[llvm] 2718a47 - [LoongArch] Lower vector select mask generation to `[X]VMSK{LT,GE,NE}Z` if possible (#142109)
via llvm-commits
llvm-commits at lists.llvm.org
Thu Jun 5 07:17:42 PDT 2025
Author: hev
Date: 2025-06-05T22:17:38+08:00
New Revision: 2718a47f4952f91f3f076242b8b6cad4b7934e9d
URL: https://github.com/llvm/llvm-project/commit/2718a47f4952f91f3f076242b8b6cad4b7934e9d
DIFF: https://github.com/llvm/llvm-project/commit/2718a47f4952f91f3f076242b8b6cad4b7934e9d.diff
LOG: [LoongArch] Lower vector select mask generation to `[X]VMSK{LT,GE,NE}Z` if possible (#142109)
This patch adds a DAG combine rule for BITCAST nodes converting from
vector `i1` masks generated by `setcc` into integer vector types. It
recognizes common select mask patterns and lowers them into efficient
LoongArch LSX/LASX mask instructions such as:
- [X]VMSKLTZ.{B,H,W,D}
- [X]VMSKGEZ.B
- [X]VMSKNEZ.B
When the vector comparison matches specific patterns (e.g., x < 0, x >=
0, x != 0, etc.), the transformation is performed pre-legalization. This
avoids scalarization and unnecessary operations, improving both
performance and code size.
Added:
Modified:
llvm/lib/Target/LoongArch/LoongArchISelLowering.cpp
llvm/lib/Target/LoongArch/LoongArchISelLowering.h
llvm/lib/Target/LoongArch/LoongArchLASXInstrInfo.td
llvm/lib/Target/LoongArch/LoongArchLSXInstrInfo.td
llvm/test/CodeGen/LoongArch/lasx/xvmskcond.ll
llvm/test/CodeGen/LoongArch/lsx/vmskcond.ll
Removed:
################################################################################
diff --git a/llvm/lib/Target/LoongArch/LoongArchISelLowering.cpp b/llvm/lib/Target/LoongArch/LoongArchISelLowering.cpp
index eda71dde620b6..b0e23c297e204 100644
--- a/llvm/lib/Target/LoongArch/LoongArchISelLowering.cpp
+++ b/llvm/lib/Target/LoongArch/LoongArchISelLowering.cpp
@@ -391,8 +391,10 @@ LoongArchTargetLowering::LoongArchTargetLowering(const TargetMachine &TM,
// Set DAG combine for 'LSX' feature.
- if (Subtarget.hasExtLSX())
+ if (Subtarget.hasExtLSX()) {
setTargetDAGCombine(ISD::INTRINSIC_WO_CHAIN);
+ setTargetDAGCombine(ISD::BITCAST);
+ }
// Compute derived properties from the register classes.
computeRegisterProperties(Subtarget.getRegisterInfo());
@@ -4329,6 +4331,85 @@ static SDValue performSRLCombine(SDNode *N, SelectionDAG &DAG,
return SDValue();
}
+static SDValue performBITCASTCombine(SDNode *N, SelectionDAG &DAG,
+ TargetLowering::DAGCombinerInfo &DCI,
+ const LoongArchSubtarget &Subtarget) {
+ SDLoc DL(N);
+ EVT VT = N->getValueType(0);
+ SDValue Src = N->getOperand(0);
+ EVT SrcVT = Src.getValueType();
+
+ if (!DCI.isBeforeLegalizeOps())
+ return SDValue();
+
+ if (!SrcVT.isSimple() || SrcVT.getScalarType() != MVT::i1)
+ return SDValue();
+
+ unsigned Opc = ISD::DELETED_NODE;
+ // Combine SETCC and BITCAST into [X]VMSK{LT,GE,NE} when possible
+ if (Src.getOpcode() == ISD::SETCC && Src.hasOneUse()) {
+ bool UseLASX;
+ EVT CmpVT = Src.getOperand(0).getValueType();
+ EVT EltVT = CmpVT.getVectorElementType();
+
+ if (Subtarget.hasExtLSX() && CmpVT.getSizeInBits() <= 128)
+ UseLASX = false;
+ else if (Subtarget.has32S() && Subtarget.hasExtLASX() &&
+ CmpVT.getSizeInBits() <= 256)
+ UseLASX = true;
+ else
+ return SDValue();
+
+ SDValue SrcN1 = Src.getOperand(1);
+ switch (cast<CondCodeSDNode>(Src.getOperand(2))->get()) {
+ default:
+ break;
+ case ISD::SETEQ:
+ // x == 0 => not (vmsknez.b x)
+ if (ISD::isBuildVectorAllZeros(SrcN1.getNode()) && EltVT == MVT::i8)
+ Opc = UseLASX ? LoongArchISD::XVMSKEQZ : LoongArchISD::VMSKEQZ;
+ break;
+ case ISD::SETGT:
+ // x > -1 => vmskgez.b x
+ if (ISD::isBuildVectorAllOnes(SrcN1.getNode()) && EltVT == MVT::i8)
+ Opc = UseLASX ? LoongArchISD::XVMSKGEZ : LoongArchISD::VMSKGEZ;
+ break;
+ case ISD::SETGE:
+ // x >= 0 => vmskgez.b x
+ if (ISD::isBuildVectorAllZeros(SrcN1.getNode()) && EltVT == MVT::i8)
+ Opc = UseLASX ? LoongArchISD::XVMSKGEZ : LoongArchISD::VMSKGEZ;
+ break;
+ case ISD::SETLT:
+ // x < 0 => vmskltz.{b,h,w,d} x
+ if (ISD::isBuildVectorAllZeros(SrcN1.getNode()) &&
+ (EltVT == MVT::i8 || EltVT == MVT::i16 || EltVT == MVT::i32 ||
+ EltVT == MVT::i64))
+ Opc = UseLASX ? LoongArchISD::XVMSKLTZ : LoongArchISD::VMSKLTZ;
+ break;
+ case ISD::SETLE:
+ // x <= -1 => vmskltz.{b,h,w,d} x
+ if (ISD::isBuildVectorAllOnes(SrcN1.getNode()) &&
+ (EltVT == MVT::i8 || EltVT == MVT::i16 || EltVT == MVT::i32 ||
+ EltVT == MVT::i64))
+ Opc = UseLASX ? LoongArchISD::XVMSKLTZ : LoongArchISD::VMSKLTZ;
+ break;
+ case ISD::SETNE:
+ // x != 0 => vmsknez.b x
+ if (ISD::isBuildVectorAllZeros(SrcN1.getNode()) && EltVT == MVT::i8)
+ Opc = UseLASX ? LoongArchISD::XVMSKNEZ : LoongArchISD::VMSKNEZ;
+ break;
+ }
+ }
+
+ if (Opc == ISD::DELETED_NODE)
+ return SDValue();
+
+ SDValue V = DAG.getNode(Opc, DL, MVT::i64, Src.getOperand(0));
+ EVT T = EVT::getIntegerVT(*DAG.getContext(), SrcVT.getVectorNumElements());
+ V = DAG.getZExtOrTrunc(V, DL, T);
+ return DAG.getBitcast(VT, V);
+}
+
static SDValue performORCombine(SDNode *N, SelectionDAG &DAG,
TargetLowering::DAGCombinerInfo &DCI,
const LoongArchSubtarget &Subtarget) {
@@ -5373,6 +5454,8 @@ SDValue LoongArchTargetLowering::PerformDAGCombine(SDNode *N,
return performSETCCCombine(N, DAG, DCI, Subtarget);
case ISD::SRL:
return performSRLCombine(N, DAG, DCI, Subtarget);
+ case ISD::BITCAST:
+ return performBITCASTCombine(N, DAG, DCI, Subtarget);
case LoongArchISD::BITREV_W:
return performBITREV_WCombine(N, DAG, DCI, Subtarget);
case ISD::INTRINSIC_WO_CHAIN:
@@ -5663,6 +5746,120 @@ static MachineBasicBlock *emitPseudoCTPOP(MachineInstr &MI,
return BB;
}
+static MachineBasicBlock *
+emitPseudoVMSKCOND(MachineInstr &MI, MachineBasicBlock *BB,
+ const LoongArchSubtarget &Subtarget) {
+ const TargetInstrInfo *TII = Subtarget.getInstrInfo();
+ const TargetRegisterClass *RC = &LoongArch::LSX128RegClass;
+ const LoongArchRegisterInfo *TRI = Subtarget.getRegisterInfo();
+ MachineRegisterInfo &MRI = BB->getParent()->getRegInfo();
+ Register Dst = MI.getOperand(0).getReg();
+ Register Src = MI.getOperand(1).getReg();
+ DebugLoc DL = MI.getDebugLoc();
+ unsigned EleBits = 8;
+ unsigned NotOpc = 0;
+ unsigned MskOpc;
+
+ switch (MI.getOpcode()) {
+ default:
+ llvm_unreachable("Unexpected opcode");
+ case LoongArch::PseudoVMSKLTZ_B:
+ MskOpc = LoongArch::VMSKLTZ_B;
+ break;
+ case LoongArch::PseudoVMSKLTZ_H:
+ MskOpc = LoongArch::VMSKLTZ_H;
+ EleBits = 16;
+ break;
+ case LoongArch::PseudoVMSKLTZ_W:
+ MskOpc = LoongArch::VMSKLTZ_W;
+ EleBits = 32;
+ break;
+ case LoongArch::PseudoVMSKLTZ_D:
+ MskOpc = LoongArch::VMSKLTZ_D;
+ EleBits = 64;
+ break;
+ case LoongArch::PseudoVMSKGEZ_B:
+ MskOpc = LoongArch::VMSKGEZ_B;
+ break;
+ case LoongArch::PseudoVMSKEQZ_B:
+ MskOpc = LoongArch::VMSKNZ_B;
+ NotOpc = LoongArch::VNOR_V;
+ break;
+ case LoongArch::PseudoVMSKNEZ_B:
+ MskOpc = LoongArch::VMSKNZ_B;
+ break;
+ case LoongArch::PseudoXVMSKLTZ_B:
+ MskOpc = LoongArch::XVMSKLTZ_B;
+ RC = &LoongArch::LASX256RegClass;
+ break;
+ case LoongArch::PseudoXVMSKLTZ_H:
+ MskOpc = LoongArch::XVMSKLTZ_H;
+ RC = &LoongArch::LASX256RegClass;
+ EleBits = 16;
+ break;
+ case LoongArch::PseudoXVMSKLTZ_W:
+ MskOpc = LoongArch::XVMSKLTZ_W;
+ RC = &LoongArch::LASX256RegClass;
+ EleBits = 32;
+ break;
+ case LoongArch::PseudoXVMSKLTZ_D:
+ MskOpc = LoongArch::XVMSKLTZ_D;
+ RC = &LoongArch::LASX256RegClass;
+ EleBits = 64;
+ break;
+ case LoongArch::PseudoXVMSKGEZ_B:
+ MskOpc = LoongArch::XVMSKGEZ_B;
+ RC = &LoongArch::LASX256RegClass;
+ break;
+ case LoongArch::PseudoXVMSKEQZ_B:
+ MskOpc = LoongArch::XVMSKNZ_B;
+ NotOpc = LoongArch::XVNOR_V;
+ RC = &LoongArch::LASX256RegClass;
+ break;
+ case LoongArch::PseudoXVMSKNEZ_B:
+ MskOpc = LoongArch::XVMSKNZ_B;
+ RC = &LoongArch::LASX256RegClass;
+ break;
+ }
+
+ Register Msk = MRI.createVirtualRegister(RC);
+ if (NotOpc) {
+ Register Tmp = MRI.createVirtualRegister(RC);
+ BuildMI(*BB, MI, DL, TII->get(MskOpc), Tmp).addReg(Src);
+ BuildMI(*BB, MI, DL, TII->get(NotOpc), Msk)
+ .addReg(Tmp, RegState::Kill)
+ .addReg(Tmp, RegState::Kill);
+ } else {
+ BuildMI(*BB, MI, DL, TII->get(MskOpc), Msk).addReg(Src);
+ }
+
+ if (TRI->getRegSizeInBits(*RC) > 128) {
+ Register Lo = MRI.createVirtualRegister(&LoongArch::GPRRegClass);
+ Register Hi = MRI.createVirtualRegister(&LoongArch::GPRRegClass);
+ BuildMI(*BB, MI, DL, TII->get(LoongArch::XVPICKVE2GR_WU), Lo)
+ .addReg(Msk)
+ .addImm(0);
+ BuildMI(*BB, MI, DL, TII->get(LoongArch::XVPICKVE2GR_WU), Hi)
+ .addReg(Msk, RegState::Kill)
+ .addImm(4);
+ BuildMI(*BB, MI, DL,
+ TII->get(Subtarget.is64Bit() ? LoongArch::BSTRINS_D
+ : LoongArch::BSTRINS_W),
+ Dst)
+ .addReg(Lo, RegState::Kill)
+ .addReg(Hi, RegState::Kill)
+ .addImm(256 / EleBits - 1)
+ .addImm(128 / EleBits);
+ } else {
+ BuildMI(*BB, MI, DL, TII->get(LoongArch::VPICKVE2GR_HU), Dst)
+ .addReg(Msk, RegState::Kill)
+ .addImm(0);
+ }
+
+ MI.eraseFromParent();
+ return BB;
+}
+
static bool isSelectPseudo(MachineInstr &MI) {
switch (MI.getOpcode()) {
default:
@@ -5869,6 +6066,21 @@ MachineBasicBlock *LoongArchTargetLowering::EmitInstrWithCustomInserter(
return emitPseudoXVINSGR2VR(MI, BB, Subtarget);
case LoongArch::PseudoCTPOP:
return emitPseudoCTPOP(MI, BB, Subtarget);
+ case LoongArch::PseudoVMSKLTZ_B:
+ case LoongArch::PseudoVMSKLTZ_H:
+ case LoongArch::PseudoVMSKLTZ_W:
+ case LoongArch::PseudoVMSKLTZ_D:
+ case LoongArch::PseudoVMSKGEZ_B:
+ case LoongArch::PseudoVMSKEQZ_B:
+ case LoongArch::PseudoVMSKNEZ_B:
+ case LoongArch::PseudoXVMSKLTZ_B:
+ case LoongArch::PseudoXVMSKLTZ_H:
+ case LoongArch::PseudoXVMSKLTZ_W:
+ case LoongArch::PseudoXVMSKLTZ_D:
+ case LoongArch::PseudoXVMSKGEZ_B:
+ case LoongArch::PseudoXVMSKEQZ_B:
+ case LoongArch::PseudoXVMSKNEZ_B:
+ return emitPseudoVMSKCOND(MI, BB, Subtarget);
case TargetOpcode::STATEPOINT:
// STATEPOINT is a pseudo instruction which has no implicit defs/uses
// while bl call instruction (where statepoint will be lowered at the
@@ -5990,6 +6202,14 @@ const char *LoongArchTargetLowering::getTargetNodeName(unsigned Opcode) const {
NODE_NAME_CASE(VBSLL)
NODE_NAME_CASE(VBSRL)
NODE_NAME_CASE(VLDREPL)
+ NODE_NAME_CASE(VMSKLTZ)
+ NODE_NAME_CASE(VMSKGEZ)
+ NODE_NAME_CASE(VMSKEQZ)
+ NODE_NAME_CASE(VMSKNEZ)
+ NODE_NAME_CASE(XVMSKLTZ)
+ NODE_NAME_CASE(XVMSKGEZ)
+ NODE_NAME_CASE(XVMSKEQZ)
+ NODE_NAME_CASE(XVMSKNEZ)
}
#undef NODE_NAME_CASE
return nullptr;
diff --git a/llvm/lib/Target/LoongArch/LoongArchISelLowering.h b/llvm/lib/Target/LoongArch/LoongArchISelLowering.h
index 8c00ec75db94b..4b6d3272db2c9 100644
--- a/llvm/lib/Target/LoongArch/LoongArchISelLowering.h
+++ b/llvm/lib/Target/LoongArch/LoongArchISelLowering.h
@@ -161,7 +161,17 @@ enum NodeType : unsigned {
VBSRL,
// Scalar load broadcast to vector
- VLDREPL
+ VLDREPL,
+
+ // Vector mask set by condition
+ VMSKLTZ,
+ VMSKGEZ,
+ VMSKEQZ,
+ VMSKNEZ,
+ XVMSKLTZ,
+ XVMSKGEZ,
+ XVMSKEQZ,
+ XVMSKNEZ,
// Intrinsic operations end =============================================
};
diff --git a/llvm/lib/Target/LoongArch/LoongArchLASXInstrInfo.td b/llvm/lib/Target/LoongArch/LoongArchLASXInstrInfo.td
index 59dec76ef1c2d..ff7b0f2ae3f25 100644
--- a/llvm/lib/Target/LoongArch/LoongArchLASXInstrInfo.td
+++ b/llvm/lib/Target/LoongArch/LoongArchLASXInstrInfo.td
@@ -12,6 +12,10 @@
// Target nodes.
def loongarch_xvpermi: SDNode<"LoongArchISD::XVPERMI", SDT_LoongArchV1RUimm>;
+def loongarch_xvmskltz: SDNode<"LoongArchISD::XVMSKLTZ", SDT_LoongArchVMSKCOND>;
+def loongarch_xvmskgez: SDNode<"LoongArchISD::XVMSKGEZ", SDT_LoongArchVMSKCOND>;
+def loongarch_xvmskeqz: SDNode<"LoongArchISD::XVMSKEQZ", SDT_LoongArchVMSKCOND>;
+def loongarch_xvmsknez: SDNode<"LoongArchISD::XVMSKNEZ", SDT_LoongArchVMSKCOND>;
def lasxsplati8
: PatFrag<(ops node:$e0),
@@ -1086,6 +1090,16 @@ def PseudoXVINSGR2VR_H
: Pseudo<(outs LASX256:$dst), (ins LASX256:$xd, GPR:$rj, uimm4:$imm)>;
} // usesCustomInserter = 1, Constraints = "$xd = $dst"
+let usesCustomInserter = 1, hasSideEffects = 0, mayLoad = 0, mayStore = 0 in {
+def PseudoXVMSKLTZ_B : Pseudo<(outs GPR:$rd), (ins LASX256:$vj)>;
+def PseudoXVMSKLTZ_H : Pseudo<(outs GPR:$rd), (ins LASX256:$vj)>;
+def PseudoXVMSKLTZ_W : Pseudo<(outs GPR:$rd), (ins LASX256:$vj)>;
+def PseudoXVMSKLTZ_D : Pseudo<(outs GPR:$rd), (ins LASX256:$vj)>;
+def PseudoXVMSKGEZ_B : Pseudo<(outs GPR:$rd), (ins LASX256:$vj)>;
+def PseudoXVMSKEQZ_B : Pseudo<(outs GPR:$rd), (ins LASX256:$vj)>;
+def PseudoXVMSKNEZ_B : Pseudo<(outs GPR:$rd), (ins LASX256:$vj)>;
+} // usesCustomInserter = 1, hasSideEffects = 0, mayLoad = 0, mayStore = 0
+
} // Predicates = [HasExtLASX]
multiclass PatXr<SDPatternOperator OpNode, string Inst> {
@@ -1856,6 +1870,15 @@ def : Pat<(vt (concat_vectors LSX128:$vd, LSX128:$vj)),
defm : PatXrXr<abds, "XVABSD">;
defm : PatXrXrU<abdu, "XVABSD">;
+// Vector mask set by condition
+def : Pat<(loongarch_xvmskltz (v32i8 LASX256:$vj)), (PseudoXVMSKLTZ_B LASX256:$vj)>;
+def : Pat<(loongarch_xvmskltz (v16i16 LASX256:$vj)), (PseudoXVMSKLTZ_H LASX256:$vj)>;
+def : Pat<(loongarch_xvmskltz (v8i32 LASX256:$vj)), (PseudoXVMSKLTZ_W LASX256:$vj)>;
+def : Pat<(loongarch_xvmskltz (v4i64 LASX256:$vj)), (PseudoXVMSKLTZ_D LASX256:$vj)>;
+def : Pat<(loongarch_xvmskgez (v32i8 LASX256:$vj)), (PseudoXVMSKGEZ_B LASX256:$vj)>;
+def : Pat<(loongarch_xvmskeqz (v32i8 LASX256:$vj)), (PseudoXVMSKEQZ_B LASX256:$vj)>;
+def : Pat<(loongarch_xvmsknez (v32i8 LASX256:$vj)), (PseudoXVMSKNEZ_B LASX256:$vj)>;
+
} // Predicates = [HasExtLASX]
/// Intrinsic pattern
diff --git a/llvm/lib/Target/LoongArch/LoongArchLSXInstrInfo.td b/llvm/lib/Target/LoongArch/LoongArchLSXInstrInfo.td
index b2b3b65155265..d73d78083ddcd 100644
--- a/llvm/lib/Target/LoongArch/LoongArchLSXInstrInfo.td
+++ b/llvm/lib/Target/LoongArch/LoongArchLSXInstrInfo.td
@@ -31,6 +31,7 @@ def SDT_LoongArchVreplgr2vr : SDTypeProfile<1, 1, [SDTCisInt<0>, SDTCisVec<0>, S
def SDT_LoongArchVFRECIPE : SDTypeProfile<1, 1, [SDTCisFP<0>, SDTCisVec<0>, SDTCisSameAs<0, 1>]>;
def SDT_LoongArchVFRSQRTE : SDTypeProfile<1, 1, [SDTCisFP<0>, SDTCisVec<0>, SDTCisSameAs<0, 1>]>;
def SDT_LoongArchVLDREPL : SDTypeProfile<1, 1, [SDTCisVec<0>, SDTCisPtrTy<1>]>;
+def SDT_LoongArchVMSKCOND : SDTypeProfile<1, 1, [SDTCisInt<0>, SDTCisVec<1>]>;
// Target nodes.
def loongarch_vreplve : SDNode<"LoongArchISD::VREPLVE", SDT_LoongArchVreplve>;
@@ -74,6 +75,11 @@ def loongarch_vldrepl
: SDNode<"LoongArchISD::VLDREPL",
SDT_LoongArchVLDREPL, [SDNPHasChain, SDNPMayLoad, SDNPMemOperand]>;
+def loongarch_vmskltz: SDNode<"LoongArchISD::VMSKLTZ", SDT_LoongArchVMSKCOND>;
+def loongarch_vmskgez: SDNode<"LoongArchISD::VMSKGEZ", SDT_LoongArchVMSKCOND>;
+def loongarch_vmskeqz: SDNode<"LoongArchISD::VMSKEQZ", SDT_LoongArchVMSKCOND>;
+def loongarch_vmsknez: SDNode<"LoongArchISD::VMSKNEZ", SDT_LoongArchVMSKCOND>;
+
def immZExt1 : ImmLeaf<i64, [{return isUInt<1>(Imm);}]>;
def immZExt2 : ImmLeaf<i64, [{return isUInt<2>(Imm);}]>;
def immZExt3 : ImmLeaf<i64, [{return isUInt<3>(Imm);}]>;
@@ -1266,6 +1272,16 @@ let usesCustomInserter = 1 in
def PseudoCTPOP : Pseudo<(outs GPR:$rd), (ins GPR:$rj),
[(set GPR:$rd, (ctpop GPR:$rj))]>;
+let usesCustomInserter = 1, hasSideEffects = 0, mayLoad = 0, mayStore = 0 in {
+def PseudoVMSKLTZ_B : Pseudo<(outs GPR:$rd), (ins LSX128:$vj)>;
+def PseudoVMSKLTZ_H : Pseudo<(outs GPR:$rd), (ins LSX128:$vj)>;
+def PseudoVMSKLTZ_W : Pseudo<(outs GPR:$rd), (ins LSX128:$vj)>;
+def PseudoVMSKLTZ_D : Pseudo<(outs GPR:$rd), (ins LSX128:$vj)>;
+def PseudoVMSKGEZ_B : Pseudo<(outs GPR:$rd), (ins LSX128:$vj)>;
+def PseudoVMSKEQZ_B : Pseudo<(outs GPR:$rd), (ins LSX128:$vj)>;
+def PseudoVMSKNEZ_B : Pseudo<(outs GPR:$rd), (ins LSX128:$vj)>;
+} // usesCustomInserter = 1, hasSideEffects = 0, mayLoad = 0, mayStore = 0
+
} // Predicates = [HasExtLSX]
multiclass PatVr<SDPatternOperator OpNode, string Inst> {
@@ -2050,6 +2066,15 @@ def : Pat<(f64 f64imm_vldi:$in),
defm : PatVrVr<abds, "VABSD">;
defm : PatVrVrU<abdu, "VABSD">;
+// Vector mask set by condition
+def : Pat<(loongarch_vmskltz (v16i8 LSX128:$vj)), (PseudoVMSKLTZ_B LSX128:$vj)>;
+def : Pat<(loongarch_vmskltz (v8i16 LSX128:$vj)), (PseudoVMSKLTZ_H LSX128:$vj)>;
+def : Pat<(loongarch_vmskltz (v4i32 LSX128:$vj)), (PseudoVMSKLTZ_W LSX128:$vj)>;
+def : Pat<(loongarch_vmskltz (v2i64 LSX128:$vj)), (PseudoVMSKLTZ_D LSX128:$vj)>;
+def : Pat<(loongarch_vmskgez (v16i8 LSX128:$vj)), (PseudoVMSKGEZ_B LSX128:$vj)>;
+def : Pat<(loongarch_vmskeqz (v16i8 LSX128:$vj)), (PseudoVMSKEQZ_B LSX128:$vj)>;
+def : Pat<(loongarch_vmsknez (v16i8 LSX128:$vj)), (PseudoVMSKNEZ_B LSX128:$vj)>;
+
} // Predicates = [HasExtLSX]
/// Intrinsic pattern
diff --git a/llvm/test/CodeGen/LoongArch/lasx/xvmskcond.ll b/llvm/test/CodeGen/LoongArch/lasx/xvmskcond.ll
index 82e2daee60f82..fb3937c476b03 100644
--- a/llvm/test/CodeGen/LoongArch/lasx/xvmskcond.ll
+++ b/llvm/test/CodeGen/LoongArch/lasx/xvmskcond.ll
@@ -1,140 +1,14 @@
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 5
-; RUN: llc --mtriple=loongarch64 --mattr=+lasx < %s | FileCheck %s
+; RUN: llc --mtriple=loongarch64 --mattr=+lasx --verify-machineinstrs < %s | FileCheck %s
define i32 @xmsk_eq_allzeros_i8(<32 x i8 > %a) {
; CHECK-LABEL: xmsk_eq_allzeros_i8:
; CHECK: # %bb.0: # %entry
-; CHECK-NEXT: addi.d $sp, $sp, -64
-; CHECK-NEXT: .cfi_def_cfa_offset 64
-; CHECK-NEXT: st.d $ra, $sp, 56 # 8-byte Folded Spill
-; CHECK-NEXT: st.d $fp, $sp, 48 # 8-byte Folded Spill
-; CHECK-NEXT: .cfi_offset 1, -8
-; CHECK-NEXT: .cfi_offset 22, -16
-; CHECK-NEXT: addi.d $fp, $sp, 64
-; CHECK-NEXT: .cfi_def_cfa 22, 0
-; CHECK-NEXT: bstrins.d $sp, $zero, 4, 0
-; CHECK-NEXT: xvseqi.b $xr0, $xr0, 0
-; CHECK-NEXT: xvst $xr0, $sp, 0
-; CHECK-NEXT: vpickve2gr.b $a0, $vr0, 0
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 1
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: bstrins.d $a0, $a1, 63, 1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 2
-; CHECK-NEXT: bstrins.d $a0, $a1, 2, 2
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 3
-; CHECK-NEXT: bstrins.d $a0, $a1, 3, 3
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 4
-; CHECK-NEXT: bstrins.d $a0, $a1, 4, 4
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 5
-; CHECK-NEXT: bstrins.d $a0, $a1, 5, 5
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 6
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 6
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 7
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 7
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 8
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 8
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 9
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 9
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 10
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 10
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 11
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 11
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 12
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 12
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 13
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 13
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 14
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 14
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 15
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: ld.bu $a2, $sp, 16
-; CHECK-NEXT: slli.d $a1, $a1, 15
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: ld.bu $a1, $sp, 17
-; CHECK-NEXT: andi $a2, $a2, 1
-; CHECK-NEXT: slli.d $a2, $a2, 16
-; CHECK-NEXT: or $a0, $a0, $a2
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: ld.bu $a2, $sp, 18
-; CHECK-NEXT: slli.d $a1, $a1, 17
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: ld.bu $a1, $sp, 19
-; CHECK-NEXT: andi $a2, $a2, 1
-; CHECK-NEXT: slli.d $a2, $a2, 18
-; CHECK-NEXT: or $a0, $a0, $a2
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: ld.bu $a2, $sp, 20
-; CHECK-NEXT: slli.d $a1, $a1, 19
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: ld.bu $a1, $sp, 21
-; CHECK-NEXT: andi $a2, $a2, 1
-; CHECK-NEXT: slli.d $a2, $a2, 20
-; CHECK-NEXT: or $a0, $a0, $a2
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: ld.bu $a2, $sp, 22
-; CHECK-NEXT: slli.d $a1, $a1, 21
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: ld.bu $a1, $sp, 23
-; CHECK-NEXT: andi $a2, $a2, 1
-; CHECK-NEXT: slli.d $a2, $a2, 22
-; CHECK-NEXT: or $a0, $a0, $a2
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: ld.bu $a2, $sp, 24
-; CHECK-NEXT: slli.d $a1, $a1, 23
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: ld.bu $a1, $sp, 25
-; CHECK-NEXT: andi $a2, $a2, 1
-; CHECK-NEXT: slli.d $a2, $a2, 24
-; CHECK-NEXT: or $a0, $a0, $a2
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: ld.bu $a2, $sp, 26
-; CHECK-NEXT: slli.d $a1, $a1, 25
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: ld.bu $a1, $sp, 27
-; CHECK-NEXT: andi $a2, $a2, 1
-; CHECK-NEXT: slli.d $a2, $a2, 26
-; CHECK-NEXT: or $a0, $a0, $a2
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: ld.bu $a2, $sp, 28
-; CHECK-NEXT: slli.d $a1, $a1, 27
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: ld.bu $a1, $sp, 29
-; CHECK-NEXT: andi $a2, $a2, 1
-; CHECK-NEXT: slli.d $a2, $a2, 28
-; CHECK-NEXT: or $a0, $a0, $a2
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: ld.bu $a2, $sp, 30
-; CHECK-NEXT: slli.d $a1, $a1, 29
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: ld.b $a1, $sp, 31
-; CHECK-NEXT: andi $a2, $a2, 1
-; CHECK-NEXT: slli.d $a2, $a2, 30
-; CHECK-NEXT: or $a0, $a0, $a2
-; CHECK-NEXT: slli.d $a1, $a1, 31
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: bstrpick.d $a0, $a0, 31, 0
-; CHECK-NEXT: addi.d $sp, $fp, -64
-; CHECK-NEXT: ld.d $fp, $sp, 48 # 8-byte Folded Reload
-; CHECK-NEXT: ld.d $ra, $sp, 56 # 8-byte Folded Reload
-; CHECK-NEXT: addi.d $sp, $sp, 64
+; CHECK-NEXT: xvmsknz.b $xr0, $xr0
+; CHECK-NEXT: xvnor.v $xr0, $xr0, $xr0
+; CHECK-NEXT: xvpickve2gr.wu $a0, $xr0, 0
+; CHECK-NEXT: xvpickve2gr.wu $a1, $xr0, 4
+; CHECK-NEXT: bstrins.d $a0, $a1, 31, 16
; CHECK-NEXT: ret
entry:
%1 = icmp eq <32 x i8> %a, splat (i8 0)
@@ -145,138 +19,10 @@ entry:
define i32 @xmsk_sgt_allones_i8(<32 x i8 > %a) {
; CHECK-LABEL: xmsk_sgt_allones_i8:
; CHECK: # %bb.0: # %entry
-; CHECK-NEXT: addi.d $sp, $sp, -64
-; CHECK-NEXT: .cfi_def_cfa_offset 64
-; CHECK-NEXT: st.d $ra, $sp, 56 # 8-byte Folded Spill
-; CHECK-NEXT: st.d $fp, $sp, 48 # 8-byte Folded Spill
-; CHECK-NEXT: .cfi_offset 1, -8
-; CHECK-NEXT: .cfi_offset 22, -16
-; CHECK-NEXT: addi.d $fp, $sp, 64
-; CHECK-NEXT: .cfi_def_cfa 22, 0
-; CHECK-NEXT: bstrins.d $sp, $zero, 4, 0
-; CHECK-NEXT: xvrepli.b $xr1, -1
-; CHECK-NEXT: xvslt.b $xr0, $xr1, $xr0
-; CHECK-NEXT: xvst $xr0, $sp, 0
-; CHECK-NEXT: vpickve2gr.b $a0, $vr0, 0
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 1
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: bstrins.d $a0, $a1, 63, 1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 2
-; CHECK-NEXT: bstrins.d $a0, $a1, 2, 2
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 3
-; CHECK-NEXT: bstrins.d $a0, $a1, 3, 3
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 4
-; CHECK-NEXT: bstrins.d $a0, $a1, 4, 4
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 5
-; CHECK-NEXT: bstrins.d $a0, $a1, 5, 5
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 6
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 6
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 7
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 7
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 8
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 8
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 9
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 9
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 10
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 10
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 11
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 11
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 12
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 12
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 13
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 13
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 14
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 14
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 15
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: ld.bu $a2, $sp, 16
-; CHECK-NEXT: slli.d $a1, $a1, 15
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: ld.bu $a1, $sp, 17
-; CHECK-NEXT: andi $a2, $a2, 1
-; CHECK-NEXT: slli.d $a2, $a2, 16
-; CHECK-NEXT: or $a0, $a0, $a2
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: ld.bu $a2, $sp, 18
-; CHECK-NEXT: slli.d $a1, $a1, 17
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: ld.bu $a1, $sp, 19
-; CHECK-NEXT: andi $a2, $a2, 1
-; CHECK-NEXT: slli.d $a2, $a2, 18
-; CHECK-NEXT: or $a0, $a0, $a2
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: ld.bu $a2, $sp, 20
-; CHECK-NEXT: slli.d $a1, $a1, 19
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: ld.bu $a1, $sp, 21
-; CHECK-NEXT: andi $a2, $a2, 1
-; CHECK-NEXT: slli.d $a2, $a2, 20
-; CHECK-NEXT: or $a0, $a0, $a2
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: ld.bu $a2, $sp, 22
-; CHECK-NEXT: slli.d $a1, $a1, 21
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: ld.bu $a1, $sp, 23
-; CHECK-NEXT: andi $a2, $a2, 1
-; CHECK-NEXT: slli.d $a2, $a2, 22
-; CHECK-NEXT: or $a0, $a0, $a2
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: ld.bu $a2, $sp, 24
-; CHECK-NEXT: slli.d $a1, $a1, 23
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: ld.bu $a1, $sp, 25
-; CHECK-NEXT: andi $a2, $a2, 1
-; CHECK-NEXT: slli.d $a2, $a2, 24
-; CHECK-NEXT: or $a0, $a0, $a2
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: ld.bu $a2, $sp, 26
-; CHECK-NEXT: slli.d $a1, $a1, 25
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: ld.bu $a1, $sp, 27
-; CHECK-NEXT: andi $a2, $a2, 1
-; CHECK-NEXT: slli.d $a2, $a2, 26
-; CHECK-NEXT: or $a0, $a0, $a2
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: ld.bu $a2, $sp, 28
-; CHECK-NEXT: slli.d $a1, $a1, 27
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: ld.bu $a1, $sp, 29
-; CHECK-NEXT: andi $a2, $a2, 1
-; CHECK-NEXT: slli.d $a2, $a2, 28
-; CHECK-NEXT: or $a0, $a0, $a2
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: ld.bu $a2, $sp, 30
-; CHECK-NEXT: slli.d $a1, $a1, 29
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: ld.b $a1, $sp, 31
-; CHECK-NEXT: andi $a2, $a2, 1
-; CHECK-NEXT: slli.d $a2, $a2, 30
-; CHECK-NEXT: or $a0, $a0, $a2
-; CHECK-NEXT: slli.d $a1, $a1, 31
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: bstrpick.d $a0, $a0, 31, 0
-; CHECK-NEXT: addi.d $sp, $fp, -64
-; CHECK-NEXT: ld.d $fp, $sp, 48 # 8-byte Folded Reload
-; CHECK-NEXT: ld.d $ra, $sp, 56 # 8-byte Folded Reload
-; CHECK-NEXT: addi.d $sp, $sp, 64
+; CHECK-NEXT: xvmskgez.b $xr0, $xr0
+; CHECK-NEXT: xvpickve2gr.wu $a0, $xr0, 0
+; CHECK-NEXT: xvpickve2gr.wu $a1, $xr0, 4
+; CHECK-NEXT: bstrins.d $a0, $a1, 31, 16
; CHECK-NEXT: ret
entry:
%1 = icmp sgt <32 x i8> %a, splat (i8 -1)
@@ -287,138 +33,10 @@ entry:
define i32 @xmsk_sge_allzeros_i8(<32 x i8 > %a) {
; CHECK-LABEL: xmsk_sge_allzeros_i8:
; CHECK: # %bb.0: # %entry
-; CHECK-NEXT: addi.d $sp, $sp, -64
-; CHECK-NEXT: .cfi_def_cfa_offset 64
-; CHECK-NEXT: st.d $ra, $sp, 56 # 8-byte Folded Spill
-; CHECK-NEXT: st.d $fp, $sp, 48 # 8-byte Folded Spill
-; CHECK-NEXT: .cfi_offset 1, -8
-; CHECK-NEXT: .cfi_offset 22, -16
-; CHECK-NEXT: addi.d $fp, $sp, 64
-; CHECK-NEXT: .cfi_def_cfa 22, 0
-; CHECK-NEXT: bstrins.d $sp, $zero, 4, 0
-; CHECK-NEXT: xvrepli.b $xr1, 0
-; CHECK-NEXT: xvsle.b $xr0, $xr1, $xr0
-; CHECK-NEXT: xvst $xr0, $sp, 0
-; CHECK-NEXT: vpickve2gr.b $a0, $vr0, 0
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 1
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: bstrins.d $a0, $a1, 63, 1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 2
-; CHECK-NEXT: bstrins.d $a0, $a1, 2, 2
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 3
-; CHECK-NEXT: bstrins.d $a0, $a1, 3, 3
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 4
-; CHECK-NEXT: bstrins.d $a0, $a1, 4, 4
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 5
-; CHECK-NEXT: bstrins.d $a0, $a1, 5, 5
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 6
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 6
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 7
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 7
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 8
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 8
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 9
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 9
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 10
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 10
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 11
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 11
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 12
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 12
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 13
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 13
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 14
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 14
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 15
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: ld.bu $a2, $sp, 16
-; CHECK-NEXT: slli.d $a1, $a1, 15
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: ld.bu $a1, $sp, 17
-; CHECK-NEXT: andi $a2, $a2, 1
-; CHECK-NEXT: slli.d $a2, $a2, 16
-; CHECK-NEXT: or $a0, $a0, $a2
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: ld.bu $a2, $sp, 18
-; CHECK-NEXT: slli.d $a1, $a1, 17
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: ld.bu $a1, $sp, 19
-; CHECK-NEXT: andi $a2, $a2, 1
-; CHECK-NEXT: slli.d $a2, $a2, 18
-; CHECK-NEXT: or $a0, $a0, $a2
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: ld.bu $a2, $sp, 20
-; CHECK-NEXT: slli.d $a1, $a1, 19
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: ld.bu $a1, $sp, 21
-; CHECK-NEXT: andi $a2, $a2, 1
-; CHECK-NEXT: slli.d $a2, $a2, 20
-; CHECK-NEXT: or $a0, $a0, $a2
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: ld.bu $a2, $sp, 22
-; CHECK-NEXT: slli.d $a1, $a1, 21
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: ld.bu $a1, $sp, 23
-; CHECK-NEXT: andi $a2, $a2, 1
-; CHECK-NEXT: slli.d $a2, $a2, 22
-; CHECK-NEXT: or $a0, $a0, $a2
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: ld.bu $a2, $sp, 24
-; CHECK-NEXT: slli.d $a1, $a1, 23
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: ld.bu $a1, $sp, 25
-; CHECK-NEXT: andi $a2, $a2, 1
-; CHECK-NEXT: slli.d $a2, $a2, 24
-; CHECK-NEXT: or $a0, $a0, $a2
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: ld.bu $a2, $sp, 26
-; CHECK-NEXT: slli.d $a1, $a1, 25
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: ld.bu $a1, $sp, 27
-; CHECK-NEXT: andi $a2, $a2, 1
-; CHECK-NEXT: slli.d $a2, $a2, 26
-; CHECK-NEXT: or $a0, $a0, $a2
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: ld.bu $a2, $sp, 28
-; CHECK-NEXT: slli.d $a1, $a1, 27
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: ld.bu $a1, $sp, 29
-; CHECK-NEXT: andi $a2, $a2, 1
-; CHECK-NEXT: slli.d $a2, $a2, 28
-; CHECK-NEXT: or $a0, $a0, $a2
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: ld.bu $a2, $sp, 30
-; CHECK-NEXT: slli.d $a1, $a1, 29
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: ld.b $a1, $sp, 31
-; CHECK-NEXT: andi $a2, $a2, 1
-; CHECK-NEXT: slli.d $a2, $a2, 30
-; CHECK-NEXT: or $a0, $a0, $a2
-; CHECK-NEXT: slli.d $a1, $a1, 31
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: bstrpick.d $a0, $a0, 31, 0
-; CHECK-NEXT: addi.d $sp, $fp, -64
-; CHECK-NEXT: ld.d $fp, $sp, 48 # 8-byte Folded Reload
-; CHECK-NEXT: ld.d $ra, $sp, 56 # 8-byte Folded Reload
-; CHECK-NEXT: addi.d $sp, $sp, 64
+; CHECK-NEXT: xvmskgez.b $xr0, $xr0
+; CHECK-NEXT: xvpickve2gr.wu $a0, $xr0, 0
+; CHECK-NEXT: xvpickve2gr.wu $a1, $xr0, 4
+; CHECK-NEXT: bstrins.d $a0, $a1, 31, 16
; CHECK-NEXT: ret
entry:
%1 = icmp sge <32 x i8> %a, splat (i8 0)
@@ -429,137 +47,10 @@ entry:
define i32 @xmsk_slt_allzeros_i8(<32 x i8 > %a) {
; CHECK-LABEL: xmsk_slt_allzeros_i8:
; CHECK: # %bb.0: # %entry
-; CHECK-NEXT: addi.d $sp, $sp, -64
-; CHECK-NEXT: .cfi_def_cfa_offset 64
-; CHECK-NEXT: st.d $ra, $sp, 56 # 8-byte Folded Spill
-; CHECK-NEXT: st.d $fp, $sp, 48 # 8-byte Folded Spill
-; CHECK-NEXT: .cfi_offset 1, -8
-; CHECK-NEXT: .cfi_offset 22, -16
-; CHECK-NEXT: addi.d $fp, $sp, 64
-; CHECK-NEXT: .cfi_def_cfa 22, 0
-; CHECK-NEXT: bstrins.d $sp, $zero, 4, 0
-; CHECK-NEXT: xvslti.b $xr0, $xr0, 0
-; CHECK-NEXT: xvst $xr0, $sp, 0
-; CHECK-NEXT: vpickve2gr.b $a0, $vr0, 0
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 1
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: bstrins.d $a0, $a1, 63, 1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 2
-; CHECK-NEXT: bstrins.d $a0, $a1, 2, 2
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 3
-; CHECK-NEXT: bstrins.d $a0, $a1, 3, 3
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 4
-; CHECK-NEXT: bstrins.d $a0, $a1, 4, 4
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 5
-; CHECK-NEXT: bstrins.d $a0, $a1, 5, 5
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 6
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 6
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 7
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 7
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 8
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 8
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 9
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 9
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 10
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 10
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 11
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 11
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 12
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 12
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 13
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 13
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 14
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 14
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 15
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: ld.bu $a2, $sp, 16
-; CHECK-NEXT: slli.d $a1, $a1, 15
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: ld.bu $a1, $sp, 17
-; CHECK-NEXT: andi $a2, $a2, 1
-; CHECK-NEXT: slli.d $a2, $a2, 16
-; CHECK-NEXT: or $a0, $a0, $a2
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: ld.bu $a2, $sp, 18
-; CHECK-NEXT: slli.d $a1, $a1, 17
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: ld.bu $a1, $sp, 19
-; CHECK-NEXT: andi $a2, $a2, 1
-; CHECK-NEXT: slli.d $a2, $a2, 18
-; CHECK-NEXT: or $a0, $a0, $a2
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: ld.bu $a2, $sp, 20
-; CHECK-NEXT: slli.d $a1, $a1, 19
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: ld.bu $a1, $sp, 21
-; CHECK-NEXT: andi $a2, $a2, 1
-; CHECK-NEXT: slli.d $a2, $a2, 20
-; CHECK-NEXT: or $a0, $a0, $a2
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: ld.bu $a2, $sp, 22
-; CHECK-NEXT: slli.d $a1, $a1, 21
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: ld.bu $a1, $sp, 23
-; CHECK-NEXT: andi $a2, $a2, 1
-; CHECK-NEXT: slli.d $a2, $a2, 22
-; CHECK-NEXT: or $a0, $a0, $a2
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: ld.bu $a2, $sp, 24
-; CHECK-NEXT: slli.d $a1, $a1, 23
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: ld.bu $a1, $sp, 25
-; CHECK-NEXT: andi $a2, $a2, 1
-; CHECK-NEXT: slli.d $a2, $a2, 24
-; CHECK-NEXT: or $a0, $a0, $a2
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: ld.bu $a2, $sp, 26
-; CHECK-NEXT: slli.d $a1, $a1, 25
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: ld.bu $a1, $sp, 27
-; CHECK-NEXT: andi $a2, $a2, 1
-; CHECK-NEXT: slli.d $a2, $a2, 26
-; CHECK-NEXT: or $a0, $a0, $a2
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: ld.bu $a2, $sp, 28
-; CHECK-NEXT: slli.d $a1, $a1, 27
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: ld.bu $a1, $sp, 29
-; CHECK-NEXT: andi $a2, $a2, 1
-; CHECK-NEXT: slli.d $a2, $a2, 28
-; CHECK-NEXT: or $a0, $a0, $a2
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: ld.bu $a2, $sp, 30
-; CHECK-NEXT: slli.d $a1, $a1, 29
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: ld.b $a1, $sp, 31
-; CHECK-NEXT: andi $a2, $a2, 1
-; CHECK-NEXT: slli.d $a2, $a2, 30
-; CHECK-NEXT: or $a0, $a0, $a2
-; CHECK-NEXT: slli.d $a1, $a1, 31
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: bstrpick.d $a0, $a0, 31, 0
-; CHECK-NEXT: addi.d $sp, $fp, -64
-; CHECK-NEXT: ld.d $fp, $sp, 48 # 8-byte Folded Reload
-; CHECK-NEXT: ld.d $ra, $sp, 56 # 8-byte Folded Reload
-; CHECK-NEXT: addi.d $sp, $sp, 64
+; CHECK-NEXT: xvmskltz.b $xr0, $xr0
+; CHECK-NEXT: xvpickve2gr.wu $a0, $xr0, 0
+; CHECK-NEXT: xvpickve2gr.wu $a1, $xr0, 4
+; CHECK-NEXT: bstrins.d $a0, $a1, 31, 16
; CHECK-NEXT: ret
entry:
%1 = icmp slt <32 x i8> %a, splat (i8 0)
@@ -570,73 +61,10 @@ entry:
define i16 @xmsk_slt_allzeros_i16(<16 x i16 > %a) {
; CHECK-LABEL: xmsk_slt_allzeros_i16:
; CHECK: # %bb.0: # %entry
-; CHECK-NEXT: addi.d $sp, $sp, -64
-; CHECK-NEXT: .cfi_def_cfa_offset 64
-; CHECK-NEXT: st.d $ra, $sp, 56 # 8-byte Folded Spill
-; CHECK-NEXT: st.d $fp, $sp, 48 # 8-byte Folded Spill
-; CHECK-NEXT: .cfi_offset 1, -8
-; CHECK-NEXT: .cfi_offset 22, -16
-; CHECK-NEXT: addi.d $fp, $sp, 64
-; CHECK-NEXT: .cfi_def_cfa 22, 0
-; CHECK-NEXT: bstrins.d $sp, $zero, 4, 0
-; CHECK-NEXT: xvslti.h $xr0, $xr0, 0
-; CHECK-NEXT: xvst $xr0, $sp, 0
-; CHECK-NEXT: vpickve2gr.h $a0, $vr0, 0
-; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 1
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: bstrins.d $a0, $a1, 63, 1
-; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 2
-; CHECK-NEXT: bstrins.d $a0, $a1, 2, 2
-; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 3
-; CHECK-NEXT: bstrins.d $a0, $a1, 3, 3
-; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 4
-; CHECK-NEXT: bstrins.d $a0, $a1, 4, 4
-; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 5
-; CHECK-NEXT: bstrins.d $a0, $a1, 5, 5
-; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 6
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 6
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 7
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: ld.hu $a2, $sp, 16
-; CHECK-NEXT: slli.d $a1, $a1, 7
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: ld.hu $a1, $sp, 18
-; CHECK-NEXT: andi $a2, $a2, 1
-; CHECK-NEXT: slli.d $a2, $a2, 8
-; CHECK-NEXT: or $a0, $a0, $a2
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: ld.hu $a2, $sp, 20
-; CHECK-NEXT: slli.d $a1, $a1, 9
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: ld.hu $a1, $sp, 22
-; CHECK-NEXT: andi $a2, $a2, 1
-; CHECK-NEXT: slli.d $a2, $a2, 10
-; CHECK-NEXT: or $a0, $a0, $a2
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: ld.hu $a2, $sp, 24
-; CHECK-NEXT: slli.d $a1, $a1, 11
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: ld.hu $a1, $sp, 26
-; CHECK-NEXT: andi $a2, $a2, 1
-; CHECK-NEXT: slli.d $a2, $a2, 12
-; CHECK-NEXT: or $a0, $a0, $a2
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: ld.hu $a2, $sp, 28
-; CHECK-NEXT: slli.d $a1, $a1, 13
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: ld.h $a1, $sp, 30
-; CHECK-NEXT: andi $a2, $a2, 1
-; CHECK-NEXT: slli.d $a2, $a2, 14
-; CHECK-NEXT: or $a0, $a0, $a2
-; CHECK-NEXT: slli.d $a1, $a1, 15
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: bstrpick.d $a0, $a0, 15, 0
-; CHECK-NEXT: addi.d $sp, $fp, -64
-; CHECK-NEXT: ld.d $fp, $sp, 48 # 8-byte Folded Reload
-; CHECK-NEXT: ld.d $ra, $sp, 56 # 8-byte Folded Reload
-; CHECK-NEXT: addi.d $sp, $sp, 64
+; CHECK-NEXT: xvmskltz.h $xr0, $xr0
+; CHECK-NEXT: xvpickve2gr.wu $a0, $xr0, 0
+; CHECK-NEXT: xvpickve2gr.wu $a1, $xr0, 4
+; CHECK-NEXT: bstrins.d $a0, $a1, 15, 8
; CHECK-NEXT: ret
entry:
%1 = icmp slt <16 x i16> %a, splat (i16 0)
@@ -647,30 +75,10 @@ entry:
define i8 @xmsk_slt_allzeros_i32(<8 x i32 > %a) {
; CHECK-LABEL: xmsk_slt_allzeros_i32:
; CHECK: # %bb.0: # %entry
-; CHECK-NEXT: addi.d $sp, $sp, -16
-; CHECK-NEXT: .cfi_def_cfa_offset 16
-; CHECK-NEXT: xvslti.w $xr0, $xr0, 0
-; CHECK-NEXT: xvpickve2gr.w $a0, $xr0, 0
-; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 1
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: bstrins.d $a0, $a1, 63, 1
-; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 2
-; CHECK-NEXT: bstrins.d $a0, $a1, 2, 2
-; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 3
-; CHECK-NEXT: bstrins.d $a0, $a1, 3, 3
-; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 4
-; CHECK-NEXT: bstrins.d $a0, $a1, 4, 4
-; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 5
-; CHECK-NEXT: bstrins.d $a0, $a1, 5, 5
-; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 6
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 6
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 7
-; CHECK-NEXT: slli.d $a1, $a1, 7
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: andi $a0, $a0, 255
-; CHECK-NEXT: addi.d $sp, $sp, 16
+; CHECK-NEXT: xvmskltz.w $xr0, $xr0
+; CHECK-NEXT: xvpickve2gr.wu $a0, $xr0, 0
+; CHECK-NEXT: xvpickve2gr.wu $a1, $xr0, 4
+; CHECK-NEXT: bstrins.d $a0, $a1, 7, 4
; CHECK-NEXT: ret
entry:
%1 = icmp slt <8 x i32> %a, splat (i32 0)
@@ -681,21 +89,10 @@ entry:
define i4 @xmsk_slt_allzeros_i64(<4 x i64 > %a) {
; CHECK-LABEL: xmsk_slt_allzeros_i64:
; CHECK: # %bb.0: # %entry
-; CHECK-NEXT: addi.d $sp, $sp, -16
-; CHECK-NEXT: .cfi_def_cfa_offset 16
-; CHECK-NEXT: xvslti.d $xr0, $xr0, 0
-; CHECK-NEXT: xvpickve2gr.d $a0, $xr0, 0
-; CHECK-NEXT: xvpickve2gr.d $a1, $xr0, 1
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 1
-; CHECK-NEXT: sub.d $a0, $a1, $a0
-; CHECK-NEXT: xvpickve2gr.d $a1, $xr0, 2
-; CHECK-NEXT: bstrins.d $a0, $a1, 2, 2
-; CHECK-NEXT: xvpickve2gr.d $a1, $xr0, 3
-; CHECK-NEXT: slli.d $a1, $a1, 3
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: andi $a0, $a0, 15
-; CHECK-NEXT: addi.d $sp, $sp, 16
+; CHECK-NEXT: xvmskltz.d $xr0, $xr0
+; CHECK-NEXT: xvpickve2gr.wu $a0, $xr0, 0
+; CHECK-NEXT: xvpickve2gr.wu $a1, $xr0, 4
+; CHECK-NEXT: bstrins.d $a0, $a1, 3, 2
; CHECK-NEXT: ret
entry:
%1 = icmp slt <4 x i64> %a, splat (i64 0)
@@ -706,137 +103,10 @@ entry:
define i32 @xmsk_sle_allones_i8(<32 x i8 > %a) {
; CHECK-LABEL: xmsk_sle_allones_i8:
; CHECK: # %bb.0: # %entry
-; CHECK-NEXT: addi.d $sp, $sp, -64
-; CHECK-NEXT: .cfi_def_cfa_offset 64
-; CHECK-NEXT: st.d $ra, $sp, 56 # 8-byte Folded Spill
-; CHECK-NEXT: st.d $fp, $sp, 48 # 8-byte Folded Spill
-; CHECK-NEXT: .cfi_offset 1, -8
-; CHECK-NEXT: .cfi_offset 22, -16
-; CHECK-NEXT: addi.d $fp, $sp, 64
-; CHECK-NEXT: .cfi_def_cfa 22, 0
-; CHECK-NEXT: bstrins.d $sp, $zero, 4, 0
-; CHECK-NEXT: xvslei.b $xr0, $xr0, -1
-; CHECK-NEXT: xvst $xr0, $sp, 0
-; CHECK-NEXT: vpickve2gr.b $a0, $vr0, 0
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 1
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: bstrins.d $a0, $a1, 63, 1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 2
-; CHECK-NEXT: bstrins.d $a0, $a1, 2, 2
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 3
-; CHECK-NEXT: bstrins.d $a0, $a1, 3, 3
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 4
-; CHECK-NEXT: bstrins.d $a0, $a1, 4, 4
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 5
-; CHECK-NEXT: bstrins.d $a0, $a1, 5, 5
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 6
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 6
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 7
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 7
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 8
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 8
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 9
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 9
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 10
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 10
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 11
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 11
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 12
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 12
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 13
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 13
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 14
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 14
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 15
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: ld.bu $a2, $sp, 16
-; CHECK-NEXT: slli.d $a1, $a1, 15
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: ld.bu $a1, $sp, 17
-; CHECK-NEXT: andi $a2, $a2, 1
-; CHECK-NEXT: slli.d $a2, $a2, 16
-; CHECK-NEXT: or $a0, $a0, $a2
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: ld.bu $a2, $sp, 18
-; CHECK-NEXT: slli.d $a1, $a1, 17
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: ld.bu $a1, $sp, 19
-; CHECK-NEXT: andi $a2, $a2, 1
-; CHECK-NEXT: slli.d $a2, $a2, 18
-; CHECK-NEXT: or $a0, $a0, $a2
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: ld.bu $a2, $sp, 20
-; CHECK-NEXT: slli.d $a1, $a1, 19
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: ld.bu $a1, $sp, 21
-; CHECK-NEXT: andi $a2, $a2, 1
-; CHECK-NEXT: slli.d $a2, $a2, 20
-; CHECK-NEXT: or $a0, $a0, $a2
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: ld.bu $a2, $sp, 22
-; CHECK-NEXT: slli.d $a1, $a1, 21
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: ld.bu $a1, $sp, 23
-; CHECK-NEXT: andi $a2, $a2, 1
-; CHECK-NEXT: slli.d $a2, $a2, 22
-; CHECK-NEXT: or $a0, $a0, $a2
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: ld.bu $a2, $sp, 24
-; CHECK-NEXT: slli.d $a1, $a1, 23
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: ld.bu $a1, $sp, 25
-; CHECK-NEXT: andi $a2, $a2, 1
-; CHECK-NEXT: slli.d $a2, $a2, 24
-; CHECK-NEXT: or $a0, $a0, $a2
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: ld.bu $a2, $sp, 26
-; CHECK-NEXT: slli.d $a1, $a1, 25
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: ld.bu $a1, $sp, 27
-; CHECK-NEXT: andi $a2, $a2, 1
-; CHECK-NEXT: slli.d $a2, $a2, 26
-; CHECK-NEXT: or $a0, $a0, $a2
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: ld.bu $a2, $sp, 28
-; CHECK-NEXT: slli.d $a1, $a1, 27
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: ld.bu $a1, $sp, 29
-; CHECK-NEXT: andi $a2, $a2, 1
-; CHECK-NEXT: slli.d $a2, $a2, 28
-; CHECK-NEXT: or $a0, $a0, $a2
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: ld.bu $a2, $sp, 30
-; CHECK-NEXT: slli.d $a1, $a1, 29
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: ld.b $a1, $sp, 31
-; CHECK-NEXT: andi $a2, $a2, 1
-; CHECK-NEXT: slli.d $a2, $a2, 30
-; CHECK-NEXT: or $a0, $a0, $a2
-; CHECK-NEXT: slli.d $a1, $a1, 31
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: bstrpick.d $a0, $a0, 31, 0
-; CHECK-NEXT: addi.d $sp, $fp, -64
-; CHECK-NEXT: ld.d $fp, $sp, 48 # 8-byte Folded Reload
-; CHECK-NEXT: ld.d $ra, $sp, 56 # 8-byte Folded Reload
-; CHECK-NEXT: addi.d $sp, $sp, 64
+; CHECK-NEXT: xvmskltz.b $xr0, $xr0
+; CHECK-NEXT: xvpickve2gr.wu $a0, $xr0, 0
+; CHECK-NEXT: xvpickve2gr.wu $a1, $xr0, 4
+; CHECK-NEXT: bstrins.d $a0, $a1, 31, 16
; CHECK-NEXT: ret
entry:
%1 = icmp sle <32 x i8> %a, splat (i8 -1)
@@ -847,73 +117,10 @@ entry:
define i16 @xmsk_sle_allones_i32(<16 x i16 > %a) {
; CHECK-LABEL: xmsk_sle_allones_i32:
; CHECK: # %bb.0: # %entry
-; CHECK-NEXT: addi.d $sp, $sp, -64
-; CHECK-NEXT: .cfi_def_cfa_offset 64
-; CHECK-NEXT: st.d $ra, $sp, 56 # 8-byte Folded Spill
-; CHECK-NEXT: st.d $fp, $sp, 48 # 8-byte Folded Spill
-; CHECK-NEXT: .cfi_offset 1, -8
-; CHECK-NEXT: .cfi_offset 22, -16
-; CHECK-NEXT: addi.d $fp, $sp, 64
-; CHECK-NEXT: .cfi_def_cfa 22, 0
-; CHECK-NEXT: bstrins.d $sp, $zero, 4, 0
-; CHECK-NEXT: xvslei.h $xr0, $xr0, -1
-; CHECK-NEXT: xvst $xr0, $sp, 0
-; CHECK-NEXT: vpickve2gr.h $a0, $vr0, 0
-; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 1
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: bstrins.d $a0, $a1, 63, 1
-; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 2
-; CHECK-NEXT: bstrins.d $a0, $a1, 2, 2
-; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 3
-; CHECK-NEXT: bstrins.d $a0, $a1, 3, 3
-; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 4
-; CHECK-NEXT: bstrins.d $a0, $a1, 4, 4
-; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 5
-; CHECK-NEXT: bstrins.d $a0, $a1, 5, 5
-; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 6
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 6
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 7
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: ld.hu $a2, $sp, 16
-; CHECK-NEXT: slli.d $a1, $a1, 7
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: ld.hu $a1, $sp, 18
-; CHECK-NEXT: andi $a2, $a2, 1
-; CHECK-NEXT: slli.d $a2, $a2, 8
-; CHECK-NEXT: or $a0, $a0, $a2
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: ld.hu $a2, $sp, 20
-; CHECK-NEXT: slli.d $a1, $a1, 9
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: ld.hu $a1, $sp, 22
-; CHECK-NEXT: andi $a2, $a2, 1
-; CHECK-NEXT: slli.d $a2, $a2, 10
-; CHECK-NEXT: or $a0, $a0, $a2
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: ld.hu $a2, $sp, 24
-; CHECK-NEXT: slli.d $a1, $a1, 11
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: ld.hu $a1, $sp, 26
-; CHECK-NEXT: andi $a2, $a2, 1
-; CHECK-NEXT: slli.d $a2, $a2, 12
-; CHECK-NEXT: or $a0, $a0, $a2
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: ld.hu $a2, $sp, 28
-; CHECK-NEXT: slli.d $a1, $a1, 13
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: ld.h $a1, $sp, 30
-; CHECK-NEXT: andi $a2, $a2, 1
-; CHECK-NEXT: slli.d $a2, $a2, 14
-; CHECK-NEXT: or $a0, $a0, $a2
-; CHECK-NEXT: slli.d $a1, $a1, 15
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: bstrpick.d $a0, $a0, 15, 0
-; CHECK-NEXT: addi.d $sp, $fp, -64
-; CHECK-NEXT: ld.d $fp, $sp, 48 # 8-byte Folded Reload
-; CHECK-NEXT: ld.d $ra, $sp, 56 # 8-byte Folded Reload
-; CHECK-NEXT: addi.d $sp, $sp, 64
+; CHECK-NEXT: xvmskltz.h $xr0, $xr0
+; CHECK-NEXT: xvpickve2gr.wu $a0, $xr0, 0
+; CHECK-NEXT: xvpickve2gr.wu $a1, $xr0, 4
+; CHECK-NEXT: bstrins.d $a0, $a1, 15, 8
; CHECK-NEXT: ret
entry:
%1 = icmp sle <16 x i16> %a, splat (i16 -1)
@@ -924,30 +131,10 @@ entry:
define i8 @xmsk_sle_allones_i16(<8 x i32 > %a) {
; CHECK-LABEL: xmsk_sle_allones_i16:
; CHECK: # %bb.0: # %entry
-; CHECK-NEXT: addi.d $sp, $sp, -16
-; CHECK-NEXT: .cfi_def_cfa_offset 16
-; CHECK-NEXT: xvslei.w $xr0, $xr0, -1
-; CHECK-NEXT: xvpickve2gr.w $a0, $xr0, 0
-; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 1
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: bstrins.d $a0, $a1, 63, 1
-; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 2
-; CHECK-NEXT: bstrins.d $a0, $a1, 2, 2
-; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 3
-; CHECK-NEXT: bstrins.d $a0, $a1, 3, 3
-; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 4
-; CHECK-NEXT: bstrins.d $a0, $a1, 4, 4
-; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 5
-; CHECK-NEXT: bstrins.d $a0, $a1, 5, 5
-; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 6
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 6
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: xvpickve2gr.w $a1, $xr0, 7
-; CHECK-NEXT: slli.d $a1, $a1, 7
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: andi $a0, $a0, 255
-; CHECK-NEXT: addi.d $sp, $sp, 16
+; CHECK-NEXT: xvmskltz.w $xr0, $xr0
+; CHECK-NEXT: xvpickve2gr.wu $a0, $xr0, 0
+; CHECK-NEXT: xvpickve2gr.wu $a1, $xr0, 4
+; CHECK-NEXT: bstrins.d $a0, $a1, 7, 4
; CHECK-NEXT: ret
entry:
%1 = icmp sle <8 x i32> %a, splat (i32 -1)
@@ -958,21 +145,10 @@ entry:
define i4 @xmsk_sle_allones_i64(<4 x i64 > %a) {
; CHECK-LABEL: xmsk_sle_allones_i64:
; CHECK: # %bb.0: # %entry
-; CHECK-NEXT: addi.d $sp, $sp, -16
-; CHECK-NEXT: .cfi_def_cfa_offset 16
-; CHECK-NEXT: xvslei.d $xr0, $xr0, -1
-; CHECK-NEXT: xvpickve2gr.d $a0, $xr0, 0
-; CHECK-NEXT: xvpickve2gr.d $a1, $xr0, 1
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 1
-; CHECK-NEXT: sub.d $a0, $a1, $a0
-; CHECK-NEXT: xvpickve2gr.d $a1, $xr0, 2
-; CHECK-NEXT: bstrins.d $a0, $a1, 2, 2
-; CHECK-NEXT: xvpickve2gr.d $a1, $xr0, 3
-; CHECK-NEXT: slli.d $a1, $a1, 3
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: andi $a0, $a0, 15
-; CHECK-NEXT: addi.d $sp, $sp, 16
+; CHECK-NEXT: xvmskltz.d $xr0, $xr0
+; CHECK-NEXT: xvpickve2gr.wu $a0, $xr0, 0
+; CHECK-NEXT: xvpickve2gr.wu $a1, $xr0, 4
+; CHECK-NEXT: bstrins.d $a0, $a1, 3, 2
; CHECK-NEXT: ret
entry:
%1 = icmp sle <4 x i64> %a, splat (i64 -1)
@@ -983,138 +159,10 @@ entry:
define i32 @xmsk_ne_allzeros_i8(<32 x i8 > %a) {
; CHECK-LABEL: xmsk_ne_allzeros_i8:
; CHECK: # %bb.0: # %entry
-; CHECK-NEXT: addi.d $sp, $sp, -64
-; CHECK-NEXT: .cfi_def_cfa_offset 64
-; CHECK-NEXT: st.d $ra, $sp, 56 # 8-byte Folded Spill
-; CHECK-NEXT: st.d $fp, $sp, 48 # 8-byte Folded Spill
-; CHECK-NEXT: .cfi_offset 1, -8
-; CHECK-NEXT: .cfi_offset 22, -16
-; CHECK-NEXT: addi.d $fp, $sp, 64
-; CHECK-NEXT: .cfi_def_cfa 22, 0
-; CHECK-NEXT: bstrins.d $sp, $zero, 4, 0
-; CHECK-NEXT: xvseqi.b $xr0, $xr0, 0
-; CHECK-NEXT: xvxori.b $xr0, $xr0, 255
-; CHECK-NEXT: xvst $xr0, $sp, 0
-; CHECK-NEXT: vpickve2gr.b $a0, $vr0, 0
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 1
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: bstrins.d $a0, $a1, 63, 1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 2
-; CHECK-NEXT: bstrins.d $a0, $a1, 2, 2
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 3
-; CHECK-NEXT: bstrins.d $a0, $a1, 3, 3
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 4
-; CHECK-NEXT: bstrins.d $a0, $a1, 4, 4
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 5
-; CHECK-NEXT: bstrins.d $a0, $a1, 5, 5
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 6
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 6
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 7
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 7
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 8
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 8
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 9
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 9
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 10
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 10
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 11
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 11
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 12
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 12
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 13
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 13
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 14
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 14
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 15
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: ld.bu $a2, $sp, 16
-; CHECK-NEXT: slli.d $a1, $a1, 15
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: ld.bu $a1, $sp, 17
-; CHECK-NEXT: andi $a2, $a2, 1
-; CHECK-NEXT: slli.d $a2, $a2, 16
-; CHECK-NEXT: or $a0, $a0, $a2
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: ld.bu $a2, $sp, 18
-; CHECK-NEXT: slli.d $a1, $a1, 17
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: ld.bu $a1, $sp, 19
-; CHECK-NEXT: andi $a2, $a2, 1
-; CHECK-NEXT: slli.d $a2, $a2, 18
-; CHECK-NEXT: or $a0, $a0, $a2
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: ld.bu $a2, $sp, 20
-; CHECK-NEXT: slli.d $a1, $a1, 19
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: ld.bu $a1, $sp, 21
-; CHECK-NEXT: andi $a2, $a2, 1
-; CHECK-NEXT: slli.d $a2, $a2, 20
-; CHECK-NEXT: or $a0, $a0, $a2
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: ld.bu $a2, $sp, 22
-; CHECK-NEXT: slli.d $a1, $a1, 21
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: ld.bu $a1, $sp, 23
-; CHECK-NEXT: andi $a2, $a2, 1
-; CHECK-NEXT: slli.d $a2, $a2, 22
-; CHECK-NEXT: or $a0, $a0, $a2
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: ld.bu $a2, $sp, 24
-; CHECK-NEXT: slli.d $a1, $a1, 23
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: ld.bu $a1, $sp, 25
-; CHECK-NEXT: andi $a2, $a2, 1
-; CHECK-NEXT: slli.d $a2, $a2, 24
-; CHECK-NEXT: or $a0, $a0, $a2
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: ld.bu $a2, $sp, 26
-; CHECK-NEXT: slli.d $a1, $a1, 25
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: ld.bu $a1, $sp, 27
-; CHECK-NEXT: andi $a2, $a2, 1
-; CHECK-NEXT: slli.d $a2, $a2, 26
-; CHECK-NEXT: or $a0, $a0, $a2
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: ld.bu $a2, $sp, 28
-; CHECK-NEXT: slli.d $a1, $a1, 27
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: ld.bu $a1, $sp, 29
-; CHECK-NEXT: andi $a2, $a2, 1
-; CHECK-NEXT: slli.d $a2, $a2, 28
-; CHECK-NEXT: or $a0, $a0, $a2
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: ld.bu $a2, $sp, 30
-; CHECK-NEXT: slli.d $a1, $a1, 29
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: ld.b $a1, $sp, 31
-; CHECK-NEXT: andi $a2, $a2, 1
-; CHECK-NEXT: slli.d $a2, $a2, 30
-; CHECK-NEXT: or $a0, $a0, $a2
-; CHECK-NEXT: slli.d $a1, $a1, 31
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: bstrpick.d $a0, $a0, 31, 0
-; CHECK-NEXT: addi.d $sp, $fp, -64
-; CHECK-NEXT: ld.d $fp, $sp, 48 # 8-byte Folded Reload
-; CHECK-NEXT: ld.d $ra, $sp, 56 # 8-byte Folded Reload
-; CHECK-NEXT: addi.d $sp, $sp, 64
+; CHECK-NEXT: xvmsknz.b $xr0, $xr0
+; CHECK-NEXT: xvpickve2gr.wu $a0, $xr0, 0
+; CHECK-NEXT: xvpickve2gr.wu $a1, $xr0, 4
+; CHECK-NEXT: bstrins.d $a0, $a1, 31, 16
; CHECK-NEXT: ret
entry:
%1 = icmp ne <32 x i8> %a, splat (i8 0)
diff --git a/llvm/test/CodeGen/LoongArch/lsx/vmskcond.ll b/llvm/test/CodeGen/LoongArch/lsx/vmskcond.ll
index eff3302145c95..5cb8f4e40b27a 100644
--- a/llvm/test/CodeGen/LoongArch/lsx/vmskcond.ll
+++ b/llvm/test/CodeGen/LoongArch/lsx/vmskcond.ll
@@ -1,65 +1,12 @@
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 5
-; RUN: llc --mtriple=loongarch64 --mattr=+lsx < %s | FileCheck %s
+; RUN: llc --mtriple=loongarch64 --mattr=+lsx --verify-machineinstrs < %s | FileCheck %s
define i16 @vmsk_eq_allzeros_i8(<16 x i8 > %a) {
; CHECK-LABEL: vmsk_eq_allzeros_i8:
; CHECK: # %bb.0: # %entry
-; CHECK-NEXT: addi.d $sp, $sp, -16
-; CHECK-NEXT: .cfi_def_cfa_offset 16
-; CHECK-NEXT: vseqi.b $vr0, $vr0, 0
-; CHECK-NEXT: vpickve2gr.b $a0, $vr0, 0
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 1
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: bstrins.d $a0, $a1, 63, 1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 2
-; CHECK-NEXT: bstrins.d $a0, $a1, 2, 2
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 3
-; CHECK-NEXT: bstrins.d $a0, $a1, 3, 3
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 4
-; CHECK-NEXT: bstrins.d $a0, $a1, 4, 4
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 5
-; CHECK-NEXT: bstrins.d $a0, $a1, 5, 5
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 6
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 6
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 7
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 7
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 8
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 8
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 9
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 9
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 10
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 10
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 11
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 11
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 12
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 12
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 13
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 13
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 14
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 14
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 15
-; CHECK-NEXT: slli.d $a1, $a1, 15
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: bstrpick.d $a0, $a0, 15, 0
-; CHECK-NEXT: addi.d $sp, $sp, 16
+; CHECK-NEXT: vmsknz.b $vr0, $vr0
+; CHECK-NEXT: vnor.v $vr0, $vr0, $vr0
+; CHECK-NEXT: vpickve2gr.hu $a0, $vr0, 0
; CHECK-NEXT: ret
entry:
%1 = icmp eq <16 x i8> %a, splat (i8 0)
@@ -70,63 +17,8 @@ entry:
define i16 @vmsk_sgt_allones_i8(<16 x i8 > %a) {
; CHECK-LABEL: vmsk_sgt_allones_i8:
; CHECK: # %bb.0: # %entry
-; CHECK-NEXT: addi.d $sp, $sp, -16
-; CHECK-NEXT: .cfi_def_cfa_offset 16
-; CHECK-NEXT: vrepli.b $vr1, -1
-; CHECK-NEXT: vslt.b $vr0, $vr1, $vr0
-; CHECK-NEXT: vpickve2gr.b $a0, $vr0, 0
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 1
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: bstrins.d $a0, $a1, 63, 1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 2
-; CHECK-NEXT: bstrins.d $a0, $a1, 2, 2
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 3
-; CHECK-NEXT: bstrins.d $a0, $a1, 3, 3
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 4
-; CHECK-NEXT: bstrins.d $a0, $a1, 4, 4
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 5
-; CHECK-NEXT: bstrins.d $a0, $a1, 5, 5
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 6
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 6
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 7
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 7
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 8
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 8
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 9
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 9
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 10
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 10
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 11
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 11
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 12
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 12
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 13
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 13
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 14
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 14
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 15
-; CHECK-NEXT: slli.d $a1, $a1, 15
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: bstrpick.d $a0, $a0, 15, 0
-; CHECK-NEXT: addi.d $sp, $sp, 16
+; CHECK-NEXT: vmskgez.b $vr0, $vr0
+; CHECK-NEXT: vpickve2gr.hu $a0, $vr0, 0
; CHECK-NEXT: ret
entry:
%1 = icmp sgt <16 x i8> %a, splat (i8 -1)
@@ -137,63 +29,8 @@ entry:
define i16 @vmsk_sge_allzeros_i8(<16 x i8 > %a) {
; CHECK-LABEL: vmsk_sge_allzeros_i8:
; CHECK: # %bb.0: # %entry
-; CHECK-NEXT: addi.d $sp, $sp, -16
-; CHECK-NEXT: .cfi_def_cfa_offset 16
-; CHECK-NEXT: vrepli.b $vr1, 0
-; CHECK-NEXT: vsle.b $vr0, $vr1, $vr0
-; CHECK-NEXT: vpickve2gr.b $a0, $vr0, 0
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 1
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: bstrins.d $a0, $a1, 63, 1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 2
-; CHECK-NEXT: bstrins.d $a0, $a1, 2, 2
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 3
-; CHECK-NEXT: bstrins.d $a0, $a1, 3, 3
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 4
-; CHECK-NEXT: bstrins.d $a0, $a1, 4, 4
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 5
-; CHECK-NEXT: bstrins.d $a0, $a1, 5, 5
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 6
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 6
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 7
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 7
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 8
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 8
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 9
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 9
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 10
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 10
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 11
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 11
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 12
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 12
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 13
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 13
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 14
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 14
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 15
-; CHECK-NEXT: slli.d $a1, $a1, 15
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: bstrpick.d $a0, $a0, 15, 0
-; CHECK-NEXT: addi.d $sp, $sp, 16
+; CHECK-NEXT: vmskgez.b $vr0, $vr0
+; CHECK-NEXT: vpickve2gr.hu $a0, $vr0, 0
; CHECK-NEXT: ret
entry:
%1 = icmp sge <16 x i8> %a, splat (i8 0)
@@ -204,62 +41,8 @@ entry:
define i16 @vmsk_slt_allzeros_i8(<16 x i8 > %a) {
; CHECK-LABEL: vmsk_slt_allzeros_i8:
; CHECK: # %bb.0: # %entry
-; CHECK-NEXT: addi.d $sp, $sp, -16
-; CHECK-NEXT: .cfi_def_cfa_offset 16
-; CHECK-NEXT: vslti.b $vr0, $vr0, 0
-; CHECK-NEXT: vpickve2gr.b $a0, $vr0, 0
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 1
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: bstrins.d $a0, $a1, 63, 1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 2
-; CHECK-NEXT: bstrins.d $a0, $a1, 2, 2
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 3
-; CHECK-NEXT: bstrins.d $a0, $a1, 3, 3
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 4
-; CHECK-NEXT: bstrins.d $a0, $a1, 4, 4
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 5
-; CHECK-NEXT: bstrins.d $a0, $a1, 5, 5
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 6
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 6
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 7
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 7
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 8
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 8
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 9
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 9
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 10
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 10
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 11
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 11
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 12
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 12
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 13
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 13
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 14
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 14
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 15
-; CHECK-NEXT: slli.d $a1, $a1, 15
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: bstrpick.d $a0, $a0, 15, 0
-; CHECK-NEXT: addi.d $sp, $sp, 16
+; CHECK-NEXT: vmskltz.b $vr0, $vr0
+; CHECK-NEXT: vpickve2gr.hu $a0, $vr0, 0
; CHECK-NEXT: ret
entry:
%1 = icmp slt <16 x i8> %a, splat (i8 0)
@@ -270,30 +53,8 @@ entry:
define i8 @vmsk_slt_allzeros_i16(<8 x i16 > %a) {
; CHECK-LABEL: vmsk_slt_allzeros_i16:
; CHECK: # %bb.0: # %entry
-; CHECK-NEXT: addi.d $sp, $sp, -16
-; CHECK-NEXT: .cfi_def_cfa_offset 16
-; CHECK-NEXT: vslti.h $vr0, $vr0, 0
-; CHECK-NEXT: vpickve2gr.h $a0, $vr0, 0
-; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 1
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: bstrins.d $a0, $a1, 63, 1
-; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 2
-; CHECK-NEXT: bstrins.d $a0, $a1, 2, 2
-; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 3
-; CHECK-NEXT: bstrins.d $a0, $a1, 3, 3
-; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 4
-; CHECK-NEXT: bstrins.d $a0, $a1, 4, 4
-; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 5
-; CHECK-NEXT: bstrins.d $a0, $a1, 5, 5
-; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 6
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 6
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 7
-; CHECK-NEXT: slli.d $a1, $a1, 7
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: andi $a0, $a0, 255
-; CHECK-NEXT: addi.d $sp, $sp, 16
+; CHECK-NEXT: vmskltz.h $vr0, $vr0
+; CHECK-NEXT: vpickve2gr.hu $a0, $vr0, 0
; CHECK-NEXT: ret
entry:
%1 = icmp slt <8 x i16> %a, splat (i16 0)
@@ -304,20 +65,8 @@ entry:
define i4 @vmsk_slt_allzeros_i32(<4 x i32 > %a) {
; CHECK-LABEL: vmsk_slt_allzeros_i32:
; CHECK: # %bb.0: # %entry
-; CHECK-NEXT: addi.d $sp, $sp, -16
-; CHECK-NEXT: .cfi_def_cfa_offset 16
-; CHECK-NEXT: vslti.w $vr0, $vr0, 0
-; CHECK-NEXT: vpickve2gr.w $a0, $vr0, 0
-; CHECK-NEXT: vpickve2gr.w $a1, $vr0, 1
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: bstrins.d $a0, $a1, 63, 1
-; CHECK-NEXT: vpickve2gr.w $a1, $vr0, 2
-; CHECK-NEXT: bstrins.d $a0, $a1, 2, 2
-; CHECK-NEXT: vpickve2gr.w $a1, $vr0, 3
-; CHECK-NEXT: slli.d $a1, $a1, 3
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: andi $a0, $a0, 15
-; CHECK-NEXT: addi.d $sp, $sp, 16
+; CHECK-NEXT: vmskltz.w $vr0, $vr0
+; CHECK-NEXT: vpickve2gr.hu $a0, $vr0, 0
; CHECK-NEXT: ret
entry:
%1 = icmp slt <4 x i32> %a, splat (i32 0)
@@ -328,15 +77,8 @@ entry:
define i2 @vmsk_slt_allzeros_i64(<2 x i64 > %a) {
; CHECK-LABEL: vmsk_slt_allzeros_i64:
; CHECK: # %bb.0: # %entry
-; CHECK-NEXT: addi.d $sp, $sp, -16
-; CHECK-NEXT: .cfi_def_cfa_offset 16
-; CHECK-NEXT: vslti.d $vr0, $vr0, 0
-; CHECK-NEXT: vpickve2gr.d $a0, $vr0, 0
-; CHECK-NEXT: vpickve2gr.d $a1, $vr0, 1
-; CHECK-NEXT: slli.d $a1, $a1, 1
-; CHECK-NEXT: sub.d $a0, $a1, $a0
-; CHECK-NEXT: andi $a0, $a0, 3
-; CHECK-NEXT: addi.d $sp, $sp, 16
+; CHECK-NEXT: vmskltz.d $vr0, $vr0
+; CHECK-NEXT: vpickve2gr.hu $a0, $vr0, 0
; CHECK-NEXT: ret
entry:
%1 = icmp slt <2 x i64> %a, splat (i64 0)
@@ -347,62 +89,8 @@ entry:
define i16 @vmsk_sle_allones_i8(<16 x i8 > %a) {
; CHECK-LABEL: vmsk_sle_allones_i8:
; CHECK: # %bb.0: # %entry
-; CHECK-NEXT: addi.d $sp, $sp, -16
-; CHECK-NEXT: .cfi_def_cfa_offset 16
-; CHECK-NEXT: vslei.b $vr0, $vr0, -1
-; CHECK-NEXT: vpickve2gr.b $a0, $vr0, 0
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 1
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: bstrins.d $a0, $a1, 63, 1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 2
-; CHECK-NEXT: bstrins.d $a0, $a1, 2, 2
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 3
-; CHECK-NEXT: bstrins.d $a0, $a1, 3, 3
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 4
-; CHECK-NEXT: bstrins.d $a0, $a1, 4, 4
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 5
-; CHECK-NEXT: bstrins.d $a0, $a1, 5, 5
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 6
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 6
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 7
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 7
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 8
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 8
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 9
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 9
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 10
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 10
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 11
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 11
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 12
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 12
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 13
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 13
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 14
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 14
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 15
-; CHECK-NEXT: slli.d $a1, $a1, 15
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: bstrpick.d $a0, $a0, 15, 0
-; CHECK-NEXT: addi.d $sp, $sp, 16
+; CHECK-NEXT: vmskltz.b $vr0, $vr0
+; CHECK-NEXT: vpickve2gr.hu $a0, $vr0, 0
; CHECK-NEXT: ret
entry:
%1 = icmp sle <16 x i8> %a, splat (i8 -1)
@@ -413,30 +101,8 @@ entry:
define i8 @vmsk_sle_allones_i16(<8 x i16 > %a) {
; CHECK-LABEL: vmsk_sle_allones_i16:
; CHECK: # %bb.0: # %entry
-; CHECK-NEXT: addi.d $sp, $sp, -16
-; CHECK-NEXT: .cfi_def_cfa_offset 16
-; CHECK-NEXT: vslei.h $vr0, $vr0, -1
-; CHECK-NEXT: vpickve2gr.h $a0, $vr0, 0
-; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 1
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: bstrins.d $a0, $a1, 63, 1
-; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 2
-; CHECK-NEXT: bstrins.d $a0, $a1, 2, 2
-; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 3
-; CHECK-NEXT: bstrins.d $a0, $a1, 3, 3
-; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 4
-; CHECK-NEXT: bstrins.d $a0, $a1, 4, 4
-; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 5
-; CHECK-NEXT: bstrins.d $a0, $a1, 5, 5
-; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 6
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 6
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.h $a1, $vr0, 7
-; CHECK-NEXT: slli.d $a1, $a1, 7
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: andi $a0, $a0, 255
-; CHECK-NEXT: addi.d $sp, $sp, 16
+; CHECK-NEXT: vmskltz.h $vr0, $vr0
+; CHECK-NEXT: vpickve2gr.hu $a0, $vr0, 0
; CHECK-NEXT: ret
entry:
%1 = icmp sle <8 x i16> %a, splat (i16 -1)
@@ -447,20 +113,8 @@ entry:
define i4 @vmsk_sle_allones_i32(<4 x i32 > %a) {
; CHECK-LABEL: vmsk_sle_allones_i32:
; CHECK: # %bb.0: # %entry
-; CHECK-NEXT: addi.d $sp, $sp, -16
-; CHECK-NEXT: .cfi_def_cfa_offset 16
-; CHECK-NEXT: vslei.w $vr0, $vr0, -1
-; CHECK-NEXT: vpickve2gr.w $a0, $vr0, 0
-; CHECK-NEXT: vpickve2gr.w $a1, $vr0, 1
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: bstrins.d $a0, $a1, 63, 1
-; CHECK-NEXT: vpickve2gr.w $a1, $vr0, 2
-; CHECK-NEXT: bstrins.d $a0, $a1, 2, 2
-; CHECK-NEXT: vpickve2gr.w $a1, $vr0, 3
-; CHECK-NEXT: slli.d $a1, $a1, 3
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: andi $a0, $a0, 15
-; CHECK-NEXT: addi.d $sp, $sp, 16
+; CHECK-NEXT: vmskltz.w $vr0, $vr0
+; CHECK-NEXT: vpickve2gr.hu $a0, $vr0, 0
; CHECK-NEXT: ret
entry:
%1 = icmp sle <4 x i32> %a, splat (i32 -1)
@@ -471,15 +125,8 @@ entry:
define i2 @vmsk_sle_allones_i64(<2 x i64 > %a) {
; CHECK-LABEL: vmsk_sle_allones_i64:
; CHECK: # %bb.0: # %entry
-; CHECK-NEXT: addi.d $sp, $sp, -16
-; CHECK-NEXT: .cfi_def_cfa_offset 16
-; CHECK-NEXT: vslei.d $vr0, $vr0, -1
-; CHECK-NEXT: vpickve2gr.d $a0, $vr0, 0
-; CHECK-NEXT: vpickve2gr.d $a1, $vr0, 1
-; CHECK-NEXT: slli.d $a1, $a1, 1
-; CHECK-NEXT: sub.d $a0, $a1, $a0
-; CHECK-NEXT: andi $a0, $a0, 3
-; CHECK-NEXT: addi.d $sp, $sp, 16
+; CHECK-NEXT: vmskltz.d $vr0, $vr0
+; CHECK-NEXT: vpickve2gr.hu $a0, $vr0, 0
; CHECK-NEXT: ret
entry:
%1 = icmp sle <2 x i64> %a, splat (i64 -1)
@@ -490,63 +137,8 @@ entry:
define i16 @vmsk_ne_allzeros_i8(<16 x i8 > %a) {
; CHECK-LABEL: vmsk_ne_allzeros_i8:
; CHECK: # %bb.0: # %entry
-; CHECK-NEXT: addi.d $sp, $sp, -16
-; CHECK-NEXT: .cfi_def_cfa_offset 16
-; CHECK-NEXT: vseqi.b $vr0, $vr0, 0
-; CHECK-NEXT: vxori.b $vr0, $vr0, 255
-; CHECK-NEXT: vpickve2gr.b $a0, $vr0, 0
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 1
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: bstrins.d $a0, $a1, 63, 1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 2
-; CHECK-NEXT: bstrins.d $a0, $a1, 2, 2
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 3
-; CHECK-NEXT: bstrins.d $a0, $a1, 3, 3
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 4
-; CHECK-NEXT: bstrins.d $a0, $a1, 4, 4
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 5
-; CHECK-NEXT: bstrins.d $a0, $a1, 5, 5
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 6
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 6
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 7
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 7
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 8
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 8
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 9
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 9
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 10
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 10
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 11
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 11
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 12
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 12
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 13
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 13
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 14
-; CHECK-NEXT: andi $a1, $a1, 1
-; CHECK-NEXT: slli.d $a1, $a1, 14
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: vpickve2gr.b $a1, $vr0, 15
-; CHECK-NEXT: slli.d $a1, $a1, 15
-; CHECK-NEXT: or $a0, $a0, $a1
-; CHECK-NEXT: bstrpick.d $a0, $a0, 15, 0
-; CHECK-NEXT: addi.d $sp, $sp, 16
+; CHECK-NEXT: vmsknz.b $vr0, $vr0
+; CHECK-NEXT: vpickve2gr.hu $a0, $vr0, 0
; CHECK-NEXT: ret
entry:
%1 = icmp ne <16 x i8> %a, splat (i8 0)
More information about the llvm-commits
mailing list