEXT:  .LBB2_49:
-; AVX2-NEXT:    vpextrb $8, %xmm3, %ecx
-; AVX2-NEXT:    vpextrb $8, %xmm1, %esi
-; AVX2-NEXT:    movl %esi, %eax
-; AVX2-NEXT:    addb %cl, %al
-; AVX2-NEXT:    setns %al
-; AVX2-NEXT:    addb %cl, %sil
-; AVX2-NEXT:    movl %ebp, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX2-NEXT:    movl %ebx, (%rsp) # 4-byte Spill
-; AVX2-NEXT:    movl %edx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX2-NEXT:    jo .LBB2_50
-; AVX2-NEXT:  # %bb.51:
-; AVX2-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX2-NEXT:    jmp .LBB2_52
-; AVX2-NEXT:  .LBB2_50:
-; AVX2-NEXT:    addb $127, %al
-; AVX2-NEXT:    # kill: def $al killed $al def $eax
-; AVX2-NEXT:    movl %eax, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX2-NEXT:  .LBB2_52:
-; AVX2-NEXT:    vpextrb $7, %xmm3, %ecx
-; AVX2-NEXT:    vpextrb $7, %xmm1, %r11d
-; AVX2-NEXT:    movl %r11d, %eax
-; AVX2-NEXT:    addb %cl, %al
-; AVX2-NEXT:    setns %al
-; AVX2-NEXT:    addb %cl, %r11b
-; AVX2-NEXT:    jno .LBB2_54
-; AVX2-NEXT:  # %bb.53:
-; AVX2-NEXT:    addb $127, %al
-; AVX2-NEXT:    movl %eax, %r11d
-; AVX2-NEXT:  .LBB2_54:
-; AVX2-NEXT:    vpextrb $6, %xmm3, %ecx
-; AVX2-NEXT:    vpextrb $6, %xmm1, %esi
-; AVX2-NEXT:    movl %esi, %eax
-; AVX2-NEXT:    addb %cl, %al
-; AVX2-NEXT:    setns %al
-; AVX2-NEXT:    addb %cl, %sil
-; AVX2-NEXT:    movl %r8d, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX2-NEXT:    movl %edi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX2-NEXT:    jno .LBB2_56
-; AVX2-NEXT:  # %bb.55:
-; AVX2-NEXT:    addb $127, %al
-; AVX2-NEXT:    movl %eax, %esi
-; AVX2-NEXT:  .LBB2_56:
-; AVX2-NEXT:    vpextrb $5, %xmm3, %ecx
-; AVX2-NEXT:    vpextrb $5, %xmm1, %edi
-; AVX2-NEXT:    movl %edi, %eax
-; AVX2-NEXT:    addb %cl, %al
-; AVX2-NEXT:    setns %al
-; AVX2-NEXT:    addb %cl, %dil
-; AVX2-NEXT:    jno .LBB2_58
-; AVX2-NEXT:  # %bb.57:
-; AVX2-NEXT:    addb $127, %al
-; AVX2-NEXT:    movl %eax, %edi
-; AVX2-NEXT:  .LBB2_58:
-; AVX2-NEXT:    vpextrb $4, %xmm3, %ecx
-; AVX2-NEXT:    vpextrb $4, %xmm1, %r13d
-; AVX2-NEXT:    movl %r13d, %eax
-; AVX2-NEXT:    addb %cl, %al
-; AVX2-NEXT:    setns %al
-; AVX2-NEXT:    addb %cl, %r13b
-; AVX2-NEXT:    movl %r13d, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX2-NEXT:    jno .LBB2_60
-; AVX2-NEXT:  # %bb.59:
-; AVX2-NEXT:    addb $127, %al
-; AVX2-NEXT:    # kill: def $al killed $al def $eax
-; AVX2-NEXT:    movl %eax, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX2-NEXT:  .LBB2_60:
-; AVX2-NEXT:    vpextrb $3, %xmm3, %ecx
-; AVX2-NEXT:    vpextrb $3, %xmm1, %ebp
-; AVX2-NEXT:    movl %ebp, %eax
-; AVX2-NEXT:    addb %cl, %al
-; AVX2-NEXT:    setns %al
-; AVX2-NEXT:    addb %cl, %bpl
-; AVX2-NEXT:    jo .LBB2_61
-; AVX2-NEXT:  # %bb.62:
-; AVX2-NEXT:    movl %ebp, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX2-NEXT:    jmp .LBB2_63
-; AVX2-NEXT:  .LBB2_61:
-; AVX2-NEXT:    addb $127, %al
-; AVX2-NEXT:    # kill: def $al killed $al def $eax
-; AVX2-NEXT:    movl %eax, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX2-NEXT:  .LBB2_63:
-; AVX2-NEXT:    vpextrb $2, %xmm3, %ecx
-; AVX2-NEXT:    vpextrb $2, %xmm1, %ebp
-; AVX2-NEXT:    movl %ebp, %eax
-; AVX2-NEXT:    addb %cl, %al
-; AVX2-NEXT:    setns %al
-; AVX2-NEXT:    addb %cl, %bpl
-; AVX2-NEXT:    movl %edi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX2-NEXT:    jno .LBB2_65
-; AVX2-NEXT:  # %bb.64:
-; AVX2-NEXT:    addb $127, %al
-; AVX2-NEXT:    movl %eax, %ebp
-; AVX2-NEXT:  .LBB2_65:
-; AVX2-NEXT:    vpextrb $0, %xmm3, %ecx
-; AVX2-NEXT:    vpextrb $0, %xmm1, %edi
-; AVX2-NEXT:    movl %edi, %eax
-; AVX2-NEXT:    addb %cl, %al
-; AVX2-NEXT:    setns %al
-; AVX2-NEXT:    addb %cl, %dil
-; AVX2-NEXT:    movl %r15d, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX2-NEXT:    jno .LBB2_67
-; AVX2-NEXT:  # %bb.66:
-; AVX2-NEXT:    addb $127, %al
-; AVX2-NEXT:    movl %eax, %edi
-; AVX2-NEXT:  .LBB2_67:
-; AVX2-NEXT:    vpextrb $1, %xmm3, %ecx
-; AVX2-NEXT:    vpextrb $1, %xmm1, %ebx
-; AVX2-NEXT:    movl %ebx, %eax
-; AVX2-NEXT:    addb %cl, %al
-; AVX2-NEXT:    setns %al
-; AVX2-NEXT:    addb %cl, %bl
-; AVX2-NEXT:    movl %r12d, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX2-NEXT:    jno .LBB2_69
-; AVX2-NEXT:  # %bb.68:
-; AVX2-NEXT:    addb $127, %al
-; AVX2-NEXT:    movl %eax, %ebx
-; AVX2-NEXT:  .LBB2_69:
-; AVX2-NEXT:    vpextrb $15, %xmm2, %ecx
-; AVX2-NEXT:    vpextrb $15, %xmm0, %edx
-; AVX2-NEXT:    movl %edx, %eax
-; AVX2-NEXT:    addb %cl, %al
-; AVX2-NEXT:    setns %al
-; AVX2-NEXT:    addb %cl, %dl
-; AVX2-NEXT:    movl %edi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX2-NEXT:    jno .LBB2_71
-; AVX2-NEXT:  # %bb.70:
-; AVX2-NEXT:    addb $127, %al
-; AVX2-NEXT:    movl %eax, %edx
-; AVX2-NEXT:  .LBB2_71:
-; AVX2-NEXT:    vpextrb $14, %xmm2, %ecx
-; AVX2-NEXT:    vpextrb $14, %xmm0, %edi
-; AVX2-NEXT:    movl %edi, %eax
-; AVX2-NEXT:    addb %cl, %al
-; AVX2-NEXT:    setns %al
-; AVX2-NEXT:    addb %cl, %dil
-; AVX2-NEXT:    jno .LBB2_73
-; AVX2-NEXT:  # %bb.72:
-; AVX2-NEXT:    addb $127, %al
-; AVX2-NEXT:    movl %eax, %edi
-; AVX2-NEXT:  .LBB2_73:
-; AVX2-NEXT:    vpextrb $13, %xmm2, %ecx
-; AVX2-NEXT:    vpextrb $13, %xmm0, %r10d
-; AVX2-NEXT:    movl %r10d, %eax
-; AVX2-NEXT:    addb %cl, %al
-; AVX2-NEXT:    setns %al
-; AVX2-NEXT:    addb %cl, %r10b
-; AVX2-NEXT:    jno .LBB2_75
-; AVX2-NEXT:  # %bb.74:
-; AVX2-NEXT:    addb $127, %al
-; AVX2-NEXT:    movl %eax, %r10d
-; AVX2-NEXT:  .LBB2_75:
-; AVX2-NEXT:    vpextrb $12, %xmm2, %ecx
-; AVX2-NEXT:    vpextrb $12, %xmm0, %r12d
-; AVX2-NEXT:    movl %r12d, %eax
-; AVX2-NEXT:    addb %cl, %al
-; AVX2-NEXT:    setns %al
-; AVX2-NEXT:    addb %cl, %r12b
-; AVX2-NEXT:    movl %r12d, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX2-NEXT:    jno .LBB2_77
-; AVX2-NEXT:  # %bb.76:
-; AVX2-NEXT:    addb $127, %al
-; AVX2-NEXT:    # kill: def $al killed $al def $eax
-; AVX2-NEXT:    movl %eax, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX2-NEXT:  .LBB2_77:
-; AVX2-NEXT:    vpextrb $11, %xmm2, %ecx
-; AVX2-NEXT:    vpextrb $11, %xmm0, %r14d
-; AVX2-NEXT:    movl %r14d, %eax
-; AVX2-NEXT:    addb %cl, %al
-; AVX2-NEXT:    setns %al
-; AVX2-NEXT:    addb %cl, %r14b
-; AVX2-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX2-NEXT:    jno .LBB2_79
-; AVX2-NEXT:  # %bb.78:
-; AVX2-NEXT:    addb $127, %al
-; AVX2-NEXT:    movl %eax, %r14d
-; AVX2-NEXT:  .LBB2_79:
-; AVX2-NEXT:    vpextrb $10, %xmm2, %ecx
-; AVX2-NEXT:    vpextrb $10, %xmm0, %r13d
-; AVX2-NEXT:    movl %r13d, %eax
-; AVX2-NEXT:    addb %cl, %al
-; AVX2-NEXT:    setns %al
-; AVX2-NEXT:    addb %cl, %r13b
-; AVX2-NEXT:    movl %r11d, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX2-NEXT:    movl %ebp, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX2-NEXT:    movl %ebx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX2-NEXT:    movl %edx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX2-NEXT:    jno .LBB2_81
-; AVX2-NEXT:  # %bb.80:
-; AVX2-NEXT:    addb $127, %al
-; AVX2-NEXT:    movl %eax, %r13d
-; AVX2-NEXT:  .LBB2_81:
-; AVX2-NEXT:    vpextrb $9, %xmm2, %ecx
-; AVX2-NEXT:    vpextrb $9, %xmm0, %r8d
-; AVX2-NEXT:    movl %r8d, %eax
-; AVX2-NEXT:    addb %cl, %al
-; AVX2-NEXT:    setns %al
-; AVX2-NEXT:    addb %cl, %r8b
-; AVX2-NEXT:    jno .LBB2_83
-; AVX2-NEXT:  # %bb.82:
-; AVX2-NEXT:    addb $127, %al
-; AVX2-NEXT:    movl %eax, %r8d
-; AVX2-NEXT:  .LBB2_83:
-; AVX2-NEXT:    vpextrb $8, %xmm2, %ecx
-; AVX2-NEXT:    vpextrb $8, %xmm0, %r15d
-; AVX2-NEXT:    movl %r15d, %eax
-; AVX2-NEXT:    addb %cl, %al
-; AVX2-NEXT:    setns %al
-; AVX2-NEXT:    addb %cl, %r15b
-; AVX2-NEXT:    jno .LBB2_85
-; AVX2-NEXT:  # %bb.84:
-; AVX2-NEXT:    addb $127, %al
-; AVX2-NEXT:    movl %eax, %r15d
-; AVX2-NEXT:  .LBB2_85:
-; AVX2-NEXT:    vpextrb $7, %xmm2, %ecx
-; AVX2-NEXT:    vpextrb $7, %xmm0, %r12d
-; AVX2-NEXT:    movl %r12d, %eax
-; AVX2-NEXT:    addb %cl, %al
-; AVX2-NEXT:    setns %al
-; AVX2-NEXT:    addb %cl, %r12b
-; AVX2-NEXT:    movl %r12d, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX2-NEXT:    jno .LBB2_87
-; AVX2-NEXT:  # %bb.86:
-; AVX2-NEXT:    addb $127, %al
-; AVX2-NEXT:    # kill: def $al killed $al def $eax
-; AVX2-NEXT:    movl %eax, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX2-NEXT:  .LBB2_87:
-; AVX2-NEXT:    vpextrb $6, %xmm2, %ecx
-; AVX2-NEXT:    vpextrb $6, %xmm0, %r12d
-; AVX2-NEXT:    movl %r12d, %eax
-; AVX2-NEXT:    addb %cl, %al
-; AVX2-NEXT:    setns %al
-; AVX2-NEXT:    addb %cl, %r12b
-; AVX2-NEXT:    jno .LBB2_89
-; AVX2-NEXT:  # %bb.88:
-; AVX2-NEXT:    addb $127, %al
-; AVX2-NEXT:    movl %eax, %r12d
-; AVX2-NEXT:  .LBB2_89:
-; AVX2-NEXT:    vpextrb $5, %xmm2, %ecx
-; AVX2-NEXT:    vpextrb $5, %xmm0, %ebp
-; AVX2-NEXT:    movl %ebp, %eax
-; AVX2-NEXT:    addb %cl, %al
-; AVX2-NEXT:    setns %al
-; AVX2-NEXT:    addb %cl, %bpl
-; AVX2-NEXT:    jno .LBB2_91
-; AVX2-NEXT:  # %bb.90:
-; AVX2-NEXT:    addb $127, %al
-; AVX2-NEXT:    movl %eax, %ebp
-; AVX2-NEXT:  .LBB2_91:
-; AVX2-NEXT:    vpextrb $4, %xmm2, %ecx
-; AVX2-NEXT:    vpextrb $4, %xmm0, %esi
-; AVX2-NEXT:    movl %esi, %eax
-; AVX2-NEXT:    addb %cl, %al
-; AVX2-NEXT:    setns %al
-; AVX2-NEXT:    addb %cl, %sil
-; AVX2-NEXT:    jno .LBB2_93
-; AVX2-NEXT:  # %bb.92:
-; AVX2-NEXT:    addb $127, %al
-; AVX2-NEXT:    movl %eax, %esi
-; AVX2-NEXT:  .LBB2_93:
-; AVX2-NEXT:    vpextrb $3, %xmm2, %ecx
-; AVX2-NEXT:    vpextrb $3, %xmm0, %edx
-; AVX2-NEXT:    movl %edx, %eax
-; AVX2-NEXT:    addb %cl, %al
-; AVX2-NEXT:    setns %al
-; AVX2-NEXT:    addb %cl, %dl
-; AVX2-NEXT:    movl %edi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX2-NEXT:    jno .LBB2_95
-; AVX2-NEXT:  # %bb.94:
-; AVX2-NEXT:    addb $127, %al
-; AVX2-NEXT:    movl %eax, %edx
-; AVX2-NEXT:  .LBB2_95:
-; AVX2-NEXT:    vpextrb $2, %xmm2, %ecx
-; AVX2-NEXT:    vpextrb $2, %xmm0, %edi
-; AVX2-NEXT:    movl %edi, %eax
-; AVX2-NEXT:    addb %cl, %al
-; AVX2-NEXT:    setns %al
-; AVX2-NEXT:    addb %cl, %dil
-; AVX2-NEXT:    jno .LBB2_97
-; AVX2-NEXT:  # %bb.96:
-; AVX2-NEXT:    addb $127, %al
-; AVX2-NEXT:    movl %eax, %edi
-; AVX2-NEXT:  .LBB2_97:
-; AVX2-NEXT:    vpextrb $0, %xmm2, %ecx
-; AVX2-NEXT:    vpextrb $0, %xmm0, %ebx
-; AVX2-NEXT:    movl %ebx, %eax
-; AVX2-NEXT:    addb %cl, %al
-; AVX2-NEXT:    setns %al
-; AVX2-NEXT:    addb %cl, %bl
-; AVX2-NEXT:    jno .LBB2_99
-; AVX2-NEXT:  # %bb.98:
-; AVX2-NEXT:    addb $127, %al
-; AVX2-NEXT:    movl %eax, %ebx
-; AVX2-NEXT:  .LBB2_99:
-; AVX2-NEXT:    vpextrb $1, %xmm2, %ecx
-; AVX2-NEXT:    vpextrb $1, %xmm0, %r11d
-; AVX2-NEXT:    movl %r11d, %eax
-; AVX2-NEXT:    addb %cl, %al
-; AVX2-NEXT:    setns %al
-; AVX2-NEXT:    addb %cl, %r11b
-; AVX2-NEXT:    movl %r11d, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX2-NEXT:    jno .LBB2_101
-; AVX2-NEXT:  # %bb.100:
-; AVX2-NEXT:    addb $127, %al
-; AVX2-NEXT:    # kill: def $al killed $al def $eax
-; AVX2-NEXT:    movl %eax, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX2-NEXT:  .LBB2_101:
-; AVX2-NEXT:    vextracti128 $1, %ymm2, %xmm1
-; AVX2-NEXT:    vpextrb $15, %xmm1, %ecx
-; AVX2-NEXT:    vextracti128 $1, %ymm0, %xmm0
-; AVX2-NEXT:    vpextrb $15, %xmm0, %r11d
-; AVX2-NEXT:    movl %r11d, %eax
-; AVX2-NEXT:    addb %cl, %al
-; AVX2-NEXT:    setns %al
-; AVX2-NEXT:    addb %cl, %r11b
-; AVX2-NEXT:    jno .LBB2_103
-; AVX2-NEXT:  # %bb.102:
-; AVX2-NEXT:    addb $127, %al
-; AVX2-NEXT:    movl %eax, %r11d
-; AVX2-NEXT:  .LBB2_103:
-; AVX2-NEXT:    vpextrb $14, %xmm1, %ecx
-; AVX2-NEXT:    vpextrb $14, %xmm0, %r9d
-; AVX2-NEXT:    movl %r9d, %eax
-; AVX2-NEXT:    addb %cl, %al
-; AVX2-NEXT:    setns %al
-; AVX2-NEXT:    addb %cl, %r9b
-; AVX2-NEXT:    movl %r9d, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX2-NEXT:    jno .LBB2_105
-; AVX2-NEXT:  # %bb.104:
-; AVX2-NEXT:    addb $127, %al
-; AVX2-NEXT:    # kill: def $al killed $al def $eax
-; AVX2-NEXT:    movl %eax, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX2-NEXT:  .LBB2_105:
-; AVX2-NEXT:    vpextrb $13, %xmm1, %ecx
-; AVX2-NEXT:    vpextrb $13, %xmm0, %r9d
-; AVX2-NEXT:    movl %r9d, %eax
-; AVX2-NEXT:    addb %cl, %al
-; AVX2-NEXT:    setns %al
-; AVX2-NEXT:    addb %cl, %r9b
-; AVX2-NEXT:    movl %r14d, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX2-NEXT:    movl %r13d, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX2-NEXT:    movl %r8d, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX2-NEXT:    movl %r15d, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX2-NEXT:    movl %r12d, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX2-NEXT:    movl %ebp, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX2-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX2-NEXT:    movl %edx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX2-NEXT:    movl %edi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX2-NEXT:    movl %ebx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX2-NEXT:    jno .LBB2_107
-; AVX2-NEXT:  # %bb.106:
-; AVX2-NEXT:    addb $127, %al
-; AVX2-NEXT:    movl %eax, %r9d
-; AVX2-NEXT:  .LBB2_107:
-; AVX2-NEXT:    vpextrb $12, %xmm1, %ecx
-; AVX2-NEXT:    vpextrb $12, %xmm0, %edx
-; AVX2-NEXT:    movl %edx, %eax
-; AVX2-NEXT:    addb %cl, %al
-; AVX2-NEXT:    setns %al
-; AVX2-NEXT:    addb %cl, %dl
-; AVX2-NEXT:    movl %r10d, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX2-NEXT:    jno .LBB2_109
-; AVX2-NEXT:  # %bb.108:
-; AVX2-NEXT:    addb $127, %al
-; AVX2-NEXT:    movl %eax, %edx
-; AVX2-NEXT:  .LBB2_109:
-; AVX2-NEXT:    vpextrb $11, %xmm1, %ecx
-; AVX2-NEXT:    vpextrb $11, %xmm0, %r13d
-; AVX2-NEXT:    movl %r13d, %eax
-; AVX2-NEXT:    addb %cl, %al
-; AVX2-NEXT:    setns %al
-; AVX2-NEXT:    addb %cl, %r13b
-; AVX2-NEXT:    jno .LBB2_111
-; AVX2-NEXT:  # %bb.110:
-; AVX2-NEXT:    addb $127, %al
-; AVX2-NEXT:    movl %eax, %r13d
-; AVX2-NEXT:  .LBB2_111:
-; AVX2-NEXT:    vpextrb $10, %xmm1, %ecx
-; AVX2-NEXT:    vpextrb $10, %xmm0, %r15d
-; AVX2-NEXT:    movl %r15d, %eax
-; AVX2-NEXT:    addb %cl, %al
-; AVX2-NEXT:    setns %al
-; AVX2-NEXT:    addb %cl, %r15b
-; AVX2-NEXT:    jno .LBB2_113
-; AVX2-NEXT:  # %bb.112:
-; AVX2-NEXT:    addb $127, %al
-; AVX2-NEXT:    movl %eax, %r15d
-; AVX2-NEXT:  .LBB2_113:
-; AVX2-NEXT:    vpextrb $9, %xmm1, %ecx
-; AVX2-NEXT:    vpextrb $9, %xmm0, %r14d
-; AVX2-NEXT:    movl %r14d, %eax
-; AVX2-NEXT:    addb %cl, %al
-; AVX2-NEXT:    setns %al
-; AVX2-NEXT:    addb %cl, %r14b
-; AVX2-NEXT:    jno .LBB2_115
-; AVX2-NEXT:  # %bb.114:
-; AVX2-NEXT:    addb $127, %al
-; AVX2-NEXT:    movl %eax, %r14d
-; AVX2-NEXT:  .LBB2_115:
-; AVX2-NEXT:    vpextrb $8, %xmm1, %ecx
-; AVX2-NEXT:    vpextrb $8, %xmm0, %ebp
-; AVX2-NEXT:    movl %ebp, %eax
-; AVX2-NEXT:    addb %cl, %al
-; AVX2-NEXT:    setns %al
-; AVX2-NEXT:    addb %cl, %bpl
-; AVX2-NEXT:    jno .LBB2_117
-; AVX2-NEXT:  # %bb.116:
-; AVX2-NEXT:    addb $127, %al
-; AVX2-NEXT:    movl %eax, %ebp
-; AVX2-NEXT:  .LBB2_117:
-; AVX2-NEXT:    vpextrb $7, %xmm1, %ecx
-; AVX2-NEXT:    vpextrb $7, %xmm0, %edi
-; AVX2-NEXT:    movl %edi, %eax
-; AVX2-NEXT:    addb %cl, %al
-; AVX2-NEXT:    setns %al
-; AVX2-NEXT:    addb %cl, %dil
-; AVX2-NEXT:    movl %edx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX2-NEXT:    jno .LBB2_119
-; AVX2-NEXT:  # %bb.118:
-; AVX2-NEXT:    addb $127, %al
-; AVX2-NEXT:    movl %eax, %edi
-; AVX2-NEXT:  .LBB2_119:
-; AVX2-NEXT:    vpextrb $6, %xmm1, %edx
-; AVX2-NEXT:    vpextrb $6, %xmm0, %eax
-; AVX2-NEXT:    movl %eax, %ecx
-; AVX2-NEXT:    addb %dl, %cl
-; AVX2-NEXT:    setns %cl
-; AVX2-NEXT:    addb %dl, %al
-; AVX2-NEXT:    jno .LBB2_121
-; AVX2-NEXT:  # %bb.120:
-; AVX2-NEXT:    addb $127, %cl
-; AVX2-NEXT:    movl %ecx, %eax
-; AVX2-NEXT:  .LBB2_121:
-; AVX2-NEXT:    vpextrb $5, %xmm1, %ebx
-; AVX2-NEXT:    vpextrb $5, %xmm0, %ecx
-; AVX2-NEXT:    movl %ecx, %edx
-; AVX2-NEXT:    addb %bl, %dl
-; AVX2-NEXT:    setns %dl
-; AVX2-NEXT:    addb %bl, %cl
-; AVX2-NEXT:    jno .LBB2_123
-; AVX2-NEXT:  # %bb.122:
-; AVX2-NEXT:    addb $127, %dl
-; AVX2-NEXT:    movl %edx, %ecx
-; AVX2-NEXT:  .LBB2_123:
-; AVX2-NEXT:    vpextrb $4, %xmm1, %esi
-; AVX2-NEXT:    vpextrb $4, %xmm0, %edx
-; AVX2-NEXT:    movl %edx, %ebx
-; AVX2-NEXT:    addb %sil, %bl
-; AVX2-NEXT:    setns %bl
-; AVX2-NEXT:    addb %sil, %dl
-; AVX2-NEXT:    jno .LBB2_125
-; AVX2-NEXT:  # %bb.124:
-; AVX2-NEXT:    addb $127, %bl
-; AVX2-NEXT:    movl %ebx, %edx
-; AVX2-NEXT:  .LBB2_125:
-; AVX2-NEXT:    vpextrb $3, %xmm1, %esi
-; AVX2-NEXT:    vpextrb $3, %xmm0, %r8d
-; AVX2-NEXT:    movl %r8d, %ebx
-; AVX2-NEXT:    addb %sil, %bl
-; AVX2-NEXT:    setns %bl
-; AVX2-NEXT:    addb %sil, %r8b
-; AVX2-NEXT:    movl %r9d, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX2-NEXT:    jno .LBB2_127
-; AVX2-NEXT:  # %bb.126:
-; AVX2-NEXT:    addb $127, %bl
-; AVX2-NEXT:    movl %ebx, %r8d
-; AVX2-NEXT:  .LBB2_127:
-; AVX2-NEXT:    vpextrb $2, %xmm1, %esi
-; AVX2-NEXT:    vpextrb $2, %xmm0, %r9d
-; AVX2-NEXT:    movl %r9d, %ebx
-; AVX2-NEXT:    addb %sil, %bl
-; AVX2-NEXT:    setns %bl
-; AVX2-NEXT:    addb %sil, %r9b
-; AVX2-NEXT:    jno .LBB2_129
-; AVX2-NEXT:  # %bb.128:
-; AVX2-NEXT:    addb $127, %bl
-; AVX2-NEXT:    movl %ebx, %r9d
-; AVX2-NEXT:  .LBB2_129:
-; AVX2-NEXT:    vpextrb $0, %xmm1, %esi
-; AVX2-NEXT:    vpextrb $0, %xmm0, %r10d
-; AVX2-NEXT:    movl %r10d, %ebx
-; AVX2-NEXT:    addb %sil, %bl
-; AVX2-NEXT:    setns %bl
-; AVX2-NEXT:    movl %r11d, %r12d
-; AVX2-NEXT:    addb %sil, %r10b
-; AVX2-NEXT:    jno .LBB2_131
-; AVX2-NEXT:  # %bb.130:
-; AVX2-NEXT:    addb $127, %bl
-; AVX2-NEXT:    movl %ebx, %r10d
-; AVX2-NEXT:  .LBB2_131:
-; AVX2-NEXT:    vpextrb $1, %xmm1, %esi
-; AVX2-NEXT:    vpextrb $1, %xmm0, %r11d
-; AVX2-NEXT:    movl %r11d, %ebx
-; AVX2-NEXT:    addb %sil, %bl
-; AVX2-NEXT:    setns %bl
-; AVX2-NEXT:    addb %sil, %r11b
-; AVX2-NEXT:    jno .LBB2_133
-; AVX2-NEXT:  # %bb.132:
-; AVX2-NEXT:    addb $127, %bl
-; AVX2-NEXT:    movl %ebx, %r11d
-; AVX2-NEXT:  .LBB2_133:
-; AVX2-NEXT:    movzbl %r10b, %esi
-; AVX2-NEXT:    vmovd %esi, %xmm0
-; AVX2-NEXT:    movzbl %r11b, %esi
-; AVX2-NEXT:    vpinsrb $1, %esi, %xmm0, %xmm0
-; AVX2-NEXT:    movzbl %r9b, %esi
-; AVX2-NEXT:    vpinsrb $2, %esi, %xmm0, %xmm0
-; AVX2-NEXT:    movzbl %r8b, %esi
-; AVX2-NEXT:    vpinsrb $3, %esi, %xmm0, %xmm0
-; AVX2-NEXT:    movzbl %dl, %edx
-; AVX2-NEXT:    vpinsrb $4, %edx, %xmm0, %xmm0
-; AVX2-NEXT:    movzbl %cl, %ecx
-; AVX2-NEXT:    vpinsrb $5, %ecx, %xmm0, %xmm0
-; AVX2-NEXT:    movzbl %al, %eax
-; AVX2-NEXT:    vpinsrb $6, %eax, %xmm0, %xmm0
-; AVX2-NEXT:    movzbl %dil, %eax
-; AVX2-NEXT:    vpinsrb $7, %eax, %xmm0, %xmm0
-; AVX2-NEXT:    movzbl %bpl, %eax
-; AVX2-NEXT:    vpinsrb $8, %eax, %xmm0, %xmm0
-; AVX2-NEXT:    movzbl %r14b, %eax
-; AVX2-NEXT:    vpinsrb $9, %eax, %xmm0, %xmm0
-; AVX2-NEXT:    movzbl %r15b, %eax
-; AVX2-NEXT:    vpinsrb $10, %eax, %xmm0, %xmm0
-; AVX2-NEXT:    movzbl %r13b, %eax
-; AVX2-NEXT:    vpinsrb $11, %eax, %xmm0, %xmm0
-; AVX2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX2-NEXT:    vpinsrb $12, %eax, %xmm0, %xmm0
-; AVX2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX2-NEXT:    vpinsrb $13, %eax, %xmm0, %xmm0
-; AVX2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX2-NEXT:    vpinsrb $14, %eax, %xmm0, %xmm0
-; AVX2-NEXT:    movzbl %r12b, %eax
-; AVX2-NEXT:    vpinsrb $15, %eax, %xmm0, %xmm0
-; AVX2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX2-NEXT:    vmovd %eax, %xmm1
-; AVX2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX2-NEXT:    vpinsrb $1, %eax, %xmm1, %xmm1
-; AVX2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX2-NEXT:    vpinsrb $2, %eax, %xmm1, %xmm1
-; AVX2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX2-NEXT:    vpinsrb $3, %eax, %xmm1, %xmm1
-; AVX2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX2-NEXT:    vpinsrb $4, %eax, %xmm1, %xmm1
-; AVX2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX2-NEXT:    vpinsrb $5, %eax, %xmm1, %xmm1
-; AVX2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX2-NEXT:    vpinsrb $6, %eax, %xmm1, %xmm1
-; AVX2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX2-NEXT:    vpinsrb $7, %eax, %xmm1, %xmm1
-; AVX2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX2-NEXT:    vpinsrb $8, %eax, %xmm1, %xmm1
-; AVX2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX2-NEXT:    vpinsrb $9, %eax, %xmm1, %xmm1
-; AVX2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX2-NEXT:    vpinsrb $10, %eax, %xmm1, %xmm1
-; AVX2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX2-NEXT:    vpinsrb $11, %eax, %xmm1, %xmm1
-; AVX2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX2-NEXT:    vpinsrb $12, %eax, %xmm1, %xmm1
-; AVX2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX2-NEXT:    vpinsrb $13, %eax, %xmm1, %xmm1
-; AVX2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX2-NEXT:    vpinsrb $14, %eax, %xmm1, %xmm1
-; AVX2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX2-NEXT:    vpinsrb $15, %eax, %xmm1, %xmm1
-; AVX2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX2-NEXT:    vmovd %eax, %xmm2
-; AVX2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX2-NEXT:    vpinsrb $1, %eax, %xmm2, %xmm2
-; AVX2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX2-NEXT:    vpinsrb $2, %eax, %xmm2, %xmm2
-; AVX2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX2-NEXT:    vpinsrb $3, %eax, %xmm2, %xmm2
-; AVX2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX2-NEXT:    vpinsrb $4, %eax, %xmm2, %xmm2
-; AVX2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX2-NEXT:    vpinsrb $5, %eax, %xmm2, %xmm2
-; AVX2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX2-NEXT:    vpinsrb $6, %eax, %xmm2, %xmm2
-; AVX2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX2-NEXT:    vpinsrb $7, %eax, %xmm2, %xmm2
-; AVX2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX2-NEXT:    vpinsrb $8, %eax, %xmm2, %xmm2
-; AVX2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX2-NEXT:    vpinsrb $9, %eax, %xmm2, %xmm2
-; AVX2-NEXT:    movzbl (%rsp), %eax # 1-byte Folded Reload
-; AVX2-NEXT:    vpinsrb $10, %eax, %xmm2, %xmm2
-; AVX2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX2-NEXT:    vpinsrb $11, %eax, %xmm2, %xmm2
-; AVX2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX2-NEXT:    vpinsrb $12, %eax, %xmm2, %xmm2
-; AVX2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX2-NEXT:    vpinsrb $13, %eax, %xmm2, %xmm2
-; AVX2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX2-NEXT:    vpinsrb $14, %eax, %xmm2, %xmm2
-; AVX2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX2-NEXT:    vpinsrb $15, %eax, %xmm2, %xmm2
-; AVX2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX2-NEXT:    vmovd %eax, %xmm3
-; AVX2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX2-NEXT:    vpinsrb $1, %eax, %xmm3, %xmm3
-; AVX2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX2-NEXT:    vpinsrb $2, %eax, %xmm3, %xmm3
-; AVX2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX2-NEXT:    vpinsrb $3, %eax, %xmm3, %xmm3
-; AVX2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX2-NEXT:    vpinsrb $4, %eax, %xmm3, %xmm3
-; AVX2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX2-NEXT:    vpinsrb $5, %eax, %xmm3, %xmm3
-; AVX2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX2-NEXT:    vpinsrb $6, %eax, %xmm3, %xmm3
-; AVX2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX2-NEXT:    vpinsrb $7, %eax, %xmm3, %xmm3
-; AVX2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX2-NEXT:    vpinsrb $8, %eax, %xmm3, %xmm3
-; AVX2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX2-NEXT:    vpinsrb $9, %eax, %xmm3, %xmm3
-; AVX2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX2-NEXT:    vpinsrb $10, %eax, %xmm3, %xmm3
-; AVX2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX2-NEXT:    vpinsrb $11, %eax, %xmm3, %xmm3
-; AVX2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX2-NEXT:    vpinsrb $12, %eax, %xmm3, %xmm3
-; AVX2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX2-NEXT:    vpinsrb $13, %eax, %xmm3, %xmm3
-; AVX2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX2-NEXT:    vpinsrb $14, %eax, %xmm3, %xmm3
-; AVX2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX2-NEXT:    vpinsrb $15, %eax, %xmm3, %xmm3
-; AVX2-NEXT:    vinserti128 $1, %xmm0, %ymm1, %ymm0
-; AVX2-NEXT:    vinserti128 $1, %xmm2, %ymm3, %ymm1
-; AVX2-NEXT:    addq $76, %rsp
-; AVX2-NEXT:    popq %rbx
-; AVX2-NEXT:    popq %r12
-; AVX2-NEXT:    popq %r13
-; AVX2-NEXT:    popq %r14
-; AVX2-NEXT:    popq %r15
-; AVX2-NEXT:    popq %rbp
+; AVX2-NEXT:    vpaddsb %ymm2, %ymm0, %ymm0
+; AVX2-NEXT:    vpaddsb %ymm3, %ymm1, %ymm1
 ; AVX2-NEXT:    retq
 ;
 ; AVX512-LABEL: v64i8:
 ; AVX512:       # %bb.0:
-; AVX512-NEXT:    pushq %rbp
-; AVX512-NEXT:    pushq %r15
-; AVX512-NEXT:    pushq %r14
-; AVX512-NEXT:    pushq %r13
-; AVX512-NEXT:    pushq %r12
-; AVX512-NEXT:    pushq %rbx
-; AVX512-NEXT:    subq $76, %rsp
-; AVX512-NEXT:    vpextrb $15, %xmm1, %ecx
-; AVX512-NEXT:    vpextrb $15, %xmm0, %edx
-; AVX512-NEXT:    movl %edx, %eax
-; AVX512-NEXT:    addb %cl, %al
-; AVX512-NEXT:    setns %al
-; AVX512-NEXT:    addb %cl, %dl
-; AVX512-NEXT:    jo .LBB2_1
-; AVX512-NEXT:  # %bb.2:
-; AVX512-NEXT:    movl %edx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX512-NEXT:    jmp .LBB2_3
-; AVX512-NEXT:  .LBB2_1:
-; AVX512-NEXT:    addb $127, %al
-; AVX512-NEXT:    # kill: def $al killed $al def $eax
-; AVX512-NEXT:    movl %eax, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX512-NEXT:  .LBB2_3:
-; AVX512-NEXT:    vpextrb $14, %xmm1, %ecx
-; AVX512-NEXT:    vpextrb $14, %xmm0, %edx
-; AVX512-NEXT:    movl %edx, %eax
-; AVX512-NEXT:    addb %cl, %al
-; AVX512-NEXT:    setns %al
-; AVX512-NEXT:    addb %cl, %dl
-; AVX512-NEXT:    jno .LBB2_5
-; AVX512-NEXT:  # %bb.4:
-; AVX512-NEXT:    addb $127, %al
-; AVX512-NEXT:    movl %eax, %edx
-; AVX512-NEXT:  .LBB2_5:
-; AVX512-NEXT:    vpextrb $13, %xmm1, %ecx
-; AVX512-NEXT:    vpextrb $13, %xmm0, %esi
-; AVX512-NEXT:    movl %esi, %eax
-; AVX512-NEXT:    addb %cl, %al
-; AVX512-NEXT:    setns %al
-; AVX512-NEXT:    addb %cl, %sil
-; AVX512-NEXT:    jo .LBB2_6
-; AVX512-NEXT:  # %bb.7:
-; AVX512-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX512-NEXT:    jmp .LBB2_8
-; AVX512-NEXT:  .LBB2_6:
-; AVX512-NEXT:    addb $127, %al
-; AVX512-NEXT:    # kill: def $al killed $al def $eax
-; AVX512-NEXT:    movl %eax, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX512-NEXT:  .LBB2_8:
-; AVX512-NEXT:    vpextrb $12, %xmm1, %ecx
-; AVX512-NEXT:    vpextrb $12, %xmm0, %esi
-; AVX512-NEXT:    movl %esi, %eax
-; AVX512-NEXT:    addb %cl, %al
-; AVX512-NEXT:    setns %al
-; AVX512-NEXT:    addb %cl, %sil
-; AVX512-NEXT:    jno .LBB2_10
-; AVX512-NEXT:  # %bb.9:
-; AVX512-NEXT:    addb $127, %al
-; AVX512-NEXT:    movl %eax, %esi
-; AVX512-NEXT:  .LBB2_10:
-; AVX512-NEXT:    vpextrb $11, %xmm1, %ecx
-; AVX512-NEXT:    vpextrb $11, %xmm0, %edi
-; AVX512-NEXT:    movl %edi, %eax
-; AVX512-NEXT:    addb %cl, %al
-; AVX512-NEXT:    setns %al
-; AVX512-NEXT:    addb %cl, %dil
-; AVX512-NEXT:    jno .LBB2_12
-; AVX512-NEXT:  # %bb.11:
-; AVX512-NEXT:    addb $127, %al
-; AVX512-NEXT:    movl %eax, %edi
-; AVX512-NEXT:  .LBB2_12:
-; AVX512-NEXT:    vpextrb $10, %xmm1, %ecx
-; AVX512-NEXT:    vpextrb $10, %xmm0, %ebp
-; AVX512-NEXT:    movl %ebp, %eax
-; AVX512-NEXT:    addb %cl, %al
-; AVX512-NEXT:    setns %al
-; AVX512-NEXT:    addb %cl, %bpl
-; AVX512-NEXT:    jno .LBB2_14
-; AVX512-NEXT:  # %bb.13:
-; AVX512-NEXT:    addb $127, %al
-; AVX512-NEXT:    movl %eax, %ebp
-; AVX512-NEXT:  .LBB2_14:
-; AVX512-NEXT:    vpextrb $9, %xmm1, %ecx
-; AVX512-NEXT:    vpextrb $9, %xmm0, %ebx
-; AVX512-NEXT:    movl %ebx, %eax
-; AVX512-NEXT:    addb %cl, %al
-; AVX512-NEXT:    setns %al
-; AVX512-NEXT:    addb %cl, %bl
-; AVX512-NEXT:    movl %ebp, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX512-NEXT:    jo .LBB2_15
-; AVX512-NEXT:  # %bb.16:
-; AVX512-NEXT:    movl %ebx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX512-NEXT:    jmp .LBB2_17
-; AVX512-NEXT:  .LBB2_15:
-; AVX512-NEXT:    addb $127, %al
-; AVX512-NEXT:    # kill: def $al killed $al def $eax
-; AVX512-NEXT:    movl %eax, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX512-NEXT:  .LBB2_17:
-; AVX512-NEXT:    vpextrb $8, %xmm1, %ecx
-; AVX512-NEXT:    vpextrb $8, %xmm0, %ebp
-; AVX512-NEXT:    movl %ebp, %eax
-; AVX512-NEXT:    addb %cl, %al
-; AVX512-NEXT:    setns %al
-; AVX512-NEXT:    addb %cl, %bpl
-; AVX512-NEXT:    jno .LBB2_19
-; AVX512-NEXT:  # %bb.18:
-; AVX512-NEXT:    addb $127, %al
-; AVX512-NEXT:    movl %eax, %ebp
-; AVX512-NEXT:  .LBB2_19:
-; AVX512-NEXT:    vpextrb $7, %xmm1, %ecx
-; AVX512-NEXT:    vpextrb $7, %xmm0, %ebx
-; AVX512-NEXT:    movl %ebx, %eax
-; AVX512-NEXT:    addb %cl, %al
-; AVX512-NEXT:    setns %al
-; AVX512-NEXT:    addb %cl, %bl
-; AVX512-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX512-NEXT:    jno .LBB2_21
-; AVX512-NEXT:  # %bb.20:
-; AVX512-NEXT:    addb $127, %al
-; AVX512-NEXT:    movl %eax, %ebx
-; AVX512-NEXT:  .LBB2_21:
-; AVX512-NEXT:    vpextrb $6, %xmm1, %ecx
-; AVX512-NEXT:    vpextrb $6, %xmm0, %esi
-; AVX512-NEXT:    movl %esi, %eax
-; AVX512-NEXT:    addb %cl, %al
-; AVX512-NEXT:    setns %al
-; AVX512-NEXT:    addb %cl, %sil
-; AVX512-NEXT:    movl %edx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX512-NEXT:    jno .LBB2_23
-; AVX512-NEXT:  # %bb.22:
-; AVX512-NEXT:    addb $127, %al
-; AVX512-NEXT:    movl %eax, %esi
-; AVX512-NEXT:  .LBB2_23:
-; AVX512-NEXT:    vpextrb $5, %xmm1, %ecx
-; AVX512-NEXT:    vpextrb $5, %xmm0, %r11d
-; AVX512-NEXT:    movl %r11d, %eax
-; AVX512-NEXT:    addb %cl, %al
-; AVX512-NEXT:    setns %al
-; AVX512-NEXT:    addb %cl, %r11b
-; AVX512-NEXT:    movl %edi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX512-NEXT:    movl %ebp, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX512-NEXT:    movl %ebx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX512-NEXT:    jno .LBB2_25
-; AVX512-NEXT:  # %bb.24:
-; AVX512-NEXT:    addb $127, %al
-; AVX512-NEXT:    movl %eax, %r11d
-; AVX512-NEXT:  .LBB2_25:
-; AVX512-NEXT:    vpextrb $4, %xmm1, %ecx
-; AVX512-NEXT:    vpextrb $4, %xmm0, %r13d
-; AVX512-NEXT:    movl %r13d, %eax
-; AVX512-NEXT:    addb %cl, %al
-; AVX512-NEXT:    setns %al
-; AVX512-NEXT:    addb %cl, %r13b
-; AVX512-NEXT:    jno .LBB2_27
-; AVX512-NEXT:  # %bb.26:
-; AVX512-NEXT:    addb $127, %al
-; AVX512-NEXT:    movl %eax, %r13d
-; AVX512-NEXT:  .LBB2_27:
-; AVX512-NEXT:    vpextrb $3, %xmm1, %ecx
-; AVX512-NEXT:    vpextrb $3, %xmm0, %r8d
-; AVX512-NEXT:    movl %r8d, %eax
-; AVX512-NEXT:    addb %cl, %al
-; AVX512-NEXT:    setns %al
-; AVX512-NEXT:    addb %cl, %r8b
-; AVX512-NEXT:    jno .LBB2_29
-; AVX512-NEXT:  # %bb.28:
-; AVX512-NEXT:    addb $127, %al
-; AVX512-NEXT:    movl %eax, %r8d
-; AVX512-NEXT:  .LBB2_29:
-; AVX512-NEXT:    vpextrb $2, %xmm1, %ecx
-; AVX512-NEXT:    vpextrb $2, %xmm0, %ebp
-; AVX512-NEXT:    movl %ebp, %eax
-; AVX512-NEXT:    addb %cl, %al
-; AVX512-NEXT:    setns %al
-; AVX512-NEXT:    addb %cl, %bpl
-; AVX512-NEXT:    jno .LBB2_31
-; AVX512-NEXT:  # %bb.30:
-; AVX512-NEXT:    addb $127, %al
-; AVX512-NEXT:    movl %eax, %ebp
-; AVX512-NEXT:  .LBB2_31:
-; AVX512-NEXT:    vpextrb $0, %xmm1, %ecx
-; AVX512-NEXT:    vpextrb $0, %xmm0, %ebx
-; AVX512-NEXT:    movl %ebx, %eax
-; AVX512-NEXT:    addb %cl, %al
-; AVX512-NEXT:    setns %al
-; AVX512-NEXT:    addb %cl, %bl
-; AVX512-NEXT:    jno .LBB2_33
-; AVX512-NEXT:  # %bb.32:
-; AVX512-NEXT:    addb $127, %al
-; AVX512-NEXT:    movl %eax, %ebx
-; AVX512-NEXT:  .LBB2_33:
-; AVX512-NEXT:    vpextrb $1, %xmm1, %ecx
-; AVX512-NEXT:    vpextrb $1, %xmm0, %edi
-; AVX512-NEXT:    movl %edi, %eax
-; AVX512-NEXT:    addb %cl, %al
-; AVX512-NEXT:    setns %al
-; AVX512-NEXT:    addb %cl, %dil
-; AVX512-NEXT:    jno .LBB2_35
-; AVX512-NEXT:  # %bb.34:
-; AVX512-NEXT:    addb $127, %al
-; AVX512-NEXT:    movl %eax, %edi
-; AVX512-NEXT:  .LBB2_35:
-; AVX512-NEXT:    vextracti128 $1, %ymm1, %xmm2
-; AVX512-NEXT:    vpextrb $15, %xmm2, %ecx
-; AVX512-NEXT:    vextracti128 $1, %ymm0, %xmm3
-; AVX512-NEXT:    vpextrb $15, %xmm3, %edx
-; AVX512-NEXT:    movl %edx, %eax
-; AVX512-NEXT:    addb %cl, %al
-; AVX512-NEXT:    setns %al
-; AVX512-NEXT:    addb %cl, %dl
-; AVX512-NEXT:    movl %ebp, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX512-NEXT:    movl %ebx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX512-NEXT:    movl %edi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX512-NEXT:    jno .LBB2_37
-; AVX512-NEXT:  # %bb.36:
-; AVX512-NEXT:    addb $127, %al
-; AVX512-NEXT:    movl %eax, %edx
-; AVX512-NEXT:  .LBB2_37:
-; AVX512-NEXT:    vpextrb $14, %xmm2, %ecx
-; AVX512-NEXT:    vpextrb $14, %xmm3, %edi
-; AVX512-NEXT:    movl %edi, %eax
-; AVX512-NEXT:    addb %cl, %al
-; AVX512-NEXT:    setns %al
-; AVX512-NEXT:    addb %cl, %dil
-; AVX512-NEXT:    movl %edx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX512-NEXT:    jno .LBB2_39
-; AVX512-NEXT:  # %bb.38:
-; AVX512-NEXT:    addb $127, %al
-; AVX512-NEXT:    movl %eax, %edi
-; AVX512-NEXT:  .LBB2_39:
-; AVX512-NEXT:    vpextrb $13, %xmm2, %ecx
-; AVX512-NEXT:    vpextrb $13, %xmm3, %r12d
-; AVX512-NEXT:    movl %r12d, %eax
-; AVX512-NEXT:    addb %cl, %al
-; AVX512-NEXT:    setns %al
-; AVX512-NEXT:    addb %cl, %r12b
-; AVX512-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX512-NEXT:    jno .LBB2_41
-; AVX512-NEXT:  # %bb.40:
-; AVX512-NEXT:    addb $127, %al
-; AVX512-NEXT:    movl %eax, %r12d
-; AVX512-NEXT:  .LBB2_41:
-; AVX512-NEXT:    vpextrb $12, %xmm2, %ecx
-; AVX512-NEXT:    vpextrb $12, %xmm3, %r15d
-; AVX512-NEXT:    movl %r15d, %eax
-; AVX512-NEXT:    addb %cl, %al
-; AVX512-NEXT:    setns %al
-; AVX512-NEXT:    addb %cl, %r15b
-; AVX512-NEXT:    movl %r11d, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX512-NEXT:    jno .LBB2_43
-; AVX512-NEXT:  # %bb.42:
-; AVX512-NEXT:    addb $127, %al
-; AVX512-NEXT:    movl %eax, %r15d
-; AVX512-NEXT:  .LBB2_43:
-; AVX512-NEXT:    vpextrb $11, %xmm2, %ecx
-; AVX512-NEXT:    vpextrb $11, %xmm3, %ebp
-; AVX512-NEXT:    movl %ebp, %eax
-; AVX512-NEXT:    addb %cl, %al
-; AVX512-NEXT:    setns %al
-; AVX512-NEXT:    addb %cl, %bpl
-; AVX512-NEXT:    movl %r13d, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX512-NEXT:    jno .LBB2_45
-; AVX512-NEXT:  # %bb.44:
-; AVX512-NEXT:    addb $127, %al
-; AVX512-NEXT:    movl %eax, %ebp
-; AVX512-NEXT:  .LBB2_45:
-; AVX512-NEXT:    vpextrb $10, %xmm2, %ecx
-; AVX512-NEXT:    vpextrb $10, %xmm3, %ebx
-; AVX512-NEXT:    movl %ebx, %eax
-; AVX512-NEXT:    addb %cl, %al
-; AVX512-NEXT:    setns %al
-; AVX512-NEXT:    addb %cl, %bl
-; AVX512-NEXT:    jno .LBB2_47
-; AVX512-NEXT:  # %bb.46:
-; AVX512-NEXT:    addb $127, %al
-; AVX512-NEXT:    movl %eax, %ebx
-; AVX512-NEXT:  .LBB2_47:
-; AVX512-NEXT:    vpextrb $9, %xmm2, %ecx
-; AVX512-NEXT:    vpextrb $9, %xmm3, %edx
-; AVX512-NEXT:    movl %edx, %eax
-; AVX512-NEXT:    addb %cl, %al
-; AVX512-NEXT:    setns %al
-; AVX512-NEXT:    addb %cl, %dl
-; AVX512-NEXT:    jno .LBB2_49
-; AVX512-NEXT:  # %bb.48:
-; AVX512-NEXT:    addb $127, %al
-; AVX512-NEXT:    movl %eax, %edx
-; AVX512-NEXT:  .LBB2_49:
-; AVX512-NEXT:    vpextrb $8, %xmm2, %ecx
-; AVX512-NEXT:    vpextrb $8, %xmm3, %esi
-; AVX512-NEXT:    movl %esi, %eax
-; AVX512-NEXT:    addb %cl, %al
-; AVX512-NEXT:    setns %al
-; AVX512-NEXT:    addb %cl, %sil
-; AVX512-NEXT:    movl %ebp, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX512-NEXT:    movl %ebx, (%rsp) # 4-byte Spill
-; AVX512-NEXT:    movl %edx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX512-NEXT:    jo .LBB2_50
-; AVX512-NEXT:  # %bb.51:
-; AVX512-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX512-NEXT:    jmp .LBB2_52
-; AVX512-NEXT:  .LBB2_50:
-; AVX512-NEXT:    addb $127, %al
-; AVX512-NEXT:    # kill: def $al killed $al def $eax
-; AVX512-NEXT:    movl %eax, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX512-NEXT:  .LBB2_52:
-; AVX512-NEXT:    vpextrb $7, %xmm2, %ecx
-; AVX512-NEXT:    vpextrb $7, %xmm3, %r11d
-; AVX512-NEXT:    movl %r11d, %eax
-; AVX512-NEXT:    addb %cl, %al
-; AVX512-NEXT:    setns %al
-; AVX512-NEXT:    addb %cl, %r11b
-; AVX512-NEXT:    jno .LBB2_54
-; AVX512-NEXT:  # %bb.53:
-; AVX512-NEXT:    addb $127, %al
-; AVX512-NEXT:    movl %eax, %r11d
-; AVX512-NEXT:  .LBB2_54:
-; AVX512-NEXT:    vpextrb $6, %xmm2, %ecx
-; AVX512-NEXT:    vpextrb $6, %xmm3, %esi
-; AVX512-NEXT:    movl %esi, %eax
-; AVX512-NEXT:    addb %cl, %al
-; AVX512-NEXT:    setns %al
-; AVX512-NEXT:    addb %cl, %sil
-; AVX512-NEXT:    movl %r8d, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX512-NEXT:    movl %edi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX512-NEXT:    jno .LBB2_56
-; AVX512-NEXT:  # %bb.55:
-; AVX512-NEXT:    addb $127, %al
-; AVX512-NEXT:    movl %eax, %esi
-; AVX512-NEXT:  .LBB2_56:
-; AVX512-NEXT:    vpextrb $5, %xmm2, %ecx
-; AVX512-NEXT:    vpextrb $5, %xmm3, %edi
-; AVX512-NEXT:    movl %edi, %eax
-; AVX512-NEXT:    addb %cl, %al
-; AVX512-NEXT:    setns %al
-; AVX512-NEXT:    addb %cl, %dil
-; AVX512-NEXT:    jno .LBB2_58
-; AVX512-NEXT:  # %bb.57:
-; AVX512-NEXT:    addb $127, %al
-; AVX512-NEXT:    movl %eax, %edi
-; AVX512-NEXT:  .LBB2_58:
-; AVX512-NEXT:    vpextrb $4, %xmm2, %ecx
-; AVX512-NEXT:    vpextrb $4, %xmm3, %r13d
-; AVX512-NEXT:    movl %r13d, %eax
-; AVX512-NEXT:    addb %cl, %al
-; AVX512-NEXT:    setns %al
-; AVX512-NEXT:    addb %cl, %r13b
-; AVX512-NEXT:    movl %r13d, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX512-NEXT:    jno .LBB2_60
-; AVX512-NEXT:  # %bb.59:
-; AVX512-NEXT:    addb $127, %al
-; AVX512-NEXT:    # kill: def $al killed $al def $eax
-; AVX512-NEXT:    movl %eax, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX512-NEXT:  .LBB2_60:
-; AVX512-NEXT:    vpextrb $3, %xmm2, %ecx
-; AVX512-NEXT:    vpextrb $3, %xmm3, %ebp
-; AVX512-NEXT:    movl %ebp, %eax
-; AVX512-NEXT:    addb %cl, %al
-; AVX512-NEXT:    setns %al
-; AVX512-NEXT:    addb %cl, %bpl
-; AVX512-NEXT:    jo .LBB2_61
-; AVX512-NEXT:  # %bb.62:
-; AVX512-NEXT:    movl %ebp, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX512-NEXT:    jmp .LBB2_63
-; AVX512-NEXT:  .LBB2_61:
-; AVX512-NEXT:    addb $127, %al
-; AVX512-NEXT:    # kill: def $al killed $al def $eax
-; AVX512-NEXT:    movl %eax, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX512-NEXT:  .LBB2_63:
-; AVX512-NEXT:    vpextrb $2, %xmm2, %ecx
-; AVX512-NEXT:    vpextrb $2, %xmm3, %ebp
-; AVX512-NEXT:    movl %ebp, %eax
-; AVX512-NEXT:    addb %cl, %al
-; AVX512-NEXT:    setns %al
-; AVX512-NEXT:    addb %cl, %bpl
-; AVX512-NEXT:    movl %edi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX512-NEXT:    jno .LBB2_65
-; AVX512-NEXT:  # %bb.64:
-; AVX512-NEXT:    addb $127, %al
-; AVX512-NEXT:    movl %eax, %ebp
-; AVX512-NEXT:  .LBB2_65:
-; AVX512-NEXT:    vpextrb $0, %xmm2, %ecx
-; AVX512-NEXT:    vpextrb $0, %xmm3, %edi
-; AVX512-NEXT:    movl %edi, %eax
-; AVX512-NEXT:    addb %cl, %al
-; AVX512-NEXT:    setns %al
-; AVX512-NEXT:    addb %cl, %dil
-; AVX512-NEXT:    movl %r15d, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX512-NEXT:    jno .LBB2_67
-; AVX512-NEXT:  # %bb.66:
-; AVX512-NEXT:    addb $127, %al
-; AVX512-NEXT:    movl %eax, %edi
-; AVX512-NEXT:  .LBB2_67:
-; AVX512-NEXT:    vpextrb $1, %xmm2, %ecx
-; AVX512-NEXT:    vpextrb $1, %xmm3, %ebx
-; AVX512-NEXT:    movl %ebx, %eax
-; AVX512-NEXT:    addb %cl, %al
-; AVX512-NEXT:    setns %al
-; AVX512-NEXT:    addb %cl, %bl
-; AVX512-NEXT:    movl %r12d, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX512-NEXT:    jno .LBB2_69
-; AVX512-NEXT:  # %bb.68:
-; AVX512-NEXT:    addb $127, %al
-; AVX512-NEXT:    movl %eax, %ebx
-; AVX512-NEXT:  .LBB2_69:
-; AVX512-NEXT:    vextracti32x4 $2, %zmm1, %xmm2
-; AVX512-NEXT:    vpextrb $15, %xmm2, %ecx
-; AVX512-NEXT:    vextracti32x4 $2, %zmm0, %xmm3
-; AVX512-NEXT:    vpextrb $15, %xmm3, %edx
-; AVX512-NEXT:    movl %edx, %eax
-; AVX512-NEXT:    addb %cl, %al
-; AVX512-NEXT:    setns %al
-; AVX512-NEXT:    addb %cl, %dl
-; AVX512-NEXT:    movl %edi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX512-NEXT:    jno .LBB2_71
-; AVX512-NEXT:  # %bb.70:
-; AVX512-NEXT:    addb $127, %al
-; AVX512-NEXT:    movl %eax, %edx
-; AVX512-NEXT:  .LBB2_71:
-; AVX512-NEXT:    vpextrb $14, %xmm2, %ecx
-; AVX512-NEXT:    vpextrb $14, %xmm3, %edi
-; AVX512-NEXT:    movl %edi, %eax
-; AVX512-NEXT:    addb %cl, %al
-; AVX512-NEXT:    setns %al
-; AVX512-NEXT:    addb %cl, %dil
-; AVX512-NEXT:    jno .LBB2_73
-; AVX512-NEXT:  # %bb.72:
-; AVX512-NEXT:    addb $127, %al
-; AVX512-NEXT:    movl %eax, %edi
-; AVX512-NEXT:  .LBB2_73:
-; AVX512-NEXT:    vpextrb $13, %xmm2, %ecx
-; AVX512-NEXT:    vpextrb $13, %xmm3, %r10d
-; AVX512-NEXT:    movl %r10d, %eax
-; AVX512-NEXT:    addb %cl, %al
-; AVX512-NEXT:    setns %al
-; AVX512-NEXT:    addb %cl, %r10b
-; AVX512-NEXT:    jno .LBB2_75
-; AVX512-NEXT:  # %bb.74:
-; AVX512-NEXT:    addb $127, %al
-; AVX512-NEXT:    movl %eax, %r10d
-; AVX512-NEXT:  .LBB2_75:
-; AVX512-NEXT:    vpextrb $12, %xmm2, %ecx
-; AVX512-NEXT:    vpextrb $12, %xmm3, %r12d
-; AVX512-NEXT:    movl %r12d, %eax
-; AVX512-NEXT:    addb %cl, %al
-; AVX512-NEXT:    setns %al
-; AVX512-NEXT:    addb %cl, %r12b
-; AVX512-NEXT:    movl %r12d, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX512-NEXT:    jno .LBB2_77
-; AVX512-NEXT:  # %bb.76:
-; AVX512-NEXT:    addb $127, %al
-; AVX512-NEXT:    # kill: def $al killed $al def $eax
-; AVX512-NEXT:    movl %eax, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX512-NEXT:  .LBB2_77:
-; AVX512-NEXT:    vpextrb $11, %xmm2, %ecx
-; AVX512-NEXT:    vpextrb $11, %xmm3, %r14d
-; AVX512-NEXT:    movl %r14d, %eax
-; AVX512-NEXT:    addb %cl, %al
-; AVX512-NEXT:    setns %al
-; AVX512-NEXT:    addb %cl, %r14b
-; AVX512-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX512-NEXT:    jno .LBB2_79
-; AVX512-NEXT:  # %bb.78:
-; AVX512-NEXT:    addb $127, %al
-; AVX512-NEXT:    movl %eax, %r14d
-; AVX512-NEXT:  .LBB2_79:
-; AVX512-NEXT:    vpextrb $10, %xmm2, %ecx
-; AVX512-NEXT:    vpextrb $10, %xmm3, %r13d
-; AVX512-NEXT:    movl %r13d, %eax
-; AVX512-NEXT:    addb %cl, %al
-; AVX512-NEXT:    setns %al
-; AVX512-NEXT:    addb %cl, %r13b
-; AVX512-NEXT:    movl %r11d, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX512-NEXT:    movl %ebp, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX512-NEXT:    movl %ebx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX512-NEXT:    movl %edx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX512-NEXT:    jno .LBB2_81
-; AVX512-NEXT:  # %bb.80:
-; AVX512-NEXT:    addb $127, %al
-; AVX512-NEXT:    movl %eax, %r13d
-; AVX512-NEXT:  .LBB2_81:
-; AVX512-NEXT:    vpextrb $9, %xmm2, %ecx
-; AVX512-NEXT:    vpextrb $9, %xmm3, %r8d
-; AVX512-NEXT:    movl %r8d, %eax
-; AVX512-NEXT:    addb %cl, %al
-; AVX512-NEXT:    setns %al
-; AVX512-NEXT:    addb %cl, %r8b
-; AVX512-NEXT:    jno .LBB2_83
-; AVX512-NEXT:  # %bb.82:
-; AVX512-NEXT:    addb $127, %al
-; AVX512-NEXT:    movl %eax, %r8d
-; AVX512-NEXT:  .LBB2_83:
-; AVX512-NEXT:    vpextrb $8, %xmm2, %ecx
-; AVX512-NEXT:    vpextrb $8, %xmm3, %r15d
-; AVX512-NEXT:    movl %r15d, %eax
-; AVX512-NEXT:    addb %cl, %al
-; AVX512-NEXT:    setns %al
-; AVX512-NEXT:    addb %cl, %r15b
-; AVX512-NEXT:    jno .LBB2_85
-; AVX512-NEXT:  # %bb.84:
-; AVX512-NEXT:    addb $127, %al
-; AVX512-NEXT:    movl %eax, %r15d
-; AVX512-NEXT:  .LBB2_85:
-; AVX512-NEXT:    vpextrb $7, %xmm2, %ecx
-; AVX512-NEXT:    vpextrb $7, %xmm3, %r12d
-; AVX512-NEXT:    movl %r12d, %eax
-; AVX512-NEXT:    addb %cl, %al
-; AVX512-NEXT:    setns %al
-; AVX512-NEXT:    addb %cl, %r12b
-; AVX512-NEXT:    movl %r12d, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX512-NEXT:    jno .LBB2_87
-; AVX512-NEXT:  # %bb.86:
-; AVX512-NEXT:    addb $127, %al
-; AVX512-NEXT:    # kill: def $al killed $al def $eax
-; AVX512-NEXT:    movl %eax, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX512-NEXT:  .LBB2_87:
-; AVX512-NEXT:    vpextrb $6, %xmm2, %ecx
-; AVX512-NEXT:    vpextrb $6, %xmm3, %r12d
-; AVX512-NEXT:    movl %r12d, %eax
-; AVX512-NEXT:    addb %cl, %al
-; AVX512-NEXT:    setns %al
-; AVX512-NEXT:    addb %cl, %r12b
-; AVX512-NEXT:    jno .LBB2_89
-; AVX512-NEXT:  # %bb.88:
-; AVX512-NEXT:    addb $127, %al
-; AVX512-NEXT:    movl %eax, %r12d
-; AVX512-NEXT:  .LBB2_89:
-; AVX512-NEXT:    vpextrb $5, %xmm2, %ecx
-; AVX512-NEXT:    vpextrb $5, %xmm3, %ebp
-; AVX512-NEXT:    movl %ebp, %eax
-; AVX512-NEXT:    addb %cl, %al
-; AVX512-NEXT:    setns %al
-; AVX512-NEXT:    addb %cl, %bpl
-; AVX512-NEXT:    jno .LBB2_91
-; AVX512-NEXT:  # %bb.90:
-; AVX512-NEXT:    addb $127, %al
-; AVX512-NEXT:    movl %eax, %ebp
-; AVX512-NEXT:  .LBB2_91:
-; AVX512-NEXT:    vpextrb $4, %xmm2, %ecx
-; AVX512-NEXT:    vpextrb $4, %xmm3, %esi
-; AVX512-NEXT:    movl %esi, %eax
-; AVX512-NEXT:    addb %cl, %al
-; AVX512-NEXT:    setns %al
-; AVX512-NEXT:    addb %cl, %sil
-; AVX512-NEXT:    jno .LBB2_93
-; AVX512-NEXT:  # %bb.92:
-; AVX512-NEXT:    addb $127, %al
-; AVX512-NEXT:    movl %eax, %esi
-; AVX512-NEXT:  .LBB2_93:
-; AVX512-NEXT:    vpextrb $3, %xmm2, %ecx
-; AVX512-NEXT:    vpextrb $3, %xmm3, %edx
-; AVX512-NEXT:    movl %edx, %eax
-; AVX512-NEXT:    addb %cl, %al
-; AVX512-NEXT:    setns %al
-; AVX512-NEXT:    addb %cl, %dl
-; AVX512-NEXT:    movl %edi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX512-NEXT:    jno .LBB2_95
-; AVX512-NEXT:  # %bb.94:
-; AVX512-NEXT:    addb $127, %al
-; AVX512-NEXT:    movl %eax, %edx
-; AVX512-NEXT:  .LBB2_95:
-; AVX512-NEXT:    vpextrb $2, %xmm2, %ecx
-; AVX512-NEXT:    vpextrb $2, %xmm3, %edi
-; AVX512-NEXT:    movl %edi, %eax
-; AVX512-NEXT:    addb %cl, %al
-; AVX512-NEXT:    setns %al
-; AVX512-NEXT:    addb %cl, %dil
-; AVX512-NEXT:    jno .LBB2_97
-; AVX512-NEXT:  # %bb.96:
-; AVX512-NEXT:    addb $127, %al
-; AVX512-NEXT:    movl %eax, %edi
-; AVX512-NEXT:  .LBB2_97:
-; AVX512-NEXT:    vpextrb $0, %xmm2, %ecx
-; AVX512-NEXT:    vpextrb $0, %xmm3, %ebx
-; AVX512-NEXT:    movl %ebx, %eax
-; AVX512-NEXT:    addb %cl, %al
-; AVX512-NEXT:    setns %al
-; AVX512-NEXT:    addb %cl, %bl
-; AVX512-NEXT:    jno .LBB2_99
-; AVX512-NEXT:  # %bb.98:
-; AVX512-NEXT:    addb $127, %al
-; AVX512-NEXT:    movl %eax, %ebx
-; AVX512-NEXT:  .LBB2_99:
-; AVX512-NEXT:    vpextrb $1, %xmm2, %ecx
-; AVX512-NEXT:    vpextrb $1, %xmm3, %r11d
-; AVX512-NEXT:    movl %r11d, %eax
-; AVX512-NEXT:    addb %cl, %al
-; AVX512-NEXT:    setns %al
-; AVX512-NEXT:    addb %cl, %r11b
-; AVX512-NEXT:    movl %r11d, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX512-NEXT:    jno .LBB2_101
-; AVX512-NEXT:  # %bb.100:
-; AVX512-NEXT:    addb $127, %al
-; AVX512-NEXT:    # kill: def $al killed $al def $eax
-; AVX512-NEXT:    movl %eax, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX512-NEXT:  .LBB2_101:
-; AVX512-NEXT:    vextracti32x4 $3, %zmm1, %xmm1
-; AVX512-NEXT:    vpextrb $15, %xmm1, %ecx
-; AVX512-NEXT:    vextracti32x4 $3, %zmm0, %xmm0
-; AVX512-NEXT:    vpextrb $15, %xmm0, %r11d
-; AVX512-NEXT:    movl %r11d, %eax
-; AVX512-NEXT:    addb %cl, %al
-; AVX512-NEXT:    setns %al
-; AVX512-NEXT:    addb %cl, %r11b
-; AVX512-NEXT:    jno .LBB2_103
-; AVX512-NEXT:  # %bb.102:
-; AVX512-NEXT:    addb $127, %al
-; AVX512-NEXT:    movl %eax, %r11d
-; AVX512-NEXT:  .LBB2_103:
-; AVX512-NEXT:    vpextrb $14, %xmm1, %ecx
-; AVX512-NEXT:    vpextrb $14, %xmm0, %r9d
-; AVX512-NEXT:    movl %r9d, %eax
-; AVX512-NEXT:    addb %cl, %al
-; AVX512-NEXT:    setns %al
-; AVX512-NEXT:    addb %cl, %r9b
-; AVX512-NEXT:    movl %r9d, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX512-NEXT:    jno .LBB2_105
-; AVX512-NEXT:  # %bb.104:
-; AVX512-NEXT:    addb $127, %al
-; AVX512-NEXT:    # kill: def $al killed $al def $eax
-; AVX512-NEXT:    movl %eax, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX512-NEXT:  .LBB2_105:
-; AVX512-NEXT:    vpextrb $13, %xmm1, %ecx
-; AVX512-NEXT:    vpextrb $13, %xmm0, %r9d
-; AVX512-NEXT:    movl %r9d, %eax
-; AVX512-NEXT:    addb %cl, %al
-; AVX512-NEXT:    setns %al
-; AVX512-NEXT:    addb %cl, %r9b
-; AVX512-NEXT:    movl %r14d, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX512-NEXT:    movl %r13d, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX512-NEXT:    movl %r8d, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX512-NEXT:    movl %r15d, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX512-NEXT:    movl %r12d, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX512-NEXT:    movl %ebp, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX512-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX512-NEXT:    movl %edx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX512-NEXT:    movl %edi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX512-NEXT:    movl %ebx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX512-NEXT:    jno .LBB2_107
-; AVX512-NEXT:  # %bb.106:
-; AVX512-NEXT:    addb $127, %al
-; AVX512-NEXT:    movl %eax, %r9d
-; AVX512-NEXT:  .LBB2_107:
-; AVX512-NEXT:    vpextrb $12, %xmm1, %ecx
-; AVX512-NEXT:    vpextrb $12, %xmm0, %edx
-; AVX512-NEXT:    movl %edx, %eax
-; AVX512-NEXT:    addb %cl, %al
-; AVX512-NEXT:    setns %al
-; AVX512-NEXT:    addb %cl, %dl
-; AVX512-NEXT:    movl %r10d, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX512-NEXT:    jno .LBB2_109
-; AVX512-NEXT:  # %bb.108:
-; AVX512-NEXT:    addb $127, %al
-; AVX512-NEXT:    movl %eax, %edx
-; AVX512-NEXT:  .LBB2_109:
-; AVX512-NEXT:    vpextrb $11, %xmm1, %ecx
-; AVX512-NEXT:    vpextrb $11, %xmm0, %r13d
-; AVX512-NEXT:    movl %r13d, %eax
-; AVX512-NEXT:    addb %cl, %al
-; AVX512-NEXT:    setns %al
-; AVX512-NEXT:    addb %cl, %r13b
-; AVX512-NEXT:    jno .LBB2_111
-; AVX512-NEXT:  # %bb.110:
-; AVX512-NEXT:    addb $127, %al
-; AVX512-NEXT:    movl %eax, %r13d
-; AVX512-NEXT:  .LBB2_111:
-; AVX512-NEXT:    vpextrb $10, %xmm1, %ecx
-; AVX512-NEXT:    vpextrb $10, %xmm0, %r15d
-; AVX512-NEXT:    movl %r15d, %eax
-; AVX512-NEXT:    addb %cl, %al
-; AVX512-NEXT:    setns %al
-; AVX512-NEXT:    addb %cl, %r15b
-; AVX512-NEXT:    jno .LBB2_113
-; AVX512-NEXT:  # %bb.112:
-; AVX512-NEXT:    addb $127, %al
-; AVX512-NEXT:    movl %eax, %r15d
-; AVX512-NEXT:  .LBB2_113:
-; AVX512-NEXT:    vpextrb $9, %xmm1, %ecx
-; AVX512-NEXT:    vpextrb $9, %xmm0, %r14d
-; AVX512-NEXT:    movl %r14d, %eax
-; AVX512-NEXT:    addb %cl, %al
-; AVX512-NEXT:    setns %al
-; AVX512-NEXT:    addb %cl, %r14b
-; AVX512-NEXT:    jno .LBB2_115
-; AVX512-NEXT:  # %bb.114:
-; AVX512-NEXT:    addb $127, %al
-; AVX512-NEXT:    movl %eax, %r14d
-; AVX512-NEXT:  .LBB2_115:
-; AVX512-NEXT:    vpextrb $8, %xmm1, %ecx
-; AVX512-NEXT:    vpextrb $8, %xmm0, %ebp
-; AVX512-NEXT:    movl %ebp, %eax
-; AVX512-NEXT:    addb %cl, %al
-; AVX512-NEXT:    setns %al
-; AVX512-NEXT:    addb %cl, %bpl
-; AVX512-NEXT:    jno .LBB2_117
-; AVX512-NEXT:  # %bb.116:
-; AVX512-NEXT:    addb $127, %al
-; AVX512-NEXT:    movl %eax, %ebp
-; AVX512-NEXT:  .LBB2_117:
-; AVX512-NEXT:    vpextrb $7, %xmm1, %ecx
-; AVX512-NEXT:    vpextrb $7, %xmm0, %edi
-; AVX512-NEXT:    movl %edi, %eax
-; AVX512-NEXT:    addb %cl, %al
-; AVX512-NEXT:    setns %al
-; AVX512-NEXT:    addb %cl, %dil
-; AVX512-NEXT:    movl %edx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX512-NEXT:    jno .LBB2_119
-; AVX512-NEXT:  # %bb.118:
-; AVX512-NEXT:    addb $127, %al
-; AVX512-NEXT:    movl %eax, %edi
-; AVX512-NEXT:  .LBB2_119:
-; AVX512-NEXT:    vpextrb $6, %xmm1, %edx
-; AVX512-NEXT:    vpextrb $6, %xmm0, %eax
-; AVX512-NEXT:    movl %eax, %ecx
-; AVX512-NEXT:    addb %dl, %cl
-; AVX512-NEXT:    setns %cl
-; AVX512-NEXT:    addb %dl, %al
-; AVX512-NEXT:    jno .LBB2_121
-; AVX512-NEXT:  # %bb.120:
-; AVX512-NEXT:    addb $127, %cl
-; AVX512-NEXT:    movl %ecx, %eax
-; AVX512-NEXT:  .LBB2_121:
-; AVX512-NEXT:    vpextrb $5, %xmm1, %ebx
-; AVX512-NEXT:    vpextrb $5, %xmm0, %ecx
-; AVX512-NEXT:    movl %ecx, %edx
-; AVX512-NEXT:    addb %bl, %dl
-; AVX512-NEXT:    setns %dl
-; AVX512-NEXT:    addb %bl, %cl
-; AVX512-NEXT:    jno .LBB2_123
-; AVX512-NEXT:  # %bb.122:
-; AVX512-NEXT:    addb $127, %dl
-; AVX512-NEXT:    movl %edx, %ecx
-; AVX512-NEXT:  .LBB2_123:
-; AVX512-NEXT:    vpextrb $4, %xmm1, %esi
-; AVX512-NEXT:    vpextrb $4, %xmm0, %edx
-; AVX512-NEXT:    movl %edx, %ebx
-; AVX512-NEXT:    addb %sil, %bl
-; AVX512-NEXT:    setns %bl
-; AVX512-NEXT:    addb %sil, %dl
-; AVX512-NEXT:    jno .LBB2_125
-; AVX512-NEXT:  # %bb.124:
-; AVX512-NEXT:    addb $127, %bl
-; AVX512-NEXT:    movl %ebx, %edx
-; AVX512-NEXT:  .LBB2_125:
-; AVX512-NEXT:    vpextrb $3, %xmm1, %esi
-; AVX512-NEXT:    vpextrb $3, %xmm0, %r8d
-; AVX512-NEXT:    movl %r8d, %ebx
-; AVX512-NEXT:    addb %sil, %bl
-; AVX512-NEXT:    setns %bl
-; AVX512-NEXT:    addb %sil, %r8b
-; AVX512-NEXT:    movl %r9d, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX512-NEXT:    jno .LBB2_127
-; AVX512-NEXT:  # %bb.126:
-; AVX512-NEXT:    addb $127, %bl
-; AVX512-NEXT:    movl %ebx, %r8d
-; AVX512-NEXT:  .LBB2_127:
-; AVX512-NEXT:    vpextrb $2, %xmm1, %esi
-; AVX512-NEXT:    vpextrb $2, %xmm0, %r9d
-; AVX512-NEXT:    movl %r9d, %ebx
-; AVX512-NEXT:    addb %sil, %bl
-; AVX512-NEXT:    setns %bl
-; AVX512-NEXT:    addb %sil, %r9b
-; AVX512-NEXT:    jno .LBB2_129
-; AVX512-NEXT:  # %bb.128:
-; AVX512-NEXT:    addb $127, %bl
-; AVX512-NEXT:    movl %ebx, %r9d
-; AVX512-NEXT:  .LBB2_129:
-; AVX512-NEXT:    vpextrb $0, %xmm1, %esi
-; AVX512-NEXT:    vpextrb $0, %xmm0, %r10d
-; AVX512-NEXT:    movl %r10d, %ebx
-; AVX512-NEXT:    addb %sil, %bl
-; AVX512-NEXT:    setns %bl
-; AVX512-NEXT:    movl %r11d, %r12d
-; AVX512-NEXT:    addb %sil, %r10b
-; AVX512-NEXT:    jno .LBB2_131
-; AVX512-NEXT:  # %bb.130:
-; AVX512-NEXT:    addb $127, %bl
-; AVX512-NEXT:    movl %ebx, %r10d
-; AVX512-NEXT:  .LBB2_131:
-; AVX512-NEXT:    vpextrb $1, %xmm1, %esi
-; AVX512-NEXT:    vpextrb $1, %xmm0, %r11d
-; AVX512-NEXT:    movl %r11d, %ebx
-; AVX512-NEXT:    addb %sil, %bl
-; AVX512-NEXT:    setns %bl
-; AVX512-NEXT:    addb %sil, %r11b
-; AVX512-NEXT:    jno .LBB2_133
-; AVX512-NEXT:  # %bb.132:
-; AVX512-NEXT:    addb $127, %bl
-; AVX512-NEXT:    movl %ebx, %r11d
-; AVX512-NEXT:  .LBB2_133:
-; AVX512-NEXT:    movzbl %r10b, %esi
-; AVX512-NEXT:    vmovd %esi, %xmm0
-; AVX512-NEXT:    movzbl %r11b, %esi
-; AVX512-NEXT:    vpinsrb $1, %esi, %xmm0, %xmm0
-; AVX512-NEXT:    movzbl %r9b, %esi
-; AVX512-NEXT:    vpinsrb $2, %esi, %xmm0, %xmm0
-; AVX512-NEXT:    movzbl %r8b, %esi
-; AVX512-NEXT:    vpinsrb $3, %esi, %xmm0, %xmm0
-; AVX512-NEXT:    movzbl %dl, %edx
-; AVX512-NEXT:    vpinsrb $4, %edx, %xmm0, %xmm0
-; AVX512-NEXT:    movzbl %cl, %ecx
-; AVX512-NEXT:    vpinsrb $5, %ecx, %xmm0, %xmm0
-; AVX512-NEXT:    movzbl %al, %eax
-; AVX512-NEXT:    vpinsrb $6, %eax, %xmm0, %xmm0
-; AVX512-NEXT:    movzbl %dil, %eax
-; AVX512-NEXT:    vpinsrb $7, %eax, %xmm0, %xmm0
-; AVX512-NEXT:    movzbl %bpl, %eax
-; AVX512-NEXT:    vpinsrb $8, %eax, %xmm0, %xmm0
-; AVX512-NEXT:    movzbl %r14b, %eax
-; AVX512-NEXT:    vpinsrb $9, %eax, %xmm0, %xmm0
-; AVX512-NEXT:    movzbl %r15b, %eax
-; AVX512-NEXT:    vpinsrb $10, %eax, %xmm0, %xmm0
-; AVX512-NEXT:    movzbl %r13b, %eax
-; AVX512-NEXT:    vpinsrb $11, %eax, %xmm0, %xmm0
-; AVX512-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX512-NEXT:    vpinsrb $12, %eax, %xmm0, %xmm0
-; AVX512-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX512-NEXT:    vpinsrb $13, %eax, %xmm0, %xmm0
-; AVX512-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX512-NEXT:    vpinsrb $14, %eax, %xmm0, %xmm0
-; AVX512-NEXT:    movzbl %r12b, %eax
-; AVX512-NEXT:    vpinsrb $15, %eax, %xmm0, %xmm0
-; AVX512-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX512-NEXT:    vmovd %eax, %xmm1
-; AVX512-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX512-NEXT:    vpinsrb $1, %eax, %xmm1, %xmm1
-; AVX512-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX512-NEXT:    vpinsrb $2, %eax, %xmm1, %xmm1
-; AVX512-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX512-NEXT:    vpinsrb $3, %eax, %xmm1, %xmm1
-; AVX512-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX512-NEXT:    vpinsrb $4, %eax, %xmm1, %xmm1
-; AVX512-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX512-NEXT:    vpinsrb $5, %eax, %xmm1, %xmm1
-; AVX512-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX512-NEXT:    vpinsrb $6, %eax, %xmm1, %xmm1
-; AVX512-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX512-NEXT:    vpinsrb $7, %eax, %xmm1, %xmm1
-; AVX512-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX512-NEXT:    vpinsrb $8, %eax, %xmm1, %xmm1
-; AVX512-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX512-NEXT:    vpinsrb $9, %eax, %xmm1, %xmm1
-; AVX512-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX512-NEXT:    vpinsrb $10, %eax, %xmm1, %xmm1
-; AVX512-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX512-NEXT:    vpinsrb $11, %eax, %xmm1, %xmm1
-; AVX512-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX512-NEXT:    vpinsrb $12, %eax, %xmm1, %xmm1
-; AVX512-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX512-NEXT:    vpinsrb $13, %eax, %xmm1, %xmm1
-; AVX512-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX512-NEXT:    vpinsrb $14, %eax, %xmm1, %xmm1
-; AVX512-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX512-NEXT:    vpinsrb $15, %eax, %xmm1, %xmm1
-; AVX512-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX512-NEXT:    vmovd %eax, %xmm2
-; AVX512-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX512-NEXT:    vpinsrb $1, %eax, %xmm2, %xmm2
-; AVX512-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX512-NEXT:    vpinsrb $2, %eax, %xmm2, %xmm2
-; AVX512-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX512-NEXT:    vpinsrb $3, %eax, %xmm2, %xmm2
-; AVX512-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX512-NEXT:    vpinsrb $4, %eax, %xmm2, %xmm2
-; AVX512-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX512-NEXT:    vpinsrb $5, %eax, %xmm2, %xmm2
-; AVX512-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX512-NEXT:    vpinsrb $6, %eax, %xmm2, %xmm2
-; AVX512-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX512-NEXT:    vpinsrb $7, %eax, %xmm2, %xmm2
-; AVX512-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX512-NEXT:    vpinsrb $8, %eax, %xmm2, %xmm2
-; AVX512-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX512-NEXT:    vpinsrb $9, %eax, %xmm2, %xmm2
-; AVX512-NEXT:    movzbl (%rsp), %eax # 1-byte Folded Reload
-; AVX512-NEXT:    vpinsrb $10, %eax, %xmm2, %xmm2
-; AVX512-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX512-NEXT:    vpinsrb $11, %eax, %xmm2, %xmm2
-; AVX512-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX512-NEXT:    vpinsrb $12, %eax, %xmm2, %xmm2
-; AVX512-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX512-NEXT:    vpinsrb $13, %eax, %xmm2, %xmm2
-; AVX512-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX512-NEXT:    vpinsrb $14, %eax, %xmm2, %xmm2
-; AVX512-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX512-NEXT:    vpinsrb $15, %eax, %xmm2, %xmm2
-; AVX512-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX512-NEXT:    vmovd %eax, %xmm3
-; AVX512-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX512-NEXT:    vpinsrb $1, %eax, %xmm3, %xmm3
-; AVX512-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX512-NEXT:    vpinsrb $2, %eax, %xmm3, %xmm3
-; AVX512-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX512-NEXT:    vpinsrb $3, %eax, %xmm3, %xmm3
-; AVX512-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX512-NEXT:    vpinsrb $4, %eax, %xmm3, %xmm3
-; AVX512-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX512-NEXT:    vpinsrb $5, %eax, %xmm3, %xmm3
-; AVX512-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX512-NEXT:    vpinsrb $6, %eax, %xmm3, %xmm3
-; AVX512-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX512-NEXT:    vpinsrb $7, %eax, %xmm3, %xmm3
-; AVX512-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX512-NEXT:    vpinsrb $8, %eax, %xmm3, %xmm3
-; AVX512-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX512-NEXT:    vpinsrb $9, %eax, %xmm3, %xmm3
-; AVX512-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX512-NEXT:    vpinsrb $10, %eax, %xmm3, %xmm3
-; AVX512-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX512-NEXT:    vpinsrb $11, %eax, %xmm3, %xmm3
-; AVX512-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX512-NEXT:    vpinsrb $12, %eax, %xmm3, %xmm3
-; AVX512-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX512-NEXT:    vpinsrb $13, %eax, %xmm3, %xmm3
-; AVX512-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX512-NEXT:    vpinsrb $14, %eax, %xmm3, %xmm3
-; AVX512-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX512-NEXT:    vpinsrb $15, %eax, %xmm3, %xmm3
-; AVX512-NEXT:    vinserti128 $1, %xmm0, %ymm1, %ymm0
-; AVX512-NEXT:    vinserti128 $1, %xmm2, %ymm3, %ymm1
-; AVX512-NEXT:    vinserti64x4 $1, %ymm0, %zmm1, %zmm0
-; AVX512-NEXT:    addq $76, %rsp
-; AVX512-NEXT:    popq %rbx
-; AVX512-NEXT:    popq %r12
-; AVX512-NEXT:    popq %r13
-; AVX512-NEXT:    popq %r14
-; AVX512-NEXT:    popq %r15
-; AVX512-NEXT:    popq %rbp
+; AVX512-NEXT:    vpaddsb %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    retq
   %z = call <64 x i8> @llvm.sadd.sat.v64i8(<64 x i8> %x, <64 x i8> %y)
   ret <64 x i8> %z
 }
 
 define <8 x i16> @v8i16(<8 x i16> %x, <8 x i16> %y) nounwind {
-; SSE2-LABEL: v8i16:
-; SSE2:       # %bb.0:
-; SSE2-NEXT:    pushq %rbp
-; SSE2-NEXT:    pushq %rbx
-; SSE2-NEXT:    movd %xmm1, %eax
-; SSE2-NEXT:    movd %xmm0, %r8d
-; SSE2-NEXT:    xorl %ecx, %ecx
-; SSE2-NEXT:    movl %r8d, %edx
-; SSE2-NEXT:    addw %ax, %dx
-; SSE2-NEXT:    setns %cl
-; SSE2-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSE2-NEXT:    addw %ax, %r8w
-; SSE2-NEXT:    cmovol %ecx, %r8d
-; SSE2-NEXT:    pextrw $1, %xmm1, %eax
-; SSE2-NEXT:    pextrw $1, %xmm0, %r9d
-; SSE2-NEXT:    xorl %ecx, %ecx
-; SSE2-NEXT:    movl %r9d, %edx
-; SSE2-NEXT:    addw %ax, %dx
-; SSE2-NEXT:    setns %cl
-; SSE2-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSE2-NEXT:    addw %ax, %r9w
-; SSE2-NEXT:    cmovol %ecx, %r9d
-; SSE2-NEXT:    pextrw $2, %xmm1, %eax
-; SSE2-NEXT:    pextrw $2, %xmm0, %r10d
-; SSE2-NEXT:    xorl %ecx, %ecx
-; SSE2-NEXT:    movl %r10d, %esi
-; SSE2-NEXT:    addw %ax, %si
-; SSE2-NEXT:    setns %cl
-; SSE2-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSE2-NEXT:    addw %ax, %r10w
-; SSE2-NEXT:    cmovol %ecx, %r10d
-; SSE2-NEXT:    pextrw $3, %xmm1, %eax
-; SSE2-NEXT:    pextrw $3, %xmm0, %r11d
-; SSE2-NEXT:    xorl %ecx, %ecx
-; SSE2-NEXT:    movl %r11d, %edi
-; SSE2-NEXT:    addw %ax, %di
-; SSE2-NEXT:    setns %cl
-; SSE2-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSE2-NEXT:    addw %ax, %r11w
-; SSE2-NEXT:    cmovol %ecx, %r11d
-; SSE2-NEXT:    pextrw $4, %xmm1, %eax
-; SSE2-NEXT:    pextrw $4, %xmm0, %edi
-; SSE2-NEXT:    xorl %ecx, %ecx
-; SSE2-NEXT:    movl %edi, %edx
-; SSE2-NEXT:    addw %ax, %dx
-; SSE2-NEXT:    setns %cl
-; SSE2-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSE2-NEXT:    addw %ax, %di
-; SSE2-NEXT:    cmovol %ecx, %edi
-; SSE2-NEXT:    pextrw $5, %xmm1, %ecx
-; SSE2-NEXT:    pextrw $5, %xmm0, %eax
-; SSE2-NEXT:    xorl %edx, %edx
-; SSE2-NEXT:    movl %eax, %esi
-; SSE2-NEXT:    addw %cx, %si
-; SSE2-NEXT:    setns %dl
-; SSE2-NEXT:    addl $32767, %edx # imm = 0x7FFF
-; SSE2-NEXT:    addw %cx, %ax
-; SSE2-NEXT:    cmovol %edx, %eax
-; SSE2-NEXT:    pextrw $6, %xmm1, %edx
-; SSE2-NEXT:    pextrw $6, %xmm0, %ecx
-; SSE2-NEXT:    xorl %esi, %esi
-; SSE2-NEXT:    movl %ecx, %ebx
-; SSE2-NEXT:    addw %dx, %bx
-; SSE2-NEXT:    setns %sil
-; SSE2-NEXT:    addl $32767, %esi # imm = 0x7FFF
-; SSE2-NEXT:    addw %dx, %cx
-; SSE2-NEXT:    cmovol %esi, %ecx
-; SSE2-NEXT:    pextrw $7, %xmm1, %edx
-; SSE2-NEXT:    pextrw $7, %xmm0, %esi
-; SSE2-NEXT:    xorl %ebx, %ebx
-; SSE2-NEXT:    movl %esi, %ebp
-; SSE2-NEXT:    addw %dx, %bp
-; SSE2-NEXT:    setns %bl
-; SSE2-NEXT:    addl $32767, %ebx # imm = 0x7FFF
-; SSE2-NEXT:    addw %dx, %si
-; SSE2-NEXT:    cmovol %ebx, %esi
-; SSE2-NEXT:    movd %esi, %xmm0
-; SSE2-NEXT:    movd %ecx, %xmm1
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3]
-; SSE2-NEXT:    movd %eax, %xmm0
-; SSE2-NEXT:    movd %edi, %xmm2
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm2 = xmm2[0],xmm0[0],xmm2[1],xmm0[1],xmm2[2],xmm0[2],xmm2[3],xmm0[3]
-; SSE2-NEXT:    punpckldq {{.*#+}} xmm2 = xmm2[0],xmm1[0],xmm2[1],xmm1[1]
-; SSE2-NEXT:    movd %r11d, %xmm0
-; SSE2-NEXT:    movd %r10d, %xmm1
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3]
-; SSE2-NEXT:    movd %r9d, %xmm3
-; SSE2-NEXT:    movd %r8d, %xmm0
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm3[0],xmm0[1],xmm3[1],xmm0[2],xmm3[2],xmm0[3],xmm3[3]
-; SSE2-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
-; SSE2-NEXT:    punpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm2[0]
-; SSE2-NEXT:    popq %rbx
-; SSE2-NEXT:    popq %rbp
-; SSE2-NEXT:    retq
-;
-; SSSE3-LABEL: v8i16:
-; SSSE3:       # %bb.0:
-; SSSE3-NEXT:    pushq %rbp
-; SSSE3-NEXT:    pushq %rbx
-; SSSE3-NEXT:    movd %xmm1, %eax
-; SSSE3-NEXT:    movd %xmm0, %r8d
-; SSSE3-NEXT:    xorl %ecx, %ecx
-; SSSE3-NEXT:    movl %r8d, %edx
-; SSSE3-NEXT:    addw %ax, %dx
-; SSSE3-NEXT:    setns %cl
-; SSSE3-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSSE3-NEXT:    addw %ax, %r8w
-; SSSE3-NEXT:    cmovol %ecx, %r8d
-; SSSE3-NEXT:    pextrw $1, %xmm1, %eax
-; SSSE3-NEXT:    pextrw $1, %xmm0, %r9d
-; SSSE3-NEXT:    xorl %ecx, %ecx
-; SSSE3-NEXT:    movl %r9d, %edx
-; SSSE3-NEXT:    addw %ax, %dx
-; SSSE3-NEXT:    setns %cl
-; SSSE3-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSSE3-NEXT:    addw %ax, %r9w
-; SSSE3-NEXT:    cmovol %ecx, %r9d
-; SSSE3-NEXT:    pextrw $2, %xmm1, %eax
-; SSSE3-NEXT:    pextrw $2, %xmm0, %r10d
-; SSSE3-NEXT:    xorl %ecx, %ecx
-; SSSE3-NEXT:    movl %r10d, %esi
-; SSSE3-NEXT:    addw %ax, %si
-; SSSE3-NEXT:    setns %cl
-; SSSE3-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSSE3-NEXT:    addw %ax, %r10w
-; SSSE3-NEXT:    cmovol %ecx, %r10d
-; SSSE3-NEXT:    pextrw $3, %xmm1, %eax
-; SSSE3-NEXT:    pextrw $3, %xmm0, %r11d
-; SSSE3-NEXT:    xorl %ecx, %ecx
-; SSSE3-NEXT:    movl %r11d, %edi
-; SSSE3-NEXT:    addw %ax, %di
-; SSSE3-NEXT:    setns %cl
-; SSSE3-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSSE3-NEXT:    addw %ax, %r11w
-; SSSE3-NEXT:    cmovol %ecx, %r11d
-; SSSE3-NEXT:    pextrw $4, %xmm1, %eax
-; SSSE3-NEXT:    pextrw $4, %xmm0, %edi
-; SSSE3-NEXT:    xorl %ecx, %ecx
-; SSSE3-NEXT:    movl %edi, %edx
-; SSSE3-NEXT:    addw %ax, %dx
-; SSSE3-NEXT:    setns %cl
-; SSSE3-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSSE3-NEXT:    addw %ax, %di
-; SSSE3-NEXT:    cmovol %ecx, %edi
-; SSSE3-NEXT:    pextrw $5, %xmm1, %ecx
-; SSSE3-NEXT:    pextrw $5, %xmm0, %eax
-; SSSE3-NEXT:    xorl %edx, %edx
-; SSSE3-NEXT:    movl %eax, %esi
-; SSSE3-NEXT:    addw %cx, %si
-; SSSE3-NEXT:    setns %dl
-; SSSE3-NEXT:    addl $32767, %edx # imm = 0x7FFF
-; SSSE3-NEXT:    addw %cx, %ax
-; SSSE3-NEXT:    cmovol %edx, %eax
-; SSSE3-NEXT:    pextrw $6, %xmm1, %edx
-; SSSE3-NEXT:    pextrw $6, %xmm0, %ecx
-; SSSE3-NEXT:    xorl %esi, %esi
-; SSSE3-NEXT:    movl %ecx, %ebx
-; SSSE3-NEXT:    addw %dx, %bx
-; SSSE3-NEXT:    setns %sil
-; SSSE3-NEXT:    addl $32767, %esi # imm = 0x7FFF
-; SSSE3-NEXT:    addw %dx, %cx
-; SSSE3-NEXT:    cmovol %esi, %ecx
-; SSSE3-NEXT:    pextrw $7, %xmm1, %edx
-; SSSE3-NEXT:    pextrw $7, %xmm0, %esi
-; SSSE3-NEXT:    xorl %ebx, %ebx
-; SSSE3-NEXT:    movl %esi, %ebp
-; SSSE3-NEXT:    addw %dx, %bp
-; SSSE3-NEXT:    setns %bl
-; SSSE3-NEXT:    addl $32767, %ebx # imm = 0x7FFF
-; SSSE3-NEXT:    addw %dx, %si
-; SSSE3-NEXT:    cmovol %ebx, %esi
-; SSSE3-NEXT:    movd %esi, %xmm0
-; SSSE3-NEXT:    movd %ecx, %xmm1
-; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3]
-; SSSE3-NEXT:    movd %eax, %xmm0
-; SSSE3-NEXT:    movd %edi, %xmm2
-; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm2 = xmm2[0],xmm0[0],xmm2[1],xmm0[1],xmm2[2],xmm0[2],xmm2[3],xmm0[3]
-; SSSE3-NEXT:    punpckldq {{.*#+}} xmm2 = xmm2[0],xmm1[0],xmm2[1],xmm1[1]
-; SSSE3-NEXT:    movd %r11d, %xmm0
-; SSSE3-NEXT:    movd %r10d, %xmm1
-; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3]
-; SSSE3-NEXT:    movd %r9d, %xmm3
-; SSSE3-NEXT:    movd %r8d, %xmm0
-; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm3[0],xmm0[1],xmm3[1],xmm0[2],xmm3[2],xmm0[3],xmm3[3]
-; SSSE3-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
-; SSSE3-NEXT:    punpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm2[0]
-; SSSE3-NEXT:    popq %rbx
-; SSSE3-NEXT:    popq %rbp
-; SSSE3-NEXT:    retq
-;
-; SSE41-LABEL: v8i16:
-; SSE41:       # %bb.0:
-; SSE41-NEXT:    pushq %rbp
-; SSE41-NEXT:    pushq %rbx
-; SSE41-NEXT:    pextrw $7, %xmm1, %eax
-; SSE41-NEXT:    pextrw $7, %xmm0, %r8d
-; SSE41-NEXT:    xorl %ecx, %ecx
-; SSE41-NEXT:    movl %r8d, %edx
-; SSE41-NEXT:    addw %ax, %dx
-; SSE41-NEXT:    setns %cl
-; SSE41-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSE41-NEXT:    addw %ax, %r8w
-; SSE41-NEXT:    cmovol %ecx, %r8d
-; SSE41-NEXT:    pextrw $6, %xmm1, %eax
-; SSE41-NEXT:    pextrw $6, %xmm0, %r9d
-; SSE41-NEXT:    xorl %edx, %edx
-; SSE41-NEXT:    movl %r9d, %esi
-; SSE41-NEXT:    addw %ax, %si
-; SSE41-NEXT:    setns %dl
-; SSE41-NEXT:    addl $32767, %edx # imm = 0x7FFF
-; SSE41-NEXT:    addw %ax, %r9w
-; SSE41-NEXT:    cmovol %edx, %r9d
-; SSE41-NEXT:    pextrw $5, %xmm1, %eax
-; SSE41-NEXT:    pextrw $5, %xmm0, %r10d
-; SSE41-NEXT:    xorl %esi, %esi
-; SSE41-NEXT:    movl %r10d, %edi
-; SSE41-NEXT:    addw %ax, %di
-; SSE41-NEXT:    setns %sil
-; SSE41-NEXT:    addl $32767, %esi # imm = 0x7FFF
-; SSE41-NEXT:    addw %ax, %r10w
-; SSE41-NEXT:    cmovol %esi, %r10d
-; SSE41-NEXT:    pextrw $4, %xmm1, %eax
-; SSE41-NEXT:    pextrw $4, %xmm0, %r11d
-; SSE41-NEXT:    xorl %edi, %edi
-; SSE41-NEXT:    movl %r11d, %ecx
-; SSE41-NEXT:    addw %ax, %cx
-; SSE41-NEXT:    setns %dil
-; SSE41-NEXT:    addl $32767, %edi # imm = 0x7FFF
-; SSE41-NEXT:    addw %ax, %r11w
-; SSE41-NEXT:    cmovol %edi, %r11d
-; SSE41-NEXT:    pextrw $3, %xmm1, %eax
-; SSE41-NEXT:    pextrw $3, %xmm0, %edi
-; SSE41-NEXT:    xorl %ecx, %ecx
-; SSE41-NEXT:    movl %edi, %edx
-; SSE41-NEXT:    addw %ax, %dx
-; SSE41-NEXT:    setns %cl
-; SSE41-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSE41-NEXT:    addw %ax, %di
-; SSE41-NEXT:    cmovol %ecx, %edi
-; SSE41-NEXT:    pextrw $2, %xmm1, %ecx
-; SSE41-NEXT:    pextrw $2, %xmm0, %eax
-; SSE41-NEXT:    xorl %edx, %edx
-; SSE41-NEXT:    movl %eax, %esi
-; SSE41-NEXT:    addw %cx, %si
-; SSE41-NEXT:    setns %dl
-; SSE41-NEXT:    addl $32767, %edx # imm = 0x7FFF
-; SSE41-NEXT:    addw %cx, %ax
-; SSE41-NEXT:    cmovol %edx, %eax
-; SSE41-NEXT:    movd %xmm1, %ecx
-; SSE41-NEXT:    movd %xmm0, %edx
-; SSE41-NEXT:    xorl %esi, %esi
-; SSE41-NEXT:    movl %edx, %ebx
-; SSE41-NEXT:    addw %cx, %bx
-; SSE41-NEXT:    setns %sil
-; SSE41-NEXT:    addl $32767, %esi # imm = 0x7FFF
-; SSE41-NEXT:    addw %cx, %dx
-; SSE41-NEXT:    cmovol %esi, %edx
-; SSE41-NEXT:    pextrw $1, %xmm1, %ecx
-; SSE41-NEXT:    pextrw $1, %xmm0, %esi
-; SSE41-NEXT:    xorl %ebx, %ebx
-; SSE41-NEXT:    movl %esi, %ebp
-; SSE41-NEXT:    addw %cx, %bp
-; SSE41-NEXT:    setns %bl
-; SSE41-NEXT:    addl $32767, %ebx # imm = 0x7FFF
-; SSE41-NEXT:    addw %cx, %si
-; SSE41-NEXT:    cmovol %ebx, %esi
-; SSE41-NEXT:    movd %edx, %xmm0
-; SSE41-NEXT:    pinsrw $1, %esi, %xmm0
-; SSE41-NEXT:    pinsrw $2, %eax, %xmm0
-; SSE41-NEXT:    pinsrw $3, %edi, %xmm0
-; SSE41-NEXT:    pinsrw $4, %r11d, %xmm0
-; SSE41-NEXT:    pinsrw $5, %r10d, %xmm0
-; SSE41-NEXT:    pinsrw $6, %r9d, %xmm0
-; SSE41-NEXT:    pinsrw $7, %r8d, %xmm0
-; SSE41-NEXT:    popq %rbx
-; SSE41-NEXT:    popq %rbp
-; SSE41-NEXT:    retq
+; SSE-LABEL: v8i16:
+; SSE:       # %bb.0:
+; SSE-NEXT:    paddsw %xmm1, %xmm0
+; SSE-NEXT:    retq
 ;
 ; AVX-LABEL: v8i16:
 ; AVX:       # %bb.0:
-; AVX-NEXT:    pushq %rbp
-; AVX-NEXT:    pushq %rbx
-; AVX-NEXT:    vpextrw $7, %xmm1, %eax
-; AVX-NEXT:    vpextrw $7, %xmm0, %r8d
-; AVX-NEXT:    xorl %ecx, %ecx
-; AVX-NEXT:    movl %r8d, %edx
-; AVX-NEXT:    addw %ax, %dx
-; AVX-NEXT:    setns %cl
-; AVX-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX-NEXT:    addw %ax, %r8w
-; AVX-NEXT:    cmovol %ecx, %r8d
-; AVX-NEXT:    vpextrw $6, %xmm1, %eax
-; AVX-NEXT:    vpextrw $6, %xmm0, %r9d
-; AVX-NEXT:    xorl %edx, %edx
-; AVX-NEXT:    movl %r9d, %esi
-; AVX-NEXT:    addw %ax, %si
-; AVX-NEXT:    setns %dl
-; AVX-NEXT:    addl $32767, %edx # imm = 0x7FFF
-; AVX-NEXT:    addw %ax, %r9w
-; AVX-NEXT:    cmovol %edx, %r9d
-; AVX-NEXT:    vpextrw $5, %xmm1, %eax
-; AVX-NEXT:    vpextrw $5, %xmm0, %r10d
-; AVX-NEXT:    xorl %esi, %esi
-; AVX-NEXT:    movl %r10d, %edi
-; AVX-NEXT:    addw %ax, %di
-; AVX-NEXT:    setns %sil
-; AVX-NEXT:    addl $32767, %esi # imm = 0x7FFF
-; AVX-NEXT:    addw %ax, %r10w
-; AVX-NEXT:    cmovol %esi, %r10d
-; AVX-NEXT:    vpextrw $4, %xmm1, %eax
-; AVX-NEXT:    vpextrw $4, %xmm0, %r11d
-; AVX-NEXT:    xorl %edi, %edi
-; AVX-NEXT:    movl %r11d, %ecx
-; AVX-NEXT:    addw %ax, %cx
-; AVX-NEXT:    setns %dil
-; AVX-NEXT:    addl $32767, %edi # imm = 0x7FFF
-; AVX-NEXT:    addw %ax, %r11w
-; AVX-NEXT:    cmovol %edi, %r11d
-; AVX-NEXT:    vpextrw $3, %xmm1, %eax
-; AVX-NEXT:    vpextrw $3, %xmm0, %edi
-; AVX-NEXT:    xorl %ecx, %ecx
-; AVX-NEXT:    movl %edi, %edx
-; AVX-NEXT:    addw %ax, %dx
-; AVX-NEXT:    setns %cl
-; AVX-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX-NEXT:    addw %ax, %di
-; AVX-NEXT:    cmovol %ecx, %edi
-; AVX-NEXT:    vpextrw $2, %xmm1, %ecx
-; AVX-NEXT:    vpextrw $2, %xmm0, %eax
-; AVX-NEXT:    xorl %edx, %edx
-; AVX-NEXT:    movl %eax, %esi
-; AVX-NEXT:    addw %cx, %si
-; AVX-NEXT:    setns %dl
-; AVX-NEXT:    addl $32767, %edx # imm = 0x7FFF
-; AVX-NEXT:    addw %cx, %ax
-; AVX-NEXT:    cmovol %edx, %eax
-; AVX-NEXT:    vmovd %xmm1, %ecx
-; AVX-NEXT:    vmovd %xmm0, %edx
-; AVX-NEXT:    xorl %esi, %esi
-; AVX-NEXT:    movl %edx, %ebx
-; AVX-NEXT:    addw %cx, %bx
-; AVX-NEXT:    setns %sil
-; AVX-NEXT:    addl $32767, %esi # imm = 0x7FFF
-; AVX-NEXT:    addw %cx, %dx
-; AVX-NEXT:    cmovol %esi, %edx
-; AVX-NEXT:    vpextrw $1, %xmm1, %ecx
-; AVX-NEXT:    vpextrw $1, %xmm0, %esi
-; AVX-NEXT:    xorl %ebx, %ebx
-; AVX-NEXT:    movl %esi, %ebp
-; AVX-NEXT:    addw %cx, %bp
-; AVX-NEXT:    setns %bl
-; AVX-NEXT:    addl $32767, %ebx # imm = 0x7FFF
-; AVX-NEXT:    addw %cx, %si
-; AVX-NEXT:    cmovol %ebx, %esi
-; AVX-NEXT:    vmovd %edx, %xmm0
-; AVX-NEXT:    vpinsrw $1, %esi, %xmm0, %xmm0
-; AVX-NEXT:    vpinsrw $2, %eax, %xmm0, %xmm0
-; AVX-NEXT:    vpinsrw $3, %edi, %xmm0, %xmm0
-; AVX-NEXT:    vpinsrw $4, %r11d, %xmm0, %xmm0
-; AVX-NEXT:    vpinsrw $5, %r10d, %xmm0, %xmm0
-; AVX-NEXT:    vpinsrw $6, %r9d, %xmm0, %xmm0
-; AVX-NEXT:    vpinsrw $7, %r8d, %xmm0, %xmm0
-; AVX-NEXT:    popq %rbx
-; AVX-NEXT:    popq %rbp
+; AVX-NEXT:    vpaddsw %xmm1, %xmm0, %xmm0
 ; AVX-NEXT:    retq
   %z = call <8 x i16> @llvm.sadd.sat.v8i16(<8 x i16> %x, <8 x i16> %y)
   ret <8 x i16> %z
 }
 
 define <16 x i16> @v16i16(<16 x i16> %x, <16 x i16> %y) nounwind {
-; SSE2-LABEL: v16i16:
-; SSE2:       # %bb.0:
-; SSE2-NEXT:    pushq %rbp
-; SSE2-NEXT:    pushq %r15
-; SSE2-NEXT:    pushq %r14
-; SSE2-NEXT:    pushq %r13
-; SSE2-NEXT:    pushq %r12
-; SSE2-NEXT:    pushq %rbx
-; SSE2-NEXT:    movd %xmm3, %eax
-; SSE2-NEXT:    movd %xmm1, %esi
-; SSE2-NEXT:    xorl %ecx, %ecx
-; SSE2-NEXT:    movl %esi, %edx
-; SSE2-NEXT:    addw %ax, %dx
-; SSE2-NEXT:    setns %cl
-; SSE2-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSE2-NEXT:    addw %ax, %si
-; SSE2-NEXT:    cmovol %ecx, %esi
-; SSE2-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE2-NEXT:    pextrw $1, %xmm3, %eax
-; SSE2-NEXT:    pextrw $1, %xmm1, %esi
-; SSE2-NEXT:    xorl %ecx, %ecx
-; SSE2-NEXT:    movl %esi, %edx
-; SSE2-NEXT:    addw %ax, %dx
-; SSE2-NEXT:    setns %cl
-; SSE2-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSE2-NEXT:    addw %ax, %si
-; SSE2-NEXT:    cmovol %ecx, %esi
-; SSE2-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE2-NEXT:    pextrw $2, %xmm3, %eax
-; SSE2-NEXT:    pextrw $2, %xmm1, %esi
-; SSE2-NEXT:    xorl %ecx, %ecx
-; SSE2-NEXT:    movl %esi, %edx
-; SSE2-NEXT:    addw %ax, %dx
-; SSE2-NEXT:    setns %cl
-; SSE2-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSE2-NEXT:    addw %ax, %si
-; SSE2-NEXT:    cmovol %ecx, %esi
-; SSE2-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE2-NEXT:    pextrw $3, %xmm3, %eax
-; SSE2-NEXT:    pextrw $3, %xmm1, %esi
-; SSE2-NEXT:    xorl %ecx, %ecx
-; SSE2-NEXT:    movl %esi, %edx
-; SSE2-NEXT:    addw %ax, %dx
-; SSE2-NEXT:    setns %cl
-; SSE2-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSE2-NEXT:    addw %ax, %si
-; SSE2-NEXT:    cmovol %ecx, %esi
-; SSE2-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE2-NEXT:    pextrw $4, %xmm3, %eax
-; SSE2-NEXT:    pextrw $4, %xmm1, %r14d
-; SSE2-NEXT:    xorl %ecx, %ecx
-; SSE2-NEXT:    movl %r14d, %edx
-; SSE2-NEXT:    addw %ax, %dx
-; SSE2-NEXT:    setns %cl
-; SSE2-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSE2-NEXT:    addw %ax, %r14w
-; SSE2-NEXT:    cmovol %ecx, %r14d
-; SSE2-NEXT:    pextrw $5, %xmm3, %eax
-; SSE2-NEXT:    pextrw $5, %xmm1, %r15d
-; SSE2-NEXT:    xorl %ecx, %ecx
-; SSE2-NEXT:    movl %r15d, %edx
-; SSE2-NEXT:    addw %ax, %dx
-; SSE2-NEXT:    setns %cl
-; SSE2-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSE2-NEXT:    addw %ax, %r15w
-; SSE2-NEXT:    cmovol %ecx, %r15d
-; SSE2-NEXT:    pextrw $6, %xmm3, %eax
-; SSE2-NEXT:    pextrw $6, %xmm1, %r12d
-; SSE2-NEXT:    xorl %ecx, %ecx
-; SSE2-NEXT:    movl %r12d, %edx
-; SSE2-NEXT:    addw %ax, %dx
-; SSE2-NEXT:    setns %cl
-; SSE2-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSE2-NEXT:    addw %ax, %r12w
-; SSE2-NEXT:    cmovol %ecx, %r12d
-; SSE2-NEXT:    pextrw $7, %xmm3, %eax
-; SSE2-NEXT:    pextrw $7, %xmm1, %r13d
-; SSE2-NEXT:    xorl %ecx, %ecx
-; SSE2-NEXT:    movl %r13d, %edx
-; SSE2-NEXT:    addw %ax, %dx
-; SSE2-NEXT:    setns %cl
-; SSE2-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSE2-NEXT:    addw %ax, %r13w
-; SSE2-NEXT:    cmovol %ecx, %r13d
-; SSE2-NEXT:    movd %xmm2, %eax
-; SSE2-NEXT:    movd %xmm0, %esi
-; SSE2-NEXT:    xorl %ecx, %ecx
-; SSE2-NEXT:    movl %esi, %edx
-; SSE2-NEXT:    addw %ax, %dx
-; SSE2-NEXT:    setns %cl
-; SSE2-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSE2-NEXT:    addw %ax, %si
-; SSE2-NEXT:    cmovol %ecx, %esi
-; SSE2-NEXT:    pextrw $1, %xmm2, %eax
-; SSE2-NEXT:    pextrw $1, %xmm0, %ebx
-; SSE2-NEXT:    xorl %ecx, %ecx
-; SSE2-NEXT:    movl %ebx, %edx
-; SSE2-NEXT:    addw %ax, %dx
-; SSE2-NEXT:    setns %cl
-; SSE2-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSE2-NEXT:    addw %ax, %bx
-; SSE2-NEXT:    cmovol %ecx, %ebx
-; SSE2-NEXT:    pextrw $2, %xmm2, %eax
-; SSE2-NEXT:    pextrw $2, %xmm0, %ebp
-; SSE2-NEXT:    xorl %ecx, %ecx
-; SSE2-NEXT:    movl %ebp, %edx
-; SSE2-NEXT:    addw %ax, %dx
-; SSE2-NEXT:    setns %cl
-; SSE2-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSE2-NEXT:    addw %ax, %bp
-; SSE2-NEXT:    cmovol %ecx, %ebp
-; SSE2-NEXT:    pextrw $3, %xmm2, %eax
-; SSE2-NEXT:    pextrw $3, %xmm0, %edi
-; SSE2-NEXT:    xorl %ecx, %ecx
-; SSE2-NEXT:    movl %edi, %edx
-; SSE2-NEXT:    addw %ax, %dx
-; SSE2-NEXT:    setns %cl
-; SSE2-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSE2-NEXT:    addw %ax, %di
-; SSE2-NEXT:    cmovol %ecx, %edi
-; SSE2-NEXT:    pextrw $4, %xmm2, %ecx
-; SSE2-NEXT:    pextrw $4, %xmm0, %eax
-; SSE2-NEXT:    xorl %edx, %edx
-; SSE2-NEXT:    movl %eax, %r8d
-; SSE2-NEXT:    addw %cx, %r8w
-; SSE2-NEXT:    setns %dl
-; SSE2-NEXT:    addl $32767, %edx # imm = 0x7FFF
-; SSE2-NEXT:    addw %cx, %ax
-; SSE2-NEXT:    cmovol %edx, %eax
-; SSE2-NEXT:    pextrw $5, %xmm2, %r8d
-; SSE2-NEXT:    pextrw $5, %xmm0, %ecx
-; SSE2-NEXT:    xorl %edx, %edx
-; SSE2-NEXT:    movl %ecx, %r9d
-; SSE2-NEXT:    addw %r8w, %r9w
-; SSE2-NEXT:    setns %dl
-; SSE2-NEXT:    addl $32767, %edx # imm = 0x7FFF
-; SSE2-NEXT:    addw %r8w, %cx
-; SSE2-NEXT:    cmovol %edx, %ecx
-; SSE2-NEXT:    pextrw $6, %xmm2, %r8d
-; SSE2-NEXT:    pextrw $6, %xmm0, %r9d
-; SSE2-NEXT:    xorl %edx, %edx
-; SSE2-NEXT:    movl %r9d, %r10d
-; SSE2-NEXT:    addw %r8w, %r10w
-; SSE2-NEXT:    setns %dl
-; SSE2-NEXT:    addl $32767, %edx # imm = 0x7FFF
-; SSE2-NEXT:    addw %r8w, %r9w
-; SSE2-NEXT:    cmovol %edx, %r9d
-; SSE2-NEXT:    pextrw $7, %xmm2, %r8d
-; SSE2-NEXT:    pextrw $7, %xmm0, %edx
-; SSE2-NEXT:    xorl %r10d, %r10d
-; SSE2-NEXT:    movl %edx, %r11d
-; SSE2-NEXT:    addw %r8w, %r11w
-; SSE2-NEXT:    setns %r10b
-; SSE2-NEXT:    addl $32767, %r10d # imm = 0x7FFF
-; SSE2-NEXT:    addw %r8w, %dx
-; SSE2-NEXT:    cmovol %r10d, %edx
-; SSE2-NEXT:    movd %edx, %xmm8
-; SSE2-NEXT:    movd %r9d, %xmm3
-; SSE2-NEXT:    movd %ecx, %xmm9
-; SSE2-NEXT:    movd %eax, %xmm4
-; SSE2-NEXT:    movd %edi, %xmm10
-; SSE2-NEXT:    movd %ebp, %xmm7
-; SSE2-NEXT:    movd %ebx, %xmm11
-; SSE2-NEXT:    movd %esi, %xmm0
-; SSE2-NEXT:    movd %r13d, %xmm12
-; SSE2-NEXT:    movd %r12d, %xmm6
-; SSE2-NEXT:    movd %r15d, %xmm13
-; SSE2-NEXT:    movd %r14d, %xmm5
-; SSE2-NEXT:    movd {{[-0-9]+}}(%r{{[sb]}}p), %xmm14 # 4-byte Folded Reload
-; SSE2-NEXT:    # xmm14 = mem[0],zero,zero,zero
-; SSE2-NEXT:    movd {{[-0-9]+}}(%r{{[sb]}}p), %xmm2 # 4-byte Folded Reload
-; SSE2-NEXT:    # xmm2 = mem[0],zero,zero,zero
-; SSE2-NEXT:    movd {{[-0-9]+}}(%r{{[sb]}}p), %xmm15 # 4-byte Folded Reload
-; SSE2-NEXT:    # xmm15 = mem[0],zero,zero,zero
-; SSE2-NEXT:    movd {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 4-byte Folded Reload
-; SSE2-NEXT:    # xmm1 = mem[0],zero,zero,zero
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm3 = xmm3[0],xmm8[0],xmm3[1],xmm8[1],xmm3[2],xmm8[2],xmm3[3],xmm8[3]
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm4 = xmm4[0],xmm9[0],xmm4[1],xmm9[1],xmm4[2],xmm9[2],xmm4[3],xmm9[3]
-; SSE2-NEXT:    punpckldq {{.*#+}} xmm4 = xmm4[0],xmm3[0],xmm4[1],xmm3[1]
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm7 = xmm7[0],xmm10[0],xmm7[1],xmm10[1],xmm7[2],xmm10[2],xmm7[3],xmm10[3]
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm11[0],xmm0[1],xmm11[1],xmm0[2],xmm11[2],xmm0[3],xmm11[3]
-; SSE2-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm7[0],xmm0[1],xmm7[1]
-; SSE2-NEXT:    punpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm4[0]
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm6 = xmm6[0],xmm12[0],xmm6[1],xmm12[1],xmm6[2],xmm12[2],xmm6[3],xmm12[3]
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm5 = xmm5[0],xmm13[0],xmm5[1],xmm13[1],xmm5[2],xmm13[2],xmm5[3],xmm13[3]
-; SSE2-NEXT:    punpckldq {{.*#+}} xmm5 = xmm5[0],xmm6[0],xmm5[1],xmm6[1]
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm2 = xmm2[0],xmm14[0],xmm2[1],xmm14[1],xmm2[2],xmm14[2],xmm2[3],xmm14[3]
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm15[0],xmm1[1],xmm15[1],xmm1[2],xmm15[2],xmm1[3],xmm15[3]
-; SSE2-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm2[0],xmm1[1],xmm2[1]
-; SSE2-NEXT:    punpcklqdq {{.*#+}} xmm1 = xmm1[0],xmm5[0]
-; SSE2-NEXT:    popq %rbx
-; SSE2-NEXT:    popq %r12
-; SSE2-NEXT:    popq %r13
-; SSE2-NEXT:    popq %r14
-; SSE2-NEXT:    popq %r15
-; SSE2-NEXT:    popq %rbp
-; SSE2-NEXT:    retq
-;
-; SSSE3-LABEL: v16i16:
-; SSSE3:       # %bb.0:
-; SSSE3-NEXT:    pushq %rbp
-; SSSE3-NEXT:    pushq %r15
-; SSSE3-NEXT:    pushq %r14
-; SSSE3-NEXT:    pushq %r13
-; SSSE3-NEXT:    pushq %r12
-; SSSE3-NEXT:    pushq %rbx
-; SSSE3-NEXT:    movd %xmm3, %eax
-; SSSE3-NEXT:    movd %xmm1, %esi
-; SSSE3-NEXT:    xorl %ecx, %ecx
-; SSSE3-NEXT:    movl %esi, %edx
-; SSSE3-NEXT:    addw %ax, %dx
-; SSSE3-NEXT:    setns %cl
-; SSSE3-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSSE3-NEXT:    addw %ax, %si
-; SSSE3-NEXT:    cmovol %ecx, %esi
-; SSSE3-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSSE3-NEXT:    pextrw $1, %xmm3, %eax
-; SSSE3-NEXT:    pextrw $1, %xmm1, %esi
-; SSSE3-NEXT:    xorl %ecx, %ecx
-; SSSE3-NEXT:    movl %esi, %edx
-; SSSE3-NEXT:    addw %ax, %dx
-; SSSE3-NEXT:    setns %cl
-; SSSE3-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSSE3-NEXT:    addw %ax, %si
-; SSSE3-NEXT:    cmovol %ecx, %esi
-; SSSE3-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSSE3-NEXT:    pextrw $2, %xmm3, %eax
-; SSSE3-NEXT:    pextrw $2, %xmm1, %esi
-; SSSE3-NEXT:    xorl %ecx, %ecx
-; SSSE3-NEXT:    movl %esi, %edx
-; SSSE3-NEXT:    addw %ax, %dx
-; SSSE3-NEXT:    setns %cl
-; SSSE3-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSSE3-NEXT:    addw %ax, %si
-; SSSE3-NEXT:    cmovol %ecx, %esi
-; SSSE3-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSSE3-NEXT:    pextrw $3, %xmm3, %eax
-; SSSE3-NEXT:    pextrw $3, %xmm1, %esi
-; SSSE3-NEXT:    xorl %ecx, %ecx
-; SSSE3-NEXT:    movl %esi, %edx
-; SSSE3-NEXT:    addw %ax, %dx
-; SSSE3-NEXT:    setns %cl
-; SSSE3-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSSE3-NEXT:    addw %ax, %si
-; SSSE3-NEXT:    cmovol %ecx, %esi
-; SSSE3-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSSE3-NEXT:    pextrw $4, %xmm3, %eax
-; SSSE3-NEXT:    pextrw $4, %xmm1, %r14d
-; SSSE3-NEXT:    xorl %ecx, %ecx
-; SSSE3-NEXT:    movl %r14d, %edx
-; SSSE3-NEXT:    addw %ax, %dx
-; SSSE3-NEXT:    setns %cl
-; SSSE3-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSSE3-NEXT:    addw %ax, %r14w
-; SSSE3-NEXT:    cmovol %ecx, %r14d
-; SSSE3-NEXT:    pextrw $5, %xmm3, %eax
-; SSSE3-NEXT:    pextrw $5, %xmm1, %r15d
-; SSSE3-NEXT:    xorl %ecx, %ecx
-; SSSE3-NEXT:    movl %r15d, %edx
-; SSSE3-NEXT:    addw %ax, %dx
-; SSSE3-NEXT:    setns %cl
-; SSSE3-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSSE3-NEXT:    addw %ax, %r15w
-; SSSE3-NEXT:    cmovol %ecx, %r15d
-; SSSE3-NEXT:    pextrw $6, %xmm3, %eax
-; SSSE3-NEXT:    pextrw $6, %xmm1, %r12d
-; SSSE3-NEXT:    xorl %ecx, %ecx
-; SSSE3-NEXT:    movl %r12d, %edx
-; SSSE3-NEXT:    addw %ax, %dx
-; SSSE3-NEXT:    setns %cl
-; SSSE3-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSSE3-NEXT:    addw %ax, %r12w
-; SSSE3-NEXT:    cmovol %ecx, %r12d
-; SSSE3-NEXT:    pextrw $7, %xmm3, %eax
-; SSSE3-NEXT:    pextrw $7, %xmm1, %r13d
-; SSSE3-NEXT:    xorl %ecx, %ecx
-; SSSE3-NEXT:    movl %r13d, %edx
-; SSSE3-NEXT:    addw %ax, %dx
-; SSSE3-NEXT:    setns %cl
-; SSSE3-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSSE3-NEXT:    addw %ax, %r13w
-; SSSE3-NEXT:    cmovol %ecx, %r13d
-; SSSE3-NEXT:    movd %xmm2, %eax
-; SSSE3-NEXT:    movd %xmm0, %esi
-; SSSE3-NEXT:    xorl %ecx, %ecx
-; SSSE3-NEXT:    movl %esi, %edx
-; SSSE3-NEXT:    addw %ax, %dx
-; SSSE3-NEXT:    setns %cl
-; SSSE3-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSSE3-NEXT:    addw %ax, %si
-; SSSE3-NEXT:    cmovol %ecx, %esi
-; SSSE3-NEXT:    pextrw $1, %xmm2, %eax
-; SSSE3-NEXT:    pextrw $1, %xmm0, %ebx
-; SSSE3-NEXT:    xorl %ecx, %ecx
-; SSSE3-NEXT:    movl %ebx, %edx
-; SSSE3-NEXT:    addw %ax, %dx
-; SSSE3-NEXT:    setns %cl
-; SSSE3-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSSE3-NEXT:    addw %ax, %bx
-; SSSE3-NEXT:    cmovol %ecx, %ebx
-; SSSE3-NEXT:    pextrw $2, %xmm2, %eax
-; SSSE3-NEXT:    pextrw $2, %xmm0, %ebp
-; SSSE3-NEXT:    xorl %ecx, %ecx
-; SSSE3-NEXT:    movl %ebp, %edx
-; SSSE3-NEXT:    addw %ax, %dx
-; SSSE3-NEXT:    setns %cl
-; SSSE3-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSSE3-NEXT:    addw %ax, %bp
-; SSSE3-NEXT:    cmovol %ecx, %ebp
-; SSSE3-NEXT:    pextrw $3, %xmm2, %eax
-; SSSE3-NEXT:    pextrw $3, %xmm0, %edi
-; SSSE3-NEXT:    xorl %ecx, %ecx
-; SSSE3-NEXT:    movl %edi, %edx
-; SSSE3-NEXT:    addw %ax, %dx
-; SSSE3-NEXT:    setns %cl
-; SSSE3-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSSE3-NEXT:    addw %ax, %di
-; SSSE3-NEXT:    cmovol %ecx, %edi
-; SSSE3-NEXT:    pextrw $4, %xmm2, %ecx
-; SSSE3-NEXT:    pextrw $4, %xmm0, %eax
-; SSSE3-NEXT:    xorl %edx, %edx
-; SSSE3-NEXT:    movl %eax, %r8d
-; SSSE3-NEXT:    addw %cx, %r8w
-; SSSE3-NEXT:    setns %dl
-; SSSE3-NEXT:    addl $32767, %edx # imm = 0x7FFF
-; SSSE3-NEXT:    addw %cx, %ax
-; SSSE3-NEXT:    cmovol %edx, %eax
-; SSSE3-NEXT:    pextrw $5, %xmm2, %r8d
-; SSSE3-NEXT:    pextrw $5, %xmm0, %ecx
-; SSSE3-NEXT:    xorl %edx, %edx
-; SSSE3-NEXT:    movl %ecx, %r9d
-; SSSE3-NEXT:    addw %r8w, %r9w
-; SSSE3-NEXT:    setns %dl
-; SSSE3-NEXT:    addl $32767, %edx # imm = 0x7FFF
-; SSSE3-NEXT:    addw %r8w, %cx
-; SSSE3-NEXT:    cmovol %edx, %ecx
-; SSSE3-NEXT:    pextrw $6, %xmm2, %r8d
-; SSSE3-NEXT:    pextrw $6, %xmm0, %r9d
-; SSSE3-NEXT:    xorl %edx, %edx
-; SSSE3-NEXT:    movl %r9d, %r10d
-; SSSE3-NEXT:    addw %r8w, %r10w
-; SSSE3-NEXT:    setns %dl
-; SSSE3-NEXT:    addl $32767, %edx # imm = 0x7FFF
-; SSSE3-NEXT:    addw %r8w, %r9w
-; SSSE3-NEXT:    cmovol %edx, %r9d
-; SSSE3-NEXT:    pextrw $7, %xmm2, %r8d
-; SSSE3-NEXT:    pextrw $7, %xmm0, %edx
-; SSSE3-NEXT:    xorl %r10d, %r10d
-; SSSE3-NEXT:    movl %edx, %r11d
-; SSSE3-NEXT:    addw %r8w, %r11w
-; SSSE3-NEXT:    setns %r10b
-; SSSE3-NEXT:    addl $32767, %r10d # imm = 0x7FFF
-; SSSE3-NEXT:    addw %r8w, %dx
-; SSSE3-NEXT:    cmovol %r10d, %edx
-; SSSE3-NEXT:    movd %edx, %xmm8
-; SSSE3-NEXT:    movd %r9d, %xmm3
-; SSSE3-NEXT:    movd %ecx, %xmm9
-; SSSE3-NEXT:    movd %eax, %xmm4
-; SSSE3-NEXT:    movd %edi, %xmm10
-; SSSE3-NEXT:    movd %ebp, %xmm7
-; SSSE3-NEXT:    movd %ebx, %xmm11
-; SSSE3-NEXT:    movd %esi, %xmm0
-; SSSE3-NEXT:    movd %r13d, %xmm12
-; SSSE3-NEXT:    movd %r12d, %xmm6
-; SSSE3-NEXT:    movd %r15d, %xmm13
-; SSSE3-NEXT:    movd %r14d, %xmm5
-; SSSE3-NEXT:    movd {{[-0-9]+}}(%r{{[sb]}}p), %xmm14 # 4-byte Folded Reload
-; SSSE3-NEXT:    # xmm14 = mem[0],zero,zero,zero
-; SSSE3-NEXT:    movd {{[-0-9]+}}(%r{{[sb]}}p), %xmm2 # 4-byte Folded Reload
-; SSSE3-NEXT:    # xmm2 = mem[0],zero,zero,zero
-; SSSE3-NEXT:    movd {{[-0-9]+}}(%r{{[sb]}}p), %xmm15 # 4-byte Folded Reload
-; SSSE3-NEXT:    # xmm15 = mem[0],zero,zero,zero
-; SSSE3-NEXT:    movd {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 4-byte Folded Reload
-; SSSE3-NEXT:    # xmm1 = mem[0],zero,zero,zero
-; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm3 = xmm3[0],xmm8[0],xmm3[1],xmm8[1],xmm3[2],xmm8[2],xmm3[3],xmm8[3]
-; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm4 = xmm4[0],xmm9[0],xmm4[1],xmm9[1],xmm4[2],xmm9[2],xmm4[3],xmm9[3]
-; SSSE3-NEXT:    punpckldq {{.*#+}} xmm4 = xmm4[0],xmm3[0],xmm4[1],xmm3[1]
-; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm7 = xmm7[0],xmm10[0],xmm7[1],xmm10[1],xmm7[2],xmm10[2],xmm7[3],xmm10[3]
-; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm11[0],xmm0[1],xmm11[1],xmm0[2],xmm11[2],xmm0[3],xmm11[3]
-; SSSE3-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm7[0],xmm0[1],xmm7[1]
-; SSSE3-NEXT:    punpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm4[0]
-; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm6 = xmm6[0],xmm12[0],xmm6[1],xmm12[1],xmm6[2],xmm12[2],xmm6[3],xmm12[3]
-; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm5 = xmm5[0],xmm13[0],xmm5[1],xmm13[1],xmm5[2],xmm13[2],xmm5[3],xmm13[3]
-; SSSE3-NEXT:    punpckldq {{.*#+}} xmm5 = xmm5[0],xmm6[0],xmm5[1],xmm6[1]
-; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm2 = xmm2[0],xmm14[0],xmm2[1],xmm14[1],xmm2[2],xmm14[2],xmm2[3],xmm14[3]
-; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm15[0],xmm1[1],xmm15[1],xmm1[2],xmm15[2],xmm1[3],xmm15[3]
-; SSSE3-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm2[0],xmm1[1],xmm2[1]
-; SSSE3-NEXT:    punpcklqdq {{.*#+}} xmm1 = xmm1[0],xmm5[0]
-; SSSE3-NEXT:    popq %rbx
-; SSSE3-NEXT:    popq %r12
-; SSSE3-NEXT:    popq %r13
-; SSSE3-NEXT:    popq %r14
-; SSSE3-NEXT:    popq %r15
-; SSSE3-NEXT:    popq %rbp
-; SSSE3-NEXT:    retq
-;
-; SSE41-LABEL: v16i16:
-; SSE41:       # %bb.0:
-; SSE41-NEXT:    pushq %rbp
-; SSE41-NEXT:    pushq %r15
-; SSE41-NEXT:    pushq %r14
-; SSE41-NEXT:    pushq %r13
-; SSE41-NEXT:    pushq %r12
-; SSE41-NEXT:    pushq %rbx
-; SSE41-NEXT:    pextrw $7, %xmm3, %eax
-; SSE41-NEXT:    pextrw $7, %xmm1, %esi
-; SSE41-NEXT:    xorl %ecx, %ecx
-; SSE41-NEXT:    movl %esi, %edx
-; SSE41-NEXT:    addw %ax, %dx
-; SSE41-NEXT:    setns %cl
-; SSE41-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSE41-NEXT:    addw %ax, %si
-; SSE41-NEXT:    cmovol %ecx, %esi
-; SSE41-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE41-NEXT:    pextrw $6, %xmm3, %eax
-; SSE41-NEXT:    pextrw $6, %xmm1, %esi
-; SSE41-NEXT:    xorl %ecx, %ecx
-; SSE41-NEXT:    movl %esi, %edx
-; SSE41-NEXT:    addw %ax, %dx
-; SSE41-NEXT:    setns %cl
-; SSE41-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSE41-NEXT:    addw %ax, %si
-; SSE41-NEXT:    cmovol %ecx, %esi
-; SSE41-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE41-NEXT:    pextrw $5, %xmm3, %eax
-; SSE41-NEXT:    pextrw $5, %xmm1, %esi
-; SSE41-NEXT:    xorl %ecx, %ecx
-; SSE41-NEXT:    movl %esi, %edx
-; SSE41-NEXT:    addw %ax, %dx
-; SSE41-NEXT:    setns %cl
-; SSE41-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSE41-NEXT:    addw %ax, %si
-; SSE41-NEXT:    cmovol %ecx, %esi
-; SSE41-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE41-NEXT:    pextrw $4, %xmm3, %eax
-; SSE41-NEXT:    pextrw $4, %xmm1, %esi
-; SSE41-NEXT:    xorl %ecx, %ecx
-; SSE41-NEXT:    movl %esi, %edx
-; SSE41-NEXT:    addw %ax, %dx
-; SSE41-NEXT:    setns %cl
-; SSE41-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSE41-NEXT:    addw %ax, %si
-; SSE41-NEXT:    cmovol %ecx, %esi
-; SSE41-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE41-NEXT:    pextrw $3, %xmm3, %eax
-; SSE41-NEXT:    pextrw $3, %xmm1, %r14d
-; SSE41-NEXT:    xorl %ecx, %ecx
-; SSE41-NEXT:    movl %r14d, %edx
-; SSE41-NEXT:    addw %ax, %dx
-; SSE41-NEXT:    setns %cl
-; SSE41-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSE41-NEXT:    addw %ax, %r14w
-; SSE41-NEXT:    cmovol %ecx, %r14d
-; SSE41-NEXT:    pextrw $2, %xmm3, %eax
-; SSE41-NEXT:    pextrw $2, %xmm1, %r15d
-; SSE41-NEXT:    xorl %ecx, %ecx
-; SSE41-NEXT:    movl %r15d, %edx
-; SSE41-NEXT:    addw %ax, %dx
-; SSE41-NEXT:    setns %cl
-; SSE41-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSE41-NEXT:    addw %ax, %r15w
-; SSE41-NEXT:    cmovol %ecx, %r15d
-; SSE41-NEXT:    movd %xmm3, %eax
-; SSE41-NEXT:    movd %xmm1, %r12d
-; SSE41-NEXT:    xorl %ecx, %ecx
-; SSE41-NEXT:    movl %r12d, %edx
-; SSE41-NEXT:    addw %ax, %dx
-; SSE41-NEXT:    setns %cl
-; SSE41-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSE41-NEXT:    addw %ax, %r12w
-; SSE41-NEXT:    cmovol %ecx, %r12d
-; SSE41-NEXT:    pextrw $1, %xmm3, %eax
-; SSE41-NEXT:    pextrw $1, %xmm1, %r13d
-; SSE41-NEXT:    xorl %ecx, %ecx
-; SSE41-NEXT:    movl %r13d, %esi
-; SSE41-NEXT:    addw %ax, %si
-; SSE41-NEXT:    setns %cl
-; SSE41-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSE41-NEXT:    addw %ax, %r13w
-; SSE41-NEXT:    cmovol %ecx, %r13d
-; SSE41-NEXT:    pextrw $7, %xmm2, %eax
-; SSE41-NEXT:    pextrw $7, %xmm0, %esi
-; SSE41-NEXT:    xorl %ecx, %ecx
-; SSE41-NEXT:    movl %esi, %edi
-; SSE41-NEXT:    addw %ax, %di
-; SSE41-NEXT:    setns %cl
-; SSE41-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSE41-NEXT:    addw %ax, %si
-; SSE41-NEXT:    cmovol %ecx, %esi
-; SSE41-NEXT:    pextrw $6, %xmm2, %eax
-; SSE41-NEXT:    pextrw $6, %xmm0, %ebx
-; SSE41-NEXT:    xorl %ecx, %ecx
-; SSE41-NEXT:    movl %ebx, %edi
-; SSE41-NEXT:    addw %ax, %di
-; SSE41-NEXT:    setns %cl
-; SSE41-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSE41-NEXT:    addw %ax, %bx
-; SSE41-NEXT:    cmovol %ecx, %ebx
-; SSE41-NEXT:    pextrw $5, %xmm2, %eax
-; SSE41-NEXT:    pextrw $5, %xmm0, %ebp
-; SSE41-NEXT:    xorl %ecx, %ecx
-; SSE41-NEXT:    movl %ebp, %edi
-; SSE41-NEXT:    addw %ax, %di
-; SSE41-NEXT:    setns %cl
-; SSE41-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSE41-NEXT:    addw %ax, %bp
-; SSE41-NEXT:    cmovol %ecx, %ebp
-; SSE41-NEXT:    pextrw $4, %xmm2, %eax
-; SSE41-NEXT:    pextrw $4, %xmm0, %edi
-; SSE41-NEXT:    xorl %ecx, %ecx
-; SSE41-NEXT:    movl %edi, %edx
-; SSE41-NEXT:    addw %ax, %dx
-; SSE41-NEXT:    setns %cl
-; SSE41-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSE41-NEXT:    addw %ax, %di
-; SSE41-NEXT:    cmovol %ecx, %edi
-; SSE41-NEXT:    pextrw $3, %xmm2, %ecx
-; SSE41-NEXT:    pextrw $3, %xmm0, %eax
-; SSE41-NEXT:    xorl %edx, %edx
-; SSE41-NEXT:    movl %eax, %r8d
-; SSE41-NEXT:    addw %cx, %r8w
-; SSE41-NEXT:    setns %dl
-; SSE41-NEXT:    addl $32767, %edx # imm = 0x7FFF
-; SSE41-NEXT:    addw %cx, %ax
-; SSE41-NEXT:    cmovol %edx, %eax
-; SSE41-NEXT:    pextrw $2, %xmm2, %r8d
-; SSE41-NEXT:    pextrw $2, %xmm0, %ecx
-; SSE41-NEXT:    xorl %edx, %edx
-; SSE41-NEXT:    movl %ecx, %r9d
-; SSE41-NEXT:    addw %r8w, %r9w
-; SSE41-NEXT:    setns %dl
-; SSE41-NEXT:    addl $32767, %edx # imm = 0x7FFF
-; SSE41-NEXT:    addw %r8w, %cx
-; SSE41-NEXT:    cmovol %edx, %ecx
-; SSE41-NEXT:    movd %xmm2, %r8d
-; SSE41-NEXT:    movd %xmm0, %r9d
-; SSE41-NEXT:    xorl %edx, %edx
-; SSE41-NEXT:    movl %r9d, %r10d
-; SSE41-NEXT:    addw %r8w, %r10w
-; SSE41-NEXT:    setns %dl
-; SSE41-NEXT:    addl $32767, %edx # imm = 0x7FFF
-; SSE41-NEXT:    addw %r8w, %r9w
-; SSE41-NEXT:    cmovol %edx, %r9d
-; SSE41-NEXT:    pextrw $1, %xmm2, %r8d
-; SSE41-NEXT:    pextrw $1, %xmm0, %edx
-; SSE41-NEXT:    xorl %r10d, %r10d
-; SSE41-NEXT:    movl %edx, %r11d
-; SSE41-NEXT:    addw %r8w, %r11w
-; SSE41-NEXT:    setns %r10b
-; SSE41-NEXT:    addl $32767, %r10d # imm = 0x7FFF
-; SSE41-NEXT:    addw %r8w, %dx
-; SSE41-NEXT:    cmovol %r10d, %edx
-; SSE41-NEXT:    movd %r9d, %xmm0
-; SSE41-NEXT:    pinsrw $1, %edx, %xmm0
-; SSE41-NEXT:    pinsrw $2, %ecx, %xmm0
-; SSE41-NEXT:    pinsrw $3, %eax, %xmm0
-; SSE41-NEXT:    pinsrw $4, %edi, %xmm0
-; SSE41-NEXT:    pinsrw $5, %ebp, %xmm0
-; SSE41-NEXT:    pinsrw $6, %ebx, %xmm0
-; SSE41-NEXT:    pinsrw $7, %esi, %xmm0
-; SSE41-NEXT:    movd %r12d, %xmm1
-; SSE41-NEXT:    pinsrw $1, %r13d, %xmm1
-; SSE41-NEXT:    pinsrw $2, %r15d, %xmm1
-; SSE41-NEXT:    pinsrw $3, %r14d, %xmm1
-; SSE41-NEXT:    pinsrw $4, {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 4-byte Folded Reload
-; SSE41-NEXT:    pinsrw $5, {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 4-byte Folded Reload
-; SSE41-NEXT:    pinsrw $6, {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 4-byte Folded Reload
-; SSE41-NEXT:    pinsrw $7, {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 4-byte Folded Reload
-; SSE41-NEXT:    popq %rbx
-; SSE41-NEXT:    popq %r12
-; SSE41-NEXT:    popq %r13
-; SSE41-NEXT:    popq %r14
-; SSE41-NEXT:    popq %r15
-; SSE41-NEXT:    popq %rbp
-; SSE41-NEXT:    retq
+; SSE-LABEL: v16i16:
+; SSE:       # %bb.0:
+; SSE-NEXT:    paddsw %xmm2, %xmm0
+; SSE-NEXT:    paddsw %xmm3, %xmm1
+; SSE-NEXT:    retq
 ;
 ; AVX1-LABEL: v16i16:
 ; AVX1:       # %bb.0:
-; AVX1-NEXT:    pushq %rbp
-; AVX1-NEXT:    pushq %r15
-; AVX1-NEXT:    pushq %r14
-; AVX1-NEXT:    pushq %r13
-; AVX1-NEXT:    pushq %r12
-; AVX1-NEXT:    pushq %rbx
-; AVX1-NEXT:    vpextrw $7, %xmm1, %eax
-; AVX1-NEXT:    vpextrw $7, %xmm0, %esi
-; AVX1-NEXT:    xorl %ecx, %ecx
-; AVX1-NEXT:    movl %esi, %edx
-; AVX1-NEXT:    addw %ax, %dx
-; AVX1-NEXT:    setns %cl
-; AVX1-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX1-NEXT:    addw %ax, %si
-; AVX1-NEXT:    cmovol %ecx, %esi
-; AVX1-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX1-NEXT:    vpextrw $6, %xmm1, %eax
-; AVX1-NEXT:    vpextrw $6, %xmm0, %esi
-; AVX1-NEXT:    xorl %ecx, %ecx
-; AVX1-NEXT:    movl %esi, %edx
-; AVX1-NEXT:    addw %ax, %dx
-; AVX1-NEXT:    setns %cl
-; AVX1-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX1-NEXT:    addw %ax, %si
-; AVX1-NEXT:    cmovol %ecx, %esi
-; AVX1-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX1-NEXT:    vpextrw $5, %xmm1, %eax
-; AVX1-NEXT:    vpextrw $5, %xmm0, %esi
-; AVX1-NEXT:    xorl %ecx, %ecx
-; AVX1-NEXT:    movl %esi, %edx
-; AVX1-NEXT:    addw %ax, %dx
-; AVX1-NEXT:    setns %cl
-; AVX1-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX1-NEXT:    addw %ax, %si
-; AVX1-NEXT:    cmovol %ecx, %esi
-; AVX1-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX1-NEXT:    vpextrw $4, %xmm1, %eax
-; AVX1-NEXT:    vpextrw $4, %xmm0, %esi
-; AVX1-NEXT:    xorl %ecx, %ecx
-; AVX1-NEXT:    movl %esi, %edx
-; AVX1-NEXT:    addw %ax, %dx
-; AVX1-NEXT:    setns %cl
-; AVX1-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX1-NEXT:    addw %ax, %si
-; AVX1-NEXT:    cmovol %ecx, %esi
-; AVX1-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX1-NEXT:    vpextrw $3, %xmm1, %eax
-; AVX1-NEXT:    vpextrw $3, %xmm0, %r14d
-; AVX1-NEXT:    xorl %ecx, %ecx
-; AVX1-NEXT:    movl %r14d, %edx
-; AVX1-NEXT:    addw %ax, %dx
-; AVX1-NEXT:    setns %cl
-; AVX1-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX1-NEXT:    addw %ax, %r14w
-; AVX1-NEXT:    cmovol %ecx, %r14d
-; AVX1-NEXT:    vpextrw $2, %xmm1, %eax
-; AVX1-NEXT:    vpextrw $2, %xmm0, %r15d
-; AVX1-NEXT:    xorl %ecx, %ecx
-; AVX1-NEXT:    movl %r15d, %edx
-; AVX1-NEXT:    addw %ax, %dx
-; AVX1-NEXT:    setns %cl
-; AVX1-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX1-NEXT:    addw %ax, %r15w
-; AVX1-NEXT:    cmovol %ecx, %r15d
-; AVX1-NEXT:    vmovd %xmm1, %eax
-; AVX1-NEXT:    vmovd %xmm0, %r12d
-; AVX1-NEXT:    xorl %ecx, %ecx
-; AVX1-NEXT:    movl %r12d, %edx
-; AVX1-NEXT:    addw %ax, %dx
-; AVX1-NEXT:    setns %cl
-; AVX1-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX1-NEXT:    addw %ax, %r12w
-; AVX1-NEXT:    cmovol %ecx, %r12d
-; AVX1-NEXT:    vpextrw $1, %xmm1, %eax
-; AVX1-NEXT:    vpextrw $1, %xmm0, %r13d
-; AVX1-NEXT:    xorl %ecx, %ecx
-; AVX1-NEXT:    movl %r13d, %esi
-; AVX1-NEXT:    addw %ax, %si
-; AVX1-NEXT:    setns %cl
-; AVX1-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX1-NEXT:    addw %ax, %r13w
-; AVX1-NEXT:    cmovol %ecx, %r13d
-; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm1
-; AVX1-NEXT:    vpextrw $7, %xmm1, %eax
-; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm0
-; AVX1-NEXT:    vpextrw $7, %xmm0, %esi
-; AVX1-NEXT:    xorl %ecx, %ecx
-; AVX1-NEXT:    movl %esi, %edi
-; AVX1-NEXT:    addw %ax, %di
-; AVX1-NEXT:    setns %cl
-; AVX1-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX1-NEXT:    addw %ax, %si
-; AVX1-NEXT:    cmovol %ecx, %esi
-; AVX1-NEXT:    vpextrw $6, %xmm1, %eax
-; AVX1-NEXT:    vpextrw $6, %xmm0, %ebx
-; AVX1-NEXT:    xorl %ecx, %ecx
-; AVX1-NEXT:    movl %ebx, %edi
-; AVX1-NEXT:    addw %ax, %di
-; AVX1-NEXT:    setns %cl
-; AVX1-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX1-NEXT:    addw %ax, %bx
-; AVX1-NEXT:    cmovol %ecx, %ebx
-; AVX1-NEXT:    vpextrw $5, %xmm1, %eax
-; AVX1-NEXT:    vpextrw $5, %xmm0, %ebp
-; AVX1-NEXT:    xorl %ecx, %ecx
-; AVX1-NEXT:    movl %ebp, %edi
-; AVX1-NEXT:    addw %ax, %di
-; AVX1-NEXT:    setns %cl
-; AVX1-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX1-NEXT:    addw %ax, %bp
-; AVX1-NEXT:    cmovol %ecx, %ebp
-; AVX1-NEXT:    vpextrw $4, %xmm1, %eax
-; AVX1-NEXT:    vpextrw $4, %xmm0, %edi
-; AVX1-NEXT:    xorl %ecx, %ecx
-; AVX1-NEXT:    movl %edi, %edx
-; AVX1-NEXT:    addw %ax, %dx
-; AVX1-NEXT:    setns %cl
-; AVX1-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX1-NEXT:    addw %ax, %di
-; AVX1-NEXT:    cmovol %ecx, %edi
-; AVX1-NEXT:    vpextrw $3, %xmm1, %ecx
-; AVX1-NEXT:    vpextrw $3, %xmm0, %eax
-; AVX1-NEXT:    xorl %edx, %edx
-; AVX1-NEXT:    movl %eax, %r8d
-; AVX1-NEXT:    addw %cx, %r8w
-; AVX1-NEXT:    setns %dl
-; AVX1-NEXT:    addl $32767, %edx # imm = 0x7FFF
-; AVX1-NEXT:    addw %cx, %ax
-; AVX1-NEXT:    cmovol %edx, %eax
-; AVX1-NEXT:    vpextrw $2, %xmm1, %r8d
-; AVX1-NEXT:    vpextrw $2, %xmm0, %ecx
-; AVX1-NEXT:    xorl %edx, %edx
-; AVX1-NEXT:    movl %ecx, %r9d
-; AVX1-NEXT:    addw %r8w, %r9w
-; AVX1-NEXT:    setns %dl
-; AVX1-NEXT:    addl $32767, %edx # imm = 0x7FFF
-; AVX1-NEXT:    addw %r8w, %cx
-; AVX1-NEXT:    cmovol %edx, %ecx
-; AVX1-NEXT:    vmovd %xmm1, %r8d
-; AVX1-NEXT:    vmovd %xmm0, %r9d
-; AVX1-NEXT:    xorl %edx, %edx
-; AVX1-NEXT:    movl %r9d, %r10d
-; AVX1-NEXT:    addw %r8w, %r10w
-; AVX1-NEXT:    setns %dl
-; AVX1-NEXT:    addl $32767, %edx # imm = 0x7FFF
-; AVX1-NEXT:    addw %r8w, %r9w
-; AVX1-NEXT:    cmovol %edx, %r9d
-; AVX1-NEXT:    vpextrw $1, %xmm1, %r8d
-; AVX1-NEXT:    vpextrw $1, %xmm0, %edx
-; AVX1-NEXT:    xorl %r10d, %r10d
-; AVX1-NEXT:    movl %edx, %r11d
-; AVX1-NEXT:    addw %r8w, %r11w
-; AVX1-NEXT:    setns %r10b
-; AVX1-NEXT:    addl $32767, %r10d # imm = 0x7FFF
-; AVX1-NEXT:    addw %r8w, %dx
-; AVX1-NEXT:    cmovol %r10d, %edx
-; AVX1-NEXT:    vmovd %r9d, %xmm0
-; AVX1-NEXT:    vpinsrw $1, %edx, %xmm0, %xmm0
-; AVX1-NEXT:    vpinsrw $2, %ecx, %xmm0, %xmm0
-; AVX1-NEXT:    vpinsrw $3, %eax, %xmm0, %xmm0
-; AVX1-NEXT:    vpinsrw $4, %edi, %xmm0, %xmm0
-; AVX1-NEXT:    vpinsrw $5, %ebp, %xmm0, %xmm0
-; AVX1-NEXT:    vpinsrw $6, %ebx, %xmm0, %xmm0
-; AVX1-NEXT:    vpinsrw $7, %esi, %xmm0, %xmm0
-; AVX1-NEXT:    vmovd %r12d, %xmm1
-; AVX1-NEXT:    vpinsrw $1, %r13d, %xmm1, %xmm1
-; AVX1-NEXT:    vpinsrw $2, %r15d, %xmm1, %xmm1
-; AVX1-NEXT:    vpinsrw $3, %r14d, %xmm1, %xmm1
-; AVX1-NEXT:    vpinsrw $4, {{[-0-9]+}}(%r{{[sb]}}p), %xmm1, %xmm1 # 4-byte Folded Reload
-; AVX1-NEXT:    vpinsrw $5, {{[-0-9]+}}(%r{{[sb]}}p), %xmm1, %xmm1 # 4-byte Folded Reload
-; AVX1-NEXT:    vpinsrw $6, {{[-0-9]+}}(%r{{[sb]}}p), %xmm1, %xmm1 # 4-byte Folded Reload
-; AVX1-NEXT:    vpinsrw $7, {{[-0-9]+}}(%r{{[sb]}}p), %xmm1, %xmm1 # 4-byte Folded Reload
-; AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm1, %ymm0
-; AVX1-NEXT:    popq %rbx
-; AVX1-NEXT:    popq %r12
-; AVX1-NEXT:    popq %r13
-; AVX1-NEXT:    popq %r14
-; AVX1-NEXT:    popq %r15
-; AVX1-NEXT:    popq %rbp
+; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm2
+; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm3
+; AVX1-NEXT:    vpaddsw %xmm2, %xmm3, %xmm2
+; AVX1-NEXT:    vpaddsw %xmm1, %xmm0, %xmm0
+; AVX1-NEXT:    vinsertf128 $1, %xmm2, %ymm0, %ymm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: v16i16:
 ; AVX2:       # %bb.0:
-; AVX2-NEXT:    pushq %rbp
-; AVX2-NEXT:    pushq %r15
-; AVX2-NEXT:    pushq %r14
-; AVX2-NEXT:    pushq %r13
-; AVX2-NEXT:    pushq %r12
-; AVX2-NEXT:    pushq %rbx
-; AVX2-NEXT:    vpextrw $7, %xmm1, %eax
-; AVX2-NEXT:    vpextrw $7, %xmm0, %esi
-; AVX2-NEXT:    xorl %ecx, %ecx
-; AVX2-NEXT:    movl %esi, %edx
-; AVX2-NEXT:    addw %ax, %dx
-; AVX2-NEXT:    setns %cl
-; AVX2-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX2-NEXT:    addw %ax, %si
-; AVX2-NEXT:    cmovol %ecx, %esi
-; AVX2-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX2-NEXT:    vpextrw $6, %xmm1, %eax
-; AVX2-NEXT:    vpextrw $6, %xmm0, %esi
-; AVX2-NEXT:    xorl %ecx, %ecx
-; AVX2-NEXT:    movl %esi, %edx
-; AVX2-NEXT:    addw %ax, %dx
-; AVX2-NEXT:    setns %cl
-; AVX2-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX2-NEXT:    addw %ax, %si
-; AVX2-NEXT:    cmovol %ecx, %esi
-; AVX2-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX2-NEXT:    vpextrw $5, %xmm1, %eax
-; AVX2-NEXT:    vpextrw $5, %xmm0, %esi
-; AVX2-NEXT:    xorl %ecx, %ecx
-; AVX2-NEXT:    movl %esi, %edx
-; AVX2-NEXT:    addw %ax, %dx
-; AVX2-NEXT:    setns %cl
-; AVX2-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX2-NEXT:    addw %ax, %si
-; AVX2-NEXT:    cmovol %ecx, %esi
-; AVX2-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX2-NEXT:    vpextrw $4, %xmm1, %eax
-; AVX2-NEXT:    vpextrw $4, %xmm0, %esi
-; AVX2-NEXT:    xorl %ecx, %ecx
-; AVX2-NEXT:    movl %esi, %edx
-; AVX2-NEXT:    addw %ax, %dx
-; AVX2-NEXT:    setns %cl
-; AVX2-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX2-NEXT:    addw %ax, %si
-; AVX2-NEXT:    cmovol %ecx, %esi
-; AVX2-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX2-NEXT:    vpextrw $3, %xmm1, %eax
-; AVX2-NEXT:    vpextrw $3, %xmm0, %r14d
-; AVX2-NEXT:    xorl %ecx, %ecx
-; AVX2-NEXT:    movl %r14d, %edx
-; AVX2-NEXT:    addw %ax, %dx
-; AVX2-NEXT:    setns %cl
-; AVX2-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX2-NEXT:    addw %ax, %r14w
-; AVX2-NEXT:    cmovol %ecx, %r14d
-; AVX2-NEXT:    vpextrw $2, %xmm1, %eax
-; AVX2-NEXT:    vpextrw $2, %xmm0, %r15d
-; AVX2-NEXT:    xorl %ecx, %ecx
-; AVX2-NEXT:    movl %r15d, %edx
-; AVX2-NEXT:    addw %ax, %dx
-; AVX2-NEXT:    setns %cl
-; AVX2-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX2-NEXT:    addw %ax, %r15w
-; AVX2-NEXT:    cmovol %ecx, %r15d
-; AVX2-NEXT:    vmovd %xmm1, %eax
-; AVX2-NEXT:    vmovd %xmm0, %r12d
-; AVX2-NEXT:    xorl %ecx, %ecx
-; AVX2-NEXT:    movl %r12d, %edx
-; AVX2-NEXT:    addw %ax, %dx
-; AVX2-NEXT:    setns %cl
-; AVX2-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX2-NEXT:    addw %ax, %r12w
-; AVX2-NEXT:    cmovol %ecx, %r12d
-; AVX2-NEXT:    vpextrw $1, %xmm1, %eax
-; AVX2-NEXT:    vpextrw $1, %xmm0, %r13d
-; AVX2-NEXT:    xorl %ecx, %ecx
-; AVX2-NEXT:    movl %r13d, %esi
-; AVX2-NEXT:    addw %ax, %si
-; AVX2-NEXT:    setns %cl
-; AVX2-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX2-NEXT:    addw %ax, %r13w
-; AVX2-NEXT:    cmovol %ecx, %r13d
-; AVX2-NEXT:    vextracti128 $1, %ymm1, %xmm1
-; AVX2-NEXT:    vpextrw $7, %xmm1, %eax
-; AVX2-NEXT:    vextracti128 $1, %ymm0, %xmm0
-; AVX2-NEXT:    vpextrw $7, %xmm0, %esi
-; AVX2-NEXT:    xorl %ecx, %ecx
-; AVX2-NEXT:    movl %esi, %edi
-; AVX2-NEXT:    addw %ax, %di
-; AVX2-NEXT:    setns %cl
-; AVX2-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX2-NEXT:    addw %ax, %si
-; AVX2-NEXT:    cmovol %ecx, %esi
-; AVX2-NEXT:    vpextrw $6, %xmm1, %eax
-; AVX2-NEXT:    vpextrw $6, %xmm0, %ebx
-; AVX2-NEXT:    xorl %ecx, %ecx
-; AVX2-NEXT:    movl %ebx, %edi
-; AVX2-NEXT:    addw %ax, %di
-; AVX2-NEXT:    setns %cl
-; AVX2-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX2-NEXT:    addw %ax, %bx
-; AVX2-NEXT:    cmovol %ecx, %ebx
-; AVX2-NEXT:    vpextrw $5, %xmm1, %eax
-; AVX2-NEXT:    vpextrw $5, %xmm0, %ebp
-; AVX2-NEXT:    xorl %ecx, %ecx
-; AVX2-NEXT:    movl %ebp, %edi
-; AVX2-NEXT:    addw %ax, %di
-; AVX2-NEXT:    setns %cl
-; AVX2-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX2-NEXT:    addw %ax, %bp
-; AVX2-NEXT:    cmovol %ecx, %ebp
-; AVX2-NEXT:    vpextrw $4, %xmm1, %eax
-; AVX2-NEXT:    vpextrw $4, %xmm0, %edi
-; AVX2-NEXT:    xorl %ecx, %ecx
-; AVX2-NEXT:    movl %edi, %edx
-; AVX2-NEXT:    addw %ax, %dx
-; AVX2-NEXT:    setns %cl
-; AVX2-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX2-NEXT:    addw %ax, %di
-; AVX2-NEXT:    cmovol %ecx, %edi
-; AVX2-NEXT:    vpextrw $3, %xmm1, %ecx
-; AVX2-NEXT:    vpextrw $3, %xmm0, %eax
-; AVX2-NEXT:    xorl %edx, %edx
-; AVX2-NEXT:    movl %eax, %r8d
-; AVX2-NEXT:    addw %cx, %r8w
-; AVX2-NEXT:    setns %dl
-; AVX2-NEXT:    addl $32767, %edx # imm = 0x7FFF
-; AVX2-NEXT:    addw %cx, %ax
-; AVX2-NEXT:    cmovol %edx, %eax
-; AVX2-NEXT:    vpextrw $2, %xmm1, %r8d
-; AVX2-NEXT:    vpextrw $2, %xmm0, %ecx
-; AVX2-NEXT:    xorl %edx, %edx
-; AVX2-NEXT:    movl %ecx, %r9d
-; AVX2-NEXT:    addw %r8w, %r9w
-; AVX2-NEXT:    setns %dl
-; AVX2-NEXT:    addl $32767, %edx # imm = 0x7FFF
-; AVX2-NEXT:    addw %r8w, %cx
-; AVX2-NEXT:    cmovol %edx, %ecx
-; AVX2-NEXT:    vmovd %xmm1, %r8d
-; AVX2-NEXT:    vmovd %xmm0, %r9d
-; AVX2-NEXT:    xorl %edx, %edx
-; AVX2-NEXT:    movl %r9d, %r10d
-; AVX2-NEXT:    addw %r8w, %r10w
-; AVX2-NEXT:    setns %dl
-; AVX2-NEXT:    addl $32767, %edx # imm = 0x7FFF
-; AVX2-NEXT:    addw %r8w, %r9w
-; AVX2-NEXT:    cmovol %edx, %r9d
-; AVX2-NEXT:    vpextrw $1, %xmm1, %r8d
-; AVX2-NEXT:    vpextrw $1, %xmm0, %edx
-; AVX2-NEXT:    xorl %r10d, %r10d
-; AVX2-NEXT:    movl %edx, %r11d
-; AVX2-NEXT:    addw %r8w, %r11w
-; AVX2-NEXT:    setns %r10b
-; AVX2-NEXT:    addl $32767, %r10d # imm = 0x7FFF
-; AVX2-NEXT:    addw %r8w, %dx
-; AVX2-NEXT:    cmovol %r10d, %edx
-; AVX2-NEXT:    vmovd %r9d, %xmm0
-; AVX2-NEXT:    vpinsrw $1, %edx, %xmm0, %xmm0
-; AVX2-NEXT:    vpinsrw $2, %ecx, %xmm0, %xmm0
-; AVX2-NEXT:    vpinsrw $3, %eax, %xmm0, %xmm0
-; AVX2-NEXT:    vpinsrw $4, %edi, %xmm0, %xmm0
-; AVX2-NEXT:    vpinsrw $5, %ebp, %xmm0, %xmm0
-; AVX2-NEXT:    vpinsrw $6, %ebx, %xmm0, %xmm0
-; AVX2-NEXT:    vpinsrw $7, %esi, %xmm0, %xmm0
-; AVX2-NEXT:    vmovd %r12d, %xmm1
-; AVX2-NEXT:    vpinsrw $1, %r13d, %xmm1, %xmm1
-; AVX2-NEXT:    vpinsrw $2, %r15d, %xmm1, %xmm1
-; AVX2-NEXT:    vpinsrw $3, %r14d, %xmm1, %xmm1
-; AVX2-NEXT:    vpinsrw $4, {{[-0-9]+}}(%r{{[sb]}}p), %xmm1, %xmm1 # 4-byte Folded Reload
-; AVX2-NEXT:    vpinsrw $5, {{[-0-9]+}}(%r{{[sb]}}p), %xmm1, %xmm1 # 4-byte Folded Reload
-; AVX2-NEXT:    vpinsrw $6, {{[-0-9]+}}(%r{{[sb]}}p), %xmm1, %xmm1 # 4-byte Folded Reload
-; AVX2-NEXT:    vpinsrw $7, {{[-0-9]+}}(%r{{[sb]}}p), %xmm1, %xmm1 # 4-byte Folded Reload
-; AVX2-NEXT:    vinserti128 $1, %xmm0, %ymm1, %ymm0
-; AVX2-NEXT:    popq %rbx
-; AVX2-NEXT:    popq %r12
-; AVX2-NEXT:    popq %r13
-; AVX2-NEXT:    popq %r14
-; AVX2-NEXT:    popq %r15
-; AVX2-NEXT:    popq %rbp
+; AVX2-NEXT:    vpaddsw %ymm1, %ymm0, %ymm0
 ; AVX2-NEXT:    retq
 ;
 ; AVX512-LABEL: v16i16:
 ; AVX512:       # %bb.0:
-; AVX512-NEXT:    pushq %rbp
-; AVX512-NEXT:    pushq %r15
-; AVX512-NEXT:    pushq %r14
-; AVX512-NEXT:    pushq %r13
-; AVX512-NEXT:    pushq %r12
-; AVX512-NEXT:    pushq %rbx
-; AVX512-NEXT:    vpextrw $7, %xmm1, %eax
-; AVX512-NEXT:    vpextrw $7, %xmm0, %esi
-; AVX512-NEXT:    xorl %ecx, %ecx
-; AVX512-NEXT:    movl %esi, %edx
-; AVX512-NEXT:    addw %ax, %dx
-; AVX512-NEXT:    setns %cl
-; AVX512-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX512-NEXT:    addw %ax, %si
-; AVX512-NEXT:    cmovol %ecx, %esi
-; AVX512-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX512-NEXT:    vpextrw $6, %xmm1, %eax
-; AVX512-NEXT:    vpextrw $6, %xmm0, %esi
-; AVX512-NEXT:    xorl %ecx, %ecx
-; AVX512-NEXT:    movl %esi, %edx
-; AVX512-NEXT:    addw %ax, %dx
-; AVX512-NEXT:    setns %cl
-; AVX512-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX512-NEXT:    addw %ax, %si
-; AVX512-NEXT:    cmovol %ecx, %esi
-; AVX512-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX512-NEXT:    vpextrw $5, %xmm1, %eax
-; AVX512-NEXT:    vpextrw $5, %xmm0, %esi
-; AVX512-NEXT:    xorl %ecx, %ecx
-; AVX512-NEXT:    movl %esi, %edx
-; AVX512-NEXT:    addw %ax, %dx
-; AVX512-NEXT:    setns %cl
-; AVX512-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX512-NEXT:    addw %ax, %si
-; AVX512-NEXT:    cmovol %ecx, %esi
-; AVX512-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX512-NEXT:    vpextrw $4, %xmm1, %eax
-; AVX512-NEXT:    vpextrw $4, %xmm0, %esi
-; AVX512-NEXT:    xorl %ecx, %ecx
-; AVX512-NEXT:    movl %esi, %edx
-; AVX512-NEXT:    addw %ax, %dx
-; AVX512-NEXT:    setns %cl
-; AVX512-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX512-NEXT:    addw %ax, %si
-; AVX512-NEXT:    cmovol %ecx, %esi
-; AVX512-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX512-NEXT:    vpextrw $3, %xmm1, %eax
-; AVX512-NEXT:    vpextrw $3, %xmm0, %r14d
-; AVX512-NEXT:    xorl %ecx, %ecx
-; AVX512-NEXT:    movl %r14d, %edx
-; AVX512-NEXT:    addw %ax, %dx
-; AVX512-NEXT:    setns %cl
-; AVX512-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX512-NEXT:    addw %ax, %r14w
-; AVX512-NEXT:    cmovol %ecx, %r14d
-; AVX512-NEXT:    vpextrw $2, %xmm1, %eax
-; AVX512-NEXT:    vpextrw $2, %xmm0, %r15d
-; AVX512-NEXT:    xorl %ecx, %ecx
-; AVX512-NEXT:    movl %r15d, %edx
-; AVX512-NEXT:    addw %ax, %dx
-; AVX512-NEXT:    setns %cl
-; AVX512-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX512-NEXT:    addw %ax, %r15w
-; AVX512-NEXT:    cmovol %ecx, %r15d
-; AVX512-NEXT:    vmovd %xmm1, %eax
-; AVX512-NEXT:    vmovd %xmm0, %r12d
-; AVX512-NEXT:    xorl %ecx, %ecx
-; AVX512-NEXT:    movl %r12d, %edx
-; AVX512-NEXT:    addw %ax, %dx
-; AVX512-NEXT:    setns %cl
-; AVX512-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX512-NEXT:    addw %ax, %r12w
-; AVX512-NEXT:    cmovol %ecx, %r12d
-; AVX512-NEXT:    vpextrw $1, %xmm1, %eax
-; AVX512-NEXT:    vpextrw $1, %xmm0, %r13d
-; AVX512-NEXT:    xorl %ecx, %ecx
-; AVX512-NEXT:    movl %r13d, %esi
-; AVX512-NEXT:    addw %ax, %si
-; AVX512-NEXT:    setns %cl
-; AVX512-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX512-NEXT:    addw %ax, %r13w
-; AVX512-NEXT:    cmovol %ecx, %r13d
-; AVX512-NEXT:    vextracti128 $1, %ymm1, %xmm1
-; AVX512-NEXT:    vpextrw $7, %xmm1, %eax
-; AVX512-NEXT:    vextracti128 $1, %ymm0, %xmm0
-; AVX512-NEXT:    vpextrw $7, %xmm0, %esi
-; AVX512-NEXT:    xorl %ecx, %ecx
-; AVX512-NEXT:    movl %esi, %edi
-; AVX512-NEXT:    addw %ax, %di
-; AVX512-NEXT:    setns %cl
-; AVX512-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX512-NEXT:    addw %ax, %si
-; AVX512-NEXT:    cmovol %ecx, %esi
-; AVX512-NEXT:    vpextrw $6, %xmm1, %eax
-; AVX512-NEXT:    vpextrw $6, %xmm0, %ebx
-; AVX512-NEXT:    xorl %ecx, %ecx
-; AVX512-NEXT:    movl %ebx, %edi
-; AVX512-NEXT:    addw %ax, %di
-; AVX512-NEXT:    setns %cl
-; AVX512-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX512-NEXT:    addw %ax, %bx
-; AVX512-NEXT:    cmovol %ecx, %ebx
-; AVX512-NEXT:    vpextrw $5, %xmm1, %eax
-; AVX512-NEXT:    vpextrw $5, %xmm0, %ebp
-; AVX512-NEXT:    xorl %ecx, %ecx
-; AVX512-NEXT:    movl %ebp, %edi
-; AVX512-NEXT:    addw %ax, %di
-; AVX512-NEXT:    setns %cl
-; AVX512-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX512-NEXT:    addw %ax, %bp
-; AVX512-NEXT:    cmovol %ecx, %ebp
-; AVX512-NEXT:    vpextrw $4, %xmm1, %eax
-; AVX512-NEXT:    vpextrw $4, %xmm0, %edi
-; AVX512-NEXT:    xorl %ecx, %ecx
-; AVX512-NEXT:    movl %edi, %edx
-; AVX512-NEXT:    addw %ax, %dx
-; AVX512-NEXT:    setns %cl
-; AVX512-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX512-NEXT:    addw %ax, %di
-; AVX512-NEXT:    cmovol %ecx, %edi
-; AVX512-NEXT:    vpextrw $3, %xmm1, %ecx
-; AVX512-NEXT:    vpextrw $3, %xmm0, %eax
-; AVX512-NEXT:    xorl %edx, %edx
-; AVX512-NEXT:    movl %eax, %r8d
-; AVX512-NEXT:    addw %cx, %r8w
-; AVX512-NEXT:    setns %dl
-; AVX512-NEXT:    addl $32767, %edx # imm = 0x7FFF
-; AVX512-NEXT:    addw %cx, %ax
-; AVX512-NEXT:    cmovol %edx, %eax
-; AVX512-NEXT:    vpextrw $2, %xmm1, %r8d
-; AVX512-NEXT:    vpextrw $2, %xmm0, %ecx
-; AVX512-NEXT:    xorl %edx, %edx
-; AVX512-NEXT:    movl %ecx, %r9d
-; AVX512-NEXT:    addw %r8w, %r9w
-; AVX512-NEXT:    setns %dl
-; AVX512-NEXT:    addl $32767, %edx # imm = 0x7FFF
-; AVX512-NEXT:    addw %r8w, %cx
-; AVX512-NEXT:    cmovol %edx, %ecx
-; AVX512-NEXT:    vmovd %xmm1, %r8d
-; AVX512-NEXT:    vmovd %xmm0, %r9d
-; AVX512-NEXT:    xorl %edx, %edx
-; AVX512-NEXT:    movl %r9d, %r10d
-; AVX512-NEXT:    addw %r8w, %r10w
-; AVX512-NEXT:    setns %dl
-; AVX512-NEXT:    addl $32767, %edx # imm = 0x7FFF
-; AVX512-NEXT:    addw %r8w, %r9w
-; AVX512-NEXT:    cmovol %edx, %r9d
-; AVX512-NEXT:    vpextrw $1, %xmm1, %r8d
-; AVX512-NEXT:    vpextrw $1, %xmm0, %edx
-; AVX512-NEXT:    xorl %r10d, %r10d
-; AVX512-NEXT:    movl %edx, %r11d
-; AVX512-NEXT:    addw %r8w, %r11w
-; AVX512-NEXT:    setns %r10b
-; AVX512-NEXT:    addl $32767, %r10d # imm = 0x7FFF
-; AVX512-NEXT:    addw %r8w, %dx
-; AVX512-NEXT:    cmovol %r10d, %edx
-; AVX512-NEXT:    vmovd %r9d, %xmm0
-; AVX512-NEXT:    vpinsrw $1, %edx, %xmm0, %xmm0
-; AVX512-NEXT:    vpinsrw $2, %ecx, %xmm0, %xmm0
-; AVX512-NEXT:    vpinsrw $3, %eax, %xmm0, %xmm0
-; AVX512-NEXT:    vpinsrw $4, %edi, %xmm0, %xmm0
-; AVX512-NEXT:    vpinsrw $5, %ebp, %xmm0, %xmm0
-; AVX512-NEXT:    vpinsrw $6, %ebx, %xmm0, %xmm0
-; AVX512-NEXT:    vpinsrw $7, %esi, %xmm0, %xmm0
-; AVX512-NEXT:    vmovd %r12d, %xmm1
-; AVX512-NEXT:    vpinsrw $1, %r13d, %xmm1, %xmm1
-; AVX512-NEXT:    vpinsrw $2, %r15d, %xmm1, %xmm1
-; AVX512-NEXT:    vpinsrw $3, %r14d, %xmm1, %xmm1
-; AVX512-NEXT:    vpinsrw $4, {{[-0-9]+}}(%r{{[sb]}}p), %xmm1, %xmm1 # 4-byte Folded Reload
-; AVX512-NEXT:    vpinsrw $5, {{[-0-9]+}}(%r{{[sb]}}p), %xmm1, %xmm1 # 4-byte Folded Reload
-; AVX512-NEXT:    vpinsrw $6, {{[-0-9]+}}(%r{{[sb]}}p), %xmm1, %xmm1 # 4-byte Folded Reload
-; AVX512-NEXT:    vpinsrw $7, {{[-0-9]+}}(%r{{[sb]}}p), %xmm1, %xmm1 # 4-byte Folded Reload
-; AVX512-NEXT:    vinserti128 $1, %xmm0, %ymm1, %ymm0
-; AVX512-NEXT:    popq %rbx
-; AVX512-NEXT:    popq %r12
-; AVX512-NEXT:    popq %r13
-; AVX512-NEXT:    popq %r14
-; AVX512-NEXT:    popq %r15
-; AVX512-NEXT:    popq %rbp
+; AVX512-NEXT:    vpaddsw %ymm1, %ymm0, %ymm0
 ; AVX512-NEXT:    retq
   %z = call <16 x i16> @llvm.sadd.sat.v16i16(<16 x i16> %x, <16 x i16> %y)
   ret <16 x i16> %z
 }
 
 define <32 x i16> @v32i16(<32 x i16> %x, <32 x i16> %y) nounwind {
-; SSE2-LABEL: v32i16:
-; SSE2:       # %bb.0:
-; SSE2-NEXT:    pushq %rbp
-; SSE2-NEXT:    pushq %r15
-; SSE2-NEXT:    pushq %r14
-; SSE2-NEXT:    pushq %r13
-; SSE2-NEXT:    pushq %r12
-; SSE2-NEXT:    pushq %rbx
-; SSE2-NEXT:    pushq %rax
-; SSE2-NEXT:    movd %xmm5, %eax
-; SSE2-NEXT:    movd %xmm1, %esi
-; SSE2-NEXT:    xorl %ecx, %ecx
-; SSE2-NEXT:    movl %esi, %edx
-; SSE2-NEXT:    addw %ax, %dx
-; SSE2-NEXT:    setns %cl
-; SSE2-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSE2-NEXT:    addw %ax, %si
-; SSE2-NEXT:    cmovol %ecx, %esi
-; SSE2-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE2-NEXT:    pextrw $1, %xmm5, %eax
-; SSE2-NEXT:    pextrw $1, %xmm1, %esi
-; SSE2-NEXT:    xorl %ecx, %ecx
-; SSE2-NEXT:    movl %esi, %edx
-; SSE2-NEXT:    addw %ax, %dx
-; SSE2-NEXT:    setns %cl
-; SSE2-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSE2-NEXT:    addw %ax, %si
-; SSE2-NEXT:    cmovol %ecx, %esi
-; SSE2-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE2-NEXT:    pextrw $2, %xmm5, %eax
-; SSE2-NEXT:    pextrw $2, %xmm1, %esi
-; SSE2-NEXT:    xorl %ecx, %ecx
-; SSE2-NEXT:    movl %esi, %edx
-; SSE2-NEXT:    addw %ax, %dx
-; SSE2-NEXT:    setns %cl
-; SSE2-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSE2-NEXT:    addw %ax, %si
-; SSE2-NEXT:    cmovol %ecx, %esi
-; SSE2-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE2-NEXT:    pextrw $3, %xmm5, %eax
-; SSE2-NEXT:    pextrw $3, %xmm1, %esi
-; SSE2-NEXT:    xorl %ecx, %ecx
-; SSE2-NEXT:    movl %esi, %edx
-; SSE2-NEXT:    addw %ax, %dx
-; SSE2-NEXT:    setns %cl
-; SSE2-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSE2-NEXT:    addw %ax, %si
-; SSE2-NEXT:    cmovol %ecx, %esi
-; SSE2-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE2-NEXT:    pextrw $4, %xmm5, %eax
-; SSE2-NEXT:    pextrw $4, %xmm1, %esi
-; SSE2-NEXT:    xorl %ecx, %ecx
-; SSE2-NEXT:    movl %esi, %edx
-; SSE2-NEXT:    addw %ax, %dx
-; SSE2-NEXT:    setns %cl
-; SSE2-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSE2-NEXT:    addw %ax, %si
-; SSE2-NEXT:    cmovol %ecx, %esi
-; SSE2-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE2-NEXT:    pextrw $5, %xmm5, %eax
-; SSE2-NEXT:    pextrw $5, %xmm1, %esi
-; SSE2-NEXT:    xorl %ecx, %ecx
-; SSE2-NEXT:    movl %esi, %edx
-; SSE2-NEXT:    addw %ax, %dx
-; SSE2-NEXT:    setns %cl
-; SSE2-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSE2-NEXT:    addw %ax, %si
-; SSE2-NEXT:    cmovol %ecx, %esi
-; SSE2-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE2-NEXT:    pextrw $6, %xmm5, %eax
-; SSE2-NEXT:    pextrw $6, %xmm1, %esi
-; SSE2-NEXT:    xorl %ecx, %ecx
-; SSE2-NEXT:    movl %esi, %edx
-; SSE2-NEXT:    addw %ax, %dx
-; SSE2-NEXT:    setns %cl
-; SSE2-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSE2-NEXT:    addw %ax, %si
-; SSE2-NEXT:    cmovol %ecx, %esi
-; SSE2-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE2-NEXT:    pextrw $7, %xmm5, %eax
-; SSE2-NEXT:    pextrw $7, %xmm1, %esi
-; SSE2-NEXT:    xorl %ecx, %ecx
-; SSE2-NEXT:    movl %esi, %edx
-; SSE2-NEXT:    addw %ax, %dx
-; SSE2-NEXT:    setns %cl
-; SSE2-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSE2-NEXT:    addw %ax, %si
-; SSE2-NEXT:    cmovol %ecx, %esi
-; SSE2-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE2-NEXT:    movd %xmm6, %eax
-; SSE2-NEXT:    movd %xmm2, %esi
-; SSE2-NEXT:    xorl %ecx, %ecx
-; SSE2-NEXT:    movl %esi, %edx
-; SSE2-NEXT:    addw %ax, %dx
-; SSE2-NEXT:    setns %cl
-; SSE2-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSE2-NEXT:    addw %ax, %si
-; SSE2-NEXT:    cmovol %ecx, %esi
-; SSE2-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE2-NEXT:    pextrw $1, %xmm6, %eax
-; SSE2-NEXT:    pextrw $1, %xmm2, %esi
-; SSE2-NEXT:    xorl %ecx, %ecx
-; SSE2-NEXT:    movl %esi, %edx
-; SSE2-NEXT:    addw %ax, %dx
-; SSE2-NEXT:    setns %cl
-; SSE2-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSE2-NEXT:    addw %ax, %si
-; SSE2-NEXT:    cmovol %ecx, %esi
-; SSE2-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE2-NEXT:    pextrw $2, %xmm6, %eax
-; SSE2-NEXT:    pextrw $2, %xmm2, %esi
-; SSE2-NEXT:    xorl %ecx, %ecx
-; SSE2-NEXT:    movl %esi, %edx
-; SSE2-NEXT:    addw %ax, %dx
-; SSE2-NEXT:    setns %cl
-; SSE2-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSE2-NEXT:    addw %ax, %si
-; SSE2-NEXT:    cmovol %ecx, %esi
-; SSE2-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE2-NEXT:    pextrw $3, %xmm6, %eax
-; SSE2-NEXT:    pextrw $3, %xmm2, %esi
-; SSE2-NEXT:    xorl %ecx, %ecx
-; SSE2-NEXT:    movl %esi, %edx
-; SSE2-NEXT:    addw %ax, %dx
-; SSE2-NEXT:    setns %cl
-; SSE2-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSE2-NEXT:    addw %ax, %si
-; SSE2-NEXT:    cmovol %ecx, %esi
-; SSE2-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE2-NEXT:    pextrw $4, %xmm6, %eax
-; SSE2-NEXT:    pextrw $4, %xmm2, %esi
-; SSE2-NEXT:    xorl %ecx, %ecx
-; SSE2-NEXT:    movl %esi, %edx
-; SSE2-NEXT:    addw %ax, %dx
-; SSE2-NEXT:    setns %cl
-; SSE2-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSE2-NEXT:    addw %ax, %si
-; SSE2-NEXT:    cmovol %ecx, %esi
-; SSE2-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE2-NEXT:    pextrw $5, %xmm6, %eax
-; SSE2-NEXT:    pextrw $5, %xmm2, %esi
-; SSE2-NEXT:    xorl %ecx, %ecx
-; SSE2-NEXT:    movl %esi, %edx
-; SSE2-NEXT:    addw %ax, %dx
-; SSE2-NEXT:    setns %cl
-; SSE2-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSE2-NEXT:    addw %ax, %si
-; SSE2-NEXT:    cmovol %ecx, %esi
-; SSE2-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE2-NEXT:    pextrw $6, %xmm6, %eax
-; SSE2-NEXT:    pextrw $6, %xmm2, %esi
-; SSE2-NEXT:    xorl %ecx, %ecx
-; SSE2-NEXT:    movl %esi, %edx
-; SSE2-NEXT:    addw %ax, %dx
-; SSE2-NEXT:    setns %cl
-; SSE2-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSE2-NEXT:    addw %ax, %si
-; SSE2-NEXT:    cmovol %ecx, %esi
-; SSE2-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE2-NEXT:    pextrw $7, %xmm6, %eax
-; SSE2-NEXT:    pextrw $7, %xmm2, %esi
-; SSE2-NEXT:    xorl %ecx, %ecx
-; SSE2-NEXT:    movl %esi, %edx
-; SSE2-NEXT:    addw %ax, %dx
-; SSE2-NEXT:    setns %cl
-; SSE2-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSE2-NEXT:    addw %ax, %si
-; SSE2-NEXT:    cmovol %ecx, %esi
-; SSE2-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE2-NEXT:    movd %xmm7, %eax
-; SSE2-NEXT:    movd %xmm3, %esi
-; SSE2-NEXT:    xorl %ecx, %ecx
-; SSE2-NEXT:    movl %esi, %edx
-; SSE2-NEXT:    addw %ax, %dx
-; SSE2-NEXT:    setns %cl
-; SSE2-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSE2-NEXT:    addw %ax, %si
-; SSE2-NEXT:    cmovol %ecx, %esi
-; SSE2-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE2-NEXT:    pextrw $1, %xmm7, %eax
-; SSE2-NEXT:    pextrw $1, %xmm3, %esi
-; SSE2-NEXT:    xorl %ecx, %ecx
-; SSE2-NEXT:    movl %esi, %edx
-; SSE2-NEXT:    addw %ax, %dx
-; SSE2-NEXT:    setns %cl
-; SSE2-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSE2-NEXT:    addw %ax, %si
-; SSE2-NEXT:    cmovol %ecx, %esi
-; SSE2-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE2-NEXT:    pextrw $2, %xmm7, %eax
-; SSE2-NEXT:    pextrw $2, %xmm3, %esi
-; SSE2-NEXT:    xorl %ecx, %ecx
-; SSE2-NEXT:    movl %esi, %edx
-; SSE2-NEXT:    addw %ax, %dx
-; SSE2-NEXT:    setns %cl
-; SSE2-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSE2-NEXT:    addw %ax, %si
-; SSE2-NEXT:    cmovol %ecx, %esi
-; SSE2-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE2-NEXT:    pextrw $3, %xmm7, %eax
-; SSE2-NEXT:    pextrw $3, %xmm3, %esi
-; SSE2-NEXT:    xorl %ecx, %ecx
-; SSE2-NEXT:    movl %esi, %edx
-; SSE2-NEXT:    addw %ax, %dx
-; SSE2-NEXT:    setns %cl
-; SSE2-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSE2-NEXT:    addw %ax, %si
-; SSE2-NEXT:    cmovol %ecx, %esi
-; SSE2-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE2-NEXT:    pextrw $4, %xmm7, %eax
-; SSE2-NEXT:    pextrw $4, %xmm3, %ebp
-; SSE2-NEXT:    xorl %ecx, %ecx
-; SSE2-NEXT:    movl %ebp, %edx
-; SSE2-NEXT:    addw %ax, %dx
-; SSE2-NEXT:    setns %cl
-; SSE2-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSE2-NEXT:    addw %ax, %bp
-; SSE2-NEXT:    cmovol %ecx, %ebp
-; SSE2-NEXT:    pextrw $5, %xmm7, %eax
-; SSE2-NEXT:    pextrw $5, %xmm3, %ebx
-; SSE2-NEXT:    xorl %ecx, %ecx
-; SSE2-NEXT:    movl %ebx, %edx
-; SSE2-NEXT:    addw %ax, %dx
-; SSE2-NEXT:    setns %cl
-; SSE2-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSE2-NEXT:    addw %ax, %bx
-; SSE2-NEXT:    cmovol %ecx, %ebx
-; SSE2-NEXT:    pextrw $6, %xmm7, %eax
-; SSE2-NEXT:    pextrw $6, %xmm3, %r11d
-; SSE2-NEXT:    xorl %ecx, %ecx
-; SSE2-NEXT:    movl %r11d, %edx
-; SSE2-NEXT:    addw %ax, %dx
-; SSE2-NEXT:    setns %cl
-; SSE2-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSE2-NEXT:    addw %ax, %r11w
-; SSE2-NEXT:    cmovol %ecx, %r11d
-; SSE2-NEXT:    pextrw $7, %xmm7, %eax
-; SSE2-NEXT:    pextrw $7, %xmm3, %r10d
-; SSE2-NEXT:    xorl %ecx, %ecx
-; SSE2-NEXT:    movl %r10d, %edx
-; SSE2-NEXT:    addw %ax, %dx
-; SSE2-NEXT:    setns %cl
-; SSE2-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSE2-NEXT:    addw %ax, %r10w
-; SSE2-NEXT:    cmovol %ecx, %r10d
-; SSE2-NEXT:    movd %xmm4, %eax
-; SSE2-NEXT:    movd %xmm0, %r9d
-; SSE2-NEXT:    xorl %ecx, %ecx
-; SSE2-NEXT:    movl %r9d, %edx
-; SSE2-NEXT:    addw %ax, %dx
-; SSE2-NEXT:    setns %cl
-; SSE2-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSE2-NEXT:    addw %ax, %r9w
-; SSE2-NEXT:    cmovol %ecx, %r9d
-; SSE2-NEXT:    pextrw $1, %xmm4, %eax
-; SSE2-NEXT:    pextrw $1, %xmm0, %r8d
-; SSE2-NEXT:    xorl %ecx, %ecx
-; SSE2-NEXT:    movl %r8d, %edx
-; SSE2-NEXT:    addw %ax, %dx
-; SSE2-NEXT:    setns %cl
-; SSE2-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSE2-NEXT:    addw %ax, %r8w
-; SSE2-NEXT:    cmovol %ecx, %r8d
-; SSE2-NEXT:    pextrw $2, %xmm4, %eax
-; SSE2-NEXT:    pextrw $2, %xmm0, %edi
-; SSE2-NEXT:    xorl %ecx, %ecx
-; SSE2-NEXT:    movl %edi, %edx
-; SSE2-NEXT:    addw %ax, %dx
-; SSE2-NEXT:    setns %cl
-; SSE2-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSE2-NEXT:    addw %ax, %di
-; SSE2-NEXT:    cmovol %ecx, %edi
-; SSE2-NEXT:    pextrw $3, %xmm4, %eax
-; SSE2-NEXT:    pextrw $3, %xmm0, %esi
-; SSE2-NEXT:    xorl %ecx, %ecx
-; SSE2-NEXT:    movl %esi, %edx
-; SSE2-NEXT:    addw %ax, %dx
-; SSE2-NEXT:    setns %cl
-; SSE2-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSE2-NEXT:    addw %ax, %si
-; SSE2-NEXT:    cmovol %ecx, %esi
-; SSE2-NEXT:    pextrw $4, %xmm4, %eax
-; SSE2-NEXT:    pextrw $4, %xmm0, %edx
-; SSE2-NEXT:    xorl %ecx, %ecx
-; SSE2-NEXT:    movl %edx, %r13d
-; SSE2-NEXT:    addw %ax, %r13w
-; SSE2-NEXT:    setns %cl
-; SSE2-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSE2-NEXT:    addw %ax, %dx
-; SSE2-NEXT:    cmovol %ecx, %edx
-; SSE2-NEXT:    pextrw $5, %xmm4, %r13d
-; SSE2-NEXT:    pextrw $5, %xmm0, %ecx
-; SSE2-NEXT:    xorl %eax, %eax
-; SSE2-NEXT:    movl %ecx, %r12d
-; SSE2-NEXT:    addw %r13w, %r12w
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addl $32767, %eax # imm = 0x7FFF
-; SSE2-NEXT:    addw %r13w, %cx
-; SSE2-NEXT:    cmovol %eax, %ecx
-; SSE2-NEXT:    pextrw $6, %xmm4, %r12d
-; SSE2-NEXT:    pextrw $6, %xmm0, %r13d
-; SSE2-NEXT:    xorl %eax, %eax
-; SSE2-NEXT:    movl %r13d, %r15d
-; SSE2-NEXT:    addw %r12w, %r15w
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addl $32767, %eax # imm = 0x7FFF
-; SSE2-NEXT:    addw %r12w, %r13w
-; SSE2-NEXT:    cmovol %eax, %r13d
-; SSE2-NEXT:    pextrw $7, %xmm4, %r15d
-; SSE2-NEXT:    pextrw $7, %xmm0, %eax
-; SSE2-NEXT:    xorl %r12d, %r12d
-; SSE2-NEXT:    movl %eax, %r14d
-; SSE2-NEXT:    addw %r15w, %r14w
-; SSE2-NEXT:    setns %r12b
-; SSE2-NEXT:    addl $32767, %r12d # imm = 0x7FFF
-; SSE2-NEXT:    addw %r15w, %ax
-; SSE2-NEXT:    cmovol %r12d, %eax
-; SSE2-NEXT:    movd %eax, %xmm10
-; SSE2-NEXT:    movd %r13d, %xmm12
-; SSE2-NEXT:    movd %ecx, %xmm8
-; SSE2-NEXT:    movd %edx, %xmm2
-; SSE2-NEXT:    movd %esi, %xmm4
-; SSE2-NEXT:    movd %edi, %xmm0
-; SSE2-NEXT:    movd %r8d, %xmm13
-; SSE2-NEXT:    movd %r9d, %xmm5
-; SSE2-NEXT:    movss {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 4-byte Reload
-; SSE2-NEXT:    # xmm1 = mem[0],zero,zero,zero
-; SSE2-NEXT:    movaps %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; SSE2-NEXT:    movd {{[-0-9]+}}(%r{{[sb]}}p), %xmm6 # 4-byte Folded Reload
-; SSE2-NEXT:    # xmm6 = mem[0],zero,zero,zero
-; SSE2-NEXT:    movd {{[-0-9]+}}(%r{{[sb]}}p), %xmm11 # 4-byte Folded Reload
-; SSE2-NEXT:    # xmm11 = mem[0],zero,zero,zero
-; SSE2-NEXT:    movd {{[-0-9]+}}(%r{{[sb]}}p), %xmm3 # 4-byte Folded Reload
-; SSE2-NEXT:    # xmm3 = mem[0],zero,zero,zero
-; SSE2-NEXT:    movd {{[-0-9]+}}(%r{{[sb]}}p), %xmm14 # 4-byte Folded Reload
-; SSE2-NEXT:    # xmm14 = mem[0],zero,zero,zero
-; SSE2-NEXT:    movd {{[-0-9]+}}(%r{{[sb]}}p), %xmm9 # 4-byte Folded Reload
-; SSE2-NEXT:    # xmm9 = mem[0],zero,zero,zero
-; SSE2-NEXT:    movd {{[-0-9]+}}(%r{{[sb]}}p), %xmm15 # 4-byte Folded Reload
-; SSE2-NEXT:    # xmm15 = mem[0],zero,zero,zero
-; SSE2-NEXT:    movd {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 4-byte Folded Reload
-; SSE2-NEXT:    # xmm1 = mem[0],zero,zero,zero
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm12 = xmm12[0],xmm10[0],xmm12[1],xmm10[1],xmm12[2],xmm10[2],xmm12[3],xmm10[3]
-; SSE2-NEXT:    movss {{[-0-9]+}}(%r{{[sb]}}p), %xmm7 # 4-byte Reload
-; SSE2-NEXT:    # xmm7 = mem[0],zero,zero,zero
-; SSE2-NEXT:    movaps %xmm7, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm2 = xmm2[0],xmm8[0],xmm2[1],xmm8[1],xmm2[2],xmm8[2],xmm2[3],xmm8[3]
-; SSE2-NEXT:    movd {{[-0-9]+}}(%r{{[sb]}}p), %xmm8 # 4-byte Folded Reload
-; SSE2-NEXT:    # xmm8 = mem[0],zero,zero,zero
-; SSE2-NEXT:    punpckldq {{.*#+}} xmm2 = xmm2[0],xmm12[0],xmm2[1],xmm12[1]
-; SSE2-NEXT:    movss {{[-0-9]+}}(%r{{[sb]}}p), %xmm7 # 4-byte Reload
-; SSE2-NEXT:    # xmm7 = mem[0],zero,zero,zero
-; SSE2-NEXT:    movaps %xmm7, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm4[0],xmm0[1],xmm4[1],xmm0[2],xmm4[2],xmm0[3],xmm4[3]
-; SSE2-NEXT:    movd {{[-0-9]+}}(%r{{[sb]}}p), %xmm4 # 4-byte Folded Reload
-; SSE2-NEXT:    # xmm4 = mem[0],zero,zero,zero
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm5 = xmm5[0],xmm13[0],xmm5[1],xmm13[1],xmm5[2],xmm13[2],xmm5[3],xmm13[3]
-; SSE2-NEXT:    movd {{[-0-9]+}}(%r{{[sb]}}p), %xmm13 # 4-byte Folded Reload
-; SSE2-NEXT:    # xmm13 = mem[0],zero,zero,zero
-; SSE2-NEXT:    punpckldq {{.*#+}} xmm5 = xmm5[0],xmm0[0],xmm5[1],xmm0[1]
-; SSE2-NEXT:    movd {{[-0-9]+}}(%r{{[sb]}}p), %xmm7 # 4-byte Folded Reload
-; SSE2-NEXT:    # xmm7 = mem[0],zero,zero,zero
-; SSE2-NEXT:    punpcklqdq {{.*#+}} xmm5 = xmm5[0],xmm2[0]
-; SSE2-NEXT:    movd {{[-0-9]+}}(%r{{[sb]}}p), %xmm12 # 4-byte Folded Reload
-; SSE2-NEXT:    # xmm12 = mem[0],zero,zero,zero
-; SSE2-NEXT:    punpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm6 # 16-byte Folded Reload
-; SSE2-NEXT:    # xmm6 = xmm6[0],mem[0],xmm6[1],mem[1],xmm6[2],mem[2],xmm6[3],mem[3]
-; SSE2-NEXT:    movd {{[-0-9]+}}(%r{{[sb]}}p), %xmm2 # 4-byte Folded Reload
-; SSE2-NEXT:    # xmm2 = mem[0],zero,zero,zero
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm3 = xmm3[0],xmm11[0],xmm3[1],xmm11[1],xmm3[2],xmm11[2],xmm3[3],xmm11[3]
-; SSE2-NEXT:    movd %r10d, %xmm11
-; SSE2-NEXT:    punpckldq {{.*#+}} xmm3 = xmm3[0],xmm6[0],xmm3[1],xmm6[1]
-; SSE2-NEXT:    movd %r11d, %xmm6
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm9 = xmm9[0],xmm14[0],xmm9[1],xmm14[1],xmm9[2],xmm14[2],xmm9[3],xmm14[3]
-; SSE2-NEXT:    movd %ebx, %xmm14
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm15[0],xmm1[1],xmm15[1],xmm1[2],xmm15[2],xmm1[3],xmm15[3]
-; SSE2-NEXT:    movd %ebp, %xmm15
-; SSE2-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm9[0],xmm1[1],xmm9[1]
-; SSE2-NEXT:    movd {{[-0-9]+}}(%r{{[sb]}}p), %xmm10 # 4-byte Folded Reload
-; SSE2-NEXT:    # xmm10 = mem[0],zero,zero,zero
-; SSE2-NEXT:    punpcklqdq {{.*#+}} xmm1 = xmm1[0],xmm3[0]
-; SSE2-NEXT:    movd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 4-byte Folded Reload
-; SSE2-NEXT:    # xmm0 = mem[0],zero,zero,zero
-; SSE2-NEXT:    punpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm8 # 16-byte Folded Reload
-; SSE2-NEXT:    # xmm8 = xmm8[0],mem[0],xmm8[1],mem[1],xmm8[2],mem[2],xmm8[3],mem[3]
-; SSE2-NEXT:    movd {{[-0-9]+}}(%r{{[sb]}}p), %xmm9 # 4-byte Folded Reload
-; SSE2-NEXT:    # xmm9 = mem[0],zero,zero,zero
-; SSE2-NEXT:    punpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm4 # 16-byte Folded Reload
-; SSE2-NEXT:    # xmm4 = xmm4[0],mem[0],xmm4[1],mem[1],xmm4[2],mem[2],xmm4[3],mem[3]
-; SSE2-NEXT:    movd {{[-0-9]+}}(%r{{[sb]}}p), %xmm3 # 4-byte Folded Reload
-; SSE2-NEXT:    # xmm3 = mem[0],zero,zero,zero
-; SSE2-NEXT:    punpckldq {{.*#+}} xmm4 = xmm4[0],xmm8[0],xmm4[1],xmm8[1]
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm7 = xmm7[0],xmm13[0],xmm7[1],xmm13[1],xmm7[2],xmm13[2],xmm7[3],xmm13[3]
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm2 = xmm2[0],xmm12[0],xmm2[1],xmm12[1],xmm2[2],xmm12[2],xmm2[3],xmm12[3]
-; SSE2-NEXT:    punpckldq {{.*#+}} xmm2 = xmm2[0],xmm7[0],xmm2[1],xmm7[1]
-; SSE2-NEXT:    punpcklqdq {{.*#+}} xmm2 = xmm2[0],xmm4[0]
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm6 = xmm6[0],xmm11[0],xmm6[1],xmm11[1],xmm6[2],xmm11[2],xmm6[3],xmm11[3]
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm15 = xmm15[0],xmm14[0],xmm15[1],xmm14[1],xmm15[2],xmm14[2],xmm15[3],xmm14[3]
-; SSE2-NEXT:    punpckldq {{.*#+}} xmm15 = xmm15[0],xmm6[0],xmm15[1],xmm6[1]
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm10[0],xmm0[1],xmm10[1],xmm0[2],xmm10[2],xmm0[3],xmm10[3]
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm3 = xmm3[0],xmm9[0],xmm3[1],xmm9[1],xmm3[2],xmm9[2],xmm3[3],xmm9[3]
-; SSE2-NEXT:    punpckldq {{.*#+}} xmm3 = xmm3[0],xmm0[0],xmm3[1],xmm0[1]
-; SSE2-NEXT:    punpcklqdq {{.*#+}} xmm3 = xmm3[0],xmm15[0]
-; SSE2-NEXT:    movdqa %xmm5, %xmm0
-; SSE2-NEXT:    addq $8, %rsp
-; SSE2-NEXT:    popq %rbx
-; SSE2-NEXT:    popq %r12
-; SSE2-NEXT:    popq %r13
-; SSE2-NEXT:    popq %r14
-; SSE2-NEXT:    popq %r15
-; SSE2-NEXT:    popq %rbp
-; SSE2-NEXT:    retq
-;
-; SSSE3-LABEL: v32i16:
-; SSSE3:       # %bb.0:
-; SSSE3-NEXT:    pushq %rbp
-; SSSE3-NEXT:    pushq %r15
-; SSSE3-NEXT:    pushq %r14
-; SSSE3-NEXT:    pushq %r13
-; SSSE3-NEXT:    pushq %r12
-; SSSE3-NEXT:    pushq %rbx
-; SSSE3-NEXT:    pushq %rax
-; SSSE3-NEXT:    movd %xmm5, %eax
-; SSSE3-NEXT:    movd %xmm1, %esi
-; SSSE3-NEXT:    xorl %ecx, %ecx
-; SSSE3-NEXT:    movl %esi, %edx
-; SSSE3-NEXT:    addw %ax, %dx
-; SSSE3-NEXT:    setns %cl
-; SSSE3-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSSE3-NEXT:    addw %ax, %si
-; SSSE3-NEXT:    cmovol %ecx, %esi
-; SSSE3-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSSE3-NEXT:    pextrw $1, %xmm5, %eax
-; SSSE3-NEXT:    pextrw $1, %xmm1, %esi
-; SSSE3-NEXT:    xorl %ecx, %ecx
-; SSSE3-NEXT:    movl %esi, %edx
-; SSSE3-NEXT:    addw %ax, %dx
-; SSSE3-NEXT:    setns %cl
-; SSSE3-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSSE3-NEXT:    addw %ax, %si
-; SSSE3-NEXT:    cmovol %ecx, %esi
-; SSSE3-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSSE3-NEXT:    pextrw $2, %xmm5, %eax
-; SSSE3-NEXT:    pextrw $2, %xmm1, %esi
-; SSSE3-NEXT:    xorl %ecx, %ecx
-; SSSE3-NEXT:    movl %esi, %edx
-; SSSE3-NEXT:    addw %ax, %dx
-; SSSE3-NEXT:    setns %cl
-; SSSE3-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSSE3-NEXT:    addw %ax, %si
-; SSSE3-NEXT:    cmovol %ecx, %esi
-; SSSE3-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSSE3-NEXT:    pextrw $3, %xmm5, %eax
-; SSSE3-NEXT:    pextrw $3, %xmm1, %esi
-; SSSE3-NEXT:    xorl %ecx, %ecx
-; SSSE3-NEXT:    movl %esi, %edx
-; SSSE3-NEXT:    addw %ax, %dx
-; SSSE3-NEXT:    setns %cl
-; SSSE3-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSSE3-NEXT:    addw %ax, %si
-; SSSE3-NEXT:    cmovol %ecx, %esi
-; SSSE3-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSSE3-NEXT:    pextrw $4, %xmm5, %eax
-; SSSE3-NEXT:    pextrw $4, %xmm1, %esi
-; SSSE3-NEXT:    xorl %ecx, %ecx
-; SSSE3-NEXT:    movl %esi, %edx
-; SSSE3-NEXT:    addw %ax, %dx
-; SSSE3-NEXT:    setns %cl
-; SSSE3-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSSE3-NEXT:    addw %ax, %si
-; SSSE3-NEXT:    cmovol %ecx, %esi
-; SSSE3-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSSE3-NEXT:    pextrw $5, %xmm5, %eax
-; SSSE3-NEXT:    pextrw $5, %xmm1, %esi
-; SSSE3-NEXT:    xorl %ecx, %ecx
-; SSSE3-NEXT:    movl %esi, %edx
-; SSSE3-NEXT:    addw %ax, %dx
-; SSSE3-NEXT:    setns %cl
-; SSSE3-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSSE3-NEXT:    addw %ax, %si
-; SSSE3-NEXT:    cmovol %ecx, %esi
-; SSSE3-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSSE3-NEXT:    pextrw $6, %xmm5, %eax
-; SSSE3-NEXT:    pextrw $6, %xmm1, %esi
-; SSSE3-NEXT:    xorl %ecx, %ecx
-; SSSE3-NEXT:    movl %esi, %edx
-; SSSE3-NEXT:    addw %ax, %dx
-; SSSE3-NEXT:    setns %cl
-; SSSE3-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSSE3-NEXT:    addw %ax, %si
-; SSSE3-NEXT:    cmovol %ecx, %esi
-; SSSE3-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSSE3-NEXT:    pextrw $7, %xmm5, %eax
-; SSSE3-NEXT:    pextrw $7, %xmm1, %esi
-; SSSE3-NEXT:    xorl %ecx, %ecx
-; SSSE3-NEXT:    movl %esi, %edx
-; SSSE3-NEXT:    addw %ax, %dx
-; SSSE3-NEXT:    setns %cl
-; SSSE3-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSSE3-NEXT:    addw %ax, %si
-; SSSE3-NEXT:    cmovol %ecx, %esi
-; SSSE3-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSSE3-NEXT:    movd %xmm6, %eax
-; SSSE3-NEXT:    movd %xmm2, %esi
-; SSSE3-NEXT:    xorl %ecx, %ecx
-; SSSE3-NEXT:    movl %esi, %edx
-; SSSE3-NEXT:    addw %ax, %dx
-; SSSE3-NEXT:    setns %cl
-; SSSE3-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSSE3-NEXT:    addw %ax, %si
-; SSSE3-NEXT:    cmovol %ecx, %esi
-; SSSE3-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSSE3-NEXT:    pextrw $1, %xmm6, %eax
-; SSSE3-NEXT:    pextrw $1, %xmm2, %esi
-; SSSE3-NEXT:    xorl %ecx, %ecx
-; SSSE3-NEXT:    movl %esi, %edx
-; SSSE3-NEXT:    addw %ax, %dx
-; SSSE3-NEXT:    setns %cl
-; SSSE3-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSSE3-NEXT:    addw %ax, %si
-; SSSE3-NEXT:    cmovol %ecx, %esi
-; SSSE3-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSSE3-NEXT:    pextrw $2, %xmm6, %eax
-; SSSE3-NEXT:    pextrw $2, %xmm2, %esi
-; SSSE3-NEXT:    xorl %ecx, %ecx
-; SSSE3-NEXT:    movl %esi, %edx
-; SSSE3-NEXT:    addw %ax, %dx
-; SSSE3-NEXT:    setns %cl
-; SSSE3-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSSE3-NEXT:    addw %ax, %si
-; SSSE3-NEXT:    cmovol %ecx, %esi
-; SSSE3-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSSE3-NEXT:    pextrw $3, %xmm6, %eax
-; SSSE3-NEXT:    pextrw $3, %xmm2, %esi
-; SSSE3-NEXT:    xorl %ecx, %ecx
-; SSSE3-NEXT:    movl %esi, %edx
-; SSSE3-NEXT:    addw %ax, %dx
-; SSSE3-NEXT:    setns %cl
-; SSSE3-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSSE3-NEXT:    addw %ax, %si
-; SSSE3-NEXT:    cmovol %ecx, %esi
-; SSSE3-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSSE3-NEXT:    pextrw $4, %xmm6, %eax
-; SSSE3-NEXT:    pextrw $4, %xmm2, %esi
-; SSSE3-NEXT:    xorl %ecx, %ecx
-; SSSE3-NEXT:    movl %esi, %edx
-; SSSE3-NEXT:    addw %ax, %dx
-; SSSE3-NEXT:    setns %cl
-; SSSE3-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSSE3-NEXT:    addw %ax, %si
-; SSSE3-NEXT:    cmovol %ecx, %esi
-; SSSE3-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSSE3-NEXT:    pextrw $5, %xmm6, %eax
-; SSSE3-NEXT:    pextrw $5, %xmm2, %esi
-; SSSE3-NEXT:    xorl %ecx, %ecx
-; SSSE3-NEXT:    movl %esi, %edx
-; SSSE3-NEXT:    addw %ax, %dx
-; SSSE3-NEXT:    setns %cl
-; SSSE3-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSSE3-NEXT:    addw %ax, %si
-; SSSE3-NEXT:    cmovol %ecx, %esi
-; SSSE3-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSSE3-NEXT:    pextrw $6, %xmm6, %eax
-; SSSE3-NEXT:    pextrw $6, %xmm2, %esi
-; SSSE3-NEXT:    xorl %ecx, %ecx
-; SSSE3-NEXT:    movl %esi, %edx
-; SSSE3-NEXT:    addw %ax, %dx
-; SSSE3-NEXT:    setns %cl
-; SSSE3-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSSE3-NEXT:    addw %ax, %si
-; SSSE3-NEXT:    cmovol %ecx, %esi
-; SSSE3-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSSE3-NEXT:    pextrw $7, %xmm6, %eax
-; SSSE3-NEXT:    pextrw $7, %xmm2, %esi
-; SSSE3-NEXT:    xorl %ecx, %ecx
-; SSSE3-NEXT:    movl %esi, %edx
-; SSSE3-NEXT:    addw %ax, %dx
-; SSSE3-NEXT:    setns %cl
-; SSSE3-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSSE3-NEXT:    addw %ax, %si
-; SSSE3-NEXT:    cmovol %ecx, %esi
-; SSSE3-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSSE3-NEXT:    movd %xmm7, %eax
-; SSSE3-NEXT:    movd %xmm3, %esi
-; SSSE3-NEXT:    xorl %ecx, %ecx
-; SSSE3-NEXT:    movl %esi, %edx
-; SSSE3-NEXT:    addw %ax, %dx
-; SSSE3-NEXT:    setns %cl
-; SSSE3-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSSE3-NEXT:    addw %ax, %si
-; SSSE3-NEXT:    cmovol %ecx, %esi
-; SSSE3-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSSE3-NEXT:    pextrw $1, %xmm7, %eax
-; SSSE3-NEXT:    pextrw $1, %xmm3, %esi
-; SSSE3-NEXT:    xorl %ecx, %ecx
-; SSSE3-NEXT:    movl %esi, %edx
-; SSSE3-NEXT:    addw %ax, %dx
-; SSSE3-NEXT:    setns %cl
-; SSSE3-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSSE3-NEXT:    addw %ax, %si
-; SSSE3-NEXT:    cmovol %ecx, %esi
-; SSSE3-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSSE3-NEXT:    pextrw $2, %xmm7, %eax
-; SSSE3-NEXT:    pextrw $2, %xmm3, %esi
-; SSSE3-NEXT:    xorl %ecx, %ecx
-; SSSE3-NEXT:    movl %esi, %edx
-; SSSE3-NEXT:    addw %ax, %dx
-; SSSE3-NEXT:    setns %cl
-; SSSE3-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSSE3-NEXT:    addw %ax, %si
-; SSSE3-NEXT:    cmovol %ecx, %esi
-; SSSE3-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSSE3-NEXT:    pextrw $3, %xmm7, %eax
-; SSSE3-NEXT:    pextrw $3, %xmm3, %esi
-; SSSE3-NEXT:    xorl %ecx, %ecx
-; SSSE3-NEXT:    movl %esi, %edx
-; SSSE3-NEXT:    addw %ax, %dx
-; SSSE3-NEXT:    setns %cl
-; SSSE3-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSSE3-NEXT:    addw %ax, %si
-; SSSE3-NEXT:    cmovol %ecx, %esi
-; SSSE3-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSSE3-NEXT:    pextrw $4, %xmm7, %eax
-; SSSE3-NEXT:    pextrw $4, %xmm3, %ebp
-; SSSE3-NEXT:    xorl %ecx, %ecx
-; SSSE3-NEXT:    movl %ebp, %edx
-; SSSE3-NEXT:    addw %ax, %dx
-; SSSE3-NEXT:    setns %cl
-; SSSE3-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSSE3-NEXT:    addw %ax, %bp
-; SSSE3-NEXT:    cmovol %ecx, %ebp
-; SSSE3-NEXT:    pextrw $5, %xmm7, %eax
-; SSSE3-NEXT:    pextrw $5, %xmm3, %ebx
-; SSSE3-NEXT:    xorl %ecx, %ecx
-; SSSE3-NEXT:    movl %ebx, %edx
-; SSSE3-NEXT:    addw %ax, %dx
-; SSSE3-NEXT:    setns %cl
-; SSSE3-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSSE3-NEXT:    addw %ax, %bx
-; SSSE3-NEXT:    cmovol %ecx, %ebx
-; SSSE3-NEXT:    pextrw $6, %xmm7, %eax
-; SSSE3-NEXT:    pextrw $6, %xmm3, %r11d
-; SSSE3-NEXT:    xorl %ecx, %ecx
-; SSSE3-NEXT:    movl %r11d, %edx
-; SSSE3-NEXT:    addw %ax, %dx
-; SSSE3-NEXT:    setns %cl
-; SSSE3-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSSE3-NEXT:    addw %ax, %r11w
-; SSSE3-NEXT:    cmovol %ecx, %r11d
-; SSSE3-NEXT:    pextrw $7, %xmm7, %eax
-; SSSE3-NEXT:    pextrw $7, %xmm3, %r10d
-; SSSE3-NEXT:    xorl %ecx, %ecx
-; SSSE3-NEXT:    movl %r10d, %edx
-; SSSE3-NEXT:    addw %ax, %dx
-; SSSE3-NEXT:    setns %cl
-; SSSE3-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSSE3-NEXT:    addw %ax, %r10w
-; SSSE3-NEXT:    cmovol %ecx, %r10d
-; SSSE3-NEXT:    movd %xmm4, %eax
-; SSSE3-NEXT:    movd %xmm0, %r9d
-; SSSE3-NEXT:    xorl %ecx, %ecx
-; SSSE3-NEXT:    movl %r9d, %edx
-; SSSE3-NEXT:    addw %ax, %dx
-; SSSE3-NEXT:    setns %cl
-; SSSE3-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSSE3-NEXT:    addw %ax, %r9w
-; SSSE3-NEXT:    cmovol %ecx, %r9d
-; SSSE3-NEXT:    pextrw $1, %xmm4, %eax
-; SSSE3-NEXT:    pextrw $1, %xmm0, %r8d
-; SSSE3-NEXT:    xorl %ecx, %ecx
-; SSSE3-NEXT:    movl %r8d, %edx
-; SSSE3-NEXT:    addw %ax, %dx
-; SSSE3-NEXT:    setns %cl
-; SSSE3-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSSE3-NEXT:    addw %ax, %r8w
-; SSSE3-NEXT:    cmovol %ecx, %r8d
-; SSSE3-NEXT:    pextrw $2, %xmm4, %eax
-; SSSE3-NEXT:    pextrw $2, %xmm0, %edi
-; SSSE3-NEXT:    xorl %ecx, %ecx
-; SSSE3-NEXT:    movl %edi, %edx
-; SSSE3-NEXT:    addw %ax, %dx
-; SSSE3-NEXT:    setns %cl
-; SSSE3-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSSE3-NEXT:    addw %ax, %di
-; SSSE3-NEXT:    cmovol %ecx, %edi
-; SSSE3-NEXT:    pextrw $3, %xmm4, %eax
-; SSSE3-NEXT:    pextrw $3, %xmm0, %esi
-; SSSE3-NEXT:    xorl %ecx, %ecx
-; SSSE3-NEXT:    movl %esi, %edx
-; SSSE3-NEXT:    addw %ax, %dx
-; SSSE3-NEXT:    setns %cl
-; SSSE3-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSSE3-NEXT:    addw %ax, %si
-; SSSE3-NEXT:    cmovol %ecx, %esi
-; SSSE3-NEXT:    pextrw $4, %xmm4, %eax
-; SSSE3-NEXT:    pextrw $4, %xmm0, %edx
-; SSSE3-NEXT:    xorl %ecx, %ecx
-; SSSE3-NEXT:    movl %edx, %r13d
-; SSSE3-NEXT:    addw %ax, %r13w
-; SSSE3-NEXT:    setns %cl
-; SSSE3-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSSE3-NEXT:    addw %ax, %dx
-; SSSE3-NEXT:    cmovol %ecx, %edx
-; SSSE3-NEXT:    pextrw $5, %xmm4, %r13d
-; SSSE3-NEXT:    pextrw $5, %xmm0, %ecx
-; SSSE3-NEXT:    xorl %eax, %eax
-; SSSE3-NEXT:    movl %ecx, %r12d
-; SSSE3-NEXT:    addw %r13w, %r12w
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addl $32767, %eax # imm = 0x7FFF
-; SSSE3-NEXT:    addw %r13w, %cx
-; SSSE3-NEXT:    cmovol %eax, %ecx
-; SSSE3-NEXT:    pextrw $6, %xmm4, %r12d
-; SSSE3-NEXT:    pextrw $6, %xmm0, %r13d
-; SSSE3-NEXT:    xorl %eax, %eax
-; SSSE3-NEXT:    movl %r13d, %r15d
-; SSSE3-NEXT:    addw %r12w, %r15w
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addl $32767, %eax # imm = 0x7FFF
-; SSSE3-NEXT:    addw %r12w, %r13w
-; SSSE3-NEXT:    cmovol %eax, %r13d
-; SSSE3-NEXT:    pextrw $7, %xmm4, %r15d
-; SSSE3-NEXT:    pextrw $7, %xmm0, %eax
-; SSSE3-NEXT:    xorl %r12d, %r12d
-; SSSE3-NEXT:    movl %eax, %r14d
-; SSSE3-NEXT:    addw %r15w, %r14w
-; SSSE3-NEXT:    setns %r12b
-; SSSE3-NEXT:    addl $32767, %r12d # imm = 0x7FFF
-; SSSE3-NEXT:    addw %r15w, %ax
-; SSSE3-NEXT:    cmovol %r12d, %eax
-; SSSE3-NEXT:    movd %eax, %xmm10
-; SSSE3-NEXT:    movd %r13d, %xmm12
-; SSSE3-NEXT:    movd %ecx, %xmm8
-; SSSE3-NEXT:    movd %edx, %xmm2
-; SSSE3-NEXT:    movd %esi, %xmm4
-; SSSE3-NEXT:    movd %edi, %xmm0
-; SSSE3-NEXT:    movd %r8d, %xmm13
-; SSSE3-NEXT:    movd %r9d, %xmm5
-; SSSE3-NEXT:    movss {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 4-byte Reload
-; SSSE3-NEXT:    # xmm1 = mem[0],zero,zero,zero
-; SSSE3-NEXT:    movaps %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; SSSE3-NEXT:    movd {{[-0-9]+}}(%r{{[sb]}}p), %xmm6 # 4-byte Folded Reload
-; SSSE3-NEXT:    # xmm6 = mem[0],zero,zero,zero
-; SSSE3-NEXT:    movd {{[-0-9]+}}(%r{{[sb]}}p), %xmm11 # 4-byte Folded Reload
-; SSSE3-NEXT:    # xmm11 = mem[0],zero,zero,zero
-; SSSE3-NEXT:    movd {{[-0-9]+}}(%r{{[sb]}}p), %xmm3 # 4-byte Folded Reload
-; SSSE3-NEXT:    # xmm3 = mem[0],zero,zero,zero
-; SSSE3-NEXT:    movd {{[-0-9]+}}(%r{{[sb]}}p), %xmm14 # 4-byte Folded Reload
-; SSSE3-NEXT:    # xmm14 = mem[0],zero,zero,zero
-; SSSE3-NEXT:    movd {{[-0-9]+}}(%r{{[sb]}}p), %xmm9 # 4-byte Folded Reload
-; SSSE3-NEXT:    # xmm9 = mem[0],zero,zero,zero
-; SSSE3-NEXT:    movd {{[-0-9]+}}(%r{{[sb]}}p), %xmm15 # 4-byte Folded Reload
-; SSSE3-NEXT:    # xmm15 = mem[0],zero,zero,zero
-; SSSE3-NEXT:    movd {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 4-byte Folded Reload
-; SSSE3-NEXT:    # xmm1 = mem[0],zero,zero,zero
-; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm12 = xmm12[0],xmm10[0],xmm12[1],xmm10[1],xmm12[2],xmm10[2],xmm12[3],xmm10[3]
-; SSSE3-NEXT:    movss {{[-0-9]+}}(%r{{[sb]}}p), %xmm7 # 4-byte Reload
-; SSSE3-NEXT:    # xmm7 = mem[0],zero,zero,zero
-; SSSE3-NEXT:    movaps %xmm7, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm2 = xmm2[0],xmm8[0],xmm2[1],xmm8[1],xmm2[2],xmm8[2],xmm2[3],xmm8[3]
-; SSSE3-NEXT:    movd {{[-0-9]+}}(%r{{[sb]}}p), %xmm8 # 4-byte Folded Reload
-; SSSE3-NEXT:    # xmm8 = mem[0],zero,zero,zero
-; SSSE3-NEXT:    punpckldq {{.*#+}} xmm2 = xmm2[0],xmm12[0],xmm2[1],xmm12[1]
-; SSSE3-NEXT:    movss {{[-0-9]+}}(%r{{[sb]}}p), %xmm7 # 4-byte Reload
-; SSSE3-NEXT:    # xmm7 = mem[0],zero,zero,zero
-; SSSE3-NEXT:    movaps %xmm7, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm4[0],xmm0[1],xmm4[1],xmm0[2],xmm4[2],xmm0[3],xmm4[3]
-; SSSE3-NEXT:    movd {{[-0-9]+}}(%r{{[sb]}}p), %xmm4 # 4-byte Folded Reload
-; SSSE3-NEXT:    # xmm4 = mem[0],zero,zero,zero
-; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm5 = xmm5[0],xmm13[0],xmm5[1],xmm13[1],xmm5[2],xmm13[2],xmm5[3],xmm13[3]
-; SSSE3-NEXT:    movd {{[-0-9]+}}(%r{{[sb]}}p), %xmm13 # 4-byte Folded Reload
-; SSSE3-NEXT:    # xmm13 = mem[0],zero,zero,zero
-; SSSE3-NEXT:    punpckldq {{.*#+}} xmm5 = xmm5[0],xmm0[0],xmm5[1],xmm0[1]
-; SSSE3-NEXT:    movd {{[-0-9]+}}(%r{{[sb]}}p), %xmm7 # 4-byte Folded Reload
-; SSSE3-NEXT:    # xmm7 = mem[0],zero,zero,zero
-; SSSE3-NEXT:    punpcklqdq {{.*#+}} xmm5 = xmm5[0],xmm2[0]
-; SSSE3-NEXT:    movd {{[-0-9]+}}(%r{{[sb]}}p), %xmm12 # 4-byte Folded Reload
-; SSSE3-NEXT:    # xmm12 = mem[0],zero,zero,zero
-; SSSE3-NEXT:    punpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm6 # 16-byte Folded Reload
-; SSSE3-NEXT:    # xmm6 = xmm6[0],mem[0],xmm6[1],mem[1],xmm6[2],mem[2],xmm6[3],mem[3]
-; SSSE3-NEXT:    movd {{[-0-9]+}}(%r{{[sb]}}p), %xmm2 # 4-byte Folded Reload
-; SSSE3-NEXT:    # xmm2 = mem[0],zero,zero,zero
-; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm3 = xmm3[0],xmm11[0],xmm3[1],xmm11[1],xmm3[2],xmm11[2],xmm3[3],xmm11[3]
-; SSSE3-NEXT:    movd %r10d, %xmm11
-; SSSE3-NEXT:    punpckldq {{.*#+}} xmm3 = xmm3[0],xmm6[0],xmm3[1],xmm6[1]
-; SSSE3-NEXT:    movd %r11d, %xmm6
-; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm9 = xmm9[0],xmm14[0],xmm9[1],xmm14[1],xmm9[2],xmm14[2],xmm9[3],xmm14[3]
-; SSSE3-NEXT:    movd %ebx, %xmm14
-; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm15[0],xmm1[1],xmm15[1],xmm1[2],xmm15[2],xmm1[3],xmm15[3]
-; SSSE3-NEXT:    movd %ebp, %xmm15
-; SSSE3-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm9[0],xmm1[1],xmm9[1]
-; SSSE3-NEXT:    movd {{[-0-9]+}}(%r{{[sb]}}p), %xmm10 # 4-byte Folded Reload
-; SSSE3-NEXT:    # xmm10 = mem[0],zero,zero,zero
-; SSSE3-NEXT:    punpcklqdq {{.*#+}} xmm1 = xmm1[0],xmm3[0]
-; SSSE3-NEXT:    movd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 4-byte Folded Reload
-; SSSE3-NEXT:    # xmm0 = mem[0],zero,zero,zero
-; SSSE3-NEXT:    punpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm8 # 16-byte Folded Reload
-; SSSE3-NEXT:    # xmm8 = xmm8[0],mem[0],xmm8[1],mem[1],xmm8[2],mem[2],xmm8[3],mem[3]
-; SSSE3-NEXT:    movd {{[-0-9]+}}(%r{{[sb]}}p), %xmm9 # 4-byte Folded Reload
-; SSSE3-NEXT:    # xmm9 = mem[0],zero,zero,zero
-; SSSE3-NEXT:    punpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm4 # 16-byte Folded Reload
-; SSSE3-NEXT:    # xmm4 = xmm4[0],mem[0],xmm4[1],mem[1],xmm4[2],mem[2],xmm4[3],mem[3]
-; SSSE3-NEXT:    movd {{[-0-9]+}}(%r{{[sb]}}p), %xmm3 # 4-byte Folded Reload
-; SSSE3-NEXT:    # xmm3 = mem[0],zero,zero,zero
-; SSSE3-NEXT:    punpckldq {{.*#+}} xmm4 = xmm4[0],xmm8[0],xmm4[1],xmm8[1]
-; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm7 = xmm7[0],xmm13[0],xmm7[1],xmm13[1],xmm7[2],xmm13[2],xmm7[3],xmm13[3]
-; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm2 = xmm2[0],xmm12[0],xmm2[1],xmm12[1],xmm2[2],xmm12[2],xmm2[3],xmm12[3]
-; SSSE3-NEXT:    punpckldq {{.*#+}} xmm2 = xmm2[0],xmm7[0],xmm2[1],xmm7[1]
-; SSSE3-NEXT:    punpcklqdq {{.*#+}} xmm2 = xmm2[0],xmm4[0]
-; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm6 = xmm6[0],xmm11[0],xmm6[1],xmm11[1],xmm6[2],xmm11[2],xmm6[3],xmm11[3]
-; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm15 = xmm15[0],xmm14[0],xmm15[1],xmm14[1],xmm15[2],xmm14[2],xmm15[3],xmm14[3]
-; SSSE3-NEXT:    punpckldq {{.*#+}} xmm15 = xmm15[0],xmm6[0],xmm15[1],xmm6[1]
-; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm10[0],xmm0[1],xmm10[1],xmm0[2],xmm10[2],xmm0[3],xmm10[3]
-; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm3 = xmm3[0],xmm9[0],xmm3[1],xmm9[1],xmm3[2],xmm9[2],xmm3[3],xmm9[3]
-; SSSE3-NEXT:    punpckldq {{.*#+}} xmm3 = xmm3[0],xmm0[0],xmm3[1],xmm0[1]
-; SSSE3-NEXT:    punpcklqdq {{.*#+}} xmm3 = xmm3[0],xmm15[0]
-; SSSE3-NEXT:    movdqa %xmm5, %xmm0
-; SSSE3-NEXT:    addq $8, %rsp
-; SSSE3-NEXT:    popq %rbx
-; SSSE3-NEXT:    popq %r12
-; SSSE3-NEXT:    popq %r13
-; SSSE3-NEXT:    popq %r14
-; SSSE3-NEXT:    popq %r15
-; SSSE3-NEXT:    popq %rbp
-; SSSE3-NEXT:    retq
-;
-; SSE41-LABEL: v32i16:
-; SSE41:       # %bb.0:
-; SSE41-NEXT:    pushq %rbp
-; SSE41-NEXT:    pushq %r15
-; SSE41-NEXT:    pushq %r14
-; SSE41-NEXT:    pushq %r13
-; SSE41-NEXT:    pushq %r12
-; SSE41-NEXT:    pushq %rbx
-; SSE41-NEXT:    pextrw $7, %xmm5, %eax
-; SSE41-NEXT:    pextrw $7, %xmm1, %esi
-; SSE41-NEXT:    xorl %ecx, %ecx
-; SSE41-NEXT:    movl %esi, %edx
-; SSE41-NEXT:    addw %ax, %dx
-; SSE41-NEXT:    setns %cl
-; SSE41-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSE41-NEXT:    addw %ax, %si
-; SSE41-NEXT:    cmovol %ecx, %esi
-; SSE41-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE41-NEXT:    pextrw $6, %xmm5, %eax
-; SSE41-NEXT:    pextrw $6, %xmm1, %esi
-; SSE41-NEXT:    xorl %ecx, %ecx
-; SSE41-NEXT:    movl %esi, %edx
-; SSE41-NEXT:    addw %ax, %dx
-; SSE41-NEXT:    setns %cl
-; SSE41-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSE41-NEXT:    addw %ax, %si
-; SSE41-NEXT:    cmovol %ecx, %esi
-; SSE41-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE41-NEXT:    pextrw $5, %xmm5, %eax
-; SSE41-NEXT:    pextrw $5, %xmm1, %esi
-; SSE41-NEXT:    xorl %ecx, %ecx
-; SSE41-NEXT:    movl %esi, %edx
-; SSE41-NEXT:    addw %ax, %dx
-; SSE41-NEXT:    setns %cl
-; SSE41-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSE41-NEXT:    addw %ax, %si
-; SSE41-NEXT:    cmovol %ecx, %esi
-; SSE41-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE41-NEXT:    pextrw $4, %xmm5, %eax
-; SSE41-NEXT:    pextrw $4, %xmm1, %esi
-; SSE41-NEXT:    xorl %ecx, %ecx
-; SSE41-NEXT:    movl %esi, %edx
-; SSE41-NEXT:    addw %ax, %dx
-; SSE41-NEXT:    setns %cl
-; SSE41-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSE41-NEXT:    addw %ax, %si
-; SSE41-NEXT:    cmovol %ecx, %esi
-; SSE41-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE41-NEXT:    pextrw $3, %xmm5, %eax
-; SSE41-NEXT:    pextrw $3, %xmm1, %esi
-; SSE41-NEXT:    xorl %ecx, %ecx
-; SSE41-NEXT:    movl %esi, %edx
-; SSE41-NEXT:    addw %ax, %dx
-; SSE41-NEXT:    setns %cl
-; SSE41-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSE41-NEXT:    addw %ax, %si
-; SSE41-NEXT:    cmovol %ecx, %esi
-; SSE41-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE41-NEXT:    pextrw $2, %xmm5, %eax
-; SSE41-NEXT:    pextrw $2, %xmm1, %esi
-; SSE41-NEXT:    xorl %ecx, %ecx
-; SSE41-NEXT:    movl %esi, %edx
-; SSE41-NEXT:    addw %ax, %dx
-; SSE41-NEXT:    setns %cl
-; SSE41-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSE41-NEXT:    addw %ax, %si
-; SSE41-NEXT:    cmovol %ecx, %esi
-; SSE41-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE41-NEXT:    movd %xmm5, %eax
-; SSE41-NEXT:    movd %xmm1, %esi
-; SSE41-NEXT:    xorl %ecx, %ecx
-; SSE41-NEXT:    movl %esi, %edx
-; SSE41-NEXT:    addw %ax, %dx
-; SSE41-NEXT:    setns %cl
-; SSE41-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSE41-NEXT:    addw %ax, %si
-; SSE41-NEXT:    cmovol %ecx, %esi
-; SSE41-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE41-NEXT:    pextrw $1, %xmm5, %eax
-; SSE41-NEXT:    pextrw $1, %xmm1, %esi
-; SSE41-NEXT:    xorl %ecx, %ecx
-; SSE41-NEXT:    movl %esi, %edx
-; SSE41-NEXT:    addw %ax, %dx
-; SSE41-NEXT:    setns %cl
-; SSE41-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSE41-NEXT:    addw %ax, %si
-; SSE41-NEXT:    cmovol %ecx, %esi
-; SSE41-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE41-NEXT:    pextrw $7, %xmm6, %eax
-; SSE41-NEXT:    pextrw $7, %xmm2, %esi
-; SSE41-NEXT:    xorl %ecx, %ecx
-; SSE41-NEXT:    movl %esi, %edx
-; SSE41-NEXT:    addw %ax, %dx
-; SSE41-NEXT:    setns %cl
-; SSE41-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSE41-NEXT:    addw %ax, %si
-; SSE41-NEXT:    cmovol %ecx, %esi
-; SSE41-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE41-NEXT:    pextrw $6, %xmm6, %eax
-; SSE41-NEXT:    pextrw $6, %xmm2, %esi
-; SSE41-NEXT:    xorl %ecx, %ecx
-; SSE41-NEXT:    movl %esi, %edx
-; SSE41-NEXT:    addw %ax, %dx
-; SSE41-NEXT:    setns %cl
-; SSE41-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSE41-NEXT:    addw %ax, %si
-; SSE41-NEXT:    cmovol %ecx, %esi
-; SSE41-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE41-NEXT:    pextrw $5, %xmm6, %eax
-; SSE41-NEXT:    pextrw $5, %xmm2, %esi
-; SSE41-NEXT:    xorl %ecx, %ecx
-; SSE41-NEXT:    movl %esi, %edx
-; SSE41-NEXT:    addw %ax, %dx
-; SSE41-NEXT:    setns %cl
-; SSE41-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSE41-NEXT:    addw %ax, %si
-; SSE41-NEXT:    cmovol %ecx, %esi
-; SSE41-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE41-NEXT:    pextrw $4, %xmm6, %eax
-; SSE41-NEXT:    pextrw $4, %xmm2, %esi
-; SSE41-NEXT:    xorl %ecx, %ecx
-; SSE41-NEXT:    movl %esi, %edx
-; SSE41-NEXT:    addw %ax, %dx
-; SSE41-NEXT:    setns %cl
-; SSE41-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSE41-NEXT:    addw %ax, %si
-; SSE41-NEXT:    cmovol %ecx, %esi
-; SSE41-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE41-NEXT:    pextrw $3, %xmm6, %eax
-; SSE41-NEXT:    pextrw $3, %xmm2, %esi
-; SSE41-NEXT:    xorl %ecx, %ecx
-; SSE41-NEXT:    movl %esi, %edx
-; SSE41-NEXT:    addw %ax, %dx
-; SSE41-NEXT:    setns %cl
-; SSE41-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSE41-NEXT:    addw %ax, %si
-; SSE41-NEXT:    cmovol %ecx, %esi
-; SSE41-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE41-NEXT:    pextrw $2, %xmm6, %eax
-; SSE41-NEXT:    pextrw $2, %xmm2, %esi
-; SSE41-NEXT:    xorl %ecx, %ecx
-; SSE41-NEXT:    movl %esi, %edx
-; SSE41-NEXT:    addw %ax, %dx
-; SSE41-NEXT:    setns %cl
-; SSE41-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSE41-NEXT:    addw %ax, %si
-; SSE41-NEXT:    cmovol %ecx, %esi
-; SSE41-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE41-NEXT:    movd %xmm6, %eax
-; SSE41-NEXT:    movd %xmm2, %esi
-; SSE41-NEXT:    xorl %ecx, %ecx
-; SSE41-NEXT:    movl %esi, %edx
-; SSE41-NEXT:    addw %ax, %dx
-; SSE41-NEXT:    setns %cl
-; SSE41-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSE41-NEXT:    addw %ax, %si
-; SSE41-NEXT:    cmovol %ecx, %esi
-; SSE41-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE41-NEXT:    pextrw $1, %xmm6, %eax
-; SSE41-NEXT:    pextrw $1, %xmm2, %esi
-; SSE41-NEXT:    xorl %ecx, %ecx
-; SSE41-NEXT:    movl %esi, %edx
-; SSE41-NEXT:    addw %ax, %dx
-; SSE41-NEXT:    setns %cl
-; SSE41-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSE41-NEXT:    addw %ax, %si
-; SSE41-NEXT:    cmovol %ecx, %esi
-; SSE41-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE41-NEXT:    pextrw $7, %xmm7, %eax
-; SSE41-NEXT:    pextrw $7, %xmm3, %esi
-; SSE41-NEXT:    xorl %ecx, %ecx
-; SSE41-NEXT:    movl %esi, %edx
-; SSE41-NEXT:    addw %ax, %dx
-; SSE41-NEXT:    setns %cl
-; SSE41-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSE41-NEXT:    addw %ax, %si
-; SSE41-NEXT:    cmovol %ecx, %esi
-; SSE41-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE41-NEXT:    pextrw $6, %xmm7, %eax
-; SSE41-NEXT:    pextrw $6, %xmm3, %esi
-; SSE41-NEXT:    xorl %ecx, %ecx
-; SSE41-NEXT:    movl %esi, %edx
-; SSE41-NEXT:    addw %ax, %dx
-; SSE41-NEXT:    setns %cl
-; SSE41-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSE41-NEXT:    addw %ax, %si
-; SSE41-NEXT:    cmovol %ecx, %esi
-; SSE41-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE41-NEXT:    pextrw $5, %xmm7, %eax
-; SSE41-NEXT:    pextrw $5, %xmm3, %esi
-; SSE41-NEXT:    xorl %ecx, %ecx
-; SSE41-NEXT:    movl %esi, %edx
-; SSE41-NEXT:    addw %ax, %dx
-; SSE41-NEXT:    setns %cl
-; SSE41-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSE41-NEXT:    addw %ax, %si
-; SSE41-NEXT:    cmovol %ecx, %esi
-; SSE41-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE41-NEXT:    pextrw $4, %xmm7, %eax
-; SSE41-NEXT:    pextrw $4, %xmm3, %esi
-; SSE41-NEXT:    xorl %ecx, %ecx
-; SSE41-NEXT:    movl %esi, %edx
-; SSE41-NEXT:    addw %ax, %dx
-; SSE41-NEXT:    setns %cl
-; SSE41-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSE41-NEXT:    addw %ax, %si
-; SSE41-NEXT:    cmovol %ecx, %esi
-; SSE41-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE41-NEXT:    pextrw $3, %xmm7, %eax
-; SSE41-NEXT:    pextrw $3, %xmm3, %ebx
-; SSE41-NEXT:    xorl %ecx, %ecx
-; SSE41-NEXT:    movl %ebx, %edx
-; SSE41-NEXT:    addw %ax, %dx
-; SSE41-NEXT:    setns %cl
-; SSE41-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSE41-NEXT:    addw %ax, %bx
-; SSE41-NEXT:    cmovol %ecx, %ebx
-; SSE41-NEXT:    pextrw $2, %xmm7, %eax
-; SSE41-NEXT:    pextrw $2, %xmm3, %r11d
-; SSE41-NEXT:    xorl %ecx, %ecx
-; SSE41-NEXT:    movl %r11d, %edx
-; SSE41-NEXT:    addw %ax, %dx
-; SSE41-NEXT:    setns %cl
-; SSE41-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSE41-NEXT:    addw %ax, %r11w
-; SSE41-NEXT:    cmovol %ecx, %r11d
-; SSE41-NEXT:    movd %xmm7, %eax
-; SSE41-NEXT:    movd %xmm3, %r10d
-; SSE41-NEXT:    xorl %ecx, %ecx
-; SSE41-NEXT:    movl %r10d, %edx
-; SSE41-NEXT:    addw %ax, %dx
-; SSE41-NEXT:    setns %cl
-; SSE41-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSE41-NEXT:    addw %ax, %r10w
-; SSE41-NEXT:    cmovol %ecx, %r10d
-; SSE41-NEXT:    pextrw $1, %xmm7, %eax
-; SSE41-NEXT:    pextrw $1, %xmm3, %r9d
-; SSE41-NEXT:    xorl %ecx, %ecx
-; SSE41-NEXT:    movl %r9d, %edx
-; SSE41-NEXT:    addw %ax, %dx
-; SSE41-NEXT:    setns %cl
-; SSE41-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSE41-NEXT:    addw %ax, %r9w
-; SSE41-NEXT:    cmovol %ecx, %r9d
-; SSE41-NEXT:    pextrw $7, %xmm4, %eax
-; SSE41-NEXT:    pextrw $7, %xmm0, %r8d
-; SSE41-NEXT:    xorl %ecx, %ecx
-; SSE41-NEXT:    movl %r8d, %edx
-; SSE41-NEXT:    addw %ax, %dx
-; SSE41-NEXT:    setns %cl
-; SSE41-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSE41-NEXT:    addw %ax, %r8w
-; SSE41-NEXT:    cmovol %ecx, %r8d
-; SSE41-NEXT:    pextrw $6, %xmm4, %eax
-; SSE41-NEXT:    pextrw $6, %xmm0, %edi
-; SSE41-NEXT:    xorl %ecx, %ecx
-; SSE41-NEXT:    movl %edi, %edx
-; SSE41-NEXT:    addw %ax, %dx
-; SSE41-NEXT:    setns %cl
-; SSE41-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSE41-NEXT:    addw %ax, %di
-; SSE41-NEXT:    cmovol %ecx, %edi
-; SSE41-NEXT:    pextrw $5, %xmm4, %eax
-; SSE41-NEXT:    pextrw $5, %xmm0, %esi
-; SSE41-NEXT:    xorl %ecx, %ecx
-; SSE41-NEXT:    movl %esi, %edx
-; SSE41-NEXT:    addw %ax, %dx
-; SSE41-NEXT:    setns %cl
-; SSE41-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSE41-NEXT:    addw %ax, %si
-; SSE41-NEXT:    cmovol %ecx, %esi
-; SSE41-NEXT:    pextrw $4, %xmm4, %eax
-; SSE41-NEXT:    pextrw $4, %xmm0, %edx
-; SSE41-NEXT:    xorl %ecx, %ecx
-; SSE41-NEXT:    movl %edx, %r13d
-; SSE41-NEXT:    addw %ax, %r13w
-; SSE41-NEXT:    setns %cl
-; SSE41-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSE41-NEXT:    addw %ax, %dx
-; SSE41-NEXT:    cmovol %ecx, %edx
-; SSE41-NEXT:    pextrw $3, %xmm4, %eax
-; SSE41-NEXT:    pextrw $3, %xmm0, %r13d
-; SSE41-NEXT:    xorl %ecx, %ecx
-; SSE41-NEXT:    movl %r13d, %r12d
-; SSE41-NEXT:    addw %ax, %r12w
-; SSE41-NEXT:    setns %cl
-; SSE41-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSE41-NEXT:    addw %ax, %r13w
-; SSE41-NEXT:    cmovol %ecx, %r13d
-; SSE41-NEXT:    pextrw $2, %xmm4, %r12d
-; SSE41-NEXT:    pextrw $2, %xmm0, %eax
-; SSE41-NEXT:    xorl %ecx, %ecx
-; SSE41-NEXT:    movl %eax, %r15d
-; SSE41-NEXT:    addw %r12w, %r15w
-; SSE41-NEXT:    setns %cl
-; SSE41-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSE41-NEXT:    addw %r12w, %ax
-; SSE41-NEXT:    cmovol %ecx, %eax
-; SSE41-NEXT:    movd %xmm4, %r15d
-; SSE41-NEXT:    movd %xmm0, %r12d
-; SSE41-NEXT:    xorl %ecx, %ecx
-; SSE41-NEXT:    movl %r12d, %r14d
-; SSE41-NEXT:    addw %r15w, %r14w
-; SSE41-NEXT:    setns %cl
-; SSE41-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSE41-NEXT:    addw %r15w, %r12w
-; SSE41-NEXT:    cmovol %ecx, %r12d
-; SSE41-NEXT:    pextrw $1, %xmm4, %r14d
-; SSE41-NEXT:    pextrw $1, %xmm0, %ecx
-; SSE41-NEXT:    xorl %r15d, %r15d
-; SSE41-NEXT:    movl %ecx, %ebp
-; SSE41-NEXT:    addw %r14w, %bp
-; SSE41-NEXT:    setns %r15b
-; SSE41-NEXT:    addl $32767, %r15d # imm = 0x7FFF
-; SSE41-NEXT:    addw %r14w, %cx
-; SSE41-NEXT:    cmovol %r15d, %ecx
-; SSE41-NEXT:    movd %r12d, %xmm0
-; SSE41-NEXT:    pinsrw $1, %ecx, %xmm0
-; SSE41-NEXT:    pinsrw $2, %eax, %xmm0
-; SSE41-NEXT:    pinsrw $3, %r13d, %xmm0
-; SSE41-NEXT:    pinsrw $4, %edx, %xmm0
-; SSE41-NEXT:    pinsrw $5, %esi, %xmm0
-; SSE41-NEXT:    pinsrw $6, %edi, %xmm0
-; SSE41-NEXT:    pinsrw $7, %r8d, %xmm0
-; SSE41-NEXT:    movd {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 4-byte Folded Reload
-; SSE41-NEXT:    # xmm1 = mem[0],zero,zero,zero
-; SSE41-NEXT:    pinsrw $1, {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 4-byte Folded Reload
-; SSE41-NEXT:    pinsrw $2, {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 4-byte Folded Reload
-; SSE41-NEXT:    pinsrw $3, {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 4-byte Folded Reload
-; SSE41-NEXT:    pinsrw $4, {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 4-byte Folded Reload
-; SSE41-NEXT:    pinsrw $5, {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 4-byte Folded Reload
-; SSE41-NEXT:    pinsrw $6, {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 4-byte Folded Reload
-; SSE41-NEXT:    pinsrw $7, {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 4-byte Folded Reload
-; SSE41-NEXT:    movd {{[-0-9]+}}(%r{{[sb]}}p), %xmm2 # 4-byte Folded Reload
-; SSE41-NEXT:    # xmm2 = mem[0],zero,zero,zero
-; SSE41-NEXT:    pinsrw $1, {{[-0-9]+}}(%r{{[sb]}}p), %xmm2 # 4-byte Folded Reload
-; SSE41-NEXT:    pinsrw $2, {{[-0-9]+}}(%r{{[sb]}}p), %xmm2 # 4-byte Folded Reload
-; SSE41-NEXT:    pinsrw $3, {{[-0-9]+}}(%r{{[sb]}}p), %xmm2 # 4-byte Folded Reload
-; SSE41-NEXT:    pinsrw $4, {{[-0-9]+}}(%r{{[sb]}}p), %xmm2 # 4-byte Folded Reload
-; SSE41-NEXT:    pinsrw $5, {{[-0-9]+}}(%r{{[sb]}}p), %xmm2 # 4-byte Folded Reload
-; SSE41-NEXT:    pinsrw $6, {{[-0-9]+}}(%r{{[sb]}}p), %xmm2 # 4-byte Folded Reload
-; SSE41-NEXT:    pinsrw $7, {{[-0-9]+}}(%r{{[sb]}}p), %xmm2 # 4-byte Folded Reload
-; SSE41-NEXT:    movd %r10d, %xmm3
-; SSE41-NEXT:    pinsrw $1, %r9d, %xmm3
-; SSE41-NEXT:    pinsrw $2, %r11d, %xmm3
-; SSE41-NEXT:    pinsrw $3, %ebx, %xmm3
-; SSE41-NEXT:    pinsrw $4, {{[-0-9]+}}(%r{{[sb]}}p), %xmm3 # 4-byte Folded Reload
-; SSE41-NEXT:    pinsrw $5, {{[-0-9]+}}(%r{{[sb]}}p), %xmm3 # 4-byte Folded Reload
-; SSE41-NEXT:    pinsrw $6, {{[-0-9]+}}(%r{{[sb]}}p), %xmm3 # 4-byte Folded Reload
-; SSE41-NEXT:    pinsrw $7, {{[-0-9]+}}(%r{{[sb]}}p), %xmm3 # 4-byte Folded Reload
-; SSE41-NEXT:    popq %rbx
-; SSE41-NEXT:    popq %r12
-; SSE41-NEXT:    popq %r13
-; SSE41-NEXT:    popq %r14
-; SSE41-NEXT:    popq %r15
-; SSE41-NEXT:    popq %rbp
-; SSE41-NEXT:    retq
+; SSE-LABEL: v32i16:
+; SSE:       # %bb.0:
+; SSE-NEXT:    paddsw %xmm4, %xmm0
+; SSE-NEXT:    paddsw %xmm5, %xmm1
+; SSE-NEXT:    paddsw %xmm6, %xmm2
+; SSE-NEXT:    paddsw %xmm7, %xmm3
+; SSE-NEXT:    retq
 ;
 ; AVX1-LABEL: v32i16:
 ; AVX1:       # %bb.0:
-; AVX1-NEXT:    pushq %rbp
-; AVX1-NEXT:    pushq %r15
-; AVX1-NEXT:    pushq %r14
-; AVX1-NEXT:    pushq %r13
-; AVX1-NEXT:    pushq %r12
-; AVX1-NEXT:    pushq %rbx
-; AVX1-NEXT:    vpextrw $7, %xmm3, %eax
-; AVX1-NEXT:    vpextrw $7, %xmm1, %esi
-; AVX1-NEXT:    xorl %ecx, %ecx
-; AVX1-NEXT:    movl %esi, %edx
-; AVX1-NEXT:    addw %ax, %dx
-; AVX1-NEXT:    setns %cl
-; AVX1-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX1-NEXT:    addw %ax, %si
-; AVX1-NEXT:    cmovol %ecx, %esi
-; AVX1-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX1-NEXT:    vpextrw $6, %xmm3, %eax
-; AVX1-NEXT:    vpextrw $6, %xmm1, %esi
-; AVX1-NEXT:    xorl %ecx, %ecx
-; AVX1-NEXT:    movl %esi, %edx
-; AVX1-NEXT:    addw %ax, %dx
-; AVX1-NEXT:    setns %cl
-; AVX1-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX1-NEXT:    addw %ax, %si
-; AVX1-NEXT:    cmovol %ecx, %esi
-; AVX1-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX1-NEXT:    vpextrw $5, %xmm3, %eax
-; AVX1-NEXT:    vpextrw $5, %xmm1, %esi
-; AVX1-NEXT:    xorl %ecx, %ecx
-; AVX1-NEXT:    movl %esi, %edx
-; AVX1-NEXT:    addw %ax, %dx
-; AVX1-NEXT:    setns %cl
-; AVX1-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX1-NEXT:    addw %ax, %si
-; AVX1-NEXT:    cmovol %ecx, %esi
-; AVX1-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX1-NEXT:    vpextrw $4, %xmm3, %eax
-; AVX1-NEXT:    vpextrw $4, %xmm1, %esi
-; AVX1-NEXT:    xorl %ecx, %ecx
-; AVX1-NEXT:    movl %esi, %edx
-; AVX1-NEXT:    addw %ax, %dx
-; AVX1-NEXT:    setns %cl
-; AVX1-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX1-NEXT:    addw %ax, %si
-; AVX1-NEXT:    cmovol %ecx, %esi
-; AVX1-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX1-NEXT:    vpextrw $3, %xmm3, %eax
-; AVX1-NEXT:    vpextrw $3, %xmm1, %esi
-; AVX1-NEXT:    xorl %ecx, %ecx
-; AVX1-NEXT:    movl %esi, %edx
-; AVX1-NEXT:    addw %ax, %dx
-; AVX1-NEXT:    setns %cl
-; AVX1-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX1-NEXT:    addw %ax, %si
-; AVX1-NEXT:    cmovol %ecx, %esi
-; AVX1-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX1-NEXT:    vpextrw $2, %xmm3, %eax
-; AVX1-NEXT:    vpextrw $2, %xmm1, %esi
-; AVX1-NEXT:    xorl %ecx, %ecx
-; AVX1-NEXT:    movl %esi, %edx
-; AVX1-NEXT:    addw %ax, %dx
-; AVX1-NEXT:    setns %cl
-; AVX1-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX1-NEXT:    addw %ax, %si
-; AVX1-NEXT:    cmovol %ecx, %esi
-; AVX1-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX1-NEXT:    vmovd %xmm3, %eax
-; AVX1-NEXT:    vmovd %xmm1, %esi
-; AVX1-NEXT:    xorl %ecx, %ecx
-; AVX1-NEXT:    movl %esi, %edx
-; AVX1-NEXT:    addw %ax, %dx
-; AVX1-NEXT:    setns %cl
-; AVX1-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX1-NEXT:    addw %ax, %si
-; AVX1-NEXT:    cmovol %ecx, %esi
-; AVX1-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX1-NEXT:    vpextrw $1, %xmm3, %eax
-; AVX1-NEXT:    vpextrw $1, %xmm1, %esi
-; AVX1-NEXT:    xorl %ecx, %ecx
-; AVX1-NEXT:    movl %esi, %edx
-; AVX1-NEXT:    addw %ax, %dx
-; AVX1-NEXT:    setns %cl
-; AVX1-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX1-NEXT:    addw %ax, %si
-; AVX1-NEXT:    cmovol %ecx, %esi
-; AVX1-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX1-NEXT:    vextractf128 $1, %ymm3, %xmm3
-; AVX1-NEXT:    vpextrw $7, %xmm3, %eax
-; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm1
-; AVX1-NEXT:    vpextrw $7, %xmm1, %esi
-; AVX1-NEXT:    xorl %ecx, %ecx
-; AVX1-NEXT:    movl %esi, %edx
-; AVX1-NEXT:    addw %ax, %dx
-; AVX1-NEXT:    setns %cl
-; AVX1-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX1-NEXT:    addw %ax, %si
-; AVX1-NEXT:    cmovol %ecx, %esi
-; AVX1-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX1-NEXT:    vpextrw $6, %xmm3, %eax
-; AVX1-NEXT:    vpextrw $6, %xmm1, %esi
-; AVX1-NEXT:    xorl %ecx, %ecx
-; AVX1-NEXT:    movl %esi, %edx
-; AVX1-NEXT:    addw %ax, %dx
-; AVX1-NEXT:    setns %cl
-; AVX1-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX1-NEXT:    addw %ax, %si
-; AVX1-NEXT:    cmovol %ecx, %esi
-; AVX1-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX1-NEXT:    vpextrw $5, %xmm3, %eax
-; AVX1-NEXT:    vpextrw $5, %xmm1, %esi
-; AVX1-NEXT:    xorl %ecx, %ecx
-; AVX1-NEXT:    movl %esi, %edx
-; AVX1-NEXT:    addw %ax, %dx
-; AVX1-NEXT:    setns %cl
-; AVX1-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX1-NEXT:    addw %ax, %si
-; AVX1-NEXT:    cmovol %ecx, %esi
-; AVX1-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX1-NEXT:    vpextrw $4, %xmm3, %eax
-; AVX1-NEXT:    vpextrw $4, %xmm1, %esi
-; AVX1-NEXT:    xorl %ecx, %ecx
-; AVX1-NEXT:    movl %esi, %edx
-; AVX1-NEXT:    addw %ax, %dx
-; AVX1-NEXT:    setns %cl
-; AVX1-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX1-NEXT:    addw %ax, %si
-; AVX1-NEXT:    cmovol %ecx, %esi
-; AVX1-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX1-NEXT:    vpextrw $3, %xmm3, %eax
-; AVX1-NEXT:    vpextrw $3, %xmm1, %esi
-; AVX1-NEXT:    xorl %ecx, %ecx
-; AVX1-NEXT:    movl %esi, %edx
-; AVX1-NEXT:    addw %ax, %dx
-; AVX1-NEXT:    setns %cl
-; AVX1-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX1-NEXT:    addw %ax, %si
-; AVX1-NEXT:    cmovol %ecx, %esi
-; AVX1-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX1-NEXT:    vpextrw $2, %xmm3, %eax
-; AVX1-NEXT:    vpextrw $2, %xmm1, %esi
-; AVX1-NEXT:    xorl %ecx, %ecx
-; AVX1-NEXT:    movl %esi, %edx
-; AVX1-NEXT:    addw %ax, %dx
-; AVX1-NEXT:    setns %cl
-; AVX1-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX1-NEXT:    addw %ax, %si
-; AVX1-NEXT:    cmovol %ecx, %esi
-; AVX1-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX1-NEXT:    vmovd %xmm3, %eax
-; AVX1-NEXT:    vmovd %xmm1, %esi
-; AVX1-NEXT:    xorl %ecx, %ecx
-; AVX1-NEXT:    movl %esi, %edx
-; AVX1-NEXT:    addw %ax, %dx
-; AVX1-NEXT:    setns %cl
-; AVX1-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX1-NEXT:    addw %ax, %si
-; AVX1-NEXT:    cmovol %ecx, %esi
-; AVX1-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX1-NEXT:    vpextrw $1, %xmm3, %eax
-; AVX1-NEXT:    vpextrw $1, %xmm1, %esi
-; AVX1-NEXT:    xorl %ecx, %ecx
-; AVX1-NEXT:    movl %esi, %edx
-; AVX1-NEXT:    addw %ax, %dx
-; AVX1-NEXT:    setns %cl
-; AVX1-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX1-NEXT:    addw %ax, %si
-; AVX1-NEXT:    cmovol %ecx, %esi
-; AVX1-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX1-NEXT:    vpextrw $7, %xmm2, %eax
-; AVX1-NEXT:    vpextrw $7, %xmm0, %esi
-; AVX1-NEXT:    xorl %ecx, %ecx
-; AVX1-NEXT:    movl %esi, %edx
-; AVX1-NEXT:    addw %ax, %dx
-; AVX1-NEXT:    setns %cl
-; AVX1-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX1-NEXT:    addw %ax, %si
-; AVX1-NEXT:    cmovol %ecx, %esi
-; AVX1-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX1-NEXT:    vpextrw $6, %xmm2, %eax
-; AVX1-NEXT:    vpextrw $6, %xmm0, %esi
-; AVX1-NEXT:    xorl %ecx, %ecx
-; AVX1-NEXT:    movl %esi, %edx
-; AVX1-NEXT:    addw %ax, %dx
-; AVX1-NEXT:    setns %cl
-; AVX1-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX1-NEXT:    addw %ax, %si
-; AVX1-NEXT:    cmovol %ecx, %esi
-; AVX1-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX1-NEXT:    vpextrw $5, %xmm2, %eax
-; AVX1-NEXT:    vpextrw $5, %xmm0, %esi
-; AVX1-NEXT:    xorl %ecx, %ecx
-; AVX1-NEXT:    movl %esi, %edx
-; AVX1-NEXT:    addw %ax, %dx
-; AVX1-NEXT:    setns %cl
-; AVX1-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX1-NEXT:    addw %ax, %si
-; AVX1-NEXT:    cmovol %ecx, %esi
-; AVX1-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX1-NEXT:    vpextrw $4, %xmm2, %eax
-; AVX1-NEXT:    vpextrw $4, %xmm0, %esi
-; AVX1-NEXT:    xorl %ecx, %ecx
-; AVX1-NEXT:    movl %esi, %edx
-; AVX1-NEXT:    addw %ax, %dx
-; AVX1-NEXT:    setns %cl
-; AVX1-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX1-NEXT:    addw %ax, %si
-; AVX1-NEXT:    cmovol %ecx, %esi
-; AVX1-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX1-NEXT:    vpextrw $3, %xmm2, %eax
-; AVX1-NEXT:    vpextrw $3, %xmm0, %ebx
-; AVX1-NEXT:    xorl %ecx, %ecx
-; AVX1-NEXT:    movl %ebx, %edx
-; AVX1-NEXT:    addw %ax, %dx
-; AVX1-NEXT:    setns %cl
-; AVX1-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX1-NEXT:    addw %ax, %bx
-; AVX1-NEXT:    cmovol %ecx, %ebx
-; AVX1-NEXT:    vpextrw $2, %xmm2, %eax
-; AVX1-NEXT:    vpextrw $2, %xmm0, %r11d
-; AVX1-NEXT:    xorl %ecx, %ecx
-; AVX1-NEXT:    movl %r11d, %edx
-; AVX1-NEXT:    addw %ax, %dx
-; AVX1-NEXT:    setns %cl
-; AVX1-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX1-NEXT:    addw %ax, %r11w
-; AVX1-NEXT:    cmovol %ecx, %r11d
-; AVX1-NEXT:    vmovd %xmm2, %eax
-; AVX1-NEXT:    vmovd %xmm0, %r10d
-; AVX1-NEXT:    xorl %ecx, %ecx
-; AVX1-NEXT:    movl %r10d, %edx
-; AVX1-NEXT:    addw %ax, %dx
-; AVX1-NEXT:    setns %cl
-; AVX1-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX1-NEXT:    addw %ax, %r10w
-; AVX1-NEXT:    cmovol %ecx, %r10d
-; AVX1-NEXT:    vpextrw $1, %xmm2, %eax
-; AVX1-NEXT:    vpextrw $1, %xmm0, %r9d
-; AVX1-NEXT:    xorl %ecx, %ecx
-; AVX1-NEXT:    movl %r9d, %edx
-; AVX1-NEXT:    addw %ax, %dx
-; AVX1-NEXT:    setns %cl
-; AVX1-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX1-NEXT:    addw %ax, %r9w
-; AVX1-NEXT:    cmovol %ecx, %r9d
-; AVX1-NEXT:    vextractf128 $1, %ymm2, %xmm1
-; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm0
-; AVX1-NEXT:    vpextrw $7, %xmm1, %eax
-; AVX1-NEXT:    vpextrw $7, %xmm0, %r8d
-; AVX1-NEXT:    xorl %ecx, %ecx
-; AVX1-NEXT:    movl %r8d, %edx
-; AVX1-NEXT:    addw %ax, %dx
-; AVX1-NEXT:    setns %cl
-; AVX1-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX1-NEXT:    addw %ax, %r8w
-; AVX1-NEXT:    cmovol %ecx, %r8d
-; AVX1-NEXT:    vpextrw $6, %xmm1, %eax
-; AVX1-NEXT:    vpextrw $6, %xmm0, %edi
-; AVX1-NEXT:    xorl %ecx, %ecx
-; AVX1-NEXT:    movl %edi, %edx
-; AVX1-NEXT:    addw %ax, %dx
-; AVX1-NEXT:    setns %cl
-; AVX1-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX1-NEXT:    addw %ax, %di
-; AVX1-NEXT:    cmovol %ecx, %edi
-; AVX1-NEXT:    vpextrw $5, %xmm1, %eax
-; AVX1-NEXT:    vpextrw $5, %xmm0, %esi
-; AVX1-NEXT:    xorl %ecx, %ecx
-; AVX1-NEXT:    movl %esi, %edx
-; AVX1-NEXT:    addw %ax, %dx
-; AVX1-NEXT:    setns %cl
-; AVX1-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX1-NEXT:    addw %ax, %si
-; AVX1-NEXT:    cmovol %ecx, %esi
-; AVX1-NEXT:    vpextrw $4, %xmm1, %eax
-; AVX1-NEXT:    vpextrw $4, %xmm0, %edx
-; AVX1-NEXT:    xorl %ecx, %ecx
-; AVX1-NEXT:    movl %edx, %r13d
-; AVX1-NEXT:    addw %ax, %r13w
-; AVX1-NEXT:    setns %cl
-; AVX1-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX1-NEXT:    addw %ax, %dx
-; AVX1-NEXT:    cmovol %ecx, %edx
-; AVX1-NEXT:    vpextrw $3, %xmm1, %eax
-; AVX1-NEXT:    vpextrw $3, %xmm0, %r13d
-; AVX1-NEXT:    xorl %ecx, %ecx
-; AVX1-NEXT:    movl %r13d, %r12d
-; AVX1-NEXT:    addw %ax, %r12w
-; AVX1-NEXT:    setns %cl
-; AVX1-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX1-NEXT:    addw %ax, %r13w
-; AVX1-NEXT:    cmovol %ecx, %r13d
-; AVX1-NEXT:    vpextrw $2, %xmm1, %r12d
-; AVX1-NEXT:    vpextrw $2, %xmm0, %eax
-; AVX1-NEXT:    xorl %ecx, %ecx
-; AVX1-NEXT:    movl %eax, %r15d
-; AVX1-NEXT:    addw %r12w, %r15w
-; AVX1-NEXT:    setns %cl
-; AVX1-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX1-NEXT:    addw %r12w, %ax
-; AVX1-NEXT:    cmovol %ecx, %eax
-; AVX1-NEXT:    vmovd %xmm1, %r15d
-; AVX1-NEXT:    vmovd %xmm0, %r12d
-; AVX1-NEXT:    xorl %ecx, %ecx
-; AVX1-NEXT:    movl %r12d, %r14d
-; AVX1-NEXT:    addw %r15w, %r14w
-; AVX1-NEXT:    setns %cl
-; AVX1-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX1-NEXT:    addw %r15w, %r12w
-; AVX1-NEXT:    cmovol %ecx, %r12d
-; AVX1-NEXT:    vpextrw $1, %xmm1, %r14d
-; AVX1-NEXT:    vpextrw $1, %xmm0, %ecx
-; AVX1-NEXT:    xorl %r15d, %r15d
-; AVX1-NEXT:    movl %ecx, %ebp
-; AVX1-NEXT:    addw %r14w, %bp
-; AVX1-NEXT:    setns %r15b
-; AVX1-NEXT:    addl $32767, %r15d # imm = 0x7FFF
-; AVX1-NEXT:    addw %r14w, %cx
-; AVX1-NEXT:    cmovol %r15d, %ecx
-; AVX1-NEXT:    vmovd %r12d, %xmm0
-; AVX1-NEXT:    vpinsrw $1, %ecx, %xmm0, %xmm0
-; AVX1-NEXT:    vpinsrw $2, %eax, %xmm0, %xmm0
-; AVX1-NEXT:    vpinsrw $3, %r13d, %xmm0, %xmm0
-; AVX1-NEXT:    vpinsrw $4, %edx, %xmm0, %xmm0
-; AVX1-NEXT:    vpinsrw $5, %esi, %xmm0, %xmm0
-; AVX1-NEXT:    vpinsrw $6, %edi, %xmm0, %xmm0
-; AVX1-NEXT:    vpinsrw $7, %r8d, %xmm0, %xmm0
-; AVX1-NEXT:    vmovd %r10d, %xmm1
-; AVX1-NEXT:    vpinsrw $1, %r9d, %xmm1, %xmm1
-; AVX1-NEXT:    vpinsrw $2, %r11d, %xmm1, %xmm1
-; AVX1-NEXT:    vpinsrw $3, %ebx, %xmm1, %xmm1
-; AVX1-NEXT:    vpinsrw $4, {{[-0-9]+}}(%r{{[sb]}}p), %xmm1, %xmm1 # 4-byte Folded Reload
-; AVX1-NEXT:    vpinsrw $5, {{[-0-9]+}}(%r{{[sb]}}p), %xmm1, %xmm1 # 4-byte Folded Reload
-; AVX1-NEXT:    vpinsrw $6, {{[-0-9]+}}(%r{{[sb]}}p), %xmm1, %xmm1 # 4-byte Folded Reload
-; AVX1-NEXT:    vpinsrw $7, {{[-0-9]+}}(%r{{[sb]}}p), %xmm1, %xmm1 # 4-byte Folded Reload
-; AVX1-NEXT:    vmovd {{[-0-9]+}}(%r{{[sb]}}p), %xmm2 # 4-byte Folded Reload
-; AVX1-NEXT:    # xmm2 = mem[0],zero,zero,zero
-; AVX1-NEXT:    vpinsrw $1, {{[-0-9]+}}(%r{{[sb]}}p), %xmm2, %xmm2 # 4-byte Folded Reload
-; AVX1-NEXT:    vpinsrw $2, {{[-0-9]+}}(%r{{[sb]}}p), %xmm2, %xmm2 # 4-byte Folded Reload
-; AVX1-NEXT:    vpinsrw $3, {{[-0-9]+}}(%r{{[sb]}}p), %xmm2, %xmm2 # 4-byte Folded Reload
-; AVX1-NEXT:    vpinsrw $4, {{[-0-9]+}}(%r{{[sb]}}p), %xmm2, %xmm2 # 4-byte Folded Reload
-; AVX1-NEXT:    vpinsrw $5, {{[-0-9]+}}(%r{{[sb]}}p), %xmm2, %xmm2 # 4-byte Folded Reload
-; AVX1-NEXT:    vpinsrw $6, {{[-0-9]+}}(%r{{[sb]}}p), %xmm2, %xmm2 # 4-byte Folded Reload
-; AVX1-NEXT:    vpinsrw $7, {{[-0-9]+}}(%r{{[sb]}}p), %xmm2, %xmm2 # 4-byte Folded Reload
-; AVX1-NEXT:    vmovd {{[-0-9]+}}(%r{{[sb]}}p), %xmm3 # 4-byte Folded Reload
-; AVX1-NEXT:    # xmm3 = mem[0],zero,zero,zero
-; AVX1-NEXT:    vpinsrw $1, {{[-0-9]+}}(%r{{[sb]}}p), %xmm3, %xmm3 # 4-byte Folded Reload
-; AVX1-NEXT:    vpinsrw $2, {{[-0-9]+}}(%r{{[sb]}}p), %xmm3, %xmm3 # 4-byte Folded Reload
-; AVX1-NEXT:    vpinsrw $3, {{[-0-9]+}}(%r{{[sb]}}p), %xmm3, %xmm3 # 4-byte Folded Reload
-; AVX1-NEXT:    vpinsrw $4, {{[-0-9]+}}(%r{{[sb]}}p), %xmm3, %xmm3 # 4-byte Folded Reload
-; AVX1-NEXT:    vpinsrw $5, {{[-0-9]+}}(%r{{[sb]}}p), %xmm3, %xmm3 # 4-byte Folded Reload
-; AVX1-NEXT:    vpinsrw $6, {{[-0-9]+}}(%r{{[sb]}}p), %xmm3, %xmm3 # 4-byte Folded Reload
-; AVX1-NEXT:    vpinsrw $7, {{[-0-9]+}}(%r{{[sb]}}p), %xmm3, %xmm3 # 4-byte Folded Reload
-; AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm1, %ymm0
-; AVX1-NEXT:    vinsertf128 $1, %xmm2, %ymm3, %ymm1
-; AVX1-NEXT:    popq %rbx
-; AVX1-NEXT:    popq %r12
-; AVX1-NEXT:    popq %r13
-; AVX1-NEXT:    popq %r14
-; AVX1-NEXT:    popq %r15
-; AVX1-NEXT:    popq %rbp
+; AVX1-NEXT:    vextractf128 $1, %ymm2, %xmm4
+; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm5
+; AVX1-NEXT:    vpaddsw %xmm4, %xmm5, %xmm4
+; AVX1-NEXT:    vpaddsw %xmm2, %xmm0, %xmm0
+; AVX1-NEXT:    vinsertf128 $1, %xmm4, %ymm0, %ymm0
+; AVX1-NEXT:    vextractf128 $1, %ymm3, %xmm2
+; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm4
+; AVX1-NEXT:    vpaddsw %xmm2, %xmm4, %xmm2
+; AVX1-NEXT:    vpaddsw %xmm3, %xmm1, %xmm1
+; AVX1-NEXT:    vinsertf128 $1, %xmm2, %ymm1, %ymm1
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: v32i16:
 ; AVX2:       # %bb.0:
-; AVX2-NEXT:    pushq %rbp
-; AVX2-NEXT:    pushq %r15
-; AVX2-NEXT:    pushq %r14
-; AVX2-NEXT:    pushq %r13
-; AVX2-NEXT:    pushq %r12
-; AVX2-NEXT:    pushq %rbx
-; AVX2-NEXT:    vpextrw $7, %xmm3, %eax
-; AVX2-NEXT:    vpextrw $7, %xmm1, %esi
-; AVX2-NEXT:    xorl %ecx, %ecx
-; AVX2-NEXT:    movl %esi, %edx
-; AVX2-NEXT:    addw %ax, %dx
-; AVX2-NEXT:    setns %cl
-; AVX2-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX2-NEXT:    addw %ax, %si
-; AVX2-NEXT:    cmovol %ecx, %esi
-; AVX2-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX2-NEXT:    vpextrw $6, %xmm3, %eax
-; AVX2-NEXT:    vpextrw $6, %xmm1, %esi
-; AVX2-NEXT:    xorl %ecx, %ecx
-; AVX2-NEXT:    movl %esi, %edx
-; AVX2-NEXT:    addw %ax, %dx
-; AVX2-NEXT:    setns %cl
-; AVX2-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX2-NEXT:    addw %ax, %si
-; AVX2-NEXT:    cmovol %ecx, %esi
-; AVX2-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX2-NEXT:    vpextrw $5, %xmm3, %eax
-; AVX2-NEXT:    vpextrw $5, %xmm1, %esi
-; AVX2-NEXT:    xorl %ecx, %ecx
-; AVX2-NEXT:    movl %esi, %edx
-; AVX2-NEXT:    addw %ax, %dx
-; AVX2-NEXT:    setns %cl
-; AVX2-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX2-NEXT:    addw %ax, %si
-; AVX2-NEXT:    cmovol %ecx, %esi
-; AVX2-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX2-NEXT:    vpextrw $4, %xmm3, %eax
-; AVX2-NEXT:    vpextrw $4, %xmm1, %esi
-; AVX2-NEXT:    xorl %ecx, %ecx
-; AVX2-NEXT:    movl %esi, %edx
-; AVX2-NEXT:    addw %ax, %dx
-; AVX2-NEXT:    setns %cl
-; AVX2-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX2-NEXT:    addw %ax, %si
-; AVX2-NEXT:    cmovol %ecx, %esi
-; AVX2-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX2-NEXT:    vpextrw $3, %xmm3, %eax
-; AVX2-NEXT:    vpextrw $3, %xmm1, %esi
-; AVX2-NEXT:    xorl %ecx, %ecx
-; AVX2-NEXT:    movl %esi, %edx
-; AVX2-NEXT:    addw %ax, %dx
-; AVX2-NEXT:    setns %cl
-; AVX2-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX2-NEXT:    addw %ax, %si
-; AVX2-NEXT:    cmovol %ecx, %esi
-; AVX2-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX2-NEXT:    vpextrw $2, %xmm3, %eax
-; AVX2-NEXT:    vpextrw $2, %xmm1, %esi
-; AVX2-NEXT:    xorl %ecx, %ecx
-; AVX2-NEXT:    movl %esi, %edx
-; AVX2-NEXT:    addw %ax, %dx
-; AVX2-NEXT:    setns %cl
-; AVX2-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX2-NEXT:    addw %ax, %si
-; AVX2-NEXT:    cmovol %ecx, %esi
-; AVX2-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX2-NEXT:    vmovd %xmm3, %eax
-; AVX2-NEXT:    vmovd %xmm1, %esi
-; AVX2-NEXT:    xorl %ecx, %ecx
-; AVX2-NEXT:    movl %esi, %edx
-; AVX2-NEXT:    addw %ax, %dx
-; AVX2-NEXT:    setns %cl
-; AVX2-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX2-NEXT:    addw %ax, %si
-; AVX2-NEXT:    cmovol %ecx, %esi
-; AVX2-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX2-NEXT:    vpextrw $1, %xmm3, %eax
-; AVX2-NEXT:    vpextrw $1, %xmm1, %esi
-; AVX2-NEXT:    xorl %ecx, %ecx
-; AVX2-NEXT:    movl %esi, %edx
-; AVX2-NEXT:    addw %ax, %dx
-; AVX2-NEXT:    setns %cl
-; AVX2-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX2-NEXT:    addw %ax, %si
-; AVX2-NEXT:    cmovol %ecx, %esi
-; AVX2-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX2-NEXT:    vextracti128 $1, %ymm3, %xmm3
-; AVX2-NEXT:    vpextrw $7, %xmm3, %eax
-; AVX2-NEXT:    vextracti128 $1, %ymm1, %xmm1
-; AVX2-NEXT:    vpextrw $7, %xmm1, %esi
-; AVX2-NEXT:    xorl %ecx, %ecx
-; AVX2-NEXT:    movl %esi, %edx
-; AVX2-NEXT:    addw %ax, %dx
-; AVX2-NEXT:    setns %cl
-; AVX2-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX2-NEXT:    addw %ax, %si
-; AVX2-NEXT:    cmovol %ecx, %esi
-; AVX2-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX2-NEXT:    vpextrw $6, %xmm3, %eax
-; AVX2-NEXT:    vpextrw $6, %xmm1, %esi
-; AVX2-NEXT:    xorl %ecx, %ecx
-; AVX2-NEXT:    movl %esi, %edx
-; AVX2-NEXT:    addw %ax, %dx
-; AVX2-NEXT:    setns %cl
-; AVX2-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX2-NEXT:    addw %ax, %si
-; AVX2-NEXT:    cmovol %ecx, %esi
-; AVX2-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX2-NEXT:    vpextrw $5, %xmm3, %eax
-; AVX2-NEXT:    vpextrw $5, %xmm1, %esi
-; AVX2-NEXT:    xorl %ecx, %ecx
-; AVX2-NEXT:    movl %esi, %edx
-; AVX2-NEXT:    addw %ax, %dx
-; AVX2-NEXT:    setns %cl
-; AVX2-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX2-NEXT:    addw %ax, %si
-; AVX2-NEXT:    cmovol %ecx, %esi
-; AVX2-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX2-NEXT:    vpextrw $4, %xmm3, %eax
-; AVX2-NEXT:    vpextrw $4, %xmm1, %esi
-; AVX2-NEXT:    xorl %ecx, %ecx
-; AVX2-NEXT:    movl %esi, %edx
-; AVX2-NEXT:    addw %ax, %dx
-; AVX2-NEXT:    setns %cl
-; AVX2-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX2-NEXT:    addw %ax, %si
-; AVX2-NEXT:    cmovol %ecx, %esi
-; AVX2-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX2-NEXT:    vpextrw $3, %xmm3, %eax
-; AVX2-NEXT:    vpextrw $3, %xmm1, %esi
-; AVX2-NEXT:    xorl %ecx, %ecx
-; AVX2-NEXT:    movl %esi, %edx
-; AVX2-NEXT:    addw %ax, %dx
-; AVX2-NEXT:    setns %cl
-; AVX2-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX2-NEXT:    addw %ax, %si
-; AVX2-NEXT:    cmovol %ecx, %esi
-; AVX2-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX2-NEXT:    vpextrw $2, %xmm3, %eax
-; AVX2-NEXT:    vpextrw $2, %xmm1, %esi
-; AVX2-NEXT:    xorl %ecx, %ecx
-; AVX2-NEXT:    movl %esi, %edx
-; AVX2-NEXT:    addw %ax, %dx
-; AVX2-NEXT:    setns %cl
-; AVX2-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX2-NEXT:    addw %ax, %si
-; AVX2-NEXT:    cmovol %ecx, %esi
-; AVX2-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX2-NEXT:    vmovd %xmm3, %eax
-; AVX2-NEXT:    vmovd %xmm1, %esi
-; AVX2-NEXT:    xorl %ecx, %ecx
-; AVX2-NEXT:    movl %esi, %edx
-; AVX2-NEXT:    addw %ax, %dx
-; AVX2-NEXT:    setns %cl
-; AVX2-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX2-NEXT:    addw %ax, %si
-; AVX2-NEXT:    cmovol %ecx, %esi
-; AVX2-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX2-NEXT:    vpextrw $1, %xmm3, %eax
-; AVX2-NEXT:    vpextrw $1, %xmm1, %esi
-; AVX2-NEXT:    xorl %ecx, %ecx
-; AVX2-NEXT:    movl %esi, %edx
-; AVX2-NEXT:    addw %ax, %dx
-; AVX2-NEXT:    setns %cl
-; AVX2-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX2-NEXT:    addw %ax, %si
-; AVX2-NEXT:    cmovol %ecx, %esi
-; AVX2-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX2-NEXT:    vpextrw $7, %xmm2, %eax
-; AVX2-NEXT:    vpextrw $7, %xmm0, %esi
-; AVX2-NEXT:    xorl %ecx, %ecx
-; AVX2-NEXT:    movl %esi, %edx
-; AVX2-NEXT:    addw %ax, %dx
-; AVX2-NEXT:    setns %cl
-; AVX2-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX2-NEXT:    addw %ax, %si
-; AVX2-NEXT:    cmovol %ecx, %esi
-; AVX2-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX2-NEXT:    vpextrw $6, %xmm2, %eax
-; AVX2-NEXT:    vpextrw $6, %xmm0, %esi
-; AVX2-NEXT:    xorl %ecx, %ecx
-; AVX2-NEXT:    movl %esi, %edx
-; AVX2-NEXT:    addw %ax, %dx
-; AVX2-NEXT:    setns %cl
-; AVX2-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX2-NEXT:    addw %ax, %si
-; AVX2-NEXT:    cmovol %ecx, %esi
-; AVX2-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX2-NEXT:    vpextrw $5, %xmm2, %eax
-; AVX2-NEXT:    vpextrw $5, %xmm0, %esi
-; AVX2-NEXT:    xorl %ecx, %ecx
-; AVX2-NEXT:    movl %esi, %edx
-; AVX2-NEXT:    addw %ax, %dx
-; AVX2-NEXT:    setns %cl
-; AVX2-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX2-NEXT:    addw %ax, %si
-; AVX2-NEXT:    cmovol %ecx, %esi
-; AVX2-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX2-NEXT:    vpextrw $4, %xmm2, %eax
-; AVX2-NEXT:    vpextrw $4, %xmm0, %esi
-; AVX2-NEXT:    xorl %ecx, %ecx
-; AVX2-NEXT:    movl %esi, %edx
-; AVX2-NEXT:    addw %ax, %dx
-; AVX2-NEXT:    setns %cl
-; AVX2-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX2-NEXT:    addw %ax, %si
-; AVX2-NEXT:    cmovol %ecx, %esi
-; AVX2-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX2-NEXT:    vpextrw $3, %xmm2, %eax
-; AVX2-NEXT:    vpextrw $3, %xmm0, %ebx
-; AVX2-NEXT:    xorl %ecx, %ecx
-; AVX2-NEXT:    movl %ebx, %edx
-; AVX2-NEXT:    addw %ax, %dx
-; AVX2-NEXT:    setns %cl
-; AVX2-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX2-NEXT:    addw %ax, %bx
-; AVX2-NEXT:    cmovol %ecx, %ebx
-; AVX2-NEXT:    vpextrw $2, %xmm2, %eax
-; AVX2-NEXT:    vpextrw $2, %xmm0, %r11d
-; AVX2-NEXT:    xorl %ecx, %ecx
-; AVX2-NEXT:    movl %r11d, %edx
-; AVX2-NEXT:    addw %ax, %dx
-; AVX2-NEXT:    setns %cl
-; AVX2-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX2-NEXT:    addw %ax, %r11w
-; AVX2-NEXT:    cmovol %ecx, %r11d
-; AVX2-NEXT:    vmovd %xmm2, %eax
-; AVX2-NEXT:    vmovd %xmm0, %r10d
-; AVX2-NEXT:    xorl %ecx, %ecx
-; AVX2-NEXT:    movl %r10d, %edx
-; AVX2-NEXT:    addw %ax, %dx
-; AVX2-NEXT:    setns %cl
-; AVX2-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX2-NEXT:    addw %ax, %r10w
-; AVX2-NEXT:    cmovol %ecx, %r10d
-; AVX2-NEXT:    vpextrw $1, %xmm2, %eax
-; AVX2-NEXT:    vpextrw $1, %xmm0, %r9d
-; AVX2-NEXT:    xorl %ecx, %ecx
-; AVX2-NEXT:    movl %r9d, %edx
-; AVX2-NEXT:    addw %ax, %dx
-; AVX2-NEXT:    setns %cl
-; AVX2-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX2-NEXT:    addw %ax, %r9w
-; AVX2-NEXT:    cmovol %ecx, %r9d
-; AVX2-NEXT:    vextracti128 $1, %ymm2, %xmm1
-; AVX2-NEXT:    vextracti128 $1, %ymm0, %xmm0
-; AVX2-NEXT:    vpextrw $7, %xmm1, %eax
-; AVX2-NEXT:    vpextrw $7, %xmm0, %r8d
-; AVX2-NEXT:    xorl %ecx, %ecx
-; AVX2-NEXT:    movl %r8d, %edx
-; AVX2-NEXT:    addw %ax, %dx
-; AVX2-NEXT:    setns %cl
-; AVX2-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX2-NEXT:    addw %ax, %r8w
-; AVX2-NEXT:    cmovol %ecx, %r8d
-; AVX2-NEXT:    vpextrw $6, %xmm1, %eax
-; AVX2-NEXT:    vpextrw $6, %xmm0, %edi
-; AVX2-NEXT:    xorl %ecx, %ecx
-; AVX2-NEXT:    movl %edi, %edx
-; AVX2-NEXT:    addw %ax, %dx
-; AVX2-NEXT:    setns %cl
-; AVX2-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX2-NEXT:    addw %ax, %di
-; AVX2-NEXT:    cmovol %ecx, %edi
-; AVX2-NEXT:    vpextrw $5, %xmm1, %eax
-; AVX2-NEXT:    vpextrw $5, %xmm0, %esi
-; AVX2-NEXT:    xorl %ecx, %ecx
-; AVX2-NEXT:    movl %esi, %edx
-; AVX2-NEXT:    addw %ax, %dx
-; AVX2-NEXT:    setns %cl
-; AVX2-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX2-NEXT:    addw %ax, %si
-; AVX2-NEXT:    cmovol %ecx, %esi
-; AVX2-NEXT:    vpextrw $4, %xmm1, %eax
-; AVX2-NEXT:    vpextrw $4, %xmm0, %edx
-; AVX2-NEXT:    xorl %ecx, %ecx
-; AVX2-NEXT:    movl %edx, %r13d
-; AVX2-NEXT:    addw %ax, %r13w
-; AVX2-NEXT:    setns %cl
-; AVX2-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX2-NEXT:    addw %ax, %dx
-; AVX2-NEXT:    cmovol %ecx, %edx
-; AVX2-NEXT:    vpextrw $3, %xmm1, %eax
-; AVX2-NEXT:    vpextrw $3, %xmm0, %r13d
-; AVX2-NEXT:    xorl %ecx, %ecx
-; AVX2-NEXT:    movl %r13d, %r12d
-; AVX2-NEXT:    addw %ax, %r12w
-; AVX2-NEXT:    setns %cl
-; AVX2-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX2-NEXT:    addw %ax, %r13w
-; AVX2-NEXT:    cmovol %ecx, %r13d
-; AVX2-NEXT:    vpextrw $2, %xmm1, %r12d
-; AVX2-NEXT:    vpextrw $2, %xmm0, %eax
-; AVX2-NEXT:    xorl %ecx, %ecx
-; AVX2-NEXT:    movl %eax, %r15d
-; AVX2-NEXT:    addw %r12w, %r15w
-; AVX2-NEXT:    setns %cl
-; AVX2-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX2-NEXT:    addw %r12w, %ax
-; AVX2-NEXT:    cmovol %ecx, %eax
-; AVX2-NEXT:    vmovd %xmm1, %r15d
-; AVX2-NEXT:    vmovd %xmm0, %r12d
-; AVX2-NEXT:    xorl %ecx, %ecx
-; AVX2-NEXT:    movl %r12d, %r14d
-; AVX2-NEXT:    addw %r15w, %r14w
-; AVX2-NEXT:    setns %cl
-; AVX2-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX2-NEXT:    addw %r15w, %r12w
-; AVX2-NEXT:    cmovol %ecx, %r12d
-; AVX2-NEXT:    vpextrw $1, %xmm1, %r14d
-; AVX2-NEXT:    vpextrw $1, %xmm0, %ecx
-; AVX2-NEXT:    xorl %r15d, %r15d
-; AVX2-NEXT:    movl %ecx, %ebp
-; AVX2-NEXT:    addw %r14w, %bp
-; AVX2-NEXT:    setns %r15b
-; AVX2-NEXT:    addl $32767, %r15d # imm = 0x7FFF
-; AVX2-NEXT:    addw %r14w, %cx
-; AVX2-NEXT:    cmovol %r15d, %ecx
-; AVX2-NEXT:    vmovd %r12d, %xmm0
-; AVX2-NEXT:    vpinsrw $1, %ecx, %xmm0, %xmm0
-; AVX2-NEXT:    vpinsrw $2, %eax, %xmm0, %xmm0
-; AVX2-NEXT:    vpinsrw $3, %r13d, %xmm0, %xmm0
-; AVX2-NEXT:    vpinsrw $4, %edx, %xmm0, %xmm0
-; AVX2-NEXT:    vpinsrw $5, %esi, %xmm0, %xmm0
-; AVX2-NEXT:    vpinsrw $6, %edi, %xmm0, %xmm0
-; AVX2-NEXT:    vpinsrw $7, %r8d, %xmm0, %xmm0
-; AVX2-NEXT:    vmovd %r10d, %xmm1
-; AVX2-NEXT:    vpinsrw $1, %r9d, %xmm1, %xmm1
-; AVX2-NEXT:    vpinsrw $2, %r11d, %xmm1, %xmm1
-; AVX2-NEXT:    vpinsrw $3, %ebx, %xmm1, %xmm1
-; AVX2-NEXT:    vpinsrw $4, {{[-0-9]+}}(%r{{[sb]}}p), %xmm1, %xmm1 # 4-byte Folded Reload
-; AVX2-NEXT:    vpinsrw $5, {{[-0-9]+}}(%r{{[sb]}}p), %xmm1, %xmm1 # 4-byte Folded Reload
-; AVX2-NEXT:    vpinsrw $6, {{[-0-9]+}}(%r{{[sb]}}p), %xmm1, %xmm1 # 4-byte Folded Reload
-; AVX2-NEXT:    vpinsrw $7, {{[-0-9]+}}(%r{{[sb]}}p), %xmm1, %xmm1 # 4-byte Folded Reload
-; AVX2-NEXT:    vmovd {{[-0-9]+}}(%r{{[sb]}}p), %xmm2 # 4-byte Folded Reload
-; AVX2-NEXT:    # xmm2 = mem[0],zero,zero,zero
-; AVX2-NEXT:    vpinsrw $1, {{[-0-9]+}}(%r{{[sb]}}p), %xmm2, %xmm2 # 4-byte Folded Reload
-; AVX2-NEXT:    vpinsrw $2, {{[-0-9]+}}(%r{{[sb]}}p), %xmm2, %xmm2 # 4-byte Folded Reload
-; AVX2-NEXT:    vpinsrw $3, {{[-0-9]+}}(%r{{[sb]}}p), %xmm2, %xmm2 # 4-byte Folded Reload
-; AVX2-NEXT:    vpinsrw $4, {{[-0-9]+}}(%r{{[sb]}}p), %xmm2, %xmm2 # 4-byte Folded Reload
-; AVX2-NEXT:    vpinsrw $5, {{[-0-9]+}}(%r{{[sb]}}p), %xmm2, %xmm2 # 4-byte Folded Reload
-; AVX2-NEXT:    vpinsrw $6, {{[-0-9]+}}(%r{{[sb]}}p), %xmm2, %xmm2 # 4-byte Folded Reload
-; AVX2-NEXT:    vpinsrw $7, {{[-0-9]+}}(%r{{[sb]}}p), %xmm2, %xmm2 # 4-byte Folded Reload
-; AVX2-NEXT:    vmovd {{[-0-9]+}}(%r{{[sb]}}p), %xmm3 # 4-byte Folded Reload
-; AVX2-NEXT:    # xmm3 = mem[0],zero,zero,zero
-; AVX2-NEXT:    vpinsrw $1, {{[-0-9]+}}(%r{{[sb]}}p), %xmm3, %xmm3 # 4-byte Folded Reload
-; AVX2-NEXT:    vpinsrw $2, {{[-0-9]+}}(%r{{[sb]}}p), %xmm3, %xmm3 # 4-byte Folded Reload
-; AVX2-NEXT:    vpinsrw $3, {{[-0-9]+}}(%r{{[sb]}}p), %xmm3, %xmm3 # 4-byte Folded Reload
-; AVX2-NEXT:    vpinsrw $4, {{[-0-9]+}}(%r{{[sb]}}p), %xmm3, %xmm3 # 4-byte Folded Reload
-; AVX2-NEXT:    vpinsrw $5, {{[-0-9]+}}(%r{{[sb]}}p), %xmm3, %xmm3 # 4-byte Folded Reload
-; AVX2-NEXT:    vpinsrw $6, {{[-0-9]+}}(%r{{[sb]}}p), %xmm3, %xmm3 # 4-byte Folded Reload
-; AVX2-NEXT:    vpinsrw $7, {{[-0-9]+}}(%r{{[sb]}}p), %xmm3, %xmm3 # 4-byte Folded Reload
-; AVX2-NEXT:    vinserti128 $1, %xmm0, %ymm1, %ymm0
-; AVX2-NEXT:    vinserti128 $1, %xmm2, %ymm3, %ymm1
-; AVX2-NEXT:    popq %rbx
-; AVX2-NEXT:    popq %r12
-; AVX2-NEXT:    popq %r13
-; AVX2-NEXT:    popq %r14
-; AVX2-NEXT:    popq %r15
-; AVX2-NEXT:    popq %rbp
+; AVX2-NEXT:    vpaddsw %ymm2, %ymm0, %ymm0
+; AVX2-NEXT:    vpaddsw %ymm3, %ymm1, %ymm1
 ; AVX2-NEXT:    retq
 ;
 ; AVX512-LABEL: v32i16:
 ; AVX512:       # %bb.0:
-; AVX512-NEXT:    pushq %rbp
-; AVX512-NEXT:    pushq %r15
-; AVX512-NEXT:    pushq %r14
-; AVX512-NEXT:    pushq %r13
-; AVX512-NEXT:    pushq %r12
-; AVX512-NEXT:    pushq %rbx
-; AVX512-NEXT:    vpextrw $7, %xmm1, %eax
-; AVX512-NEXT:    vpextrw $7, %xmm0, %esi
-; AVX512-NEXT:    xorl %ecx, %ecx
-; AVX512-NEXT:    movl %esi, %edx
-; AVX512-NEXT:    addw %ax, %dx
-; AVX512-NEXT:    setns %cl
-; AVX512-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX512-NEXT:    addw %ax, %si
-; AVX512-NEXT:    cmovol %ecx, %esi
-; AVX512-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX512-NEXT:    vpextrw $6, %xmm1, %eax
-; AVX512-NEXT:    vpextrw $6, %xmm0, %esi
-; AVX512-NEXT:    xorl %ecx, %ecx
-; AVX512-NEXT:    movl %esi, %edx
-; AVX512-NEXT:    addw %ax, %dx
-; AVX512-NEXT:    setns %cl
-; AVX512-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX512-NEXT:    addw %ax, %si
-; AVX512-NEXT:    cmovol %ecx, %esi
-; AVX512-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX512-NEXT:    vpextrw $5, %xmm1, %eax
-; AVX512-NEXT:    vpextrw $5, %xmm0, %esi
-; AVX512-NEXT:    xorl %ecx, %ecx
-; AVX512-NEXT:    movl %esi, %edx
-; AVX512-NEXT:    addw %ax, %dx
-; AVX512-NEXT:    setns %cl
-; AVX512-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX512-NEXT:    addw %ax, %si
-; AVX512-NEXT:    cmovol %ecx, %esi
-; AVX512-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX512-NEXT:    vpextrw $4, %xmm1, %eax
-; AVX512-NEXT:    vpextrw $4, %xmm0, %esi
-; AVX512-NEXT:    xorl %ecx, %ecx
-; AVX512-NEXT:    movl %esi, %edx
-; AVX512-NEXT:    addw %ax, %dx
-; AVX512-NEXT:    setns %cl
-; AVX512-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX512-NEXT:    addw %ax, %si
-; AVX512-NEXT:    cmovol %ecx, %esi
-; AVX512-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX512-NEXT:    vpextrw $3, %xmm1, %eax
-; AVX512-NEXT:    vpextrw $3, %xmm0, %esi
-; AVX512-NEXT:    xorl %ecx, %ecx
-; AVX512-NEXT:    movl %esi, %edx
-; AVX512-NEXT:    addw %ax, %dx
-; AVX512-NEXT:    setns %cl
-; AVX512-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX512-NEXT:    addw %ax, %si
-; AVX512-NEXT:    cmovol %ecx, %esi
-; AVX512-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX512-NEXT:    vpextrw $2, %xmm1, %eax
-; AVX512-NEXT:    vpextrw $2, %xmm0, %esi
-; AVX512-NEXT:    xorl %ecx, %ecx
-; AVX512-NEXT:    movl %esi, %edx
-; AVX512-NEXT:    addw %ax, %dx
-; AVX512-NEXT:    setns %cl
-; AVX512-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX512-NEXT:    addw %ax, %si
-; AVX512-NEXT:    cmovol %ecx, %esi
-; AVX512-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX512-NEXT:    vmovd %xmm1, %eax
-; AVX512-NEXT:    vmovd %xmm0, %esi
-; AVX512-NEXT:    xorl %ecx, %ecx
-; AVX512-NEXT:    movl %esi, %edx
-; AVX512-NEXT:    addw %ax, %dx
-; AVX512-NEXT:    setns %cl
-; AVX512-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX512-NEXT:    addw %ax, %si
-; AVX512-NEXT:    cmovol %ecx, %esi
-; AVX512-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX512-NEXT:    vpextrw $1, %xmm1, %eax
-; AVX512-NEXT:    vpextrw $1, %xmm0, %esi
-; AVX512-NEXT:    xorl %ecx, %ecx
-; AVX512-NEXT:    movl %esi, %edx
-; AVX512-NEXT:    addw %ax, %dx
-; AVX512-NEXT:    setns %cl
-; AVX512-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX512-NEXT:    addw %ax, %si
-; AVX512-NEXT:    cmovol %ecx, %esi
-; AVX512-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX512-NEXT:    vextracti128 $1, %ymm1, %xmm2
-; AVX512-NEXT:    vpextrw $7, %xmm2, %eax
-; AVX512-NEXT:    vextracti128 $1, %ymm0, %xmm3
-; AVX512-NEXT:    vpextrw $7, %xmm3, %esi
-; AVX512-NEXT:    xorl %ecx, %ecx
-; AVX512-NEXT:    movl %esi, %edx
-; AVX512-NEXT:    addw %ax, %dx
-; AVX512-NEXT:    setns %cl
-; AVX512-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX512-NEXT:    addw %ax, %si
-; AVX512-NEXT:    cmovol %ecx, %esi
-; AVX512-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX512-NEXT:    vpextrw $6, %xmm2, %eax
-; AVX512-NEXT:    vpextrw $6, %xmm3, %esi
-; AVX512-NEXT:    xorl %ecx, %ecx
-; AVX512-NEXT:    movl %esi, %edx
-; AVX512-NEXT:    addw %ax, %dx
-; AVX512-NEXT:    setns %cl
-; AVX512-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX512-NEXT:    addw %ax, %si
-; AVX512-NEXT:    cmovol %ecx, %esi
-; AVX512-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX512-NEXT:    vpextrw $5, %xmm2, %eax
-; AVX512-NEXT:    vpextrw $5, %xmm3, %esi
-; AVX512-NEXT:    xorl %ecx, %ecx
-; AVX512-NEXT:    movl %esi, %edx
-; AVX512-NEXT:    addw %ax, %dx
-; AVX512-NEXT:    setns %cl
-; AVX512-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX512-NEXT:    addw %ax, %si
-; AVX512-NEXT:    cmovol %ecx, %esi
-; AVX512-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX512-NEXT:    vpextrw $4, %xmm2, %eax
-; AVX512-NEXT:    vpextrw $4, %xmm3, %esi
-; AVX512-NEXT:    xorl %ecx, %ecx
-; AVX512-NEXT:    movl %esi, %edx
-; AVX512-NEXT:    addw %ax, %dx
-; AVX512-NEXT:    setns %cl
-; AVX512-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX512-NEXT:    addw %ax, %si
-; AVX512-NEXT:    cmovol %ecx, %esi
-; AVX512-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX512-NEXT:    vpextrw $3, %xmm2, %eax
-; AVX512-NEXT:    vpextrw $3, %xmm3, %esi
-; AVX512-NEXT:    xorl %ecx, %ecx
-; AVX512-NEXT:    movl %esi, %edx
-; AVX512-NEXT:    addw %ax, %dx
-; AVX512-NEXT:    setns %cl
-; AVX512-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX512-NEXT:    addw %ax, %si
-; AVX512-NEXT:    cmovol %ecx, %esi
-; AVX512-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX512-NEXT:    vpextrw $2, %xmm2, %eax
-; AVX512-NEXT:    vpextrw $2, %xmm3, %esi
-; AVX512-NEXT:    xorl %ecx, %ecx
-; AVX512-NEXT:    movl %esi, %edx
-; AVX512-NEXT:    addw %ax, %dx
-; AVX512-NEXT:    setns %cl
-; AVX512-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX512-NEXT:    addw %ax, %si
-; AVX512-NEXT:    cmovol %ecx, %esi
-; AVX512-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX512-NEXT:    vmovd %xmm2, %eax
-; AVX512-NEXT:    vmovd %xmm3, %esi
-; AVX512-NEXT:    xorl %ecx, %ecx
-; AVX512-NEXT:    movl %esi, %edx
-; AVX512-NEXT:    addw %ax, %dx
-; AVX512-NEXT:    setns %cl
-; AVX512-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX512-NEXT:    addw %ax, %si
-; AVX512-NEXT:    cmovol %ecx, %esi
-; AVX512-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX512-NEXT:    vpextrw $1, %xmm2, %eax
-; AVX512-NEXT:    vpextrw $1, %xmm3, %esi
-; AVX512-NEXT:    xorl %ecx, %ecx
-; AVX512-NEXT:    movl %esi, %edx
-; AVX512-NEXT:    addw %ax, %dx
-; AVX512-NEXT:    setns %cl
-; AVX512-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX512-NEXT:    addw %ax, %si
-; AVX512-NEXT:    cmovol %ecx, %esi
-; AVX512-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX512-NEXT:    vextracti32x4 $2, %zmm1, %xmm2
-; AVX512-NEXT:    vextracti32x4 $2, %zmm0, %xmm3
-; AVX512-NEXT:    vpextrw $7, %xmm2, %eax
-; AVX512-NEXT:    vpextrw $7, %xmm3, %esi
-; AVX512-NEXT:    xorl %ecx, %ecx
-; AVX512-NEXT:    movl %esi, %edx
-; AVX512-NEXT:    addw %ax, %dx
-; AVX512-NEXT:    setns %cl
-; AVX512-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX512-NEXT:    addw %ax, %si
-; AVX512-NEXT:    cmovol %ecx, %esi
-; AVX512-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX512-NEXT:    vpextrw $6, %xmm2, %eax
-; AVX512-NEXT:    vpextrw $6, %xmm3, %esi
-; AVX512-NEXT:    xorl %ecx, %ecx
-; AVX512-NEXT:    movl %esi, %edx
-; AVX512-NEXT:    addw %ax, %dx
-; AVX512-NEXT:    setns %cl
-; AVX512-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX512-NEXT:    addw %ax, %si
-; AVX512-NEXT:    cmovol %ecx, %esi
-; AVX512-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX512-NEXT:    vpextrw $5, %xmm2, %eax
-; AVX512-NEXT:    vpextrw $5, %xmm3, %esi
-; AVX512-NEXT:    xorl %ecx, %ecx
-; AVX512-NEXT:    movl %esi, %edx
-; AVX512-NEXT:    addw %ax, %dx
-; AVX512-NEXT:    setns %cl
-; AVX512-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX512-NEXT:    addw %ax, %si
-; AVX512-NEXT:    cmovol %ecx, %esi
-; AVX512-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX512-NEXT:    vpextrw $4, %xmm2, %eax
-; AVX512-NEXT:    vpextrw $4, %xmm3, %esi
-; AVX512-NEXT:    xorl %ecx, %ecx
-; AVX512-NEXT:    movl %esi, %edx
-; AVX512-NEXT:    addw %ax, %dx
-; AVX512-NEXT:    setns %cl
-; AVX512-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX512-NEXT:    addw %ax, %si
-; AVX512-NEXT:    cmovol %ecx, %esi
-; AVX512-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX512-NEXT:    vpextrw $3, %xmm2, %eax
-; AVX512-NEXT:    vpextrw $3, %xmm3, %ebx
-; AVX512-NEXT:    xorl %ecx, %ecx
-; AVX512-NEXT:    movl %ebx, %edx
-; AVX512-NEXT:    addw %ax, %dx
-; AVX512-NEXT:    setns %cl
-; AVX512-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX512-NEXT:    addw %ax, %bx
-; AVX512-NEXT:    cmovol %ecx, %ebx
-; AVX512-NEXT:    vpextrw $2, %xmm2, %eax
-; AVX512-NEXT:    vpextrw $2, %xmm3, %r11d
-; AVX512-NEXT:    xorl %ecx, %ecx
-; AVX512-NEXT:    movl %r11d, %edx
-; AVX512-NEXT:    addw %ax, %dx
-; AVX512-NEXT:    setns %cl
-; AVX512-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX512-NEXT:    addw %ax, %r11w
-; AVX512-NEXT:    cmovol %ecx, %r11d
-; AVX512-NEXT:    vmovd %xmm2, %eax
-; AVX512-NEXT:    vmovd %xmm3, %r10d
-; AVX512-NEXT:    xorl %ecx, %ecx
-; AVX512-NEXT:    movl %r10d, %edx
-; AVX512-NEXT:    addw %ax, %dx
-; AVX512-NEXT:    setns %cl
-; AVX512-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX512-NEXT:    addw %ax, %r10w
-; AVX512-NEXT:    cmovol %ecx, %r10d
-; AVX512-NEXT:    vpextrw $1, %xmm2, %eax
-; AVX512-NEXT:    vpextrw $1, %xmm3, %r9d
-; AVX512-NEXT:    xorl %ecx, %ecx
-; AVX512-NEXT:    movl %r9d, %edx
-; AVX512-NEXT:    addw %ax, %dx
-; AVX512-NEXT:    setns %cl
-; AVX512-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX512-NEXT:    addw %ax, %r9w
-; AVX512-NEXT:    cmovol %ecx, %r9d
-; AVX512-NEXT:    vextracti32x4 $3, %zmm1, %xmm1
-; AVX512-NEXT:    vextracti32x4 $3, %zmm0, %xmm0
-; AVX512-NEXT:    vpextrw $7, %xmm1, %eax
-; AVX512-NEXT:    vpextrw $7, %xmm0, %r8d
-; AVX512-NEXT:    xorl %ecx, %ecx
-; AVX512-NEXT:    movl %r8d, %edx
-; AVX512-NEXT:    addw %ax, %dx
-; AVX512-NEXT:    setns %cl
-; AVX512-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX512-NEXT:    addw %ax, %r8w
-; AVX512-NEXT:    cmovol %ecx, %r8d
-; AVX512-NEXT:    vpextrw $6, %xmm1, %eax
-; AVX512-NEXT:    vpextrw $6, %xmm0, %edi
-; AVX512-NEXT:    xorl %ecx, %ecx
-; AVX512-NEXT:    movl %edi, %edx
-; AVX512-NEXT:    addw %ax, %dx
-; AVX512-NEXT:    setns %cl
-; AVX512-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX512-NEXT:    addw %ax, %di
-; AVX512-NEXT:    cmovol %ecx, %edi
-; AVX512-NEXT:    vpextrw $5, %xmm1, %eax
-; AVX512-NEXT:    vpextrw $5, %xmm0, %esi
-; AVX512-NEXT:    xorl %ecx, %ecx
-; AVX512-NEXT:    movl %esi, %edx
-; AVX512-NEXT:    addw %ax, %dx
-; AVX512-NEXT:    setns %cl
-; AVX512-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX512-NEXT:    addw %ax, %si
-; AVX512-NEXT:    cmovol %ecx, %esi
-; AVX512-NEXT:    vpextrw $4, %xmm1, %eax
-; AVX512-NEXT:    vpextrw $4, %xmm0, %edx
-; AVX512-NEXT:    xorl %ecx, %ecx
-; AVX512-NEXT:    movl %edx, %r13d
-; AVX512-NEXT:    addw %ax, %r13w
-; AVX512-NEXT:    setns %cl
-; AVX512-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX512-NEXT:    addw %ax, %dx
-; AVX512-NEXT:    cmovol %ecx, %edx
-; AVX512-NEXT:    vpextrw $3, %xmm1, %eax
-; AVX512-NEXT:    vpextrw $3, %xmm0, %r13d
-; AVX512-NEXT:    xorl %ecx, %ecx
-; AVX512-NEXT:    movl %r13d, %r12d
-; AVX512-NEXT:    addw %ax, %r12w
-; AVX512-NEXT:    setns %cl
-; AVX512-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX512-NEXT:    addw %ax, %r13w
-; AVX512-NEXT:    cmovol %ecx, %r13d
-; AVX512-NEXT:    vpextrw $2, %xmm1, %r12d
-; AVX512-NEXT:    vpextrw $2, %xmm0, %eax
-; AVX512-NEXT:    xorl %ecx, %ecx
-; AVX512-NEXT:    movl %eax, %r15d
-; AVX512-NEXT:    addw %r12w, %r15w
-; AVX512-NEXT:    setns %cl
-; AVX512-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX512-NEXT:    addw %r12w, %ax
-; AVX512-NEXT:    cmovol %ecx, %eax
-; AVX512-NEXT:    vmovd %xmm1, %r15d
-; AVX512-NEXT:    vmovd %xmm0, %r12d
-; AVX512-NEXT:    xorl %ecx, %ecx
-; AVX512-NEXT:    movl %r12d, %r14d
-; AVX512-NEXT:    addw %r15w, %r14w
-; AVX512-NEXT:    setns %cl
-; AVX512-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX512-NEXT:    addw %r15w, %r12w
-; AVX512-NEXT:    cmovol %ecx, %r12d
-; AVX512-NEXT:    vpextrw $1, %xmm1, %r14d
-; AVX512-NEXT:    vpextrw $1, %xmm0, %ecx
-; AVX512-NEXT:    xorl %r15d, %r15d
-; AVX512-NEXT:    movl %ecx, %ebp
-; AVX512-NEXT:    addw %r14w, %bp
-; AVX512-NEXT:    setns %r15b
-; AVX512-NEXT:    addl $32767, %r15d # imm = 0x7FFF
-; AVX512-NEXT:    addw %r14w, %cx
-; AVX512-NEXT:    cmovol %r15d, %ecx
-; AVX512-NEXT:    vmovd %r12d, %xmm0
-; AVX512-NEXT:    vpinsrw $1, %ecx, %xmm0, %xmm0
-; AVX512-NEXT:    vpinsrw $2, %eax, %xmm0, %xmm0
-; AVX512-NEXT:    vpinsrw $3, %r13d, %xmm0, %xmm0
-; AVX512-NEXT:    vpinsrw $4, %edx, %xmm0, %xmm0
-; AVX512-NEXT:    vpinsrw $5, %esi, %xmm0, %xmm0
-; AVX512-NEXT:    vpinsrw $6, %edi, %xmm0, %xmm0
-; AVX512-NEXT:    vpinsrw $7, %r8d, %xmm0, %xmm0
-; AVX512-NEXT:    vmovd %r10d, %xmm1
-; AVX512-NEXT:    vpinsrw $1, %r9d, %xmm1, %xmm1
-; AVX512-NEXT:    vpinsrw $2, %r11d, %xmm1, %xmm1
-; AVX512-NEXT:    vpinsrw $3, %ebx, %xmm1, %xmm1
-; AVX512-NEXT:    vpinsrw $4, {{[-0-9]+}}(%r{{[sb]}}p), %xmm1, %xmm1 # 4-byte Folded Reload
-; AVX512-NEXT:    vpinsrw $5, {{[-0-9]+}}(%r{{[sb]}}p), %xmm1, %xmm1 # 4-byte Folded Reload
-; AVX512-NEXT:    vpinsrw $6, {{[-0-9]+}}(%r{{[sb]}}p), %xmm1, %xmm1 # 4-byte Folded Reload
-; AVX512-NEXT:    vpinsrw $7, {{[-0-9]+}}(%r{{[sb]}}p), %xmm1, %xmm1 # 4-byte Folded Reload
-; AVX512-NEXT:    vmovd {{[-0-9]+}}(%r{{[sb]}}p), %xmm2 # 4-byte Folded Reload
-; AVX512-NEXT:    # xmm2 = mem[0],zero,zero,zero
-; AVX512-NEXT:    vpinsrw $1, {{[-0-9]+}}(%r{{[sb]}}p), %xmm2, %xmm2 # 4-byte Folded Reload
-; AVX512-NEXT:    vpinsrw $2, {{[-0-9]+}}(%r{{[sb]}}p), %xmm2, %xmm2 # 4-byte Folded Reload
-; AVX512-NEXT:    vpinsrw $3, {{[-0-9]+}}(%r{{[sb]}}p), %xmm2, %xmm2 # 4-byte Folded Reload
-; AVX512-NEXT:    vpinsrw $4, {{[-0-9]+}}(%r{{[sb]}}p), %xmm2, %xmm2 # 4-byte Folded Reload
-; AVX512-NEXT:    vpinsrw $5, {{[-0-9]+}}(%r{{[sb]}}p), %xmm2, %xmm2 # 4-byte Folded Reload
-; AVX512-NEXT:    vpinsrw $6, {{[-0-9]+}}(%r{{[sb]}}p), %xmm2, %xmm2 # 4-byte Folded Reload
-; AVX512-NEXT:    vpinsrw $7, {{[-0-9]+}}(%r{{[sb]}}p), %xmm2, %xmm2 # 4-byte Folded Reload
-; AVX512-NEXT:    vmovd {{[-0-9]+}}(%r{{[sb]}}p), %xmm3 # 4-byte Folded Reload
-; AVX512-NEXT:    # xmm3 = mem[0],zero,zero,zero
-; AVX512-NEXT:    vpinsrw $1, {{[-0-9]+}}(%r{{[sb]}}p), %xmm3, %xmm3 # 4-byte Folded Reload
-; AVX512-NEXT:    vpinsrw $2, {{[-0-9]+}}(%r{{[sb]}}p), %xmm3, %xmm3 # 4-byte Folded Reload
-; AVX512-NEXT:    vpinsrw $3, {{[-0-9]+}}(%r{{[sb]}}p), %xmm3, %xmm3 # 4-byte Folded Reload
-; AVX512-NEXT:    vpinsrw $4, {{[-0-9]+}}(%r{{[sb]}}p), %xmm3, %xmm3 # 4-byte Folded Reload
-; AVX512-NEXT:    vpinsrw $5, {{[-0-9]+}}(%r{{[sb]}}p), %xmm3, %xmm3 # 4-byte Folded Reload
-; AVX512-NEXT:    vpinsrw $6, {{[-0-9]+}}(%r{{[sb]}}p), %xmm3, %xmm3 # 4-byte Folded Reload
-; AVX512-NEXT:    vpinsrw $7, {{[-0-9]+}}(%r{{[sb]}}p), %xmm3, %xmm3 # 4-byte Folded Reload
-; AVX512-NEXT:    vinserti128 $1, %xmm0, %ymm1, %ymm0
-; AVX512-NEXT:    vinserti128 $1, %xmm2, %ymm3, %ymm1
-; AVX512-NEXT:    vinserti64x4 $1, %ymm0, %zmm1, %zmm0
-; AVX512-NEXT:    popq %rbx
-; AVX512-NEXT:    popq %r12
-; AVX512-NEXT:    popq %r13
-; AVX512-NEXT:    popq %r14
-; AVX512-NEXT:    popq %r15
-; AVX512-NEXT:    popq %rbp
+; AVX512-NEXT:    vpaddsw %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    retq
   %z = call <32 x i16> @llvm.sadd.sat.v32i16(<32 x i16> %x, <32 x i16> %y)
   ret <32 x i16> %z
@@ -13489,608 +196,37 @@ define <32 x i16> @v32i16(<32 x i16> %x,
 ; Too narrow vectors, legalized by widening.
 
 define void @v8i8(<8 x i8>* %px, <8 x i8>* %py, <8 x i8>* %pz) nounwind {
-; SSE2-LABEL: v8i8:
-; SSE2:       # %bb.0:
-; SSE2-NEXT:    pushq %rbp
-; SSE2-NEXT:    pushq %r14
-; SSE2-NEXT:    pushq %rbx
-; SSE2-NEXT:    movq {{.*#+}} xmm2 = mem[0],zero
-; SSE2-NEXT:    movq {{.*#+}} xmm3 = mem[0],zero
-; SSE2-NEXT:    pxor %xmm0, %xmm0
-; SSE2-NEXT:    pxor %xmm1, %xmm1
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm3[0],xmm1[1],xmm3[1],xmm1[2],xmm3[2],xmm1[3],xmm3[3],xmm1[4],xmm3[4],xmm1[5],xmm3[5],xmm1[6],xmm3[6],xmm1[7],xmm3[7]
-; SSE2-NEXT:    movd %xmm1, %eax
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1],xmm0[2],xmm2[2],xmm0[3],xmm2[3],xmm0[4],xmm2[4],xmm0[5],xmm2[5],xmm0[6],xmm2[6],xmm0[7],xmm2[7]
-; SSE2-NEXT:    movd %xmm0, %r8d
-; SSE2-NEXT:    xorl %ecx, %ecx
-; SSE2-NEXT:    movl %r8d, %esi
-; SSE2-NEXT:    addw %ax, %si
-; SSE2-NEXT:    setns %cl
-; SSE2-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSE2-NEXT:    addw %ax, %r8w
-; SSE2-NEXT:    cmovol %ecx, %r8d
-; SSE2-NEXT:    pextrw $1, %xmm1, %eax
-; SSE2-NEXT:    pextrw $1, %xmm0, %r9d
-; SSE2-NEXT:    xorl %ecx, %ecx
-; SSE2-NEXT:    movl %r9d, %esi
-; SSE2-NEXT:    addw %ax, %si
-; SSE2-NEXT:    setns %cl
-; SSE2-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSE2-NEXT:    addw %ax, %r9w
-; SSE2-NEXT:    cmovol %ecx, %r9d
-; SSE2-NEXT:    pextrw $2, %xmm1, %eax
-; SSE2-NEXT:    pextrw $2, %xmm0, %r10d
-; SSE2-NEXT:    xorl %ecx, %ecx
-; SSE2-NEXT:    movl %r10d, %esi
-; SSE2-NEXT:    addw %ax, %si
-; SSE2-NEXT:    setns %cl
-; SSE2-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSE2-NEXT:    addw %ax, %r10w
-; SSE2-NEXT:    cmovol %ecx, %r10d
-; SSE2-NEXT:    pextrw $3, %xmm1, %eax
-; SSE2-NEXT:    pextrw $3, %xmm0, %r11d
-; SSE2-NEXT:    xorl %ecx, %ecx
-; SSE2-NEXT:    movl %r11d, %esi
-; SSE2-NEXT:    addw %ax, %si
-; SSE2-NEXT:    setns %cl
-; SSE2-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSE2-NEXT:    addw %ax, %r11w
-; SSE2-NEXT:    cmovol %ecx, %r11d
-; SSE2-NEXT:    pextrw $4, %xmm1, %ecx
-; SSE2-NEXT:    pextrw $4, %xmm0, %r14d
-; SSE2-NEXT:    xorl %esi, %esi
-; SSE2-NEXT:    movl %r14d, %edi
-; SSE2-NEXT:    addw %cx, %di
-; SSE2-NEXT:    setns %sil
-; SSE2-NEXT:    addl $32767, %esi # imm = 0x7FFF
-; SSE2-NEXT:    addw %cx, %r14w
-; SSE2-NEXT:    cmovol %esi, %r14d
-; SSE2-NEXT:    pextrw $5, %xmm1, %esi
-; SSE2-NEXT:    pextrw $5, %xmm0, %ecx
-; SSE2-NEXT:    xorl %edi, %edi
-; SSE2-NEXT:    movl %ecx, %ebx
-; SSE2-NEXT:    addw %si, %bx
-; SSE2-NEXT:    setns %dil
-; SSE2-NEXT:    addl $32767, %edi # imm = 0x7FFF
-; SSE2-NEXT:    addw %si, %cx
-; SSE2-NEXT:    cmovol %edi, %ecx
-; SSE2-NEXT:    pextrw $6, %xmm1, %edi
-; SSE2-NEXT:    pextrw $6, %xmm0, %esi
-; SSE2-NEXT:    xorl %ebx, %ebx
-; SSE2-NEXT:    movl %esi, %ebp
-; SSE2-NEXT:    addw %di, %bp
-; SSE2-NEXT:    setns %bl
-; SSE2-NEXT:    addl $32767, %ebx # imm = 0x7FFF
-; SSE2-NEXT:    addw %di, %si
-; SSE2-NEXT:    cmovol %ebx, %esi
-; SSE2-NEXT:    pextrw $7, %xmm1, %edi
-; SSE2-NEXT:    pextrw $7, %xmm0, %ebx
-; SSE2-NEXT:    xorl %ebp, %ebp
-; SSE2-NEXT:    movl %ebx, %eax
-; SSE2-NEXT:    addw %di, %ax
-; SSE2-NEXT:    setns %bpl
-; SSE2-NEXT:    addl $32767, %ebp # imm = 0x7FFF
-; SSE2-NEXT:    addw %di, %bx
-; SSE2-NEXT:    cmovol %ebp, %ebx
-; SSE2-NEXT:    movd %ebx, %xmm0
-; SSE2-NEXT:    movd %esi, %xmm1
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3]
-; SSE2-NEXT:    movd %ecx, %xmm0
-; SSE2-NEXT:    movd %r14d, %xmm2
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm2 = xmm2[0],xmm0[0],xmm2[1],xmm0[1],xmm2[2],xmm0[2],xmm2[3],xmm0[3]
-; SSE2-NEXT:    punpckldq {{.*#+}} xmm2 = xmm2[0],xmm1[0],xmm2[1],xmm1[1]
-; SSE2-NEXT:    movd %r11d, %xmm0
-; SSE2-NEXT:    movd %r10d, %xmm1
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3]
-; SSE2-NEXT:    movd %r9d, %xmm0
-; SSE2-NEXT:    movd %r8d, %xmm3
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm3 = xmm3[0],xmm0[0],xmm3[1],xmm0[1],xmm3[2],xmm0[2],xmm3[3],xmm0[3]
-; SSE2-NEXT:    punpckldq {{.*#+}} xmm3 = xmm3[0],xmm1[0],xmm3[1],xmm1[1]
-; SSE2-NEXT:    punpcklqdq {{.*#+}} xmm3 = xmm3[0],xmm2[0]
-; SSE2-NEXT:    psrlw $8, %xmm3
-; SSE2-NEXT:    packuswb %xmm0, %xmm3
-; SSE2-NEXT:    movq %xmm3, (%rdx)
-; SSE2-NEXT:    popq %rbx
-; SSE2-NEXT:    popq %r14
-; SSE2-NEXT:    popq %rbp
-; SSE2-NEXT:    retq
-;
-; SSSE3-LABEL: v8i8:
-; SSSE3:       # %bb.0:
-; SSSE3-NEXT:    pushq %rbp
-; SSSE3-NEXT:    pushq %r14
-; SSSE3-NEXT:    pushq %rbx
-; SSSE3-NEXT:    movq {{.*#+}} xmm2 = mem[0],zero
-; SSSE3-NEXT:    movq {{.*#+}} xmm3 = mem[0],zero
-; SSSE3-NEXT:    pxor %xmm0, %xmm0
-; SSSE3-NEXT:    pxor %xmm1, %xmm1
-; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm3[0],xmm1[1],xmm3[1],xmm1[2],xmm3[2],xmm1[3],xmm3[3],xmm1[4],xmm3[4],xmm1[5],xmm3[5],xmm1[6],xmm3[6],xmm1[7],xmm3[7]
-; SSSE3-NEXT:    movd %xmm1, %eax
-; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1],xmm0[2],xmm2[2],xmm0[3],xmm2[3],xmm0[4],xmm2[4],xmm0[5],xmm2[5],xmm0[6],xmm2[6],xmm0[7],xmm2[7]
-; SSSE3-NEXT:    movd %xmm0, %r8d
-; SSSE3-NEXT:    xorl %ecx, %ecx
-; SSSE3-NEXT:    movl %r8d, %esi
-; SSSE3-NEXT:    addw %ax, %si
-; SSSE3-NEXT:    setns %cl
-; SSSE3-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSSE3-NEXT:    addw %ax, %r8w
-; SSSE3-NEXT:    cmovol %ecx, %r8d
-; SSSE3-NEXT:    pextrw $1, %xmm1, %eax
-; SSSE3-NEXT:    pextrw $1, %xmm0, %r9d
-; SSSE3-NEXT:    xorl %ecx, %ecx
-; SSSE3-NEXT:    movl %r9d, %esi
-; SSSE3-NEXT:    addw %ax, %si
-; SSSE3-NEXT:    setns %cl
-; SSSE3-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSSE3-NEXT:    addw %ax, %r9w
-; SSSE3-NEXT:    cmovol %ecx, %r9d
-; SSSE3-NEXT:    pextrw $2, %xmm1, %eax
-; SSSE3-NEXT:    pextrw $2, %xmm0, %r10d
-; SSSE3-NEXT:    xorl %ecx, %ecx
-; SSSE3-NEXT:    movl %r10d, %esi
-; SSSE3-NEXT:    addw %ax, %si
-; SSSE3-NEXT:    setns %cl
-; SSSE3-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSSE3-NEXT:    addw %ax, %r10w
-; SSSE3-NEXT:    cmovol %ecx, %r10d
-; SSSE3-NEXT:    pextrw $3, %xmm1, %eax
-; SSSE3-NEXT:    pextrw $3, %xmm0, %r11d
-; SSSE3-NEXT:    xorl %ecx, %ecx
-; SSSE3-NEXT:    movl %r11d, %esi
-; SSSE3-NEXT:    addw %ax, %si
-; SSSE3-NEXT:    setns %cl
-; SSSE3-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSSE3-NEXT:    addw %ax, %r11w
-; SSSE3-NEXT:    cmovol %ecx, %r11d
-; SSSE3-NEXT:    pextrw $4, %xmm1, %ecx
-; SSSE3-NEXT:    pextrw $4, %xmm0, %r14d
-; SSSE3-NEXT:    xorl %esi, %esi
-; SSSE3-NEXT:    movl %r14d, %edi
-; SSSE3-NEXT:    addw %cx, %di
-; SSSE3-NEXT:    setns %sil
-; SSSE3-NEXT:    addl $32767, %esi # imm = 0x7FFF
-; SSSE3-NEXT:    addw %cx, %r14w
-; SSSE3-NEXT:    cmovol %esi, %r14d
-; SSSE3-NEXT:    pextrw $5, %xmm1, %esi
-; SSSE3-NEXT:    pextrw $5, %xmm0, %ecx
-; SSSE3-NEXT:    xorl %edi, %edi
-; SSSE3-NEXT:    movl %ecx, %ebx
-; SSSE3-NEXT:    addw %si, %bx
-; SSSE3-NEXT:    setns %dil
-; SSSE3-NEXT:    addl $32767, %edi # imm = 0x7FFF
-; SSSE3-NEXT:    addw %si, %cx
-; SSSE3-NEXT:    cmovol %edi, %ecx
-; SSSE3-NEXT:    pextrw $6, %xmm1, %edi
-; SSSE3-NEXT:    pextrw $6, %xmm0, %esi
-; SSSE3-NEXT:    xorl %ebx, %ebx
-; SSSE3-NEXT:    movl %esi, %ebp
-; SSSE3-NEXT:    addw %di, %bp
-; SSSE3-NEXT:    setns %bl
-; SSSE3-NEXT:    addl $32767, %ebx # imm = 0x7FFF
-; SSSE3-NEXT:    addw %di, %si
-; SSSE3-NEXT:    cmovol %ebx, %esi
-; SSSE3-NEXT:    pextrw $7, %xmm1, %edi
-; SSSE3-NEXT:    pextrw $7, %xmm0, %ebx
-; SSSE3-NEXT:    xorl %ebp, %ebp
-; SSSE3-NEXT:    movl %ebx, %eax
-; SSSE3-NEXT:    addw %di, %ax
-; SSSE3-NEXT:    setns %bpl
-; SSSE3-NEXT:    addl $32767, %ebp # imm = 0x7FFF
-; SSSE3-NEXT:    addw %di, %bx
-; SSSE3-NEXT:    cmovol %ebp, %ebx
-; SSSE3-NEXT:    movd %ebx, %xmm0
-; SSSE3-NEXT:    movd %esi, %xmm1
-; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3]
-; SSSE3-NEXT:    movd %ecx, %xmm0
-; SSSE3-NEXT:    movd %r14d, %xmm2
-; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm2 = xmm2[0],xmm0[0],xmm2[1],xmm0[1],xmm2[2],xmm0[2],xmm2[3],xmm0[3]
-; SSSE3-NEXT:    punpckldq {{.*#+}} xmm2 = xmm2[0],xmm1[0],xmm2[1],xmm1[1]
-; SSSE3-NEXT:    movd %r11d, %xmm0
-; SSSE3-NEXT:    movd %r10d, %xmm1
-; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3]
-; SSSE3-NEXT:    movd %r9d, %xmm0
-; SSSE3-NEXT:    movd %r8d, %xmm3
-; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm3 = xmm3[0],xmm0[0],xmm3[1],xmm0[1],xmm3[2],xmm0[2],xmm3[3],xmm0[3]
-; SSSE3-NEXT:    punpckldq {{.*#+}} xmm3 = xmm3[0],xmm1[0],xmm3[1],xmm1[1]
-; SSSE3-NEXT:    punpcklqdq {{.*#+}} xmm3 = xmm3[0],xmm2[0]
-; SSSE3-NEXT:    psrlw $8, %xmm3
-; SSSE3-NEXT:    packuswb %xmm0, %xmm3
-; SSSE3-NEXT:    movq %xmm3, (%rdx)
-; SSSE3-NEXT:    popq %rbx
-; SSSE3-NEXT:    popq %r14
-; SSSE3-NEXT:    popq %rbp
-; SSSE3-NEXT:    retq
-;
-; SSE41-LABEL: v8i8:
-; SSE41:       # %bb.0:
-; SSE41-NEXT:    pushq %rbp
-; SSE41-NEXT:    pushq %r14
-; SSE41-NEXT:    pushq %rbx
-; SSE41-NEXT:    movq {{.*#+}} xmm2 = mem[0],zero
-; SSE41-NEXT:    movq {{.*#+}} xmm3 = mem[0],zero
-; SSE41-NEXT:    pxor %xmm0, %xmm0
-; SSE41-NEXT:    pxor %xmm1, %xmm1
-; SSE41-NEXT:    punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm3[0],xmm1[1],xmm3[1],xmm1[2],xmm3[2],xmm1[3],xmm3[3],xmm1[4],xmm3[4],xmm1[5],xmm3[5],xmm1[6],xmm3[6],xmm1[7],xmm3[7]
-; SSE41-NEXT:    pextrw $7, %xmm1, %eax
-; SSE41-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1],xmm0[2],xmm2[2],xmm0[3],xmm2[3],xmm0[4],xmm2[4],xmm0[5],xmm2[5],xmm0[6],xmm2[6],xmm0[7],xmm2[7]
-; SSE41-NEXT:    pextrw $7, %xmm0, %r8d
-; SSE41-NEXT:    xorl %ecx, %ecx
-; SSE41-NEXT:    movl %r8d, %esi
-; SSE41-NEXT:    addw %ax, %si
-; SSE41-NEXT:    setns %cl
-; SSE41-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSE41-NEXT:    addw %ax, %r8w
-; SSE41-NEXT:    cmovol %ecx, %r8d
-; SSE41-NEXT:    pextrw $6, %xmm1, %eax
-; SSE41-NEXT:    pextrw $6, %xmm0, %r9d
-; SSE41-NEXT:    xorl %ecx, %ecx
-; SSE41-NEXT:    movl %r9d, %esi
-; SSE41-NEXT:    addw %ax, %si
-; SSE41-NEXT:    setns %cl
-; SSE41-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSE41-NEXT:    addw %ax, %r9w
-; SSE41-NEXT:    cmovol %ecx, %r9d
-; SSE41-NEXT:    pextrw $5, %xmm1, %eax
-; SSE41-NEXT:    pextrw $5, %xmm0, %r10d
-; SSE41-NEXT:    xorl %ecx, %ecx
-; SSE41-NEXT:    movl %r10d, %edi
-; SSE41-NEXT:    addw %ax, %di
-; SSE41-NEXT:    setns %cl
-; SSE41-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSE41-NEXT:    addw %ax, %r10w
-; SSE41-NEXT:    cmovol %ecx, %r10d
-; SSE41-NEXT:    pextrw $4, %xmm1, %eax
-; SSE41-NEXT:    pextrw $4, %xmm0, %r11d
-; SSE41-NEXT:    xorl %ecx, %ecx
-; SSE41-NEXT:    movl %r11d, %esi
-; SSE41-NEXT:    addw %ax, %si
-; SSE41-NEXT:    setns %cl
-; SSE41-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSE41-NEXT:    addw %ax, %r11w
-; SSE41-NEXT:    cmovol %ecx, %r11d
-; SSE41-NEXT:    pextrw $3, %xmm1, %ecx
-; SSE41-NEXT:    pextrw $3, %xmm0, %r14d
-; SSE41-NEXT:    xorl %esi, %esi
-; SSE41-NEXT:    movl %r14d, %edi
-; SSE41-NEXT:    addw %cx, %di
-; SSE41-NEXT:    setns %sil
-; SSE41-NEXT:    addl $32767, %esi # imm = 0x7FFF
-; SSE41-NEXT:    addw %cx, %r14w
-; SSE41-NEXT:    cmovol %esi, %r14d
-; SSE41-NEXT:    pextrw $2, %xmm1, %esi
-; SSE41-NEXT:    pextrw $2, %xmm0, %ecx
-; SSE41-NEXT:    xorl %edi, %edi
-; SSE41-NEXT:    movl %ecx, %ebx
-; SSE41-NEXT:    addw %si, %bx
-; SSE41-NEXT:    setns %dil
-; SSE41-NEXT:    addl $32767, %edi # imm = 0x7FFF
-; SSE41-NEXT:    addw %si, %cx
-; SSE41-NEXT:    cmovol %edi, %ecx
-; SSE41-NEXT:    movd %xmm1, %esi
-; SSE41-NEXT:    movd %xmm0, %edi
-; SSE41-NEXT:    xorl %ebx, %ebx
-; SSE41-NEXT:    movl %edi, %ebp
-; SSE41-NEXT:    addw %si, %bp
-; SSE41-NEXT:    setns %bl
-; SSE41-NEXT:    addl $32767, %ebx # imm = 0x7FFF
-; SSE41-NEXT:    addw %si, %di
-; SSE41-NEXT:    cmovol %ebx, %edi
-; SSE41-NEXT:    pextrw $1, %xmm1, %esi
-; SSE41-NEXT:    pextrw $1, %xmm0, %ebx
-; SSE41-NEXT:    xorl %ebp, %ebp
-; SSE41-NEXT:    movl %ebx, %eax
-; SSE41-NEXT:    addw %si, %ax
-; SSE41-NEXT:    setns %bpl
-; SSE41-NEXT:    addl $32767, %ebp # imm = 0x7FFF
-; SSE41-NEXT:    addw %si, %bx
-; SSE41-NEXT:    cmovol %ebp, %ebx
-; SSE41-NEXT:    movd %edi, %xmm0
-; SSE41-NEXT:    pinsrw $1, %ebx, %xmm0
-; SSE41-NEXT:    pinsrw $2, %ecx, %xmm0
-; SSE41-NEXT:    pinsrw $3, %r14d, %xmm0
-; SSE41-NEXT:    pinsrw $4, %r11d, %xmm0
-; SSE41-NEXT:    pinsrw $5, %r10d, %xmm0
-; SSE41-NEXT:    pinsrw $6, %r9d, %xmm0
-; SSE41-NEXT:    pinsrw $7, %r8d, %xmm0
-; SSE41-NEXT:    psrlw $8, %xmm0
-; SSE41-NEXT:    packuswb %xmm0, %xmm0
-; SSE41-NEXT:    movq %xmm0, (%rdx)
-; SSE41-NEXT:    popq %rbx
-; SSE41-NEXT:    popq %r14
-; SSE41-NEXT:    popq %rbp
-; SSE41-NEXT:    retq
+; SSE-LABEL: v8i8:
+; SSE:       # %bb.0:
+; SSE-NEXT:    movq {{.*#+}} xmm0 = mem[0],zero
+; SSE-NEXT:    movq {{.*#+}} xmm1 = mem[0],zero
+; SSE-NEXT:    paddsb %xmm0, %xmm1
+; SSE-NEXT:    movq %xmm1, (%rdx)
+; SSE-NEXT:    retq
 ;
 ; AVX1-LABEL: v8i8:
 ; AVX1:       # %bb.0:
-; AVX1-NEXT:    pushq %rbp
-; AVX1-NEXT:    pushq %r14
-; AVX1-NEXT:    pushq %rbx
-; AVX1-NEXT:    vmovq {{.*#+}} xmm1 = mem[0],zero
 ; AVX1-NEXT:    vmovq {{.*#+}} xmm0 = mem[0],zero
-; AVX1-NEXT:    vpxor %xmm2, %xmm2, %xmm2
-; AVX1-NEXT:    vpunpcklbw {{.*#+}} xmm0 = xmm2[0],xmm0[0],xmm2[1],xmm0[1],xmm2[2],xmm0[2],xmm2[3],xmm0[3],xmm2[4],xmm0[4],xmm2[5],xmm0[5],xmm2[6],xmm0[6],xmm2[7],xmm0[7]
-; AVX1-NEXT:    vpextrw $7, %xmm0, %eax
-; AVX1-NEXT:    vpunpcklbw {{.*#+}} xmm1 = xmm2[0],xmm1[0],xmm2[1],xmm1[1],xmm2[2],xmm1[2],xmm2[3],xmm1[3],xmm2[4],xmm1[4],xmm2[5],xmm1[5],xmm2[6],xmm1[6],xmm2[7],xmm1[7]
-; AVX1-NEXT:    vpextrw $7, %xmm1, %r8d
-; AVX1-NEXT:    xorl %ecx, %ecx
-; AVX1-NEXT:    movl %r8d, %esi
-; AVX1-NEXT:    addw %ax, %si
-; AVX1-NEXT:    setns %cl
-; AVX1-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX1-NEXT:    addw %ax, %r8w
-; AVX1-NEXT:    cmovol %ecx, %r8d
-; AVX1-NEXT:    vpextrw $6, %xmm0, %eax
-; AVX1-NEXT:    vpextrw $6, %xmm1, %r9d
-; AVX1-NEXT:    xorl %ecx, %ecx
-; AVX1-NEXT:    movl %r9d, %esi
-; AVX1-NEXT:    addw %ax, %si
-; AVX1-NEXT:    setns %cl
-; AVX1-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX1-NEXT:    addw %ax, %r9w
-; AVX1-NEXT:    cmovol %ecx, %r9d
-; AVX1-NEXT:    vpextrw $5, %xmm0, %eax
-; AVX1-NEXT:    vpextrw $5, %xmm1, %r10d
-; AVX1-NEXT:    xorl %ecx, %ecx
-; AVX1-NEXT:    movl %r10d, %edi
-; AVX1-NEXT:    addw %ax, %di
-; AVX1-NEXT:    setns %cl
-; AVX1-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX1-NEXT:    addw %ax, %r10w
-; AVX1-NEXT:    cmovol %ecx, %r10d
-; AVX1-NEXT:    vpextrw $4, %xmm0, %eax
-; AVX1-NEXT:    vpextrw $4, %xmm1, %r11d
-; AVX1-NEXT:    xorl %ecx, %ecx
-; AVX1-NEXT:    movl %r11d, %esi
-; AVX1-NEXT:    addw %ax, %si
-; AVX1-NEXT:    setns %cl
-; AVX1-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX1-NEXT:    addw %ax, %r11w
-; AVX1-NEXT:    cmovol %ecx, %r11d
-; AVX1-NEXT:    vpextrw $3, %xmm0, %ecx
-; AVX1-NEXT:    vpextrw $3, %xmm1, %r14d
-; AVX1-NEXT:    xorl %esi, %esi
-; AVX1-NEXT:    movl %r14d, %edi
-; AVX1-NEXT:    addw %cx, %di
-; AVX1-NEXT:    setns %sil
-; AVX1-NEXT:    addl $32767, %esi # imm = 0x7FFF
-; AVX1-NEXT:    addw %cx, %r14w
-; AVX1-NEXT:    cmovol %esi, %r14d
-; AVX1-NEXT:    vpextrw $2, %xmm0, %esi
-; AVX1-NEXT:    vpextrw $2, %xmm1, %ecx
-; AVX1-NEXT:    xorl %edi, %edi
-; AVX1-NEXT:    movl %ecx, %ebx
-; AVX1-NEXT:    addw %si, %bx
-; AVX1-NEXT:    setns %dil
-; AVX1-NEXT:    addl $32767, %edi # imm = 0x7FFF
-; AVX1-NEXT:    addw %si, %cx
-; AVX1-NEXT:    cmovol %edi, %ecx
-; AVX1-NEXT:    vmovd %xmm0, %esi
-; AVX1-NEXT:    vmovd %xmm1, %edi
-; AVX1-NEXT:    xorl %ebx, %ebx
-; AVX1-NEXT:    movl %edi, %ebp
-; AVX1-NEXT:    addw %si, %bp
-; AVX1-NEXT:    setns %bl
-; AVX1-NEXT:    addl $32767, %ebx # imm = 0x7FFF
-; AVX1-NEXT:    addw %si, %di
-; AVX1-NEXT:    cmovol %ebx, %edi
-; AVX1-NEXT:    vpextrw $1, %xmm0, %esi
-; AVX1-NEXT:    vpextrw $1, %xmm1, %ebx
-; AVX1-NEXT:    xorl %ebp, %ebp
-; AVX1-NEXT:    movl %ebx, %eax
-; AVX1-NEXT:    addw %si, %ax
-; AVX1-NEXT:    setns %bpl
-; AVX1-NEXT:    addl $32767, %ebp # imm = 0x7FFF
-; AVX1-NEXT:    addw %si, %bx
-; AVX1-NEXT:    cmovol %ebp, %ebx
-; AVX1-NEXT:    vmovd %edi, %xmm0
-; AVX1-NEXT:    vpinsrw $1, %ebx, %xmm0, %xmm0
-; AVX1-NEXT:    vpinsrw $2, %ecx, %xmm0, %xmm0
-; AVX1-NEXT:    vpinsrw $3, %r14d, %xmm0, %xmm0
-; AVX1-NEXT:    vpinsrw $4, %r11d, %xmm0, %xmm0
-; AVX1-NEXT:    vpinsrw $5, %r10d, %xmm0, %xmm0
-; AVX1-NEXT:    vpinsrw $6, %r9d, %xmm0, %xmm0
-; AVX1-NEXT:    vpinsrw $7, %r8d, %xmm0, %xmm0
-; AVX1-NEXT:    vpsrlw $8, %xmm0, %xmm0
-; AVX1-NEXT:    vpackuswb %xmm0, %xmm0, %xmm0
+; AVX1-NEXT:    vmovq {{.*#+}} xmm1 = mem[0],zero
+; AVX1-NEXT:    vpaddsb %xmm1, %xmm0, %xmm0
 ; AVX1-NEXT:    vmovq %xmm0, (%rdx)
-; AVX1-NEXT:    popq %rbx
-; AVX1-NEXT:    popq %r14
-; AVX1-NEXT:    popq %rbp
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: v8i8:
 ; AVX2:       # %bb.0:
-; AVX2-NEXT:    pushq %rbp
-; AVX2-NEXT:    pushq %r14
-; AVX2-NEXT:    pushq %rbx
-; AVX2-NEXT:    vmovq {{.*#+}} xmm1 = mem[0],zero
 ; AVX2-NEXT:    vmovq {{.*#+}} xmm0 = mem[0],zero
-; AVX2-NEXT:    vpxor %xmm2, %xmm2, %xmm2
-; AVX2-NEXT:    vpunpcklbw {{.*#+}} xmm0 = xmm2[0],xmm0[0],xmm2[1],xmm0[1],xmm2[2],xmm0[2],xmm2[3],xmm0[3],xmm2[4],xmm0[4],xmm2[5],xmm0[5],xmm2[6],xmm0[6],xmm2[7],xmm0[7]
-; AVX2-NEXT:    vpextrw $7, %xmm0, %eax
-; AVX2-NEXT:    vpunpcklbw {{.*#+}} xmm1 = xmm2[0],xmm1[0],xmm2[1],xmm1[1],xmm2[2],xmm1[2],xmm2[3],xmm1[3],xmm2[4],xmm1[4],xmm2[5],xmm1[5],xmm2[6],xmm1[6],xmm2[7],xmm1[7]
-; AVX2-NEXT:    vpextrw $7, %xmm1, %r8d
-; AVX2-NEXT:    xorl %ecx, %ecx
-; AVX2-NEXT:    movl %r8d, %esi
-; AVX2-NEXT:    addw %ax, %si
-; AVX2-NEXT:    setns %cl
-; AVX2-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX2-NEXT:    addw %ax, %r8w
-; AVX2-NEXT:    cmovol %ecx, %r8d
-; AVX2-NEXT:    vpextrw $6, %xmm0, %eax
-; AVX2-NEXT:    vpextrw $6, %xmm1, %r9d
-; AVX2-NEXT:    xorl %ecx, %ecx
-; AVX2-NEXT:    movl %r9d, %esi
-; AVX2-NEXT:    addw %ax, %si
-; AVX2-NEXT:    setns %cl
-; AVX2-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX2-NEXT:    addw %ax, %r9w
-; AVX2-NEXT:    cmovol %ecx, %r9d
-; AVX2-NEXT:    vpextrw $5, %xmm0, %eax
-; AVX2-NEXT:    vpextrw $5, %xmm1, %r10d
-; AVX2-NEXT:    xorl %ecx, %ecx
-; AVX2-NEXT:    movl %r10d, %edi
-; AVX2-NEXT:    addw %ax, %di
-; AVX2-NEXT:    setns %cl
-; AVX2-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX2-NEXT:    addw %ax, %r10w
-; AVX2-NEXT:    cmovol %ecx, %r10d
-; AVX2-NEXT:    vpextrw $4, %xmm0, %eax
-; AVX2-NEXT:    vpextrw $4, %xmm1, %r11d
-; AVX2-NEXT:    xorl %ecx, %ecx
-; AVX2-NEXT:    movl %r11d, %esi
-; AVX2-NEXT:    addw %ax, %si
-; AVX2-NEXT:    setns %cl
-; AVX2-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX2-NEXT:    addw %ax, %r11w
-; AVX2-NEXT:    cmovol %ecx, %r11d
-; AVX2-NEXT:    vpextrw $3, %xmm0, %ecx
-; AVX2-NEXT:    vpextrw $3, %xmm1, %r14d
-; AVX2-NEXT:    xorl %esi, %esi
-; AVX2-NEXT:    movl %r14d, %edi
-; AVX2-NEXT:    addw %cx, %di
-; AVX2-NEXT:    setns %sil
-; AVX2-NEXT:    addl $32767, %esi # imm = 0x7FFF
-; AVX2-NEXT:    addw %cx, %r14w
-; AVX2-NEXT:    cmovol %esi, %r14d
-; AVX2-NEXT:    vpextrw $2, %xmm0, %esi
-; AVX2-NEXT:    vpextrw $2, %xmm1, %ecx
-; AVX2-NEXT:    xorl %edi, %edi
-; AVX2-NEXT:    movl %ecx, %ebx
-; AVX2-NEXT:    addw %si, %bx
-; AVX2-NEXT:    setns %dil
-; AVX2-NEXT:    addl $32767, %edi # imm = 0x7FFF
-; AVX2-NEXT:    addw %si, %cx
-; AVX2-NEXT:    cmovol %edi, %ecx
-; AVX2-NEXT:    vmovd %xmm0, %esi
-; AVX2-NEXT:    vmovd %xmm1, %edi
-; AVX2-NEXT:    xorl %ebx, %ebx
-; AVX2-NEXT:    movl %edi, %ebp
-; AVX2-NEXT:    addw %si, %bp
-; AVX2-NEXT:    setns %bl
-; AVX2-NEXT:    addl $32767, %ebx # imm = 0x7FFF
-; AVX2-NEXT:    addw %si, %di
-; AVX2-NEXT:    cmovol %ebx, %edi
-; AVX2-NEXT:    vpextrw $1, %xmm0, %esi
-; AVX2-NEXT:    vpextrw $1, %xmm1, %ebx
-; AVX2-NEXT:    xorl %ebp, %ebp
-; AVX2-NEXT:    movl %ebx, %eax
-; AVX2-NEXT:    addw %si, %ax
-; AVX2-NEXT:    setns %bpl
-; AVX2-NEXT:    addl $32767, %ebp # imm = 0x7FFF
-; AVX2-NEXT:    addw %si, %bx
-; AVX2-NEXT:    cmovol %ebp, %ebx
-; AVX2-NEXT:    vmovd %edi, %xmm0
-; AVX2-NEXT:    vpinsrw $1, %ebx, %xmm0, %xmm0
-; AVX2-NEXT:    vpinsrw $2, %ecx, %xmm0, %xmm0
-; AVX2-NEXT:    vpinsrw $3, %r14d, %xmm0, %xmm0
-; AVX2-NEXT:    vpinsrw $4, %r11d, %xmm0, %xmm0
-; AVX2-NEXT:    vpinsrw $5, %r10d, %xmm0, %xmm0
-; AVX2-NEXT:    vpinsrw $6, %r9d, %xmm0, %xmm0
-; AVX2-NEXT:    vpinsrw $7, %r8d, %xmm0, %xmm0
-; AVX2-NEXT:    vpsrlw $8, %xmm0, %xmm0
-; AVX2-NEXT:    vpackuswb %xmm0, %xmm0, %xmm0
+; AVX2-NEXT:    vmovq {{.*#+}} xmm1 = mem[0],zero
+; AVX2-NEXT:    vpaddsb %xmm1, %xmm0, %xmm0
 ; AVX2-NEXT:    vmovq %xmm0, (%rdx)
-; AVX2-NEXT:    popq %rbx
-; AVX2-NEXT:    popq %r14
-; AVX2-NEXT:    popq %rbp
 ; AVX2-NEXT:    retq
 ;
 ; AVX512-LABEL: v8i8:
 ; AVX512:       # %bb.0:
-; AVX512-NEXT:    pushq %rbp
-; AVX512-NEXT:    pushq %r14
-; AVX512-NEXT:    pushq %rbx
-; AVX512-NEXT:    vmovq {{.*#+}} xmm1 = mem[0],zero
 ; AVX512-NEXT:    vmovq {{.*#+}} xmm0 = mem[0],zero
-; AVX512-NEXT:    vpxor %xmm2, %xmm2, %xmm2
-; AVX512-NEXT:    vpunpcklbw {{.*#+}} xmm0 = xmm2[0],xmm0[0],xmm2[1],xmm0[1],xmm2[2],xmm0[2],xmm2[3],xmm0[3],xmm2[4],xmm0[4],xmm2[5],xmm0[5],xmm2[6],xmm0[6],xmm2[7],xmm0[7]
-; AVX512-NEXT:    vpextrw $7, %xmm0, %eax
-; AVX512-NEXT:    vpunpcklbw {{.*#+}} xmm1 = xmm2[0],xmm1[0],xmm2[1],xmm1[1],xmm2[2],xmm1[2],xmm2[3],xmm1[3],xmm2[4],xmm1[4],xmm2[5],xmm1[5],xmm2[6],xmm1[6],xmm2[7],xmm1[7]
-; AVX512-NEXT:    vpextrw $7, %xmm1, %r8d
-; AVX512-NEXT:    xorl %ecx, %ecx
-; AVX512-NEXT:    movl %r8d, %esi
-; AVX512-NEXT:    addw %ax, %si
-; AVX512-NEXT:    setns %cl
-; AVX512-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX512-NEXT:    addw %ax, %r8w
-; AVX512-NEXT:    cmovol %ecx, %r8d
-; AVX512-NEXT:    vpextrw $6, %xmm0, %eax
-; AVX512-NEXT:    vpextrw $6, %xmm1, %r9d
-; AVX512-NEXT:    xorl %ecx, %ecx
-; AVX512-NEXT:    movl %r9d, %esi
-; AVX512-NEXT:    addw %ax, %si
-; AVX512-NEXT:    setns %cl
-; AVX512-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX512-NEXT:    addw %ax, %r9w
-; AVX512-NEXT:    cmovol %ecx, %r9d
-; AVX512-NEXT:    vpextrw $5, %xmm0, %eax
-; AVX512-NEXT:    vpextrw $5, %xmm1, %r10d
-; AVX512-NEXT:    xorl %ecx, %ecx
-; AVX512-NEXT:    movl %r10d, %edi
-; AVX512-NEXT:    addw %ax, %di
-; AVX512-NEXT:    setns %cl
-; AVX512-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX512-NEXT:    addw %ax, %r10w
-; AVX512-NEXT:    cmovol %ecx, %r10d
-; AVX512-NEXT:    vpextrw $4, %xmm0, %eax
-; AVX512-NEXT:    vpextrw $4, %xmm1, %r11d
-; AVX512-NEXT:    xorl %ecx, %ecx
-; AVX512-NEXT:    movl %r11d, %esi
-; AVX512-NEXT:    addw %ax, %si
-; AVX512-NEXT:    setns %cl
-; AVX512-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX512-NEXT:    addw %ax, %r11w
-; AVX512-NEXT:    cmovol %ecx, %r11d
-; AVX512-NEXT:    vpextrw $3, %xmm0, %ecx
-; AVX512-NEXT:    vpextrw $3, %xmm1, %r14d
-; AVX512-NEXT:    xorl %esi, %esi
-; AVX512-NEXT:    movl %r14d, %edi
-; AVX512-NEXT:    addw %cx, %di
-; AVX512-NEXT:    setns %sil
-; AVX512-NEXT:    addl $32767, %esi # imm = 0x7FFF
-; AVX512-NEXT:    addw %cx, %r14w
-; AVX512-NEXT:    cmovol %esi, %r14d
-; AVX512-NEXT:    vpextrw $2, %xmm0, %esi
-; AVX512-NEXT:    vpextrw $2, %xmm1, %ecx
-; AVX512-NEXT:    xorl %edi, %edi
-; AVX512-NEXT:    movl %ecx, %ebx
-; AVX512-NEXT:    addw %si, %bx
-; AVX512-NEXT:    setns %dil
-; AVX512-NEXT:    addl $32767, %edi # imm = 0x7FFF
-; AVX512-NEXT:    addw %si, %cx
-; AVX512-NEXT:    cmovol %edi, %ecx
-; AVX512-NEXT:    vmovd %xmm0, %esi
-; AVX512-NEXT:    vmovd %xmm1, %edi
-; AVX512-NEXT:    xorl %ebx, %ebx
-; AVX512-NEXT:    movl %edi, %ebp
-; AVX512-NEXT:    addw %si, %bp
-; AVX512-NEXT:    setns %bl
-; AVX512-NEXT:    addl $32767, %ebx # imm = 0x7FFF
-; AVX512-NEXT:    addw %si, %di
-; AVX512-NEXT:    cmovol %ebx, %edi
-; AVX512-NEXT:    vpextrw $1, %xmm0, %esi
-; AVX512-NEXT:    vpextrw $1, %xmm1, %ebx
-; AVX512-NEXT:    xorl %ebp, %ebp
-; AVX512-NEXT:    movl %ebx, %eax
-; AVX512-NEXT:    addw %si, %ax
-; AVX512-NEXT:    setns %bpl
-; AVX512-NEXT:    addl $32767, %ebp # imm = 0x7FFF
-; AVX512-NEXT:    addw %si, %bx
-; AVX512-NEXT:    cmovol %ebp, %ebx
-; AVX512-NEXT:    vmovd %edi, %xmm0
-; AVX512-NEXT:    vpinsrw $1, %ebx, %xmm0, %xmm0
-; AVX512-NEXT:    vpinsrw $2, %ecx, %xmm0, %xmm0
-; AVX512-NEXT:    vpinsrw $3, %r14d, %xmm0, %xmm0
-; AVX512-NEXT:    vpinsrw $4, %r11d, %xmm0, %xmm0
-; AVX512-NEXT:    vpinsrw $5, %r10d, %xmm0, %xmm0
-; AVX512-NEXT:    vpinsrw $6, %r9d, %xmm0, %xmm0
-; AVX512-NEXT:    vpinsrw $7, %r8d, %xmm0, %xmm0
-; AVX512-NEXT:    vpsrlw $8, %xmm0, %xmm0
+; AVX512-NEXT:    vmovq {{.*#+}} xmm1 = mem[0],zero
+; AVX512-NEXT:    vpaddsb %xmm1, %xmm0, %xmm0
+; AVX512-NEXT:    vpmovzxbw {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
 ; AVX512-NEXT:    vpmovwb %xmm0, (%rdx)
-; AVX512-NEXT:    popq %rbx
-; AVX512-NEXT:    popq %r14
-; AVX512-NEXT:    popq %rbp
 ; AVX512-NEXT:    retq
   %x = load <8 x i8>, <8 x i8>* %px
   %y = load <8 x i8>, <8 x i8>* %py
@@ -14100,329 +236,36 @@ define void @v8i8(<8 x i8>* %px, <8 x i8
 }
 
 define void @v4i8(<4 x i8>* %px, <4 x i8>* %py, <4 x i8>* %pz) nounwind {
-; SSE2-LABEL: v4i8:
-; SSE2:       # %bb.0:
-; SSE2-NEXT:    movd {{.*#+}} xmm0 = mem[0],zero,zero,zero
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7]
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3]
-; SSE2-NEXT:    movd {{.*#+}} xmm1 = mem[0],zero,zero,zero
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3],xmm1[4],xmm0[4],xmm1[5],xmm0[5],xmm1[6],xmm0[6],xmm1[7],xmm0[7]
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3]
-; SSE2-NEXT:    pslld $24, %xmm1
-; SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm1[1,1,2,3]
-; SSE2-NEXT:    movd %xmm2, %ecx
-; SSE2-NEXT:    pslld $24, %xmm0
-; SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm0[1,1,2,3]
-; SSE2-NEXT:    movd %xmm2, %r8d
-; SSE2-NEXT:    xorl %esi, %esi
-; SSE2-NEXT:    movl %r8d, %edi
-; SSE2-NEXT:    addl %ecx, %edi
-; SSE2-NEXT:    setns %sil
-; SSE2-NEXT:    addl $2147483647, %esi # imm = 0x7FFFFFFF
-; SSE2-NEXT:    addl %ecx, %r8d
-; SSE2-NEXT:    cmovol %esi, %r8d
-; SSE2-NEXT:    movd %xmm1, %esi
-; SSE2-NEXT:    movd %xmm0, %r10d
-; SSE2-NEXT:    xorl %edi, %edi
-; SSE2-NEXT:    movl %r10d, %ecx
-; SSE2-NEXT:    addl %esi, %ecx
-; SSE2-NEXT:    setns %dil
-; SSE2-NEXT:    addl $2147483647, %edi # imm = 0x7FFFFFFF
-; SSE2-NEXT:    addl %esi, %r10d
-; SSE2-NEXT:    cmovol %edi, %r10d
-; SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm1[2,3,0,1]
-; SSE2-NEXT:    movd %xmm2, %r9d
-; SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm0[2,3,0,1]
-; SSE2-NEXT:    movd %xmm2, %ecx
-; SSE2-NEXT:    xorl %edi, %edi
-; SSE2-NEXT:    movl %ecx, %esi
-; SSE2-NEXT:    addl %r9d, %esi
-; SSE2-NEXT:    setns %dil
-; SSE2-NEXT:    addl $2147483647, %edi # imm = 0x7FFFFFFF
-; SSE2-NEXT:    addl %r9d, %ecx
-; SSE2-NEXT:    cmovol %edi, %ecx
-; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[3,1,2,3]
-; SSE2-NEXT:    movd %xmm1, %r9d
-; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[3,1,2,3]
-; SSE2-NEXT:    movd %xmm0, %eax
-; SSE2-NEXT:    xorl %esi, %esi
-; SSE2-NEXT:    movl %eax, %edi
-; SSE2-NEXT:    addl %r9d, %edi
-; SSE2-NEXT:    setns %sil
-; SSE2-NEXT:    addl $2147483647, %esi # imm = 0x7FFFFFFF
-; SSE2-NEXT:    addl %r9d, %eax
-; SSE2-NEXT:    cmovol %esi, %eax
-; SSE2-NEXT:    movd %eax, %xmm0
-; SSE2-NEXT:    movd %ecx, %xmm1
-; SSE2-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1]
-; SSE2-NEXT:    movd %r10d, %xmm0
-; SSE2-NEXT:    movd %r8d, %xmm2
-; SSE2-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1]
-; SSE2-NEXT:    punpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm1[0]
-; SSE2-NEXT:    psrld $24, %xmm0
-; SSE2-NEXT:    packuswb %xmm0, %xmm0
-; SSE2-NEXT:    packuswb %xmm0, %xmm0
-; SSE2-NEXT:    movd %xmm0, (%rdx)
-; SSE2-NEXT:    retq
-;
-; SSSE3-LABEL: v4i8:
-; SSSE3:       # %bb.0:
-; SSSE3-NEXT:    movd {{.*#+}} xmm0 = mem[0],zero,zero,zero
-; SSSE3-NEXT:    movd {{.*#+}} xmm1 = mem[0],zero,zero,zero
-; SSSE3-NEXT:    movdqa {{.*#+}} xmm2 = [255,255,255,0,255,255,255,1,255,255,255,2,255,255,255,3]
-; SSSE3-NEXT:    pshufb %xmm2, %xmm1
-; SSSE3-NEXT:    pshufd {{.*#+}} xmm3 = xmm1[1,1,2,3]
-; SSSE3-NEXT:    movd %xmm3, %ecx
-; SSSE3-NEXT:    pshufb %xmm2, %xmm0
-; SSSE3-NEXT:    pshufd {{.*#+}} xmm2 = xmm0[1,1,2,3]
-; SSSE3-NEXT:    movd %xmm2, %r8d
-; SSSE3-NEXT:    xorl %esi, %esi
-; SSSE3-NEXT:    movl %r8d, %edi
-; SSSE3-NEXT:    addl %ecx, %edi
-; SSSE3-NEXT:    setns %sil
-; SSSE3-NEXT:    addl $2147483647, %esi # imm = 0x7FFFFFFF
-; SSSE3-NEXT:    addl %ecx, %r8d
-; SSSE3-NEXT:    cmovol %esi, %r8d
-; SSSE3-NEXT:    movd %xmm1, %esi
-; SSSE3-NEXT:    movd %xmm0, %r10d
-; SSSE3-NEXT:    xorl %edi, %edi
-; SSSE3-NEXT:    movl %r10d, %ecx
-; SSSE3-NEXT:    addl %esi, %ecx
-; SSSE3-NEXT:    setns %dil
-; SSSE3-NEXT:    addl $2147483647, %edi # imm = 0x7FFFFFFF
-; SSSE3-NEXT:    addl %esi, %r10d
-; SSSE3-NEXT:    cmovol %edi, %r10d
-; SSSE3-NEXT:    pshufd {{.*#+}} xmm2 = xmm1[2,3,0,1]
-; SSSE3-NEXT:    movd %xmm2, %r9d
-; SSSE3-NEXT:    pshufd {{.*#+}} xmm2 = xmm0[2,3,0,1]
-; SSSE3-NEXT:    movd %xmm2, %ecx
-; SSSE3-NEXT:    xorl %edi, %edi
-; SSSE3-NEXT:    movl %ecx, %esi
-; SSSE3-NEXT:    addl %r9d, %esi
-; SSSE3-NEXT:    setns %dil
-; SSSE3-NEXT:    addl $2147483647, %edi # imm = 0x7FFFFFFF
-; SSSE3-NEXT:    addl %r9d, %ecx
-; SSSE3-NEXT:    cmovol %edi, %ecx
-; SSSE3-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[3,1,2,3]
-; SSSE3-NEXT:    movd %xmm1, %r9d
-; SSSE3-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[3,1,2,3]
-; SSSE3-NEXT:    movd %xmm0, %eax
-; SSSE3-NEXT:    xorl %esi, %esi
-; SSSE3-NEXT:    movl %eax, %edi
-; SSSE3-NEXT:    addl %r9d, %edi
-; SSSE3-NEXT:    setns %sil
-; SSSE3-NEXT:    addl $2147483647, %esi # imm = 0x7FFFFFFF
-; SSSE3-NEXT:    addl %r9d, %eax
-; SSSE3-NEXT:    cmovol %esi, %eax
-; SSSE3-NEXT:    movd %eax, %xmm0
-; SSSE3-NEXT:    movd %ecx, %xmm1
-; SSSE3-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1]
-; SSSE3-NEXT:    movd %r10d, %xmm0
-; SSSE3-NEXT:    movd %r8d, %xmm2
-; SSSE3-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1]
-; SSSE3-NEXT:    punpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm1[0]
-; SSSE3-NEXT:    pshufb {{.*#+}} xmm0 = xmm0[3,7,11,15,u,u,u,u,u,u,u,u,u,u,u,u]
-; SSSE3-NEXT:    movd %xmm0, (%rdx)
-; SSSE3-NEXT:    retq
-;
-; SSE41-LABEL: v4i8:
-; SSE41:       # %bb.0:
-; SSE41-NEXT:    pmovzxbd {{.*#+}} xmm0 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
-; SSE41-NEXT:    pmovzxbd {{.*#+}} xmm1 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
-; SSE41-NEXT:    pslld $24, %xmm1
-; SSE41-NEXT:    pextrd $3, %xmm1, %ecx
-; SSE41-NEXT:    pslld $24, %xmm0
-; SSE41-NEXT:    pextrd $3, %xmm0, %r8d
-; SSE41-NEXT:    xorl %esi, %esi
-; SSE41-NEXT:    movl %r8d, %edi
-; SSE41-NEXT:    addl %ecx, %edi
-; SSE41-NEXT:    setns %sil
-; SSE41-NEXT:    addl $2147483647, %esi # imm = 0x7FFFFFFF
-; SSE41-NEXT:    addl %ecx, %r8d
-; SSE41-NEXT:    cmovol %esi, %r8d
-; SSE41-NEXT:    pextrd $2, %xmm1, %esi
-; SSE41-NEXT:    pextrd $2, %xmm0, %r10d
-; SSE41-NEXT:    xorl %edi, %edi
-; SSE41-NEXT:    movl %r10d, %ecx
-; SSE41-NEXT:    addl %esi, %ecx
-; SSE41-NEXT:    setns %dil
-; SSE41-NEXT:    addl $2147483647, %edi # imm = 0x7FFFFFFF
-; SSE41-NEXT:    addl %esi, %r10d
-; SSE41-NEXT:    cmovol %edi, %r10d
-; SSE41-NEXT:    movd %xmm1, %r9d
-; SSE41-NEXT:    movd %xmm0, %ecx
-; SSE41-NEXT:    xorl %edi, %edi
-; SSE41-NEXT:    movl %ecx, %esi
-; SSE41-NEXT:    addl %r9d, %esi
-; SSE41-NEXT:    setns %dil
-; SSE41-NEXT:    addl $2147483647, %edi # imm = 0x7FFFFFFF
-; SSE41-NEXT:    addl %r9d, %ecx
-; SSE41-NEXT:    cmovol %edi, %ecx
-; SSE41-NEXT:    pextrd $1, %xmm1, %r9d
-; SSE41-NEXT:    pextrd $1, %xmm0, %eax
-; SSE41-NEXT:    xorl %esi, %esi
-; SSE41-NEXT:    movl %eax, %edi
-; SSE41-NEXT:    addl %r9d, %edi
-; SSE41-NEXT:    setns %sil
-; SSE41-NEXT:    addl $2147483647, %esi # imm = 0x7FFFFFFF
-; SSE41-NEXT:    addl %r9d, %eax
-; SSE41-NEXT:    cmovol %esi, %eax
-; SSE41-NEXT:    movd %ecx, %xmm0
-; SSE41-NEXT:    pinsrd $1, %eax, %xmm0
-; SSE41-NEXT:    pinsrd $2, %r10d, %xmm0
-; SSE41-NEXT:    pinsrd $3, %r8d, %xmm0
-; SSE41-NEXT:    pshufb {{.*#+}} xmm0 = xmm0[3,7,11,15,u,u,u,u,u,u,u,u,u,u,u,u]
-; SSE41-NEXT:    movd %xmm0, (%rdx)
-; SSE41-NEXT:    retq
+; SSE-LABEL: v4i8:
+; SSE:       # %bb.0:
+; SSE-NEXT:    movd {{.*#+}} xmm0 = mem[0],zero,zero,zero
+; SSE-NEXT:    movd {{.*#+}} xmm1 = mem[0],zero,zero,zero
+; SSE-NEXT:    paddsb %xmm0, %xmm1
+; SSE-NEXT:    movd %xmm1, (%rdx)
+; SSE-NEXT:    retq
 ;
 ; AVX1-LABEL: v4i8:
 ; AVX1:       # %bb.0:
-; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm1 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
-; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm0 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
-; AVX1-NEXT:    vpslld $24, %xmm0, %xmm0
-; AVX1-NEXT:    vpextrd $3, %xmm0, %ecx
-; AVX1-NEXT:    vpslld $24, %xmm1, %xmm1
-; AVX1-NEXT:    vpextrd $3, %xmm1, %r9d
-; AVX1-NEXT:    xorl %esi, %esi
-; AVX1-NEXT:    movl %r9d, %edi
-; AVX1-NEXT:    addl %ecx, %edi
-; AVX1-NEXT:    setns %sil
-; AVX1-NEXT:    addl $2147483647, %esi # imm = 0x7FFFFFFF
-; AVX1-NEXT:    addl %ecx, %r9d
-; AVX1-NEXT:    cmovol %esi, %r9d
-; AVX1-NEXT:    vpextrd $2, %xmm0, %r8d
-; AVX1-NEXT:    vpextrd $2, %xmm1, %r10d
-; AVX1-NEXT:    xorl %edi, %edi
-; AVX1-NEXT:    movl %r10d, %ecx
-; AVX1-NEXT:    addl %r8d, %ecx
-; AVX1-NEXT:    setns %dil
-; AVX1-NEXT:    addl $2147483647, %edi # imm = 0x7FFFFFFF
-; AVX1-NEXT:    addl %r8d, %r10d
-; AVX1-NEXT:    cmovol %edi, %r10d
-; AVX1-NEXT:    vmovd %xmm0, %r8d
-; AVX1-NEXT:    vmovd %xmm1, %eax
-; AVX1-NEXT:    xorl %ecx, %ecx
-; AVX1-NEXT:    movl %eax, %edi
-; AVX1-NEXT:    addl %r8d, %edi
-; AVX1-NEXT:    setns %cl
-; AVX1-NEXT:    addl $2147483647, %ecx # imm = 0x7FFFFFFF
-; AVX1-NEXT:    addl %r8d, %eax
-; AVX1-NEXT:    cmovol %ecx, %eax
-; AVX1-NEXT:    vpextrd $1, %xmm0, %r8d
-; AVX1-NEXT:    vpextrd $1, %xmm1, %esi
-; AVX1-NEXT:    xorl %ecx, %ecx
-; AVX1-NEXT:    movl %esi, %edi
-; AVX1-NEXT:    addl %r8d, %edi
-; AVX1-NEXT:    setns %cl
-; AVX1-NEXT:    addl $2147483647, %ecx # imm = 0x7FFFFFFF
-; AVX1-NEXT:    addl %r8d, %esi
-; AVX1-NEXT:    cmovol %ecx, %esi
-; AVX1-NEXT:    vmovd %eax, %xmm0
-; AVX1-NEXT:    vpinsrd $1, %esi, %xmm0, %xmm0
-; AVX1-NEXT:    vpinsrd $2, %r10d, %xmm0, %xmm0
-; AVX1-NEXT:    vpinsrd $3, %r9d, %xmm0, %xmm0
-; AVX1-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[3,7,11,15,u,u,u,u,u,u,u,u,u,u,u,u]
+; AVX1-NEXT:    vmovd {{.*#+}} xmm0 = mem[0],zero,zero,zero
+; AVX1-NEXT:    vmovd {{.*#+}} xmm1 = mem[0],zero,zero,zero
+; AVX1-NEXT:    vpaddsb %xmm1, %xmm0, %xmm0
 ; AVX1-NEXT:    vmovd %xmm0, (%rdx)
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: v4i8:
 ; AVX2:       # %bb.0:
-; AVX2-NEXT:    vpmovzxbd {{.*#+}} xmm1 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
-; AVX2-NEXT:    vpmovzxbd {{.*#+}} xmm0 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
-; AVX2-NEXT:    vpslld $24, %xmm0, %xmm0
-; AVX2-NEXT:    vpextrd $3, %xmm0, %ecx
-; AVX2-NEXT:    vpslld $24, %xmm1, %xmm1
-; AVX2-NEXT:    vpextrd $3, %xmm1, %r9d
-; AVX2-NEXT:    xorl %esi, %esi
-; AVX2-NEXT:    movl %r9d, %edi
-; AVX2-NEXT:    addl %ecx, %edi
-; AVX2-NEXT:    setns %sil
-; AVX2-NEXT:    addl $2147483647, %esi # imm = 0x7FFFFFFF
-; AVX2-NEXT:    addl %ecx, %r9d
-; AVX2-NEXT:    cmovol %esi, %r9d
-; AVX2-NEXT:    vpextrd $2, %xmm0, %r8d
-; AVX2-NEXT:    vpextrd $2, %xmm1, %r10d
-; AVX2-NEXT:    xorl %edi, %edi
-; AVX2-NEXT:    movl %r10d, %ecx
-; AVX2-NEXT:    addl %r8d, %ecx
-; AVX2-NEXT:    setns %dil
-; AVX2-NEXT:    addl $2147483647, %edi # imm = 0x7FFFFFFF
-; AVX2-NEXT:    addl %r8d, %r10d
-; AVX2-NEXT:    cmovol %edi, %r10d
-; AVX2-NEXT:    vmovd %xmm0, %r8d
-; AVX2-NEXT:    vmovd %xmm1, %eax
-; AVX2-NEXT:    xorl %ecx, %ecx
-; AVX2-NEXT:    movl %eax, %edi
-; AVX2-NEXT:    addl %r8d, %edi
-; AVX2-NEXT:    setns %cl
-; AVX2-NEXT:    addl $2147483647, %ecx # imm = 0x7FFFFFFF
-; AVX2-NEXT:    addl %r8d, %eax
-; AVX2-NEXT:    cmovol %ecx, %eax
-; AVX2-NEXT:    vpextrd $1, %xmm0, %r8d
-; AVX2-NEXT:    vpextrd $1, %xmm1, %esi
-; AVX2-NEXT:    xorl %ecx, %ecx
-; AVX2-NEXT:    movl %esi, %edi
-; AVX2-NEXT:    addl %r8d, %edi
-; AVX2-NEXT:    setns %cl
-; AVX2-NEXT:    addl $2147483647, %ecx # imm = 0x7FFFFFFF
-; AVX2-NEXT:    addl %r8d, %esi
-; AVX2-NEXT:    cmovol %ecx, %esi
-; AVX2-NEXT:    vmovd %eax, %xmm0
-; AVX2-NEXT:    vpinsrd $1, %esi, %xmm0, %xmm0
-; AVX2-NEXT:    vpinsrd $2, %r10d, %xmm0, %xmm0
-; AVX2-NEXT:    vpinsrd $3, %r9d, %xmm0, %xmm0
-; AVX2-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[3,7,11,15,u,u,u,u,u,u,u,u,u,u,u,u]
+; AVX2-NEXT:    vmovd {{.*#+}} xmm0 = mem[0],zero,zero,zero
+; AVX2-NEXT:    vmovd {{.*#+}} xmm1 = mem[0],zero,zero,zero
+; AVX2-NEXT:    vpaddsb %xmm1, %xmm0, %xmm0
 ; AVX2-NEXT:    vmovd %xmm0, (%rdx)
 ; AVX2-NEXT:    retq
 ;
 ; AVX512-LABEL: v4i8:
 ; AVX512:       # %bb.0:
-; AVX512-NEXT:    vmovd {{.*#+}} xmm1 = mem[0],zero,zero,zero
 ; AVX512-NEXT:    vmovd {{.*#+}} xmm0 = mem[0],zero,zero,zero
-; AVX512-NEXT:    vmovdqa {{.*#+}} xmm2 = [255,255,255,0,255,255,255,1,255,255,255,2,255,255,255,3]
-; AVX512-NEXT:    vpshufb %xmm2, %xmm0, %xmm0
-; AVX512-NEXT:    vpextrd $3, %xmm0, %ecx
-; AVX512-NEXT:    vpshufb %xmm2, %xmm1, %xmm1
-; AVX512-NEXT:    vpextrd $3, %xmm1, %r9d
-; AVX512-NEXT:    xorl %esi, %esi
-; AVX512-NEXT:    movl %r9d, %edi
-; AVX512-NEXT:    addl %ecx, %edi
-; AVX512-NEXT:    setns %sil
-; AVX512-NEXT:    addl $2147483647, %esi # imm = 0x7FFFFFFF
-; AVX512-NEXT:    addl %ecx, %r9d
-; AVX512-NEXT:    cmovol %esi, %r9d
-; AVX512-NEXT:    vpextrd $2, %xmm0, %r8d
-; AVX512-NEXT:    vpextrd $2, %xmm1, %r10d
-; AVX512-NEXT:    xorl %edi, %edi
-; AVX512-NEXT:    movl %r10d, %ecx
-; AVX512-NEXT:    addl %r8d, %ecx
-; AVX512-NEXT:    setns %dil
-; AVX512-NEXT:    addl $2147483647, %edi # imm = 0x7FFFFFFF
-; AVX512-NEXT:    addl %r8d, %r10d
-; AVX512-NEXT:    cmovol %edi, %r10d
-; AVX512-NEXT:    vmovd %xmm0, %r8d
-; AVX512-NEXT:    vmovd %xmm1, %eax
-; AVX512-NEXT:    xorl %ecx, %ecx
-; AVX512-NEXT:    movl %eax, %edi
-; AVX512-NEXT:    addl %r8d, %edi
-; AVX512-NEXT:    setns %cl
-; AVX512-NEXT:    addl $2147483647, %ecx # imm = 0x7FFFFFFF
-; AVX512-NEXT:    addl %r8d, %eax
-; AVX512-NEXT:    cmovol %ecx, %eax
-; AVX512-NEXT:    vpextrd $1, %xmm0, %r8d
-; AVX512-NEXT:    vpextrd $1, %xmm1, %esi
-; AVX512-NEXT:    xorl %ecx, %ecx
-; AVX512-NEXT:    movl %esi, %edi
-; AVX512-NEXT:    addl %r8d, %edi
-; AVX512-NEXT:    setns %cl
-; AVX512-NEXT:    addl $2147483647, %ecx # imm = 0x7FFFFFFF
-; AVX512-NEXT:    addl %r8d, %esi
-; AVX512-NEXT:    cmovol %ecx, %esi
-; AVX512-NEXT:    vmovd %eax, %xmm0
-; AVX512-NEXT:    vpinsrd $1, %esi, %xmm0, %xmm0
-; AVX512-NEXT:    vpinsrd $2, %r10d, %xmm0, %xmm0
-; AVX512-NEXT:    vpinsrd $3, %r9d, %xmm0, %xmm0
-; AVX512-NEXT:    vpsrld $24, %xmm0, %xmm0
+; AVX512-NEXT:    vmovd {{.*#+}} xmm1 = mem[0],zero,zero,zero
+; AVX512-NEXT:    vpaddsb %xmm1, %xmm0, %xmm0
+; AVX512-NEXT:    vpmovzxbd {{.*#+}} xmm0 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero,xmm0[2],zero,zero,zero,xmm0[3],zero,zero,zero
 ; AVX512-NEXT:    vpmovdb %xmm0, (%rdx)
 ; AVX512-NEXT:    retq
   %x = load <4 x i8>, <4 x i8>* %px
@@ -14437,45 +280,10 @@ define void @v2i8(<2 x i8>* %px, <2 x i8
 ; SSE2:       # %bb.0:
 ; SSE2-NEXT:    movzwl (%rdi), %eax
 ; SSE2-NEXT:    movd %eax, %xmm0
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7]
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3]
-; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[0,1,1,3]
 ; SSE2-NEXT:    movzwl (%rsi), %eax
 ; SSE2-NEXT:    movd %eax, %xmm1
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3],xmm1[4],xmm0[4],xmm1[5],xmm0[5],xmm1[6],xmm0[6],xmm1[7],xmm0[7]
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3]
-; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[0,1,1,3]
-; SSE2-NEXT:    psllq $56, %xmm1
-; SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm1[2,3,0,1]
-; SSE2-NEXT:    movq %xmm2, %rax
-; SSE2-NEXT:    psllq $56, %xmm0
-; SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm0[2,3,0,1]
-; SSE2-NEXT:    movq %xmm2, %rcx
-; SSE2-NEXT:    xorl %esi, %esi
-; SSE2-NEXT:    movq %rcx, %rdi
-; SSE2-NEXT:    addq %rax, %rdi
-; SSE2-NEXT:    setns %sil
-; SSE2-NEXT:    movabsq $9223372036854775807, %r8 # imm = 0x7FFFFFFFFFFFFFFF
-; SSE2-NEXT:    addq %r8, %rsi
-; SSE2-NEXT:    addq %rax, %rcx
-; SSE2-NEXT:    cmovoq %rsi, %rcx
-; SSE2-NEXT:    movq %xmm1, %r9
-; SSE2-NEXT:    movq %xmm0, %rax
-; SSE2-NEXT:    xorl %edi, %edi
-; SSE2-NEXT:    movq %rax, %rsi
-; SSE2-NEXT:    addq %r9, %rsi
-; SSE2-NEXT:    setns %dil
-; SSE2-NEXT:    addq %r8, %rdi
-; SSE2-NEXT:    addq %r9, %rax
-; SSE2-NEXT:    cmovoq %rdi, %rax
-; SSE2-NEXT:    movq %rax, %xmm0
-; SSE2-NEXT:    movq %rcx, %xmm1
-; SSE2-NEXT:    punpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm1[0]
-; SSE2-NEXT:    psrlq $56, %xmm0
-; SSE2-NEXT:    packuswb %xmm0, %xmm0
-; SSE2-NEXT:    packuswb %xmm0, %xmm0
-; SSE2-NEXT:    packuswb %xmm0, %xmm0
-; SSE2-NEXT:    movd %xmm0, %eax
+; SSE2-NEXT:    paddsb %xmm0, %xmm1
+; SSE2-NEXT:    movd %xmm1, %eax
 ; SSE2-NEXT:    movw %ax, (%rdx)
 ; SSE2-NEXT:    retq
 ;
@@ -14485,131 +293,38 @@ define void @v2i8(<2 x i8>* %px, <2 x i8
 ; SSSE3-NEXT:    movd %eax, %xmm0
 ; SSSE3-NEXT:    movzwl (%rsi), %eax
 ; SSSE3-NEXT:    movd %eax, %xmm1
-; SSSE3-NEXT:    movdqa {{.*#+}} xmm2 = [255,255,255,255,255,255,255,0,255,255,255,255,255,255,255,1]
-; SSSE3-NEXT:    pshufb %xmm2, %xmm1
-; SSSE3-NEXT:    pshufd {{.*#+}} xmm3 = xmm1[2,3,0,1]
-; SSSE3-NEXT:    movq %xmm3, %rax
-; SSSE3-NEXT:    pshufb %xmm2, %xmm0
-; SSSE3-NEXT:    pshufd {{.*#+}} xmm2 = xmm0[2,3,0,1]
-; SSSE3-NEXT:    movq %xmm2, %rcx
-; SSSE3-NEXT:    xorl %esi, %esi
-; SSSE3-NEXT:    movq %rcx, %rdi
-; SSSE3-NEXT:    addq %rax, %rdi
-; SSSE3-NEXT:    setns %sil
-; SSSE3-NEXT:    movabsq $9223372036854775807, %r8 # imm = 0x7FFFFFFFFFFFFFFF
-; SSSE3-NEXT:    addq %r8, %rsi
-; SSSE3-NEXT:    addq %rax, %rcx
-; SSSE3-NEXT:    cmovoq %rsi, %rcx
-; SSSE3-NEXT:    movq %xmm1, %r9
-; SSSE3-NEXT:    movq %xmm0, %rax
-; SSSE3-NEXT:    xorl %edi, %edi
-; SSSE3-NEXT:    movq %rax, %rsi
-; SSSE3-NEXT:    addq %r9, %rsi
-; SSSE3-NEXT:    setns %dil
-; SSSE3-NEXT:    addq %r8, %rdi
-; SSSE3-NEXT:    addq %r9, %rax
-; SSSE3-NEXT:    cmovoq %rdi, %rax
-; SSSE3-NEXT:    movq %rax, %xmm0
-; SSSE3-NEXT:    movq %rcx, %xmm1
-; SSSE3-NEXT:    punpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm1[0]
-; SSSE3-NEXT:    pshufb {{.*#+}} xmm0 = xmm0[7,15,u,u,u,u,u,u,u,u,u,u,u,u,u,u]
-; SSSE3-NEXT:    movd %xmm0, %eax
+; SSSE3-NEXT:    paddsb %xmm0, %xmm1
+; SSSE3-NEXT:    movd %xmm1, %eax
 ; SSSE3-NEXT:    movw %ax, (%rdx)
 ; SSSE3-NEXT:    retq
 ;
 ; SSE41-LABEL: v2i8:
 ; SSE41:       # %bb.0:
-; SSE41-NEXT:    pmovzxbq {{.*#+}} xmm0 = mem[0],zero,zero,zero,zero,zero,zero,zero,mem[1],zero,zero,zero,zero,zero,zero,zero
-; SSE41-NEXT:    pmovzxbq {{.*#+}} xmm1 = mem[0],zero,zero,zero,zero,zero,zero,zero,mem[1],zero,zero,zero,zero,zero,zero,zero
-; SSE41-NEXT:    psllq $56, %xmm1
-; SSE41-NEXT:    movq %xmm1, %rax
-; SSE41-NEXT:    psllq $56, %xmm0
-; SSE41-NEXT:    movq %xmm0, %rcx
-; SSE41-NEXT:    xorl %esi, %esi
-; SSE41-NEXT:    movq %rcx, %rdi
-; SSE41-NEXT:    addq %rax, %rdi
-; SSE41-NEXT:    setns %sil
-; SSE41-NEXT:    movabsq $9223372036854775807, %r8 # imm = 0x7FFFFFFFFFFFFFFF
-; SSE41-NEXT:    addq %r8, %rsi
-; SSE41-NEXT:    addq %rax, %rcx
-; SSE41-NEXT:    cmovoq %rsi, %rcx
-; SSE41-NEXT:    pextrq $1, %xmm1, %r9
-; SSE41-NEXT:    pextrq $1, %xmm0, %rax
-; SSE41-NEXT:    xorl %edi, %edi
-; SSE41-NEXT:    movq %rax, %rsi
-; SSE41-NEXT:    addq %r9, %rsi
-; SSE41-NEXT:    setns %dil
-; SSE41-NEXT:    addq %r8, %rdi
-; SSE41-NEXT:    addq %r9, %rax
-; SSE41-NEXT:    cmovoq %rdi, %rax
-; SSE41-NEXT:    movq %rax, %xmm0
-; SSE41-NEXT:    movq %rcx, %xmm1
-; SSE41-NEXT:    punpcklqdq {{.*#+}} xmm1 = xmm1[0],xmm0[0]
-; SSE41-NEXT:    pshufb {{.*#+}} xmm1 = xmm1[7,15,u,u,u,u,u,u,u,u,u,u,u,u,u,u]
+; SSE41-NEXT:    movzwl (%rdi), %eax
+; SSE41-NEXT:    movd %eax, %xmm0
+; SSE41-NEXT:    movzwl (%rsi), %eax
+; SSE41-NEXT:    movd %eax, %xmm1
+; SSE41-NEXT:    paddsb %xmm0, %xmm1
 ; SSE41-NEXT:    pextrw $0, %xmm1, (%rdx)
 ; SSE41-NEXT:    retq
 ;
 ; AVX1-LABEL: v2i8:
 ; AVX1:       # %bb.0:
-; AVX1-NEXT:    vpmovzxbq {{.*#+}} xmm0 = mem[0],zero,zero,zero,zero,zero,zero,zero,mem[1],zero,zero,zero,zero,zero,zero,zero
-; AVX1-NEXT:    vpmovzxbq {{.*#+}} xmm1 = mem[0],zero,zero,zero,zero,zero,zero,zero,mem[1],zero,zero,zero,zero,zero,zero,zero
-; AVX1-NEXT:    vpsllq $56, %xmm1, %xmm1
-; AVX1-NEXT:    vmovq %xmm1, %rax
-; AVX1-NEXT:    vpsllq $56, %xmm0, %xmm0
-; AVX1-NEXT:    vmovq %xmm0, %rcx
-; AVX1-NEXT:    xorl %esi, %esi
-; AVX1-NEXT:    movq %rcx, %rdi
-; AVX1-NEXT:    addq %rax, %rdi
-; AVX1-NEXT:    setns %sil
-; AVX1-NEXT:    movabsq $9223372036854775807, %r8 # imm = 0x7FFFFFFFFFFFFFFF
-; AVX1-NEXT:    addq %r8, %rsi
-; AVX1-NEXT:    addq %rax, %rcx
-; AVX1-NEXT:    cmovoq %rsi, %rcx
-; AVX1-NEXT:    vpextrq $1, %xmm1, %r9
-; AVX1-NEXT:    vpextrq $1, %xmm0, %rax
-; AVX1-NEXT:    xorl %edi, %edi
-; AVX1-NEXT:    movq %rax, %rsi
-; AVX1-NEXT:    addq %r9, %rsi
-; AVX1-NEXT:    setns %dil
-; AVX1-NEXT:    addq %r8, %rdi
-; AVX1-NEXT:    addq %r9, %rax
-; AVX1-NEXT:    cmovoq %rdi, %rax
-; AVX1-NEXT:    vmovq %rax, %xmm0
-; AVX1-NEXT:    vmovq %rcx, %xmm1
-; AVX1-NEXT:    vpunpcklqdq {{.*#+}} xmm0 = xmm1[0],xmm0[0]
-; AVX1-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[7,15,u,u,u,u,u,u,u,u,u,u,u,u,u,u]
+; AVX1-NEXT:    movzwl (%rdi), %eax
+; AVX1-NEXT:    vmovd %eax, %xmm0
+; AVX1-NEXT:    movzwl (%rsi), %eax
+; AVX1-NEXT:    vmovd %eax, %xmm1
+; AVX1-NEXT:    vpaddsb %xmm1, %xmm0, %xmm0
 ; AVX1-NEXT:    vpextrw $0, %xmm0, (%rdx)
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: v2i8:
 ; AVX2:       # %bb.0:
-; AVX2-NEXT:    vpmovzxbq {{.*#+}} xmm0 = mem[0],zero,zero,zero,zero,zero,zero,zero,mem[1],zero,zero,zero,zero,zero,zero,zero
-; AVX2-NEXT:    vpmovzxbq {{.*#+}} xmm1 = mem[0],zero,zero,zero,zero,zero,zero,zero,mem[1],zero,zero,zero,zero,zero,zero,zero
-; AVX2-NEXT:    vpsllq $56, %xmm1, %xmm1
-; AVX2-NEXT:    vmovq %xmm1, %rax
-; AVX2-NEXT:    vpsllq $56, %xmm0, %xmm0
-; AVX2-NEXT:    vmovq %xmm0, %rcx
-; AVX2-NEXT:    xorl %esi, %esi
-; AVX2-NEXT:    movq %rcx, %rdi
-; AVX2-NEXT:    addq %rax, %rdi
-; AVX2-NEXT:    setns %sil
-; AVX2-NEXT:    movabsq $9223372036854775807, %r8 # imm = 0x7FFFFFFFFFFFFFFF
-; AVX2-NEXT:    addq %r8, %rsi
-; AVX2-NEXT:    addq %rax, %rcx
-; AVX2-NEXT:    cmovoq %rsi, %rcx
-; AVX2-NEXT:    vpextrq $1, %xmm1, %r9
-; AVX2-NEXT:    vpextrq $1, %xmm0, %rax
-; AVX2-NEXT:    xorl %edi, %edi
-; AVX2-NEXT:    movq %rax, %rsi
-; AVX2-NEXT:    addq %r9, %rsi
-; AVX2-NEXT:    setns %dil
-; AVX2-NEXT:    addq %r8, %rdi
-; AVX2-NEXT:    addq %r9, %rax
-; AVX2-NEXT:    cmovoq %rdi, %rax
-; AVX2-NEXT:    vmovq %rax, %xmm0
-; AVX2-NEXT:    vmovq %rcx, %xmm1
-; AVX2-NEXT:    vpunpcklqdq {{.*#+}} xmm0 = xmm1[0],xmm0[0]
-; AVX2-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[7,15,u,u,u,u,u,u,u,u,u,u,u,u,u,u]
+; AVX2-NEXT:    movzwl (%rdi), %eax
+; AVX2-NEXT:    vmovd %eax, %xmm0
+; AVX2-NEXT:    movzwl (%rsi), %eax
+; AVX2-NEXT:    vmovd %eax, %xmm1
+; AVX2-NEXT:    vpaddsb %xmm1, %xmm0, %xmm0
 ; AVX2-NEXT:    vpextrw $0, %xmm0, (%rdx)
 ; AVX2-NEXT:    retq
 ;
@@ -14619,32 +334,8 @@ define void @v2i8(<2 x i8>* %px, <2 x i8
 ; AVX512-NEXT:    vmovd %eax, %xmm0
 ; AVX512-NEXT:    movzwl (%rsi), %eax
 ; AVX512-NEXT:    vmovd %eax, %xmm1
-; AVX512-NEXT:    vmovdqa {{.*#+}} xmm2 = [255,255,255,255,255,255,255,0,255,255,255,255,255,255,255,1]
-; AVX512-NEXT:    vpshufb %xmm2, %xmm1, %xmm1
-; AVX512-NEXT:    vmovq %xmm1, %rax
-; AVX512-NEXT:    vpshufb %xmm2, %xmm0, %xmm0
-; AVX512-NEXT:    vmovq %xmm0, %rcx
-; AVX512-NEXT:    xorl %esi, %esi
-; AVX512-NEXT:    movq %rcx, %rdi
-; AVX512-NEXT:    addq %rax, %rdi
-; AVX512-NEXT:    setns %sil
-; AVX512-NEXT:    movabsq $9223372036854775807, %r8 # imm = 0x7FFFFFFFFFFFFFFF
-; AVX512-NEXT:    addq %r8, %rsi
-; AVX512-NEXT:    addq %rax, %rcx
-; AVX512-NEXT:    cmovoq %rsi, %rcx
-; AVX512-NEXT:    vpextrq $1, %xmm1, %r9
-; AVX512-NEXT:    vpextrq $1, %xmm0, %rax
-; AVX512-NEXT:    xorl %edi, %edi
-; AVX512-NEXT:    movq %rax, %rsi
-; AVX512-NEXT:    addq %r9, %rsi
-; AVX512-NEXT:    setns %dil
-; AVX512-NEXT:    addq %r8, %rdi
-; AVX512-NEXT:    addq %r9, %rax
-; AVX512-NEXT:    cmovoq %rdi, %rax
-; AVX512-NEXT:    vmovq %rax, %xmm0
-; AVX512-NEXT:    vmovq %rcx, %xmm1
-; AVX512-NEXT:    vpunpcklqdq {{.*#+}} xmm0 = xmm1[0],xmm0[0]
-; AVX512-NEXT:    vpsrlq $56, %xmm0, %xmm0
+; AVX512-NEXT:    vpaddsb %xmm1, %xmm0, %xmm0
+; AVX512-NEXT:    vpmovzxbq {{.*#+}} xmm0 = xmm0[0],zero,zero,zero,zero,zero,zero,zero,xmm0[1],zero,zero,zero,zero,zero,zero,zero
 ; AVX512-NEXT:    vpmovqb %xmm0, (%rdx)
 ; AVX512-NEXT:    retq
   %x = load <2 x i8>, <2 x i8>* %px
@@ -14655,336 +346,36 @@ define void @v2i8(<2 x i8>* %px, <2 x i8
 }
 
 define void @v4i16(<4 x i16>* %px, <4 x i16>* %py, <4 x i16>* %pz) nounwind {
-; SSE2-LABEL: v4i16:
-; SSE2:       # %bb.0:
-; SSE2-NEXT:    movq {{.*#+}} xmm2 = mem[0],zero
-; SSE2-NEXT:    movq {{.*#+}} xmm3 = mem[0],zero
-; SSE2-NEXT:    pxor %xmm0, %xmm0
-; SSE2-NEXT:    pxor %xmm1, %xmm1
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm3[0],xmm1[1],xmm3[1],xmm1[2],xmm3[2],xmm1[3],xmm3[3]
-; SSE2-NEXT:    pshufd {{.*#+}} xmm3 = xmm1[1,1,2,3]
-; SSE2-NEXT:    movd %xmm3, %ecx
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1],xmm0[2],xmm2[2],xmm0[3],xmm2[3]
-; SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm0[1,1,2,3]
-; SSE2-NEXT:    movd %xmm2, %r8d
-; SSE2-NEXT:    xorl %esi, %esi
-; SSE2-NEXT:    movl %r8d, %edi
-; SSE2-NEXT:    addl %ecx, %edi
-; SSE2-NEXT:    setns %sil
-; SSE2-NEXT:    addl $2147483647, %esi # imm = 0x7FFFFFFF
-; SSE2-NEXT:    addl %ecx, %r8d
-; SSE2-NEXT:    cmovol %esi, %r8d
-; SSE2-NEXT:    movd %xmm1, %esi
-; SSE2-NEXT:    movd %xmm0, %r10d
-; SSE2-NEXT:    xorl %edi, %edi
-; SSE2-NEXT:    movl %r10d, %ecx
-; SSE2-NEXT:    addl %esi, %ecx
-; SSE2-NEXT:    setns %dil
-; SSE2-NEXT:    addl $2147483647, %edi # imm = 0x7FFFFFFF
-; SSE2-NEXT:    addl %esi, %r10d
-; SSE2-NEXT:    cmovol %edi, %r10d
-; SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm1[2,3,0,1]
-; SSE2-NEXT:    movd %xmm2, %r9d
-; SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm0[2,3,0,1]
-; SSE2-NEXT:    movd %xmm2, %ecx
-; SSE2-NEXT:    xorl %edi, %edi
-; SSE2-NEXT:    movl %ecx, %esi
-; SSE2-NEXT:    addl %r9d, %esi
-; SSE2-NEXT:    setns %dil
-; SSE2-NEXT:    addl $2147483647, %edi # imm = 0x7FFFFFFF
-; SSE2-NEXT:    addl %r9d, %ecx
-; SSE2-NEXT:    cmovol %edi, %ecx
-; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[3,1,2,3]
-; SSE2-NEXT:    movd %xmm1, %r9d
-; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[3,1,2,3]
-; SSE2-NEXT:    movd %xmm0, %eax
-; SSE2-NEXT:    xorl %esi, %esi
-; SSE2-NEXT:    movl %eax, %edi
-; SSE2-NEXT:    addl %r9d, %edi
-; SSE2-NEXT:    setns %sil
-; SSE2-NEXT:    addl $2147483647, %esi # imm = 0x7FFFFFFF
-; SSE2-NEXT:    addl %r9d, %eax
-; SSE2-NEXT:    cmovol %esi, %eax
-; SSE2-NEXT:    movd %eax, %xmm0
-; SSE2-NEXT:    movd %ecx, %xmm1
-; SSE2-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1]
-; SSE2-NEXT:    movd %r10d, %xmm0
-; SSE2-NEXT:    movd %r8d, %xmm2
-; SSE2-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1]
-; SSE2-NEXT:    punpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm1[0]
-; SSE2-NEXT:    psrld $16, %xmm0
-; SSE2-NEXT:    pshuflw {{.*#+}} xmm0 = xmm0[0,2,2,3,4,5,6,7]
-; SSE2-NEXT:    pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,6,6,7]
-; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
-; SSE2-NEXT:    movq %xmm0, (%rdx)
-; SSE2-NEXT:    retq
-;
-; SSSE3-LABEL: v4i16:
-; SSSE3:       # %bb.0:
-; SSSE3-NEXT:    movq {{.*#+}} xmm2 = mem[0],zero
-; SSSE3-NEXT:    movq {{.*#+}} xmm3 = mem[0],zero
-; SSSE3-NEXT:    pxor %xmm0, %xmm0
-; SSSE3-NEXT:    pxor %xmm1, %xmm1
-; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm3[0],xmm1[1],xmm3[1],xmm1[2],xmm3[2],xmm1[3],xmm3[3]
-; SSSE3-NEXT:    pshufd {{.*#+}} xmm3 = xmm1[1,1,2,3]
-; SSSE3-NEXT:    movd %xmm3, %ecx
-; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1],xmm0[2],xmm2[2],xmm0[3],xmm2[3]
-; SSSE3-NEXT:    pshufd {{.*#+}} xmm2 = xmm0[1,1,2,3]
-; SSSE3-NEXT:    movd %xmm2, %r8d
-; SSSE3-NEXT:    xorl %esi, %esi
-; SSSE3-NEXT:    movl %r8d, %edi
-; SSSE3-NEXT:    addl %ecx, %edi
-; SSSE3-NEXT:    setns %sil
-; SSSE3-NEXT:    addl $2147483647, %esi # imm = 0x7FFFFFFF
-; SSSE3-NEXT:    addl %ecx, %r8d
-; SSSE3-NEXT:    cmovol %esi, %r8d
-; SSSE3-NEXT:    movd %xmm1, %esi
-; SSSE3-NEXT:    movd %xmm0, %r10d
-; SSSE3-NEXT:    xorl %edi, %edi
-; SSSE3-NEXT:    movl %r10d, %ecx
-; SSSE3-NEXT:    addl %esi, %ecx
-; SSSE3-NEXT:    setns %dil
-; SSSE3-NEXT:    addl $2147483647, %edi # imm = 0x7FFFFFFF
-; SSSE3-NEXT:    addl %esi, %r10d
-; SSSE3-NEXT:    cmovol %edi, %r10d
-; SSSE3-NEXT:    pshufd {{.*#+}} xmm2 = xmm1[2,3,0,1]
-; SSSE3-NEXT:    movd %xmm2, %r9d
-; SSSE3-NEXT:    pshufd {{.*#+}} xmm2 = xmm0[2,3,0,1]
-; SSSE3-NEXT:    movd %xmm2, %ecx
-; SSSE3-NEXT:    xorl %edi, %edi
-; SSSE3-NEXT:    movl %ecx, %esi
-; SSSE3-NEXT:    addl %r9d, %esi
-; SSSE3-NEXT:    setns %dil
-; SSSE3-NEXT:    addl $2147483647, %edi # imm = 0x7FFFFFFF
-; SSSE3-NEXT:    addl %r9d, %ecx
-; SSSE3-NEXT:    cmovol %edi, %ecx
-; SSSE3-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[3,1,2,3]
-; SSSE3-NEXT:    movd %xmm1, %r9d
-; SSSE3-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[3,1,2,3]
-; SSSE3-NEXT:    movd %xmm0, %eax
-; SSSE3-NEXT:    xorl %esi, %esi
-; SSSE3-NEXT:    movl %eax, %edi
-; SSSE3-NEXT:    addl %r9d, %edi
-; SSSE3-NEXT:    setns %sil
-; SSSE3-NEXT:    addl $2147483647, %esi # imm = 0x7FFFFFFF
-; SSSE3-NEXT:    addl %r9d, %eax
-; SSSE3-NEXT:    cmovol %esi, %eax
-; SSSE3-NEXT:    movd %eax, %xmm0
-; SSSE3-NEXT:    movd %ecx, %xmm1
-; SSSE3-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1]
-; SSSE3-NEXT:    movd %r10d, %xmm0
-; SSSE3-NEXT:    movd %r8d, %xmm2
-; SSSE3-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1]
-; SSSE3-NEXT:    punpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm1[0]
-; SSSE3-NEXT:    pshufb {{.*#+}} xmm0 = xmm0[2,3,6,7,10,11,14,15,10,11,14,15,14,15],zero,zero
-; SSSE3-NEXT:    movq %xmm0, (%rdx)
-; SSSE3-NEXT:    retq
-;
-; SSE41-LABEL: v4i16:
-; SSE41:       # %bb.0:
-; SSE41-NEXT:    movq {{.*#+}} xmm2 = mem[0],zero
-; SSE41-NEXT:    movq {{.*#+}} xmm3 = mem[0],zero
-; SSE41-NEXT:    pxor %xmm0, %xmm0
-; SSE41-NEXT:    pxor %xmm1, %xmm1
-; SSE41-NEXT:    punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm3[0],xmm1[1],xmm3[1],xmm1[2],xmm3[2],xmm1[3],xmm3[3]
-; SSE41-NEXT:    pextrd $3, %xmm1, %ecx
-; SSE41-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1],xmm0[2],xmm2[2],xmm0[3],xmm2[3]
-; SSE41-NEXT:    pextrd $3, %xmm0, %r8d
-; SSE41-NEXT:    xorl %esi, %esi
-; SSE41-NEXT:    movl %r8d, %edi
-; SSE41-NEXT:    addl %ecx, %edi
-; SSE41-NEXT:    setns %sil
-; SSE41-NEXT:    addl $2147483647, %esi # imm = 0x7FFFFFFF
-; SSE41-NEXT:    addl %ecx, %r8d
-; SSE41-NEXT:    cmovol %esi, %r8d
-; SSE41-NEXT:    pextrd $2, %xmm1, %esi
-; SSE41-NEXT:    pextrd $2, %xmm0, %r10d
-; SSE41-NEXT:    xorl %edi, %edi
-; SSE41-NEXT:    movl %r10d, %ecx
-; SSE41-NEXT:    addl %esi, %ecx
-; SSE41-NEXT:    setns %dil
-; SSE41-NEXT:    addl $2147483647, %edi # imm = 0x7FFFFFFF
-; SSE41-NEXT:    addl %esi, %r10d
-; SSE41-NEXT:    cmovol %edi, %r10d
-; SSE41-NEXT:    movd %xmm1, %r9d
-; SSE41-NEXT:    movd %xmm0, %ecx
-; SSE41-NEXT:    xorl %edi, %edi
-; SSE41-NEXT:    movl %ecx, %esi
-; SSE41-NEXT:    addl %r9d, %esi
-; SSE41-NEXT:    setns %dil
-; SSE41-NEXT:    addl $2147483647, %edi # imm = 0x7FFFFFFF
-; SSE41-NEXT:    addl %r9d, %ecx
-; SSE41-NEXT:    cmovol %edi, %ecx
-; SSE41-NEXT:    pextrd $1, %xmm1, %r9d
-; SSE41-NEXT:    pextrd $1, %xmm0, %eax
-; SSE41-NEXT:    xorl %esi, %esi
-; SSE41-NEXT:    movl %eax, %edi
-; SSE41-NEXT:    addl %r9d, %edi
-; SSE41-NEXT:    setns %sil
-; SSE41-NEXT:    addl $2147483647, %esi # imm = 0x7FFFFFFF
-; SSE41-NEXT:    addl %r9d, %eax
-; SSE41-NEXT:    cmovol %esi, %eax
-; SSE41-NEXT:    movd %ecx, %xmm0
-; SSE41-NEXT:    pinsrd $1, %eax, %xmm0
-; SSE41-NEXT:    pinsrd $2, %r10d, %xmm0
-; SSE41-NEXT:    pinsrd $3, %r8d, %xmm0
-; SSE41-NEXT:    psrld $16, %xmm0
-; SSE41-NEXT:    packusdw %xmm0, %xmm0
-; SSE41-NEXT:    movq %xmm0, (%rdx)
-; SSE41-NEXT:    retq
+; SSE-LABEL: v4i16:
+; SSE:       # %bb.0:
+; SSE-NEXT:    movq {{.*#+}} xmm0 = mem[0],zero
+; SSE-NEXT:    movq {{.*#+}} xmm1 = mem[0],zero
+; SSE-NEXT:    paddsw %xmm0, %xmm1
+; SSE-NEXT:    movq %xmm1, (%rdx)
+; SSE-NEXT:    retq
 ;
 ; AVX1-LABEL: v4i16:
 ; AVX1:       # %bb.0:
-; AVX1-NEXT:    vmovq {{.*#+}} xmm1 = mem[0],zero
 ; AVX1-NEXT:    vmovq {{.*#+}} xmm0 = mem[0],zero
-; AVX1-NEXT:    vpxor %xmm2, %xmm2, %xmm2
-; AVX1-NEXT:    vpunpcklwd {{.*#+}} xmm0 = xmm2[0],xmm0[0],xmm2[1],xmm0[1],xmm2[2],xmm0[2],xmm2[3],xmm0[3]
-; AVX1-NEXT:    vpextrd $3, %xmm0, %ecx
-; AVX1-NEXT:    vpunpcklwd {{.*#+}} xmm1 = xmm2[0],xmm1[0],xmm2[1],xmm1[1],xmm2[2],xmm1[2],xmm2[3],xmm1[3]
-; AVX1-NEXT:    vpextrd $3, %xmm1, %r9d
-; AVX1-NEXT:    xorl %esi, %esi
-; AVX1-NEXT:    movl %r9d, %edi
-; AVX1-NEXT:    addl %ecx, %edi
-; AVX1-NEXT:    setns %sil
-; AVX1-NEXT:    addl $2147483647, %esi # imm = 0x7FFFFFFF
-; AVX1-NEXT:    addl %ecx, %r9d
-; AVX1-NEXT:    cmovol %esi, %r9d
-; AVX1-NEXT:    vpextrd $2, %xmm0, %r8d
-; AVX1-NEXT:    vpextrd $2, %xmm1, %r10d
-; AVX1-NEXT:    xorl %edi, %edi
-; AVX1-NEXT:    movl %r10d, %ecx
-; AVX1-NEXT:    addl %r8d, %ecx
-; AVX1-NEXT:    setns %dil
-; AVX1-NEXT:    addl $2147483647, %edi # imm = 0x7FFFFFFF
-; AVX1-NEXT:    addl %r8d, %r10d
-; AVX1-NEXT:    cmovol %edi, %r10d
-; AVX1-NEXT:    vmovd %xmm0, %r8d
-; AVX1-NEXT:    vmovd %xmm1, %eax
-; AVX1-NEXT:    xorl %ecx, %ecx
-; AVX1-NEXT:    movl %eax, %edi
-; AVX1-NEXT:    addl %r8d, %edi
-; AVX1-NEXT:    setns %cl
-; AVX1-NEXT:    addl $2147483647, %ecx # imm = 0x7FFFFFFF
-; AVX1-NEXT:    addl %r8d, %eax
-; AVX1-NEXT:    cmovol %ecx, %eax
-; AVX1-NEXT:    vpextrd $1, %xmm0, %r8d
-; AVX1-NEXT:    vpextrd $1, %xmm1, %esi
-; AVX1-NEXT:    xorl %ecx, %ecx
-; AVX1-NEXT:    movl %esi, %edi
-; AVX1-NEXT:    addl %r8d, %edi
-; AVX1-NEXT:    setns %cl
-; AVX1-NEXT:    addl $2147483647, %ecx # imm = 0x7FFFFFFF
-; AVX1-NEXT:    addl %r8d, %esi
-; AVX1-NEXT:    cmovol %ecx, %esi
-; AVX1-NEXT:    vmovd %eax, %xmm0
-; AVX1-NEXT:    vpinsrd $1, %esi, %xmm0, %xmm0
-; AVX1-NEXT:    vpinsrd $2, %r10d, %xmm0, %xmm0
-; AVX1-NEXT:    vpinsrd $3, %r9d, %xmm0, %xmm0
-; AVX1-NEXT:    vpsrld $16, %xmm0, %xmm0
-; AVX1-NEXT:    vpackusdw %xmm0, %xmm0, %xmm0
+; AVX1-NEXT:    vmovq {{.*#+}} xmm1 = mem[0],zero
+; AVX1-NEXT:    vpaddsw %xmm1, %xmm0, %xmm0
 ; AVX1-NEXT:    vmovq %xmm0, (%rdx)
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: v4i16:
 ; AVX2:       # %bb.0:
-; AVX2-NEXT:    vmovq {{.*#+}} xmm1 = mem[0],zero
 ; AVX2-NEXT:    vmovq {{.*#+}} xmm0 = mem[0],zero
-; AVX2-NEXT:    vpxor %xmm2, %xmm2, %xmm2
-; AVX2-NEXT:    vpunpcklwd {{.*#+}} xmm0 = xmm2[0],xmm0[0],xmm2[1],xmm0[1],xmm2[2],xmm0[2],xmm2[3],xmm0[3]
-; AVX2-NEXT:    vpextrd $3, %xmm0, %ecx
-; AVX2-NEXT:    vpunpcklwd {{.*#+}} xmm1 = xmm2[0],xmm1[0],xmm2[1],xmm1[1],xmm2[2],xmm1[2],xmm2[3],xmm1[3]
-; AVX2-NEXT:    vpextrd $3, %xmm1, %r9d
-; AVX2-NEXT:    xorl %esi, %esi
-; AVX2-NEXT:    movl %r9d, %edi
-; AVX2-NEXT:    addl %ecx, %edi
-; AVX2-NEXT:    setns %sil
-; AVX2-NEXT:    addl $2147483647, %esi # imm = 0x7FFFFFFF
-; AVX2-NEXT:    addl %ecx, %r9d
-; AVX2-NEXT:    cmovol %esi, %r9d
-; AVX2-NEXT:    vpextrd $2, %xmm0, %r8d
-; AVX2-NEXT:    vpextrd $2, %xmm1, %r10d
-; AVX2-NEXT:    xorl %edi, %edi
-; AVX2-NEXT:    movl %r10d, %ecx
-; AVX2-NEXT:    addl %r8d, %ecx
-; AVX2-NEXT:    setns %dil
-; AVX2-NEXT:    addl $2147483647, %edi # imm = 0x7FFFFFFF
-; AVX2-NEXT:    addl %r8d, %r10d
-; AVX2-NEXT:    cmovol %edi, %r10d
-; AVX2-NEXT:    vmovd %xmm0, %r8d
-; AVX2-NEXT:    vmovd %xmm1, %eax
-; AVX2-NEXT:    xorl %ecx, %ecx
-; AVX2-NEXT:    movl %eax, %edi
-; AVX2-NEXT:    addl %r8d, %edi
-; AVX2-NEXT:    setns %cl
-; AVX2-NEXT:    addl $2147483647, %ecx # imm = 0x7FFFFFFF
-; AVX2-NEXT:    addl %r8d, %eax
-; AVX2-NEXT:    cmovol %ecx, %eax
-; AVX2-NEXT:    vpextrd $1, %xmm0, %r8d
-; AVX2-NEXT:    vpextrd $1, %xmm1, %esi
-; AVX2-NEXT:    xorl %ecx, %ecx
-; AVX2-NEXT:    movl %esi, %edi
-; AVX2-NEXT:    addl %r8d, %edi
-; AVX2-NEXT:    setns %cl
-; AVX2-NEXT:    addl $2147483647, %ecx # imm = 0x7FFFFFFF
-; AVX2-NEXT:    addl %r8d, %esi
-; AVX2-NEXT:    cmovol %ecx, %esi
-; AVX2-NEXT:    vmovd %eax, %xmm0
-; AVX2-NEXT:    vpinsrd $1, %esi, %xmm0, %xmm0
-; AVX2-NEXT:    vpinsrd $2, %r10d, %xmm0, %xmm0
-; AVX2-NEXT:    vpinsrd $3, %r9d, %xmm0, %xmm0
-; AVX2-NEXT:    vpsrld $16, %xmm0, %xmm0
-; AVX2-NEXT:    vpackusdw %xmm0, %xmm0, %xmm0
+; AVX2-NEXT:    vmovq {{.*#+}} xmm1 = mem[0],zero
+; AVX2-NEXT:    vpaddsw %xmm1, %xmm0, %xmm0
 ; AVX2-NEXT:    vmovq %xmm0, (%rdx)
 ; AVX2-NEXT:    retq
 ;
 ; AVX512-LABEL: v4i16:
 ; AVX512:       # %bb.0:
-; AVX512-NEXT:    vmovq {{.*#+}} xmm1 = mem[0],zero
 ; AVX512-NEXT:    vmovq {{.*#+}} xmm0 = mem[0],zero
-; AVX512-NEXT:    vmovdqa {{.*#+}} xmm2 = [255,255,0,1,255,255,2,3,255,255,4,5,255,255,6,7]
-; AVX512-NEXT:    vpshufb %xmm2, %xmm0, %xmm0
-; AVX512-NEXT:    vpextrd $3, %xmm0, %ecx
-; AVX512-NEXT:    vpshufb %xmm2, %xmm1, %xmm1
-; AVX512-NEXT:    vpextrd $3, %xmm1, %r9d
-; AVX512-NEXT:    xorl %esi, %esi
-; AVX512-NEXT:    movl %r9d, %edi
-; AVX512-NEXT:    addl %ecx, %edi
-; AVX512-NEXT:    setns %sil
-; AVX512-NEXT:    addl $2147483647, %esi # imm = 0x7FFFFFFF
-; AVX512-NEXT:    addl %ecx, %r9d
-; AVX512-NEXT:    cmovol %esi, %r9d
-; AVX512-NEXT:    vpextrd $2, %xmm0, %r8d
-; AVX512-NEXT:    vpextrd $2, %xmm1, %r10d
-; AVX512-NEXT:    xorl %edi, %edi
-; AVX512-NEXT:    movl %r10d, %ecx
-; AVX512-NEXT:    addl %r8d, %ecx
-; AVX512-NEXT:    setns %dil
-; AVX512-NEXT:    addl $2147483647, %edi # imm = 0x7FFFFFFF
-; AVX512-NEXT:    addl %r8d, %r10d
-; AVX512-NEXT:    cmovol %edi, %r10d
-; AVX512-NEXT:    vmovd %xmm0, %r8d
-; AVX512-NEXT:    vmovd %xmm1, %eax
-; AVX512-NEXT:    xorl %ecx, %ecx
-; AVX512-NEXT:    movl %eax, %edi
-; AVX512-NEXT:    addl %r8d, %edi
-; AVX512-NEXT:    setns %cl
-; AVX512-NEXT:    addl $2147483647, %ecx # imm = 0x7FFFFFFF
-; AVX512-NEXT:    addl %r8d, %eax
-; AVX512-NEXT:    cmovol %ecx, %eax
-; AVX512-NEXT:    vpextrd $1, %xmm0, %r8d
-; AVX512-NEXT:    vpextrd $1, %xmm1, %esi
-; AVX512-NEXT:    xorl %ecx, %ecx
-; AVX512-NEXT:    movl %esi, %edi
-; AVX512-NEXT:    addl %r8d, %edi
-; AVX512-NEXT:    setns %cl
-; AVX512-NEXT:    addl $2147483647, %ecx # imm = 0x7FFFFFFF
-; AVX512-NEXT:    addl %r8d, %esi
-; AVX512-NEXT:    cmovol %ecx, %esi
-; AVX512-NEXT:    vmovd %eax, %xmm0
-; AVX512-NEXT:    vpinsrd $1, %esi, %xmm0, %xmm0
-; AVX512-NEXT:    vpinsrd $2, %r10d, %xmm0, %xmm0
-; AVX512-NEXT:    vpinsrd $3, %r9d, %xmm0, %xmm0
-; AVX512-NEXT:    vpsrld $16, %xmm0, %xmm0
+; AVX512-NEXT:    vmovq {{.*#+}} xmm1 = mem[0],zero
+; AVX512-NEXT:    vpaddsw %xmm1, %xmm0, %xmm0
+; AVX512-NEXT:    vpmovzxwd {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero
 ; AVX512-NEXT:    vpmovdw %xmm0, (%rdx)
 ; AVX512-NEXT:    retq
   %x = load <4 x i16>, <4 x i16>* %px
@@ -14995,174 +386,27 @@ define void @v4i16(<4 x i16>* %px, <4 x
 }
 
 define void @v2i16(<2 x i16>* %px, <2 x i16>* %py, <2 x i16>* %pz) nounwind {
-; SSE2-LABEL: v2i16:
-; SSE2:       # %bb.0:
-; SSE2-NEXT:    movd {{.*#+}} xmm0 = mem[0],zero,zero,zero
-; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[0,1,0,3]
-; SSE2-NEXT:    pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,5,5,6,7]
-; SSE2-NEXT:    movd {{.*#+}} xmm1 = mem[0],zero,zero,zero
-; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[0,1,0,3]
-; SSE2-NEXT:    pshufhw {{.*#+}} xmm1 = xmm1[0,1,2,3,5,5,6,7]
-; SSE2-NEXT:    psllq $48, %xmm1
-; SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm1[2,3,0,1]
-; SSE2-NEXT:    movq %xmm2, %rax
-; SSE2-NEXT:    psllq $48, %xmm0
-; SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm0[2,3,0,1]
-; SSE2-NEXT:    movq %xmm2, %rcx
-; SSE2-NEXT:    xorl %esi, %esi
-; SSE2-NEXT:    movq %rcx, %rdi
-; SSE2-NEXT:    addq %rax, %rdi
-; SSE2-NEXT:    setns %sil
-; SSE2-NEXT:    movabsq $9223372036854775807, %r8 # imm = 0x7FFFFFFFFFFFFFFF
-; SSE2-NEXT:    addq %r8, %rsi
-; SSE2-NEXT:    addq %rax, %rcx
-; SSE2-NEXT:    cmovoq %rsi, %rcx
-; SSE2-NEXT:    movq %xmm1, %r9
-; SSE2-NEXT:    movq %xmm0, %rax
-; SSE2-NEXT:    xorl %edi, %edi
-; SSE2-NEXT:    movq %rax, %rsi
-; SSE2-NEXT:    addq %r9, %rsi
-; SSE2-NEXT:    setns %dil
-; SSE2-NEXT:    addq %r8, %rdi
-; SSE2-NEXT:    addq %r9, %rax
-; SSE2-NEXT:    cmovoq %rdi, %rax
-; SSE2-NEXT:    movq %rax, %xmm0
-; SSE2-NEXT:    movq %rcx, %xmm1
-; SSE2-NEXT:    punpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm1[0]
-; SSE2-NEXT:    psrlq $48, %xmm0
-; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
-; SSE2-NEXT:    pshuflw {{.*#+}} xmm0 = xmm0[0,2,2,3,4,5,6,7]
-; SSE2-NEXT:    movd %xmm0, (%rdx)
-; SSE2-NEXT:    retq
-;
-; SSSE3-LABEL: v2i16:
-; SSSE3:       # %bb.0:
-; SSSE3-NEXT:    movd {{.*#+}} xmm0 = mem[0],zero,zero,zero
-; SSSE3-NEXT:    movd {{.*#+}} xmm1 = mem[0],zero,zero,zero
-; SSSE3-NEXT:    movdqa {{.*#+}} xmm2 = [255,255,255,255,255,255,0,1,255,255,255,255,255,255,2,3]
-; SSSE3-NEXT:    pshufb %xmm2, %xmm1
-; SSSE3-NEXT:    pshufd {{.*#+}} xmm3 = xmm1[2,3,0,1]
-; SSSE3-NEXT:    movq %xmm3, %rax
-; SSSE3-NEXT:    pshufb %xmm2, %xmm0
-; SSSE3-NEXT:    pshufd {{.*#+}} xmm2 = xmm0[2,3,0,1]
-; SSSE3-NEXT:    movq %xmm2, %rcx
-; SSSE3-NEXT:    xorl %esi, %esi
-; SSSE3-NEXT:    movq %rcx, %rdi
-; SSSE3-NEXT:    addq %rax, %rdi
-; SSSE3-NEXT:    setns %sil
-; SSSE3-NEXT:    movabsq $9223372036854775807, %r8 # imm = 0x7FFFFFFFFFFFFFFF
-; SSSE3-NEXT:    addq %r8, %rsi
-; SSSE3-NEXT:    addq %rax, %rcx
-; SSSE3-NEXT:    cmovoq %rsi, %rcx
-; SSSE3-NEXT:    movq %xmm1, %r9
-; SSSE3-NEXT:    movq %xmm0, %rax
-; SSSE3-NEXT:    xorl %edi, %edi
-; SSSE3-NEXT:    movq %rax, %rsi
-; SSSE3-NEXT:    addq %r9, %rsi
-; SSSE3-NEXT:    setns %dil
-; SSSE3-NEXT:    addq %r8, %rdi
-; SSSE3-NEXT:    addq %r9, %rax
-; SSSE3-NEXT:    cmovoq %rdi, %rax
-; SSSE3-NEXT:    movq %rax, %xmm0
-; SSSE3-NEXT:    movq %rcx, %xmm1
-; SSSE3-NEXT:    punpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm1[0]
-; SSSE3-NEXT:    pshufb {{.*#+}} xmm0 = xmm0[6,7,14,15,14,15],zero,zero,xmm0[14,15],zero,zero,zero,zero,zero,zero
-; SSSE3-NEXT:    movd %xmm0, (%rdx)
-; SSSE3-NEXT:    retq
-;
-; SSE41-LABEL: v2i16:
-; SSE41:       # %bb.0:
-; SSE41-NEXT:    pmovzxwq {{.*#+}} xmm0 = mem[0],zero,zero,zero,mem[1],zero,zero,zero
-; SSE41-NEXT:    pmovzxwq {{.*#+}} xmm1 = mem[0],zero,zero,zero,mem[1],zero,zero,zero
-; SSE41-NEXT:    psllq $48, %xmm1
-; SSE41-NEXT:    movq %xmm1, %rax
-; SSE41-NEXT:    psllq $48, %xmm0
-; SSE41-NEXT:    movq %xmm0, %rcx
-; SSE41-NEXT:    xorl %esi, %esi
-; SSE41-NEXT:    movq %rcx, %rdi
-; SSE41-NEXT:    addq %rax, %rdi
-; SSE41-NEXT:    setns %sil
-; SSE41-NEXT:    movabsq $9223372036854775807, %r8 # imm = 0x7FFFFFFFFFFFFFFF
-; SSE41-NEXT:    addq %r8, %rsi
-; SSE41-NEXT:    addq %rax, %rcx
-; SSE41-NEXT:    cmovoq %rsi, %rcx
-; SSE41-NEXT:    pextrq $1, %xmm1, %r9
-; SSE41-NEXT:    pextrq $1, %xmm0, %rax
-; SSE41-NEXT:    xorl %edi, %edi
-; SSE41-NEXT:    movq %rax, %rsi
-; SSE41-NEXT:    addq %r9, %rsi
-; SSE41-NEXT:    setns %dil
-; SSE41-NEXT:    addq %r8, %rdi
-; SSE41-NEXT:    addq %r9, %rax
-; SSE41-NEXT:    cmovoq %rdi, %rax
-; SSE41-NEXT:    movq %rax, %xmm0
-; SSE41-NEXT:    movq %rcx, %xmm1
-; SSE41-NEXT:    punpcklqdq {{.*#+}} xmm1 = xmm1[0],xmm0[0]
-; SSE41-NEXT:    pshufb {{.*#+}} xmm1 = xmm1[6,7,14,15,14,15],zero,zero,xmm1[14,15],zero,zero,zero,zero,zero,zero
-; SSE41-NEXT:    movd %xmm1, (%rdx)
-; SSE41-NEXT:    retq
+; SSE-LABEL: v2i16:
+; SSE:       # %bb.0:
+; SSE-NEXT:    movd {{.*#+}} xmm0 = mem[0],zero,zero,zero
+; SSE-NEXT:    movd {{.*#+}} xmm1 = mem[0],zero,zero,zero
+; SSE-NEXT:    paddsw %xmm0, %xmm1
+; SSE-NEXT:    movd %xmm1, (%rdx)
+; SSE-NEXT:    retq
 ;
 ; AVX1-LABEL: v2i16:
 ; AVX1:       # %bb.0:
-; AVX1-NEXT:    vpmovzxwq {{.*#+}} xmm0 = mem[0],zero,zero,zero,mem[1],zero,zero,zero
-; AVX1-NEXT:    vpmovzxwq {{.*#+}} xmm1 = mem[0],zero,zero,zero,mem[1],zero,zero,zero
-; AVX1-NEXT:    vpsllq $48, %xmm1, %xmm1
-; AVX1-NEXT:    vmovq %xmm1, %rax
-; AVX1-NEXT:    vpsllq $48, %xmm0, %xmm0
-; AVX1-NEXT:    vmovq %xmm0, %rcx
-; AVX1-NEXT:    xorl %esi, %esi
-; AVX1-NEXT:    movq %rcx, %rdi
-; AVX1-NEXT:    addq %rax, %rdi
-; AVX1-NEXT:    setns %sil
-; AVX1-NEXT:    movabsq $9223372036854775807, %r8 # imm = 0x7FFFFFFFFFFFFFFF
-; AVX1-NEXT:    addq %r8, %rsi
-; AVX1-NEXT:    addq %rax, %rcx
-; AVX1-NEXT:    cmovoq %rsi, %rcx
-; AVX1-NEXT:    vpextrq $1, %xmm1, %r9
-; AVX1-NEXT:    vpextrq $1, %xmm0, %rax
-; AVX1-NEXT:    xorl %edi, %edi
-; AVX1-NEXT:    movq %rax, %rsi
-; AVX1-NEXT:    addq %r9, %rsi
-; AVX1-NEXT:    setns %dil
-; AVX1-NEXT:    addq %r8, %rdi
-; AVX1-NEXT:    addq %r9, %rax
-; AVX1-NEXT:    cmovoq %rdi, %rax
-; AVX1-NEXT:    vmovq %rax, %xmm0
-; AVX1-NEXT:    vmovq %rcx, %xmm1
-; AVX1-NEXT:    vpunpcklqdq {{.*#+}} xmm0 = xmm1[0],xmm0[0]
-; AVX1-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[6,7,14,15,14,15],zero,zero,xmm0[14,15],zero,zero,zero,zero,zero,zero
+; AVX1-NEXT:    vmovd {{.*#+}} xmm0 = mem[0],zero,zero,zero
+; AVX1-NEXT:    vmovd {{.*#+}} xmm1 = mem[0],zero,zero,zero
+; AVX1-NEXT:    vpaddsw %xmm1, %xmm0, %xmm0
 ; AVX1-NEXT:    vmovd %xmm0, (%rdx)
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: v2i16:
 ; AVX2:       # %bb.0:
-; AVX2-NEXT:    vpmovzxwq {{.*#+}} xmm0 = mem[0],zero,zero,zero,mem[1],zero,zero,zero
-; AVX2-NEXT:    vpmovzxwq {{.*#+}} xmm1 = mem[0],zero,zero,zero,mem[1],zero,zero,zero
-; AVX2-NEXT:    vpsllq $48, %xmm1, %xmm1
-; AVX2-NEXT:    vmovq %xmm1, %rax
-; AVX2-NEXT:    vpsllq $48, %xmm0, %xmm0
-; AVX2-NEXT:    vmovq %xmm0, %rcx
-; AVX2-NEXT:    xorl %esi, %esi
-; AVX2-NEXT:    movq %rcx, %rdi
-; AVX2-NEXT:    addq %rax, %rdi
-; AVX2-NEXT:    setns %sil
-; AVX2-NEXT:    movabsq $9223372036854775807, %r8 # imm = 0x7FFFFFFFFFFFFFFF
-; AVX2-NEXT:    addq %r8, %rsi
-; AVX2-NEXT:    addq %rax, %rcx
-; AVX2-NEXT:    cmovoq %rsi, %rcx
-; AVX2-NEXT:    vpextrq $1, %xmm1, %r9
-; AVX2-NEXT:    vpextrq $1, %xmm0, %rax
-; AVX2-NEXT:    xorl %edi, %edi
-; AVX2-NEXT:    movq %rax, %rsi
-; AVX2-NEXT:    addq %r9, %rsi
-; AVX2-NEXT:    setns %dil
-; AVX2-NEXT:    addq %r8, %rdi
-; AVX2-NEXT:    addq %r9, %rax
-; AVX2-NEXT:    cmovoq %rdi, %rax
-; AVX2-NEXT:    vmovq %rax, %xmm0
-; AVX2-NEXT:    vmovq %rcx, %xmm1
-; AVX2-NEXT:    vpunpcklqdq {{.*#+}} xmm0 = xmm1[0],xmm0[0]
-; AVX2-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[6,7,14,15,14,15],zero,zero,xmm0[14,15],zero,zero,zero,zero,zero,zero
+; AVX2-NEXT:    vmovd {{.*#+}} xmm0 = mem[0],zero,zero,zero
+; AVX2-NEXT:    vmovd {{.*#+}} xmm1 = mem[0],zero,zero,zero
+; AVX2-NEXT:    vpaddsw %xmm1, %xmm0, %xmm0
 ; AVX2-NEXT:    vmovd %xmm0, (%rdx)
 ; AVX2-NEXT:    retq
 ;
@@ -15170,32 +414,8 @@ define void @v2i16(<2 x i16>* %px, <2 x
 ; AVX512:       # %bb.0:
 ; AVX512-NEXT:    vmovd {{.*#+}} xmm0 = mem[0],zero,zero,zero
 ; AVX512-NEXT:    vmovd {{.*#+}} xmm1 = mem[0],zero,zero,zero
-; AVX512-NEXT:    vmovdqa {{.*#+}} xmm2 = [255,255,255,255,255,255,0,1,255,255,255,255,255,255,2,3]
-; AVX512-NEXT:    vpshufb %xmm2, %xmm1, %xmm1
-; AVX512-NEXT:    vmovq %xmm1, %rax
-; AVX512-NEXT:    vpshufb %xmm2, %xmm0, %xmm0
-; AVX512-NEXT:    vmovq %xmm0, %rcx
-; AVX512-NEXT:    xorl %esi, %esi
-; AVX512-NEXT:    movq %rcx, %rdi
-; AVX512-NEXT:    addq %rax, %rdi
-; AVX512-NEXT:    setns %sil
-; AVX512-NEXT:    movabsq $9223372036854775807, %r8 # imm = 0x7FFFFFFFFFFFFFFF
-; AVX512-NEXT:    addq %r8, %rsi
-; AVX512-NEXT:    addq %rax, %rcx
-; AVX512-NEXT:    cmovoq %rsi, %rcx
-; AVX512-NEXT:    vpextrq $1, %xmm1, %r9
-; AVX512-NEXT:    vpextrq $1, %xmm0, %rax
-; AVX512-NEXT:    xorl %edi, %edi
-; AVX512-NEXT:    movq %rax, %rsi
-; AVX512-NEXT:    addq %r9, %rsi
-; AVX512-NEXT:    setns %dil
-; AVX512-NEXT:    addq %r8, %rdi
-; AVX512-NEXT:    addq %r9, %rax
-; AVX512-NEXT:    cmovoq %rdi, %rax
-; AVX512-NEXT:    vmovq %rax, %xmm0
-; AVX512-NEXT:    vmovq %rcx, %xmm1
-; AVX512-NEXT:    vpunpcklqdq {{.*#+}} xmm0 = xmm1[0],xmm0[0]
-; AVX512-NEXT:    vpsrlq $48, %xmm0, %xmm0
+; AVX512-NEXT:    vpaddsw %xmm1, %xmm0, %xmm0
+; AVX512-NEXT:    vpmovzxwq {{.*#+}} xmm0 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero
 ; AVX512-NEXT:    vpmovqw %xmm0, (%rdx)
 ; AVX512-NEXT:    retq
   %x = load <2 x i16>, <2 x i16>* %px
@@ -15206,1586 +426,59 @@ define void @v2i16(<2 x i16>* %px, <2 x
 }
 
 define <12 x i8> @v12i8(<12 x i8> %x, <12 x i8> %y) nounwind {
-; SSE2-LABEL: v12i8:
-; SSE2:       # %bb.0:
-; SSE2-NEXT:    pushq %rbp
-; SSE2-NEXT:    pushq %r15
-; SSE2-NEXT:    pushq %r14
-; SSE2-NEXT:    pushq %r13
-; SSE2-NEXT:    pushq %r12
-; SSE2-NEXT:    pushq %rbx
-; SSE2-NEXT:    movaps %xmm1, -{{[0-9]+}}(%rsp)
-; SSE2-NEXT:    movaps %xmm0, -{{[0-9]+}}(%rsp)
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %r9b
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %sil
-; SSE2-NEXT:    movl %r9d, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %r9b
-; SSE2-NEXT:    jno .LBB11_2
-; SSE2-NEXT:  # %bb.1:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %r9d
-; SSE2-NEXT:  .LBB11_2:
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movl %esi, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %sil
-; SSE2-NEXT:    jno .LBB11_4
-; SSE2-NEXT:  # %bb.3:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %esi
-; SSE2-NEXT:  .LBB11_4:
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %bl
-; SSE2-NEXT:    movl %ebx, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %bl
-; SSE2-NEXT:    jno .LBB11_6
-; SSE2-NEXT:  # %bb.5:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %ebx
-; SSE2-NEXT:  .LBB11_6:
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %dl
-; SSE2-NEXT:    movl %edx, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %dl
-; SSE2-NEXT:    jno .LBB11_8
-; SSE2-NEXT:  # %bb.7:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %edx
-; SSE2-NEXT:  .LBB11_8:
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %r10b
-; SSE2-NEXT:    movl %r10d, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %r10b
-; SSE2-NEXT:    jno .LBB11_10
-; SSE2-NEXT:  # %bb.9:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %r10d
-; SSE2-NEXT:  .LBB11_10:
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %r11b
-; SSE2-NEXT:    movl %r11d, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %r11b
-; SSE2-NEXT:    jno .LBB11_12
-; SSE2-NEXT:  # %bb.11:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %r11d
-; SSE2-NEXT:  .LBB11_12:
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %bpl
-; SSE2-NEXT:    movl %ebp, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %bpl
-; SSE2-NEXT:    jno .LBB11_14
-; SSE2-NEXT:  # %bb.13:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %ebp
-; SSE2-NEXT:  .LBB11_14:
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %r14b
-; SSE2-NEXT:    movl %r14d, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %r14b
-; SSE2-NEXT:    jno .LBB11_16
-; SSE2-NEXT:  # %bb.15:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %r14d
-; SSE2-NEXT:  .LBB11_16:
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %r15b
-; SSE2-NEXT:    movl %r15d, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %r15b
-; SSE2-NEXT:    jno .LBB11_18
-; SSE2-NEXT:  # %bb.17:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %r15d
-; SSE2-NEXT:  .LBB11_18:
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %r12b
-; SSE2-NEXT:    movl %r12d, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %r12b
-; SSE2-NEXT:    jno .LBB11_20
-; SSE2-NEXT:  # %bb.19:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %r12d
-; SSE2-NEXT:  .LBB11_20:
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %r13b
-; SSE2-NEXT:    movl %r13d, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %r13b
-; SSE2-NEXT:    movb %bl, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSE2-NEXT:    movb %dl, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSE2-NEXT:    jno .LBB11_22
-; SSE2-NEXT:  # %bb.21:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %r13d
-; SSE2-NEXT:  .LBB11_22:
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %dil
-; SSE2-NEXT:    movl %edi, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %dil
-; SSE2-NEXT:    jno .LBB11_24
-; SSE2-NEXT:  # %bb.23:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %edi
-; SSE2-NEXT:  .LBB11_24:
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %r8b
-; SSE2-NEXT:    movl %r8d, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %r8b
-; SSE2-NEXT:    jno .LBB11_26
-; SSE2-NEXT:  # %bb.25:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %r8d
-; SSE2-NEXT:  .LBB11_26:
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %bl
-; SSE2-NEXT:    movl %ebx, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %bl
-; SSE2-NEXT:    jno .LBB11_28
-; SSE2-NEXT:  # %bb.27:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %ebx
-; SSE2-NEXT:  .LBB11_28:
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %dl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
-; SSE2-NEXT:    movl %eax, %ecx
-; SSE2-NEXT:    addb %dl, %cl
-; SSE2-NEXT:    setns %cl
-; SSE2-NEXT:    addb %dl, %al
-; SSE2-NEXT:    movb %sil, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSE2-NEXT:    jno .LBB11_30
-; SSE2-NEXT:  # %bb.29:
-; SSE2-NEXT:    addb $127, %cl
-; SSE2-NEXT:    movl %ecx, %eax
-; SSE2-NEXT:  .LBB11_30:
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %dl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %sil
-; SSE2-NEXT:    movl %esi, %ecx
-; SSE2-NEXT:    addb %dl, %cl
-; SSE2-NEXT:    setns %cl
-; SSE2-NEXT:    addb %dl, %sil
-; SSE2-NEXT:    jno .LBB11_32
-; SSE2-NEXT:  # %bb.31:
-; SSE2-NEXT:    addb $127, %cl
-; SSE2-NEXT:    movl %ecx, %esi
-; SSE2-NEXT:  .LBB11_32:
-; SSE2-NEXT:    movzbl %sil, %ecx
-; SSE2-NEXT:    movd %ecx, %xmm0
-; SSE2-NEXT:    movzbl %al, %eax
-; SSE2-NEXT:    movd %eax, %xmm1
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3],xmm1[4],xmm0[4],xmm1[5],xmm0[5],xmm1[6],xmm0[6],xmm1[7],xmm0[7]
-; SSE2-NEXT:    movzbl %bl, %eax
-; SSE2-NEXT:    movd %eax, %xmm0
-; SSE2-NEXT:    movzbl %r8b, %eax
-; SSE2-NEXT:    movd %eax, %xmm2
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm2 = xmm2[0],xmm0[0],xmm2[1],xmm0[1],xmm2[2],xmm0[2],xmm2[3],xmm0[3],xmm2[4],xmm0[4],xmm2[5],xmm0[5],xmm2[6],xmm0[6],xmm2[7],xmm0[7]
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm2 = xmm2[0],xmm1[0],xmm2[1],xmm1[1],xmm2[2],xmm1[2],xmm2[3],xmm1[3]
-; SSE2-NEXT:    movzbl %dil, %eax
-; SSE2-NEXT:    movd %eax, %xmm0
-; SSE2-NEXT:    movzbl %r13b, %eax
-; SSE2-NEXT:    movd %eax, %xmm3
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm3 = xmm3[0],xmm0[0],xmm3[1],xmm0[1],xmm3[2],xmm0[2],xmm3[3],xmm0[3],xmm3[4],xmm0[4],xmm3[5],xmm0[5],xmm3[6],xmm0[6],xmm3[7],xmm0[7]
-; SSE2-NEXT:    movzbl %r12b, %eax
-; SSE2-NEXT:    movd %eax, %xmm0
-; SSE2-NEXT:    movzbl %r15b, %eax
-; SSE2-NEXT:    movd %eax, %xmm1
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3],xmm1[4],xmm0[4],xmm1[5],xmm0[5],xmm1[6],xmm0[6],xmm1[7],xmm0[7]
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm3[0],xmm1[1],xmm3[1],xmm1[2],xmm3[2],xmm1[3],xmm3[3]
-; SSE2-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm2[0],xmm1[1],xmm2[1]
-; SSE2-NEXT:    movzbl %r14b, %eax
-; SSE2-NEXT:    movd %eax, %xmm0
-; SSE2-NEXT:    movzbl %bpl, %eax
-; SSE2-NEXT:    movd %eax, %xmm2
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm2 = xmm2[0],xmm0[0],xmm2[1],xmm0[1],xmm2[2],xmm0[2],xmm2[3],xmm0[3],xmm2[4],xmm0[4],xmm2[5],xmm0[5],xmm2[6],xmm0[6],xmm2[7],xmm0[7]
-; SSE2-NEXT:    movzbl %r11b, %eax
-; SSE2-NEXT:    movd %eax, %xmm0
-; SSE2-NEXT:    movzbl %r10b, %eax
-; SSE2-NEXT:    movd %eax, %xmm3
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm3 = xmm3[0],xmm0[0],xmm3[1],xmm0[1],xmm3[2],xmm0[2],xmm3[3],xmm0[3],xmm3[4],xmm0[4],xmm3[5],xmm0[5],xmm3[6],xmm0[6],xmm3[7],xmm0[7]
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm3 = xmm3[0],xmm2[0],xmm3[1],xmm2[1],xmm3[2],xmm2[2],xmm3[3],xmm2[3]
-; SSE2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; SSE2-NEXT:    movd %eax, %xmm0
-; SSE2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; SSE2-NEXT:    movd %eax, %xmm2
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm2 = xmm2[0],xmm0[0],xmm2[1],xmm0[1],xmm2[2],xmm0[2],xmm2[3],xmm0[3],xmm2[4],xmm0[4],xmm2[5],xmm0[5],xmm2[6],xmm0[6],xmm2[7],xmm0[7]
-; SSE2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; SSE2-NEXT:    movd %eax, %xmm4
-; SSE2-NEXT:    movzbl %r9b, %eax
-; SSE2-NEXT:    movd %eax, %xmm0
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm4[0],xmm0[1],xmm4[1],xmm0[2],xmm4[2],xmm0[3],xmm4[3],xmm0[4],xmm4[4],xmm0[5],xmm4[5],xmm0[6],xmm4[6],xmm0[7],xmm4[7]
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1],xmm0[2],xmm2[2],xmm0[3],xmm2[3]
-; SSE2-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm3[0],xmm0[1],xmm3[1]
-; SSE2-NEXT:    punpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm1[0]
-; SSE2-NEXT:    popq %rbx
-; SSE2-NEXT:    popq %r12
-; SSE2-NEXT:    popq %r13
-; SSE2-NEXT:    popq %r14
-; SSE2-NEXT:    popq %r15
-; SSE2-NEXT:    popq %rbp
-; SSE2-NEXT:    retq
-;
-; SSSE3-LABEL: v12i8:
-; SSSE3:       # %bb.0:
-; SSSE3-NEXT:    pushq %rbp
-; SSSE3-NEXT:    pushq %r15
-; SSSE3-NEXT:    pushq %r14
-; SSSE3-NEXT:    pushq %r13
-; SSSE3-NEXT:    pushq %r12
-; SSSE3-NEXT:    pushq %rbx
-; SSSE3-NEXT:    movaps %xmm1, -{{[0-9]+}}(%rsp)
-; SSSE3-NEXT:    movaps %xmm0, -{{[0-9]+}}(%rsp)
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %r9b
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %sil
-; SSSE3-NEXT:    movl %r9d, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %r9b
-; SSSE3-NEXT:    jno .LBB11_2
-; SSSE3-NEXT:  # %bb.1:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %r9d
-; SSSE3-NEXT:  .LBB11_2:
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movl %esi, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %sil
-; SSSE3-NEXT:    jno .LBB11_4
-; SSSE3-NEXT:  # %bb.3:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %esi
-; SSSE3-NEXT:  .LBB11_4:
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %bl
-; SSSE3-NEXT:    movl %ebx, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %bl
-; SSSE3-NEXT:    jno .LBB11_6
-; SSSE3-NEXT:  # %bb.5:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %ebx
-; SSSE3-NEXT:  .LBB11_6:
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %dl
-; SSSE3-NEXT:    movl %edx, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %dl
-; SSSE3-NEXT:    jno .LBB11_8
-; SSSE3-NEXT:  # %bb.7:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %edx
-; SSSE3-NEXT:  .LBB11_8:
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %r10b
-; SSSE3-NEXT:    movl %r10d, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %r10b
-; SSSE3-NEXT:    jno .LBB11_10
-; SSSE3-NEXT:  # %bb.9:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %r10d
-; SSSE3-NEXT:  .LBB11_10:
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %r11b
-; SSSE3-NEXT:    movl %r11d, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %r11b
-; SSSE3-NEXT:    jno .LBB11_12
-; SSSE3-NEXT:  # %bb.11:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %r11d
-; SSSE3-NEXT:  .LBB11_12:
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %bpl
-; SSSE3-NEXT:    movl %ebp, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %bpl
-; SSSE3-NEXT:    jno .LBB11_14
-; SSSE3-NEXT:  # %bb.13:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %ebp
-; SSSE3-NEXT:  .LBB11_14:
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %r14b
-; SSSE3-NEXT:    movl %r14d, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %r14b
-; SSSE3-NEXT:    jno .LBB11_16
-; SSSE3-NEXT:  # %bb.15:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %r14d
-; SSSE3-NEXT:  .LBB11_16:
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %r15b
-; SSSE3-NEXT:    movl %r15d, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %r15b
-; SSSE3-NEXT:    jno .LBB11_18
-; SSSE3-NEXT:  # %bb.17:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %r15d
-; SSSE3-NEXT:  .LBB11_18:
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %r12b
-; SSSE3-NEXT:    movl %r12d, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %r12b
-; SSSE3-NEXT:    jno .LBB11_20
-; SSSE3-NEXT:  # %bb.19:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %r12d
-; SSSE3-NEXT:  .LBB11_20:
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %r13b
-; SSSE3-NEXT:    movl %r13d, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %r13b
-; SSSE3-NEXT:    movb %bl, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSSE3-NEXT:    movb %dl, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSSE3-NEXT:    jno .LBB11_22
-; SSSE3-NEXT:  # %bb.21:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %r13d
-; SSSE3-NEXT:  .LBB11_22:
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %dil
-; SSSE3-NEXT:    movl %edi, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %dil
-; SSSE3-NEXT:    jno .LBB11_24
-; SSSE3-NEXT:  # %bb.23:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %edi
-; SSSE3-NEXT:  .LBB11_24:
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %r8b
-; SSSE3-NEXT:    movl %r8d, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %r8b
-; SSSE3-NEXT:    jno .LBB11_26
-; SSSE3-NEXT:  # %bb.25:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %r8d
-; SSSE3-NEXT:  .LBB11_26:
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %bl
-; SSSE3-NEXT:    movl %ebx, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %bl
-; SSSE3-NEXT:    jno .LBB11_28
-; SSSE3-NEXT:  # %bb.27:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %ebx
-; SSSE3-NEXT:  .LBB11_28:
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %dl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %al
-; SSSE3-NEXT:    movl %eax, %ecx
-; SSSE3-NEXT:    addb %dl, %cl
-; SSSE3-NEXT:    setns %cl
-; SSSE3-NEXT:    addb %dl, %al
-; SSSE3-NEXT:    movb %sil, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSSE3-NEXT:    jno .LBB11_30
-; SSSE3-NEXT:  # %bb.29:
-; SSSE3-NEXT:    addb $127, %cl
-; SSSE3-NEXT:    movl %ecx, %eax
-; SSSE3-NEXT:  .LBB11_30:
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %dl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %sil
-; SSSE3-NEXT:    movl %esi, %ecx
-; SSSE3-NEXT:    addb %dl, %cl
-; SSSE3-NEXT:    setns %cl
-; SSSE3-NEXT:    addb %dl, %sil
-; SSSE3-NEXT:    jno .LBB11_32
-; SSSE3-NEXT:  # %bb.31:
-; SSSE3-NEXT:    addb $127, %cl
-; SSSE3-NEXT:    movl %ecx, %esi
-; SSSE3-NEXT:  .LBB11_32:
-; SSSE3-NEXT:    movzbl %sil, %ecx
-; SSSE3-NEXT:    movd %ecx, %xmm0
-; SSSE3-NEXT:    movzbl %al, %eax
-; SSSE3-NEXT:    movd %eax, %xmm1
-; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3],xmm1[4],xmm0[4],xmm1[5],xmm0[5],xmm1[6],xmm0[6],xmm1[7],xmm0[7]
-; SSSE3-NEXT:    movzbl %bl, %eax
-; SSSE3-NEXT:    movd %eax, %xmm0
-; SSSE3-NEXT:    movzbl %r8b, %eax
-; SSSE3-NEXT:    movd %eax, %xmm2
-; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm2 = xmm2[0],xmm0[0],xmm2[1],xmm0[1],xmm2[2],xmm0[2],xmm2[3],xmm0[3],xmm2[4],xmm0[4],xmm2[5],xmm0[5],xmm2[6],xmm0[6],xmm2[7],xmm0[7]
-; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm2 = xmm2[0],xmm1[0],xmm2[1],xmm1[1],xmm2[2],xmm1[2],xmm2[3],xmm1[3]
-; SSSE3-NEXT:    movzbl %dil, %eax
-; SSSE3-NEXT:    movd %eax, %xmm0
-; SSSE3-NEXT:    movzbl %r13b, %eax
-; SSSE3-NEXT:    movd %eax, %xmm3
-; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm3 = xmm3[0],xmm0[0],xmm3[1],xmm0[1],xmm3[2],xmm0[2],xmm3[3],xmm0[3],xmm3[4],xmm0[4],xmm3[5],xmm0[5],xmm3[6],xmm0[6],xmm3[7],xmm0[7]
-; SSSE3-NEXT:    movzbl %r12b, %eax
-; SSSE3-NEXT:    movd %eax, %xmm0
-; SSSE3-NEXT:    movzbl %r15b, %eax
-; SSSE3-NEXT:    movd %eax, %xmm1
-; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3],xmm1[4],xmm0[4],xmm1[5],xmm0[5],xmm1[6],xmm0[6],xmm1[7],xmm0[7]
-; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm3[0],xmm1[1],xmm3[1],xmm1[2],xmm3[2],xmm1[3],xmm3[3]
-; SSSE3-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm2[0],xmm1[1],xmm2[1]
-; SSSE3-NEXT:    movzbl %r14b, %eax
-; SSSE3-NEXT:    movd %eax, %xmm0
-; SSSE3-NEXT:    movzbl %bpl, %eax
-; SSSE3-NEXT:    movd %eax, %xmm2
-; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm2 = xmm2[0],xmm0[0],xmm2[1],xmm0[1],xmm2[2],xmm0[2],xmm2[3],xmm0[3],xmm2[4],xmm0[4],xmm2[5],xmm0[5],xmm2[6],xmm0[6],xmm2[7],xmm0[7]
-; SSSE3-NEXT:    movzbl %r11b, %eax
-; SSSE3-NEXT:    movd %eax, %xmm0
-; SSSE3-NEXT:    movzbl %r10b, %eax
-; SSSE3-NEXT:    movd %eax, %xmm3
-; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm3 = xmm3[0],xmm0[0],xmm3[1],xmm0[1],xmm3[2],xmm0[2],xmm3[3],xmm0[3],xmm3[4],xmm0[4],xmm3[5],xmm0[5],xmm3[6],xmm0[6],xmm3[7],xmm0[7]
-; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm3 = xmm3[0],xmm2[0],xmm3[1],xmm2[1],xmm3[2],xmm2[2],xmm3[3],xmm2[3]
-; SSSE3-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; SSSE3-NEXT:    movd %eax, %xmm0
-; SSSE3-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; SSSE3-NEXT:    movd %eax, %xmm2
-; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm2 = xmm2[0],xmm0[0],xmm2[1],xmm0[1],xmm2[2],xmm0[2],xmm2[3],xmm0[3],xmm2[4],xmm0[4],xmm2[5],xmm0[5],xmm2[6],xmm0[6],xmm2[7],xmm0[7]
-; SSSE3-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; SSSE3-NEXT:    movd %eax, %xmm4
-; SSSE3-NEXT:    movzbl %r9b, %eax
-; SSSE3-NEXT:    movd %eax, %xmm0
-; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm4[0],xmm0[1],xmm4[1],xmm0[2],xmm4[2],xmm0[3],xmm4[3],xmm0[4],xmm4[4],xmm0[5],xmm4[5],xmm0[6],xmm4[6],xmm0[7],xmm4[7]
-; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1],xmm0[2],xmm2[2],xmm0[3],xmm2[3]
-; SSSE3-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm3[0],xmm0[1],xmm3[1]
-; SSSE3-NEXT:    punpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm1[0]
-; SSSE3-NEXT:    popq %rbx
-; SSSE3-NEXT:    popq %r12
-; SSSE3-NEXT:    popq %r13
-; SSSE3-NEXT:    popq %r14
-; SSSE3-NEXT:    popq %r15
-; SSSE3-NEXT:    popq %rbp
-; SSSE3-NEXT:    retq
-;
-; SSE41-LABEL: v12i8:
-; SSE41:       # %bb.0:
-; SSE41-NEXT:    pextrb $15, %xmm1, %ecx
-; SSE41-NEXT:    pextrb $15, %xmm0, %edx
-; SSE41-NEXT:    movl %edx, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %dl
-; SSE41-NEXT:    jno .LBB11_2
-; SSE41-NEXT:  # %bb.1:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %edx
-; SSE41-NEXT:  .LBB11_2:
-; SSE41-NEXT:    pextrb $14, %xmm1, %ecx
-; SSE41-NEXT:    pextrb $14, %xmm0, %r11d
-; SSE41-NEXT:    movl %r11d, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %r11b
-; SSE41-NEXT:    jno .LBB11_4
-; SSE41-NEXT:  # %bb.3:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %r11d
-; SSE41-NEXT:  .LBB11_4:
-; SSE41-NEXT:    pextrb $13, %xmm1, %ecx
-; SSE41-NEXT:    pextrb $13, %xmm0, %edi
-; SSE41-NEXT:    movl %edi, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %dil
-; SSE41-NEXT:    jno .LBB11_6
-; SSE41-NEXT:  # %bb.5:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %edi
-; SSE41-NEXT:  .LBB11_6:
-; SSE41-NEXT:    pushq %rbp
-; SSE41-NEXT:    pushq %r15
-; SSE41-NEXT:    pushq %r14
-; SSE41-NEXT:    pushq %r13
-; SSE41-NEXT:    pushq %r12
-; SSE41-NEXT:    pushq %rbx
-; SSE41-NEXT:    pextrb $12, %xmm1, %ecx
-; SSE41-NEXT:    pextrb $12, %xmm0, %r14d
-; SSE41-NEXT:    movl %r14d, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %r14b
-; SSE41-NEXT:    jno .LBB11_8
-; SSE41-NEXT:  # %bb.7:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %r14d
-; SSE41-NEXT:  .LBB11_8:
-; SSE41-NEXT:    pextrb $11, %xmm1, %ecx
-; SSE41-NEXT:    pextrb $11, %xmm0, %ebp
-; SSE41-NEXT:    movl %ebp, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %bpl
-; SSE41-NEXT:    jno .LBB11_10
-; SSE41-NEXT:  # %bb.9:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %ebp
-; SSE41-NEXT:  .LBB11_10:
-; SSE41-NEXT:    pextrb $10, %xmm1, %ecx
-; SSE41-NEXT:    pextrb $10, %xmm0, %r15d
-; SSE41-NEXT:    movl %r15d, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %r15b
-; SSE41-NEXT:    jno .LBB11_12
-; SSE41-NEXT:  # %bb.11:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %r15d
-; SSE41-NEXT:  .LBB11_12:
-; SSE41-NEXT:    pextrb $9, %xmm1, %ecx
-; SSE41-NEXT:    pextrb $9, %xmm0, %r12d
-; SSE41-NEXT:    movl %r12d, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %r12b
-; SSE41-NEXT:    jno .LBB11_14
-; SSE41-NEXT:  # %bb.13:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %r12d
-; SSE41-NEXT:  .LBB11_14:
-; SSE41-NEXT:    pextrb $8, %xmm1, %ecx
-; SSE41-NEXT:    pextrb $8, %xmm0, %r13d
-; SSE41-NEXT:    movl %r13d, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %r13b
-; SSE41-NEXT:    jno .LBB11_16
-; SSE41-NEXT:  # %bb.15:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %r13d
-; SSE41-NEXT:  .LBB11_16:
-; SSE41-NEXT:    pextrb $7, %xmm1, %ecx
-; SSE41-NEXT:    pextrb $7, %xmm0, %r10d
-; SSE41-NEXT:    movl %r10d, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %r10b
-; SSE41-NEXT:    movl %ebp, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE41-NEXT:    jno .LBB11_18
-; SSE41-NEXT:  # %bb.17:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %r10d
-; SSE41-NEXT:  .LBB11_18:
-; SSE41-NEXT:    pextrb $6, %xmm1, %ecx
-; SSE41-NEXT:    pextrb $6, %xmm0, %r9d
-; SSE41-NEXT:    movl %r9d, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %r9b
-; SSE41-NEXT:    jno .LBB11_20
-; SSE41-NEXT:  # %bb.19:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %r9d
-; SSE41-NEXT:  .LBB11_20:
-; SSE41-NEXT:    pextrb $5, %xmm1, %ecx
-; SSE41-NEXT:    pextrb $5, %xmm0, %ebp
-; SSE41-NEXT:    movl %ebp, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %bpl
-; SSE41-NEXT:    movl %edi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE41-NEXT:    jno .LBB11_22
-; SSE41-NEXT:  # %bb.21:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %ebp
-; SSE41-NEXT:  .LBB11_22:
-; SSE41-NEXT:    pextrb $4, %xmm1, %ecx
-; SSE41-NEXT:    pextrb $4, %xmm0, %edi
-; SSE41-NEXT:    movl %edi, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %dil
-; SSE41-NEXT:    movl %edx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE41-NEXT:    jno .LBB11_24
-; SSE41-NEXT:  # %bb.23:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %edi
-; SSE41-NEXT:  .LBB11_24:
-; SSE41-NEXT:    pextrb $3, %xmm1, %edx
-; SSE41-NEXT:    pextrb $3, %xmm0, %eax
-; SSE41-NEXT:    movl %eax, %ecx
-; SSE41-NEXT:    addb %dl, %cl
-; SSE41-NEXT:    setns %cl
-; SSE41-NEXT:    addb %dl, %al
-; SSE41-NEXT:    jno .LBB11_26
-; SSE41-NEXT:  # %bb.25:
-; SSE41-NEXT:    addb $127, %cl
-; SSE41-NEXT:    movl %ecx, %eax
-; SSE41-NEXT:  .LBB11_26:
-; SSE41-NEXT:    pextrb $2, %xmm1, %ebx
-; SSE41-NEXT:    pextrb $2, %xmm0, %ecx
-; SSE41-NEXT:    movl %ecx, %edx
-; SSE41-NEXT:    addb %bl, %dl
-; SSE41-NEXT:    setns %dl
-; SSE41-NEXT:    addb %bl, %cl
-; SSE41-NEXT:    jno .LBB11_28
-; SSE41-NEXT:  # %bb.27:
-; SSE41-NEXT:    addb $127, %dl
-; SSE41-NEXT:    movl %edx, %ecx
-; SSE41-NEXT:  .LBB11_28:
-; SSE41-NEXT:    pextrb $0, %xmm1, %esi
-; SSE41-NEXT:    pextrb $0, %xmm0, %edx
-; SSE41-NEXT:    movl %edx, %ebx
-; SSE41-NEXT:    addb %sil, %bl
-; SSE41-NEXT:    setns %bl
-; SSE41-NEXT:    addb %sil, %dl
-; SSE41-NEXT:    jno .LBB11_30
-; SSE41-NEXT:  # %bb.29:
-; SSE41-NEXT:    addb $127, %bl
-; SSE41-NEXT:    movl %ebx, %edx
-; SSE41-NEXT:  .LBB11_30:
-; SSE41-NEXT:    pextrb $1, %xmm1, %esi
-; SSE41-NEXT:    pextrb $1, %xmm0, %r8d
-; SSE41-NEXT:    movl %r8d, %ebx
-; SSE41-NEXT:    addb %sil, %bl
-; SSE41-NEXT:    setns %bl
-; SSE41-NEXT:    addb %sil, %r8b
-; SSE41-NEXT:    jno .LBB11_32
-; SSE41-NEXT:  # %bb.31:
-; SSE41-NEXT:    addb $127, %bl
-; SSE41-NEXT:    movl %ebx, %r8d
-; SSE41-NEXT:  .LBB11_32:
-; SSE41-NEXT:    movzbl %dl, %edx
-; SSE41-NEXT:    movd %edx, %xmm0
-; SSE41-NEXT:    movzbl %r8b, %edx
-; SSE41-NEXT:    pinsrb $1, %edx, %xmm0
-; SSE41-NEXT:    movzbl %cl, %ecx
-; SSE41-NEXT:    pinsrb $2, %ecx, %xmm0
-; SSE41-NEXT:    movzbl %al, %eax
-; SSE41-NEXT:    pinsrb $3, %eax, %xmm0
-; SSE41-NEXT:    movzbl %dil, %eax
-; SSE41-NEXT:    pinsrb $4, %eax, %xmm0
-; SSE41-NEXT:    movzbl %bpl, %eax
-; SSE41-NEXT:    pinsrb $5, %eax, %xmm0
-; SSE41-NEXT:    movzbl %r9b, %eax
-; SSE41-NEXT:    pinsrb $6, %eax, %xmm0
-; SSE41-NEXT:    movzbl %r10b, %eax
-; SSE41-NEXT:    pinsrb $7, %eax, %xmm0
-; SSE41-NEXT:    movzbl %r13b, %eax
-; SSE41-NEXT:    pinsrb $8, %eax, %xmm0
-; SSE41-NEXT:    movzbl %r12b, %eax
-; SSE41-NEXT:    pinsrb $9, %eax, %xmm0
-; SSE41-NEXT:    movzbl %r15b, %eax
-; SSE41-NEXT:    pinsrb $10, %eax, %xmm0
-; SSE41-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; SSE41-NEXT:    pinsrb $11, %eax, %xmm0
-; SSE41-NEXT:    movzbl %r14b, %eax
-; SSE41-NEXT:    pinsrb $12, %eax, %xmm0
-; SSE41-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; SSE41-NEXT:    pinsrb $13, %eax, %xmm0
-; SSE41-NEXT:    movzbl %r11b, %eax
-; SSE41-NEXT:    pinsrb $14, %eax, %xmm0
-; SSE41-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; SSE41-NEXT:    pinsrb $15, %eax, %xmm0
-; SSE41-NEXT:    popq %rbx
-; SSE41-NEXT:    popq %r12
-; SSE41-NEXT:    popq %r13
-; SSE41-NEXT:    popq %r14
-; SSE41-NEXT:    popq %r15
-; SSE41-NEXT:    popq %rbp
-; SSE41-NEXT:    retq
+; SSE-LABEL: v12i8:
+; SSE:       # %bb.0:
+; SSE-NEXT:    paddsb %xmm1, %xmm0
+; SSE-NEXT:    retq
 ;
 ; AVX-LABEL: v12i8:
 ; AVX:       # %bb.0:
-; AVX-NEXT:    vpextrb $15, %xmm1, %ecx
-; AVX-NEXT:    vpextrb $15, %xmm0, %edx
-; AVX-NEXT:    movl %edx, %eax
-; AVX-NEXT:    addb %cl, %al
-; AVX-NEXT:    setns %al
-; AVX-NEXT:    addb %cl, %dl
-; AVX-NEXT:    jno .LBB11_2
-; AVX-NEXT:  # %bb.1:
-; AVX-NEXT:    addb $127, %al
-; AVX-NEXT:    movl %eax, %edx
-; AVX-NEXT:  .LBB11_2:
-; AVX-NEXT:    vpextrb $14, %xmm1, %ecx
-; AVX-NEXT:    vpextrb $14, %xmm0, %r11d
-; AVX-NEXT:    movl %r11d, %eax
-; AVX-NEXT:    addb %cl, %al
-; AVX-NEXT:    setns %al
-; AVX-NEXT:    addb %cl, %r11b
-; AVX-NEXT:    jno .LBB11_4
-; AVX-NEXT:  # %bb.3:
-; AVX-NEXT:    addb $127, %al
-; AVX-NEXT:    movl %eax, %r11d
-; AVX-NEXT:  .LBB11_4:
-; AVX-NEXT:    vpextrb $13, %xmm1, %ecx
-; AVX-NEXT:    vpextrb $13, %xmm0, %edi
-; AVX-NEXT:    movl %edi, %eax
-; AVX-NEXT:    addb %cl, %al
-; AVX-NEXT:    setns %al
-; AVX-NEXT:    addb %cl, %dil
-; AVX-NEXT:    jno .LBB11_6
-; AVX-NEXT:  # %bb.5:
-; AVX-NEXT:    addb $127, %al
-; AVX-NEXT:    movl %eax, %edi
-; AVX-NEXT:  .LBB11_6:
-; AVX-NEXT:    pushq %rbp
-; AVX-NEXT:    pushq %r15
-; AVX-NEXT:    pushq %r14
-; AVX-NEXT:    pushq %r13
-; AVX-NEXT:    pushq %r12
-; AVX-NEXT:    pushq %rbx
-; AVX-NEXT:    vpextrb $12, %xmm1, %ecx
-; AVX-NEXT:    vpextrb $12, %xmm0, %r14d
-; AVX-NEXT:    movl %r14d, %eax
-; AVX-NEXT:    addb %cl, %al
-; AVX-NEXT:    setns %al
-; AVX-NEXT:    addb %cl, %r14b
-; AVX-NEXT:    jno .LBB11_8
-; AVX-NEXT:  # %bb.7:
-; AVX-NEXT:    addb $127, %al
-; AVX-NEXT:    movl %eax, %r14d
-; AVX-NEXT:  .LBB11_8:
-; AVX-NEXT:    vpextrb $11, %xmm1, %ecx
-; AVX-NEXT:    vpextrb $11, %xmm0, %ebp
-; AVX-NEXT:    movl %ebp, %eax
-; AVX-NEXT:    addb %cl, %al
-; AVX-NEXT:    setns %al
-; AVX-NEXT:    addb %cl, %bpl
-; AVX-NEXT:    jno .LBB11_10
-; AVX-NEXT:  # %bb.9:
-; AVX-NEXT:    addb $127, %al
-; AVX-NEXT:    movl %eax, %ebp
-; AVX-NEXT:  .LBB11_10:
-; AVX-NEXT:    vpextrb $10, %xmm1, %ecx
-; AVX-NEXT:    vpextrb $10, %xmm0, %r15d
-; AVX-NEXT:    movl %r15d, %eax
-; AVX-NEXT:    addb %cl, %al
-; AVX-NEXT:    setns %al
-; AVX-NEXT:    addb %cl, %r15b
-; AVX-NEXT:    jno .LBB11_12
-; AVX-NEXT:  # %bb.11:
-; AVX-NEXT:    addb $127, %al
-; AVX-NEXT:    movl %eax, %r15d
-; AVX-NEXT:  .LBB11_12:
-; AVX-NEXT:    vpextrb $9, %xmm1, %ecx
-; AVX-NEXT:    vpextrb $9, %xmm0, %r12d
-; AVX-NEXT:    movl %r12d, %eax
-; AVX-NEXT:    addb %cl, %al
-; AVX-NEXT:    setns %al
-; AVX-NEXT:    addb %cl, %r12b
-; AVX-NEXT:    jno .LBB11_14
-; AVX-NEXT:  # %bb.13:
-; AVX-NEXT:    addb $127, %al
-; AVX-NEXT:    movl %eax, %r12d
-; AVX-NEXT:  .LBB11_14:
-; AVX-NEXT:    vpextrb $8, %xmm1, %ecx
-; AVX-NEXT:    vpextrb $8, %xmm0, %r13d
-; AVX-NEXT:    movl %r13d, %eax
-; AVX-NEXT:    addb %cl, %al
-; AVX-NEXT:    setns %al
-; AVX-NEXT:    addb %cl, %r13b
-; AVX-NEXT:    jno .LBB11_16
-; AVX-NEXT:  # %bb.15:
-; AVX-NEXT:    addb $127, %al
-; AVX-NEXT:    movl %eax, %r13d
-; AVX-NEXT:  .LBB11_16:
-; AVX-NEXT:    vpextrb $7, %xmm1, %ecx
-; AVX-NEXT:    vpextrb $7, %xmm0, %r10d
-; AVX-NEXT:    movl %r10d, %eax
-; AVX-NEXT:    addb %cl, %al
-; AVX-NEXT:    setns %al
-; AVX-NEXT:    addb %cl, %r10b
-; AVX-NEXT:    movl %ebp, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX-NEXT:    jno .LBB11_18
-; AVX-NEXT:  # %bb.17:
-; AVX-NEXT:    addb $127, %al
-; AVX-NEXT:    movl %eax, %r10d
-; AVX-NEXT:  .LBB11_18:
-; AVX-NEXT:    vpextrb $6, %xmm1, %ecx
-; AVX-NEXT:    vpextrb $6, %xmm0, %r9d
-; AVX-NEXT:    movl %r9d, %eax
-; AVX-NEXT:    addb %cl, %al
-; AVX-NEXT:    setns %al
-; AVX-NEXT:    addb %cl, %r9b
-; AVX-NEXT:    jno .LBB11_20
-; AVX-NEXT:  # %bb.19:
-; AVX-NEXT:    addb $127, %al
-; AVX-NEXT:    movl %eax, %r9d
-; AVX-NEXT:  .LBB11_20:
-; AVX-NEXT:    vpextrb $5, %xmm1, %ecx
-; AVX-NEXT:    vpextrb $5, %xmm0, %ebp
-; AVX-NEXT:    movl %ebp, %eax
-; AVX-NEXT:    addb %cl, %al
-; AVX-NEXT:    setns %al
-; AVX-NEXT:    addb %cl, %bpl
-; AVX-NEXT:    movl %edi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX-NEXT:    jno .LBB11_22
-; AVX-NEXT:  # %bb.21:
-; AVX-NEXT:    addb $127, %al
-; AVX-NEXT:    movl %eax, %ebp
-; AVX-NEXT:  .LBB11_22:
-; AVX-NEXT:    vpextrb $4, %xmm1, %ecx
-; AVX-NEXT:    vpextrb $4, %xmm0, %edi
-; AVX-NEXT:    movl %edi, %eax
-; AVX-NEXT:    addb %cl, %al
-; AVX-NEXT:    setns %al
-; AVX-NEXT:    addb %cl, %dil
-; AVX-NEXT:    movl %edx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX-NEXT:    jno .LBB11_24
-; AVX-NEXT:  # %bb.23:
-; AVX-NEXT:    addb $127, %al
-; AVX-NEXT:    movl %eax, %edi
-; AVX-NEXT:  .LBB11_24:
-; AVX-NEXT:    vpextrb $3, %xmm1, %edx
-; AVX-NEXT:    vpextrb $3, %xmm0, %eax
-; AVX-NEXT:    movl %eax, %ecx
-; AVX-NEXT:    addb %dl, %cl
-; AVX-NEXT:    setns %cl
-; AVX-NEXT:    addb %dl, %al
-; AVX-NEXT:    jno .LBB11_26
-; AVX-NEXT:  # %bb.25:
-; AVX-NEXT:    addb $127, %cl
-; AVX-NEXT:    movl %ecx, %eax
-; AVX-NEXT:  .LBB11_26:
-; AVX-NEXT:    vpextrb $2, %xmm1, %ebx
-; AVX-NEXT:    vpextrb $2, %xmm0, %ecx
-; AVX-NEXT:    movl %ecx, %edx
-; AVX-NEXT:    addb %bl, %dl
-; AVX-NEXT:    setns %dl
-; AVX-NEXT:    addb %bl, %cl
-; AVX-NEXT:    jno .LBB11_28
-; AVX-NEXT:  # %bb.27:
-; AVX-NEXT:    addb $127, %dl
-; AVX-NEXT:    movl %edx, %ecx
-; AVX-NEXT:  .LBB11_28:
-; AVX-NEXT:    vpextrb $0, %xmm1, %esi
-; AVX-NEXT:    vpextrb $0, %xmm0, %edx
-; AVX-NEXT:    movl %edx, %ebx
-; AVX-NEXT:    addb %sil, %bl
-; AVX-NEXT:    setns %bl
-; AVX-NEXT:    addb %sil, %dl
-; AVX-NEXT:    jno .LBB11_30
-; AVX-NEXT:  # %bb.29:
-; AVX-NEXT:    addb $127, %bl
-; AVX-NEXT:    movl %ebx, %edx
-; AVX-NEXT:  .LBB11_30:
-; AVX-NEXT:    vpextrb $1, %xmm1, %esi
-; AVX-NEXT:    vpextrb $1, %xmm0, %r8d
-; AVX-NEXT:    movl %r8d, %ebx
-; AVX-NEXT:    addb %sil, %bl
-; AVX-NEXT:    setns %bl
-; AVX-NEXT:    addb %sil, %r8b
-; AVX-NEXT:    jno .LBB11_32
-; AVX-NEXT:  # %bb.31:
-; AVX-NEXT:    addb $127, %bl
-; AVX-NEXT:    movl %ebx, %r8d
-; AVX-NEXT:  .LBB11_32:
-; AVX-NEXT:    movzbl %dl, %edx
-; AVX-NEXT:    vmovd %edx, %xmm0
-; AVX-NEXT:    movzbl %r8b, %edx
-; AVX-NEXT:    vpinsrb $1, %edx, %xmm0, %xmm0
-; AVX-NEXT:    movzbl %cl, %ecx
-; AVX-NEXT:    vpinsrb $2, %ecx, %xmm0, %xmm0
-; AVX-NEXT:    movzbl %al, %eax
-; AVX-NEXT:    vpinsrb $3, %eax, %xmm0, %xmm0
-; AVX-NEXT:    movzbl %dil, %eax
-; AVX-NEXT:    vpinsrb $4, %eax, %xmm0, %xmm0
-; AVX-NEXT:    movzbl %bpl, %eax
-; AVX-NEXT:    vpinsrb $5, %eax, %xmm0, %xmm0
-; AVX-NEXT:    movzbl %r9b, %eax
-; AVX-NEXT:    vpinsrb $6, %eax, %xmm0, %xmm0
-; AVX-NEXT:    movzbl %r10b, %eax
-; AVX-NEXT:    vpinsrb $7, %eax, %xmm0, %xmm0
-; AVX-NEXT:    movzbl %r13b, %eax
-; AVX-NEXT:    vpinsrb $8, %eax, %xmm0, %xmm0
-; AVX-NEXT:    movzbl %r12b, %eax
-; AVX-NEXT:    vpinsrb $9, %eax, %xmm0, %xmm0
-; AVX-NEXT:    movzbl %r15b, %eax
-; AVX-NEXT:    vpinsrb $10, %eax, %xmm0, %xmm0
-; AVX-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX-NEXT:    vpinsrb $11, %eax, %xmm0, %xmm0
-; AVX-NEXT:    movzbl %r14b, %eax
-; AVX-NEXT:    vpinsrb $12, %eax, %xmm0, %xmm0
-; AVX-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX-NEXT:    vpinsrb $13, %eax, %xmm0, %xmm0
-; AVX-NEXT:    movzbl %r11b, %eax
-; AVX-NEXT:    vpinsrb $14, %eax, %xmm0, %xmm0
-; AVX-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX-NEXT:    vpinsrb $15, %eax, %xmm0, %xmm0
-; AVX-NEXT:    popq %rbx
-; AVX-NEXT:    popq %r12
-; AVX-NEXT:    popq %r13
-; AVX-NEXT:    popq %r14
-; AVX-NEXT:    popq %r15
-; AVX-NEXT:    popq %rbp
+; AVX-NEXT:    vpaddsb %xmm1, %xmm0, %xmm0
 ; AVX-NEXT:    retq
   %z = call <12 x i8> @llvm.sadd.sat.v12i8(<12 x i8> %x, <12 x i8> %y)
   ret <12 x i8> %z
 }
 
 define void @v12i16(<12 x i16>* %px, <12 x i16>* %py, <12 x i16>* %pz) nounwind {
-; SSE2-LABEL: v12i16:
-; SSE2:       # %bb.0:
-; SSE2-NEXT:    pushq %rbp
-; SSE2-NEXT:    pushq %r15
-; SSE2-NEXT:    pushq %r14
-; SSE2-NEXT:    pushq %r13
-; SSE2-NEXT:    pushq %r12
-; SSE2-NEXT:    pushq %rbx
-; SSE2-NEXT:    movq %rdx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill
-; SSE2-NEXT:    movdqa (%rdi), %xmm0
-; SSE2-NEXT:    movdqa 16(%rdi), %xmm2
-; SSE2-NEXT:    movdqa (%rsi), %xmm1
-; SSE2-NEXT:    movdqa 16(%rsi), %xmm3
-; SSE2-NEXT:    pextrw $3, %xmm3, %eax
-; SSE2-NEXT:    pextrw $3, %xmm2, %edx
-; SSE2-NEXT:    xorl %ecx, %ecx
-; SSE2-NEXT:    movl %edx, %esi
-; SSE2-NEXT:    addw %ax, %si
-; SSE2-NEXT:    setns %cl
-; SSE2-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSE2-NEXT:    addw %ax, %dx
-; SSE2-NEXT:    cmovol %ecx, %edx
-; SSE2-NEXT:    pextrw $2, %xmm3, %eax
-; SSE2-NEXT:    pextrw $2, %xmm2, %r9d
-; SSE2-NEXT:    xorl %ecx, %ecx
-; SSE2-NEXT:    movl %r9d, %esi
-; SSE2-NEXT:    addw %ax, %si
-; SSE2-NEXT:    setns %cl
-; SSE2-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSE2-NEXT:    addw %ax, %r9w
-; SSE2-NEXT:    cmovol %ecx, %r9d
-; SSE2-NEXT:    movd %xmm3, %eax
-; SSE2-NEXT:    movd %xmm2, %r10d
-; SSE2-NEXT:    xorl %ecx, %ecx
-; SSE2-NEXT:    movl %r10d, %esi
-; SSE2-NEXT:    addw %ax, %si
-; SSE2-NEXT:    setns %cl
-; SSE2-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSE2-NEXT:    addw %ax, %r10w
-; SSE2-NEXT:    cmovol %ecx, %r10d
-; SSE2-NEXT:    pextrw $1, %xmm3, %eax
-; SSE2-NEXT:    pextrw $1, %xmm2, %r11d
-; SSE2-NEXT:    xorl %ecx, %ecx
-; SSE2-NEXT:    movl %r11d, %esi
-; SSE2-NEXT:    addw %ax, %si
-; SSE2-NEXT:    setns %cl
-; SSE2-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSE2-NEXT:    addw %ax, %r11w
-; SSE2-NEXT:    cmovol %ecx, %r11d
-; SSE2-NEXT:    movd %xmm1, %eax
-; SSE2-NEXT:    movd %xmm0, %r14d
-; SSE2-NEXT:    xorl %ecx, %ecx
-; SSE2-NEXT:    movl %r14d, %esi
-; SSE2-NEXT:    addw %ax, %si
-; SSE2-NEXT:    setns %cl
-; SSE2-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSE2-NEXT:    addw %ax, %r14w
-; SSE2-NEXT:    cmovol %ecx, %r14d
-; SSE2-NEXT:    pextrw $1, %xmm1, %eax
-; SSE2-NEXT:    pextrw $1, %xmm0, %r15d
-; SSE2-NEXT:    xorl %esi, %esi
-; SSE2-NEXT:    movl %r15d, %edi
-; SSE2-NEXT:    addw %ax, %di
-; SSE2-NEXT:    setns %sil
-; SSE2-NEXT:    addl $32767, %esi # imm = 0x7FFF
-; SSE2-NEXT:    addw %ax, %r15w
-; SSE2-NEXT:    cmovol %esi, %r15d
-; SSE2-NEXT:    pextrw $2, %xmm1, %eax
-; SSE2-NEXT:    pextrw $2, %xmm0, %r12d
-; SSE2-NEXT:    xorl %edi, %edi
-; SSE2-NEXT:    movl %r12d, %ebx
-; SSE2-NEXT:    addw %ax, %bx
-; SSE2-NEXT:    setns %dil
-; SSE2-NEXT:    addl $32767, %edi # imm = 0x7FFF
-; SSE2-NEXT:    addw %ax, %r12w
-; SSE2-NEXT:    cmovol %edi, %r12d
-; SSE2-NEXT:    pextrw $3, %xmm1, %eax
-; SSE2-NEXT:    pextrw $3, %xmm0, %r13d
-; SSE2-NEXT:    xorl %ebx, %ebx
-; SSE2-NEXT:    movl %r13d, %ebp
-; SSE2-NEXT:    addw %ax, %bp
-; SSE2-NEXT:    setns %bl
-; SSE2-NEXT:    addl $32767, %ebx # imm = 0x7FFF
-; SSE2-NEXT:    addw %ax, %r13w
-; SSE2-NEXT:    cmovol %ebx, %r13d
-; SSE2-NEXT:    pextrw $4, %xmm1, %eax
-; SSE2-NEXT:    pextrw $4, %xmm0, %ebx
-; SSE2-NEXT:    xorl %ebp, %ebp
-; SSE2-NEXT:    movl %ebx, %ecx
-; SSE2-NEXT:    addw %ax, %cx
-; SSE2-NEXT:    setns %bpl
-; SSE2-NEXT:    addl $32767, %ebp # imm = 0x7FFF
-; SSE2-NEXT:    addw %ax, %bx
-; SSE2-NEXT:    cmovol %ebp, %ebx
-; SSE2-NEXT:    pextrw $5, %xmm1, %eax
-; SSE2-NEXT:    pextrw $5, %xmm0, %ebp
-; SSE2-NEXT:    xorl %ecx, %ecx
-; SSE2-NEXT:    movl %ebp, %esi
-; SSE2-NEXT:    addw %ax, %si
-; SSE2-NEXT:    setns %cl
-; SSE2-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSE2-NEXT:    addw %ax, %bp
-; SSE2-NEXT:    cmovol %ecx, %ebp
-; SSE2-NEXT:    pextrw $6, %xmm1, %ecx
-; SSE2-NEXT:    pextrw $6, %xmm0, %eax
-; SSE2-NEXT:    xorl %esi, %esi
-; SSE2-NEXT:    movl %eax, %edi
-; SSE2-NEXT:    addw %cx, %di
-; SSE2-NEXT:    setns %sil
-; SSE2-NEXT:    addl $32767, %esi # imm = 0x7FFF
-; SSE2-NEXT:    addw %cx, %ax
-; SSE2-NEXT:    cmovol %esi, %eax
-; SSE2-NEXT:    pextrw $7, %xmm1, %ecx
-; SSE2-NEXT:    pextrw $7, %xmm0, %esi
-; SSE2-NEXT:    xorl %edi, %edi
-; SSE2-NEXT:    movl %esi, %r8d
-; SSE2-NEXT:    addw %cx, %r8w
-; SSE2-NEXT:    setns %dil
-; SSE2-NEXT:    addl $32767, %edi # imm = 0x7FFF
-; SSE2-NEXT:    addw %cx, %si
-; SSE2-NEXT:    cmovol %edi, %esi
-; SSE2-NEXT:    movd %esi, %xmm8
-; SSE2-NEXT:    movd %eax, %xmm1
-; SSE2-NEXT:    movd %ebp, %xmm2
-; SSE2-NEXT:    movd %ebx, %xmm3
-; SSE2-NEXT:    movd %r13d, %xmm4
-; SSE2-NEXT:    movd %r12d, %xmm5
-; SSE2-NEXT:    movd %r15d, %xmm6
-; SSE2-NEXT:    movd %r14d, %xmm7
-; SSE2-NEXT:    movd %r10d, %xmm0
-; SSE2-NEXT:    pinsrw $1, %r11d, %xmm0
-; SSE2-NEXT:    pinsrw $2, %r9d, %xmm0
-; SSE2-NEXT:    pinsrw $3, %edx, %xmm0
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm8[0],xmm1[1],xmm8[1],xmm1[2],xmm8[2],xmm1[3],xmm8[3]
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm3 = xmm3[0],xmm2[0],xmm3[1],xmm2[1],xmm3[2],xmm2[2],xmm3[3],xmm2[3]
-; SSE2-NEXT:    punpckldq {{.*#+}} xmm3 = xmm3[0],xmm1[0],xmm3[1],xmm1[1]
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm5 = xmm5[0],xmm4[0],xmm5[1],xmm4[1],xmm5[2],xmm4[2],xmm5[3],xmm4[3]
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm7 = xmm7[0],xmm6[0],xmm7[1],xmm6[1],xmm7[2],xmm6[2],xmm7[3],xmm6[3]
-; SSE2-NEXT:    punpckldq {{.*#+}} xmm7 = xmm7[0],xmm5[0],xmm7[1],xmm5[1]
-; SSE2-NEXT:    punpcklqdq {{.*#+}} xmm7 = xmm7[0],xmm3[0]
-; SSE2-NEXT:    movq {{[-0-9]+}}(%r{{[sb]}}p), %rax # 8-byte Reload
-; SSE2-NEXT:    movq %xmm0, 16(%rax)
-; SSE2-NEXT:    movdqa %xmm7, (%rax)
-; SSE2-NEXT:    popq %rbx
-; SSE2-NEXT:    popq %r12
-; SSE2-NEXT:    popq %r13
-; SSE2-NEXT:    popq %r14
-; SSE2-NEXT:    popq %r15
-; SSE2-NEXT:    popq %rbp
-; SSE2-NEXT:    retq
+; SSE-LABEL: v12i16:
+; SSE:       # %bb.0:
+; SSE-NEXT:    movdqa (%rdi), %xmm0
+; SSE-NEXT:    movdqa 16(%rdi), %xmm1
+; SSE-NEXT:    paddsw (%rsi), %xmm0
+; SSE-NEXT:    paddsw 16(%rsi), %xmm1
+; SSE-NEXT:    movq %xmm1, 16(%rdx)
+; SSE-NEXT:    movdqa %xmm0, (%rdx)
+; SSE-NEXT:    retq
 ;
-; SSSE3-LABEL: v12i16:
-; SSSE3:       # %bb.0:
-; SSSE3-NEXT:    pushq %rbp
-; SSSE3-NEXT:    pushq %r15
-; SSSE3-NEXT:    pushq %r14
-; SSSE3-NEXT:    pushq %r13
-; SSSE3-NEXT:    pushq %r12
-; SSSE3-NEXT:    pushq %rbx
-; SSSE3-NEXT:    movq %rdx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill
-; SSSE3-NEXT:    movdqa (%rdi), %xmm0
-; SSSE3-NEXT:    movdqa 16(%rdi), %xmm2
-; SSSE3-NEXT:    movdqa (%rsi), %xmm1
-; SSSE3-NEXT:    movdqa 16(%rsi), %xmm3
-; SSSE3-NEXT:    pextrw $3, %xmm3, %eax
-; SSSE3-NEXT:    pextrw $3, %xmm2, %edx
-; SSSE3-NEXT:    xorl %ecx, %ecx
-; SSSE3-NEXT:    movl %edx, %esi
-; SSSE3-NEXT:    addw %ax, %si
-; SSSE3-NEXT:    setns %cl
-; SSSE3-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSSE3-NEXT:    addw %ax, %dx
-; SSSE3-NEXT:    cmovol %ecx, %edx
-; SSSE3-NEXT:    pextrw $2, %xmm3, %eax
-; SSSE3-NEXT:    pextrw $2, %xmm2, %r9d
-; SSSE3-NEXT:    xorl %ecx, %ecx
-; SSSE3-NEXT:    movl %r9d, %esi
-; SSSE3-NEXT:    addw %ax, %si
-; SSSE3-NEXT:    setns %cl
-; SSSE3-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSSE3-NEXT:    addw %ax, %r9w
-; SSSE3-NEXT:    cmovol %ecx, %r9d
-; SSSE3-NEXT:    movd %xmm3, %eax
-; SSSE3-NEXT:    movd %xmm2, %r10d
-; SSSE3-NEXT:    xorl %ecx, %ecx
-; SSSE3-NEXT:    movl %r10d, %esi
-; SSSE3-NEXT:    addw %ax, %si
-; SSSE3-NEXT:    setns %cl
-; SSSE3-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSSE3-NEXT:    addw %ax, %r10w
-; SSSE3-NEXT:    cmovol %ecx, %r10d
-; SSSE3-NEXT:    pextrw $1, %xmm3, %eax
-; SSSE3-NEXT:    pextrw $1, %xmm2, %r11d
-; SSSE3-NEXT:    xorl %ecx, %ecx
-; SSSE3-NEXT:    movl %r11d, %esi
-; SSSE3-NEXT:    addw %ax, %si
-; SSSE3-NEXT:    setns %cl
-; SSSE3-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSSE3-NEXT:    addw %ax, %r11w
-; SSSE3-NEXT:    cmovol %ecx, %r11d
-; SSSE3-NEXT:    movd %xmm1, %eax
-; SSSE3-NEXT:    movd %xmm0, %r14d
-; SSSE3-NEXT:    xorl %ecx, %ecx
-; SSSE3-NEXT:    movl %r14d, %esi
-; SSSE3-NEXT:    addw %ax, %si
-; SSSE3-NEXT:    setns %cl
-; SSSE3-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSSE3-NEXT:    addw %ax, %r14w
-; SSSE3-NEXT:    cmovol %ecx, %r14d
-; SSSE3-NEXT:    pextrw $1, %xmm1, %eax
-; SSSE3-NEXT:    pextrw $1, %xmm0, %r15d
-; SSSE3-NEXT:    xorl %esi, %esi
-; SSSE3-NEXT:    movl %r15d, %edi
-; SSSE3-NEXT:    addw %ax, %di
-; SSSE3-NEXT:    setns %sil
-; SSSE3-NEXT:    addl $32767, %esi # imm = 0x7FFF
-; SSSE3-NEXT:    addw %ax, %r15w
-; SSSE3-NEXT:    cmovol %esi, %r15d
-; SSSE3-NEXT:    pextrw $2, %xmm1, %eax
-; SSSE3-NEXT:    pextrw $2, %xmm0, %r12d
-; SSSE3-NEXT:    xorl %edi, %edi
-; SSSE3-NEXT:    movl %r12d, %ebx
-; SSSE3-NEXT:    addw %ax, %bx
-; SSSE3-NEXT:    setns %dil
-; SSSE3-NEXT:    addl $32767, %edi # imm = 0x7FFF
-; SSSE3-NEXT:    addw %ax, %r12w
-; SSSE3-NEXT:    cmovol %edi, %r12d
-; SSSE3-NEXT:    pextrw $3, %xmm1, %eax
-; SSSE3-NEXT:    pextrw $3, %xmm0, %r13d
-; SSSE3-NEXT:    xorl %ebx, %ebx
-; SSSE3-NEXT:    movl %r13d, %ebp
-; SSSE3-NEXT:    addw %ax, %bp
-; SSSE3-NEXT:    setns %bl
-; SSSE3-NEXT:    addl $32767, %ebx # imm = 0x7FFF
-; SSSE3-NEXT:    addw %ax, %r13w
-; SSSE3-NEXT:    cmovol %ebx, %r13d
-; SSSE3-NEXT:    pextrw $4, %xmm1, %eax
-; SSSE3-NEXT:    pextrw $4, %xmm0, %ebx
-; SSSE3-NEXT:    xorl %ebp, %ebp
-; SSSE3-NEXT:    movl %ebx, %ecx
-; SSSE3-NEXT:    addw %ax, %cx
-; SSSE3-NEXT:    setns %bpl
-; SSSE3-NEXT:    addl $32767, %ebp # imm = 0x7FFF
-; SSSE3-NEXT:    addw %ax, %bx
-; SSSE3-NEXT:    cmovol %ebp, %ebx
-; SSSE3-NEXT:    pextrw $5, %xmm1, %eax
-; SSSE3-NEXT:    pextrw $5, %xmm0, %ebp
-; SSSE3-NEXT:    xorl %ecx, %ecx
-; SSSE3-NEXT:    movl %ebp, %esi
-; SSSE3-NEXT:    addw %ax, %si
-; SSSE3-NEXT:    setns %cl
-; SSSE3-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSSE3-NEXT:    addw %ax, %bp
-; SSSE3-NEXT:    cmovol %ecx, %ebp
-; SSSE3-NEXT:    pextrw $6, %xmm1, %ecx
-; SSSE3-NEXT:    pextrw $6, %xmm0, %eax
-; SSSE3-NEXT:    xorl %esi, %esi
-; SSSE3-NEXT:    movl %eax, %edi
-; SSSE3-NEXT:    addw %cx, %di
-; SSSE3-NEXT:    setns %sil
-; SSSE3-NEXT:    addl $32767, %esi # imm = 0x7FFF
-; SSSE3-NEXT:    addw %cx, %ax
-; SSSE3-NEXT:    cmovol %esi, %eax
-; SSSE3-NEXT:    pextrw $7, %xmm1, %ecx
-; SSSE3-NEXT:    pextrw $7, %xmm0, %esi
-; SSSE3-NEXT:    xorl %edi, %edi
-; SSSE3-NEXT:    movl %esi, %r8d
-; SSSE3-NEXT:    addw %cx, %r8w
-; SSSE3-NEXT:    setns %dil
-; SSSE3-NEXT:    addl $32767, %edi # imm = 0x7FFF
-; SSSE3-NEXT:    addw %cx, %si
-; SSSE3-NEXT:    cmovol %edi, %esi
-; SSSE3-NEXT:    movd %esi, %xmm8
-; SSSE3-NEXT:    movd %eax, %xmm1
-; SSSE3-NEXT:    movd %ebp, %xmm2
-; SSSE3-NEXT:    movd %ebx, %xmm3
-; SSSE3-NEXT:    movd %r13d, %xmm4
-; SSSE3-NEXT:    movd %r12d, %xmm5
-; SSSE3-NEXT:    movd %r15d, %xmm6
-; SSSE3-NEXT:    movd %r14d, %xmm7
-; SSSE3-NEXT:    movd %r10d, %xmm0
-; SSSE3-NEXT:    pinsrw $1, %r11d, %xmm0
-; SSSE3-NEXT:    pinsrw $2, %r9d, %xmm0
-; SSSE3-NEXT:    pinsrw $3, %edx, %xmm0
-; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm8[0],xmm1[1],xmm8[1],xmm1[2],xmm8[2],xmm1[3],xmm8[3]
-; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm3 = xmm3[0],xmm2[0],xmm3[1],xmm2[1],xmm3[2],xmm2[2],xmm3[3],xmm2[3]
-; SSSE3-NEXT:    punpckldq {{.*#+}} xmm3 = xmm3[0],xmm1[0],xmm3[1],xmm1[1]
-; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm5 = xmm5[0],xmm4[0],xmm5[1],xmm4[1],xmm5[2],xmm4[2],xmm5[3],xmm4[3]
-; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm7 = xmm7[0],xmm6[0],xmm7[1],xmm6[1],xmm7[2],xmm6[2],xmm7[3],xmm6[3]
-; SSSE3-NEXT:    punpckldq {{.*#+}} xmm7 = xmm7[0],xmm5[0],xmm7[1],xmm5[1]
-; SSSE3-NEXT:    punpcklqdq {{.*#+}} xmm7 = xmm7[0],xmm3[0]
-; SSSE3-NEXT:    movq {{[-0-9]+}}(%r{{[sb]}}p), %rax # 8-byte Reload
-; SSSE3-NEXT:    movq %xmm0, 16(%rax)
-; SSSE3-NEXT:    movdqa %xmm7, (%rax)
-; SSSE3-NEXT:    popq %rbx
-; SSSE3-NEXT:    popq %r12
-; SSSE3-NEXT:    popq %r13
-; SSSE3-NEXT:    popq %r14
-; SSSE3-NEXT:    popq %r15
-; SSSE3-NEXT:    popq %rbp
-; SSSE3-NEXT:    retq
+; AVX1-LABEL: v12i16:
+; AVX1:       # %bb.0:
+; AVX1-NEXT:    vmovdqa (%rdi), %xmm0
+; AVX1-NEXT:    vmovdqa 16(%rdi), %xmm1
+; AVX1-NEXT:    vpaddsw (%rsi), %xmm0, %xmm0
+; AVX1-NEXT:    vpaddsw 16(%rsi), %xmm1, %xmm1
+; AVX1-NEXT:    vmovq %xmm1, 16(%rdx)
+; AVX1-NEXT:    vmovdqa %xmm0, (%rdx)
+; AVX1-NEXT:    retq
 ;
-; SSE41-LABEL: v12i16:
-; SSE41:       # %bb.0:
-; SSE41-NEXT:    pushq %rbp
-; SSE41-NEXT:    pushq %r15
-; SSE41-NEXT:    pushq %r14
-; SSE41-NEXT:    pushq %r13
-; SSE41-NEXT:    pushq %r12
-; SSE41-NEXT:    pushq %rbx
-; SSE41-NEXT:    movq %rdx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill
-; SSE41-NEXT:    movdqa (%rdi), %xmm0
-; SSE41-NEXT:    movdqa 16(%rdi), %xmm2
-; SSE41-NEXT:    movdqa (%rsi), %xmm1
-; SSE41-NEXT:    movdqa 16(%rsi), %xmm3
-; SSE41-NEXT:    pextrw $3, %xmm3, %eax
-; SSE41-NEXT:    pextrw $3, %xmm2, %edx
-; SSE41-NEXT:    xorl %ecx, %ecx
-; SSE41-NEXT:    movl %edx, %esi
-; SSE41-NEXT:    addw %ax, %si
-; SSE41-NEXT:    setns %cl
-; SSE41-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSE41-NEXT:    addw %ax, %dx
-; SSE41-NEXT:    cmovol %ecx, %edx
-; SSE41-NEXT:    pextrw $2, %xmm3, %eax
-; SSE41-NEXT:    pextrw $2, %xmm2, %r9d
-; SSE41-NEXT:    xorl %ecx, %ecx
-; SSE41-NEXT:    movl %r9d, %esi
-; SSE41-NEXT:    addw %ax, %si
-; SSE41-NEXT:    setns %cl
-; SSE41-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSE41-NEXT:    addw %ax, %r9w
-; SSE41-NEXT:    cmovol %ecx, %r9d
-; SSE41-NEXT:    movd %xmm3, %eax
-; SSE41-NEXT:    movd %xmm2, %r10d
-; SSE41-NEXT:    xorl %ecx, %ecx
-; SSE41-NEXT:    movl %r10d, %esi
-; SSE41-NEXT:    addw %ax, %si
-; SSE41-NEXT:    setns %cl
-; SSE41-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSE41-NEXT:    addw %ax, %r10w
-; SSE41-NEXT:    cmovol %ecx, %r10d
-; SSE41-NEXT:    pextrw $1, %xmm3, %eax
-; SSE41-NEXT:    pextrw $1, %xmm2, %r11d
-; SSE41-NEXT:    xorl %ecx, %ecx
-; SSE41-NEXT:    movl %r11d, %esi
-; SSE41-NEXT:    addw %ax, %si
-; SSE41-NEXT:    setns %cl
-; SSE41-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSE41-NEXT:    addw %ax, %r11w
-; SSE41-NEXT:    cmovol %ecx, %r11d
-; SSE41-NEXT:    pextrw $7, %xmm1, %ecx
-; SSE41-NEXT:    pextrw $7, %xmm0, %r14d
-; SSE41-NEXT:    xorl %esi, %esi
-; SSE41-NEXT:    movl %r14d, %edi
-; SSE41-NEXT:    addw %cx, %di
-; SSE41-NEXT:    setns %sil
-; SSE41-NEXT:    addl $32767, %esi # imm = 0x7FFF
-; SSE41-NEXT:    addw %cx, %r14w
-; SSE41-NEXT:    cmovol %esi, %r14d
-; SSE41-NEXT:    pextrw $6, %xmm1, %esi
-; SSE41-NEXT:    pextrw $6, %xmm0, %r15d
-; SSE41-NEXT:    xorl %edi, %edi
-; SSE41-NEXT:    movl %r15d, %ebx
-; SSE41-NEXT:    addw %si, %bx
-; SSE41-NEXT:    setns %dil
-; SSE41-NEXT:    addl $32767, %edi # imm = 0x7FFF
-; SSE41-NEXT:    addw %si, %r15w
-; SSE41-NEXT:    cmovol %edi, %r15d
-; SSE41-NEXT:    pextrw $5, %xmm1, %edi
-; SSE41-NEXT:    pextrw $5, %xmm0, %r12d
-; SSE41-NEXT:    xorl %ebx, %ebx
-; SSE41-NEXT:    movl %r12d, %ebp
-; SSE41-NEXT:    addw %di, %bp
-; SSE41-NEXT:    setns %bl
-; SSE41-NEXT:    addl $32767, %ebx # imm = 0x7FFF
-; SSE41-NEXT:    addw %di, %r12w
-; SSE41-NEXT:    cmovol %ebx, %r12d
-; SSE41-NEXT:    pextrw $4, %xmm1, %ebx
-; SSE41-NEXT:    pextrw $4, %xmm0, %r13d
-; SSE41-NEXT:    xorl %ebp, %ebp
-; SSE41-NEXT:    movl %r13d, %eax
-; SSE41-NEXT:    addw %bx, %ax
-; SSE41-NEXT:    setns %bpl
-; SSE41-NEXT:    addl $32767, %ebp # imm = 0x7FFF
-; SSE41-NEXT:    addw %bx, %r13w
-; SSE41-NEXT:    cmovol %ebp, %r13d
-; SSE41-NEXT:    pextrw $3, %xmm1, %eax
-; SSE41-NEXT:    pextrw $3, %xmm0, %ebx
-; SSE41-NEXT:    xorl %ebp, %ebp
-; SSE41-NEXT:    movl %ebx, %ecx
-; SSE41-NEXT:    addw %ax, %cx
-; SSE41-NEXT:    setns %bpl
-; SSE41-NEXT:    addl $32767, %ebp # imm = 0x7FFF
-; SSE41-NEXT:    addw %ax, %bx
-; SSE41-NEXT:    cmovol %ebp, %ebx
-; SSE41-NEXT:    pextrw $2, %xmm1, %eax
-; SSE41-NEXT:    pextrw $2, %xmm0, %ebp
-; SSE41-NEXT:    xorl %ecx, %ecx
-; SSE41-NEXT:    movl %ebp, %esi
-; SSE41-NEXT:    addw %ax, %si
-; SSE41-NEXT:    setns %cl
-; SSE41-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; SSE41-NEXT:    addw %ax, %bp
-; SSE41-NEXT:    cmovol %ecx, %ebp
-; SSE41-NEXT:    movd %xmm1, %eax
-; SSE41-NEXT:    movd %xmm0, %ecx
-; SSE41-NEXT:    xorl %esi, %esi
-; SSE41-NEXT:    movl %ecx, %edi
-; SSE41-NEXT:    addw %ax, %di
-; SSE41-NEXT:    setns %sil
-; SSE41-NEXT:    addl $32767, %esi # imm = 0x7FFF
-; SSE41-NEXT:    addw %ax, %cx
-; SSE41-NEXT:    cmovol %esi, %ecx
-; SSE41-NEXT:    pextrw $1, %xmm1, %eax
-; SSE41-NEXT:    pextrw $1, %xmm0, %esi
-; SSE41-NEXT:    xorl %edi, %edi
-; SSE41-NEXT:    movl %esi, %r8d
-; SSE41-NEXT:    addw %ax, %r8w
-; SSE41-NEXT:    setns %dil
-; SSE41-NEXT:    addl $32767, %edi # imm = 0x7FFF
-; SSE41-NEXT:    addw %ax, %si
-; SSE41-NEXT:    cmovol %edi, %esi
-; SSE41-NEXT:    movd %ecx, %xmm0
-; SSE41-NEXT:    pinsrw $1, %esi, %xmm0
-; SSE41-NEXT:    pinsrw $2, %ebp, %xmm0
-; SSE41-NEXT:    pinsrw $3, %ebx, %xmm0
-; SSE41-NEXT:    pinsrw $4, %r13d, %xmm0
-; SSE41-NEXT:    pinsrw $5, %r12d, %xmm0
-; SSE41-NEXT:    pinsrw $6, %r15d, %xmm0
-; SSE41-NEXT:    pinsrw $7, %r14d, %xmm0
-; SSE41-NEXT:    movd %r10d, %xmm1
-; SSE41-NEXT:    pinsrw $1, %r11d, %xmm1
-; SSE41-NEXT:    pinsrw $2, %r9d, %xmm1
-; SSE41-NEXT:    pinsrw $3, %edx, %xmm1
-; SSE41-NEXT:    movq {{[-0-9]+}}(%r{{[sb]}}p), %rax # 8-byte Reload
-; SSE41-NEXT:    movq %xmm1, 16(%rax)
-; SSE41-NEXT:    movdqa %xmm0, (%rax)
-; SSE41-NEXT:    popq %rbx
-; SSE41-NEXT:    popq %r12
-; SSE41-NEXT:    popq %r13
-; SSE41-NEXT:    popq %r14
-; SSE41-NEXT:    popq %r15
-; SSE41-NEXT:    popq %rbp
-; SSE41-NEXT:    retq
+; AVX2-LABEL: v12i16:
+; AVX2:       # %bb.0:
+; AVX2-NEXT:    vmovdqa (%rdi), %ymm0
+; AVX2-NEXT:    vpaddsw (%rsi), %ymm0, %ymm0
+; AVX2-NEXT:    vextracti128 $1, %ymm0, %xmm1
+; AVX2-NEXT:    vmovq %xmm1, 16(%rdx)
+; AVX2-NEXT:    vmovdqa %xmm0, (%rdx)
+; AVX2-NEXT:    vzeroupper
+; AVX2-NEXT:    retq
 ;
-; AVX-LABEL: v12i16:
-; AVX:       # %bb.0:
-; AVX-NEXT:    pushq %rbp
-; AVX-NEXT:    pushq %r15
-; AVX-NEXT:    pushq %r14
-; AVX-NEXT:    pushq %r13
-; AVX-NEXT:    pushq %r12
-; AVX-NEXT:    pushq %rbx
-; AVX-NEXT:    movq %rdx, {{[-0-9]+}}(%r{{[sb]}}p) # 8-byte Spill
-; AVX-NEXT:    vmovdqa (%rsi), %xmm2
-; AVX-NEXT:    vmovdqa 16(%rsi), %xmm0
-; AVX-NEXT:    vmovd %xmm2, %eax
-; AVX-NEXT:    vmovdqa (%rdi), %xmm3
-; AVX-NEXT:    vmovdqa 16(%rdi), %xmm1
-; AVX-NEXT:    vmovd %xmm3, %edx
-; AVX-NEXT:    xorl %ecx, %ecx
-; AVX-NEXT:    movl %edx, %esi
-; AVX-NEXT:    addw %ax, %si
-; AVX-NEXT:    setns %cl
-; AVX-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX-NEXT:    addw %ax, %dx
-; AVX-NEXT:    cmovol %ecx, %edx
-; AVX-NEXT:    movl %edx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX-NEXT:    vpextrw $1, %xmm2, %eax
-; AVX-NEXT:    vpextrw $1, %xmm3, %edx
-; AVX-NEXT:    xorl %ecx, %ecx
-; AVX-NEXT:    movl %edx, %esi
-; AVX-NEXT:    addw %ax, %si
-; AVX-NEXT:    setns %cl
-; AVX-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX-NEXT:    addw %ax, %dx
-; AVX-NEXT:    cmovol %ecx, %edx
-; AVX-NEXT:    movl %edx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX-NEXT:    vpextrw $2, %xmm2, %eax
-; AVX-NEXT:    vpextrw $2, %xmm3, %edx
-; AVX-NEXT:    xorl %ecx, %ecx
-; AVX-NEXT:    movl %edx, %esi
-; AVX-NEXT:    addw %ax, %si
-; AVX-NEXT:    setns %cl
-; AVX-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX-NEXT:    addw %ax, %dx
-; AVX-NEXT:    cmovol %ecx, %edx
-; AVX-NEXT:    movl %edx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX-NEXT:    vpextrw $3, %xmm2, %eax
-; AVX-NEXT:    vpextrw $3, %xmm3, %edx
-; AVX-NEXT:    xorl %ecx, %ecx
-; AVX-NEXT:    movl %edx, %esi
-; AVX-NEXT:    addw %ax, %si
-; AVX-NEXT:    setns %cl
-; AVX-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX-NEXT:    addw %ax, %dx
-; AVX-NEXT:    cmovol %ecx, %edx
-; AVX-NEXT:    movl %edx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX-NEXT:    vpextrw $4, %xmm2, %eax
-; AVX-NEXT:    vpextrw $4, %xmm3, %r14d
-; AVX-NEXT:    xorl %ecx, %ecx
-; AVX-NEXT:    movl %r14d, %esi
-; AVX-NEXT:    addw %ax, %si
-; AVX-NEXT:    setns %cl
-; AVX-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX-NEXT:    addw %ax, %r14w
-; AVX-NEXT:    cmovol %ecx, %r14d
-; AVX-NEXT:    vpextrw $5, %xmm2, %eax
-; AVX-NEXT:    vpextrw $5, %xmm3, %r15d
-; AVX-NEXT:    xorl %ecx, %ecx
-; AVX-NEXT:    movl %r15d, %esi
-; AVX-NEXT:    addw %ax, %si
-; AVX-NEXT:    setns %cl
-; AVX-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX-NEXT:    addw %ax, %r15w
-; AVX-NEXT:    cmovol %ecx, %r15d
-; AVX-NEXT:    vpextrw $6, %xmm2, %eax
-; AVX-NEXT:    vpextrw $6, %xmm3, %r12d
-; AVX-NEXT:    xorl %ecx, %ecx
-; AVX-NEXT:    movl %r12d, %esi
-; AVX-NEXT:    addw %ax, %si
-; AVX-NEXT:    setns %cl
-; AVX-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX-NEXT:    addw %ax, %r12w
-; AVX-NEXT:    cmovol %ecx, %r12d
-; AVX-NEXT:    vpextrw $7, %xmm2, %eax
-; AVX-NEXT:    vpextrw $7, %xmm3, %r13d
-; AVX-NEXT:    xorl %ecx, %ecx
-; AVX-NEXT:    movl %r13d, %esi
-; AVX-NEXT:    addw %ax, %si
-; AVX-NEXT:    setns %cl
-; AVX-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX-NEXT:    addw %ax, %r13w
-; AVX-NEXT:    cmovol %ecx, %r13d
-; AVX-NEXT:    vpextrw $7, %xmm0, %eax
-; AVX-NEXT:    vpextrw $7, %xmm1, %ebx
-; AVX-NEXT:    xorl %ecx, %ecx
-; AVX-NEXT:    movl %ebx, %esi
-; AVX-NEXT:    addw %ax, %si
-; AVX-NEXT:    setns %cl
-; AVX-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX-NEXT:    addw %ax, %bx
-; AVX-NEXT:    cmovol %ecx, %ebx
-; AVX-NEXT:    vpextrw $6, %xmm0, %eax
-; AVX-NEXT:    vpextrw $6, %xmm1, %ebp
-; AVX-NEXT:    xorl %ecx, %ecx
-; AVX-NEXT:    movl %ebp, %esi
-; AVX-NEXT:    addw %ax, %si
-; AVX-NEXT:    setns %cl
-; AVX-NEXT:    addl $32767, %ecx # imm = 0x7FFF
-; AVX-NEXT:    addw %ax, %bp
-; AVX-NEXT:    cmovol %ecx, %ebp
-; AVX-NEXT:    vpextrw $5, %xmm0, %ecx
-; AVX-NEXT:    vpextrw $5, %xmm1, %eax
-; AVX-NEXT:    xorl %esi, %esi
-; AVX-NEXT:    movl %eax, %edi
-; AVX-NEXT:    addw %cx, %di
-; AVX-NEXT:    setns %sil
-; AVX-NEXT:    addl $32767, %esi # imm = 0x7FFF
-; AVX-NEXT:    addw %cx, %ax
-; AVX-NEXT:    cmovol %esi, %eax
-; AVX-NEXT:    vpextrw $4, %xmm0, %esi
-; AVX-NEXT:    vpextrw $4, %xmm1, %ecx
-; AVX-NEXT:    xorl %edi, %edi
-; AVX-NEXT:    movl %ecx, %r8d
-; AVX-NEXT:    addw %si, %r8w
-; AVX-NEXT:    setns %dil
-; AVX-NEXT:    addl $32767, %edi # imm = 0x7FFF
-; AVX-NEXT:    addw %si, %cx
-; AVX-NEXT:    cmovol %edi, %ecx
-; AVX-NEXT:    vpextrw $3, %xmm0, %edi
-; AVX-NEXT:    vpextrw $3, %xmm1, %r8d
-; AVX-NEXT:    xorl %esi, %esi
-; AVX-NEXT:    movl %r8d, %edx
-; AVX-NEXT:    addw %di, %dx
-; AVX-NEXT:    setns %sil
-; AVX-NEXT:    addl $32767, %esi # imm = 0x7FFF
-; AVX-NEXT:    addw %di, %r8w
-; AVX-NEXT:    cmovol %esi, %r8d
-; AVX-NEXT:    vpextrw $2, %xmm0, %edx
-; AVX-NEXT:    vpextrw $2, %xmm1, %edi
-; AVX-NEXT:    xorl %esi, %esi
-; AVX-NEXT:    movl %edi, %r9d
-; AVX-NEXT:    addw %dx, %r9w
-; AVX-NEXT:    setns %sil
-; AVX-NEXT:    addl $32767, %esi # imm = 0x7FFF
-; AVX-NEXT:    addw %dx, %di
-; AVX-NEXT:    cmovol %esi, %edi
-; AVX-NEXT:    vmovd %xmm0, %r9d
-; AVX-NEXT:    vmovd %xmm1, %esi
-; AVX-NEXT:    xorl %edx, %edx
-; AVX-NEXT:    movl %esi, %r10d
-; AVX-NEXT:    addw %r9w, %r10w
-; AVX-NEXT:    setns %dl
-; AVX-NEXT:    addl $32767, %edx # imm = 0x7FFF
-; AVX-NEXT:    addw %r9w, %si
-; AVX-NEXT:    cmovol %edx, %esi
-; AVX-NEXT:    vpextrw $1, %xmm0, %r9d
-; AVX-NEXT:    vpextrw $1, %xmm1, %edx
-; AVX-NEXT:    xorl %r10d, %r10d
-; AVX-NEXT:    movl %edx, %r11d
-; AVX-NEXT:    addw %r9w, %r11w
-; AVX-NEXT:    setns %r10b
-; AVX-NEXT:    addl $32767, %r10d # imm = 0x7FFF
-; AVX-NEXT:    addw %r9w, %dx
-; AVX-NEXT:    cmovol %r10d, %edx
-; AVX-NEXT:    vmovd %esi, %xmm0
-; AVX-NEXT:    vpinsrw $1, %edx, %xmm0, %xmm0
-; AVX-NEXT:    vpinsrw $2, %edi, %xmm0, %xmm0
-; AVX-NEXT:    vpinsrw $3, %r8d, %xmm0, %xmm0
-; AVX-NEXT:    vpinsrw $4, %ecx, %xmm0, %xmm0
-; AVX-NEXT:    vpinsrw $5, %eax, %xmm0, %xmm0
-; AVX-NEXT:    vpinsrw $6, %ebp, %xmm0, %xmm0
-; AVX-NEXT:    vpinsrw $7, %ebx, %xmm0, %xmm0
-; AVX-NEXT:    vmovd {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 4-byte Folded Reload
-; AVX-NEXT:    # xmm1 = mem[0],zero,zero,zero
-; AVX-NEXT:    vpinsrw $1, {{[-0-9]+}}(%r{{[sb]}}p), %xmm1, %xmm1 # 4-byte Folded Reload
-; AVX-NEXT:    vpinsrw $2, {{[-0-9]+}}(%r{{[sb]}}p), %xmm1, %xmm1 # 4-byte Folded Reload
-; AVX-NEXT:    vpinsrw $3, {{[-0-9]+}}(%r{{[sb]}}p), %xmm1, %xmm1 # 4-byte Folded Reload
-; AVX-NEXT:    vpinsrw $4, %r14d, %xmm1, %xmm1
-; AVX-NEXT:    vpinsrw $5, %r15d, %xmm1, %xmm1
-; AVX-NEXT:    vpinsrw $6, %r12d, %xmm1, %xmm1
-; AVX-NEXT:    vpinsrw $7, %r13d, %xmm1, %xmm1
-; AVX-NEXT:    movq {{[-0-9]+}}(%r{{[sb]}}p), %rax # 8-byte Reload
-; AVX-NEXT:    vmovq %xmm0, 16(%rax)
-; AVX-NEXT:    vmovdqa %xmm1, (%rax)
-; AVX-NEXT:    popq %rbx
-; AVX-NEXT:    popq %r12
-; AVX-NEXT:    popq %r13
-; AVX-NEXT:    popq %r14
-; AVX-NEXT:    popq %r15
-; AVX-NEXT:    popq %rbp
-; AVX-NEXT:    retq
+; AVX512-LABEL: v12i16:
+; AVX512:       # %bb.0:
+; AVX512-NEXT:    vmovdqa (%rdi), %ymm0
+; AVX512-NEXT:    vpaddsw (%rsi), %ymm0, %ymm0
+; AVX512-NEXT:    vextracti128 $1, %ymm0, %xmm1
+; AVX512-NEXT:    vmovq %xmm1, 16(%rdx)
+; AVX512-NEXT:    vmovdqa %xmm0, (%rdx)
+; AVX512-NEXT:    vzeroupper
+; AVX512-NEXT:    retq
   %x = load <12 x i16>, <12 x i16>* %px
   %y = load <12 x i16>, <12 x i16>* %py
   %z = call <12 x i16> @llvm.sadd.sat.v12i16(<12 x i16> %x, <12 x i16> %y)
@@ -16872,1960 +565,63 @@ define void @v1i16(<1 x i16>* %px, <1 x
 ; Promotion
 
 define <16 x i4> @v16i4(<16 x i4> %x, <16 x i4> %y) nounwind {
-; SSE2-LABEL: v16i4:
-; SSE2:       # %bb.0:
-; SSE2-NEXT:    pushq %rbp
-; SSE2-NEXT:    pushq %r15
-; SSE2-NEXT:    pushq %r14
-; SSE2-NEXT:    pushq %r13
-; SSE2-NEXT:    pushq %r12
-; SSE2-NEXT:    pushq %rbx
-; SSE2-NEXT:    psllw $4, %xmm1
-; SSE2-NEXT:    movdqa {{.*#+}} xmm2 = [240,240,240,240,240,240,240,240,240,240,240,240,240,240,240,240]
-; SSE2-NEXT:    pand %xmm2, %xmm1
-; SSE2-NEXT:    movdqa %xmm1, -{{[0-9]+}}(%rsp)
-; SSE2-NEXT:    psllw $4, %xmm0
-; SSE2-NEXT:    pand %xmm2, %xmm0
-; SSE2-NEXT:    movdqa %xmm0, -{{[0-9]+}}(%rsp)
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %r9b
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %sil
-; SSE2-NEXT:    movl %r9d, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %r9b
-; SSE2-NEXT:    jno .LBB15_2
-; SSE2-NEXT:  # %bb.1:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %r9d
-; SSE2-NEXT:  .LBB15_2:
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movl %esi, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %sil
-; SSE2-NEXT:    jno .LBB15_4
-; SSE2-NEXT:  # %bb.3:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %esi
-; SSE2-NEXT:  .LBB15_4:
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %bl
-; SSE2-NEXT:    movl %ebx, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %bl
-; SSE2-NEXT:    jno .LBB15_6
-; SSE2-NEXT:  # %bb.5:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %ebx
-; SSE2-NEXT:  .LBB15_6:
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %dl
-; SSE2-NEXT:    movl %edx, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %dl
-; SSE2-NEXT:    jno .LBB15_8
-; SSE2-NEXT:  # %bb.7:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %edx
-; SSE2-NEXT:  .LBB15_8:
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %r10b
-; SSE2-NEXT:    movl %r10d, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %r10b
-; SSE2-NEXT:    jno .LBB15_10
-; SSE2-NEXT:  # %bb.9:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %r10d
-; SSE2-NEXT:  .LBB15_10:
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %r11b
-; SSE2-NEXT:    movl %r11d, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %r11b
-; SSE2-NEXT:    jno .LBB15_12
-; SSE2-NEXT:  # %bb.11:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %r11d
-; SSE2-NEXT:  .LBB15_12:
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %bpl
-; SSE2-NEXT:    movl %ebp, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %bpl
-; SSE2-NEXT:    jno .LBB15_14
-; SSE2-NEXT:  # %bb.13:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %ebp
-; SSE2-NEXT:  .LBB15_14:
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %r14b
-; SSE2-NEXT:    movl %r14d, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %r14b
-; SSE2-NEXT:    jno .LBB15_16
-; SSE2-NEXT:  # %bb.15:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %r14d
-; SSE2-NEXT:  .LBB15_16:
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %r15b
-; SSE2-NEXT:    movl %r15d, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %r15b
-; SSE2-NEXT:    jno .LBB15_18
-; SSE2-NEXT:  # %bb.17:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %r15d
-; SSE2-NEXT:  .LBB15_18:
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %r12b
-; SSE2-NEXT:    movl %r12d, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %r12b
-; SSE2-NEXT:    jno .LBB15_20
-; SSE2-NEXT:  # %bb.19:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %r12d
-; SSE2-NEXT:  .LBB15_20:
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %r13b
-; SSE2-NEXT:    movl %r13d, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %r13b
-; SSE2-NEXT:    movb %bl, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSE2-NEXT:    movb %dl, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSE2-NEXT:    jno .LBB15_22
-; SSE2-NEXT:  # %bb.21:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %r13d
-; SSE2-NEXT:  .LBB15_22:
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %dil
-; SSE2-NEXT:    movl %edi, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %dil
-; SSE2-NEXT:    jno .LBB15_24
-; SSE2-NEXT:  # %bb.23:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %edi
-; SSE2-NEXT:  .LBB15_24:
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %r8b
-; SSE2-NEXT:    movl %r8d, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %r8b
-; SSE2-NEXT:    jno .LBB15_26
-; SSE2-NEXT:  # %bb.25:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %r8d
-; SSE2-NEXT:  .LBB15_26:
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %bl
-; SSE2-NEXT:    movl %ebx, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %bl
-; SSE2-NEXT:    jno .LBB15_28
-; SSE2-NEXT:  # %bb.27:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %ebx
-; SSE2-NEXT:  .LBB15_28:
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %dl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
-; SSE2-NEXT:    movl %eax, %ecx
-; SSE2-NEXT:    addb %dl, %cl
-; SSE2-NEXT:    setns %cl
-; SSE2-NEXT:    addb %dl, %al
-; SSE2-NEXT:    movb %sil, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSE2-NEXT:    jno .LBB15_30
-; SSE2-NEXT:  # %bb.29:
-; SSE2-NEXT:    addb $127, %cl
-; SSE2-NEXT:    movl %ecx, %eax
-; SSE2-NEXT:  .LBB15_30:
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %dl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %sil
-; SSE2-NEXT:    movl %esi, %ecx
-; SSE2-NEXT:    addb %dl, %cl
-; SSE2-NEXT:    setns %cl
-; SSE2-NEXT:    addb %dl, %sil
-; SSE2-NEXT:    jno .LBB15_32
-; SSE2-NEXT:  # %bb.31:
-; SSE2-NEXT:    addb $127, %cl
-; SSE2-NEXT:    movl %ecx, %esi
-; SSE2-NEXT:  .LBB15_32:
-; SSE2-NEXT:    movzbl %sil, %ecx
-; SSE2-NEXT:    movd %ecx, %xmm0
-; SSE2-NEXT:    movzbl %al, %eax
-; SSE2-NEXT:    movd %eax, %xmm1
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3],xmm1[4],xmm0[4],xmm1[5],xmm0[5],xmm1[6],xmm0[6],xmm1[7],xmm0[7]
-; SSE2-NEXT:    movzbl %bl, %eax
-; SSE2-NEXT:    movd %eax, %xmm0
-; SSE2-NEXT:    movzbl %r8b, %eax
-; SSE2-NEXT:    movd %eax, %xmm2
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm2 = xmm2[0],xmm0[0],xmm2[1],xmm0[1],xmm2[2],xmm0[2],xmm2[3],xmm0[3],xmm2[4],xmm0[4],xmm2[5],xmm0[5],xmm2[6],xmm0[6],xmm2[7],xmm0[7]
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm2 = xmm2[0],xmm1[0],xmm2[1],xmm1[1],xmm2[2],xmm1[2],xmm2[3],xmm1[3]
-; SSE2-NEXT:    movzbl %dil, %eax
-; SSE2-NEXT:    movd %eax, %xmm0
-; SSE2-NEXT:    movzbl %r13b, %eax
-; SSE2-NEXT:    movd %eax, %xmm3
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm3 = xmm3[0],xmm0[0],xmm3[1],xmm0[1],xmm3[2],xmm0[2],xmm3[3],xmm0[3],xmm3[4],xmm0[4],xmm3[5],xmm0[5],xmm3[6],xmm0[6],xmm3[7],xmm0[7]
-; SSE2-NEXT:    movzbl %r12b, %eax
-; SSE2-NEXT:    movd %eax, %xmm0
-; SSE2-NEXT:    movzbl %r15b, %eax
-; SSE2-NEXT:    movd %eax, %xmm1
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3],xmm1[4],xmm0[4],xmm1[5],xmm0[5],xmm1[6],xmm0[6],xmm1[7],xmm0[7]
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm3[0],xmm1[1],xmm3[1],xmm1[2],xmm3[2],xmm1[3],xmm3[3]
-; SSE2-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm2[0],xmm1[1],xmm2[1]
-; SSE2-NEXT:    movzbl %r14b, %eax
-; SSE2-NEXT:    movd %eax, %xmm0
-; SSE2-NEXT:    movzbl %bpl, %eax
-; SSE2-NEXT:    movd %eax, %xmm2
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm2 = xmm2[0],xmm0[0],xmm2[1],xmm0[1],xmm2[2],xmm0[2],xmm2[3],xmm0[3],xmm2[4],xmm0[4],xmm2[5],xmm0[5],xmm2[6],xmm0[6],xmm2[7],xmm0[7]
-; SSE2-NEXT:    movzbl %r11b, %eax
-; SSE2-NEXT:    movd %eax, %xmm0
-; SSE2-NEXT:    movzbl %r10b, %eax
-; SSE2-NEXT:    movd %eax, %xmm3
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm3 = xmm3[0],xmm0[0],xmm3[1],xmm0[1],xmm3[2],xmm0[2],xmm3[3],xmm0[3],xmm3[4],xmm0[4],xmm3[5],xmm0[5],xmm3[6],xmm0[6],xmm3[7],xmm0[7]
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm3 = xmm3[0],xmm2[0],xmm3[1],xmm2[1],xmm3[2],xmm2[2],xmm3[3],xmm2[3]
-; SSE2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; SSE2-NEXT:    movd %eax, %xmm0
-; SSE2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; SSE2-NEXT:    movd %eax, %xmm2
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm2 = xmm2[0],xmm0[0],xmm2[1],xmm0[1],xmm2[2],xmm0[2],xmm2[3],xmm0[3],xmm2[4],xmm0[4],xmm2[5],xmm0[5],xmm2[6],xmm0[6],xmm2[7],xmm0[7]
-; SSE2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; SSE2-NEXT:    movd %eax, %xmm4
-; SSE2-NEXT:    movzbl %r9b, %eax
-; SSE2-NEXT:    movd %eax, %xmm0
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm4[0],xmm0[1],xmm4[1],xmm0[2],xmm4[2],xmm0[3],xmm4[3],xmm0[4],xmm4[4],xmm0[5],xmm4[5],xmm0[6],xmm4[6],xmm0[7],xmm4[7]
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1],xmm0[2],xmm2[2],xmm0[3],xmm2[3]
-; SSE2-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm3[0],xmm0[1],xmm3[1]
-; SSE2-NEXT:    punpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm1[0]
-; SSE2-NEXT:    psrlw $4, %xmm0
-; SSE2-NEXT:    pand {{.*}}(%rip), %xmm0
-; SSE2-NEXT:    movdqa {{.*#+}} xmm1 = [8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8]
-; SSE2-NEXT:    pxor %xmm1, %xmm0
-; SSE2-NEXT:    psubb %xmm1, %xmm0
-; SSE2-NEXT:    popq %rbx
-; SSE2-NEXT:    popq %r12
-; SSE2-NEXT:    popq %r13
-; SSE2-NEXT:    popq %r14
-; SSE2-NEXT:    popq %r15
-; SSE2-NEXT:    popq %rbp
-; SSE2-NEXT:    retq
-;
-; SSSE3-LABEL: v16i4:
-; SSSE3:       # %bb.0:
-; SSSE3-NEXT:    pushq %rbp
-; SSSE3-NEXT:    pushq %r15
-; SSSE3-NEXT:    pushq %r14
-; SSSE3-NEXT:    pushq %r13
-; SSSE3-NEXT:    pushq %r12
-; SSSE3-NEXT:    pushq %rbx
-; SSSE3-NEXT:    psllw $4, %xmm1
-; SSSE3-NEXT:    movdqa {{.*#+}} xmm2 = [240,240,240,240,240,240,240,240,240,240,240,240,240,240,240,240]
-; SSSE3-NEXT:    pand %xmm2, %xmm1
-; SSSE3-NEXT:    movdqa %xmm1, -{{[0-9]+}}(%rsp)
-; SSSE3-NEXT:    psllw $4, %xmm0
-; SSSE3-NEXT:    pand %xmm2, %xmm0
-; SSSE3-NEXT:    movdqa %xmm0, -{{[0-9]+}}(%rsp)
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %r9b
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %sil
-; SSSE3-NEXT:    movl %r9d, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %r9b
-; SSSE3-NEXT:    jno .LBB15_2
-; SSSE3-NEXT:  # %bb.1:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %r9d
-; SSSE3-NEXT:  .LBB15_2:
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movl %esi, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %sil
-; SSSE3-NEXT:    jno .LBB15_4
-; SSSE3-NEXT:  # %bb.3:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %esi
-; SSSE3-NEXT:  .LBB15_4:
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %bl
-; SSSE3-NEXT:    movl %ebx, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %bl
-; SSSE3-NEXT:    jno .LBB15_6
-; SSSE3-NEXT:  # %bb.5:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %ebx
-; SSSE3-NEXT:  .LBB15_6:
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %dl
-; SSSE3-NEXT:    movl %edx, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %dl
-; SSSE3-NEXT:    jno .LBB15_8
-; SSSE3-NEXT:  # %bb.7:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %edx
-; SSSE3-NEXT:  .LBB15_8:
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %r10b
-; SSSE3-NEXT:    movl %r10d, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %r10b
-; SSSE3-NEXT:    jno .LBB15_10
-; SSSE3-NEXT:  # %bb.9:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %r10d
-; SSSE3-NEXT:  .LBB15_10:
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %r11b
-; SSSE3-NEXT:    movl %r11d, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %r11b
-; SSSE3-NEXT:    jno .LBB15_12
-; SSSE3-NEXT:  # %bb.11:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %r11d
-; SSSE3-NEXT:  .LBB15_12:
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %bpl
-; SSSE3-NEXT:    movl %ebp, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %bpl
-; SSSE3-NEXT:    jno .LBB15_14
-; SSSE3-NEXT:  # %bb.13:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %ebp
-; SSSE3-NEXT:  .LBB15_14:
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %r14b
-; SSSE3-NEXT:    movl %r14d, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %r14b
-; SSSE3-NEXT:    jno .LBB15_16
-; SSSE3-NEXT:  # %bb.15:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %r14d
-; SSSE3-NEXT:  .LBB15_16:
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %r15b
-; SSSE3-NEXT:    movl %r15d, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %r15b
-; SSSE3-NEXT:    jno .LBB15_18
-; SSSE3-NEXT:  # %bb.17:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %r15d
-; SSSE3-NEXT:  .LBB15_18:
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %r12b
-; SSSE3-NEXT:    movl %r12d, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %r12b
-; SSSE3-NEXT:    jno .LBB15_20
-; SSSE3-NEXT:  # %bb.19:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %r12d
-; SSSE3-NEXT:  .LBB15_20:
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %r13b
-; SSSE3-NEXT:    movl %r13d, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %r13b
-; SSSE3-NEXT:    movb %bl, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSSE3-NEXT:    movb %dl, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSSE3-NEXT:    jno .LBB15_22
-; SSSE3-NEXT:  # %bb.21:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %r13d
-; SSSE3-NEXT:  .LBB15_22:
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %dil
-; SSSE3-NEXT:    movl %edi, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %dil
-; SSSE3-NEXT:    jno .LBB15_24
-; SSSE3-NEXT:  # %bb.23:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %edi
-; SSSE3-NEXT:  .LBB15_24:
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %r8b
-; SSSE3-NEXT:    movl %r8d, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %r8b
-; SSSE3-NEXT:    jno .LBB15_26
-; SSSE3-NEXT:  # %bb.25:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %r8d
-; SSSE3-NEXT:  .LBB15_26:
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %bl
-; SSSE3-NEXT:    movl %ebx, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %bl
-; SSSE3-NEXT:    jno .LBB15_28
-; SSSE3-NEXT:  # %bb.27:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %ebx
-; SSSE3-NEXT:  .LBB15_28:
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %dl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %al
-; SSSE3-NEXT:    movl %eax, %ecx
-; SSSE3-NEXT:    addb %dl, %cl
-; SSSE3-NEXT:    setns %cl
-; SSSE3-NEXT:    addb %dl, %al
-; SSSE3-NEXT:    movb %sil, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSSE3-NEXT:    jno .LBB15_30
-; SSSE3-NEXT:  # %bb.29:
-; SSSE3-NEXT:    addb $127, %cl
-; SSSE3-NEXT:    movl %ecx, %eax
-; SSSE3-NEXT:  .LBB15_30:
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %dl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %sil
-; SSSE3-NEXT:    movl %esi, %ecx
-; SSSE3-NEXT:    addb %dl, %cl
-; SSSE3-NEXT:    setns %cl
-; SSSE3-NEXT:    addb %dl, %sil
-; SSSE3-NEXT:    jno .LBB15_32
-; SSSE3-NEXT:  # %bb.31:
-; SSSE3-NEXT:    addb $127, %cl
-; SSSE3-NEXT:    movl %ecx, %esi
-; SSSE3-NEXT:  .LBB15_32:
-; SSSE3-NEXT:    movzbl %sil, %ecx
-; SSSE3-NEXT:    movd %ecx, %xmm0
-; SSSE3-NEXT:    movzbl %al, %eax
-; SSSE3-NEXT:    movd %eax, %xmm1
-; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3],xmm1[4],xmm0[4],xmm1[5],xmm0[5],xmm1[6],xmm0[6],xmm1[7],xmm0[7]
-; SSSE3-NEXT:    movzbl %bl, %eax
-; SSSE3-NEXT:    movd %eax, %xmm0
-; SSSE3-NEXT:    movzbl %r8b, %eax
-; SSSE3-NEXT:    movd %eax, %xmm2
-; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm2 = xmm2[0],xmm0[0],xmm2[1],xmm0[1],xmm2[2],xmm0[2],xmm2[3],xmm0[3],xmm2[4],xmm0[4],xmm2[5],xmm0[5],xmm2[6],xmm0[6],xmm2[7],xmm0[7]
-; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm2 = xmm2[0],xmm1[0],xmm2[1],xmm1[1],xmm2[2],xmm1[2],xmm2[3],xmm1[3]
-; SSSE3-NEXT:    movzbl %dil, %eax
-; SSSE3-NEXT:    movd %eax, %xmm0
-; SSSE3-NEXT:    movzbl %r13b, %eax
-; SSSE3-NEXT:    movd %eax, %xmm3
-; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm3 = xmm3[0],xmm0[0],xmm3[1],xmm0[1],xmm3[2],xmm0[2],xmm3[3],xmm0[3],xmm3[4],xmm0[4],xmm3[5],xmm0[5],xmm3[6],xmm0[6],xmm3[7],xmm0[7]
-; SSSE3-NEXT:    movzbl %r12b, %eax
-; SSSE3-NEXT:    movd %eax, %xmm0
-; SSSE3-NEXT:    movzbl %r15b, %eax
-; SSSE3-NEXT:    movd %eax, %xmm1
-; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3],xmm1[4],xmm0[4],xmm1[5],xmm0[5],xmm1[6],xmm0[6],xmm1[7],xmm0[7]
-; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm3[0],xmm1[1],xmm3[1],xmm1[2],xmm3[2],xmm1[3],xmm3[3]
-; SSSE3-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm2[0],xmm1[1],xmm2[1]
-; SSSE3-NEXT:    movzbl %r14b, %eax
-; SSSE3-NEXT:    movd %eax, %xmm0
-; SSSE3-NEXT:    movzbl %bpl, %eax
-; SSSE3-NEXT:    movd %eax, %xmm2
-; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm2 = xmm2[0],xmm0[0],xmm2[1],xmm0[1],xmm2[2],xmm0[2],xmm2[3],xmm0[3],xmm2[4],xmm0[4],xmm2[5],xmm0[5],xmm2[6],xmm0[6],xmm2[7],xmm0[7]
-; SSSE3-NEXT:    movzbl %r11b, %eax
-; SSSE3-NEXT:    movd %eax, %xmm0
-; SSSE3-NEXT:    movzbl %r10b, %eax
-; SSSE3-NEXT:    movd %eax, %xmm3
-; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm3 = xmm3[0],xmm0[0],xmm3[1],xmm0[1],xmm3[2],xmm0[2],xmm3[3],xmm0[3],xmm3[4],xmm0[4],xmm3[5],xmm0[5],xmm3[6],xmm0[6],xmm3[7],xmm0[7]
-; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm3 = xmm3[0],xmm2[0],xmm3[1],xmm2[1],xmm3[2],xmm2[2],xmm3[3],xmm2[3]
-; SSSE3-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; SSSE3-NEXT:    movd %eax, %xmm0
-; SSSE3-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; SSSE3-NEXT:    movd %eax, %xmm2
-; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm2 = xmm2[0],xmm0[0],xmm2[1],xmm0[1],xmm2[2],xmm0[2],xmm2[3],xmm0[3],xmm2[4],xmm0[4],xmm2[5],xmm0[5],xmm2[6],xmm0[6],xmm2[7],xmm0[7]
-; SSSE3-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; SSSE3-NEXT:    movd %eax, %xmm4
-; SSSE3-NEXT:    movzbl %r9b, %eax
-; SSSE3-NEXT:    movd %eax, %xmm0
-; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm4[0],xmm0[1],xmm4[1],xmm0[2],xmm4[2],xmm0[3],xmm4[3],xmm0[4],xmm4[4],xmm0[5],xmm4[5],xmm0[6],xmm4[6],xmm0[7],xmm4[7]
-; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1],xmm0[2],xmm2[2],xmm0[3],xmm2[3]
-; SSSE3-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm3[0],xmm0[1],xmm3[1]
-; SSSE3-NEXT:    punpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm1[0]
-; SSSE3-NEXT:    psrlw $4, %xmm0
-; SSSE3-NEXT:    pand {{.*}}(%rip), %xmm0
-; SSSE3-NEXT:    movdqa {{.*#+}} xmm1 = [8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8]
-; SSSE3-NEXT:    pxor %xmm1, %xmm0
-; SSSE3-NEXT:    psubb %xmm1, %xmm0
-; SSSE3-NEXT:    popq %rbx
-; SSSE3-NEXT:    popq %r12
-; SSSE3-NEXT:    popq %r13
-; SSSE3-NEXT:    popq %r14
-; SSSE3-NEXT:    popq %r15
-; SSSE3-NEXT:    popq %rbp
-; SSSE3-NEXT:    retq
-;
-; SSE41-LABEL: v16i4:
-; SSE41:       # %bb.0:
-; SSE41-NEXT:    psllw $4, %xmm1
-; SSE41-NEXT:    movdqa {{.*#+}} xmm2 = [240,240,240,240,240,240,240,240,240,240,240,240,240,240,240,240]
-; SSE41-NEXT:    pand %xmm2, %xmm1
-; SSE41-NEXT:    pextrb $15, %xmm1, %ecx
-; SSE41-NEXT:    psllw $4, %xmm0
-; SSE41-NEXT:    pand %xmm2, %xmm0
-; SSE41-NEXT:    pextrb $15, %xmm0, %edx
-; SSE41-NEXT:    movl %edx, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %dl
-; SSE41-NEXT:    jno .LBB15_2
-; SSE41-NEXT:  # %bb.1:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %edx
-; SSE41-NEXT:  .LBB15_2:
-; SSE41-NEXT:    pextrb $14, %xmm1, %ecx
-; SSE41-NEXT:    pextrb $14, %xmm0, %r11d
-; SSE41-NEXT:    movl %r11d, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %r11b
-; SSE41-NEXT:    jno .LBB15_4
-; SSE41-NEXT:  # %bb.3:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %r11d
-; SSE41-NEXT:  .LBB15_4:
-; SSE41-NEXT:    pextrb $13, %xmm1, %ecx
-; SSE41-NEXT:    pextrb $13, %xmm0, %edi
-; SSE41-NEXT:    movl %edi, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %dil
-; SSE41-NEXT:    jno .LBB15_6
-; SSE41-NEXT:  # %bb.5:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %edi
-; SSE41-NEXT:  .LBB15_6:
-; SSE41-NEXT:    pushq %rbp
-; SSE41-NEXT:    pushq %r15
-; SSE41-NEXT:    pushq %r14
-; SSE41-NEXT:    pushq %r13
-; SSE41-NEXT:    pushq %r12
-; SSE41-NEXT:    pushq %rbx
-; SSE41-NEXT:    pextrb $12, %xmm1, %ecx
-; SSE41-NEXT:    pextrb $12, %xmm0, %r14d
-; SSE41-NEXT:    movl %r14d, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %r14b
-; SSE41-NEXT:    jno .LBB15_8
-; SSE41-NEXT:  # %bb.7:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %r14d
-; SSE41-NEXT:  .LBB15_8:
-; SSE41-NEXT:    pextrb $11, %xmm1, %ecx
-; SSE41-NEXT:    pextrb $11, %xmm0, %ebp
-; SSE41-NEXT:    movl %ebp, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %bpl
-; SSE41-NEXT:    jno .LBB15_10
-; SSE41-NEXT:  # %bb.9:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %ebp
-; SSE41-NEXT:  .LBB15_10:
-; SSE41-NEXT:    pextrb $10, %xmm1, %ecx
-; SSE41-NEXT:    pextrb $10, %xmm0, %r15d
-; SSE41-NEXT:    movl %r15d, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %r15b
-; SSE41-NEXT:    jno .LBB15_12
-; SSE41-NEXT:  # %bb.11:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %r15d
-; SSE41-NEXT:  .LBB15_12:
-; SSE41-NEXT:    pextrb $9, %xmm1, %ecx
-; SSE41-NEXT:    pextrb $9, %xmm0, %r12d
-; SSE41-NEXT:    movl %r12d, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %r12b
-; SSE41-NEXT:    jno .LBB15_14
-; SSE41-NEXT:  # %bb.13:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %r12d
-; SSE41-NEXT:  .LBB15_14:
-; SSE41-NEXT:    pextrb $8, %xmm1, %ecx
-; SSE41-NEXT:    pextrb $8, %xmm0, %r13d
-; SSE41-NEXT:    movl %r13d, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %r13b
-; SSE41-NEXT:    jno .LBB15_16
-; SSE41-NEXT:  # %bb.15:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %r13d
-; SSE41-NEXT:  .LBB15_16:
-; SSE41-NEXT:    pextrb $7, %xmm1, %ecx
-; SSE41-NEXT:    pextrb $7, %xmm0, %r10d
-; SSE41-NEXT:    movl %r10d, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %r10b
-; SSE41-NEXT:    movl %ebp, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE41-NEXT:    jno .LBB15_18
-; SSE41-NEXT:  # %bb.17:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %r10d
-; SSE41-NEXT:  .LBB15_18:
-; SSE41-NEXT:    pextrb $6, %xmm1, %ecx
-; SSE41-NEXT:    pextrb $6, %xmm0, %r9d
-; SSE41-NEXT:    movl %r9d, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %r9b
-; SSE41-NEXT:    jno .LBB15_20
-; SSE41-NEXT:  # %bb.19:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %r9d
-; SSE41-NEXT:  .LBB15_20:
-; SSE41-NEXT:    pextrb $5, %xmm1, %ecx
-; SSE41-NEXT:    pextrb $5, %xmm0, %ebp
-; SSE41-NEXT:    movl %ebp, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %bpl
-; SSE41-NEXT:    movl %edi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE41-NEXT:    jno .LBB15_22
-; SSE41-NEXT:  # %bb.21:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %ebp
-; SSE41-NEXT:  .LBB15_22:
-; SSE41-NEXT:    pextrb $4, %xmm1, %ecx
-; SSE41-NEXT:    pextrb $4, %xmm0, %edi
-; SSE41-NEXT:    movl %edi, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %dil
-; SSE41-NEXT:    movl %edx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE41-NEXT:    jno .LBB15_24
-; SSE41-NEXT:  # %bb.23:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %edi
-; SSE41-NEXT:  .LBB15_24:
-; SSE41-NEXT:    pextrb $3, %xmm1, %edx
-; SSE41-NEXT:    pextrb $3, %xmm0, %eax
-; SSE41-NEXT:    movl %eax, %ecx
-; SSE41-NEXT:    addb %dl, %cl
-; SSE41-NEXT:    setns %cl
-; SSE41-NEXT:    addb %dl, %al
-; SSE41-NEXT:    jno .LBB15_26
-; SSE41-NEXT:  # %bb.25:
-; SSE41-NEXT:    addb $127, %cl
-; SSE41-NEXT:    movl %ecx, %eax
-; SSE41-NEXT:  .LBB15_26:
-; SSE41-NEXT:    pextrb $2, %xmm1, %ebx
-; SSE41-NEXT:    pextrb $2, %xmm0, %ecx
-; SSE41-NEXT:    movl %ecx, %edx
-; SSE41-NEXT:    addb %bl, %dl
-; SSE41-NEXT:    setns %dl
-; SSE41-NEXT:    addb %bl, %cl
-; SSE41-NEXT:    jno .LBB15_28
-; SSE41-NEXT:  # %bb.27:
-; SSE41-NEXT:    addb $127, %dl
-; SSE41-NEXT:    movl %edx, %ecx
-; SSE41-NEXT:  .LBB15_28:
-; SSE41-NEXT:    pextrb $0, %xmm1, %esi
-; SSE41-NEXT:    pextrb $0, %xmm0, %edx
-; SSE41-NEXT:    movl %edx, %ebx
-; SSE41-NEXT:    addb %sil, %bl
-; SSE41-NEXT:    setns %bl
-; SSE41-NEXT:    addb %sil, %dl
-; SSE41-NEXT:    jno .LBB15_30
-; SSE41-NEXT:  # %bb.29:
-; SSE41-NEXT:    addb $127, %bl
-; SSE41-NEXT:    movl %ebx, %edx
-; SSE41-NEXT:  .LBB15_30:
-; SSE41-NEXT:    pextrb $1, %xmm1, %esi
-; SSE41-NEXT:    pextrb $1, %xmm0, %r8d
-; SSE41-NEXT:    movl %r8d, %ebx
-; SSE41-NEXT:    addb %sil, %bl
-; SSE41-NEXT:    setns %bl
-; SSE41-NEXT:    addb %sil, %r8b
-; SSE41-NEXT:    jno .LBB15_32
-; SSE41-NEXT:  # %bb.31:
-; SSE41-NEXT:    addb $127, %bl
-; SSE41-NEXT:    movl %ebx, %r8d
-; SSE41-NEXT:  .LBB15_32:
-; SSE41-NEXT:    movzbl %dl, %edx
-; SSE41-NEXT:    movd %edx, %xmm0
-; SSE41-NEXT:    movzbl %r8b, %edx
-; SSE41-NEXT:    pinsrb $1, %edx, %xmm0
-; SSE41-NEXT:    movzbl %cl, %ecx
-; SSE41-NEXT:    pinsrb $2, %ecx, %xmm0
-; SSE41-NEXT:    movzbl %al, %eax
-; SSE41-NEXT:    pinsrb $3, %eax, %xmm0
-; SSE41-NEXT:    movzbl %dil, %eax
-; SSE41-NEXT:    pinsrb $4, %eax, %xmm0
-; SSE41-NEXT:    movzbl %bpl, %eax
-; SSE41-NEXT:    pinsrb $5, %eax, %xmm0
-; SSE41-NEXT:    movzbl %r9b, %eax
-; SSE41-NEXT:    pinsrb $6, %eax, %xmm0
-; SSE41-NEXT:    movzbl %r10b, %eax
-; SSE41-NEXT:    pinsrb $7, %eax, %xmm0
-; SSE41-NEXT:    movzbl %r13b, %eax
-; SSE41-NEXT:    pinsrb $8, %eax, %xmm0
-; SSE41-NEXT:    movzbl %r12b, %eax
-; SSE41-NEXT:    pinsrb $9, %eax, %xmm0
-; SSE41-NEXT:    movzbl %r15b, %eax
-; SSE41-NEXT:    pinsrb $10, %eax, %xmm0
-; SSE41-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; SSE41-NEXT:    pinsrb $11, %eax, %xmm0
-; SSE41-NEXT:    movzbl %r14b, %eax
-; SSE41-NEXT:    pinsrb $12, %eax, %xmm0
-; SSE41-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; SSE41-NEXT:    pinsrb $13, %eax, %xmm0
-; SSE41-NEXT:    movzbl %r11b, %eax
-; SSE41-NEXT:    pinsrb $14, %eax, %xmm0
-; SSE41-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; SSE41-NEXT:    pinsrb $15, %eax, %xmm0
-; SSE41-NEXT:    psrlw $4, %xmm0
-; SSE41-NEXT:    pand {{.*}}(%rip), %xmm0
-; SSE41-NEXT:    movdqa {{.*#+}} xmm1 = [8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8]
-; SSE41-NEXT:    pxor %xmm1, %xmm0
-; SSE41-NEXT:    psubb %xmm1, %xmm0
-; SSE41-NEXT:    popq %rbx
-; SSE41-NEXT:    popq %r12
-; SSE41-NEXT:    popq %r13
-; SSE41-NEXT:    popq %r14
-; SSE41-NEXT:    popq %r15
-; SSE41-NEXT:    popq %rbp
-; SSE41-NEXT:    retq
+; SSE-LABEL: v16i4:
+; SSE:       # %bb.0:
+; SSE-NEXT:    psllw $4, %xmm1
+; SSE-NEXT:    movdqa {{.*#+}} xmm2 = [240,240,240,240,240,240,240,240,240,240,240,240,240,240,240,240]
+; SSE-NEXT:    pand %xmm2, %xmm1
+; SSE-NEXT:    psllw $4, %xmm0
+; SSE-NEXT:    pand %xmm2, %xmm0
+; SSE-NEXT:    paddsb %xmm1, %xmm0
+; SSE-NEXT:    psrlw $4, %xmm0
+; SSE-NEXT:    pand {{.*}}(%rip), %xmm0
+; SSE-NEXT:    movdqa {{.*#+}} xmm1 = [8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8]
+; SSE-NEXT:    pxor %xmm1, %xmm0
+; SSE-NEXT:    psubb %xmm1, %xmm0
+; SSE-NEXT:    retq
 ;
 ; AVX-LABEL: v16i4:
 ; AVX:       # %bb.0:
 ; AVX-NEXT:    vpsllw $4, %xmm1, %xmm1
 ; AVX-NEXT:    vmovdqa {{.*#+}} xmm2 = [240,240,240,240,240,240,240,240,240,240,240,240,240,240,240,240]
 ; AVX-NEXT:    vpand %xmm2, %xmm1, %xmm1
-; AVX-NEXT:    vpextrb $15, %xmm1, %ecx
 ; AVX-NEXT:    vpsllw $4, %xmm0, %xmm0
 ; AVX-NEXT:    vpand %xmm2, %xmm0, %xmm0
-; AVX-NEXT:    vpextrb $15, %xmm0, %edx
-; AVX-NEXT:    movl %edx, %eax
-; AVX-NEXT:    addb %cl, %al
-; AVX-NEXT:    setns %al
-; AVX-NEXT:    addb %cl, %dl
-; AVX-NEXT:    jno .LBB15_2
-; AVX-NEXT:  # %bb.1:
-; AVX-NEXT:    addb $127, %al
-; AVX-NEXT:    movl %eax, %edx
-; AVX-NEXT:  .LBB15_2:
-; AVX-NEXT:    vpextrb $14, %xmm1, %ecx
-; AVX-NEXT:    vpextrb $14, %xmm0, %r11d
-; AVX-NEXT:    movl %r11d, %eax
-; AVX-NEXT:    addb %cl, %al
-; AVX-NEXT:    setns %al
-; AVX-NEXT:    addb %cl, %r11b
-; AVX-NEXT:    jno .LBB15_4
-; AVX-NEXT:  # %bb.3:
-; AVX-NEXT:    addb $127, %al
-; AVX-NEXT:    movl %eax, %r11d
-; AVX-NEXT:  .LBB15_4:
-; AVX-NEXT:    vpextrb $13, %xmm1, %ecx
-; AVX-NEXT:    vpextrb $13, %xmm0, %edi
-; AVX-NEXT:    movl %edi, %eax
-; AVX-NEXT:    addb %cl, %al
-; AVX-NEXT:    setns %al
-; AVX-NEXT:    addb %cl, %dil
-; AVX-NEXT:    jno .LBB15_6
-; AVX-NEXT:  # %bb.5:
-; AVX-NEXT:    addb $127, %al
-; AVX-NEXT:    movl %eax, %edi
-; AVX-NEXT:  .LBB15_6:
-; AVX-NEXT:    pushq %rbp
-; AVX-NEXT:    pushq %r15
-; AVX-NEXT:    pushq %r14
-; AVX-NEXT:    pushq %r13
-; AVX-NEXT:    pushq %r12
-; AVX-NEXT:    pushq %rbx
-; AVX-NEXT:    vpextrb $12, %xmm1, %ecx
-; AVX-NEXT:    vpextrb $12, %xmm0, %r14d
-; AVX-NEXT:    movl %r14d, %eax
-; AVX-NEXT:    addb %cl, %al
-; AVX-NEXT:    setns %al
-; AVX-NEXT:    addb %cl, %r14b
-; AVX-NEXT:    jno .LBB15_8
-; AVX-NEXT:  # %bb.7:
-; AVX-NEXT:    addb $127, %al
-; AVX-NEXT:    movl %eax, %r14d
-; AVX-NEXT:  .LBB15_8:
-; AVX-NEXT:    vpextrb $11, %xmm1, %ecx
-; AVX-NEXT:    vpextrb $11, %xmm0, %ebp
-; AVX-NEXT:    movl %ebp, %eax
-; AVX-NEXT:    addb %cl, %al
-; AVX-NEXT:    setns %al
-; AVX-NEXT:    addb %cl, %bpl
-; AVX-NEXT:    jno .LBB15_10
-; AVX-NEXT:  # %bb.9:
-; AVX-NEXT:    addb $127, %al
-; AVX-NEXT:    movl %eax, %ebp
-; AVX-NEXT:  .LBB15_10:
-; AVX-NEXT:    vpextrb $10, %xmm1, %ecx
-; AVX-NEXT:    vpextrb $10, %xmm0, %r15d
-; AVX-NEXT:    movl %r15d, %eax
-; AVX-NEXT:    addb %cl, %al
-; AVX-NEXT:    setns %al
-; AVX-NEXT:    addb %cl, %r15b
-; AVX-NEXT:    jno .LBB15_12
-; AVX-NEXT:  # %bb.11:
-; AVX-NEXT:    addb $127, %al
-; AVX-NEXT:    movl %eax, %r15d
-; AVX-NEXT:  .LBB15_12:
-; AVX-NEXT:    vpextrb $9, %xmm1, %ecx
-; AVX-NEXT:    vpextrb $9, %xmm0, %r12d
-; AVX-NEXT:    movl %r12d, %eax
-; AVX-NEXT:    addb %cl, %al
-; AVX-NEXT:    setns %al
-; AVX-NEXT:    addb %cl, %r12b
-; AVX-NEXT:    jno .LBB15_14
-; AVX-NEXT:  # %bb.13:
-; AVX-NEXT:    addb $127, %al
-; AVX-NEXT:    movl %eax, %r12d
-; AVX-NEXT:  .LBB15_14:
-; AVX-NEXT:    vpextrb $8, %xmm1, %ecx
-; AVX-NEXT:    vpextrb $8, %xmm0, %r13d
-; AVX-NEXT:    movl %r13d, %eax
-; AVX-NEXT:    addb %cl, %al
-; AVX-NEXT:    setns %al
-; AVX-NEXT:    addb %cl, %r13b
-; AVX-NEXT:    jno .LBB15_16
-; AVX-NEXT:  # %bb.15:
-; AVX-NEXT:    addb $127, %al
-; AVX-NEXT:    movl %eax, %r13d
-; AVX-NEXT:  .LBB15_16:
-; AVX-NEXT:    vpextrb $7, %xmm1, %ecx
-; AVX-NEXT:    vpextrb $7, %xmm0, %r10d
-; AVX-NEXT:    movl %r10d, %eax
-; AVX-NEXT:    addb %cl, %al
-; AVX-NEXT:    setns %al
-; AVX-NEXT:    addb %cl, %r10b
-; AVX-NEXT:    movl %ebp, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX-NEXT:    jno .LBB15_18
-; AVX-NEXT:  # %bb.17:
-; AVX-NEXT:    addb $127, %al
-; AVX-NEXT:    movl %eax, %r10d
-; AVX-NEXT:  .LBB15_18:
-; AVX-NEXT:    vpextrb $6, %xmm1, %ecx
-; AVX-NEXT:    vpextrb $6, %xmm0, %r9d
-; AVX-NEXT:    movl %r9d, %eax
-; AVX-NEXT:    addb %cl, %al
-; AVX-NEXT:    setns %al
-; AVX-NEXT:    addb %cl, %r9b
-; AVX-NEXT:    jno .LBB15_20
-; AVX-NEXT:  # %bb.19:
-; AVX-NEXT:    addb $127, %al
-; AVX-NEXT:    movl %eax, %r9d
-; AVX-NEXT:  .LBB15_20:
-; AVX-NEXT:    vpextrb $5, %xmm1, %ecx
-; AVX-NEXT:    vpextrb $5, %xmm0, %ebp
-; AVX-NEXT:    movl %ebp, %eax
-; AVX-NEXT:    addb %cl, %al
-; AVX-NEXT:    setns %al
-; AVX-NEXT:    addb %cl, %bpl
-; AVX-NEXT:    movl %edi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX-NEXT:    jno .LBB15_22
-; AVX-NEXT:  # %bb.21:
-; AVX-NEXT:    addb $127, %al
-; AVX-NEXT:    movl %eax, %ebp
-; AVX-NEXT:  .LBB15_22:
-; AVX-NEXT:    vpextrb $4, %xmm1, %ecx
-; AVX-NEXT:    vpextrb $4, %xmm0, %edi
-; AVX-NEXT:    movl %edi, %eax
-; AVX-NEXT:    addb %cl, %al
-; AVX-NEXT:    setns %al
-; AVX-NEXT:    addb %cl, %dil
-; AVX-NEXT:    movl %edx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX-NEXT:    jno .LBB15_24
-; AVX-NEXT:  # %bb.23:
-; AVX-NEXT:    addb $127, %al
-; AVX-NEXT:    movl %eax, %edi
-; AVX-NEXT:  .LBB15_24:
-; AVX-NEXT:    vpextrb $3, %xmm1, %edx
-; AVX-NEXT:    vpextrb $3, %xmm0, %eax
-; AVX-NEXT:    movl %eax, %ecx
-; AVX-NEXT:    addb %dl, %cl
-; AVX-NEXT:    setns %cl
-; AVX-NEXT:    addb %dl, %al
-; AVX-NEXT:    jno .LBB15_26
-; AVX-NEXT:  # %bb.25:
-; AVX-NEXT:    addb $127, %cl
-; AVX-NEXT:    movl %ecx, %eax
-; AVX-NEXT:  .LBB15_26:
-; AVX-NEXT:    vpextrb $2, %xmm1, %ebx
-; AVX-NEXT:    vpextrb $2, %xmm0, %ecx
-; AVX-NEXT:    movl %ecx, %edx
-; AVX-NEXT:    addb %bl, %dl
-; AVX-NEXT:    setns %dl
-; AVX-NEXT:    addb %bl, %cl
-; AVX-NEXT:    jno .LBB15_28
-; AVX-NEXT:  # %bb.27:
-; AVX-NEXT:    addb $127, %dl
-; AVX-NEXT:    movl %edx, %ecx
-; AVX-NEXT:  .LBB15_28:
-; AVX-NEXT:    vpextrb $0, %xmm1, %esi
-; AVX-NEXT:    vpextrb $0, %xmm0, %edx
-; AVX-NEXT:    movl %edx, %ebx
-; AVX-NEXT:    addb %sil, %bl
-; AVX-NEXT:    setns %bl
-; AVX-NEXT:    addb %sil, %dl
-; AVX-NEXT:    jno .LBB15_30
-; AVX-NEXT:  # %bb.29:
-; AVX-NEXT:    addb $127, %bl
-; AVX-NEXT:    movl %ebx, %edx
-; AVX-NEXT:  .LBB15_30:
-; AVX-NEXT:    vpextrb $1, %xmm1, %esi
-; AVX-NEXT:    vpextrb $1, %xmm0, %r8d
-; AVX-NEXT:    movl %r8d, %ebx
-; AVX-NEXT:    addb %sil, %bl
-; AVX-NEXT:    setns %bl
-; AVX-NEXT:    addb %sil, %r8b
-; AVX-NEXT:    jno .LBB15_32
-; AVX-NEXT:  # %bb.31:
-; AVX-NEXT:    addb $127, %bl
-; AVX-NEXT:    movl %ebx, %r8d
-; AVX-NEXT:  .LBB15_32:
-; AVX-NEXT:    movzbl %dl, %edx
-; AVX-NEXT:    vmovd %edx, %xmm0
-; AVX-NEXT:    movzbl %r8b, %edx
-; AVX-NEXT:    vpinsrb $1, %edx, %xmm0, %xmm0
-; AVX-NEXT:    movzbl %cl, %ecx
-; AVX-NEXT:    vpinsrb $2, %ecx, %xmm0, %xmm0
-; AVX-NEXT:    movzbl %al, %eax
-; AVX-NEXT:    vpinsrb $3, %eax, %xmm0, %xmm0
-; AVX-NEXT:    movzbl %dil, %eax
-; AVX-NEXT:    vpinsrb $4, %eax, %xmm0, %xmm0
-; AVX-NEXT:    movzbl %bpl, %eax
-; AVX-NEXT:    vpinsrb $5, %eax, %xmm0, %xmm0
-; AVX-NEXT:    movzbl %r9b, %eax
-; AVX-NEXT:    vpinsrb $6, %eax, %xmm0, %xmm0
-; AVX-NEXT:    movzbl %r10b, %eax
-; AVX-NEXT:    vpinsrb $7, %eax, %xmm0, %xmm0
-; AVX-NEXT:    movzbl %r13b, %eax
-; AVX-NEXT:    vpinsrb $8, %eax, %xmm0, %xmm0
-; AVX-NEXT:    movzbl %r12b, %eax
-; AVX-NEXT:    vpinsrb $9, %eax, %xmm0, %xmm0
-; AVX-NEXT:    movzbl %r15b, %eax
-; AVX-NEXT:    vpinsrb $10, %eax, %xmm0, %xmm0
-; AVX-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX-NEXT:    vpinsrb $11, %eax, %xmm0, %xmm0
-; AVX-NEXT:    movzbl %r14b, %eax
-; AVX-NEXT:    vpinsrb $12, %eax, %xmm0, %xmm0
-; AVX-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX-NEXT:    vpinsrb $13, %eax, %xmm0, %xmm0
-; AVX-NEXT:    movzbl %r11b, %eax
-; AVX-NEXT:    vpinsrb $14, %eax, %xmm0, %xmm0
-; AVX-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX-NEXT:    vpinsrb $15, %eax, %xmm0, %xmm0
+; AVX-NEXT:    vpaddsb %xmm1, %xmm0, %xmm0
 ; AVX-NEXT:    vpsrlw $4, %xmm0, %xmm0
 ; AVX-NEXT:    vpand {{.*}}(%rip), %xmm0, %xmm0
 ; AVX-NEXT:    vmovdqa {{.*#+}} xmm1 = [8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8]
 ; AVX-NEXT:    vpxor %xmm1, %xmm0, %xmm0
 ; AVX-NEXT:    vpsubb %xmm1, %xmm0, %xmm0
-; AVX-NEXT:    popq %rbx
-; AVX-NEXT:    popq %r12
-; AVX-NEXT:    popq %r13
-; AVX-NEXT:    popq %r14
-; AVX-NEXT:    popq %r15
-; AVX-NEXT:    popq %rbp
 ; AVX-NEXT:    retq
   %z = call <16 x i4> @llvm.sadd.sat.v16i4(<16 x i4> %x, <16 x i4> %y)
   ret <16 x i4> %z
 }
 
 define <16 x i1> @v16i1(<16 x i1> %x, <16 x i1> %y) nounwind {
-; SSE2-LABEL: v16i1:
-; SSE2:       # %bb.0:
-; SSE2-NEXT:    pushq %rbp
-; SSE2-NEXT:    pushq %r15
-; SSE2-NEXT:    pushq %r14
-; SSE2-NEXT:    pushq %r13
-; SSE2-NEXT:    pushq %r12
-; SSE2-NEXT:    pushq %rbx
-; SSE2-NEXT:    psllw $7, %xmm1
-; SSE2-NEXT:    movdqa {{.*#+}} xmm2 = [128,128,128,128,128,128,128,128,128,128,128,128,128,128,128,128]
-; SSE2-NEXT:    pand %xmm2, %xmm1
-; SSE2-NEXT:    movdqa %xmm1, -{{[0-9]+}}(%rsp)
-; SSE2-NEXT:    psllw $7, %xmm0
-; SSE2-NEXT:    pand %xmm2, %xmm0
-; SSE2-NEXT:    movdqa %xmm0, -{{[0-9]+}}(%rsp)
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %r9b
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %sil
-; SSE2-NEXT:    movl %r9d, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %r9b
-; SSE2-NEXT:    jno .LBB16_2
-; SSE2-NEXT:  # %bb.1:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %r9d
-; SSE2-NEXT:  .LBB16_2:
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movl %esi, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %sil
-; SSE2-NEXT:    jno .LBB16_4
-; SSE2-NEXT:  # %bb.3:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %esi
-; SSE2-NEXT:  .LBB16_4:
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %bl
-; SSE2-NEXT:    movl %ebx, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %bl
-; SSE2-NEXT:    jno .LBB16_6
-; SSE2-NEXT:  # %bb.5:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %ebx
-; SSE2-NEXT:  .LBB16_6:
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %dl
-; SSE2-NEXT:    movl %edx, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %dl
-; SSE2-NEXT:    jno .LBB16_8
-; SSE2-NEXT:  # %bb.7:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %edx
-; SSE2-NEXT:  .LBB16_8:
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %r10b
-; SSE2-NEXT:    movl %r10d, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %r10b
-; SSE2-NEXT:    jno .LBB16_10
-; SSE2-NEXT:  # %bb.9:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %r10d
-; SSE2-NEXT:  .LBB16_10:
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %r11b
-; SSE2-NEXT:    movl %r11d, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %r11b
-; SSE2-NEXT:    jno .LBB16_12
-; SSE2-NEXT:  # %bb.11:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %r11d
-; SSE2-NEXT:  .LBB16_12:
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %bpl
-; SSE2-NEXT:    movl %ebp, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %bpl
-; SSE2-NEXT:    jno .LBB16_14
-; SSE2-NEXT:  # %bb.13:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %ebp
-; SSE2-NEXT:  .LBB16_14:
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %r14b
-; SSE2-NEXT:    movl %r14d, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %r14b
-; SSE2-NEXT:    jno .LBB16_16
-; SSE2-NEXT:  # %bb.15:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %r14d
-; SSE2-NEXT:  .LBB16_16:
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %r15b
-; SSE2-NEXT:    movl %r15d, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %r15b
-; SSE2-NEXT:    jno .LBB16_18
-; SSE2-NEXT:  # %bb.17:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %r15d
-; SSE2-NEXT:  .LBB16_18:
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %r12b
-; SSE2-NEXT:    movl %r12d, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %r12b
-; SSE2-NEXT:    jno .LBB16_20
-; SSE2-NEXT:  # %bb.19:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %r12d
-; SSE2-NEXT:  .LBB16_20:
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %r13b
-; SSE2-NEXT:    movl %r13d, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %r13b
-; SSE2-NEXT:    movb %bl, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSE2-NEXT:    movb %dl, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSE2-NEXT:    jno .LBB16_22
-; SSE2-NEXT:  # %bb.21:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %r13d
-; SSE2-NEXT:  .LBB16_22:
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %dil
-; SSE2-NEXT:    movl %edi, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %dil
-; SSE2-NEXT:    jno .LBB16_24
-; SSE2-NEXT:  # %bb.23:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %edi
-; SSE2-NEXT:  .LBB16_24:
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %r8b
-; SSE2-NEXT:    movl %r8d, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %r8b
-; SSE2-NEXT:    jno .LBB16_26
-; SSE2-NEXT:  # %bb.25:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %r8d
-; SSE2-NEXT:  .LBB16_26:
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %bl
-; SSE2-NEXT:    movl %ebx, %eax
-; SSE2-NEXT:    addb %cl, %al
-; SSE2-NEXT:    setns %al
-; SSE2-NEXT:    addb %cl, %bl
-; SSE2-NEXT:    jno .LBB16_28
-; SSE2-NEXT:  # %bb.27:
-; SSE2-NEXT:    addb $127, %al
-; SSE2-NEXT:    movl %eax, %ebx
-; SSE2-NEXT:  .LBB16_28:
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %dl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
-; SSE2-NEXT:    movl %eax, %ecx
-; SSE2-NEXT:    addb %dl, %cl
-; SSE2-NEXT:    setns %cl
-; SSE2-NEXT:    addb %dl, %al
-; SSE2-NEXT:    movb %sil, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSE2-NEXT:    jno .LBB16_30
-; SSE2-NEXT:  # %bb.29:
-; SSE2-NEXT:    addb $127, %cl
-; SSE2-NEXT:    movl %ecx, %eax
-; SSE2-NEXT:  .LBB16_30:
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %dl
-; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %sil
-; SSE2-NEXT:    movl %esi, %ecx
-; SSE2-NEXT:    addb %dl, %cl
-; SSE2-NEXT:    setns %cl
-; SSE2-NEXT:    addb %dl, %sil
-; SSE2-NEXT:    jno .LBB16_32
-; SSE2-NEXT:  # %bb.31:
-; SSE2-NEXT:    addb $127, %cl
-; SSE2-NEXT:    movl %ecx, %esi
-; SSE2-NEXT:  .LBB16_32:
-; SSE2-NEXT:    movzbl %sil, %ecx
-; SSE2-NEXT:    movd %ecx, %xmm0
-; SSE2-NEXT:    movzbl %al, %eax
-; SSE2-NEXT:    movd %eax, %xmm1
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3],xmm1[4],xmm0[4],xmm1[5],xmm0[5],xmm1[6],xmm0[6],xmm1[7],xmm0[7]
-; SSE2-NEXT:    movzbl %bl, %eax
-; SSE2-NEXT:    movd %eax, %xmm0
-; SSE2-NEXT:    movzbl %r8b, %eax
-; SSE2-NEXT:    movd %eax, %xmm2
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm2 = xmm2[0],xmm0[0],xmm2[1],xmm0[1],xmm2[2],xmm0[2],xmm2[3],xmm0[3],xmm2[4],xmm0[4],xmm2[5],xmm0[5],xmm2[6],xmm0[6],xmm2[7],xmm0[7]
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm2 = xmm2[0],xmm1[0],xmm2[1],xmm1[1],xmm2[2],xmm1[2],xmm2[3],xmm1[3]
-; SSE2-NEXT:    movzbl %dil, %eax
-; SSE2-NEXT:    movd %eax, %xmm0
-; SSE2-NEXT:    movzbl %r13b, %eax
-; SSE2-NEXT:    movd %eax, %xmm1
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3],xmm1[4],xmm0[4],xmm1[5],xmm0[5],xmm1[6],xmm0[6],xmm1[7],xmm0[7]
-; SSE2-NEXT:    movzbl %r12b, %eax
-; SSE2-NEXT:    movd %eax, %xmm3
-; SSE2-NEXT:    movzbl %r15b, %eax
-; SSE2-NEXT:    movd %eax, %xmm0
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm3[0],xmm0[1],xmm3[1],xmm0[2],xmm3[2],xmm0[3],xmm3[3],xmm0[4],xmm3[4],xmm0[5],xmm3[5],xmm0[6],xmm3[6],xmm0[7],xmm3[7]
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
-; SSE2-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1]
-; SSE2-NEXT:    movzbl %r14b, %eax
-; SSE2-NEXT:    movd %eax, %xmm1
-; SSE2-NEXT:    movzbl %bpl, %eax
-; SSE2-NEXT:    movd %eax, %xmm2
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm2 = xmm2[0],xmm1[0],xmm2[1],xmm1[1],xmm2[2],xmm1[2],xmm2[3],xmm1[3],xmm2[4],xmm1[4],xmm2[5],xmm1[5],xmm2[6],xmm1[6],xmm2[7],xmm1[7]
-; SSE2-NEXT:    movzbl %r11b, %eax
-; SSE2-NEXT:    movd %eax, %xmm1
-; SSE2-NEXT:    movzbl %r10b, %eax
-; SSE2-NEXT:    movd %eax, %xmm3
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm3 = xmm3[0],xmm1[0],xmm3[1],xmm1[1],xmm3[2],xmm1[2],xmm3[3],xmm1[3],xmm3[4],xmm1[4],xmm3[5],xmm1[5],xmm3[6],xmm1[6],xmm3[7],xmm1[7]
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm3 = xmm3[0],xmm2[0],xmm3[1],xmm2[1],xmm3[2],xmm2[2],xmm3[3],xmm2[3]
-; SSE2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; SSE2-NEXT:    movd %eax, %xmm1
-; SSE2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; SSE2-NEXT:    movd %eax, %xmm2
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm2 = xmm2[0],xmm1[0],xmm2[1],xmm1[1],xmm2[2],xmm1[2],xmm2[3],xmm1[3],xmm2[4],xmm1[4],xmm2[5],xmm1[5],xmm2[6],xmm1[6],xmm2[7],xmm1[7]
-; SSE2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; SSE2-NEXT:    movd %eax, %xmm1
-; SSE2-NEXT:    movzbl %r9b, %eax
-; SSE2-NEXT:    movd %eax, %xmm4
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm4 = xmm4[0],xmm1[0],xmm4[1],xmm1[1],xmm4[2],xmm1[2],xmm4[3],xmm1[3],xmm4[4],xmm1[4],xmm4[5],xmm1[5],xmm4[6],xmm1[6],xmm4[7],xmm1[7]
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm4 = xmm4[0],xmm2[0],xmm4[1],xmm2[1],xmm4[2],xmm2[2],xmm4[3],xmm2[3]
-; SSE2-NEXT:    punpckldq {{.*#+}} xmm4 = xmm4[0],xmm3[0],xmm4[1],xmm3[1]
-; SSE2-NEXT:    punpcklqdq {{.*#+}} xmm4 = xmm4[0],xmm0[0]
-; SSE2-NEXT:    pxor %xmm0, %xmm0
-; SSE2-NEXT:    pcmpgtb %xmm4, %xmm0
-; SSE2-NEXT:    popq %rbx
-; SSE2-NEXT:    popq %r12
-; SSE2-NEXT:    popq %r13
-; SSE2-NEXT:    popq %r14
-; SSE2-NEXT:    popq %r15
-; SSE2-NEXT:    popq %rbp
-; SSE2-NEXT:    retq
-;
-; SSSE3-LABEL: v16i1:
-; SSSE3:       # %bb.0:
-; SSSE3-NEXT:    pushq %rbp
-; SSSE3-NEXT:    pushq %r15
-; SSSE3-NEXT:    pushq %r14
-; SSSE3-NEXT:    pushq %r13
-; SSSE3-NEXT:    pushq %r12
-; SSSE3-NEXT:    pushq %rbx
-; SSSE3-NEXT:    psllw $7, %xmm1
-; SSSE3-NEXT:    movdqa {{.*#+}} xmm2 = [128,128,128,128,128,128,128,128,128,128,128,128,128,128,128,128]
-; SSSE3-NEXT:    pand %xmm2, %xmm1
-; SSSE3-NEXT:    movdqa %xmm1, -{{[0-9]+}}(%rsp)
-; SSSE3-NEXT:    psllw $7, %xmm0
-; SSSE3-NEXT:    pand %xmm2, %xmm0
-; SSSE3-NEXT:    movdqa %xmm0, -{{[0-9]+}}(%rsp)
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %r9b
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %sil
-; SSSE3-NEXT:    movl %r9d, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %r9b
-; SSSE3-NEXT:    jno .LBB16_2
-; SSSE3-NEXT:  # %bb.1:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %r9d
-; SSSE3-NEXT:  .LBB16_2:
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movl %esi, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %sil
-; SSSE3-NEXT:    jno .LBB16_4
-; SSSE3-NEXT:  # %bb.3:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %esi
-; SSSE3-NEXT:  .LBB16_4:
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %bl
-; SSSE3-NEXT:    movl %ebx, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %bl
-; SSSE3-NEXT:    jno .LBB16_6
-; SSSE3-NEXT:  # %bb.5:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %ebx
-; SSSE3-NEXT:  .LBB16_6:
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %dl
-; SSSE3-NEXT:    movl %edx, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %dl
-; SSSE3-NEXT:    jno .LBB16_8
-; SSSE3-NEXT:  # %bb.7:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %edx
-; SSSE3-NEXT:  .LBB16_8:
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %r10b
-; SSSE3-NEXT:    movl %r10d, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %r10b
-; SSSE3-NEXT:    jno .LBB16_10
-; SSSE3-NEXT:  # %bb.9:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %r10d
-; SSSE3-NEXT:  .LBB16_10:
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %r11b
-; SSSE3-NEXT:    movl %r11d, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %r11b
-; SSSE3-NEXT:    jno .LBB16_12
-; SSSE3-NEXT:  # %bb.11:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %r11d
-; SSSE3-NEXT:  .LBB16_12:
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %bpl
-; SSSE3-NEXT:    movl %ebp, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %bpl
-; SSSE3-NEXT:    jno .LBB16_14
-; SSSE3-NEXT:  # %bb.13:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %ebp
-; SSSE3-NEXT:  .LBB16_14:
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %r14b
-; SSSE3-NEXT:    movl %r14d, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %r14b
-; SSSE3-NEXT:    jno .LBB16_16
-; SSSE3-NEXT:  # %bb.15:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %r14d
-; SSSE3-NEXT:  .LBB16_16:
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %r15b
-; SSSE3-NEXT:    movl %r15d, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %r15b
-; SSSE3-NEXT:    jno .LBB16_18
-; SSSE3-NEXT:  # %bb.17:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %r15d
-; SSSE3-NEXT:  .LBB16_18:
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %r12b
-; SSSE3-NEXT:    movl %r12d, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %r12b
-; SSSE3-NEXT:    jno .LBB16_20
-; SSSE3-NEXT:  # %bb.19:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %r12d
-; SSSE3-NEXT:  .LBB16_20:
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %r13b
-; SSSE3-NEXT:    movl %r13d, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %r13b
-; SSSE3-NEXT:    movb %bl, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSSE3-NEXT:    movb %dl, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSSE3-NEXT:    jno .LBB16_22
-; SSSE3-NEXT:  # %bb.21:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %r13d
-; SSSE3-NEXT:  .LBB16_22:
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %dil
-; SSSE3-NEXT:    movl %edi, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %dil
-; SSSE3-NEXT:    jno .LBB16_24
-; SSSE3-NEXT:  # %bb.23:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %edi
-; SSSE3-NEXT:  .LBB16_24:
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %r8b
-; SSSE3-NEXT:    movl %r8d, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %r8b
-; SSSE3-NEXT:    jno .LBB16_26
-; SSSE3-NEXT:  # %bb.25:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %r8d
-; SSSE3-NEXT:  .LBB16_26:
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %bl
-; SSSE3-NEXT:    movl %ebx, %eax
-; SSSE3-NEXT:    addb %cl, %al
-; SSSE3-NEXT:    setns %al
-; SSSE3-NEXT:    addb %cl, %bl
-; SSSE3-NEXT:    jno .LBB16_28
-; SSSE3-NEXT:  # %bb.27:
-; SSSE3-NEXT:    addb $127, %al
-; SSSE3-NEXT:    movl %eax, %ebx
-; SSSE3-NEXT:  .LBB16_28:
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %dl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %al
-; SSSE3-NEXT:    movl %eax, %ecx
-; SSSE3-NEXT:    addb %dl, %cl
-; SSSE3-NEXT:    setns %cl
-; SSSE3-NEXT:    addb %dl, %al
-; SSSE3-NEXT:    movb %sil, {{[-0-9]+}}(%r{{[sb]}}p) # 1-byte Spill
-; SSSE3-NEXT:    jno .LBB16_30
-; SSSE3-NEXT:  # %bb.29:
-; SSSE3-NEXT:    addb $127, %cl
-; SSSE3-NEXT:    movl %ecx, %eax
-; SSSE3-NEXT:  .LBB16_30:
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %dl
-; SSSE3-NEXT:    movb -{{[0-9]+}}(%rsp), %sil
-; SSSE3-NEXT:    movl %esi, %ecx
-; SSSE3-NEXT:    addb %dl, %cl
-; SSSE3-NEXT:    setns %cl
-; SSSE3-NEXT:    addb %dl, %sil
-; SSSE3-NEXT:    jno .LBB16_32
-; SSSE3-NEXT:  # %bb.31:
-; SSSE3-NEXT:    addb $127, %cl
-; SSSE3-NEXT:    movl %ecx, %esi
-; SSSE3-NEXT:  .LBB16_32:
-; SSSE3-NEXT:    movzbl %sil, %ecx
-; SSSE3-NEXT:    movd %ecx, %xmm0
-; SSSE3-NEXT:    movzbl %al, %eax
-; SSSE3-NEXT:    movd %eax, %xmm1
-; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3],xmm1[4],xmm0[4],xmm1[5],xmm0[5],xmm1[6],xmm0[6],xmm1[7],xmm0[7]
-; SSSE3-NEXT:    movzbl %bl, %eax
-; SSSE3-NEXT:    movd %eax, %xmm0
-; SSSE3-NEXT:    movzbl %r8b, %eax
-; SSSE3-NEXT:    movd %eax, %xmm2
-; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm2 = xmm2[0],xmm0[0],xmm2[1],xmm0[1],xmm2[2],xmm0[2],xmm2[3],xmm0[3],xmm2[4],xmm0[4],xmm2[5],xmm0[5],xmm2[6],xmm0[6],xmm2[7],xmm0[7]
-; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm2 = xmm2[0],xmm1[0],xmm2[1],xmm1[1],xmm2[2],xmm1[2],xmm2[3],xmm1[3]
-; SSSE3-NEXT:    movzbl %dil, %eax
-; SSSE3-NEXT:    movd %eax, %xmm0
-; SSSE3-NEXT:    movzbl %r13b, %eax
-; SSSE3-NEXT:    movd %eax, %xmm1
-; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3],xmm1[4],xmm0[4],xmm1[5],xmm0[5],xmm1[6],xmm0[6],xmm1[7],xmm0[7]
-; SSSE3-NEXT:    movzbl %r12b, %eax
-; SSSE3-NEXT:    movd %eax, %xmm3
-; SSSE3-NEXT:    movzbl %r15b, %eax
-; SSSE3-NEXT:    movd %eax, %xmm0
-; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm3[0],xmm0[1],xmm3[1],xmm0[2],xmm3[2],xmm0[3],xmm3[3],xmm0[4],xmm3[4],xmm0[5],xmm3[5],xmm0[6],xmm3[6],xmm0[7],xmm3[7]
-; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
-; SSSE3-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1]
-; SSSE3-NEXT:    movzbl %r14b, %eax
-; SSSE3-NEXT:    movd %eax, %xmm1
-; SSSE3-NEXT:    movzbl %bpl, %eax
-; SSSE3-NEXT:    movd %eax, %xmm2
-; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm2 = xmm2[0],xmm1[0],xmm2[1],xmm1[1],xmm2[2],xmm1[2],xmm2[3],xmm1[3],xmm2[4],xmm1[4],xmm2[5],xmm1[5],xmm2[6],xmm1[6],xmm2[7],xmm1[7]
-; SSSE3-NEXT:    movzbl %r11b, %eax
-; SSSE3-NEXT:    movd %eax, %xmm1
-; SSSE3-NEXT:    movzbl %r10b, %eax
-; SSSE3-NEXT:    movd %eax, %xmm3
-; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm3 = xmm3[0],xmm1[0],xmm3[1],xmm1[1],xmm3[2],xmm1[2],xmm3[3],xmm1[3],xmm3[4],xmm1[4],xmm3[5],xmm1[5],xmm3[6],xmm1[6],xmm3[7],xmm1[7]
-; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm3 = xmm3[0],xmm2[0],xmm3[1],xmm2[1],xmm3[2],xmm2[2],xmm3[3],xmm2[3]
-; SSSE3-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; SSSE3-NEXT:    movd %eax, %xmm1
-; SSSE3-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; SSSE3-NEXT:    movd %eax, %xmm2
-; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm2 = xmm2[0],xmm1[0],xmm2[1],xmm1[1],xmm2[2],xmm1[2],xmm2[3],xmm1[3],xmm2[4],xmm1[4],xmm2[5],xmm1[5],xmm2[6],xmm1[6],xmm2[7],xmm1[7]
-; SSSE3-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; SSSE3-NEXT:    movd %eax, %xmm1
-; SSSE3-NEXT:    movzbl %r9b, %eax
-; SSSE3-NEXT:    movd %eax, %xmm4
-; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm4 = xmm4[0],xmm1[0],xmm4[1],xmm1[1],xmm4[2],xmm1[2],xmm4[3],xmm1[3],xmm4[4],xmm1[4],xmm4[5],xmm1[5],xmm4[6],xmm1[6],xmm4[7],xmm1[7]
-; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm4 = xmm4[0],xmm2[0],xmm4[1],xmm2[1],xmm4[2],xmm2[2],xmm4[3],xmm2[3]
-; SSSE3-NEXT:    punpckldq {{.*#+}} xmm4 = xmm4[0],xmm3[0],xmm4[1],xmm3[1]
-; SSSE3-NEXT:    punpcklqdq {{.*#+}} xmm4 = xmm4[0],xmm0[0]
-; SSSE3-NEXT:    pxor %xmm0, %xmm0
-; SSSE3-NEXT:    pcmpgtb %xmm4, %xmm0
-; SSSE3-NEXT:    popq %rbx
-; SSSE3-NEXT:    popq %r12
-; SSSE3-NEXT:    popq %r13
-; SSSE3-NEXT:    popq %r14
-; SSSE3-NEXT:    popq %r15
-; SSSE3-NEXT:    popq %rbp
-; SSSE3-NEXT:    retq
-;
-; SSE41-LABEL: v16i1:
-; SSE41:       # %bb.0:
-; SSE41-NEXT:    psllw $7, %xmm1
-; SSE41-NEXT:    movdqa {{.*#+}} xmm2 = [128,128,128,128,128,128,128,128,128,128,128,128,128,128,128,128]
-; SSE41-NEXT:    pand %xmm2, %xmm1
-; SSE41-NEXT:    pextrb $15, %xmm1, %ecx
-; SSE41-NEXT:    psllw $7, %xmm0
-; SSE41-NEXT:    pand %xmm2, %xmm0
-; SSE41-NEXT:    pextrb $15, %xmm0, %edx
-; SSE41-NEXT:    movl %edx, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %dl
-; SSE41-NEXT:    jno .LBB16_2
-; SSE41-NEXT:  # %bb.1:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %edx
-; SSE41-NEXT:  .LBB16_2:
-; SSE41-NEXT:    pextrb $14, %xmm1, %ecx
-; SSE41-NEXT:    pextrb $14, %xmm0, %r11d
-; SSE41-NEXT:    movl %r11d, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %r11b
-; SSE41-NEXT:    jno .LBB16_4
-; SSE41-NEXT:  # %bb.3:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %r11d
-; SSE41-NEXT:  .LBB16_4:
-; SSE41-NEXT:    pextrb $13, %xmm1, %ecx
-; SSE41-NEXT:    pextrb $13, %xmm0, %edi
-; SSE41-NEXT:    movl %edi, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %dil
-; SSE41-NEXT:    jno .LBB16_6
-; SSE41-NEXT:  # %bb.5:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %edi
-; SSE41-NEXT:  .LBB16_6:
-; SSE41-NEXT:    pushq %rbp
-; SSE41-NEXT:    pushq %r15
-; SSE41-NEXT:    pushq %r14
-; SSE41-NEXT:    pushq %r13
-; SSE41-NEXT:    pushq %r12
-; SSE41-NEXT:    pushq %rbx
-; SSE41-NEXT:    pextrb $12, %xmm1, %ecx
-; SSE41-NEXT:    pextrb $12, %xmm0, %r14d
-; SSE41-NEXT:    movl %r14d, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %r14b
-; SSE41-NEXT:    jno .LBB16_8
-; SSE41-NEXT:  # %bb.7:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %r14d
-; SSE41-NEXT:  .LBB16_8:
-; SSE41-NEXT:    pextrb $11, %xmm1, %ecx
-; SSE41-NEXT:    pextrb $11, %xmm0, %ebp
-; SSE41-NEXT:    movl %ebp, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %bpl
-; SSE41-NEXT:    jno .LBB16_10
-; SSE41-NEXT:  # %bb.9:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %ebp
-; SSE41-NEXT:  .LBB16_10:
-; SSE41-NEXT:    pextrb $10, %xmm1, %ecx
-; SSE41-NEXT:    pextrb $10, %xmm0, %r15d
-; SSE41-NEXT:    movl %r15d, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %r15b
-; SSE41-NEXT:    jno .LBB16_12
-; SSE41-NEXT:  # %bb.11:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %r15d
-; SSE41-NEXT:  .LBB16_12:
-; SSE41-NEXT:    pextrb $9, %xmm1, %ecx
-; SSE41-NEXT:    pextrb $9, %xmm0, %r12d
-; SSE41-NEXT:    movl %r12d, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %r12b
-; SSE41-NEXT:    jno .LBB16_14
-; SSE41-NEXT:  # %bb.13:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %r12d
-; SSE41-NEXT:  .LBB16_14:
-; SSE41-NEXT:    pextrb $8, %xmm1, %ecx
-; SSE41-NEXT:    pextrb $8, %xmm0, %r13d
-; SSE41-NEXT:    movl %r13d, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %r13b
-; SSE41-NEXT:    jno .LBB16_16
-; SSE41-NEXT:  # %bb.15:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %r13d
-; SSE41-NEXT:  .LBB16_16:
-; SSE41-NEXT:    pextrb $7, %xmm1, %ecx
-; SSE41-NEXT:    pextrb $7, %xmm0, %r10d
-; SSE41-NEXT:    movl %r10d, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %r10b
-; SSE41-NEXT:    movl %ebp, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE41-NEXT:    jno .LBB16_18
-; SSE41-NEXT:  # %bb.17:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %r10d
-; SSE41-NEXT:  .LBB16_18:
-; SSE41-NEXT:    pextrb $6, %xmm1, %ecx
-; SSE41-NEXT:    pextrb $6, %xmm0, %r9d
-; SSE41-NEXT:    movl %r9d, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %r9b
-; SSE41-NEXT:    jno .LBB16_20
-; SSE41-NEXT:  # %bb.19:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %r9d
-; SSE41-NEXT:  .LBB16_20:
-; SSE41-NEXT:    pextrb $5, %xmm1, %ecx
-; SSE41-NEXT:    pextrb $5, %xmm0, %ebp
-; SSE41-NEXT:    movl %ebp, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %bpl
-; SSE41-NEXT:    movl %edi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE41-NEXT:    jno .LBB16_22
-; SSE41-NEXT:  # %bb.21:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %ebp
-; SSE41-NEXT:  .LBB16_22:
-; SSE41-NEXT:    pextrb $4, %xmm1, %ecx
-; SSE41-NEXT:    pextrb $4, %xmm0, %edi
-; SSE41-NEXT:    movl %edi, %eax
-; SSE41-NEXT:    addb %cl, %al
-; SSE41-NEXT:    setns %al
-; SSE41-NEXT:    addb %cl, %dil
-; SSE41-NEXT:    movl %edx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; SSE41-NEXT:    jno .LBB16_24
-; SSE41-NEXT:  # %bb.23:
-; SSE41-NEXT:    addb $127, %al
-; SSE41-NEXT:    movl %eax, %edi
-; SSE41-NEXT:  .LBB16_24:
-; SSE41-NEXT:    pextrb $3, %xmm1, %edx
-; SSE41-NEXT:    pextrb $3, %xmm0, %eax
-; SSE41-NEXT:    movl %eax, %ecx
-; SSE41-NEXT:    addb %dl, %cl
-; SSE41-NEXT:    setns %cl
-; SSE41-NEXT:    addb %dl, %al
-; SSE41-NEXT:    jno .LBB16_26
-; SSE41-NEXT:  # %bb.25:
-; SSE41-NEXT:    addb $127, %cl
-; SSE41-NEXT:    movl %ecx, %eax
-; SSE41-NEXT:  .LBB16_26:
-; SSE41-NEXT:    pextrb $2, %xmm1, %ebx
-; SSE41-NEXT:    pextrb $2, %xmm0, %ecx
-; SSE41-NEXT:    movl %ecx, %edx
-; SSE41-NEXT:    addb %bl, %dl
-; SSE41-NEXT:    setns %dl
-; SSE41-NEXT:    addb %bl, %cl
-; SSE41-NEXT:    jno .LBB16_28
-; SSE41-NEXT:  # %bb.27:
-; SSE41-NEXT:    addb $127, %dl
-; SSE41-NEXT:    movl %edx, %ecx
-; SSE41-NEXT:  .LBB16_28:
-; SSE41-NEXT:    pextrb $0, %xmm1, %esi
-; SSE41-NEXT:    pextrb $0, %xmm0, %edx
-; SSE41-NEXT:    movl %edx, %ebx
-; SSE41-NEXT:    addb %sil, %bl
-; SSE41-NEXT:    setns %bl
-; SSE41-NEXT:    addb %sil, %dl
-; SSE41-NEXT:    jno .LBB16_30
-; SSE41-NEXT:  # %bb.29:
-; SSE41-NEXT:    addb $127, %bl
-; SSE41-NEXT:    movl %ebx, %edx
-; SSE41-NEXT:  .LBB16_30:
-; SSE41-NEXT:    pextrb $1, %xmm1, %esi
-; SSE41-NEXT:    pextrb $1, %xmm0, %r8d
-; SSE41-NEXT:    movl %r8d, %ebx
-; SSE41-NEXT:    addb %sil, %bl
-; SSE41-NEXT:    setns %bl
-; SSE41-NEXT:    addb %sil, %r8b
-; SSE41-NEXT:    jno .LBB16_32
-; SSE41-NEXT:  # %bb.31:
-; SSE41-NEXT:    addb $127, %bl
-; SSE41-NEXT:    movl %ebx, %r8d
-; SSE41-NEXT:  .LBB16_32:
-; SSE41-NEXT:    movzbl %dl, %edx
-; SSE41-NEXT:    movd %edx, %xmm1
-; SSE41-NEXT:    movzbl %r8b, %edx
-; SSE41-NEXT:    pinsrb $1, %edx, %xmm1
-; SSE41-NEXT:    movzbl %cl, %ecx
-; SSE41-NEXT:    pinsrb $2, %ecx, %xmm1
-; SSE41-NEXT:    movzbl %al, %eax
-; SSE41-NEXT:    pinsrb $3, %eax, %xmm1
-; SSE41-NEXT:    movzbl %dil, %eax
-; SSE41-NEXT:    pinsrb $4, %eax, %xmm1
-; SSE41-NEXT:    movzbl %bpl, %eax
-; SSE41-NEXT:    pinsrb $5, %eax, %xmm1
-; SSE41-NEXT:    movzbl %r9b, %eax
-; SSE41-NEXT:    pinsrb $6, %eax, %xmm1
-; SSE41-NEXT:    movzbl %r10b, %eax
-; SSE41-NEXT:    pinsrb $7, %eax, %xmm1
-; SSE41-NEXT:    movzbl %r13b, %eax
-; SSE41-NEXT:    pinsrb $8, %eax, %xmm1
-; SSE41-NEXT:    movzbl %r12b, %eax
-; SSE41-NEXT:    pinsrb $9, %eax, %xmm1
-; SSE41-NEXT:    movzbl %r15b, %eax
-; SSE41-NEXT:    pinsrb $10, %eax, %xmm1
-; SSE41-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; SSE41-NEXT:    pinsrb $11, %eax, %xmm1
-; SSE41-NEXT:    movzbl %r14b, %eax
-; SSE41-NEXT:    pinsrb $12, %eax, %xmm1
-; SSE41-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; SSE41-NEXT:    pinsrb $13, %eax, %xmm1
-; SSE41-NEXT:    movzbl %r11b, %eax
-; SSE41-NEXT:    pinsrb $14, %eax, %xmm1
-; SSE41-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; SSE41-NEXT:    pinsrb $15, %eax, %xmm1
-; SSE41-NEXT:    pxor %xmm0, %xmm0
-; SSE41-NEXT:    pcmpgtb %xmm1, %xmm0
-; SSE41-NEXT:    popq %rbx
-; SSE41-NEXT:    popq %r12
-; SSE41-NEXT:    popq %r13
-; SSE41-NEXT:    popq %r14
-; SSE41-NEXT:    popq %r15
-; SSE41-NEXT:    popq %rbp
-; SSE41-NEXT:    retq
+; SSE-LABEL: v16i1:
+; SSE:       # %bb.0:
+; SSE-NEXT:    psllw $7, %xmm1
+; SSE-NEXT:    movdqa {{.*#+}} xmm2 = [128,128,128,128,128,128,128,128,128,128,128,128,128,128,128,128]
+; SSE-NEXT:    pand %xmm2, %xmm1
+; SSE-NEXT:    psllw $7, %xmm0
+; SSE-NEXT:    pand %xmm2, %xmm0
+; SSE-NEXT:    paddsb %xmm1, %xmm0
+; SSE-NEXT:    pxor %xmm1, %xmm1
+; SSE-NEXT:    pcmpgtb %xmm0, %xmm1
+; SSE-NEXT:    movdqa %xmm1, %xmm0
+; SSE-NEXT:    retq
 ;
 ; AVX1-LABEL: v16i1:
 ; AVX1:       # %bb.0:
 ; AVX1-NEXT:    vpsllw $7, %xmm1, %xmm1
 ; AVX1-NEXT:    vmovdqa {{.*#+}} xmm2 = [128,128,128,128,128,128,128,128,128,128,128,128,128,128,128,128]
 ; AVX1-NEXT:    vpand %xmm2, %xmm1, %xmm1
-; AVX1-NEXT:    vpextrb $15, %xmm1, %ecx
 ; AVX1-NEXT:    vpsllw $7, %xmm0, %xmm0
 ; AVX1-NEXT:    vpand %xmm2, %xmm0, %xmm0
-; AVX1-NEXT:    vpextrb $15, %xmm0, %edx
-; AVX1-NEXT:    movl %edx, %eax
-; AVX1-NEXT:    addb %cl, %al
-; AVX1-NEXT:    setns %al
-; AVX1-NEXT:    addb %cl, %dl
-; AVX1-NEXT:    jno .LBB16_2
-; AVX1-NEXT:  # %bb.1:
-; AVX1-NEXT:    addb $127, %al
-; AVX1-NEXT:    movl %eax, %edx
-; AVX1-NEXT:  .LBB16_2:
-; AVX1-NEXT:    vpextrb $14, %xmm1, %ecx
-; AVX1-NEXT:    vpextrb $14, %xmm0, %r11d
-; AVX1-NEXT:    movl %r11d, %eax
-; AVX1-NEXT:    addb %cl, %al
-; AVX1-NEXT:    setns %al
-; AVX1-NEXT:    addb %cl, %r11b
-; AVX1-NEXT:    jno .LBB16_4
-; AVX1-NEXT:  # %bb.3:
-; AVX1-NEXT:    addb $127, %al
-; AVX1-NEXT:    movl %eax, %r11d
-; AVX1-NEXT:  .LBB16_4:
-; AVX1-NEXT:    vpextrb $13, %xmm1, %ecx
-; AVX1-NEXT:    vpextrb $13, %xmm0, %edi
-; AVX1-NEXT:    movl %edi, %eax
-; AVX1-NEXT:    addb %cl, %al
-; AVX1-NEXT:    setns %al
-; AVX1-NEXT:    addb %cl, %dil
-; AVX1-NEXT:    jno .LBB16_6
-; AVX1-NEXT:  # %bb.5:
-; AVX1-NEXT:    addb $127, %al
-; AVX1-NEXT:    movl %eax, %edi
-; AVX1-NEXT:  .LBB16_6:
-; AVX1-NEXT:    pushq %rbp
-; AVX1-NEXT:    pushq %r15
-; AVX1-NEXT:    pushq %r14
-; AVX1-NEXT:    pushq %r13
-; AVX1-NEXT:    pushq %r12
-; AVX1-NEXT:    pushq %rbx
-; AVX1-NEXT:    vpextrb $12, %xmm1, %ecx
-; AVX1-NEXT:    vpextrb $12, %xmm0, %r14d
-; AVX1-NEXT:    movl %r14d, %eax
-; AVX1-NEXT:    addb %cl, %al
-; AVX1-NEXT:    setns %al
-; AVX1-NEXT:    addb %cl, %r14b
-; AVX1-NEXT:    jno .LBB16_8
-; AVX1-NEXT:  # %bb.7:
-; AVX1-NEXT:    addb $127, %al
-; AVX1-NEXT:    movl %eax, %r14d
-; AVX1-NEXT:  .LBB16_8:
-; AVX1-NEXT:    vpextrb $11, %xmm1, %ecx
-; AVX1-NEXT:    vpextrb $11, %xmm0, %ebp
-; AVX1-NEXT:    movl %ebp, %eax
-; AVX1-NEXT:    addb %cl, %al
-; AVX1-NEXT:    setns %al
-; AVX1-NEXT:    addb %cl, %bpl
-; AVX1-NEXT:    jno .LBB16_10
-; AVX1-NEXT:  # %bb.9:
-; AVX1-NEXT:    addb $127, %al
-; AVX1-NEXT:    movl %eax, %ebp
-; AVX1-NEXT:  .LBB16_10:
-; AVX1-NEXT:    vpextrb $10, %xmm1, %ecx
-; AVX1-NEXT:    vpextrb $10, %xmm0, %r15d
-; AVX1-NEXT:    movl %r15d, %eax
-; AVX1-NEXT:    addb %cl, %al
-; AVX1-NEXT:    setns %al
-; AVX1-NEXT:    addb %cl, %r15b
-; AVX1-NEXT:    jno .LBB16_12
-; AVX1-NEXT:  # %bb.11:
-; AVX1-NEXT:    addb $127, %al
-; AVX1-NEXT:    movl %eax, %r15d
-; AVX1-NEXT:  .LBB16_12:
-; AVX1-NEXT:    vpextrb $9, %xmm1, %ecx
-; AVX1-NEXT:    vpextrb $9, %xmm0, %r12d
-; AVX1-NEXT:    movl %r12d, %eax
-; AVX1-NEXT:    addb %cl, %al
-; AVX1-NEXT:    setns %al
-; AVX1-NEXT:    addb %cl, %r12b
-; AVX1-NEXT:    jno .LBB16_14
-; AVX1-NEXT:  # %bb.13:
-; AVX1-NEXT:    addb $127, %al
-; AVX1-NEXT:    movl %eax, %r12d
-; AVX1-NEXT:  .LBB16_14:
-; AVX1-NEXT:    vpextrb $8, %xmm1, %ecx
-; AVX1-NEXT:    vpextrb $8, %xmm0, %r13d
-; AVX1-NEXT:    movl %r13d, %eax
-; AVX1-NEXT:    addb %cl, %al
-; AVX1-NEXT:    setns %al
-; AVX1-NEXT:    addb %cl, %r13b
-; AVX1-NEXT:    jno .LBB16_16
-; AVX1-NEXT:  # %bb.15:
-; AVX1-NEXT:    addb $127, %al
-; AVX1-NEXT:    movl %eax, %r13d
-; AVX1-NEXT:  .LBB16_16:
-; AVX1-NEXT:    vpextrb $7, %xmm1, %ecx
-; AVX1-NEXT:    vpextrb $7, %xmm0, %r10d
-; AVX1-NEXT:    movl %r10d, %eax
-; AVX1-NEXT:    addb %cl, %al
-; AVX1-NEXT:    setns %al
-; AVX1-NEXT:    addb %cl, %r10b
-; AVX1-NEXT:    movl %ebp, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX1-NEXT:    jno .LBB16_18
-; AVX1-NEXT:  # %bb.17:
-; AVX1-NEXT:    addb $127, %al
-; AVX1-NEXT:    movl %eax, %r10d
-; AVX1-NEXT:  .LBB16_18:
-; AVX1-NEXT:    vpextrb $6, %xmm1, %ecx
-; AVX1-NEXT:    vpextrb $6, %xmm0, %r9d
-; AVX1-NEXT:    movl %r9d, %eax
-; AVX1-NEXT:    addb %cl, %al
-; AVX1-NEXT:    setns %al
-; AVX1-NEXT:    addb %cl, %r9b
-; AVX1-NEXT:    jno .LBB16_20
-; AVX1-NEXT:  # %bb.19:
-; AVX1-NEXT:    addb $127, %al
-; AVX1-NEXT:    movl %eax, %r9d
-; AVX1-NEXT:  .LBB16_20:
-; AVX1-NEXT:    vpextrb $5, %xmm1, %ecx
-; AVX1-NEXT:    vpextrb $5, %xmm0, %ebp
-; AVX1-NEXT:    movl %ebp, %eax
-; AVX1-NEXT:    addb %cl, %al
-; AVX1-NEXT:    setns %al
-; AVX1-NEXT:    addb %cl, %bpl
-; AVX1-NEXT:    movl %edi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX1-NEXT:    jno .LBB16_22
-; AVX1-NEXT:  # %bb.21:
-; AVX1-NEXT:    addb $127, %al
-; AVX1-NEXT:    movl %eax, %ebp
-; AVX1-NEXT:  .LBB16_22:
-; AVX1-NEXT:    vpextrb $4, %xmm1, %ecx
-; AVX1-NEXT:    vpextrb $4, %xmm0, %edi
-; AVX1-NEXT:    movl %edi, %eax
-; AVX1-NEXT:    addb %cl, %al
-; AVX1-NEXT:    setns %al
-; AVX1-NEXT:    addb %cl, %dil
-; AVX1-NEXT:    movl %edx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX1-NEXT:    jno .LBB16_24
-; AVX1-NEXT:  # %bb.23:
-; AVX1-NEXT:    addb $127, %al
-; AVX1-NEXT:    movl %eax, %edi
-; AVX1-NEXT:  .LBB16_24:
-; AVX1-NEXT:    vpextrb $3, %xmm1, %edx
-; AVX1-NEXT:    vpextrb $3, %xmm0, %eax
-; AVX1-NEXT:    movl %eax, %ecx
-; AVX1-NEXT:    addb %dl, %cl
-; AVX1-NEXT:    setns %cl
-; AVX1-NEXT:    addb %dl, %al
-; AVX1-NEXT:    jno .LBB16_26
-; AVX1-NEXT:  # %bb.25:
-; AVX1-NEXT:    addb $127, %cl
-; AVX1-NEXT:    movl %ecx, %eax
-; AVX1-NEXT:  .LBB16_26:
-; AVX1-NEXT:    vpextrb $2, %xmm1, %ebx
-; AVX1-NEXT:    vpextrb $2, %xmm0, %ecx
-; AVX1-NEXT:    movl %ecx, %edx
-; AVX1-NEXT:    addb %bl, %dl
-; AVX1-NEXT:    setns %dl
-; AVX1-NEXT:    addb %bl, %cl
-; AVX1-NEXT:    jno .LBB16_28
-; AVX1-NEXT:  # %bb.27:
-; AVX1-NEXT:    addb $127, %dl
-; AVX1-NEXT:    movl %edx, %ecx
-; AVX1-NEXT:  .LBB16_28:
-; AVX1-NEXT:    vpextrb $0, %xmm1, %esi
-; AVX1-NEXT:    vpextrb $0, %xmm0, %edx
-; AVX1-NEXT:    movl %edx, %ebx
-; AVX1-NEXT:    addb %sil, %bl
-; AVX1-NEXT:    setns %bl
-; AVX1-NEXT:    addb %sil, %dl
-; AVX1-NEXT:    jno .LBB16_30
-; AVX1-NEXT:  # %bb.29:
-; AVX1-NEXT:    addb $127, %bl
-; AVX1-NEXT:    movl %ebx, %edx
-; AVX1-NEXT:  .LBB16_30:
-; AVX1-NEXT:    vpextrb $1, %xmm1, %esi
-; AVX1-NEXT:    vpextrb $1, %xmm0, %r8d
-; AVX1-NEXT:    movl %r8d, %ebx
-; AVX1-NEXT:    addb %sil, %bl
-; AVX1-NEXT:    setns %bl
-; AVX1-NEXT:    addb %sil, %r8b
-; AVX1-NEXT:    jno .LBB16_32
-; AVX1-NEXT:  # %bb.31:
-; AVX1-NEXT:    addb $127, %bl
-; AVX1-NEXT:    movl %ebx, %r8d
-; AVX1-NEXT:  .LBB16_32:
-; AVX1-NEXT:    movzbl %dl, %edx
-; AVX1-NEXT:    vmovd %edx, %xmm0
-; AVX1-NEXT:    movzbl %r8b, %edx
-; AVX1-NEXT:    vpinsrb $1, %edx, %xmm0, %xmm0
-; AVX1-NEXT:    movzbl %cl, %ecx
-; AVX1-NEXT:    vpinsrb $2, %ecx, %xmm0, %xmm0
-; AVX1-NEXT:    movzbl %al, %eax
-; AVX1-NEXT:    vpinsrb $3, %eax, %xmm0, %xmm0
-; AVX1-NEXT:    movzbl %dil, %eax
-; AVX1-NEXT:    vpinsrb $4, %eax, %xmm0, %xmm0
-; AVX1-NEXT:    movzbl %bpl, %eax
-; AVX1-NEXT:    vpinsrb $5, %eax, %xmm0, %xmm0
-; AVX1-NEXT:    movzbl %r9b, %eax
-; AVX1-NEXT:    vpinsrb $6, %eax, %xmm0, %xmm0
-; AVX1-NEXT:    movzbl %r10b, %eax
-; AVX1-NEXT:    vpinsrb $7, %eax, %xmm0, %xmm0
-; AVX1-NEXT:    movzbl %r13b, %eax
-; AVX1-NEXT:    vpinsrb $8, %eax, %xmm0, %xmm0
-; AVX1-NEXT:    movzbl %r12b, %eax
-; AVX1-NEXT:    vpinsrb $9, %eax, %xmm0, %xmm0
-; AVX1-NEXT:    movzbl %r15b, %eax
-; AVX1-NEXT:    vpinsrb $10, %eax, %xmm0, %xmm0
-; AVX1-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX1-NEXT:    vpinsrb $11, %eax, %xmm0, %xmm0
-; AVX1-NEXT:    movzbl %r14b, %eax
-; AVX1-NEXT:    vpinsrb $12, %eax, %xmm0, %xmm0
-; AVX1-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX1-NEXT:    vpinsrb $13, %eax, %xmm0, %xmm0
-; AVX1-NEXT:    movzbl %r11b, %eax
-; AVX1-NEXT:    vpinsrb $14, %eax, %xmm0, %xmm0
-; AVX1-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX1-NEXT:    vpinsrb $15, %eax, %xmm0, %xmm0
+; AVX1-NEXT:    vpaddsb %xmm1, %xmm0, %xmm0
 ; AVX1-NEXT:    vpxor %xmm1, %xmm1, %xmm1
 ; AVX1-NEXT:    vpcmpgtb %xmm0, %xmm1, %xmm0
-; AVX1-NEXT:    popq %rbx
-; AVX1-NEXT:    popq %r12
-; AVX1-NEXT:    popq %r13
-; AVX1-NEXT:    popq %r14
-; AVX1-NEXT:    popq %r15
-; AVX1-NEXT:    popq %rbp
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: v16i1:
@@ -18833,604 +629,21 @@ define <16 x i1> @v16i1(<16 x i1> %x, <1
 ; AVX2-NEXT:    vpsllw $7, %xmm1, %xmm1
 ; AVX2-NEXT:    vmovdqa {{.*#+}} xmm2 = [128,128,128,128,128,128,128,128,128,128,128,128,128,128,128,128]
 ; AVX2-NEXT:    vpand %xmm2, %xmm1, %xmm1
-; AVX2-NEXT:    vpextrb $15, %xmm1, %ecx
 ; AVX2-NEXT:    vpsllw $7, %xmm0, %xmm0
 ; AVX2-NEXT:    vpand %xmm2, %xmm0, %xmm0
-; AVX2-NEXT:    vpextrb $15, %xmm0, %edx
-; AVX2-NEXT:    movl %edx, %eax
-; AVX2-NEXT:    addb %cl, %al
-; AVX2-NEXT:    setns %al
-; AVX2-NEXT:    addb %cl, %dl
-; AVX2-NEXT:    jno .LBB16_2
-; AVX2-NEXT:  # %bb.1:
-; AVX2-NEXT:    addb $127, %al
-; AVX2-NEXT:    movl %eax, %edx
-; AVX2-NEXT:  .LBB16_2:
-; AVX2-NEXT:    vpextrb $14, %xmm1, %ecx
-; AVX2-NEXT:    vpextrb $14, %xmm0, %r11d
-; AVX2-NEXT:    movl %r11d, %eax
-; AVX2-NEXT:    addb %cl, %al
-; AVX2-NEXT:    setns %al
-; AVX2-NEXT:    addb %cl, %r11b
-; AVX2-NEXT:    jno .LBB16_4
-; AVX2-NEXT:  # %bb.3:
-; AVX2-NEXT:    addb $127, %al
-; AVX2-NEXT:    movl %eax, %r11d
-; AVX2-NEXT:  .LBB16_4:
-; AVX2-NEXT:    vpextrb $13, %xmm1, %ecx
-; AVX2-NEXT:    vpextrb $13, %xmm0, %edi
-; AVX2-NEXT:    movl %edi, %eax
-; AVX2-NEXT:    addb %cl, %al
-; AVX2-NEXT:    setns %al
-; AVX2-NEXT:    addb %cl, %dil
-; AVX2-NEXT:    jno .LBB16_6
-; AVX2-NEXT:  # %bb.5:
-; AVX2-NEXT:    addb $127, %al
-; AVX2-NEXT:    movl %eax, %edi
-; AVX2-NEXT:  .LBB16_6:
-; AVX2-NEXT:    pushq %rbp
-; AVX2-NEXT:    pushq %r15
-; AVX2-NEXT:    pushq %r14
-; AVX2-NEXT:    pushq %r13
-; AVX2-NEXT:    pushq %r12
-; AVX2-NEXT:    pushq %rbx
-; AVX2-NEXT:    vpextrb $12, %xmm1, %ecx
-; AVX2-NEXT:    vpextrb $12, %xmm0, %r14d
-; AVX2-NEXT:    movl %r14d, %eax
-; AVX2-NEXT:    addb %cl, %al
-; AVX2-NEXT:    setns %al
-; AVX2-NEXT:    addb %cl, %r14b
-; AVX2-NEXT:    jno .LBB16_8
-; AVX2-NEXT:  # %bb.7:
-; AVX2-NEXT:    addb $127, %al
-; AVX2-NEXT:    movl %eax, %r14d
-; AVX2-NEXT:  .LBB16_8:
-; AVX2-NEXT:    vpextrb $11, %xmm1, %ecx
-; AVX2-NEXT:    vpextrb $11, %xmm0, %ebp
-; AVX2-NEXT:    movl %ebp, %eax
-; AVX2-NEXT:    addb %cl, %al
-; AVX2-NEXT:    setns %al
-; AVX2-NEXT:    addb %cl, %bpl
-; AVX2-NEXT:    jno .LBB16_10
-; AVX2-NEXT:  # %bb.9:
-; AVX2-NEXT:    addb $127, %al
-; AVX2-NEXT:    movl %eax, %ebp
-; AVX2-NEXT:  .LBB16_10:
-; AVX2-NEXT:    vpextrb $10, %xmm1, %ecx
-; AVX2-NEXT:    vpextrb $10, %xmm0, %r15d
-; AVX2-NEXT:    movl %r15d, %eax
-; AVX2-NEXT:    addb %cl, %al
-; AVX2-NEXT:    setns %al
-; AVX2-NEXT:    addb %cl, %r15b
-; AVX2-NEXT:    jno .LBB16_12
-; AVX2-NEXT:  # %bb.11:
-; AVX2-NEXT:    addb $127, %al
-; AVX2-NEXT:    movl %eax, %r15d
-; AVX2-NEXT:  .LBB16_12:
-; AVX2-NEXT:    vpextrb $9, %xmm1, %ecx
-; AVX2-NEXT:    vpextrb $9, %xmm0, %r12d
-; AVX2-NEXT:    movl %r12d, %eax
-; AVX2-NEXT:    addb %cl, %al
-; AVX2-NEXT:    setns %al
-; AVX2-NEXT:    addb %cl, %r12b
-; AVX2-NEXT:    jno .LBB16_14
-; AVX2-NEXT:  # %bb.13:
-; AVX2-NEXT:    addb $127, %al
-; AVX2-NEXT:    movl %eax, %r12d
-; AVX2-NEXT:  .LBB16_14:
-; AVX2-NEXT:    vpextrb $8, %xmm1, %ecx
-; AVX2-NEXT:    vpextrb $8, %xmm0, %r13d
-; AVX2-NEXT:    movl %r13d, %eax
-; AVX2-NEXT:    addb %cl, %al
-; AVX2-NEXT:    setns %al
-; AVX2-NEXT:    addb %cl, %r13b
-; AVX2-NEXT:    jno .LBB16_16
-; AVX2-NEXT:  # %bb.15:
-; AVX2-NEXT:    addb $127, %al
-; AVX2-NEXT:    movl %eax, %r13d
-; AVX2-NEXT:  .LBB16_16:
-; AVX2-NEXT:    vpextrb $7, %xmm1, %ecx
-; AVX2-NEXT:    vpextrb $7, %xmm0, %r10d
-; AVX2-NEXT:    movl %r10d, %eax
-; AVX2-NEXT:    addb %cl, %al
-; AVX2-NEXT:    setns %al
-; AVX2-NEXT:    addb %cl, %r10b
-; AVX2-NEXT:    movl %ebp, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX2-NEXT:    jno .LBB16_18
-; AVX2-NEXT:  # %bb.17:
-; AVX2-NEXT:    addb $127, %al
-; AVX2-NEXT:    movl %eax, %r10d
-; AVX2-NEXT:  .LBB16_18:
-; AVX2-NEXT:    vpextrb $6, %xmm1, %ecx
-; AVX2-NEXT:    vpextrb $6, %xmm0, %r9d
-; AVX2-NEXT:    movl %r9d, %eax
-; AVX2-NEXT:    addb %cl, %al
-; AVX2-NEXT:    setns %al
-; AVX2-NEXT:    addb %cl, %r9b
-; AVX2-NEXT:    jno .LBB16_20
-; AVX2-NEXT:  # %bb.19:
-; AVX2-NEXT:    addb $127, %al
-; AVX2-NEXT:    movl %eax, %r9d
-; AVX2-NEXT:  .LBB16_20:
-; AVX2-NEXT:    vpextrb $5, %xmm1, %ecx
-; AVX2-NEXT:    vpextrb $5, %xmm0, %ebp
-; AVX2-NEXT:    movl %ebp, %eax
-; AVX2-NEXT:    addb %cl, %al
-; AVX2-NEXT:    setns %al
-; AVX2-NEXT:    addb %cl, %bpl
-; AVX2-NEXT:    movl %edi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX2-NEXT:    jno .LBB16_22
-; AVX2-NEXT:  # %bb.21:
-; AVX2-NEXT:    addb $127, %al
-; AVX2-NEXT:    movl %eax, %ebp
-; AVX2-NEXT:  .LBB16_22:
-; AVX2-NEXT:    vpextrb $4, %xmm1, %ecx
-; AVX2-NEXT:    vpextrb $4, %xmm0, %edi
-; AVX2-NEXT:    movl %edi, %eax
-; AVX2-NEXT:    addb %cl, %al
-; AVX2-NEXT:    setns %al
-; AVX2-NEXT:    addb %cl, %dil
-; AVX2-NEXT:    movl %edx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX2-NEXT:    jno .LBB16_24
-; AVX2-NEXT:  # %bb.23:
-; AVX2-NEXT:    addb $127, %al
-; AVX2-NEXT:    movl %eax, %edi
-; AVX2-NEXT:  .LBB16_24:
-; AVX2-NEXT:    vpextrb $3, %xmm1, %edx
-; AVX2-NEXT:    vpextrb $3, %xmm0, %eax
-; AVX2-NEXT:    movl %eax, %ecx
-; AVX2-NEXT:    addb %dl, %cl
-; AVX2-NEXT:    setns %cl
-; AVX2-NEXT:    addb %dl, %al
-; AVX2-NEXT:    jno .LBB16_26
-; AVX2-NEXT:  # %bb.25:
-; AVX2-NEXT:    addb $127, %cl
-; AVX2-NEXT:    movl %ecx, %eax
-; AVX2-NEXT:  .LBB16_26:
-; AVX2-NEXT:    vpextrb $2, %xmm1, %ebx
-; AVX2-NEXT:    vpextrb $2, %xmm0, %ecx
-; AVX2-NEXT:    movl %ecx, %edx
-; AVX2-NEXT:    addb %bl, %dl
-; AVX2-NEXT:    setns %dl
-; AVX2-NEXT:    addb %bl, %cl
-; AVX2-NEXT:    jno .LBB16_28
-; AVX2-NEXT:  # %bb.27:
-; AVX2-NEXT:    addb $127, %dl
-; AVX2-NEXT:    movl %edx, %ecx
-; AVX2-NEXT:  .LBB16_28:
-; AVX2-NEXT:    vpextrb $0, %xmm1, %esi
-; AVX2-NEXT:    vpextrb $0, %xmm0, %edx
-; AVX2-NEXT:    movl %edx, %ebx
-; AVX2-NEXT:    addb %sil, %bl
-; AVX2-NEXT:    setns %bl
-; AVX2-NEXT:    addb %sil, %dl
-; AVX2-NEXT:    jno .LBB16_30
-; AVX2-NEXT:  # %bb.29:
-; AVX2-NEXT:    addb $127, %bl
-; AVX2-NEXT:    movl %ebx, %edx
-; AVX2-NEXT:  .LBB16_30:
-; AVX2-NEXT:    vpextrb $1, %xmm1, %esi
-; AVX2-NEXT:    vpextrb $1, %xmm0, %r8d
-; AVX2-NEXT:    movl %r8d, %ebx
-; AVX2-NEXT:    addb %sil, %bl
-; AVX2-NEXT:    setns %bl
-; AVX2-NEXT:    addb %sil, %r8b
-; AVX2-NEXT:    jno .LBB16_32
-; AVX2-NEXT:  # %bb.31:
-; AVX2-NEXT:    addb $127, %bl
-; AVX2-NEXT:    movl %ebx, %r8d
-; AVX2-NEXT:  .LBB16_32:
-; AVX2-NEXT:    movzbl %dl, %edx
-; AVX2-NEXT:    vmovd %edx, %xmm0
-; AVX2-NEXT:    movzbl %r8b, %edx
-; AVX2-NEXT:    vpinsrb $1, %edx, %xmm0, %xmm0
-; AVX2-NEXT:    movzbl %cl, %ecx
-; AVX2-NEXT:    vpinsrb $2, %ecx, %xmm0, %xmm0
-; AVX2-NEXT:    movzbl %al, %eax
-; AVX2-NEXT:    vpinsrb $3, %eax, %xmm0, %xmm0
-; AVX2-NEXT:    movzbl %dil, %eax
-; AVX2-NEXT:    vpinsrb $4, %eax, %xmm0, %xmm0
-; AVX2-NEXT:    movzbl %bpl, %eax
-; AVX2-NEXT:    vpinsrb $5, %eax, %xmm0, %xmm0
-; AVX2-NEXT:    movzbl %r9b, %eax
-; AVX2-NEXT:    vpinsrb $6, %eax, %xmm0, %xmm0
-; AVX2-NEXT:    movzbl %r10b, %eax
-; AVX2-NEXT:    vpinsrb $7, %eax, %xmm0, %xmm0
-; AVX2-NEXT:    movzbl %r13b, %eax
-; AVX2-NEXT:    vpinsrb $8, %eax, %xmm0, %xmm0
-; AVX2-NEXT:    movzbl %r12b, %eax
-; AVX2-NEXT:    vpinsrb $9, %eax, %xmm0, %xmm0
-; AVX2-NEXT:    movzbl %r15b, %eax
-; AVX2-NEXT:    vpinsrb $10, %eax, %xmm0, %xmm0
-; AVX2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX2-NEXT:    vpinsrb $11, %eax, %xmm0, %xmm0
-; AVX2-NEXT:    movzbl %r14b, %eax
-; AVX2-NEXT:    vpinsrb $12, %eax, %xmm0, %xmm0
-; AVX2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX2-NEXT:    vpinsrb $13, %eax, %xmm0, %xmm0
-; AVX2-NEXT:    movzbl %r11b, %eax
-; AVX2-NEXT:    vpinsrb $14, %eax, %xmm0, %xmm0
-; AVX2-NEXT:    movzbl {{[-0-9]+}}(%r{{[sb]}}p), %eax # 1-byte Folded Reload
-; AVX2-NEXT:    vpinsrb $15, %eax, %xmm0, %xmm0
+; AVX2-NEXT:    vpaddsb %xmm1, %xmm0, %xmm0
 ; AVX2-NEXT:    vpxor %xmm1, %xmm1, %xmm1
 ; AVX2-NEXT:    vpcmpgtb %xmm0, %xmm1, %xmm0
-; AVX2-NEXT:    popq %rbx
-; AVX2-NEXT:    popq %r12
-; AVX2-NEXT:    popq %r13
-; AVX2-NEXT:    popq %r14
-; AVX2-NEXT:    popq %r15
-; AVX2-NEXT:    popq %rbp
 ; AVX2-NEXT:    retq
 ;
 ; AVX512-LABEL: v16i1:
 ; AVX512:       # %bb.0:
+; AVX512-NEXT:    vpsllw $7, %xmm1, %xmm1
+; AVX512-NEXT:    vpmovb2m %xmm1, %k0
 ; AVX512-NEXT:    vpsllw $7, %xmm0, %xmm0
-; AVX512-NEXT:    vpmovb2m %xmm0, %k0
-; AVX512-NEXT:    kshiftrw $1, %k0, %k1
-; AVX512-NEXT:    kmovd %k1, %edx
-; AVX512-NEXT:    vpsllw $7, %xmm1, %xmm0
 ; AVX512-NEXT:    vpmovb2m %xmm0, %k1
-; AVX512-NEXT:    kshiftrw $1, %k1, %k2
-; AVX512-NEXT:    kmovd %k2, %eax
-; AVX512-NEXT:    shlb $7, %al
-; AVX512-NEXT:    shlb $7, %dl
-; AVX512-NEXT:    movl %edx, %ecx
-; AVX512-NEXT:    addb %al, %cl
-; AVX512-NEXT:    setns %cl
-; AVX512-NEXT:    addb %al, %dl
-; AVX512-NEXT:    kmovd %k0, %esi
-; AVX512-NEXT:    kmovd %k1, %eax
-; AVX512-NEXT:    jno .LBB16_2
-; AVX512-NEXT:  # %bb.1:
-; AVX512-NEXT:    addb $127, %cl
-; AVX512-NEXT:    movl %ecx, %edx
-; AVX512-NEXT:  .LBB16_2:
-; AVX512-NEXT:    shlb $7, %al
-; AVX512-NEXT:    shlb $7, %sil
-; AVX512-NEXT:    movl %esi, %ecx
-; AVX512-NEXT:    addb %al, %cl
-; AVX512-NEXT:    setns %cl
-; AVX512-NEXT:    addb %al, %sil
-; AVX512-NEXT:    kshiftrw $2, %k0, %k2
-; AVX512-NEXT:    kmovd %k2, %edi
-; AVX512-NEXT:    kshiftrw $2, %k1, %k2
-; AVX512-NEXT:    kmovd %k2, %eax
-; AVX512-NEXT:    jno .LBB16_4
-; AVX512-NEXT:  # %bb.3:
-; AVX512-NEXT:    addb $127, %cl
-; AVX512-NEXT:    movl %ecx, %esi
-; AVX512-NEXT:  .LBB16_4:
-; AVX512-NEXT:    shlb $7, %al
-; AVX512-NEXT:    shlb $7, %dil
-; AVX512-NEXT:    movl %edi, %ecx
-; AVX512-NEXT:    addb %al, %cl
-; AVX512-NEXT:    setns %cl
-; AVX512-NEXT:    addb %al, %dil
-; AVX512-NEXT:    kshiftrw $3, %k0, %k2
-; AVX512-NEXT:    kmovd %k2, %r11d
-; AVX512-NEXT:    kshiftrw $3, %k1, %k2
-; AVX512-NEXT:    kmovd %k2, %eax
-; AVX512-NEXT:    jno .LBB16_6
-; AVX512-NEXT:  # %bb.5:
-; AVX512-NEXT:    addb $127, %cl
-; AVX512-NEXT:    movl %ecx, %edi
-; AVX512-NEXT:  .LBB16_6:
-; AVX512-NEXT:    pushq %rbp
-; AVX512-NEXT:    pushq %r15
-; AVX512-NEXT:    pushq %r14
-; AVX512-NEXT:    pushq %r13
-; AVX512-NEXT:    pushq %r12
-; AVX512-NEXT:    pushq %rbx
-; AVX512-NEXT:    shlb $7, %al
-; AVX512-NEXT:    shlb $7, %r11b
-; AVX512-NEXT:    movl %r11d, %ecx
-; AVX512-NEXT:    addb %al, %cl
-; AVX512-NEXT:    setns %cl
-; AVX512-NEXT:    addb %al, %r11b
-; AVX512-NEXT:    kshiftrw $4, %k0, %k2
-; AVX512-NEXT:    kmovd %k2, %r14d
-; AVX512-NEXT:    kshiftrw $4, %k1, %k2
-; AVX512-NEXT:    kmovd %k2, %eax
-; AVX512-NEXT:    jno .LBB16_8
-; AVX512-NEXT:  # %bb.7:
-; AVX512-NEXT:    addb $127, %cl
-; AVX512-NEXT:    movl %ecx, %r11d
-; AVX512-NEXT:  .LBB16_8:
-; AVX512-NEXT:    shlb $7, %al
-; AVX512-NEXT:    shlb $7, %r14b
-; AVX512-NEXT:    movl %r14d, %ecx
-; AVX512-NEXT:    addb %al, %cl
-; AVX512-NEXT:    setns %cl
-; AVX512-NEXT:    addb %al, %r14b
-; AVX512-NEXT:    kshiftrw $5, %k0, %k2
-; AVX512-NEXT:    kmovd %k2, %r15d
-; AVX512-NEXT:    kshiftrw $5, %k1, %k2
-; AVX512-NEXT:    kmovd %k2, %eax
-; AVX512-NEXT:    jno .LBB16_10
-; AVX512-NEXT:  # %bb.9:
-; AVX512-NEXT:    addb $127, %cl
-; AVX512-NEXT:    movl %ecx, %r14d
-; AVX512-NEXT:  .LBB16_10:
-; AVX512-NEXT:    shlb $7, %al
-; AVX512-NEXT:    shlb $7, %r15b
-; AVX512-NEXT:    movl %r15d, %ecx
-; AVX512-NEXT:    addb %al, %cl
-; AVX512-NEXT:    setns %cl
-; AVX512-NEXT:    addb %al, %r15b
-; AVX512-NEXT:    kshiftrw $6, %k0, %k2
-; AVX512-NEXT:    kmovd %k2, %r12d
-; AVX512-NEXT:    kshiftrw $6, %k1, %k2
-; AVX512-NEXT:    kmovd %k2, %eax
-; AVX512-NEXT:    jno .LBB16_12
-; AVX512-NEXT:  # %bb.11:
-; AVX512-NEXT:    addb $127, %cl
-; AVX512-NEXT:    movl %ecx, %r15d
-; AVX512-NEXT:  .LBB16_12:
-; AVX512-NEXT:    shlb $7, %al
-; AVX512-NEXT:    shlb $7, %r12b
-; AVX512-NEXT:    movl %r12d, %ecx
-; AVX512-NEXT:    addb %al, %cl
-; AVX512-NEXT:    setns %cl
-; AVX512-NEXT:    addb %al, %r12b
-; AVX512-NEXT:    kshiftrw $7, %k0, %k2
-; AVX512-NEXT:    kmovd %k2, %r13d
-; AVX512-NEXT:    kshiftrw $7, %k1, %k2
-; AVX512-NEXT:    kmovd %k2, %eax
-; AVX512-NEXT:    jno .LBB16_14
-; AVX512-NEXT:  # %bb.13:
-; AVX512-NEXT:    addb $127, %cl
-; AVX512-NEXT:    movl %ecx, %r12d
-; AVX512-NEXT:  .LBB16_14:
-; AVX512-NEXT:    shlb $7, %al
-; AVX512-NEXT:    shlb $7, %r13b
-; AVX512-NEXT:    movl %r13d, %ecx
-; AVX512-NEXT:    addb %al, %cl
-; AVX512-NEXT:    setns %cl
-; AVX512-NEXT:    addb %al, %r13b
-; AVX512-NEXT:    kshiftrw $8, %k0, %k2
-; AVX512-NEXT:    kmovd %k2, %r9d
-; AVX512-NEXT:    kshiftrw $8, %k1, %k2
-; AVX512-NEXT:    kmovd %k2, %eax
-; AVX512-NEXT:    jno .LBB16_16
-; AVX512-NEXT:  # %bb.15:
-; AVX512-NEXT:    addb $127, %cl
-; AVX512-NEXT:    movl %ecx, %r13d
-; AVX512-NEXT:  .LBB16_16:
-; AVX512-NEXT:    shlb $7, %al
-; AVX512-NEXT:    shlb $7, %r9b
-; AVX512-NEXT:    movl %r9d, %ecx
-; AVX512-NEXT:    addb %al, %cl
-; AVX512-NEXT:    setns %cl
-; AVX512-NEXT:    addb %al, %r9b
-; AVX512-NEXT:    kshiftrw $9, %k0, %k2
-; AVX512-NEXT:    kmovd %k2, %r10d
-; AVX512-NEXT:    kshiftrw $9, %k1, %k2
-; AVX512-NEXT:    kmovd %k2, %eax
-; AVX512-NEXT:    jno .LBB16_18
-; AVX512-NEXT:  # %bb.17:
-; AVX512-NEXT:    addb $127, %cl
-; AVX512-NEXT:    movl %ecx, %r9d
-; AVX512-NEXT:  .LBB16_18:
-; AVX512-NEXT:    shlb $7, %al
-; AVX512-NEXT:    shlb $7, %r10b
-; AVX512-NEXT:    movl %r10d, %ecx
-; AVX512-NEXT:    addb %al, %cl
-; AVX512-NEXT:    setns %cl
-; AVX512-NEXT:    addb %al, %r10b
-; AVX512-NEXT:    kshiftrw $10, %k0, %k2
-; AVX512-NEXT:    kmovd %k2, %ebp
-; AVX512-NEXT:    kshiftrw $10, %k1, %k2
-; AVX512-NEXT:    kmovd %k2, %eax
-; AVX512-NEXT:    movl %edi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX512-NEXT:    jno .LBB16_20
-; AVX512-NEXT:  # %bb.19:
-; AVX512-NEXT:    addb $127, %cl
-; AVX512-NEXT:    movl %ecx, %r10d
-; AVX512-NEXT:  .LBB16_20:
-; AVX512-NEXT:    shlb $7, %al
-; AVX512-NEXT:    shlb $7, %bpl
-; AVX512-NEXT:    movl %ebp, %ecx
-; AVX512-NEXT:    addb %al, %cl
-; AVX512-NEXT:    setns %cl
-; AVX512-NEXT:    addb %al, %bpl
-; AVX512-NEXT:    kshiftrw $11, %k0, %k2
-; AVX512-NEXT:    kmovd %k2, %edi
-; AVX512-NEXT:    kshiftrw $11, %k1, %k2
-; AVX512-NEXT:    kmovd %k2, %eax
-; AVX512-NEXT:    movl %edx, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX512-NEXT:    jno .LBB16_22
-; AVX512-NEXT:  # %bb.21:
-; AVX512-NEXT:    addb $127, %cl
-; AVX512-NEXT:    movl %ecx, %ebp
-; AVX512-NEXT:  .LBB16_22:
-; AVX512-NEXT:    shlb $7, %al
-; AVX512-NEXT:    shlb $7, %dil
-; AVX512-NEXT:    movl %edi, %ecx
-; AVX512-NEXT:    addb %al, %cl
-; AVX512-NEXT:    setns %dl
-; AVX512-NEXT:    addb %al, %dil
-; AVX512-NEXT:    kshiftrw $12, %k0, %k2
-; AVX512-NEXT:    kmovd %k2, %eax
-; AVX512-NEXT:    kshiftrw $12, %k1, %k2
-; AVX512-NEXT:    kmovd %k2, %ecx
-; AVX512-NEXT:    jno .LBB16_24
-; AVX512-NEXT:  # %bb.23:
-; AVX512-NEXT:    addb $127, %dl
-; AVX512-NEXT:    movl %edx, %edi
-; AVX512-NEXT:  .LBB16_24:
-; AVX512-NEXT:    shlb $7, %cl
-; AVX512-NEXT:    shlb $7, %al
-; AVX512-NEXT:    movl %eax, %edx
-; AVX512-NEXT:    addb %cl, %dl
-; AVX512-NEXT:    setns %bl
-; AVX512-NEXT:    addb %cl, %al
-; AVX512-NEXT:    kshiftrw $13, %k0, %k2
-; AVX512-NEXT:    kmovd %k2, %ecx
-; AVX512-NEXT:    kshiftrw $13, %k1, %k2
-; AVX512-NEXT:    kmovd %k2, %edx
-; AVX512-NEXT:    movl %esi, {{[-0-9]+}}(%r{{[sb]}}p) # 4-byte Spill
-; AVX512-NEXT:    jno .LBB16_26
-; AVX512-NEXT:  # %bb.25:
-; AVX512-NEXT:    addb $127, %bl
-; AVX512-NEXT:    movl %ebx, %eax
-; AVX512-NEXT:  .LBB16_26:
-; AVX512-NEXT:    shlb $7, %dl
-; AVX512-NEXT:    shlb $7, %cl
-; AVX512-NEXT:    movl %ecx, %ebx
-; AVX512-NEXT:    addb %dl, %bl
-; AVX512-NEXT:    setns %bl
-; AVX512-NEXT:    addb %dl, %cl
-; AVX512-NEXT:    kshiftrw $14, %k0, %k2
-; AVX512-NEXT:    kmovd %k2, %edx
-; AVX512-NEXT:    kshiftrw $14, %k1, %k2
-; AVX512-NEXT:    kmovd %k2, %esi
-; AVX512-NEXT:    jno .LBB16_28
-; AVX512-NEXT:  # %bb.27:
-; AVX512-NEXT:    addb $127, %bl
-; AVX512-NEXT:    movl %ebx, %ecx
-; AVX512-NEXT:  .LBB16_28:
-; AVX512-NEXT:    shlb $7, %sil
-; AVX512-NEXT:    shlb $7, %dl
-; AVX512-NEXT:    movl %edx, %ebx
-; AVX512-NEXT:    addb %sil, %bl
-; AVX512-NEXT:    setns %bl
-; AVX512-NEXT:    addb %sil, %dl
-; AVX512-NEXT:    kshiftrw $15, %k0, %k0
-; AVX512-NEXT:    kmovd %k0, %r8d
-; AVX512-NEXT:    kshiftrw $15, %k1, %k0
-; AVX512-NEXT:    kmovd %k0, %esi
-; AVX512-NEXT:    jno .LBB16_30
-; AVX512-NEXT:  # %bb.29:
-; AVX512-NEXT:    addb $127, %bl
-; AVX512-NEXT:    movl %ebx, %edx
-; AVX512-NEXT:  .LBB16_30:
-; AVX512-NEXT:    shlb $7, %sil
-; AVX512-NEXT:    shlb $7, %r8b
-; AVX512-NEXT:    movl %r8d, %ebx
-; AVX512-NEXT:    addb %sil, %bl
-; AVX512-NEXT:    setns %bl
-; AVX512-NEXT:    addb %sil, %r8b
-; AVX512-NEXT:    jno .LBB16_32
-; AVX512-NEXT:  # %bb.31:
-; AVX512-NEXT:    addb $127, %bl
-; AVX512-NEXT:    movl %ebx, %r8d
-; AVX512-NEXT:  .LBB16_32:
-; AVX512-NEXT:    movl {{[-0-9]+}}(%r{{[sb]}}p), %esi # 4-byte Reload
-; AVX512-NEXT:    sarb $7, %sil
-; AVX512-NEXT:    kmovd %esi, %k1
-; AVX512-NEXT:    movl {{[-0-9]+}}(%r{{[sb]}}p), %esi # 4-byte Reload
-; AVX512-NEXT:    sarb $7, %sil
-; AVX512-NEXT:    kmovd %esi, %k0
-; AVX512-NEXT:    movl {{[-0-9]+}}(%r{{[sb]}}p), %esi # 4-byte Reload
-; AVX512-NEXT:    sarb $7, %sil
-; AVX512-NEXT:    kmovd %esi, %k2
-; AVX512-NEXT:    sarb $7, %r11b
-; AVX512-NEXT:    kmovd %r11d, %k3
-; AVX512-NEXT:    sarb $7, %r14b
-; AVX512-NEXT:    kmovd %r14d, %k4
-; AVX512-NEXT:    sarb $7, %r15b
-; AVX512-NEXT:    kmovd %r15d, %k5
-; AVX512-NEXT:    sarb $7, %r12b
-; AVX512-NEXT:    kmovd %r12d, %k6
-; AVX512-NEXT:    kshiftrw $1, %k0, %k7
-; AVX512-NEXT:    kxorw %k1, %k7, %k7
-; AVX512-NEXT:    sarb $7, %r13b
-; AVX512-NEXT:    kmovd %r13d, %k1
-; AVX512-NEXT:    kshiftlw $15, %k7, %k7
-; AVX512-NEXT:    kshiftrw $14, %k7, %k7
-; AVX512-NEXT:    kxorw %k7, %k0, %k0
-; AVX512-NEXT:    kshiftrw $2, %k0, %k7
-; AVX512-NEXT:    kxorw %k2, %k7, %k7
-; AVX512-NEXT:    sarb $7, %r9b
-; AVX512-NEXT:    kmovd %r9d, %k2
-; AVX512-NEXT:    kshiftlw $15, %k7, %k7
-; AVX512-NEXT:    kshiftrw $13, %k7, %k7
-; AVX512-NEXT:    kxorw %k7, %k0, %k0
-; AVX512-NEXT:    kshiftrw $3, %k0, %k7
-; AVX512-NEXT:    kxorw %k3, %k7, %k7
-; AVX512-NEXT:    sarb $7, %r10b
-; AVX512-NEXT:    kmovd %r10d, %k3
-; AVX512-NEXT:    kshiftlw $15, %k7, %k7
-; AVX512-NEXT:    kshiftrw $12, %k7, %k7
-; AVX512-NEXT:    kxorw %k7, %k0, %k7
-; AVX512-NEXT:    kshiftrw $4, %k7, %k0
-; AVX512-NEXT:    kxorw %k4, %k0, %k4
-; AVX512-NEXT:    sarb $7, %bpl
-; AVX512-NEXT:    kmovd %ebp, %k0
-; AVX512-NEXT:    kshiftlw $15, %k4, %k4
-; AVX512-NEXT:    kshiftrw $11, %k4, %k4
-; AVX512-NEXT:    kxorw %k4, %k7, %k7
-; AVX512-NEXT:    kshiftrw $5, %k7, %k4
-; AVX512-NEXT:    kxorw %k5, %k4, %k5
-; AVX512-NEXT:    sarb $7, %dil
-; AVX512-NEXT:    kmovd %edi, %k4
-; AVX512-NEXT:    kshiftlw $15, %k5, %k5
-; AVX512-NEXT:    kshiftrw $10, %k5, %k5
-; AVX512-NEXT:    kxorw %k5, %k7, %k7
-; AVX512-NEXT:    kshiftrw $6, %k7, %k5
-; AVX512-NEXT:    kxorw %k6, %k5, %k6
-; AVX512-NEXT:    sarb $7, %al
-; AVX512-NEXT:    kmovd %eax, %k5
-; AVX512-NEXT:    kshiftlw $15, %k6, %k6
-; AVX512-NEXT:    kshiftrw $9, %k6, %k6
-; AVX512-NEXT:    kxorw %k6, %k7, %k6
-; AVX512-NEXT:    kshiftrw $7, %k6, %k7
-; AVX512-NEXT:    kxorw %k1, %k7, %k7
-; AVX512-NEXT:    sarb $7, %cl
-; AVX512-NEXT:    kmovd %ecx, %k1
-; AVX512-NEXT:    kshiftlw $15, %k7, %k7
-; AVX512-NEXT:    kshiftrw $8, %k7, %k7
-; AVX512-NEXT:    kxorw %k7, %k6, %k6
-; AVX512-NEXT:    kshiftrw $8, %k6, %k7
-; AVX512-NEXT:    kxorw %k2, %k7, %k7
-; AVX512-NEXT:    sarb $7, %dl
-; AVX512-NEXT:    kmovd %edx, %k2
-; AVX512-NEXT:    kshiftlw $15, %k7, %k7
-; AVX512-NEXT:    kshiftrw $7, %k7, %k7
-; AVX512-NEXT:    kxorw %k7, %k6, %k6
-; AVX512-NEXT:    kshiftrw $9, %k6, %k7
-; AVX512-NEXT:    kxorw %k3, %k7, %k3
-; AVX512-NEXT:    sarb $7, %r8b
-; AVX512-NEXT:    kmovd %r8d, %k7
-; AVX512-NEXT:    kshiftlw $15, %k3, %k3
-; AVX512-NEXT:    kshiftrw $6, %k3, %k3
-; AVX512-NEXT:    kxorw %k3, %k6, %k3
-; AVX512-NEXT:    kshiftrw $10, %k3, %k6
-; AVX512-NEXT:    kxorw %k0, %k6, %k0
-; AVX512-NEXT:    kshiftlw $15, %k0, %k0
-; AVX512-NEXT:    kshiftrw $5, %k0, %k0
-; AVX512-NEXT:    kxorw %k0, %k3, %k0
-; AVX512-NEXT:    kshiftrw $11, %k0, %k3
-; AVX512-NEXT:    kxorw %k4, %k3, %k3
-; AVX512-NEXT:    kshiftlw $15, %k3, %k3
-; AVX512-NEXT:    kshiftrw $4, %k3, %k3
-; AVX512-NEXT:    kxorw %k3, %k0, %k0
-; AVX512-NEXT:    kshiftrw $12, %k0, %k3
-; AVX512-NEXT:    kxorw %k5, %k3, %k3
-; AVX512-NEXT:    kshiftlw $15, %k3, %k3
-; AVX512-NEXT:    kshiftrw $3, %k3, %k3
-; AVX512-NEXT:    kxorw %k3, %k0, %k0
-; AVX512-NEXT:    kshiftrw $13, %k0, %k3
-; AVX512-NEXT:    kxorw %k1, %k3, %k1
-; AVX512-NEXT:    kshiftlw $15, %k1, %k1
-; AVX512-NEXT:    kshiftrw $2, %k1, %k1
-; AVX512-NEXT:    kxorw %k1, %k0, %k0
-; AVX512-NEXT:    kshiftrw $14, %k0, %k1
-; AVX512-NEXT:    kxorw %k2, %k1, %k1
-; AVX512-NEXT:    kshiftlw $15, %k1, %k1
-; AVX512-NEXT:    kshiftrw $1, %k1, %k1
-; AVX512-NEXT:    kxorw %k1, %k0, %k0
-; AVX512-NEXT:    kshiftlw $1, %k0, %k0
-; AVX512-NEXT:    kshiftrw $1, %k0, %k0
-; AVX512-NEXT:    kshiftlw $15, %k7, %k1
-; AVX512-NEXT:    korw %k1, %k0, %k0
+; AVX512-NEXT:    korw %k0, %k1, %k0
 ; AVX512-NEXT:    vpmovm2b %k0, %xmm0
-; AVX512-NEXT:    popq %rbx
-; AVX512-NEXT:    popq %r12
-; AVX512-NEXT:    popq %r13
-; AVX512-NEXT:    popq %r14
-; AVX512-NEXT:    popq %r15
-; AVX512-NEXT:    popq %rbp
 ; AVX512-NEXT:    retq
   %z = call <16 x i1> @llvm.sadd.sat.v16i1(<16 x i1> %x, <16 x i1> %y)
   ret <16 x i1> %z




More information about the llvm-commits mailing list