[llvm] r318993 - [x86][icelake]GFNI
Coby Tayree via llvm-commits
llvm-commits at lists.llvm.org
Sun Nov 26 01:36:41 PST 2017
Author: coby
Date: Sun Nov 26 01:36:41 2017
New Revision: 318993
URL: http://llvm.org/viewvc/llvm-project?rev=318993&view=rev
Log:
[x86][icelake]GFNI
galois field arithmetic (GF(2^8)) insns:
gf2p8affineinvqb
gf2p8affineqb
gf2p8mulb
Differential Revision: https://reviews.llvm.org/D40373
Added:
llvm/trunk/test/CodeGen/X86/avx-gfni-intrinsics.ll
llvm/trunk/test/CodeGen/X86/avx512-gfni-intrinsics.ll
llvm/trunk/test/CodeGen/X86/gfni-intrinsics.ll
llvm/trunk/test/MC/X86/avx512gfni-encoding.s
llvm/trunk/test/MC/X86/avx512vl_gfni-encoding.s
llvm/trunk/test/MC/X86/gfni-encoding.s
Modified:
llvm/trunk/include/llvm/IR/IntrinsicsX86.td
llvm/trunk/lib/Support/Host.cpp
llvm/trunk/lib/Target/X86/X86.td
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/X86InstrInfo.td
llvm/trunk/lib/Target/X86/X86InstrSSE.td
llvm/trunk/lib/Target/X86/X86IntrinsicsInfo.h
llvm/trunk/lib/Target/X86/X86Subtarget.cpp
llvm/trunk/lib/Target/X86/X86Subtarget.h
Modified: llvm/trunk/include/llvm/IR/IntrinsicsX86.td
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/include/llvm/IR/IntrinsicsX86.td?rev=318993&r1=318992&r2=318993&view=diff
==============================================================================
--- llvm/trunk/include/llvm/IR/IntrinsicsX86.td (original)
+++ llvm/trunk/include/llvm/IR/IntrinsicsX86.td Sun Nov 26 01:36:41 2017
@@ -1341,6 +1341,57 @@ let TargetPrefix = "x86" in { // All in
}
+// GFNI Instructions
+let TargetPrefix = "x86" in { // All intrinsics start with "llvm.x86.".
+ def int_x86_vgf2p8affineinvqb_128 :
+ GCCBuiltin<"__builtin_ia32_vgf2p8affineinvqb_v16qi">,
+ Intrinsic<[llvm_v16i8_ty],
+ [llvm_v16i8_ty, llvm_v16i8_ty, llvm_i8_ty],
+ [IntrNoMem]>;
+ def int_x86_vgf2p8affineinvqb_256 :
+ GCCBuiltin<"__builtin_ia32_vgf2p8affineinvqb_v32qi">,
+ Intrinsic<[llvm_v32i8_ty],
+ [llvm_v32i8_ty, llvm_v32i8_ty, llvm_i8_ty],
+ [IntrNoMem]>;
+ def int_x86_vgf2p8affineinvqb_512 :
+ GCCBuiltin<"__builtin_ia32_vgf2p8affineinvqb_v64qi">,
+ Intrinsic<[llvm_v64i8_ty],
+ [llvm_v64i8_ty, llvm_v64i8_ty, llvm_i8_ty],
+ [IntrNoMem]>;
+
+ def int_x86_vgf2p8affineqb_128 :
+ GCCBuiltin<"__builtin_ia32_vgf2p8affineqb_v16qi">,
+ Intrinsic<[llvm_v16i8_ty],
+ [llvm_v16i8_ty, llvm_v16i8_ty, llvm_i8_ty],
+ [IntrNoMem]>;
+ def int_x86_vgf2p8affineqb_256 :
+ GCCBuiltin<"__builtin_ia32_vgf2p8affineqb_v32qi">,
+ Intrinsic<[llvm_v32i8_ty],
+ [llvm_v32i8_ty, llvm_v32i8_ty, llvm_i8_ty],
+ [IntrNoMem]>;
+ def int_x86_vgf2p8affineqb_512 :
+ GCCBuiltin<"__builtin_ia32_vgf2p8affineqb_v64qi">,
+ Intrinsic<[llvm_v64i8_ty],
+ [llvm_v64i8_ty, llvm_v64i8_ty, llvm_i8_ty],
+ [IntrNoMem]>;
+
+ def int_x86_vgf2p8mulb_128 :
+ GCCBuiltin<"__builtin_ia32_vgf2p8mulb_v16qi">,
+ Intrinsic<[llvm_v16i8_ty],
+ [llvm_v16i8_ty, llvm_v16i8_ty],
+ [IntrNoMem]>;
+ def int_x86_vgf2p8mulb_256 :
+ GCCBuiltin<"__builtin_ia32_vgf2p8mulb_v32qi">,
+ Intrinsic<[llvm_v32i8_ty],
+ [llvm_v32i8_ty, llvm_v32i8_ty],
+ [IntrNoMem]>;
+ def int_x86_vgf2p8mulb_512 :
+ GCCBuiltin<"__builtin_ia32_vgf2p8mulb_v64qi">,
+ Intrinsic<[llvm_v64i8_ty],
+ [llvm_v64i8_ty, llvm_v64i8_ty],
+ [IntrNoMem]>;
+}
+
// Vector blend
let TargetPrefix = "x86" in { // All intrinsics start with "llvm.x86.".
def int_x86_avx_blendv_pd_256 : GCCBuiltin<"__builtin_ia32_blendvpd256">,
Modified: llvm/trunk/lib/Support/Host.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Support/Host.cpp?rev=318993&r1=318992&r2=318993&view=diff
==============================================================================
--- llvm/trunk/lib/Support/Host.cpp (original)
+++ llvm/trunk/lib/Support/Host.cpp Sun Nov 26 01:36:41 2017
@@ -1217,6 +1217,7 @@ bool sys::getHostCPUFeatures(StringMap<b
Features["avx512vbmi"] = HasLeaf7 && ((ECX >> 1) & 1) && HasAVX512Save;
Features["pku"] = HasLeaf7 && ((ECX >> 4) & 1);
Features["avx512vbmi2"] = HasLeaf7 && ((ECX >> 6) & 1) && HasAVX512Save;
+ Features["gfni"] = HasLeaf7 && ((ECX >> 8) & 1);
Features["vaes"] = HasLeaf7 && ((ECX >> 9) & 1) && HasAVXSave;
Features["vpclmulqdq"] = HasLeaf7 && ((ECX >> 10) & 1) && HasAVXSave;
Features["avx512vnni"] = HasLeaf7 && ((ECX >> 11) & 1) && HasAVX512Save;
Modified: llvm/trunk/lib/Target/X86/X86.td
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/X86/X86.td?rev=318993&r1=318992&r2=318993&view=diff
==============================================================================
--- llvm/trunk/lib/Target/X86/X86.td (original)
+++ llvm/trunk/lib/Target/X86/X86.td Sun Nov 26 01:36:41 2017
@@ -169,6 +169,9 @@ def FeatureBITALG : SubtargetFeature<"a
def FeaturePCLMUL : SubtargetFeature<"pclmul", "HasPCLMUL", "true",
"Enable packed carry-less multiplication instructions",
[FeatureSSE2]>;
+def FeatureGFNI : SubtargetFeature<"gfni", "HasGFNI", "true",
+ "Enable Galois Field Arithmetic Instructions",
+ [FeatureSSE2]>;
def FeatureVPCLMULQDQ : SubtargetFeature<"vpclmulqdq", "HasVPCLMULQDQ", "true",
"Enable vpclmulqdq instructions",
[FeatureAVX, FeaturePCLMUL]>;
@@ -698,8 +701,8 @@ def ICLFeatures : ProcessorFeatures<CNLF
FeatureVBMI2,
FeatureVNNI,
FeatureVPCLMULQDQ,
- FeatureVPOPCNTDQ
- // TODO: Add GFNI when it is implemented.
+ FeatureVPOPCNTDQ,
+ FeatureGFNI
]>;
class IcelakeProc<string Name> : ProcModel<Name, SkylakeServerModel,
Modified: llvm/trunk/lib/Target/X86/X86ISelLowering.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/X86/X86ISelLowering.cpp?rev=318993&r1=318992&r2=318993&view=diff
==============================================================================
--- llvm/trunk/lib/Target/X86/X86ISelLowering.cpp (original)
+++ llvm/trunk/lib/Target/X86/X86ISelLowering.cpp Sun Nov 26 01:36:41 2017
@@ -25254,6 +25254,9 @@ const char *X86TargetLowering::getTarget
case X86ISD::VPDPWSSD: return "X86ISD::VPDPWSSD";
case X86ISD::VPDPWSSDS: return "X86ISD::VPDPWSSDS";
case X86ISD::VPSHUFBITQMB: return "X86ISD::VPSHUFBITQMB";
+ case X86ISD::GF2P8MULB: return "X86ISD::GF2P8MULB";
+ case X86ISD::GF2P8AFFINEQB: return "X86ISD::GF2P8AFFINEQB";
+ case X86ISD::GF2P8AFFINEINVQB: return "X86ISD::GF2P8AFFINEINVQB";
}
return nullptr;
}
Modified: llvm/trunk/lib/Target/X86/X86ISelLowering.h
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/X86/X86ISelLowering.h?rev=318993&r1=318992&r2=318993&view=diff
==============================================================================
--- llvm/trunk/lib/Target/X86/X86ISelLowering.h (original)
+++ llvm/trunk/lib/Target/X86/X86ISelLowering.h Sun Nov 26 01:36:41 2017
@@ -587,6 +587,9 @@ namespace llvm {
// Conversions between float and half-float.
CVTPS2PH, CVTPH2PS, CVTPH2PS_RND,
+ // Galois Field Arithmetic Instructions
+ GF2P8AFFINEINVQB, GF2P8AFFINEQB, GF2P8MULB,
+
// LWP insert record.
LWPINS,
Modified: llvm/trunk/lib/Target/X86/X86InstrAVX512.td
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/X86/X86InstrAVX512.td?rev=318993&r1=318992&r2=318993&view=diff
==============================================================================
--- llvm/trunk/lib/Target/X86/X86InstrAVX512.td (original)
+++ llvm/trunk/lib/Target/X86/X86InstrAVX512.td Sun Nov 26 01:36:41 2017
@@ -10242,3 +10242,55 @@ multiclass VPSHUFBITQMB_common<AVX512VLV
defm VPSHUFBITQMB : VPSHUFBITQMB_common<avx512vl_i8_info>;
+//===----------------------------------------------------------------------===//
+// GFNI
+//===----------------------------------------------------------------------===//
+
+multiclass GF2P8MULB_avx512_common<bits<8> Op, string OpStr, SDNode OpNode> {
+ let Predicates = [HasGFNI, HasAVX512, HasBWI] in
+ defm Z : avx512_binop_rm<Op, OpStr, OpNode, v64i8_info,
+ SSE_INTALU_ITINS_P, 1>, EVEX_V512;
+ let Predicates = [HasGFNI, HasVLX, HasBWI] in {
+ defm Z256 : avx512_binop_rm<Op, OpStr, OpNode, v32i8x_info,
+ SSE_INTALU_ITINS_P, 1>, EVEX_V256;
+ defm Z128 : avx512_binop_rm<Op, OpStr, OpNode, v16i8x_info,
+ SSE_INTALU_ITINS_P, 1>, EVEX_V128;
+ }
+}
+
+defm GF2P8MULB : GF2P8MULB_avx512_common<0xCF, "vgf2p8mulb", X86GF2P8mulb>,
+ EVEX_CD8<8, CD8VF>, T8PD;
+
+multiclass GF2P8AFFINE_avx512_rmb_imm<bits<8> Op, string OpStr, SDNode OpNode,
+ X86VectorVTInfo VTI,
+ X86VectorVTInfo BcstVTI>
+ : avx512_3Op_rm_imm8<Op, OpStr, OpNode, VTI, VTI> {
+ let ExeDomain = VTI.ExeDomain in
+ defm rmbi : AVX512_maskable<Op, MRMSrcMem, VTI, (outs VTI.RC:$dst),
+ (ins VTI.RC:$src1, VTI.ScalarMemOp:$src2, u8imm:$src3),
+ OpStr, "$src3, ${src2}"##BcstVTI.BroadcastStr##", $src1",
+ "$src1, ${src2}"##BcstVTI.BroadcastStr##", $src3",
+ (OpNode (VTI.VT VTI.RC:$src1),
+ (bitconvert (BcstVTI.VT (X86VBroadcast (loadi64 addr:$src2)))),
+ (i8 imm:$src3))>, EVEX_B;
+}
+
+multiclass GF2P8AFFINE_avx512_common<bits<8> Op, string OpStr, SDNode OpNode> {
+ let Predicates = [HasGFNI, HasAVX512, HasBWI] in
+ defm Z : GF2P8AFFINE_avx512_rmb_imm<Op, OpStr, OpNode, v64i8_info,
+ v8i64_info>, EVEX_V512;
+ let Predicates = [HasGFNI, HasVLX, HasBWI] in {
+ defm Z256 : GF2P8AFFINE_avx512_rmb_imm<Op, OpStr, OpNode, v32i8x_info,
+ v4i64x_info>, EVEX_V256;
+ defm Z128 : GF2P8AFFINE_avx512_rmb_imm<Op, OpStr, OpNode, v16i8x_info,
+ v2i64x_info>, EVEX_V128;
+ }
+}
+
+defm GF2P8AFFINEINVQB : GF2P8AFFINE_avx512_common<0xCF, "vgf2p8affineinvqb",
+ X86GF2P8affineinvqb>,
+ EVEX_4V, EVEX_CD8<8, CD8VF>, VEX_W, AVX512AIi8Base;
+defm GF2P8AFFINEQB : GF2P8AFFINE_avx512_common<0xCE, "vgf2p8affineqb",
+ X86GF2P8affineqb>,
+ EVEX_4V, EVEX_CD8<8, CD8VF>, VEX_W, AVX512AIi8Base;
+
Modified: llvm/trunk/lib/Target/X86/X86InstrFragmentsSIMD.td
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/X86/X86InstrFragmentsSIMD.td?rev=318993&r1=318992&r2=318993&view=diff
==============================================================================
--- llvm/trunk/lib/Target/X86/X86InstrFragmentsSIMD.td (original)
+++ llvm/trunk/lib/Target/X86/X86InstrFragmentsSIMD.td Sun Nov 26 01:36:41 2017
@@ -672,6 +672,11 @@ def X86vfproundRnd: SDNode<"X86ISD::VFPR
def X86cvt2mask : SDNode<"X86ISD::CVT2MASK", SDTIntTruncOp>;
+// galois field arithmetic
+def X86GF2P8affineinvqb : SDNode<"X86ISD::GF2P8AFFINEINVQB", SDTBlend>;
+def X86GF2P8affineqb : SDNode<"X86ISD::GF2P8AFFINEQB", SDTBlend>;
+def X86GF2P8mulb : SDNode<"X86ISD::GF2P8MULB", SDTIntBinOp>;
+
//===----------------------------------------------------------------------===//
// SSE Complex Patterns
//===----------------------------------------------------------------------===//
Modified: llvm/trunk/lib/Target/X86/X86InstrInfo.td
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/X86/X86InstrInfo.td?rev=318993&r1=318992&r2=318993&view=diff
==============================================================================
--- llvm/trunk/lib/Target/X86/X86InstrInfo.td (original)
+++ llvm/trunk/lib/Target/X86/X86InstrInfo.td Sun Nov 26 01:36:41 2017
@@ -848,6 +848,7 @@ def HasPCLMUL : Predicate<"Subtarget-
def NoVLX_Or_NoVPCLMULQDQ :
Predicate<"!Subtarget->hasVLX() || !Subtarget->hasVPCLMULQDQ()">;
def HasVPCLMULQDQ : Predicate<"Subtarget->hasVPCLMULQDQ()">;
+def HasGFNI : Predicate<"Subtarget->hasGFNI()">;
def HasFMA : Predicate<"Subtarget->hasFMA()">;
def HasFMA4 : Predicate<"Subtarget->hasFMA4()">;
def NoFMA4 : Predicate<"!Subtarget->hasFMA4()">;
Modified: llvm/trunk/lib/Target/X86/X86InstrSSE.td
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/X86/X86InstrSSE.td?rev=318993&r1=318992&r2=318993&view=diff
==============================================================================
--- llvm/trunk/lib/Target/X86/X86InstrSSE.td (original)
+++ llvm/trunk/lib/Target/X86/X86InstrSSE.td Sun Nov 26 01:36:41 2017
@@ -8466,3 +8466,82 @@ def : Pat<(xor FR128:$src1, FR128:$src2)
(COPY_TO_REGCLASS
(XORPSrr (COPY_TO_REGCLASS FR128:$src1, VR128),
(COPY_TO_REGCLASS FR128:$src2, VR128)), FR128)>;
+
+//===----------------------------------------------------------------------===//
+// GFNI instructions
+//===----------------------------------------------------------------------===//
+
+multiclass GF2P8MULB_rm<string OpcodeStr, ValueType OpVT,
+ RegisterClass RC, PatFrag MemOpFrag,
+ X86MemOperand X86MemOp, bit Is2Addr = 0> {
+ let ExeDomain = SSEPackedInt,
+ AsmString = !if(Is2Addr,
+ OpcodeStr##"\t{$src2, $dst|$dst, $src2}",
+ OpcodeStr##"\t{$src2, $src1, $dst|$dst, $src1, $src2}") in {
+ let isCommutable = 1 in
+ def rr : PDI<0xCF, MRMSrcReg, (outs RC:$dst), (ins RC:$src1, RC:$src2), "",
+ [(set RC:$dst, (OpVT (X86GF2P8mulb RC:$src1, RC:$src2)))],
+ SSE_INTALU_ITINS_P.rr>,
+ Sched<[SSE_INTALU_ITINS_P.Sched]>, T8PD;
+
+ def rm : PDI<0xCF, MRMSrcMem, (outs RC:$dst), (ins RC:$src1, X86MemOp:$src2), "",
+ [(set RC:$dst, (OpVT (X86GF2P8mulb RC:$src1,
+ (bitconvert (MemOpFrag addr:$src2)))))],
+ SSE_INTALU_ITINS_P.rm>,
+ Sched<[SSE_INTALU_ITINS_P.Sched.Folded, ReadAfterLd]>, T8PD;
+ }
+}
+
+multiclass GF2P8AFFINE_rmi<bits<8> Op, string OpStr, ValueType OpVT,
+ SDNode OpNode, RegisterClass RC, PatFrag MemOpFrag,
+ X86MemOperand X86MemOp, bit Is2Addr = 0> {
+ let AsmString = !if(Is2Addr,
+ OpStr##"\t{$src3, $src2, $dst|$dst, $src2, $src3}",
+ OpStr##"\t{$src3, $src2, $src1, $dst|$dst, $src1, $src2, $src3}") in {
+ def rri : Ii8<Op, MRMSrcReg, (outs RC:$dst),
+ (ins RC:$src1, RC:$src2, u8imm:$src3), "",
+ [(set RC:$dst, (OpVT (OpNode RC:$src1, RC:$src2, imm:$src3)))],
+ SSE_INTALU_ITINS_P.rr, SSEPackedInt>,
+ Sched<[WriteVecALU]>;
+ def rmi : Ii8<Op, MRMSrcMem, (outs RC:$dst),
+ (ins RC:$src1, X86MemOp:$src2, u8imm:$src3), "",
+ [(set RC:$dst, (OpVT (OpNode RC:$src1,
+ (bitconvert (MemOpFrag addr:$src2)),
+ imm:$src3)))],
+ SSE_INTALU_ITINS_P.rm, SSEPackedInt>,
+ Sched<[WriteVecALU.Folded, ReadAfterLd]>;
+ }
+}
+
+multiclass GF2P8AFFINE_common<bits<8> Op, string OpStr, SDNode OpNode> {
+ let Constraints = "$src1 = $dst",
+ Predicates = [HasGFNI, UseSSE2] in
+ defm NAME : GF2P8AFFINE_rmi<Op, OpStr, v16i8, OpNode,
+ VR128, loadv2i64, i128mem, 1>;
+ let Predicates = [HasGFNI, HasAVX, NoVLX_Or_NoBWI] in {
+ defm V##NAME : GF2P8AFFINE_rmi<Op, "v"##OpStr, v16i8, OpNode, VR128,
+ loadv2i64, i128mem>, VEX_4V, VEX_W;
+ defm V##NAME##Y : GF2P8AFFINE_rmi<Op, "v"##OpStr, v32i8, OpNode, VR256,
+ loadv4i64, i256mem>, VEX_4V, VEX_L, VEX_W;
+ }
+}
+
+// GF2P8MULB
+let Constraints = "$src1 = $dst",
+ Predicates = [HasGFNI, UseSSE2] in
+defm GF2P8MULB : GF2P8MULB_rm<"gf2p8mulb", v16i8, VR128, memopv2i64,
+ i128mem, 1>;
+let Predicates = [HasGFNI, HasAVX, NoVLX_Or_NoBWI] in {
+ defm VGF2P8MULB : GF2P8MULB_rm<"vgf2p8mulb", v16i8, VR128, loadv2i64,
+ i128mem>, VEX_4V;
+ defm VGF2P8MULBY : GF2P8MULB_rm<"vgf2p8mulb", v32i8, VR256, loadv4i64,
+ i256mem>, VEX_4V, VEX_L;
+}
+// GF2P8AFFINEINVQB, GF2P8AFFINEQB
+let isCommutable = 0 in {
+ defm GF2P8AFFINEINVQB : GF2P8AFFINE_common<0xCF, "gf2p8affineinvqb",
+ X86GF2P8affineinvqb>, TAPD;
+ defm GF2P8AFFINEQB : GF2P8AFFINE_common<0xCE, "gf2p8affineqb",
+ X86GF2P8affineqb>, TAPD;
+}
+
Modified: llvm/trunk/lib/Target/X86/X86IntrinsicsInfo.h
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/X86/X86IntrinsicsInfo.h?rev=318993&r1=318992&r2=318993&view=diff
==============================================================================
--- llvm/trunk/lib/Target/X86/X86IntrinsicsInfo.h (original)
+++ llvm/trunk/lib/Target/X86/X86IntrinsicsInfo.h Sun Nov 26 01:36:41 2017
@@ -1170,7 +1170,7 @@ static const IntrinsicData IntrinsicsWi
X86_INTRINSIC_DATA(avx512_mask_vpdpwssds_256, FMA_OP_MASK, X86ISD::VPDPWSSDS, 0),
X86_INTRINSIC_DATA(avx512_mask_vpdpwssds_512, FMA_OP_MASK, X86ISD::VPDPWSSDS, 0),
- X86_INTRINSIC_DATA(avx512_mask_vpermi2var_d_128, VPERM_3OP_MASK,
+ X86_INTRINSIC_DATA(avx512_mask_vpermi2var_d_128, VPERM_3OP_MASK,
X86ISD::VPERMIV3, 0),
X86_INTRINSIC_DATA(avx512_mask_vpermi2var_d_256, VPERM_3OP_MASK,
X86ISD::VPERMIV3, 0),
@@ -1700,6 +1700,26 @@ static const IntrinsicData IntrinsicsWi
X86_INTRINSIC_DATA(vcvtph2ps_256, INTR_TYPE_1OP, X86ISD::CVTPH2PS, 0),
X86_INTRINSIC_DATA(vcvtps2ph_128, INTR_TYPE_2OP, X86ISD::CVTPS2PH, 0),
X86_INTRINSIC_DATA(vcvtps2ph_256, INTR_TYPE_2OP, X86ISD::CVTPS2PH, 0),
+
+ X86_INTRINSIC_DATA(vgf2p8affineinvqb_128, INTR_TYPE_3OP,
+ X86ISD::GF2P8AFFINEINVQB, 0),
+ X86_INTRINSIC_DATA(vgf2p8affineinvqb_256, INTR_TYPE_3OP,
+ X86ISD::GF2P8AFFINEINVQB, 0),
+ X86_INTRINSIC_DATA(vgf2p8affineinvqb_512, INTR_TYPE_3OP,
+ X86ISD::GF2P8AFFINEINVQB, 0),
+ X86_INTRINSIC_DATA(vgf2p8affineqb_128, INTR_TYPE_3OP,
+ X86ISD::GF2P8AFFINEQB, 0),
+ X86_INTRINSIC_DATA(vgf2p8affineqb_256, INTR_TYPE_3OP,
+ X86ISD::GF2P8AFFINEQB, 0),
+ X86_INTRINSIC_DATA(vgf2p8affineqb_512, INTR_TYPE_3OP,
+ X86ISD::GF2P8AFFINEQB, 0),
+ X86_INTRINSIC_DATA(vgf2p8mulb_128, INTR_TYPE_2OP,
+ X86ISD::GF2P8MULB, 0),
+ X86_INTRINSIC_DATA(vgf2p8mulb_256, INTR_TYPE_2OP,
+ X86ISD::GF2P8MULB, 0),
+ X86_INTRINSIC_DATA(vgf2p8mulb_512, INTR_TYPE_2OP,
+ X86ISD::GF2P8MULB, 0),
+
X86_INTRINSIC_DATA(xop_vpcomb, INTR_TYPE_3OP, X86ISD::VPCOM, 0),
X86_INTRINSIC_DATA(xop_vpcomd, INTR_TYPE_3OP, X86ISD::VPCOM, 0),
X86_INTRINSIC_DATA(xop_vpcomq, INTR_TYPE_3OP, X86ISD::VPCOM, 0),
Modified: llvm/trunk/lib/Target/X86/X86Subtarget.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/X86/X86Subtarget.cpp?rev=318993&r1=318992&r2=318993&view=diff
==============================================================================
--- llvm/trunk/lib/Target/X86/X86Subtarget.cpp (original)
+++ llvm/trunk/lib/Target/X86/X86Subtarget.cpp Sun Nov 26 01:36:41 2017
@@ -299,6 +299,7 @@ void X86Subtarget::initializeEnvironment
HasXSAVES = false;
HasPCLMUL = false;
HasVPCLMULQDQ = false;
+ HasGFNI = false;
HasFMA = false;
HasFMA4 = false;
HasXOP = false;
Modified: llvm/trunk/lib/Target/X86/X86Subtarget.h
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/X86/X86Subtarget.h?rev=318993&r1=318992&r2=318993&view=diff
==============================================================================
--- llvm/trunk/lib/Target/X86/X86Subtarget.h (original)
+++ llvm/trunk/lib/Target/X86/X86Subtarget.h Sun Nov 26 01:36:41 2017
@@ -128,6 +128,9 @@ protected:
bool HasPCLMUL;
bool HasVPCLMULQDQ;
+ /// Target has Galois Field Arithmetic instructions
+ bool HasGFNI;
+
/// Target has 3-operand fused multiply-add
bool HasFMA;
@@ -480,6 +483,7 @@ public:
bool hasXSAVES() const { return HasXSAVES; }
bool hasPCLMUL() const { return HasPCLMUL; }
bool hasVPCLMULQDQ() const { return HasVPCLMULQDQ; }
+ bool hasGFNI() const { return HasGFNI; }
// Prefer FMA4 to FMA - its better for commutation/memory folding and
// has equal or better performance on all supported targets.
bool hasFMA() const { return HasFMA; }
Added: llvm/trunk/test/CodeGen/X86/avx-gfni-intrinsics.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/X86/avx-gfni-intrinsics.ll?rev=318993&view=auto
==============================================================================
--- llvm/trunk/test/CodeGen/X86/avx-gfni-intrinsics.ll (added)
+++ llvm/trunk/test/CodeGen/X86/avx-gfni-intrinsics.ll Sun Nov 26 01:36:41 2017
@@ -0,0 +1,63 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc < %s -mtriple=i386-apple-darwin -mattr=+gfni,+avx -show-mc-encoding | FileCheck %s
+
+declare <16 x i8> @llvm.x86.vgf2p8affineinvqb.128(<16 x i8>, <16 x i8>, i8)
+define <16 x i8> @test_vgf2p8affineinvqb_128(<16 x i8> %src1, <16 x i8> %src2) {
+; CHECK-LABEL: test_vgf2p8affineinvqb_128:
+; CHECK: ## BB#0:
+; CHECK-NEXT: vgf2p8affineinvqb $11, %xmm1, %xmm0, %xmm0 ## encoding: [0xc4,0xe3,0xf9,0xcf,0xc1,0x0b]
+; CHECK-NEXT: retl ## encoding: [0xc3]
+ %1 = call <16 x i8> @llvm.x86.vgf2p8affineinvqb.128(<16 x i8> %src1, <16 x i8> %src2, i8 11)
+ ret <16 x i8> %1
+}
+
+declare <32 x i8> @llvm.x86.vgf2p8affineinvqb.256(<32 x i8>, <32 x i8>, i8)
+define <32 x i8> @test_vgf2p8affineinvqb_256(<32 x i8> %src1, <32 x i8> %src2) {
+; CHECK-LABEL: test_vgf2p8affineinvqb_256:
+; CHECK: ## BB#0:
+; CHECK-NEXT: vgf2p8affineinvqb $11, %ymm1, %ymm0, %ymm0 ## encoding: [0xc4,0xe3,0xfd,0xcf,0xc1,0x0b]
+; CHECK-NEXT: retl ## encoding: [0xc3]
+ %1 = call <32 x i8> @llvm.x86.vgf2p8affineinvqb.256(<32 x i8> %src1, <32 x i8> %src2, i8 11)
+ ret <32 x i8> %1
+}
+
+declare <16 x i8> @llvm.x86.vgf2p8affineqb.128(<16 x i8>, <16 x i8>, i8)
+define <16 x i8> @test_vgf2p8affineqb(<16 x i8> %src1, <16 x i8> %src2) {
+; CHECK-LABEL: test_vgf2p8affineqb:
+; CHECK: ## BB#0:
+; CHECK-NEXT: vgf2p8affineqb $11, %xmm1, %xmm0, %xmm0 ## encoding: [0xc4,0xe3,0xf9,0xce,0xc1,0x0b]
+; CHECK-NEXT: retl ## encoding: [0xc3]
+ %1 = call <16 x i8> @llvm.x86.vgf2p8affineqb.128(<16 x i8> %src1, <16 x i8> %src2, i8 11)
+ ret <16 x i8> %1
+}
+
+declare <32 x i8> @llvm.x86.vgf2p8affineqb.256(<32 x i8>, <32 x i8>, i8)
+define <32 x i8> @test_vgf2p8affineqb_256(<32 x i8> %src1, <32 x i8> %src2) {
+; CHECK-LABEL: test_vgf2p8affineqb_256:
+; CHECK: ## BB#0:
+; CHECK-NEXT: vgf2p8affineqb $11, %ymm1, %ymm0, %ymm0 ## encoding: [0xc4,0xe3,0xfd,0xce,0xc1,0x0b]
+; CHECK-NEXT: retl ## encoding: [0xc3]
+ %1 = call <32 x i8> @llvm.x86.vgf2p8affineqb.256(<32 x i8> %src1, <32 x i8> %src2, i8 11)
+ ret <32 x i8> %1
+}
+
+declare <16 x i8> @llvm.x86.vgf2p8mulb.128(<16 x i8>, <16 x i8>)
+define <16 x i8> @test_vgf2p8mulb_128(<16 x i8> %src1, <16 x i8> %src2) {
+; CHECK-LABEL: test_vgf2p8mulb_128:
+; CHECK: ## BB#0:
+; CHECK-NEXT: vgf2p8mulb %xmm1, %xmm0, %xmm0 ## encoding: [0xc4,0xe2,0x79,0xcf,0xc1]
+; CHECK-NEXT: retl ## encoding: [0xc3]
+ %1 = call <16 x i8> @llvm.x86.vgf2p8mulb.128(<16 x i8> %src1, <16 x i8> %src2)
+ ret <16 x i8> %1
+}
+
+declare <32 x i8> @llvm.x86.vgf2p8mulb.256(<32 x i8>, <32 x i8>)
+define <32 x i8> @test_vgf2p8mulb_256(<32 x i8> %src1, <32 x i8> %src2) {
+; CHECK-LABEL: test_vgf2p8mulb_256:
+; CHECK: ## BB#0:
+; CHECK-NEXT: vgf2p8mulb %ymm1, %ymm0, %ymm0 ## encoding: [0xc4,0xe2,0x7d,0xcf,0xc1]
+; CHECK-NEXT: retl ## encoding: [0xc3]
+ %1 = call <32 x i8> @llvm.x86.vgf2p8mulb.256(<32 x i8> %src1, <32 x i8> %src2)
+ ret <32 x i8> %1
+}
+
Added: llvm/trunk/test/CodeGen/X86/avx512-gfni-intrinsics.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/X86/avx512-gfni-intrinsics.ll?rev=318993&view=auto
==============================================================================
--- llvm/trunk/test/CodeGen/X86/avx512-gfni-intrinsics.ll (added)
+++ llvm/trunk/test/CodeGen/X86/avx512-gfni-intrinsics.ll Sun Nov 26 01:36:41 2017
@@ -0,0 +1,183 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc < %s -mtriple=x86_64-apple-darwin -mattr=+avx512vl,+gfni,+avx512bw --show-mc-encoding | FileCheck %s
+
+declare <16 x i8> @llvm.x86.vgf2p8affineinvqb.128(<16 x i8>, <16 x i8>, i8)
+define <16 x i8> @test_vgf2p8affineinvqb_128(<16 x i8> %src1, <16 x i8> %src2, <16 x i8> %passthru, i16 %mask) {
+; CHECK-LABEL: test_vgf2p8affineinvqb_128:
+; CHECK: ## BB#0:
+; CHECK-NEXT: kmovd %edi, %k1 ## encoding: [0xc5,0xfb,0x92,0xcf]
+; CHECK-NEXT: vgf2p8affineinvqb $3, %xmm1, %xmm0, %xmm3 ## EVEX TO VEX Compression encoding: [0xc4,0xe3,0xf9,0xcf,0xd9,0x03]
+; CHECK-NEXT: vgf2p8affineinvqb $3, %xmm1, %xmm0, %xmm4 {%k1} {z} ## encoding: [0x62,0xf3,0xfd,0x89,0xcf,0xe1,0x03]
+; CHECK-NEXT: vgf2p8affineinvqb $3, %xmm1, %xmm0, %xmm2 {%k1} ## encoding: [0x62,0xf3,0xfd,0x09,0xcf,0xd1,0x03]
+; CHECK-NEXT: vpxor %xmm3, %xmm2, %xmm0 ## EVEX TO VEX Compression encoding: [0xc5,0xe9,0xef,0xc3]
+; CHECK-NEXT: vpxor %xmm0, %xmm4, %xmm0 ## EVEX TO VEX Compression encoding: [0xc5,0xd9,0xef,0xc0]
+; CHECK-NEXT: retq ## encoding: [0xc3]
+ %1 = bitcast i16 %mask to <16 x i1>
+ %2 = call <16 x i8> @llvm.x86.vgf2p8affineinvqb.128(<16 x i8> %src1, <16 x i8> %src2, i8 3)
+ %3 = select <16 x i1> %1, <16 x i8> %2, <16 x i8> zeroinitializer
+ %4 = select <16 x i1> %1, <16 x i8> %2, <16 x i8> %passthru
+ %5 = xor <16 x i8> %3, %4
+ %6 = xor <16 x i8> %5, %2
+ ret <16 x i8> %6
+}
+
+declare <32 x i8> @llvm.x86.vgf2p8affineinvqb.256(<32 x i8>, <32 x i8>, i8)
+define <32 x i8> @test_vgf2p8affineinvqb_256(<32 x i8> %src1, <32 x i8> %src2, <32 x i8> %passthru, i32 %mask) {
+; CHECK-LABEL: test_vgf2p8affineinvqb_256:
+; CHECK: ## BB#0:
+; CHECK-NEXT: kmovd %edi, %k1 ## encoding: [0xc5,0xfb,0x92,0xcf]
+; CHECK-NEXT: vgf2p8affineinvqb $3, %ymm1, %ymm0, %ymm3 ## EVEX TO VEX Compression encoding: [0xc4,0xe3,0xfd,0xcf,0xd9,0x03]
+; CHECK-NEXT: vgf2p8affineinvqb $3, %ymm1, %ymm0, %ymm4 {%k1} {z} ## encoding: [0x62,0xf3,0xfd,0xa9,0xcf,0xe1,0x03]
+; CHECK-NEXT: vgf2p8affineinvqb $3, %ymm1, %ymm0, %ymm2 {%k1} ## encoding: [0x62,0xf3,0xfd,0x29,0xcf,0xd1,0x03]
+; CHECK-NEXT: vpxor %ymm3, %ymm2, %ymm0 ## EVEX TO VEX Compression encoding: [0xc5,0xed,0xef,0xc3]
+; CHECK-NEXT: vpxor %ymm0, %ymm4, %ymm0 ## EVEX TO VEX Compression encoding: [0xc5,0xdd,0xef,0xc0]
+; CHECK-NEXT: retq ## encoding: [0xc3]
+ %1 = bitcast i32 %mask to <32 x i1>
+ %2 = call <32 x i8> @llvm.x86.vgf2p8affineinvqb.256(<32 x i8> %src1, <32 x i8> %src2, i8 3)
+ %3 = select <32 x i1> %1, <32 x i8> %2, <32 x i8> zeroinitializer
+ %4 = select <32 x i1> %1, <32 x i8> %2, <32 x i8> %passthru
+ %5 = xor <32 x i8> %3, %4
+ %6 = xor <32 x i8> %5, %2
+ ret <32 x i8> %6
+}
+
+declare <64 x i8> @llvm.x86.vgf2p8affineinvqb.512(<64 x i8>, <64 x i8>, i8)
+define <64 x i8> @test_vgf2p8affineinvqb_512(<64 x i8> %src1, <64 x i8> %src2, <64 x i8> %passthru, i64 %mask) {
+; CHECK-LABEL: test_vgf2p8affineinvqb_512:
+; CHECK: ## BB#0:
+; CHECK-NEXT: kmovq %rdi, %k1 ## encoding: [0xc4,0xe1,0xfb,0x92,0xcf]
+; CHECK-NEXT: vgf2p8affineinvqb $3, %zmm1, %zmm0, %zmm3 ## encoding: [0x62,0xf3,0xfd,0x48,0xcf,0xd9,0x03]
+; CHECK-NEXT: vgf2p8affineinvqb $3, %zmm1, %zmm0, %zmm4 {%k1} {z} ## encoding: [0x62,0xf3,0xfd,0xc9,0xcf,0xe1,0x03]
+; CHECK-NEXT: vgf2p8affineinvqb $3, %zmm1, %zmm0, %zmm2 {%k1} ## encoding: [0x62,0xf3,0xfd,0x49,0xcf,0xd1,0x03]
+; CHECK-NEXT: vpxorq %zmm3, %zmm2, %zmm0 ## encoding: [0x62,0xf1,0xed,0x48,0xef,0xc3]
+; CHECK-NEXT: vpxorq %zmm0, %zmm4, %zmm0 ## encoding: [0x62,0xf1,0xdd,0x48,0xef,0xc0]
+; CHECK-NEXT: retq ## encoding: [0xc3]
+ %1 = bitcast i64 %mask to <64 x i1>
+ %2 = call <64 x i8> @llvm.x86.vgf2p8affineinvqb.512(<64 x i8> %src1, <64 x i8> %src2, i8 3)
+ %3 = select <64 x i1> %1, <64 x i8> %2, <64 x i8> zeroinitializer
+ %4 = select <64 x i1> %1, <64 x i8> %2, <64 x i8> %passthru
+ %5 = xor <64 x i8> %3, %4
+ %6 = xor <64 x i8> %5, %2
+ ret <64 x i8> %6
+}
+
+declare <16 x i8> @llvm.x86.vgf2p8affineqb.128(<16 x i8>, <16 x i8>, i8)
+define <16 x i8> @test_vgf2p8affineqb_128(<16 x i8> %src1, <16 x i8> %src2, <16 x i8> %passthru, i16 %mask) {
+; CHECK-LABEL: test_vgf2p8affineqb_128:
+; CHECK: ## BB#0:
+; CHECK-NEXT: kmovd %edi, %k1 ## encoding: [0xc5,0xfb,0x92,0xcf]
+; CHECK-NEXT: vgf2p8affineqb $3, %xmm1, %xmm0, %xmm3 ## EVEX TO VEX Compression encoding: [0xc4,0xe3,0xf9,0xce,0xd9,0x03]
+; CHECK-NEXT: vgf2p8affineqb $3, %xmm1, %xmm0, %xmm4 {%k1} {z} ## encoding: [0x62,0xf3,0xfd,0x89,0xce,0xe1,0x03]
+; CHECK-NEXT: vgf2p8affineqb $3, %xmm1, %xmm0, %xmm2 {%k1} ## encoding: [0x62,0xf3,0xfd,0x09,0xce,0xd1,0x03]
+; CHECK-NEXT: vpxor %xmm3, %xmm2, %xmm0 ## EVEX TO VEX Compression encoding: [0xc5,0xe9,0xef,0xc3]
+; CHECK-NEXT: vpxor %xmm0, %xmm4, %xmm0 ## EVEX TO VEX Compression encoding: [0xc5,0xd9,0xef,0xc0]
+; CHECK-NEXT: retq ## encoding: [0xc3]
+ %1 = bitcast i16 %mask to <16 x i1>
+ %2 = call <16 x i8> @llvm.x86.vgf2p8affineqb.128(<16 x i8> %src1, <16 x i8> %src2, i8 3)
+ %3 = select <16 x i1> %1, <16 x i8> %2, <16 x i8> zeroinitializer
+ %4 = select <16 x i1> %1, <16 x i8> %2, <16 x i8> %passthru
+ %5 = xor <16 x i8> %3, %4
+ %6 = xor <16 x i8> %5, %2
+ ret <16 x i8> %6
+}
+
+declare <32 x i8> @llvm.x86.vgf2p8affineqb.256(<32 x i8>, <32 x i8>, i8)
+define <32 x i8> @test_vgf2p8affineqb_256(<32 x i8> %src1, <32 x i8> %src2, <32 x i8> %passthru, i32 %mask) {
+; CHECK-LABEL: test_vgf2p8affineqb_256:
+; CHECK: ## BB#0:
+; CHECK-NEXT: kmovd %edi, %k1 ## encoding: [0xc5,0xfb,0x92,0xcf]
+; CHECK-NEXT: vgf2p8affineqb $3, %ymm1, %ymm0, %ymm3 ## EVEX TO VEX Compression encoding: [0xc4,0xe3,0xfd,0xce,0xd9,0x03]
+; CHECK-NEXT: vgf2p8affineqb $3, %ymm1, %ymm0, %ymm4 {%k1} {z} ## encoding: [0x62,0xf3,0xfd,0xa9,0xce,0xe1,0x03]
+; CHECK-NEXT: vgf2p8affineqb $3, %ymm1, %ymm0, %ymm2 {%k1} ## encoding: [0x62,0xf3,0xfd,0x29,0xce,0xd1,0x03]
+; CHECK-NEXT: vpxor %ymm3, %ymm2, %ymm0 ## EVEX TO VEX Compression encoding: [0xc5,0xed,0xef,0xc3]
+; CHECK-NEXT: vpxor %ymm0, %ymm4, %ymm0 ## EVEX TO VEX Compression encoding: [0xc5,0xdd,0xef,0xc0]
+; CHECK-NEXT: retq ## encoding: [0xc3]
+ %1 = bitcast i32 %mask to <32 x i1>
+ %2 = call <32 x i8> @llvm.x86.vgf2p8affineqb.256(<32 x i8> %src1, <32 x i8> %src2, i8 3)
+ %3 = select <32 x i1> %1, <32 x i8> %2, <32 x i8> zeroinitializer
+ %4 = select <32 x i1> %1, <32 x i8> %2, <32 x i8> %passthru
+ %5 = xor <32 x i8> %3, %4
+ %6 = xor <32 x i8> %5, %2
+ ret <32 x i8> %6
+}
+
+declare <64 x i8> @llvm.x86.vgf2p8affineqb.512(<64 x i8>, <64 x i8>, i8)
+define <64 x i8> @test_vgf2p8affineqb_512(<64 x i8> %src1, <64 x i8> %src2, <64 x i8> %passthru, i64 %mask) {
+; CHECK-LABEL: test_vgf2p8affineqb_512:
+; CHECK: ## BB#0:
+; CHECK-NEXT: kmovq %rdi, %k1 ## encoding: [0xc4,0xe1,0xfb,0x92,0xcf]
+; CHECK-NEXT: vgf2p8affineqb $3, %zmm1, %zmm0, %zmm3 ## encoding: [0x62,0xf3,0xfd,0x48,0xce,0xd9,0x03]
+; CHECK-NEXT: vgf2p8affineqb $3, %zmm1, %zmm0, %zmm4 {%k1} {z} ## encoding: [0x62,0xf3,0xfd,0xc9,0xce,0xe1,0x03]
+; CHECK-NEXT: vgf2p8affineqb $3, %zmm1, %zmm0, %zmm2 {%k1} ## encoding: [0x62,0xf3,0xfd,0x49,0xce,0xd1,0x03]
+; CHECK-NEXT: vpxorq %zmm3, %zmm2, %zmm0 ## encoding: [0x62,0xf1,0xed,0x48,0xef,0xc3]
+; CHECK-NEXT: vpxorq %zmm0, %zmm4, %zmm0 ## encoding: [0x62,0xf1,0xdd,0x48,0xef,0xc0]
+; CHECK-NEXT: retq ## encoding: [0xc3]
+ %1 = bitcast i64 %mask to <64 x i1>
+ %2 = call <64 x i8> @llvm.x86.vgf2p8affineqb.512(<64 x i8> %src1, <64 x i8> %src2, i8 3)
+ %3 = select <64 x i1> %1, <64 x i8> %2, <64 x i8> zeroinitializer
+ %4 = select <64 x i1> %1, <64 x i8> %2, <64 x i8> %passthru
+ %5 = xor <64 x i8> %3, %4
+ %6 = xor <64 x i8> %5, %2
+ ret <64 x i8> %6
+}
+
+declare <16 x i8> @llvm.x86.vgf2p8mulb.128(<16 x i8>, <16 x i8>)
+define <16 x i8> @test_vgf2p8mulb_128(<16 x i8> %src1, <16 x i8> %src2, <16 x i8> %passthru, i16 %mask) {
+; CHECK-LABEL: test_vgf2p8mulb_128:
+; CHECK: ## BB#0:
+; CHECK-NEXT: kmovd %edi, %k1 ## encoding: [0xc5,0xfb,0x92,0xcf]
+; CHECK-NEXT: vgf2p8mulb %xmm1, %xmm0, %xmm3 ## EVEX TO VEX Compression encoding: [0xc4,0xe2,0x79,0xcf,0xd9]
+; CHECK-NEXT: vgf2p8mulb %xmm1, %xmm0, %xmm4 {%k1} {z} ## encoding: [0x62,0xf2,0x7d,0x89,0xcf,0xe1]
+; CHECK-NEXT: vgf2p8mulb %xmm1, %xmm0, %xmm2 {%k1} ## encoding: [0x62,0xf2,0x7d,0x09,0xcf,0xd1]
+; CHECK-NEXT: vpxor %xmm3, %xmm2, %xmm0 ## EVEX TO VEX Compression encoding: [0xc5,0xe9,0xef,0xc3]
+; CHECK-NEXT: vpxor %xmm0, %xmm4, %xmm0 ## EVEX TO VEX Compression encoding: [0xc5,0xd9,0xef,0xc0]
+; CHECK-NEXT: retq ## encoding: [0xc3]
+ %1 = bitcast i16 %mask to <16 x i1>
+ %2 = call <16 x i8> @llvm.x86.vgf2p8mulb.128(<16 x i8> %src1, <16 x i8> %src2)
+ %3 = select <16 x i1> %1, <16 x i8> %2, <16 x i8> zeroinitializer
+ %4 = select <16 x i1> %1, <16 x i8> %2, <16 x i8> %passthru
+ %5 = xor <16 x i8> %3, %4
+ %6 = xor <16 x i8> %5, %2
+ ret <16 x i8> %6
+}
+
+declare <32 x i8> @llvm.x86.vgf2p8mulb.256(<32 x i8>, <32 x i8>)
+define <32 x i8> @test_vgf2p8mulb_256(<32 x i8> %src1, <32 x i8> %src2, <32 x i8> %passthru, i32 %mask) {
+; CHECK-LABEL: test_vgf2p8mulb_256:
+; CHECK: ## BB#0:
+; CHECK-NEXT: kmovd %edi, %k1 ## encoding: [0xc5,0xfb,0x92,0xcf]
+; CHECK-NEXT: vgf2p8mulb %ymm1, %ymm0, %ymm3 ## EVEX TO VEX Compression encoding: [0xc4,0xe2,0x7d,0xcf,0xd9]
+; CHECK-NEXT: vgf2p8mulb %ymm1, %ymm0, %ymm4 {%k1} {z} ## encoding: [0x62,0xf2,0x7d,0xa9,0xcf,0xe1]
+; CHECK-NEXT: vgf2p8mulb %ymm1, %ymm0, %ymm2 {%k1} ## encoding: [0x62,0xf2,0x7d,0x29,0xcf,0xd1]
+; CHECK-NEXT: vpxor %ymm3, %ymm2, %ymm0 ## EVEX TO VEX Compression encoding: [0xc5,0xed,0xef,0xc3]
+; CHECK-NEXT: vpxor %ymm0, %ymm4, %ymm0 ## EVEX TO VEX Compression encoding: [0xc5,0xdd,0xef,0xc0]
+; CHECK-NEXT: retq ## encoding: [0xc3]
+ %1 = bitcast i32 %mask to <32 x i1>
+ %2 = call <32 x i8> @llvm.x86.vgf2p8mulb.256(<32 x i8> %src1, <32 x i8> %src2)
+ %3 = select <32 x i1> %1, <32 x i8> %2, <32 x i8> zeroinitializer
+ %4 = select <32 x i1> %1, <32 x i8> %2, <32 x i8> %passthru
+ %5 = xor <32 x i8> %3, %4
+ %6 = xor <32 x i8> %5, %2
+ ret <32 x i8> %6
+}
+
+declare <64 x i8> @llvm.x86.vgf2p8mulb.512(<64 x i8>, <64 x i8>)
+define <64 x i8> @test_vgf2p8mulb_512(<64 x i8> %src1, <64 x i8> %src2, <64 x i8> %passthru, i64 %mask) {
+; CHECK-LABEL: test_vgf2p8mulb_512:
+; CHECK: ## BB#0:
+; CHECK-NEXT: kmovq %rdi, %k1 ## encoding: [0xc4,0xe1,0xfb,0x92,0xcf]
+; CHECK-NEXT: vgf2p8mulb %zmm1, %zmm0, %zmm3 ## encoding: [0x62,0xf2,0x7d,0x48,0xcf,0xd9]
+; CHECK-NEXT: vgf2p8mulb %zmm1, %zmm0, %zmm4 {%k1} {z} ## encoding: [0x62,0xf2,0x7d,0xc9,0xcf,0xe1]
+; CHECK-NEXT: vgf2p8mulb %zmm1, %zmm0, %zmm2 {%k1} ## encoding: [0x62,0xf2,0x7d,0x49,0xcf,0xd1]
+; CHECK-NEXT: vpxorq %zmm3, %zmm2, %zmm0 ## encoding: [0x62,0xf1,0xed,0x48,0xef,0xc3]
+; CHECK-NEXT: vpxorq %zmm0, %zmm4, %zmm0 ## encoding: [0x62,0xf1,0xdd,0x48,0xef,0xc0]
+; CHECK-NEXT: retq ## encoding: [0xc3]
+ %1 = bitcast i64 %mask to <64 x i1>
+ %2 = call <64 x i8> @llvm.x86.vgf2p8mulb.512(<64 x i8> %src1, <64 x i8> %src2)
+ %3 = select <64 x i1> %1, <64 x i8> %2, <64 x i8> zeroinitializer
+ %4 = select <64 x i1> %1, <64 x i8> %2, <64 x i8> %passthru
+ %5 = xor <64 x i8> %3, %4
+ %6 = xor <64 x i8> %5, %2
+ ret <64 x i8> %6
+}
+
Added: llvm/trunk/test/CodeGen/X86/gfni-intrinsics.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/X86/gfni-intrinsics.ll?rev=318993&view=auto
==============================================================================
--- llvm/trunk/test/CodeGen/X86/gfni-intrinsics.ll (added)
+++ llvm/trunk/test/CodeGen/X86/gfni-intrinsics.ll Sun Nov 26 01:36:41 2017
@@ -0,0 +1,33 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc < %s -mtriple=i386-apple-darwin -mattr=+gfni -show-mc-encoding | FileCheck %s
+
+declare <16 x i8> @llvm.x86.vgf2p8affineinvqb.128(<16 x i8>, <16 x i8>, i8)
+define <16 x i8> @test_gf2p8affineinvqb_128(<16 x i8> %src1, <16 x i8> %src2) {
+; CHECK-LABEL: test_gf2p8affineinvqb_128:
+; CHECK: ## BB#0:
+; CHECK-NEXT: gf2p8affineinvqb $11, %xmm1, %xmm0 ## encoding: [0x66,0x0f,0x3a,0xcf,0xc1,0x0b]
+; CHECK-NEXT: retl ## encoding: [0xc3]
+ %1 = call <16 x i8> @llvm.x86.vgf2p8affineinvqb.128(<16 x i8> %src1, <16 x i8> %src2, i8 11)
+ ret <16 x i8> %1
+}
+
+declare <16 x i8> @llvm.x86.vgf2p8affineqb.128(<16 x i8>, <16 x i8>, i8)
+define <16 x i8> @test_gf2p8affineqb_128(<16 x i8> %src1, <16 x i8> %src2) {
+; CHECK-LABEL: test_gf2p8affineqb_128:
+; CHECK: ## BB#0:
+; CHECK-NEXT: gf2p8affineqb $11, %xmm1, %xmm0 ## encoding: [0x66,0x0f,0x3a,0xce,0xc1,0x0b]
+; CHECK-NEXT: retl ## encoding: [0xc3]
+ %1 = call <16 x i8> @llvm.x86.vgf2p8affineqb.128(<16 x i8> %src1, <16 x i8> %src2, i8 11)
+ ret <16 x i8> %1
+}
+
+declare <16 x i8> @llvm.x86.vgf2p8mulb.128(<16 x i8>, <16 x i8>)
+define <16 x i8> @test_gf2p8mulb_128(<16 x i8> %src1, <16 x i8> %src2) {
+; CHECK-LABEL: test_gf2p8mulb_128:
+; CHECK: ## BB#0:
+; CHECK-NEXT: gf2p8mulb %xmm1, %xmm0 ## encoding: [0x66,0x0f,0x38,0xcf,0xc1]
+; CHECK-NEXT: retl ## encoding: [0xc3]
+ %1 = call <16 x i8> @llvm.x86.vgf2p8mulb.128(<16 x i8> %src1, <16 x i8> %src2)
+ ret <16 x i8> %1
+}
+
Added: llvm/trunk/test/MC/X86/avx512gfni-encoding.s
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/MC/X86/avx512gfni-encoding.s?rev=318993&view=auto
==============================================================================
--- llvm/trunk/test/MC/X86/avx512gfni-encoding.s (added)
+++ llvm/trunk/test/MC/X86/avx512gfni-encoding.s Sun Nov 26 01:36:41 2017
@@ -0,0 +1,178 @@
+// RUN: llvm-mc -triple x86_64-unknown-unknown -mattr=+gfni,+avx512f,+avx512bw --show-encoding < %s | FileCheck %s
+
+// CHECK: vgf2p8affineinvqb $7, %zmm2, %zmm20, %zmm1
+// CHECK: encoding: [0x62,0xf3,0xdd,0x40,0xcf,0xca,0x07]
+ vgf2p8affineinvqb $7, %zmm2, %zmm20, %zmm1
+
+// CHECK: vgf2p8affineqb $7, %zmm2, %zmm20, %zmm1
+// CHECK: encoding: [0x62,0xf3,0xdd,0x40,0xce,0xca,0x07]
+ vgf2p8affineqb $7, %zmm2, %zmm20, %zmm1
+
+// CHECK: vgf2p8affineinvqb $7, %zmm2, %zmm20, %zmm1 {%k2}
+// CHECK: encoding: [0x62,0xf3,0xdd,0x42,0xcf,0xca,0x07]
+ vgf2p8affineinvqb $7, %zmm2, %zmm20, %zmm1 {%k2}
+
+// CHECK: vgf2p8affineqb $7, %zmm2, %zmm20, %zmm1 {%k2}
+// CHECK: encoding: [0x62,0xf3,0xdd,0x42,0xce,0xca,0x07]
+ vgf2p8affineqb $7, %zmm2, %zmm20, %zmm1 {%k2}
+
+// CHECK: vgf2p8affineinvqb $7, (%rcx), %zmm20, %zmm1
+// CHECK: encoding: [0x62,0xf3,0xdd,0x40,0xcf,0x09,0x07]
+ vgf2p8affineinvqb $7, (%rcx), %zmm20, %zmm1
+
+// CHECK: vgf2p8affineinvqb $7, -256(%rsp), %zmm20, %zmm1
+// CHECK: encoding: [0x62,0xf3,0xdd,0x40,0xcf,0x4c,0x24,0xfc,0x07]
+ vgf2p8affineinvqb $7, -256(%rsp), %zmm20, %zmm1
+
+// CHECK: vgf2p8affineinvqb $7, 256(%rsp), %zmm20, %zmm1
+// CHECK: encoding: [0x62,0xf3,0xdd,0x40,0xcf,0x4c,0x24,0x04,0x07]
+ vgf2p8affineinvqb $7, 256(%rsp), %zmm20, %zmm1
+
+// CHECK: vgf2p8affineinvqb $7, 268435456(%rcx,%r14,8), %zmm20, %zmm1
+// CHECK: encoding: [0x62,0xb3,0xdd,0x40,0xcf,0x8c,0xf1,0x00,0x00,0x00,0x10,0x07]
+ vgf2p8affineinvqb $7, 268435456(%rcx,%r14,8), %zmm20, %zmm1
+
+// CHECK: vgf2p8affineinvqb $7, -536870912(%rcx,%r14,8), %zmm20, %zmm1
+// CHECK: encoding: [0x62,0xb3,0xdd,0x40,0xcf,0x8c,0xf1,0x00,0x00,0x00,0xe0,0x07]
+ vgf2p8affineinvqb $7, -536870912(%rcx,%r14,8), %zmm20, %zmm1
+
+// CHECK: vgf2p8affineinvqb $7, -536870910(%rcx,%r14,8), %zmm20, %zmm1
+// CHECK: encoding: [0x62,0xb3,0xdd,0x40,0xcf,0x8c,0xf1,0x02,0x00,0x00,0xe0,0x07]
+ vgf2p8affineinvqb $7, -536870910(%rcx,%r14,8), %zmm20, %zmm1
+
+// CHECK: vgf2p8affineqb $7, (%rcx), %zmm20, %zmm1
+// CHECK: encoding: [0x62,0xf3,0xdd,0x40,0xce,0x09,0x07]
+ vgf2p8affineqb $7, (%rcx), %zmm20, %zmm1
+
+// CHECK: vgf2p8affineqb $7, -256(%rsp), %zmm20, %zmm1
+// CHECK: encoding: [0x62,0xf3,0xdd,0x40,0xce,0x4c,0x24,0xfc,0x07]
+ vgf2p8affineqb $7, -256(%rsp), %zmm20, %zmm1
+
+// CHECK: vgf2p8affineqb $7, 256(%rsp), %zmm20, %zmm1
+// CHECK: encoding: [0x62,0xf3,0xdd,0x40,0xce,0x4c,0x24,0x04,0x07]
+ vgf2p8affineqb $7, 256(%rsp), %zmm20, %zmm1
+
+// CHECK: vgf2p8affineqb $7, 268435456(%rcx,%r14,8), %zmm20, %zmm1
+// CHECK: encoding: [0x62,0xb3,0xdd,0x40,0xce,0x8c,0xf1,0x00,0x00,0x00,0x10,0x07]
+ vgf2p8affineqb $7, 268435456(%rcx,%r14,8), %zmm20, %zmm1
+
+// CHECK: vgf2p8affineqb $7, -536870912(%rcx,%r14,8), %zmm20, %zmm1
+// CHECK: encoding: [0x62,0xb3,0xdd,0x40,0xce,0x8c,0xf1,0x00,0x00,0x00,0xe0,0x07]
+ vgf2p8affineqb $7, -536870912(%rcx,%r14,8), %zmm20, %zmm1
+
+// CHECK: vgf2p8affineqb $7, -536870910(%rcx,%r14,8), %zmm20, %zmm1
+// CHECK: encoding: [0x62,0xb3,0xdd,0x40,0xce,0x8c,0xf1,0x02,0x00,0x00,0xe0,0x07]
+ vgf2p8affineqb $7, -536870910(%rcx,%r14,8), %zmm20, %zmm1
+
+// CHECK: vgf2p8affineinvqb $7, (%rcx), %zmm20, %zmm1 {%k2}
+// CHECK: encoding: [0x62,0xf3,0xdd,0x42,0xcf,0x09,0x07]
+ vgf2p8affineinvqb $7, (%rcx), %zmm20, %zmm1 {%k2}
+
+// CHECK: vgf2p8affineinvqb $7, -256(%rsp), %zmm20, %zmm1 {%k2}
+// CHECK: encoding: [0x62,0xf3,0xdd,0x42,0xcf,0x4c,0x24,0xfc,0x07]
+ vgf2p8affineinvqb $7, -256(%rsp), %zmm20, %zmm1 {%k2}
+
+// CHECK: vgf2p8affineinvqb $7, 256(%rsp), %zmm20, %zmm1 {%k2}
+// CHECK: encoding: [0x62,0xf3,0xdd,0x42,0xcf,0x4c,0x24,0x04,0x07]
+ vgf2p8affineinvqb $7, 256(%rsp), %zmm20, %zmm1 {%k2}
+
+// CHECK: vgf2p8affineinvqb $7, 268435456(%rcx,%r14,8), %zmm20, %zmm1 {%k2}
+// CHECK: encoding: [0x62,0xb3,0xdd,0x42,0xcf,0x8c,0xf1,0x00,0x00,0x00,0x10,0x07]
+ vgf2p8affineinvqb $7, 268435456(%rcx,%r14,8), %zmm20, %zmm1 {%k2}
+
+// CHECK: vgf2p8affineinvqb $7, -536870912(%rcx,%r14,8), %zmm20, %zmm1 {%k2}
+// CHECK: encoding: [0x62,0xb3,0xdd,0x42,0xcf,0x8c,0xf1,0x00,0x00,0x00,0xe0,0x07]
+ vgf2p8affineinvqb $7, -536870912(%rcx,%r14,8), %zmm20, %zmm1 {%k2}
+
+// CHECK: vgf2p8affineinvqb $7, -536870910(%rcx,%r14,8), %zmm20, %zmm1 {%k2}
+// CHECK: encoding: [0x62,0xb3,0xdd,0x42,0xcf,0x8c,0xf1,0x02,0x00,0x00,0xe0,0x07]
+ vgf2p8affineinvqb $7, -536870910(%rcx,%r14,8), %zmm20, %zmm1 {%k2}
+
+// CHECK: vgf2p8affineqb $7, (%rcx), %zmm20, %zmm1 {%k2}
+// CHECK: encoding: [0x62,0xf3,0xdd,0x42,0xce,0x09,0x07]
+ vgf2p8affineqb $7, (%rcx), %zmm20, %zmm1 {%k2}
+
+// CHECK: vgf2p8affineqb $7, -256(%rsp), %zmm20, %zmm1 {%k2}
+// CHECK: encoding: [0x62,0xf3,0xdd,0x42,0xce,0x4c,0x24,0xfc,0x07]
+ vgf2p8affineqb $7, -256(%rsp), %zmm20, %zmm1 {%k2}
+
+// CHECK: vgf2p8affineqb $7, 256(%rsp), %zmm20, %zmm1 {%k2}
+// CHECK: encoding: [0x62,0xf3,0xdd,0x42,0xce,0x4c,0x24,0x04,0x07]
+ vgf2p8affineqb $7, 256(%rsp), %zmm20, %zmm1 {%k2}
+
+// CHECK: vgf2p8affineqb $7, 268435456(%rcx,%r14,8), %zmm20, %zmm1 {%k2}
+// CHECK: encoding: [0x62,0xb3,0xdd,0x42,0xce,0x8c,0xf1,0x00,0x00,0x00,0x10,0x07]
+ vgf2p8affineqb $7, 268435456(%rcx,%r14,8), %zmm20, %zmm1 {%k2}
+
+// CHECK: vgf2p8affineqb $7, -536870912(%rcx,%r14,8), %zmm20, %zmm1 {%k2}
+// CHECK: encoding: [0x62,0xb3,0xdd,0x42,0xce,0x8c,0xf1,0x00,0x00,0x00,0xe0,0x07]
+ vgf2p8affineqb $7, -536870912(%rcx,%r14,8), %zmm20, %zmm1 {%k2}
+
+// CHECK: vgf2p8affineqb $7, -536870910(%rcx,%r14,8), %zmm20, %zmm1 {%k2}
+// CHECK: encoding: [0x62,0xb3,0xdd,0x42,0xce,0x8c,0xf1,0x02,0x00,0x00,0xe0,0x07]
+ vgf2p8affineqb $7, -536870910(%rcx,%r14,8), %zmm20, %zmm1 {%k2}
+
+// CHECK: vgf2p8mulb %zmm2, %zmm20, %zmm1
+// CHECK: encoding: [0x62,0xf2,0x5d,0x40,0xcf,0xca]
+ vgf2p8mulb %zmm2, %zmm20, %zmm1
+
+// CHECK: vgf2p8mulb %zmm2, %zmm20, %zmm1 {%k2}
+// CHECK: encoding: [0x62,0xf2,0x5d,0x42,0xcf,0xca]
+ vgf2p8mulb %zmm2, %zmm20, %zmm1 {%k2}
+
+// CHECK: vgf2p8mulb (%rcx), %zmm20, %zmm1
+// CHECK: encoding: [0x62,0xf2,0x5d,0x40,0xcf,0x09]
+ vgf2p8mulb (%rcx), %zmm20, %zmm1
+
+// CHECK: vgf2p8mulb -256(%rsp), %zmm20, %zmm1
+// CHECK: encoding: [0x62,0xf2,0x5d,0x40,0xcf,0x4c,0x24,0xfc]
+ vgf2p8mulb -256(%rsp), %zmm20, %zmm1
+
+// CHECK: vgf2p8mulb 256(%rsp), %zmm20, %zmm1
+// CHECK: encoding: [0x62,0xf2,0x5d,0x40,0xcf,0x4c,0x24,0x04]
+ vgf2p8mulb 256(%rsp), %zmm20, %zmm1
+
+// CHECK: vgf2p8mulb 268435456(%rcx,%r14,8), %zmm20, %zmm1
+// CHECK: encoding: [0x62,0xb2,0x5d,0x40,0xcf,0x8c,0xf1,0x00,0x00,0x00,0x10]
+ vgf2p8mulb 268435456(%rcx,%r14,8), %zmm20, %zmm1
+
+// CHECK: vgf2p8mulb -536870912(%rcx,%r14,8), %zmm20, %zmm1
+// CHECK: encoding: [0x62,0xb2,0x5d,0x40,0xcf,0x8c,0xf1,0x00,0x00,0x00,0xe0]
+ vgf2p8mulb -536870912(%rcx,%r14,8), %zmm20, %zmm1
+
+// CHECK: vgf2p8mulb -536870910(%rcx,%r14,8), %zmm20, %zmm1
+// CHECK: encoding: [0x62,0xb2,0x5d,0x40,0xcf,0x8c,0xf1,0x02,0x00,0x00,0xe0]
+ vgf2p8mulb -536870910(%rcx,%r14,8), %zmm20, %zmm1
+
+// CHECK: vgf2p8mulb (%rcx), %zmm20, %zmm1 {%k2}
+// CHECK: encoding: [0x62,0xf2,0x5d,0x42,0xcf,0x09]
+ vgf2p8mulb (%rcx), %zmm20, %zmm1 {%k2}
+
+// CHECK: vgf2p8mulb -256(%rsp), %zmm20, %zmm1 {%k2}
+// CHECK: encoding: [0x62,0xf2,0x5d,0x42,0xcf,0x4c,0x24,0xfc]
+ vgf2p8mulb -256(%rsp), %zmm20, %zmm1 {%k2}
+
+// CHECK: vgf2p8mulb 256(%rsp), %zmm20, %zmm1 {%k2}
+// CHECK: encoding: [0x62,0xf2,0x5d,0x42,0xcf,0x4c,0x24,0x04]
+ vgf2p8mulb 256(%rsp), %zmm20, %zmm1 {%k2}
+
+// CHECK: vgf2p8mulb 268435456(%rcx,%r14,8), %zmm20, %zmm1 {%k2}
+// CHECK: encoding: [0x62,0xb2,0x5d,0x42,0xcf,0x8c,0xf1,0x00,0x00,0x00,0x10]
+ vgf2p8mulb 268435456(%rcx,%r14,8), %zmm20, %zmm1 {%k2}
+
+// CHECK: vgf2p8mulb -536870912(%rcx,%r14,8), %zmm20, %zmm1 {%k2}
+// CHECK: encoding: [0x62,0xb2,0x5d,0x42,0xcf,0x8c,0xf1,0x00,0x00,0x00,0xe0]
+ vgf2p8mulb -536870912(%rcx,%r14,8), %zmm20, %zmm1 {%k2}
+
+// CHECK: vgf2p8mulb -536870910(%rcx,%r14,8), %zmm20, %zmm1 {%k2}
+// CHECK: encoding: [0x62,0xb2,0x5d,0x42,0xcf,0x8c,0xf1,0x02,0x00,0x00,0xe0]
+ vgf2p8mulb -536870910(%rcx,%r14,8), %zmm20, %zmm1 {%k2}
+
+// CHECK: vgf2p8affineinvqb $7, (%rcx){1to8}, %zmm20, %zmm1
+// CHECK: encoding: [0x62,0xf3,0xdd,0x50,0xcf,0x09,0x07]
+ vgf2p8affineinvqb $7, (%rcx){1to8}, %zmm20, %zmm1
+
+// CHECK: vgf2p8affineqb $7, (%rcx){1to8}, %zmm20, %zmm1
+// CHECK: encoding: [0x62,0xf3,0xdd,0x50,0xce,0x09,0x07]
+ vgf2p8affineqb $7, (%rcx){1to8}, %zmm20, %zmm1
+
Added: llvm/trunk/test/MC/X86/avx512vl_gfni-encoding.s
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/MC/X86/avx512vl_gfni-encoding.s?rev=318993&view=auto
==============================================================================
--- llvm/trunk/test/MC/X86/avx512vl_gfni-encoding.s (added)
+++ llvm/trunk/test/MC/X86/avx512vl_gfni-encoding.s Sun Nov 26 01:36:41 2017
@@ -0,0 +1,354 @@
+// RUN: llvm-mc -triple x86_64-unknown-unknown -mattr=+gfni,+avx512vl,+avx512bw --show-encoding < %s | FileCheck %s
+
+// CHECK: vgf2p8affineinvqb $7, %xmm2, %xmm20, %xmm1
+// CHECK: encoding: [0x62,0xf3,0xdd,0x00,0xcf,0xca,0x07]
+ vgf2p8affineinvqb $7, %xmm2, %xmm20, %xmm1
+
+// CHECK: vgf2p8affineqb $7, %xmm2, %xmm20, %xmm1
+// CHECK: encoding: [0x62,0xf3,0xdd,0x00,0xce,0xca,0x07]
+ vgf2p8affineqb $7, %xmm2, %xmm20, %xmm1
+
+// CHECK: vgf2p8affineinvqb $7, %xmm2, %xmm20, %xmm1 {%k2}
+// CHECK: encoding: [0x62,0xf3,0xdd,0x02,0xcf,0xca,0x07]
+ vgf2p8affineinvqb $7, %xmm2, %xmm20, %xmm1 {%k2}
+
+// CHECK: vgf2p8affineqb $7, %xmm2, %xmm20, %xmm1 {%k2}
+// CHECK: encoding: [0x62,0xf3,0xdd,0x02,0xce,0xca,0x07]
+ vgf2p8affineqb $7, %xmm2, %xmm20, %xmm1 {%k2}
+
+// CHECK: vgf2p8affineinvqb $7, (%rcx), %xmm20, %xmm1
+// CHECK: encoding: [0x62,0xf3,0xdd,0x00,0xcf,0x09,0x07]
+ vgf2p8affineinvqb $7, (%rcx), %xmm20, %xmm1
+
+// CHECK: vgf2p8affineinvqb $7, -64(%rsp), %xmm20, %xmm1
+// CHECK: encoding: [0x62,0xf3,0xdd,0x00,0xcf,0x4c,0x24,0xfc,0x07]
+ vgf2p8affineinvqb $7, -64(%rsp), %xmm20, %xmm1
+
+// CHECK: vgf2p8affineinvqb $7, 64(%rsp), %xmm20, %xmm1
+// CHECK: encoding: [0x62,0xf3,0xdd,0x00,0xcf,0x4c,0x24,0x04,0x07]
+ vgf2p8affineinvqb $7, 64(%rsp), %xmm20, %xmm1
+
+// CHECK: vgf2p8affineinvqb $7, 268435456(%rcx,%r14,8), %xmm20, %xmm1
+// CHECK: encoding: [0x62,0xb3,0xdd,0x00,0xcf,0x8c,0xf1,0x00,0x00,0x00,0x10,0x07]
+ vgf2p8affineinvqb $7, 268435456(%rcx,%r14,8), %xmm20, %xmm1
+
+// CHECK: vgf2p8affineinvqb $7, -536870912(%rcx,%r14,8), %xmm20, %xmm1
+// CHECK: encoding: [0x62,0xb3,0xdd,0x00,0xcf,0x8c,0xf1,0x00,0x00,0x00,0xe0,0x07]
+ vgf2p8affineinvqb $7, -536870912(%rcx,%r14,8), %xmm20, %xmm1
+
+// CHECK: vgf2p8affineinvqb $7, -536870910(%rcx,%r14,8), %xmm20, %xmm1
+// CHECK: encoding: [0x62,0xb3,0xdd,0x00,0xcf,0x8c,0xf1,0x02,0x00,0x00,0xe0,0x07]
+ vgf2p8affineinvqb $7, -536870910(%rcx,%r14,8), %xmm20, %xmm1
+
+// CHECK: vgf2p8affineqb $7, (%rcx), %xmm20, %xmm1
+// CHECK: encoding: [0x62,0xf3,0xdd,0x00,0xce,0x09,0x07]
+ vgf2p8affineqb $7, (%rcx), %xmm20, %xmm1
+
+// CHECK: vgf2p8affineqb $7, -64(%rsp), %xmm20, %xmm1
+// CHECK: encoding: [0x62,0xf3,0xdd,0x00,0xce,0x4c,0x24,0xfc,0x07]
+ vgf2p8affineqb $7, -64(%rsp), %xmm20, %xmm1
+
+// CHECK: vgf2p8affineqb $7, 64(%rsp), %xmm20, %xmm1
+// CHECK: encoding: [0x62,0xf3,0xdd,0x00,0xce,0x4c,0x24,0x04,0x07]
+ vgf2p8affineqb $7, 64(%rsp), %xmm20, %xmm1
+
+// CHECK: vgf2p8affineqb $7, 268435456(%rcx,%r14,8), %xmm20, %xmm1
+// CHECK: encoding: [0x62,0xb3,0xdd,0x00,0xce,0x8c,0xf1,0x00,0x00,0x00,0x10,0x07]
+ vgf2p8affineqb $7, 268435456(%rcx,%r14,8), %xmm20, %xmm1
+
+// CHECK: vgf2p8affineqb $7, -536870912(%rcx,%r14,8), %xmm20, %xmm1
+// CHECK: encoding: [0x62,0xb3,0xdd,0x00,0xce,0x8c,0xf1,0x00,0x00,0x00,0xe0,0x07]
+ vgf2p8affineqb $7, -536870912(%rcx,%r14,8), %xmm20, %xmm1
+
+// CHECK: vgf2p8affineqb $7, -536870910(%rcx,%r14,8), %xmm20, %xmm1
+// CHECK: encoding: [0x62,0xb3,0xdd,0x00,0xce,0x8c,0xf1,0x02,0x00,0x00,0xe0,0x07]
+ vgf2p8affineqb $7, -536870910(%rcx,%r14,8), %xmm20, %xmm1
+
+// CHECK: vgf2p8affineinvqb $7, (%rcx), %xmm20, %xmm1 {%k2}
+// CHECK: encoding: [0x62,0xf3,0xdd,0x02,0xcf,0x09,0x07]
+ vgf2p8affineinvqb $7, (%rcx), %xmm20, %xmm1 {%k2}
+
+// CHECK: vgf2p8affineinvqb $7, -64(%rsp), %xmm20, %xmm1 {%k2}
+// CHECK: encoding: [0x62,0xf3,0xdd,0x02,0xcf,0x4c,0x24,0xfc,0x07]
+ vgf2p8affineinvqb $7, -64(%rsp), %xmm20, %xmm1 {%k2}
+
+// CHECK: vgf2p8affineinvqb $7, 64(%rsp), %xmm20, %xmm1 {%k2}
+// CHECK: encoding: [0x62,0xf3,0xdd,0x02,0xcf,0x4c,0x24,0x04,0x07]
+ vgf2p8affineinvqb $7, 64(%rsp), %xmm20, %xmm1 {%k2}
+
+// CHECK: vgf2p8affineinvqb $7, 268435456(%rcx,%r14,8), %xmm20, %xmm1 {%k2}
+// CHECK: encoding: [0x62,0xb3,0xdd,0x02,0xcf,0x8c,0xf1,0x00,0x00,0x00,0x10,0x07]
+ vgf2p8affineinvqb $7, 268435456(%rcx,%r14,8), %xmm20, %xmm1 {%k2}
+
+// CHECK: vgf2p8affineinvqb $7, -536870912(%rcx,%r14,8), %xmm20, %xmm1 {%k2}
+// CHECK: encoding: [0x62,0xb3,0xdd,0x02,0xcf,0x8c,0xf1,0x00,0x00,0x00,0xe0,0x07]
+ vgf2p8affineinvqb $7, -536870912(%rcx,%r14,8), %xmm20, %xmm1 {%k2}
+
+// CHECK: vgf2p8affineinvqb $7, -536870910(%rcx,%r14,8), %xmm20, %xmm1 {%k2}
+// CHECK: encoding: [0x62,0xb3,0xdd,0x02,0xcf,0x8c,0xf1,0x02,0x00,0x00,0xe0,0x07]
+ vgf2p8affineinvqb $7, -536870910(%rcx,%r14,8), %xmm20, %xmm1 {%k2}
+
+// CHECK: vgf2p8affineqb $7, (%rcx), %xmm20, %xmm1 {%k2}
+// CHECK: encoding: [0x62,0xf3,0xdd,0x02,0xce,0x09,0x07]
+ vgf2p8affineqb $7, (%rcx), %xmm20, %xmm1 {%k2}
+
+// CHECK: vgf2p8affineqb $7, -64(%rsp), %xmm20, %xmm1 {%k2}
+// CHECK: encoding: [0x62,0xf3,0xdd,0x02,0xce,0x4c,0x24,0xfc,0x07]
+ vgf2p8affineqb $7, -64(%rsp), %xmm20, %xmm1 {%k2}
+
+// CHECK: vgf2p8affineqb $7, 64(%rsp), %xmm20, %xmm1 {%k2}
+// CHECK: encoding: [0x62,0xf3,0xdd,0x02,0xce,0x4c,0x24,0x04,0x07]
+ vgf2p8affineqb $7, 64(%rsp), %xmm20, %xmm1 {%k2}
+
+// CHECK: vgf2p8affineqb $7, 268435456(%rcx,%r14,8), %xmm20, %xmm1 {%k2}
+// CHECK: encoding: [0x62,0xb3,0xdd,0x02,0xce,0x8c,0xf1,0x00,0x00,0x00,0x10,0x07]
+ vgf2p8affineqb $7, 268435456(%rcx,%r14,8), %xmm20, %xmm1 {%k2}
+
+// CHECK: vgf2p8affineqb $7, -536870912(%rcx,%r14,8), %xmm20, %xmm1 {%k2}
+// CHECK: encoding: [0x62,0xb3,0xdd,0x02,0xce,0x8c,0xf1,0x00,0x00,0x00,0xe0,0x07]
+ vgf2p8affineqb $7, -536870912(%rcx,%r14,8), %xmm20, %xmm1 {%k2}
+
+// CHECK: vgf2p8affineqb $7, -536870910(%rcx,%r14,8), %xmm20, %xmm1 {%k2}
+// CHECK: encoding: [0x62,0xb3,0xdd,0x02,0xce,0x8c,0xf1,0x02,0x00,0x00,0xe0,0x07]
+ vgf2p8affineqb $7, -536870910(%rcx,%r14,8), %xmm20, %xmm1 {%k2}
+
+// CHECK: vgf2p8affineinvqb $7, %ymm2, %ymm20, %ymm1
+// CHECK: encoding: [0x62,0xf3,0xdd,0x20,0xcf,0xca,0x07]
+ vgf2p8affineinvqb $7, %ymm2, %ymm20, %ymm1
+
+// CHECK: vgf2p8affineqb $7, %ymm2, %ymm20, %ymm1
+// CHECK: encoding: [0x62,0xf3,0xdd,0x20,0xce,0xca,0x07]
+ vgf2p8affineqb $7, %ymm2, %ymm20, %ymm1
+
+// CHECK: vgf2p8affineinvqb $7, %ymm2, %ymm20, %ymm1 {%k2}
+// CHECK: encoding: [0x62,0xf3,0xdd,0x22,0xcf,0xca,0x07]
+ vgf2p8affineinvqb $7, %ymm2, %ymm20, %ymm1 {%k2}
+
+// CHECK: vgf2p8affineqb $7, %ymm2, %ymm20, %ymm1 {%k2}
+// CHECK: encoding: [0x62,0xf3,0xdd,0x22,0xce,0xca,0x07]
+ vgf2p8affineqb $7, %ymm2, %ymm20, %ymm1 {%k2}
+
+// CHECK: vgf2p8affineinvqb $7, (%rcx), %ymm20, %ymm1
+// CHECK: encoding: [0x62,0xf3,0xdd,0x20,0xcf,0x09,0x07]
+ vgf2p8affineinvqb $7, (%rcx), %ymm20, %ymm1
+
+// CHECK: vgf2p8affineinvqb $7, -128(%rsp), %ymm20, %ymm1
+// CHECK: encoding: [0x62,0xf3,0xdd,0x20,0xcf,0x4c,0x24,0xfc,0x07]
+ vgf2p8affineinvqb $7, -128(%rsp), %ymm20, %ymm1
+
+// CHECK: vgf2p8affineinvqb $7, 128(%rsp), %ymm20, %ymm1
+// CHECK: encoding: [0x62,0xf3,0xdd,0x20,0xcf,0x4c,0x24,0x04,0x07]
+ vgf2p8affineinvqb $7, 128(%rsp), %ymm20, %ymm1
+
+// CHECK: vgf2p8affineinvqb $7, 268435456(%rcx,%r14,8), %ymm20, %ymm1
+// CHECK: encoding: [0x62,0xb3,0xdd,0x20,0xcf,0x8c,0xf1,0x00,0x00,0x00,0x10,0x07]
+ vgf2p8affineinvqb $7, 268435456(%rcx,%r14,8), %ymm20, %ymm1
+
+// CHECK: vgf2p8affineinvqb $7, -536870912(%rcx,%r14,8), %ymm20, %ymm1
+// CHECK: encoding: [0x62,0xb3,0xdd,0x20,0xcf,0x8c,0xf1,0x00,0x00,0x00,0xe0,0x07]
+ vgf2p8affineinvqb $7, -536870912(%rcx,%r14,8), %ymm20, %ymm1
+
+// CHECK: vgf2p8affineinvqb $7, -536870910(%rcx,%r14,8), %ymm20, %ymm1
+// CHECK: encoding: [0x62,0xb3,0xdd,0x20,0xcf,0x8c,0xf1,0x02,0x00,0x00,0xe0,0x07]
+ vgf2p8affineinvqb $7, -536870910(%rcx,%r14,8), %ymm20, %ymm1
+
+// CHECK: vgf2p8affineqb $7, (%rcx), %ymm20, %ymm1
+// CHECK: encoding: [0x62,0xf3,0xdd,0x20,0xce,0x09,0x07]
+ vgf2p8affineqb $7, (%rcx), %ymm20, %ymm1
+
+// CHECK: vgf2p8affineqb $7, -128(%rsp), %ymm20, %ymm1
+// CHECK: encoding: [0x62,0xf3,0xdd,0x20,0xce,0x4c,0x24,0xfc,0x07]
+ vgf2p8affineqb $7, -128(%rsp), %ymm20, %ymm1
+
+// CHECK: vgf2p8affineqb $7, 128(%rsp), %ymm20, %ymm1
+// CHECK: encoding: [0x62,0xf3,0xdd,0x20,0xce,0x4c,0x24,0x04,0x07]
+ vgf2p8affineqb $7, 128(%rsp), %ymm20, %ymm1
+
+// CHECK: vgf2p8affineqb $7, 268435456(%rcx,%r14,8), %ymm20, %ymm1
+// CHECK: encoding: [0x62,0xb3,0xdd,0x20,0xce,0x8c,0xf1,0x00,0x00,0x00,0x10,0x07]
+ vgf2p8affineqb $7, 268435456(%rcx,%r14,8), %ymm20, %ymm1
+
+// CHECK: vgf2p8affineqb $7, -536870912(%rcx,%r14,8), %ymm20, %ymm1
+// CHECK: encoding: [0x62,0xb3,0xdd,0x20,0xce,0x8c,0xf1,0x00,0x00,0x00,0xe0,0x07]
+ vgf2p8affineqb $7, -536870912(%rcx,%r14,8), %ymm20, %ymm1
+
+// CHECK: vgf2p8affineqb $7, -536870910(%rcx,%r14,8), %ymm20, %ymm1
+// CHECK: encoding: [0x62,0xb3,0xdd,0x20,0xce,0x8c,0xf1,0x02,0x00,0x00,0xe0,0x07]
+ vgf2p8affineqb $7, -536870910(%rcx,%r14,8), %ymm20, %ymm1
+
+// CHECK: vgf2p8affineinvqb $7, (%rcx), %ymm20, %ymm1 {%k2}
+// CHECK: encoding: [0x62,0xf3,0xdd,0x22,0xcf,0x09,0x07]
+ vgf2p8affineinvqb $7, (%rcx), %ymm20, %ymm1 {%k2}
+
+// CHECK: vgf2p8affineinvqb $7, -128(%rsp), %ymm20, %ymm1 {%k2}
+// CHECK: encoding: [0x62,0xf3,0xdd,0x22,0xcf,0x4c,0x24,0xfc,0x07]
+ vgf2p8affineinvqb $7, -128(%rsp), %ymm20, %ymm1 {%k2}
+
+// CHECK: vgf2p8affineinvqb $7, 128(%rsp), %ymm20, %ymm1 {%k2}
+// CHECK: encoding: [0x62,0xf3,0xdd,0x22,0xcf,0x4c,0x24,0x04,0x07]
+ vgf2p8affineinvqb $7, 128(%rsp), %ymm20, %ymm1 {%k2}
+
+// CHECK: vgf2p8affineinvqb $7, 268435456(%rcx,%r14,8), %ymm20, %ymm1 {%k2}
+// CHECK: encoding: [0x62,0xb3,0xdd,0x22,0xcf,0x8c,0xf1,0x00,0x00,0x00,0x10,0x07]
+ vgf2p8affineinvqb $7, 268435456(%rcx,%r14,8), %ymm20, %ymm1 {%k2}
+
+// CHECK: vgf2p8affineinvqb $7, -536870912(%rcx,%r14,8), %ymm20, %ymm1 {%k2}
+// CHECK: encoding: [0x62,0xb3,0xdd,0x22,0xcf,0x8c,0xf1,0x00,0x00,0x00,0xe0,0x07]
+ vgf2p8affineinvqb $7, -536870912(%rcx,%r14,8), %ymm20, %ymm1 {%k2}
+
+// CHECK: vgf2p8affineinvqb $7, -536870910(%rcx,%r14,8), %ymm20, %ymm1 {%k2}
+// CHECK: encoding: [0x62,0xb3,0xdd,0x22,0xcf,0x8c,0xf1,0x02,0x00,0x00,0xe0,0x07]
+ vgf2p8affineinvqb $7, -536870910(%rcx,%r14,8), %ymm20, %ymm1 {%k2}
+
+// CHECK: vgf2p8affineqb $7, (%rcx), %ymm20, %ymm1 {%k2}
+// CHECK: encoding: [0x62,0xf3,0xdd,0x22,0xce,0x09,0x07]
+ vgf2p8affineqb $7, (%rcx), %ymm20, %ymm1 {%k2}
+
+// CHECK: vgf2p8affineqb $7, -128(%rsp), %ymm20, %ymm1 {%k2}
+// CHECK: encoding: [0x62,0xf3,0xdd,0x22,0xce,0x4c,0x24,0xfc,0x07]
+ vgf2p8affineqb $7, -128(%rsp), %ymm20, %ymm1 {%k2}
+
+// CHECK: vgf2p8affineqb $7, 128(%rsp), %ymm20, %ymm1 {%k2}
+// CHECK: encoding: [0x62,0xf3,0xdd,0x22,0xce,0x4c,0x24,0x04,0x07]
+ vgf2p8affineqb $7, 128(%rsp), %ymm20, %ymm1 {%k2}
+
+// CHECK: vgf2p8affineqb $7, 268435456(%rcx,%r14,8), %ymm20, %ymm1 {%k2}
+// CHECK: encoding: [0x62,0xb3,0xdd,0x22,0xce,0x8c,0xf1,0x00,0x00,0x00,0x10,0x07]
+ vgf2p8affineqb $7, 268435456(%rcx,%r14,8), %ymm20, %ymm1 {%k2}
+
+// CHECK: vgf2p8affineqb $7, -536870912(%rcx,%r14,8), %ymm20, %ymm1 {%k2}
+// CHECK: encoding: [0x62,0xb3,0xdd,0x22,0xce,0x8c,0xf1,0x00,0x00,0x00,0xe0,0x07]
+ vgf2p8affineqb $7, -536870912(%rcx,%r14,8), %ymm20, %ymm1 {%k2}
+
+// CHECK: vgf2p8affineqb $7, -536870910(%rcx,%r14,8), %ymm20, %ymm1 {%k2}
+// CHECK: encoding: [0x62,0xb3,0xdd,0x22,0xce,0x8c,0xf1,0x02,0x00,0x00,0xe0,0x07]
+ vgf2p8affineqb $7, -536870910(%rcx,%r14,8), %ymm20, %ymm1 {%k2}
+
+// CHECK: vgf2p8mulb %xmm2, %xmm20, %xmm1
+// CHECK: encoding: [0x62,0xf2,0x5d,0x00,0xcf,0xca]
+ vgf2p8mulb %xmm2, %xmm20, %xmm1
+
+// CHECK: vgf2p8mulb %xmm2, %xmm20, %xmm1 {%k2}
+// CHECK: encoding: [0x62,0xf2,0x5d,0x02,0xcf,0xca]
+ vgf2p8mulb %xmm2, %xmm20, %xmm1 {%k2}
+
+// CHECK: vgf2p8mulb (%rcx), %xmm20, %xmm1
+// CHECK: encoding: [0x62,0xf2,0x5d,0x00,0xcf,0x09]
+ vgf2p8mulb (%rcx), %xmm20, %xmm1
+
+// CHECK: vgf2p8mulb -64(%rsp), %xmm20, %xmm1
+// CHECK: encoding: [0x62,0xf2,0x5d,0x00,0xcf,0x4c,0x24,0xfc]
+ vgf2p8mulb -64(%rsp), %xmm20, %xmm1
+
+// CHECK: vgf2p8mulb 64(%rsp), %xmm20, %xmm1
+// CHECK: encoding: [0x62,0xf2,0x5d,0x00,0xcf,0x4c,0x24,0x04]
+ vgf2p8mulb 64(%rsp), %xmm20, %xmm1
+
+// CHECK: vgf2p8mulb 268435456(%rcx,%r14,8), %xmm20, %xmm1
+// CHECK: encoding: [0x62,0xb2,0x5d,0x00,0xcf,0x8c,0xf1,0x00,0x00,0x00,0x10]
+ vgf2p8mulb 268435456(%rcx,%r14,8), %xmm20, %xmm1
+
+// CHECK: vgf2p8mulb -536870912(%rcx,%r14,8), %xmm20, %xmm1
+// CHECK: encoding: [0x62,0xb2,0x5d,0x00,0xcf,0x8c,0xf1,0x00,0x00,0x00,0xe0]
+ vgf2p8mulb -536870912(%rcx,%r14,8), %xmm20, %xmm1
+
+// CHECK: vgf2p8mulb -536870910(%rcx,%r14,8), %xmm20, %xmm1
+// CHECK: encoding: [0x62,0xb2,0x5d,0x00,0xcf,0x8c,0xf1,0x02,0x00,0x00,0xe0]
+ vgf2p8mulb -536870910(%rcx,%r14,8), %xmm20, %xmm1
+
+// CHECK: vgf2p8mulb (%rcx), %xmm20, %xmm1 {%k2}
+// CHECK: encoding: [0x62,0xf2,0x5d,0x02,0xcf,0x09]
+ vgf2p8mulb (%rcx), %xmm20, %xmm1 {%k2}
+
+// CHECK: vgf2p8mulb -64(%rsp), %xmm20, %xmm1 {%k2}
+// CHECK: encoding: [0x62,0xf2,0x5d,0x02,0xcf,0x4c,0x24,0xfc]
+ vgf2p8mulb -64(%rsp), %xmm20, %xmm1 {%k2}
+
+// CHECK: vgf2p8mulb 64(%rsp), %xmm20, %xmm1 {%k2}
+// CHECK: encoding: [0x62,0xf2,0x5d,0x02,0xcf,0x4c,0x24,0x04]
+ vgf2p8mulb 64(%rsp), %xmm20, %xmm1 {%k2}
+
+// CHECK: vgf2p8mulb 268435456(%rcx,%r14,8), %xmm20, %xmm1 {%k2}
+// CHECK: encoding: [0x62,0xb2,0x5d,0x02,0xcf,0x8c,0xf1,0x00,0x00,0x00,0x10]
+ vgf2p8mulb 268435456(%rcx,%r14,8), %xmm20, %xmm1 {%k2}
+
+// CHECK: vgf2p8mulb -536870912(%rcx,%r14,8), %xmm20, %xmm1 {%k2}
+// CHECK: encoding: [0x62,0xb2,0x5d,0x02,0xcf,0x8c,0xf1,0x00,0x00,0x00,0xe0]
+ vgf2p8mulb -536870912(%rcx,%r14,8), %xmm20, %xmm1 {%k2}
+
+// CHECK: vgf2p8mulb -536870910(%rcx,%r14,8), %xmm20, %xmm1 {%k2}
+// CHECK: encoding: [0x62,0xb2,0x5d,0x02,0xcf,0x8c,0xf1,0x02,0x00,0x00,0xe0]
+ vgf2p8mulb -536870910(%rcx,%r14,8), %xmm20, %xmm1 {%k2}
+
+// CHECK: vgf2p8mulb %ymm2, %ymm20, %ymm1
+// CHECK: encoding: [0x62,0xf2,0x5d,0x20,0xcf,0xca]
+ vgf2p8mulb %ymm2, %ymm20, %ymm1
+
+// CHECK: vgf2p8mulb %ymm2, %ymm20, %ymm1 {%k2}
+// CHECK: encoding: [0x62,0xf2,0x5d,0x22,0xcf,0xca]
+ vgf2p8mulb %ymm2, %ymm20, %ymm1 {%k2}
+
+// CHECK: vgf2p8mulb (%rcx), %ymm20, %ymm1
+// CHECK: encoding: [0x62,0xf2,0x5d,0x20,0xcf,0x09]
+ vgf2p8mulb (%rcx), %ymm20, %ymm1
+
+// CHECK: vgf2p8mulb -128(%rsp), %ymm20, %ymm1
+// CHECK: encoding: [0x62,0xf2,0x5d,0x20,0xcf,0x4c,0x24,0xfc]
+ vgf2p8mulb -128(%rsp), %ymm20, %ymm1
+
+// CHECK: vgf2p8mulb 128(%rsp), %ymm20, %ymm1
+// CHECK: encoding: [0x62,0xf2,0x5d,0x20,0xcf,0x4c,0x24,0x04]
+ vgf2p8mulb 128(%rsp), %ymm20, %ymm1
+
+// CHECK: vgf2p8mulb 268435456(%rcx,%r14,8), %ymm20, %ymm1
+// CHECK: encoding: [0x62,0xb2,0x5d,0x20,0xcf,0x8c,0xf1,0x00,0x00,0x00,0x10]
+ vgf2p8mulb 268435456(%rcx,%r14,8), %ymm20, %ymm1
+
+// CHECK: vgf2p8mulb -536870912(%rcx,%r14,8), %ymm20, %ymm1
+// CHECK: encoding: [0x62,0xb2,0x5d,0x20,0xcf,0x8c,0xf1,0x00,0x00,0x00,0xe0]
+ vgf2p8mulb -536870912(%rcx,%r14,8), %ymm20, %ymm1
+
+// CHECK: vgf2p8mulb -536870910(%rcx,%r14,8), %ymm20, %ymm1
+// CHECK: encoding: [0x62,0xb2,0x5d,0x20,0xcf,0x8c,0xf1,0x02,0x00,0x00,0xe0]
+ vgf2p8mulb -536870910(%rcx,%r14,8), %ymm20, %ymm1
+
+// CHECK: vgf2p8mulb (%rcx), %ymm20, %ymm1 {%k2}
+// CHECK: encoding: [0x62,0xf2,0x5d,0x22,0xcf,0x09]
+ vgf2p8mulb (%rcx), %ymm20, %ymm1 {%k2}
+
+// CHECK: vgf2p8mulb -128(%rsp), %ymm20, %ymm1 {%k2}
+// CHECK: encoding: [0x62,0xf2,0x5d,0x22,0xcf,0x4c,0x24,0xfc]
+ vgf2p8mulb -128(%rsp), %ymm20, %ymm1 {%k2}
+
+// CHECK: vgf2p8mulb 128(%rsp), %ymm20, %ymm1 {%k2}
+// CHECK: encoding: [0x62,0xf2,0x5d,0x22,0xcf,0x4c,0x24,0x04]
+ vgf2p8mulb 128(%rsp), %ymm20, %ymm1 {%k2}
+
+// CHECK: vgf2p8mulb 268435456(%rcx,%r14,8), %ymm20, %ymm1 {%k2}
+// CHECK: encoding: [0x62,0xb2,0x5d,0x22,0xcf,0x8c,0xf1,0x00,0x00,0x00,0x10]
+ vgf2p8mulb 268435456(%rcx,%r14,8), %ymm20, %ymm1 {%k2}
+
+// CHECK: vgf2p8mulb -536870912(%rcx,%r14,8), %ymm20, %ymm1 {%k2}
+// CHECK: encoding: [0x62,0xb2,0x5d,0x22,0xcf,0x8c,0xf1,0x00,0x00,0x00,0xe0]
+ vgf2p8mulb -536870912(%rcx,%r14,8), %ymm20, %ymm1 {%k2}
+
+// CHECK: vgf2p8mulb -536870910(%rcx,%r14,8), %ymm20, %ymm1 {%k2}
+// CHECK: encoding: [0x62,0xb2,0x5d,0x22,0xcf,0x8c,0xf1,0x02,0x00,0x00,0xe0]
+ vgf2p8mulb -536870910(%rcx,%r14,8), %ymm20, %ymm1 {%k2}
+
+// CHECK: vgf2p8affineinvqb $7, (%rcx){1to2}, %xmm20, %xmm1
+// CHECK: encoding: [0x62,0xf3,0xdd,0x10,0xcf,0x09,0x07]
+ vgf2p8affineinvqb $7, (%rcx){1to2}, %xmm20, %xmm1
+
+// CHECK: vgf2p8affineinvqb $7, (%rcx){1to4}, %ymm20, %ymm1
+// CHECK: encoding: [0x62,0xf3,0xdd,0x30,0xcf,0x09,0x07]
+ vgf2p8affineinvqb $7, (%rcx){1to4}, %ymm20, %ymm1
+
+// CHECK: vgf2p8affineqb $7, (%rcx){1to2}, %xmm20, %xmm1
+// CHECK: encoding: [0x62,0xf3,0xdd,0x10,0xce,0x09,0x07]
+ vgf2p8affineqb $7, (%rcx){1to2}, %xmm20, %xmm1
+
+// CHECK: vgf2p8affineqb $7, (%rcx){1to4}, %ymm20, %ymm1
+// CHECK: encoding: [0x62,0xf3,0xdd,0x30,0xce,0x09,0x07]
+ vgf2p8affineqb $7, (%rcx){1to4}, %ymm20, %ymm1
+
Added: llvm/trunk/test/MC/X86/gfni-encoding.s
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/MC/X86/gfni-encoding.s?rev=318993&view=auto
==============================================================================
--- llvm/trunk/test/MC/X86/gfni-encoding.s (added)
+++ llvm/trunk/test/MC/X86/gfni-encoding.s Sun Nov 26 01:36:41 2017
@@ -0,0 +1,254 @@
+// RUN: llvm-mc -triple x86_64-unknown-unknown -mattr=+gfni --show-encoding < %s | FileCheck %s
+
+// CHECK: gf2p8affineinvqb $7, %xmm2, %xmm1
+// CHECK: encoding: [0x66,0x0f,0x3a,0xcf,0xca,0x07]
+ gf2p8affineinvqb $7, %xmm2, %xmm1
+
+// CHECK: gf2p8affineqb $7, %xmm2, %xmm1
+// CHECK: encoding: [0x66,0x0f,0x3a,0xce,0xca,0x07]
+ gf2p8affineqb $7, %xmm2, %xmm1
+
+// CHECK: gf2p8affineinvqb $7, (%rcx), %xmm1
+// CHECK: encoding: [0x66,0x0f,0x3a,0xcf,0x09,0x07]
+ gf2p8affineinvqb $7, (%rcx), %xmm1
+
+// CHECK: gf2p8affineinvqb $7, -4(%rsp), %xmm1
+// CHECK: encoding: [0x66,0x0f,0x3a,0xcf,0x4c,0x24,0xfc,0x07]
+ gf2p8affineinvqb $7, -4(%rsp), %xmm1
+
+// CHECK: gf2p8affineinvqb $7, 4(%rsp), %xmm1
+// CHECK: encoding: [0x66,0x0f,0x3a,0xcf,0x4c,0x24,0x04,0x07]
+ gf2p8affineinvqb $7, 4(%rsp), %xmm1
+
+// CHECK: gf2p8affineinvqb $7, 268435456(%rcx,%r14,8), %xmm1
+// CHECK: encoding: [0x66,0x42,0x0f,0x3a,0xcf,0x8c,0xf1,0x00,0x00,0x00,0x10,0x07]
+ gf2p8affineinvqb $7, 268435456(%rcx,%r14,8), %xmm1
+
+// CHECK: gf2p8affineinvqb $7, -536870912(%rcx,%r14,8), %xmm1
+// CHECK: encoding: [0x66,0x42,0x0f,0x3a,0xcf,0x8c,0xf1,0x00,0x00,0x00,0xe0,0x07]
+ gf2p8affineinvqb $7, -536870912(%rcx,%r14,8), %xmm1
+
+// CHECK: gf2p8affineinvqb $7, -536870910(%rcx,%r14,8), %xmm1
+// CHECK: encoding: [0x66,0x42,0x0f,0x3a,0xcf,0x8c,0xf1,0x02,0x00,0x00,0xe0,0x07]
+ gf2p8affineinvqb $7, -536870910(%rcx,%r14,8), %xmm1
+
+// CHECK: gf2p8affineqb $7, (%rcx), %xmm1
+// CHECK: encoding: [0x66,0x0f,0x3a,0xce,0x09,0x07]
+ gf2p8affineqb $7, (%rcx), %xmm1
+
+// CHECK: gf2p8affineqb $7, -4(%rsp), %xmm1
+// CHECK: encoding: [0x66,0x0f,0x3a,0xce,0x4c,0x24,0xfc,0x07]
+ gf2p8affineqb $7, -4(%rsp), %xmm1
+
+// CHECK: gf2p8affineqb $7, 4(%rsp), %xmm1
+// CHECK: encoding: [0x66,0x0f,0x3a,0xce,0x4c,0x24,0x04,0x07]
+ gf2p8affineqb $7, 4(%rsp), %xmm1
+
+// CHECK: gf2p8affineqb $7, 268435456(%rcx,%r14,8), %xmm1
+// CHECK: encoding: [0x66,0x42,0x0f,0x3a,0xce,0x8c,0xf1,0x00,0x00,0x00,0x10,0x07]
+ gf2p8affineqb $7, 268435456(%rcx,%r14,8), %xmm1
+
+// CHECK: gf2p8affineqb $7, -536870912(%rcx,%r14,8), %xmm1
+// CHECK: encoding: [0x66,0x42,0x0f,0x3a,0xce,0x8c,0xf1,0x00,0x00,0x00,0xe0,0x07]
+ gf2p8affineqb $7, -536870912(%rcx,%r14,8), %xmm1
+
+// CHECK: gf2p8affineqb $7, -536870910(%rcx,%r14,8), %xmm1
+// CHECK: encoding: [0x66,0x42,0x0f,0x3a,0xce,0x8c,0xf1,0x02,0x00,0x00,0xe0,0x07]
+ gf2p8affineqb $7, -536870910(%rcx,%r14,8), %xmm1
+
+// CHECK: gf2p8mulb %xmm2, %xmm1
+// CHECK: encoding: [0x66,0x0f,0x38,0xcf,0xca]
+ gf2p8mulb %xmm2, %xmm1
+
+// CHECK: gf2p8mulb (%rcx), %xmm1
+// CHECK: encoding: [0x66,0x0f,0x38,0xcf,0x09]
+ gf2p8mulb (%rcx), %xmm1
+
+// CHECK: gf2p8mulb -4(%rsp), %xmm1
+// CHECK: encoding: [0x66,0x0f,0x38,0xcf,0x4c,0x24,0xfc]
+ gf2p8mulb -4(%rsp), %xmm1
+
+// CHECK: gf2p8mulb 4(%rsp), %xmm1
+// CHECK: encoding: [0x66,0x0f,0x38,0xcf,0x4c,0x24,0x04]
+ gf2p8mulb 4(%rsp), %xmm1
+
+// CHECK: gf2p8mulb 268435456(%rcx,%r14,8), %xmm1
+// CHECK: encoding: [0x66,0x42,0x0f,0x38,0xcf,0x8c,0xf1,0x00,0x00,0x00,0x10]
+ gf2p8mulb 268435456(%rcx,%r14,8), %xmm1
+
+// CHECK: gf2p8mulb -536870912(%rcx,%r14,8), %xmm1
+// CHECK: encoding: [0x66,0x42,0x0f,0x38,0xcf,0x8c,0xf1,0x00,0x00,0x00,0xe0]
+ gf2p8mulb -536870912(%rcx,%r14,8), %xmm1
+
+// CHECK: gf2p8mulb -536870910(%rcx,%r14,8), %xmm1
+// CHECK: encoding: [0x66,0x42,0x0f,0x38,0xcf,0x8c,0xf1,0x02,0x00,0x00,0xe0]
+ gf2p8mulb -536870910(%rcx,%r14,8), %xmm1
+
+// CHECK: vgf2p8affineinvqb $7, %xmm2, %xmm10, %xmm1
+// CHECK: encoding: [0xc4,0xe3,0xa9,0xcf,0xca,0x07]
+ vgf2p8affineinvqb $7, %xmm2, %xmm10, %xmm1
+
+// CHECK: vgf2p8affineqb $7, %xmm2, %xmm10, %xmm1
+// CHECK: encoding: [0xc4,0xe3,0xa9,0xce,0xca,0x07]
+ vgf2p8affineqb $7, %xmm2, %xmm10, %xmm1
+
+// CHECK: vgf2p8affineinvqb $7, (%rcx), %xmm10, %xmm1
+// CHECK: encoding: [0xc4,0xe3,0xa9,0xcf,0x09,0x07]
+ vgf2p8affineinvqb $7, (%rcx), %xmm10, %xmm1
+
+// CHECK: vgf2p8affineinvqb $7, -4(%rsp), %xmm10, %xmm1
+// CHECK: encoding: [0xc4,0xe3,0xa9,0xcf,0x4c,0x24,0xfc,0x07]
+ vgf2p8affineinvqb $7, -4(%rsp), %xmm10, %xmm1
+
+// CHECK: vgf2p8affineinvqb $7, 4(%rsp), %xmm10, %xmm1
+// CHECK: encoding: [0xc4,0xe3,0xa9,0xcf,0x4c,0x24,0x04,0x07]
+ vgf2p8affineinvqb $7, 4(%rsp), %xmm10, %xmm1
+
+// CHECK: vgf2p8affineinvqb $7, 268435456(%rcx,%r14,8), %xmm10, %xmm1
+// CHECK: encoding: [0xc4,0xa3,0xa9,0xcf,0x8c,0xf1,0x00,0x00,0x00,0x10,0x07]
+ vgf2p8affineinvqb $7, 268435456(%rcx,%r14,8), %xmm10, %xmm1
+
+// CHECK: vgf2p8affineinvqb $7, -536870912(%rcx,%r14,8), %xmm10, %xmm1
+// CHECK: encoding: [0xc4,0xa3,0xa9,0xcf,0x8c,0xf1,0x00,0x00,0x00,0xe0,0x07]
+ vgf2p8affineinvqb $7, -536870912(%rcx,%r14,8), %xmm10, %xmm1
+
+// CHECK: vgf2p8affineinvqb $7, -536870910(%rcx,%r14,8), %xmm10, %xmm1
+// CHECK: encoding: [0xc4,0xa3,0xa9,0xcf,0x8c,0xf1,0x02,0x00,0x00,0xe0,0x07]
+ vgf2p8affineinvqb $7, -536870910(%rcx,%r14,8), %xmm10, %xmm1
+
+// CHECK: vgf2p8affineqb $7, (%rcx), %xmm10, %xmm1
+// CHECK: encoding: [0xc4,0xe3,0xa9,0xce,0x09,0x07]
+ vgf2p8affineqb $7, (%rcx), %xmm10, %xmm1
+
+// CHECK: vgf2p8affineqb $7, -4(%rsp), %xmm10, %xmm1
+// CHECK: encoding: [0xc4,0xe3,0xa9,0xce,0x4c,0x24,0xfc,0x07]
+ vgf2p8affineqb $7, -4(%rsp), %xmm10, %xmm1
+
+// CHECK: vgf2p8affineqb $7, 4(%rsp), %xmm10, %xmm1
+// CHECK: encoding: [0xc4,0xe3,0xa9,0xce,0x4c,0x24,0x04,0x07]
+ vgf2p8affineqb $7, 4(%rsp), %xmm10, %xmm1
+
+// CHECK: vgf2p8affineqb $7, 268435456(%rcx,%r14,8), %xmm10, %xmm1
+// CHECK: encoding: [0xc4,0xa3,0xa9,0xce,0x8c,0xf1,0x00,0x00,0x00,0x10,0x07]
+ vgf2p8affineqb $7, 268435456(%rcx,%r14,8), %xmm10, %xmm1
+
+// CHECK: vgf2p8affineqb $7, -536870912(%rcx,%r14,8), %xmm10, %xmm1
+// CHECK: encoding: [0xc4,0xa3,0xa9,0xce,0x8c,0xf1,0x00,0x00,0x00,0xe0,0x07]
+ vgf2p8affineqb $7, -536870912(%rcx,%r14,8), %xmm10, %xmm1
+
+// CHECK: vgf2p8affineqb $7, -536870910(%rcx,%r14,8), %xmm10, %xmm1
+// CHECK: encoding: [0xc4,0xa3,0xa9,0xce,0x8c,0xf1,0x02,0x00,0x00,0xe0,0x07]
+ vgf2p8affineqb $7, -536870910(%rcx,%r14,8), %xmm10, %xmm1
+
+// CHECK: vgf2p8affineinvqb $7, %ymm2, %ymm10, %ymm1
+// CHECK: encoding: [0xc4,0xe3,0xad,0xcf,0xca,0x07]
+ vgf2p8affineinvqb $7, %ymm2, %ymm10, %ymm1
+
+// CHECK: vgf2p8affineqb $7, %ymm2, %ymm10, %ymm1
+// CHECK: encoding: [0xc4,0xe3,0xad,0xce,0xca,0x07]
+ vgf2p8affineqb $7, %ymm2, %ymm10, %ymm1
+
+// CHECK: vgf2p8affineinvqb $7, (%rcx), %ymm10, %ymm1
+// CHECK: encoding: [0xc4,0xe3,0xad,0xcf,0x09,0x07]
+ vgf2p8affineinvqb $7, (%rcx), %ymm10, %ymm1
+
+// CHECK: vgf2p8affineinvqb $7, -4(%rsp), %ymm10, %ymm1
+// CHECK: encoding: [0xc4,0xe3,0xad,0xcf,0x4c,0x24,0xfc,0x07]
+ vgf2p8affineinvqb $7, -4(%rsp), %ymm10, %ymm1
+
+// CHECK: vgf2p8affineinvqb $7, 4(%rsp), %ymm10, %ymm1
+// CHECK: encoding: [0xc4,0xe3,0xad,0xcf,0x4c,0x24,0x04,0x07]
+ vgf2p8affineinvqb $7, 4(%rsp), %ymm10, %ymm1
+
+// CHECK: vgf2p8affineinvqb $7, 268435456(%rcx,%r14,8), %ymm10, %ymm1
+// CHECK: encoding: [0xc4,0xa3,0xad,0xcf,0x8c,0xf1,0x00,0x00,0x00,0x10,0x07]
+ vgf2p8affineinvqb $7, 268435456(%rcx,%r14,8), %ymm10, %ymm1
+
+// CHECK: vgf2p8affineinvqb $7, -536870912(%rcx,%r14,8), %ymm10, %ymm1
+// CHECK: encoding: [0xc4,0xa3,0xad,0xcf,0x8c,0xf1,0x00,0x00,0x00,0xe0,0x07]
+ vgf2p8affineinvqb $7, -536870912(%rcx,%r14,8), %ymm10, %ymm1
+
+// CHECK: vgf2p8affineinvqb $7, -536870910(%rcx,%r14,8), %ymm10, %ymm1
+// CHECK: encoding: [0xc4,0xa3,0xad,0xcf,0x8c,0xf1,0x02,0x00,0x00,0xe0,0x07]
+ vgf2p8affineinvqb $7, -536870910(%rcx,%r14,8), %ymm10, %ymm1
+
+// CHECK: vgf2p8affineqb $7, (%rcx), %ymm10, %ymm1
+// CHECK: encoding: [0xc4,0xe3,0xad,0xce,0x09,0x07]
+ vgf2p8affineqb $7, (%rcx), %ymm10, %ymm1
+
+// CHECK: vgf2p8affineqb $7, -4(%rsp), %ymm10, %ymm1
+// CHECK: encoding: [0xc4,0xe3,0xad,0xce,0x4c,0x24,0xfc,0x07]
+ vgf2p8affineqb $7, -4(%rsp), %ymm10, %ymm1
+
+// CHECK: vgf2p8affineqb $7, 4(%rsp), %ymm10, %ymm1
+// CHECK: encoding: [0xc4,0xe3,0xad,0xce,0x4c,0x24,0x04,0x07]
+ vgf2p8affineqb $7, 4(%rsp), %ymm10, %ymm1
+
+// CHECK: vgf2p8affineqb $7, 268435456(%rcx,%r14,8), %ymm10, %ymm1
+// CHECK: encoding: [0xc4,0xa3,0xad,0xce,0x8c,0xf1,0x00,0x00,0x00,0x10,0x07]
+ vgf2p8affineqb $7, 268435456(%rcx,%r14,8), %ymm10, %ymm1
+
+// CHECK: vgf2p8affineqb $7, -536870912(%rcx,%r14,8), %ymm10, %ymm1
+// CHECK: encoding: [0xc4,0xa3,0xad,0xce,0x8c,0xf1,0x00,0x00,0x00,0xe0,0x07]
+ vgf2p8affineqb $7, -536870912(%rcx,%r14,8), %ymm10, %ymm1
+
+// CHECK: vgf2p8affineqb $7, -536870910(%rcx,%r14,8), %ymm10, %ymm1
+// CHECK: encoding: [0xc4,0xa3,0xad,0xce,0x8c,0xf1,0x02,0x00,0x00,0xe0,0x07]
+ vgf2p8affineqb $7, -536870910(%rcx,%r14,8), %ymm10, %ymm1
+
+// CHECK: vgf2p8mulb %xmm2, %xmm10, %xmm1
+// CHECK: encoding: [0xc4,0xe2,0x29,0xcf,0xca]
+ vgf2p8mulb %xmm2, %xmm10, %xmm1
+
+// CHECK: vgf2p8mulb (%rcx), %xmm10, %xmm1
+// CHECK: encoding: [0xc4,0xe2,0x29,0xcf,0x09]
+ vgf2p8mulb (%rcx), %xmm10, %xmm1
+
+// CHECK: vgf2p8mulb -4(%rsp), %xmm10, %xmm1
+// CHECK: encoding: [0xc4,0xe2,0x29,0xcf,0x4c,0x24,0xfc]
+ vgf2p8mulb -4(%rsp), %xmm10, %xmm1
+
+// CHECK: vgf2p8mulb 4(%rsp), %xmm10, %xmm1
+// CHECK: encoding: [0xc4,0xe2,0x29,0xcf,0x4c,0x24,0x04]
+ vgf2p8mulb 4(%rsp), %xmm10, %xmm1
+
+// CHECK: vgf2p8mulb 268435456(%rcx,%r14,8), %xmm10, %xmm1
+// CHECK: encoding: [0xc4,0xa2,0x29,0xcf,0x8c,0xf1,0x00,0x00,0x00,0x10]
+ vgf2p8mulb 268435456(%rcx,%r14,8), %xmm10, %xmm1
+
+// CHECK: vgf2p8mulb -536870912(%rcx,%r14,8), %xmm10, %xmm1
+// CHECK: encoding: [0xc4,0xa2,0x29,0xcf,0x8c,0xf1,0x00,0x00,0x00,0xe0]
+ vgf2p8mulb -536870912(%rcx,%r14,8), %xmm10, %xmm1
+
+// CHECK: vgf2p8mulb -536870910(%rcx,%r14,8), %xmm10, %xmm1
+// CHECK: encoding: [0xc4,0xa2,0x29,0xcf,0x8c,0xf1,0x02,0x00,0x00,0xe0]
+ vgf2p8mulb -536870910(%rcx,%r14,8), %xmm10, %xmm1
+
+// CHECK: vgf2p8mulb %ymm2, %ymm10, %ymm1
+// CHECK: encoding: [0xc4,0xe2,0x2d,0xcf,0xca]
+ vgf2p8mulb %ymm2, %ymm10, %ymm1
+
+// CHECK: vgf2p8mulb (%rcx), %ymm10, %ymm1
+// CHECK: encoding: [0xc4,0xe2,0x2d,0xcf,0x09]
+ vgf2p8mulb (%rcx), %ymm10, %ymm1
+
+// CHECK: vgf2p8mulb -4(%rsp), %ymm10, %ymm1
+// CHECK: encoding: [0xc4,0xe2,0x2d,0xcf,0x4c,0x24,0xfc]
+ vgf2p8mulb -4(%rsp), %ymm10, %ymm1
+
+// CHECK: vgf2p8mulb 4(%rsp), %ymm10, %ymm1
+// CHECK: encoding: [0xc4,0xe2,0x2d,0xcf,0x4c,0x24,0x04]
+ vgf2p8mulb 4(%rsp), %ymm10, %ymm1
+
+// CHECK: vgf2p8mulb 268435456(%rcx,%r14,8), %ymm10, %ymm1
+// CHECK: encoding: [0xc4,0xa2,0x2d,0xcf,0x8c,0xf1,0x00,0x00,0x00,0x10]
+ vgf2p8mulb 268435456(%rcx,%r14,8), %ymm10, %ymm1
+
+// CHECK: vgf2p8mulb -536870912(%rcx,%r14,8), %ymm10, %ymm1
+// CHECK: encoding: [0xc4,0xa2,0x2d,0xcf,0x8c,0xf1,0x00,0x00,0x00,0xe0]
+ vgf2p8mulb -536870912(%rcx,%r14,8), %ymm10, %ymm1
+
+// CHECK: vgf2p8mulb -536870910(%rcx,%r14,8), %ymm10, %ymm1
+// CHECK: encoding: [0xc4,0xa2,0x2d,0xcf,0x8c,0xf1,0x02,0x00,0x00,0xe0]
+ vgf2p8mulb -536870910(%rcx,%r14,8), %ymm10, %ymm1
+
More information about the llvm-commits
mailing list