[RISCV][VP] Add basic RVV codegen for vp.icmp

[frasercrmck]

Summary

This patch adds the minimum required to successfully lower vp.icmp via
the new ISD::VP_SETCC node to RVV instructions.

Regular ISD::SETCC goes through a lot of canonicalization which targets
may rely on which has not hereto been ported to VP_SETCC. It also
supports expansion of individual condition codes and a non-boolean
return type. Support for all of that will follow in later patches.

In the case of RVV this largely isn't a problem as the vector integer
comparison instructions are plentiful enough that it can lower all
VP_SETCC nodes on legal integer vectors except for boolean vectors,
which regular SETCC folds away immediately into logical operations.

Floating-point VP_SETCC operations aren't as well supported in RVV and
the backend relies on condition code expansion, so support for those
operations will come in later patches.

Portions of this code were taken from the VP reference patches.

[frasercrmck]

Note I'm only testing i8, i32 and i64 with vscale x 1 and vscale x 8 because the test was getting silly. I thought it was better to test all the CCs and various LHS/RHS swappings than each vector type explicitly. If there are concerns about that then let me know!

Oh and I'll update the patch to test fixed-length vectors, which I'd forgotten.

[frasercrmck]

• add fixed-length tests

[rogfer01]

llvm/include/llvm/IR/VPIntrinsics.def

312

Now that we're at it, could we replace this block with the following one?

BEGIN_REGISTER_VP_INTRINSIC(vp_icmp, 3, 4)
VP_PROPERTY_FUNCTIONAL_OPC(ICmp)
VP_PROPERTY_CMP(2, false)
END_REGISTER_VP_INTRINSIC(vp_icmp)

Should be equivalent and avoids defining an unnecessary ISD::VP_ICMP which we are not going to use.

(A similar thing can be done for VP_FCMP but we can do that in the patch with the lowering)

[craig.topper]

llvm/include/llvm/CodeGen/SelectionDAG.h

1081

This could be stricter. Everything needs to be a vector.

llvm/lib/CodeGen/SelectionDAG/SelectionDAGBuilder.cpp

7521

Is the vp.fcmp intrinsic considered an FPMathOperator or not? I wouldn't expect it to be dynamic. I'm guess it's not because it doesn't return an FP type which is what the dyn_cast is checking

[frasercrmck]

llvm/include/llvm/CodeGen/SelectionDAG.h

1081

Yep, good catch.

llvm/include/llvm/IR/VPIntrinsics.def

312

Good shout, thanks.

llvm/lib/CodeGen/SelectionDAG/SelectionDAGBuilder.cpp

7521

I'd imagine that vp.fcmp should in theory be as much a FPMathOperator as fcmp is, but you're right about this never actually being a FPMathOperator due to the return type. I copied this particular code from the reference patch so I don't know if @simoll has other ideas. Looking again at the reference patch, I notice a change to classof to support VP intrinsics:

// Judge based on function.
auto *VPIntrin = dyn_cast<VPIntrinsic>(V);
if (VPIntrin) {
  auto OCOpt = VPIntrin->getFunctionalOpcode();
  Opcode = OCOpt ? *OCOpt : (unsigned) Instruction::Call;
}

Also fcmp being a FPMathOperator seems to be questionable:

// FIXME: To clean up and correct the semantics of fast-math-flags, FCmp
//        should not be treated as a math op, but the other opcodes should.
//        This would make things consistent with Select/PHI (FP value type
//        determines whether they are math ops and, therefore, capable of
//        having fast-math-flags).
case Instruction::FCmp:
  return true;

My initial reaction is to remove this bit for now and come back to it? This patch is only nominally supporting icmp anyway. What do you think?

[frasercrmck]

rebase & address feedback

[craig.topper]

llvm/lib/CodeGen/SelectionDAG/SelectionDAGBuilder.cpp

7521

Let's remove it from this patch.

That FIXME on fcmp fast math flgs has been there for a while. I don't think we're close to fixing it yet.

[frasercrmck]

• drop impossible cast to FPMathOperator (with FIXME)

[frasercrmck]

llvm/lib/CodeGen/SelectionDAG/SelectionDAGBuilder.cpp

7521

Alright, sounds good.

[frasercrmck]

• rebase on top of already-committed setcc pats

[craig.topper]

LGTM

llvm/include/llvm/CodeGen/SelectionDAG.h

SDValue getSetCC(const SDLoc &DL, EVT VT, SDValue LHS, SDValue RHS,
     assert(Cond != ISD::SETCC_INVALID &&
            "Cannot create a setCC of an invalid node.");
     if (Chain)
       return getNode(IsSignaling ? ISD::STRICT_FSETCCS : ISD::STRICT_FSETCC, DL,
                      {VT, MVT::Other}, {Chain, LHS, RHS, getCondCode(Cond)});
     return getNode(ISD::SETCC, DL, VT, LHS, RHS, getCondCode(Cond));
   }
 
+  /// Helper function to make it easier to build VP_SETCCs if you just have an
+  /// ISD::CondCode instead of an SDValue.
+  SDValue getSetCCVP(const SDLoc &DL, EVT VT, SDValue LHS, SDValue RHS,
+                     ISD::CondCode Cond, SDValue Mask, SDValue EVL) {
+    assert(LHS.getValueType().isVector() && RHS.getValueType().isVector() &&
+           "Cannot compare scalars");
+    assert(Cond != ISD::SETCC_INVALID &&
+           "Cannot create a setCC of an invalid node.");
+    return getNode(ISD::VP_SETCC, DL, VT, LHS, RHS, getCondCode(Cond), Mask,
+                   EVL);
+  }
+
   /// Helper function to make it easier to build Select's if you just have
   /// operands and don't want to check for vector.
   SDValue getSelect(const SDLoc &DL, EVT VT, SDValue Cond, SDValue LHS,
                     SDValue RHS) {
     assert(LHS.getValueType() == RHS.getValueType() &&
            "Cannot use select on differing types");
     assert(VT.isVector() == LHS.getValueType().isVector() &&
            "Cannot mix vectors and scalars");

llvm/include/llvm/IR/VPIntrinsics.def

 // llvm.vp.fcmp(x,y,cc,mask,vlen)
 BEGIN_REGISTER_VP(vp_fcmp, 3, 4, VP_FCMP, -1)
 VP_PROPERTY_FUNCTIONAL_OPC(FCmp)
 VP_PROPERTY_CMP(2, true)
 VP_PROPERTY_CONSTRAINEDFP(0, 1, experimental_constrained_fcmp)
 END_REGISTER_VP(vp_fcmp, VP_FCMP)
 
 // llvm.vp.icmp(x,y,cc,mask,vlen)
-BEGIN_REGISTER_VP(vp_icmp, 3, 4, VP_ICMP, -1)
+BEGIN_REGISTER_VP_INTRINSIC(vp_icmp, 3, 4)
 VP_PROPERTY_FUNCTIONAL_OPC(ICmp)
 VP_PROPERTY_CMP(2, false)
-END_REGISTER_VP(vp_icmp, VP_ICMP)
+END_REGISTER_VP_INTRINSIC(vp_icmp)
+
+// VP_SETCC (ISel only)
+BEGIN_REGISTER_VP_SDNODE(VP_SETCC, -1, vp_setcc, 3, 4)
+END_REGISTER_VP_SDNODE(VP_SETCC)
 
 ///// } Comparisons
 
 ///// Memory Operations {
 // llvm.vp.store(val,ptr,mask,vlen)
 BEGIN_REGISTER_VP_INTRINSIC(vp_store, 2, 3)
 // chain = VP_STORE chain,val,base,offset,mask,evl
 BEGIN_REGISTER_VP_SDNODE(VP_STORE, 0, vp_store, 4, 5)

llvm/lib/CodeGen/SelectionDAG/LegalizeDAG.cpp

#endif
   case ISD::ATOMIC_STORE:
     Action = TLI.getOperationAction(Node->getOpcode(),
                                     Node->getOperand(2).getValueType());
     break;
   case ISD::SELECT_CC:
   case ISD::STRICT_FSETCC:
   case ISD::STRICT_FSETCCS:
   case ISD::SETCC:
+  case ISD::VP_SETCC:
   case ISD::BR_CC: {
-    unsigned CCOperand = Node->getOpcode() == ISD::SELECT_CC ? 4 :
-                         Node->getOpcode() == ISD::STRICT_FSETCC ? 3 :
-                         Node->getOpcode() == ISD::STRICT_FSETCCS ? 3 :
-                         Node->getOpcode() == ISD::SETCC ? 2 : 1;
-    unsigned CompareOperand = Node->getOpcode() == ISD::BR_CC ? 2 :
-                              Node->getOpcode() == ISD::STRICT_FSETCC ? 1 :
-                              Node->getOpcode() == ISD::STRICT_FSETCCS ? 1 : 0;
+    unsigned Opc = Node->getOpcode();
+    unsigned CCOperand = Opc == ISD::SELECT_CC                         ? 4
+                         : Opc == ISD::STRICT_FSETCC                   ? 3
+                         : Opc == ISD::STRICT_FSETCCS                  ? 3
+                         : (Opc == ISD::SETCC || Opc == ISD::VP_SETCC) ? 2
+                                                                       : 1;
+    unsigned CompareOperand = Opc == ISD::BR_CC            ? 2
+                              : Opc == ISD::STRICT_FSETCC  ? 1
+                              : Opc == ISD::STRICT_FSETCCS ? 1
+                                                           : 0;
     MVT OpVT = Node->getOperand(CompareOperand).getSimpleValueType();
     ISD::CondCode CCCode =
         cast<CondCodeSDNode>(Node->getOperand(CCOperand))->get();
     Action = TLI.getCondCodeAction(CCCode, OpVT);
     if (Action == TargetLowering::Legal) {
       if (Node->getOpcode() == ISD::SELECT_CC)
         Action = TLI.getOperationAction(Node->getOpcode(),
                                         Node->getValueType(0));

llvm/lib/CodeGen/SelectionDAG/SelectionDAGBuilder.h

private:
   void visitVPLoadGather(const VPIntrinsic &VPIntrin, EVT VT,
                          SmallVector<SDValue, 7> &OpValues, bool IsGather);
   void visitVPStoreScatter(const VPIntrinsic &VPIntrin,
                            SmallVector<SDValue, 7> &OpValues, bool IsScatter);
   void visitVPStridedLoad(const VPIntrinsic &VPIntrin, EVT VT,
                           SmallVectorImpl<SDValue> &OpValues);
   void visitVPStridedStore(const VPIntrinsic &VPIntrin,
                            SmallVectorImpl<SDValue> &OpValues);
+  void visitVPCmp(const VPCmpIntrinsic &VPIntrin);
   void visitVectorPredicationIntrinsic(const VPIntrinsic &VPIntrin);
 
   void visitVAStart(const CallInst &I);
   void visitVAArg(const VAArgInst &I);
   void visitVAEnd(const CallInst &I);
   void visitVACopy(const CallInst &I);
   void visitStackmap(const CallInst &I);
   void visitPatchpoint(const CallBase &CB, const BasicBlock *EHPadBB = nullptr);

llvm/lib/CodeGen/SelectionDAG/SelectionDAGBuilder.cpp

#include "llvm/IR/ConstrainedOps.def"
   pushOutChain(Result, EB);
 
   SDValue FPResult = Result.getValue(0);
   setValue(&FPI, FPResult);
 }
 
 static unsigned getISDForVPIntrinsic(const VPIntrinsic &VPIntrin) {
   Optional<unsigned> ResOPC;
-  switch (VPIntrin.getIntrinsicID()) {
+  auto IID = VPIntrin.getIntrinsicID();
+  // vp.fcmp and vp.icmp are handled specially
+  if (IID == Intrinsic::vp_fcmp || IID == Intrinsic::vp_icmp)
+    return ISD::VP_SETCC;
+
+  switch (IID) {
 #define BEGIN_REGISTER_VP_INTRINSIC(VPID, ...) case Intrinsic::VPID:
 #define BEGIN_REGISTER_VP_SDNODE(VPSD, ...) ResOPC = ISD::VPSD;
 #define END_REGISTER_VP_INTRINSIC(VPID) break;
 #include "llvm/IR/VPIntrinsics.def"
   }
 
   if (!ResOPC.hasValue())
     llvm_unreachable(
SDValue ST = DAG.getStridedStoreVP(
       DAG.getUNDEF(OpValues[1].getValueType()), OpValues[2], OpValues[3],
       OpValues[4], VT, MMO, ISD::UNINDEXED, /*IsTruncating*/ false,
       /*IsCompressing*/ false);
 
   DAG.setRoot(ST);
   setValue(&VPIntrin, ST);
 }
 
+void SelectionDAGBuilder::visitVPCmp(const VPCmpIntrinsic &VPIntrin) {
+  const TargetLowering &TLI = DAG.getTargetLoweringInfo();
+  SDLoc DL = getCurSDLoc();
+
+  ISD::CondCode Condition;
+  CmpInst::Predicate CondCode = VPIntrin.getPredicate();
+  bool IsFP = VPIntrin.getOperand(0)->getType()->isFPOrFPVectorTy();
+  if (IsFP) {
+    // FIXME: Regular fcmps are FPMathOperators which may have fast-math (nnan)
+    // flags, but calls that don't return floating-point types can't be
+    // FPMathOperators, like vp.fcmp. This affects constrained fcmp too.
+    Condition = getFCmpCondCode(CondCode);
+    if (TM.Options.NoNaNsFPMath)
+      Condition = getFCmpCodeWithoutNaN(Condition);
+  } else {
+    Condition = getICmpCondCode(CondCode);
+  }
+
+  SDValue Op1 = getValue(VPIntrin.getOperand(0));
+  SDValue Op2 = getValue(VPIntrin.getOperand(1));
+  // #2 is the condition code
+  SDValue MaskOp = getValue(VPIntrin.getOperand(3));
+  SDValue EVL = getValue(VPIntrin.getOperand(4));
+  MVT EVLParamVT = TLI.getVPExplicitVectorLengthTy();
+  assert(EVLParamVT.isScalarInteger() && EVLParamVT.bitsGE(MVT::i32) &&
+         "Unexpected target EVL type");
+  EVL = DAG.getNode(ISD::ZERO_EXTEND, DL, EVLParamVT, EVL);
+
+  EVT DestVT = DAG.getTargetLoweringInfo().getValueType(DAG.getDataLayout(),
+                                                        VPIntrin.getType());
+  setValue(&VPIntrin,
+           DAG.getSetCCVP(DL, DestVT, Op1, Op2, Condition, MaskOp, EVL));
+}
+
 void SelectionDAGBuilder::visitVectorPredicationIntrinsic(
     const VPIntrinsic &VPIntrin) {
   SDLoc DL = getCurSDLoc();
   unsigned Opcode = getISDForVPIntrinsic(VPIntrin);
 
+  auto IID = VPIntrin.getIntrinsicID();
+
+  if (const auto *CmpI = dyn_cast<VPCmpIntrinsic>(&VPIntrin))
+    return visitVPCmp(*CmpI);
+
   SmallVector<EVT, 4> ValueVTs;
   const TargetLowering &TLI = DAG.getTargetLoweringInfo();
   ComputeValueVTs(TLI, DAG.getDataLayout(), VPIntrin.getType(), ValueVTs);
   SDVTList VTs = DAG.getVTList(ValueVTs);
 
-  auto EVLParamPos =
-      VPIntrinsic::getVectorLengthParamPos(VPIntrin.getIntrinsicID());
+  auto EVLParamPos = VPIntrinsic::getVectorLengthParamPos(IID);
 
   MVT EVLParamVT = TLI.getVPExplicitVectorLengthTy();
   assert(EVLParamVT.isScalarInteger() && EVLParamVT.bitsGE(MVT::i32) &&
          "Unexpected target EVL type");
 
   // Request operands.
   SmallVector<SDValue, 7> OpValues;
   for (unsigned I = 0; I < VPIntrin.arg_size(); ++I) {

llvm/lib/Target/RISCV/RISCVISelLowering.cpp

static const unsigned IntegerVPOps[] = {
         ISD::VP_ADD,         ISD::VP_SUB,         ISD::VP_MUL,
         ISD::VP_SDIV,        ISD::VP_UDIV,        ISD::VP_SREM,
         ISD::VP_UREM,        ISD::VP_AND,         ISD::VP_OR,
         ISD::VP_XOR,         ISD::VP_ASHR,        ISD::VP_LSHR,
         ISD::VP_SHL,         ISD::VP_REDUCE_ADD,  ISD::VP_REDUCE_AND,
         ISD::VP_REDUCE_OR,   ISD::VP_REDUCE_XOR,  ISD::VP_REDUCE_SMAX,
         ISD::VP_REDUCE_SMIN, ISD::VP_REDUCE_UMAX, ISD::VP_REDUCE_UMIN,
         ISD::VP_MERGE,       ISD::VP_SELECT,      ISD::VP_FPTOSI,
-        ISD::VP_FPTOUI};
+        ISD::VP_FPTOUI,      ISD::VP_SETCC};
 
     static const unsigned FloatingPointVPOps[] = {
-        ISD::VP_FADD,        ISD::VP_FSUB,        ISD::VP_FMUL,
-        ISD::VP_FDIV,        ISD::VP_FNEG,        ISD::VP_FMA,
-        ISD::VP_REDUCE_FADD, ISD::VP_REDUCE_SEQ_FADD, ISD::VP_REDUCE_FMIN,
-        ISD::VP_REDUCE_FMAX, ISD::VP_MERGE,       ISD::VP_SELECT,
-        ISD::VP_SITOFP,      ISD::VP_UITOFP};
+        ISD::VP_FADD,        ISD::VP_FSUB,
+        ISD::VP_FMUL,        ISD::VP_FDIV,
+        ISD::VP_FNEG,        ISD::VP_FMA,
+        ISD::VP_REDUCE_FADD, ISD::VP_REDUCE_SEQ_FADD,
+        ISD::VP_REDUCE_FMIN, ISD::VP_REDUCE_FMAX,
+        ISD::VP_MERGE,       ISD::VP_SELECT,
+        ISD::VP_SITOFP,      ISD::VP_UITOFP,
+        ISD::VP_SETCC};
 
     if (!Subtarget.is64Bit()) {
       // We must custom-lower certain vXi64 operations on RV32 due to the vector
       // element type being illegal.
       setOperationAction(ISD::INSERT_VECTOR_ELT, MVT::i64, Custom);
       setOperationAction(ISD::EXTRACT_VECTOR_ELT, MVT::i64, Custom);
 
       setOperationAction(ISD::VECREDUCE_ADD, MVT::i64, Custom);
if (Subtarget.useRVVForFixedLengthVectors()) {
           setOperationAction(ISD::VP_OR, VT, Custom);
           setOperationAction(ISD::VP_XOR, VT, Custom);
           setOperationAction(ISD::AND, VT, Custom);
           setOperationAction(ISD::OR, VT, Custom);
           setOperationAction(ISD::XOR, VT, Custom);
 
           setOperationAction(ISD::VP_FPTOSI, VT, Custom);
           setOperationAction(ISD::VP_FPTOUI, VT, Custom);
+          setOperationAction(ISD::VP_SETCC, VT, Custom);
           continue;
         }
 
         // Make SPLAT_VECTOR Legal so DAGCombine will convert splat vectors to
         // it before type legalization for i64 vectors on RV32. It will then be
         // type legalized to SPLAT_VECTOR_PARTS which we need to Custom handle.
         // FIXME: Use SPLAT_VECTOR for all types? DAGCombine probably needs
         // improvements first.
SDValue RISCVTargetLowering::LowerOperation(SDValue Op,
   case ISD::VP_FPTOSI:
     return lowerVPFPIntConvOp(Op, DAG, RISCVISD::FP_TO_SINT_VL);
   case ISD::VP_FPTOUI:
     return lowerVPFPIntConvOp(Op, DAG, RISCVISD::FP_TO_UINT_VL);
   case ISD::VP_SITOFP:
     return lowerVPFPIntConvOp(Op, DAG, RISCVISD::SINT_TO_FP_VL);
   case ISD::VP_UITOFP:
     return lowerVPFPIntConvOp(Op, DAG, RISCVISD::UINT_TO_FP_VL);
+  case ISD::VP_SETCC:
+    return lowerVPOp(Op, DAG, RISCVISD::SETCC_VL);
   }
 }
 
 static SDValue getTargetNode(GlobalAddressSDNode *N, SDLoc DL, EVT Ty,
                              SelectionDAG &DAG, unsigned Flags) {
   return DAG.getTargetGlobalAddress(N->getGlobal(), DL, Ty, 0, Flags);
 }
 

llvm/lib/Target/RISCV/RISCVInstrInfoVVLPatterns.td

multiclass VPatIntegerSetCCVL_VV<VTypeInfo vti, string instruction_name,
                                  CondCode cc> {
   def : Pat<(vti.Mask (riscv_setcc_vl (vti.Vector vti.RegClass:$rs1),
                                       vti.RegClass:$rs2, cc,
                                       (vti.Mask true_mask),
                                       VLOpFrag)),
             (!cast<Instruction>(instruction_name#"_VV_"#vti.LMul.MX)
                          vti.RegClass:$rs1, vti.RegClass:$rs2, GPR:$vl,
                          vti.Log2SEW)>;
+  def : Pat<(vti.Mask (riscv_setcc_vl (vti.Vector vti.RegClass:$rs1),
+                                      vti.RegClass:$rs2, cc,
+                                      (vti.Mask V0),
+                                      VLOpFrag)),
+            (!cast<Instruction>(instruction_name#"_VV_"#vti.LMul.MX#"_MASK")
+                         (vti.Mask (IMPLICIT_DEF)),
+                         vti.RegClass:$rs1, vti.RegClass:$rs2, (vti.Mask V0),
+                         GPR:$vl, vti.Log2SEW)>;
 }
 
 // Inherits from VPatIntegerSetCCVL_VV and adds a pattern with operands swapped.
 multiclass VPatIntegerSetCCVL_VV_Swappable<VTypeInfo vti, string instruction_name,
                                            CondCode cc, CondCode invcc> :
   VPatIntegerSetCCVL_VV<vti, instruction_name, cc> {
   def : Pat<(vti.Mask (riscv_setcc_vl (vti.Vector vti.RegClass:$rs2),
                                       vti.RegClass:$rs1, invcc,
                                       (vti.Mask true_mask),
                                       VLOpFrag)),
             (!cast<Instruction>(instruction_name#"_VV_"#vti.LMul.MX)
                          vti.RegClass:$rs1, vti.RegClass:$rs2, GPR:$vl,
                          vti.Log2SEW)>;
+  def : Pat<(vti.Mask (riscv_setcc_vl (vti.Vector vti.RegClass:$rs2),
+                                      vti.RegClass:$rs1, invcc,
+                                      (vti.Mask V0),
+                                      VLOpFrag)),
+            (!cast<Instruction>(instruction_name#"_VV_"#vti.LMul.MX#"_MASK")
+                         (vti.Mask (IMPLICIT_DEF)),
+                         vti.RegClass:$rs1, vti.RegClass:$rs2, (vti.Mask V0),
+                         GPR:$vl, vti.Log2SEW)>;
 }
 
 multiclass VPatIntegerSetCCVL_VX_Swappable<VTypeInfo vti, string instruction_name,
                                            CondCode cc, CondCode invcc> {
   defvar instruction = !cast<Instruction>(instruction_name#"_VX_"#vti.LMul.MX);
+  defvar mask_instruction = !cast<Instruction>(instruction_name#"_VX_"#vti.LMul.MX#"_MASK");
   def : Pat<(vti.Mask (riscv_setcc_vl (vti.Vector vti.RegClass:$rs1),
                                       (SplatPat (XLenVT GPR:$rs2)), cc,
                                       (vti.Mask true_mask),
                                       VLOpFrag)),
             (instruction vti.RegClass:$rs1, GPR:$rs2, GPR:$vl, vti.Log2SEW)>;
+  def : Pat<(vti.Mask (riscv_setcc_vl (vti.Vector vti.RegClass:$rs1),
+                                      (SplatPat (XLenVT GPR:$rs2)), cc,
+                                      (vti.Mask V0), VLOpFrag)),
+            (mask_instruction (vti.Mask (IMPLICIT_DEF)), vti.RegClass:$rs1, GPR:$rs2,
+                              (vti.Mask V0), GPR:$vl, vti.Log2SEW)>;
   def : Pat<(vti.Mask (riscv_setcc_vl (SplatPat (XLenVT GPR:$rs2)),
                                       (vti.Vector vti.RegClass:$rs1), invcc,
                                       (vti.Mask true_mask),
                                       VLOpFrag)),
             (instruction vti.RegClass:$rs1, GPR:$rs2, GPR:$vl, vti.Log2SEW)>;
+  def : Pat<(vti.Mask (riscv_setcc_vl (SplatPat (XLenVT GPR:$rs2)),
+                                      (vti.Vector vti.RegClass:$rs1), invcc,
+                                      (vti.Mask V0), VLOpFrag)),
+            (mask_instruction (vti.Mask (IMPLICIT_DEF)), vti.RegClass:$rs1, GPR:$rs2,
+                              (vti.Mask V0), GPR:$vl, vti.Log2SEW)>;
 }
 
 multiclass VPatIntegerSetCCVL_VI_Swappable<VTypeInfo vti, string instruction_name,
                                            CondCode cc, CondCode invcc> {
   defvar instruction = !cast<Instruction>(instruction_name#"_VI_"#vti.LMul.MX);
+  defvar mask_instruction = !cast<Instruction>(instruction_name#"_VI_"#vti.LMul.MX#"_MASK");
   def : Pat<(vti.Mask (riscv_setcc_vl (vti.Vector vti.RegClass:$rs1),
                                       (SplatPat_simm5 simm5:$rs2), cc,
                                       (vti.Mask true_mask),
                                       VLOpFrag)),
             (instruction vti.RegClass:$rs1, XLenVT:$rs2, GPR:$vl, vti.Log2SEW)>;
+  def : Pat<(vti.Mask (riscv_setcc_vl (vti.Vector vti.RegClass:$rs1),
+                                      (SplatPat_simm5 simm5:$rs2), cc,
+                                      (vti.Mask V0), VLOpFrag)),
+            (mask_instruction (vti.Mask (IMPLICIT_DEF)), vti.RegClass:$rs1, XLenVT:$rs2,
+                              (vti.Mask V0), GPR:$vl, vti.Log2SEW)>;
   def : Pat<(vti.Mask (riscv_setcc_vl (SplatPat_simm5 simm5:$rs2),
                                       (vti.Vector vti.RegClass:$rs1), invcc,
                                       (vti.Mask true_mask),
                                       VLOpFrag)),
             (instruction vti.RegClass:$rs1, simm5:$rs2, GPR:$vl, vti.Log2SEW)>;
+  def : Pat<(vti.Mask (riscv_setcc_vl (SplatPat_simm5 simm5:$rs2),
+                                      (vti.Vector vti.RegClass:$rs1), invcc,
+                                      (vti.Mask V0), VLOpFrag)),
+            (mask_instruction (vti.Mask (IMPLICIT_DEF)), vti.RegClass:$rs1, simm5:$rs2,
+                              (vti.Mask V0), GPR:$vl, vti.Log2SEW)>;
 }
 
 multiclass VPatIntegerSetCCVL_VIPlus1<VTypeInfo vti, string instruction_name,
                                       CondCode cc, ComplexPattern splatpat_kind> {
   defvar instruction = !cast<Instruction>(instruction_name#"_VI_"#vti.LMul.MX);
+  defvar mask_instruction = !cast<Instruction>(instruction_name#"_VI_"#vti.LMul.MX#"_MASK");
   def : Pat<(vti.Mask (riscv_setcc_vl (vti.Vector vti.RegClass:$rs1),
                                       (splatpat_kind simm5:$rs2), cc,
                                       (vti.Mask true_mask),
                                       VLOpFrag)),
             (instruction vti.RegClass:$rs1, (DecImm simm5:$rs2),
                          GPR:$vl, vti.Log2SEW)>;
+  def : Pat<(vti.Mask (riscv_setcc_vl (vti.Vector vti.RegClass:$rs1),
+                                      (splatpat_kind simm5:$rs2), cc,
+                                      (vti.Mask V0), VLOpFrag)),
+            (mask_instruction (vti.Mask (IMPLICIT_DEF)), vti.RegClass:$rs1,
+                              (DecImm simm5:$rs2), (vti.Mask V0),
+                              GPR:$vl, vti.Log2SEW)>;
 }
 
 multiclass VPatFPSetCCVL_VV_VF_FV<CondCode cc,
                                   string inst_name,
                                   string swapped_op_inst_name> {
   foreach fvti = AllFloatVectors in {
     def : Pat<(fvti.Mask (riscv_setcc_vl (fvti.Vector fvti.RegClass:$rs1),
                                          fvti.RegClass:$rs2,

llvm/test/CodeGen/RISCV/rvv/fixed-vectors-setcc-int-vp.ll

+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc -mtriple=riscv32 -mattr=+v -riscv-v-vector-bits-min=128 -verify-machineinstrs < %s \
+; RUN:   | FileCheck %s --check-prefixes=CHECK,RV32
+; RUN: llc -mtriple=riscv64 -mattr=+v -riscv-v-vector-bits-min=128 -verify-machineinstrs < %s \
+; RUN:   | FileCheck %s --check-prefixes=CHECK,RV64
+
+; FIXME: We're missing canonicalizations of ISD::VP_SETCC equivalent to those
+; for ISD::SETCC, e.g., splats aren't moved to the RHS.
+
+declare <8 x i1> @llvm.vp.icmp.v8i8(<8 x i8>, <8 x i8>, metadata, <8 x i1>, i32)
+
+define <8 x i1> @icmp_eq_vv_v8i8(<8 x i8> %va, <8 x i8> %vb, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_eq_vv_v8i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e8, mf2, ta, ma
+; CHECK-NEXT:    vmseq.vv v0, v8, v9, v0.t
+; CHECK-NEXT:    ret
+  %v = call <8 x i1> @llvm.vp.icmp.v8i8(<8 x i8> %va, <8 x i8> %vb, metadata !"eq", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_eq_vx_v8i8(<8 x i8> %va, i8 %b, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_eq_vx_v8i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a1, e8, mf2, ta, ma
+; CHECK-NEXT:    vmseq.vx v0, v8, a0, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x i8> poison, i8 %b, i32 0
+  %vb = shufflevector <8 x i8> %elt.head, <8 x i8> poison, <8 x i32> zeroinitializer
+  %v = call <8 x i1> @llvm.vp.icmp.v8i8(<8 x i8> %va, <8 x i8> %vb, metadata !"eq", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_eq_vx_swap_v8i8(<8 x i8> %va, i8 %b, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_eq_vx_swap_v8i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a1, e8, mf2, ta, ma
+; CHECK-NEXT:    vmseq.vx v0, v8, a0, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x i8> poison, i8 %b, i32 0
+  %vb = shufflevector <8 x i8> %elt.head, <8 x i8> poison, <8 x i32> zeroinitializer
+  %v = call <8 x i1> @llvm.vp.icmp.v8i8(<8 x i8> %vb, <8 x i8> %va, metadata !"eq", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_eq_vi_v8i8(<8 x i8> %va, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_eq_vi_v8i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e8, mf2, ta, ma
+; CHECK-NEXT:    vmseq.vi v0, v8, 4, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x i8> poison, i8 4, i32 0
+  %vb = shufflevector <8 x i8> %elt.head, <8 x i8> poison, <8 x i32> zeroinitializer
+  %v = call <8 x i1> @llvm.vp.icmp.v8i8(<8 x i8> %va, <8 x i8> %vb, metadata !"eq", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_eq_vi_swap_v8i8(<8 x i8> %va, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_eq_vi_swap_v8i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e8, mf2, ta, ma
+; CHECK-NEXT:    vmseq.vi v0, v8, 4, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x i8> poison, i8 4, i32 0
+  %vb = shufflevector <8 x i8> %elt.head, <8 x i8> poison, <8 x i32> zeroinitializer
+  %v = call <8 x i1> @llvm.vp.icmp.v8i8(<8 x i8> %vb, <8 x i8> %va, metadata !"eq", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_ne_vv_v8i8(<8 x i8> %va, <8 x i8> %vb, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ne_vv_v8i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e8, mf2, ta, ma
+; CHECK-NEXT:    vmsne.vv v0, v8, v9, v0.t
+; CHECK-NEXT:    ret
+  %v = call <8 x i1> @llvm.vp.icmp.v8i8(<8 x i8> %va, <8 x i8> %vb, metadata !"ne", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_ne_vx_v8i8(<8 x i8> %va, i8 %b, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ne_vx_v8i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a1, e8, mf2, ta, ma
+; CHECK-NEXT:    vmsne.vx v0, v8, a0, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x i8> poison, i8 %b, i32 0
+  %vb = shufflevector <8 x i8> %elt.head, <8 x i8> poison, <8 x i32> zeroinitializer
+  %v = call <8 x i1> @llvm.vp.icmp.v8i8(<8 x i8> %va, <8 x i8> %vb, metadata !"ne", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_ne_vx_swap_v8i8(<8 x i8> %va, i8 %b, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ne_vx_swap_v8i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a1, e8, mf2, ta, ma
+; CHECK-NEXT:    vmsne.vx v0, v8, a0, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x i8> poison, i8 %b, i32 0
+  %vb = shufflevector <8 x i8> %elt.head, <8 x i8> poison, <8 x i32> zeroinitializer
+  %v = call <8 x i1> @llvm.vp.icmp.v8i8(<8 x i8> %vb, <8 x i8> %va, metadata !"ne", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_ne_vi_v8i8(<8 x i8> %va, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ne_vi_v8i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e8, mf2, ta, ma
+; CHECK-NEXT:    vmsne.vi v0, v8, 4, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x i8> poison, i8 4, i32 0
+  %vb = shufflevector <8 x i8> %elt.head, <8 x i8> poison, <8 x i32> zeroinitializer
+  %v = call <8 x i1> @llvm.vp.icmp.v8i8(<8 x i8> %va, <8 x i8> %vb, metadata !"ne", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_ne_vi_swap_v8i8(<8 x i8> %va, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ne_vi_swap_v8i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e8, mf2, ta, ma
+; CHECK-NEXT:    vmsne.vi v0, v8, 4, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x i8> poison, i8 4, i32 0
+  %vb = shufflevector <8 x i8> %elt.head, <8 x i8> poison, <8 x i32> zeroinitializer
+  %v = call <8 x i1> @llvm.vp.icmp.v8i8(<8 x i8> %vb, <8 x i8> %va, metadata !"ne", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_ugt_vv_v8i8(<8 x i8> %va, <8 x i8> %vb, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ugt_vv_v8i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e8, mf2, ta, ma
+; CHECK-NEXT:    vmsltu.vv v0, v9, v8, v0.t
+; CHECK-NEXT:    ret
+  %v = call <8 x i1> @llvm.vp.icmp.v8i8(<8 x i8> %va, <8 x i8> %vb, metadata !"ugt", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_ugt_vx_v8i8(<8 x i8> %va, i8 %b, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ugt_vx_v8i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a1, e8, mf2, ta, ma
+; CHECK-NEXT:    vmsgtu.vx v0, v8, a0, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x i8> poison, i8 %b, i32 0
+  %vb = shufflevector <8 x i8> %elt.head, <8 x i8> poison, <8 x i32> zeroinitializer
+  %v = call <8 x i1> @llvm.vp.icmp.v8i8(<8 x i8> %va, <8 x i8> %vb, metadata !"ugt", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_ugt_vx_swap_v8i8(<8 x i8> %va, i8 %b, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ugt_vx_swap_v8i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a1, e8, mf2, ta, ma
+; CHECK-NEXT:    vmsltu.vx v0, v8, a0, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x i8> poison, i8 %b, i32 0
+  %vb = shufflevector <8 x i8> %elt.head, <8 x i8> poison, <8 x i32> zeroinitializer
+  %v = call <8 x i1> @llvm.vp.icmp.v8i8(<8 x i8> %vb, <8 x i8> %va, metadata !"ugt", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_ugt_vi_v8i8(<8 x i8> %va, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ugt_vi_v8i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e8, mf2, ta, ma
+; CHECK-NEXT:    vmsgtu.vi v0, v8, 4, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x i8> poison, i8 4, i32 0
+  %vb = shufflevector <8 x i8> %elt.head, <8 x i8> poison, <8 x i32> zeroinitializer
+  %v = call <8 x i1> @llvm.vp.icmp.v8i8(<8 x i8> %va, <8 x i8> %vb, metadata !"ugt", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_ugt_vi_swap_v8i8(<8 x i8> %va, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ugt_vi_swap_v8i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    li a1, 4
+; CHECK-NEXT:    vsetvli zero, a0, e8, mf2, ta, ma
+; CHECK-NEXT:    vmsltu.vx v0, v8, a1, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x i8> poison, i8 4, i32 0
+  %vb = shufflevector <8 x i8> %elt.head, <8 x i8> poison, <8 x i32> zeroinitializer
+  %v = call <8 x i1> @llvm.vp.icmp.v8i8(<8 x i8> %vb, <8 x i8> %va, metadata !"ugt", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_uge_vv_v8i8(<8 x i8> %va, <8 x i8> %vb, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_uge_vv_v8i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e8, mf2, ta, ma
+; CHECK-NEXT:    vmsleu.vv v0, v9, v8, v0.t
+; CHECK-NEXT:    ret
+  %v = call <8 x i1> @llvm.vp.icmp.v8i8(<8 x i8> %va, <8 x i8> %vb, metadata !"uge", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_uge_vx_v8i8(<8 x i8> %va, i8 %b, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_uge_vx_v8i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli zero, 8, e8, mf2, ta, mu
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, a1, e8, mf2, ta, ma
+; CHECK-NEXT:    vmsleu.vv v0, v9, v8, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x i8> poison, i8 %b, i32 0
+  %vb = shufflevector <8 x i8> %elt.head, <8 x i8> poison, <8 x i32> zeroinitializer
+  %v = call <8 x i1> @llvm.vp.icmp.v8i8(<8 x i8> %va, <8 x i8> %vb, metadata !"uge", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_uge_vx_swap_v8i8(<8 x i8> %va, i8 %b, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_uge_vx_swap_v8i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a1, e8, mf2, ta, ma
+; CHECK-NEXT:    vmsleu.vx v0, v8, a0, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x i8> poison, i8 %b, i32 0
+  %vb = shufflevector <8 x i8> %elt.head, <8 x i8> poison, <8 x i32> zeroinitializer
+  %v = call <8 x i1> @llvm.vp.icmp.v8i8(<8 x i8> %vb, <8 x i8> %va, metadata !"uge", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_uge_vi_v8i8(<8 x i8> %va, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_uge_vi_v8i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e8, mf2, ta, ma
+; CHECK-NEXT:    vmsgtu.vi v0, v8, 3, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x i8> poison, i8 4, i32 0
+  %vb = shufflevector <8 x i8> %elt.head, <8 x i8> poison, <8 x i32> zeroinitializer
+  %v = call <8 x i1> @llvm.vp.icmp.v8i8(<8 x i8> %va, <8 x i8> %vb, metadata !"uge", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_uge_vi_swap_v8i8(<8 x i8> %va, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_uge_vi_swap_v8i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e8, mf2, ta, ma
+; CHECK-NEXT:    vmsleu.vi v0, v8, 4, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x i8> poison, i8 4, i32 0
+  %vb = shufflevector <8 x i8> %elt.head, <8 x i8> poison, <8 x i32> zeroinitializer
+  %v = call <8 x i1> @llvm.vp.icmp.v8i8(<8 x i8> %vb, <8 x i8> %va, metadata !"uge", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_ult_vv_v8i8(<8 x i8> %va, <8 x i8> %vb, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ult_vv_v8i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e8, mf2, ta, ma
+; CHECK-NEXT:    vmsltu.vv v0, v8, v9, v0.t
+; CHECK-NEXT:    ret
+  %v = call <8 x i1> @llvm.vp.icmp.v8i8(<8 x i8> %va, <8 x i8> %vb, metadata !"ult", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_ult_vx_v8i8(<8 x i8> %va, i8 %b, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ult_vx_v8i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a1, e8, mf2, ta, ma
+; CHECK-NEXT:    vmsltu.vx v0, v8, a0, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x i8> poison, i8 %b, i32 0
+  %vb = shufflevector <8 x i8> %elt.head, <8 x i8> poison, <8 x i32> zeroinitializer
+  %v = call <8 x i1> @llvm.vp.icmp.v8i8(<8 x i8> %va, <8 x i8> %vb, metadata !"ult", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_ult_vx_swap_v8i8(<8 x i8> %va, i8 %b, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ult_vx_swap_v8i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a1, e8, mf2, ta, ma
+; CHECK-NEXT:    vmsgtu.vx v0, v8, a0, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x i8> poison, i8 %b, i32 0
+  %vb = shufflevector <8 x i8> %elt.head, <8 x i8> poison, <8 x i32> zeroinitializer
+  %v = call <8 x i1> @llvm.vp.icmp.v8i8(<8 x i8> %vb, <8 x i8> %va, metadata !"ult", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_ult_vi_v8i8(<8 x i8> %va, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ult_vi_v8i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e8, mf2, ta, ma
+; CHECK-NEXT:    vmsleu.vi v0, v8, 3, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x i8> poison, i8 4, i32 0
+  %vb = shufflevector <8 x i8> %elt.head, <8 x i8> poison, <8 x i32> zeroinitializer
+  %v = call <8 x i1> @llvm.vp.icmp.v8i8(<8 x i8> %va, <8 x i8> %vb, metadata !"ult", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_ult_vi_swap_v8i8(<8 x i8> %va, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ult_vi_swap_v8i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e8, mf2, ta, ma
+; CHECK-NEXT:    vmsgtu.vi v0, v8, 4, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x i8> poison, i8 4, i32 0
+  %vb = shufflevector <8 x i8> %elt.head, <8 x i8> poison, <8 x i32> zeroinitializer
+  %v = call <8 x i1> @llvm.vp.icmp.v8i8(<8 x i8> %vb, <8 x i8> %va, metadata !"ult", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_sgt_vv_v8i8(<8 x i8> %va, <8 x i8> %vb, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sgt_vv_v8i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e8, mf2, ta, ma
+; CHECK-NEXT:    vmslt.vv v0, v9, v8, v0.t
+; CHECK-NEXT:    ret
+  %v = call <8 x i1> @llvm.vp.icmp.v8i8(<8 x i8> %va, <8 x i8> %vb, metadata !"sgt", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_sgt_vx_v8i8(<8 x i8> %va, i8 %b, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sgt_vx_v8i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a1, e8, mf2, ta, ma
+; CHECK-NEXT:    vmsgt.vx v0, v8, a0, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x i8> poison, i8 %b, i32 0
+  %vb = shufflevector <8 x i8> %elt.head, <8 x i8> poison, <8 x i32> zeroinitializer
+  %v = call <8 x i1> @llvm.vp.icmp.v8i8(<8 x i8> %va, <8 x i8> %vb, metadata !"sgt", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_sgt_vx_swap_v8i8(<8 x i8> %va, i8 %b, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sgt_vx_swap_v8i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a1, e8, mf2, ta, ma
+; CHECK-NEXT:    vmslt.vx v0, v8, a0, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x i8> poison, i8 %b, i32 0
+  %vb = shufflevector <8 x i8> %elt.head, <8 x i8> poison, <8 x i32> zeroinitializer
+  %v = call <8 x i1> @llvm.vp.icmp.v8i8(<8 x i8> %vb, <8 x i8> %va, metadata !"sgt", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_sgt_vi_v8i8(<8 x i8> %va, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sgt_vi_v8i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e8, mf2, ta, ma
+; CHECK-NEXT:    vmsgt.vi v0, v8, 4, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x i8> poison, i8 4, i32 0
+  %vb = shufflevector <8 x i8> %elt.head, <8 x i8> poison, <8 x i32> zeroinitializer
+  %v = call <8 x i1> @llvm.vp.icmp.v8i8(<8 x i8> %va, <8 x i8> %vb, metadata !"sgt", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_sgt_vi_swap_v8i8(<8 x i8> %va, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sgt_vi_swap_v8i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    li a1, 4
+; CHECK-NEXT:    vsetvli zero, a0, e8, mf2, ta, ma
+; CHECK-NEXT:    vmslt.vx v0, v8, a1, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x i8> poison, i8 4, i32 0
+  %vb = shufflevector <8 x i8> %elt.head, <8 x i8> poison, <8 x i32> zeroinitializer
+  %v = call <8 x i1> @llvm.vp.icmp.v8i8(<8 x i8> %vb, <8 x i8> %va, metadata !"sgt", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_sge_vv_v8i8(<8 x i8> %va, <8 x i8> %vb, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sge_vv_v8i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e8, mf2, ta, ma
+; CHECK-NEXT:    vmsle.vv v0, v9, v8, v0.t
+; CHECK-NEXT:    ret
+  %v = call <8 x i1> @llvm.vp.icmp.v8i8(<8 x i8> %va, <8 x i8> %vb, metadata !"sge", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_sge_vx_v8i8(<8 x i8> %va, i8 %b, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sge_vx_v8i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli zero, 8, e8, mf2, ta, mu
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, a1, e8, mf2, ta, ma
+; CHECK-NEXT:    vmsle.vv v0, v9, v8, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x i8> poison, i8 %b, i32 0
+  %vb = shufflevector <8 x i8> %elt.head, <8 x i8> poison, <8 x i32> zeroinitializer
+  %v = call <8 x i1> @llvm.vp.icmp.v8i8(<8 x i8> %va, <8 x i8> %vb, metadata !"sge", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_sge_vx_swap_v8i8(<8 x i8> %va, i8 %b, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sge_vx_swap_v8i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a1, e8, mf2, ta, ma
+; CHECK-NEXT:    vmsle.vx v0, v8, a0, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x i8> poison, i8 %b, i32 0
+  %vb = shufflevector <8 x i8> %elt.head, <8 x i8> poison, <8 x i32> zeroinitializer
+  %v = call <8 x i1> @llvm.vp.icmp.v8i8(<8 x i8> %vb, <8 x i8> %va, metadata !"sge", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_sge_vi_v8i8(<8 x i8> %va, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sge_vi_v8i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e8, mf2, ta, ma
+; CHECK-NEXT:    vmsgt.vi v0, v8, 3, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x i8> poison, i8 4, i32 0
+  %vb = shufflevector <8 x i8> %elt.head, <8 x i8> poison, <8 x i32> zeroinitializer
+  %v = call <8 x i1> @llvm.vp.icmp.v8i8(<8 x i8> %va, <8 x i8> %vb, metadata !"sge", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_sge_vi_swap_v8i8(<8 x i8> %va, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sge_vi_swap_v8i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e8, mf2, ta, ma
+; CHECK-NEXT:    vmsle.vi v0, v8, 4, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x i8> poison, i8 4, i32 0
+  %vb = shufflevector <8 x i8> %elt.head, <8 x i8> poison, <8 x i32> zeroinitializer
+  %v = call <8 x i1> @llvm.vp.icmp.v8i8(<8 x i8> %vb, <8 x i8> %va, metadata !"sge", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_slt_vv_v8i8(<8 x i8> %va, <8 x i8> %vb, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_slt_vv_v8i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e8, mf2, ta, ma
+; CHECK-NEXT:    vmslt.vv v0, v8, v9, v0.t
+; CHECK-NEXT:    ret
+  %v = call <8 x i1> @llvm.vp.icmp.v8i8(<8 x i8> %va, <8 x i8> %vb, metadata !"slt", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_slt_vx_v8i8(<8 x i8> %va, i8 %b, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_slt_vx_v8i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a1, e8, mf2, ta, ma
+; CHECK-NEXT:    vmslt.vx v0, v8, a0, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x i8> poison, i8 %b, i32 0
+  %vb = shufflevector <8 x i8> %elt.head, <8 x i8> poison, <8 x i32> zeroinitializer
+  %v = call <8 x i1> @llvm.vp.icmp.v8i8(<8 x i8> %va, <8 x i8> %vb, metadata !"slt", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_slt_vx_swap_v8i8(<8 x i8> %va, i8 %b, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_slt_vx_swap_v8i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a1, e8, mf2, ta, ma
+; CHECK-NEXT:    vmsgt.vx v0, v8, a0, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x i8> poison, i8 %b, i32 0
+  %vb = shufflevector <8 x i8> %elt.head, <8 x i8> poison, <8 x i32> zeroinitializer
+  %v = call <8 x i1> @llvm.vp.icmp.v8i8(<8 x i8> %vb, <8 x i8> %va, metadata !"slt", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_slt_vi_v8i8(<8 x i8> %va, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_slt_vi_v8i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e8, mf2, ta, ma
+; CHECK-NEXT:    vmsle.vi v0, v8, 3, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x i8> poison, i8 4, i32 0
+  %vb = shufflevector <8 x i8> %elt.head, <8 x i8> poison, <8 x i32> zeroinitializer
+  %v = call <8 x i1> @llvm.vp.icmp.v8i8(<8 x i8> %va, <8 x i8> %vb, metadata !"slt", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_slt_vi_swap_v8i8(<8 x i8> %va, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_slt_vi_swap_v8i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e8, mf2, ta, ma
+; CHECK-NEXT:    vmsgt.vi v0, v8, 4, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x i8> poison, i8 4, i32 0
+  %vb = shufflevector <8 x i8> %elt.head, <8 x i8> poison, <8 x i32> zeroinitializer
+  %v = call <8 x i1> @llvm.vp.icmp.v8i8(<8 x i8> %vb, <8 x i8> %va, metadata !"slt", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_sle_vv_v8i8(<8 x i8> %va, <8 x i8> %vb, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sle_vv_v8i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e8, mf2, ta, ma
+; CHECK-NEXT:    vmsle.vv v0, v8, v9, v0.t
+; CHECK-NEXT:    ret
+  %v = call <8 x i1> @llvm.vp.icmp.v8i8(<8 x i8> %va, <8 x i8> %vb, metadata !"sle", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_sle_vx_v8i8(<8 x i8> %va, i8 %b, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sle_vx_v8i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a1, e8, mf2, ta, ma
+; CHECK-NEXT:    vmsle.vx v0, v8, a0, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x i8> poison, i8 %b, i32 0
+  %vb = shufflevector <8 x i8> %elt.head, <8 x i8> poison, <8 x i32> zeroinitializer
+  %v = call <8 x i1> @llvm.vp.icmp.v8i8(<8 x i8> %va, <8 x i8> %vb, metadata !"sle", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_sle_vx_swap_v8i8(<8 x i8> %va, i8 %b, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sle_vx_swap_v8i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli zero, 8, e8, mf2, ta, mu
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, a1, e8, mf2, ta, ma
+; CHECK-NEXT:    vmsle.vv v0, v9, v8, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x i8> poison, i8 %b, i32 0
+  %vb = shufflevector <8 x i8> %elt.head, <8 x i8> poison, <8 x i32> zeroinitializer
+  %v = call <8 x i1> @llvm.vp.icmp.v8i8(<8 x i8> %vb, <8 x i8> %va, metadata !"sle", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_sle_vi_v8i8(<8 x i8> %va, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sle_vi_v8i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e8, mf2, ta, ma
+; CHECK-NEXT:    vmsle.vi v0, v8, 4, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x i8> poison, i8 4, i32 0
+  %vb = shufflevector <8 x i8> %elt.head, <8 x i8> poison, <8 x i32> zeroinitializer
+  %v = call <8 x i1> @llvm.vp.icmp.v8i8(<8 x i8> %va, <8 x i8> %vb, metadata !"sle", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_sle_vi_swap_v8i8(<8 x i8> %va, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sle_vi_swap_v8i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli zero, 8, e8, mf2, ta, mu
+; CHECK-NEXT:    vmv.v.i v9, 4
+; CHECK-NEXT:    vsetvli zero, a0, e8, mf2, ta, ma
+; CHECK-NEXT:    vmsle.vv v0, v9, v8, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x i8> poison, i8 4, i32 0
+  %vb = shufflevector <8 x i8> %elt.head, <8 x i8> poison, <8 x i32> zeroinitializer
+  %v = call <8 x i1> @llvm.vp.icmp.v8i8(<8 x i8> %vb, <8 x i8> %va, metadata !"sle", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+declare <8 x i1> @llvm.vp.icmp.v8i32(<8 x i32>, <8 x i32>, metadata, <8 x i1>, i32)
+
+define <8 x i1> @icmp_eq_vv_v8i32(<8 x i32> %va, <8 x i32> %vb, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_eq_vv_v8i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m2, ta, ma
+; CHECK-NEXT:    vmseq.vv v12, v8, v10, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v12
+; CHECK-NEXT:    ret
+  %v = call <8 x i1> @llvm.vp.icmp.v8i32(<8 x i32> %va, <8 x i32> %vb, metadata !"eq", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_eq_vx_v8i32(<8 x i32> %va, i32 %b, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_eq_vx_v8i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a1, e32, m2, ta, ma
+; CHECK-NEXT:    vmseq.vx v10, v8, a0, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v10
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x i32> poison, i32 %b, i32 0
+  %vb = shufflevector <8 x i32> %elt.head, <8 x i32> poison, <8 x i32> zeroinitializer
+  %v = call <8 x i1> @llvm.vp.icmp.v8i32(<8 x i32> %va, <8 x i32> %vb, metadata !"eq", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_eq_vx_swap_v8i32(<8 x i32> %va, i32 %b, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_eq_vx_swap_v8i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a1, e32, m2, ta, ma
+; CHECK-NEXT:    vmseq.vx v10, v8, a0, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v10
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x i32> poison, i32 %b, i32 0
+  %vb = shufflevector <8 x i32> %elt.head, <8 x i32> poison, <8 x i32> zeroinitializer
+  %v = call <8 x i1> @llvm.vp.icmp.v8i32(<8 x i32> %vb, <8 x i32> %va, metadata !"eq", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_eq_vi_v8i32(<8 x i32> %va, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_eq_vi_v8i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m2, ta, ma
+; CHECK-NEXT:    vmseq.vi v10, v8, 4, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v10
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x i32> poison, i32 4, i32 0
+  %vb = shufflevector <8 x i32> %elt.head, <8 x i32> poison, <8 x i32> zeroinitializer
+  %v = call <8 x i1> @llvm.vp.icmp.v8i32(<8 x i32> %va, <8 x i32> %vb, metadata !"eq", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_eq_vi_swap_v8i32(<8 x i32> %va, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_eq_vi_swap_v8i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m2, ta, ma
+; CHECK-NEXT:    vmseq.vi v10, v8, 4, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v10
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x i32> poison, i32 4, i32 0
+  %vb = shufflevector <8 x i32> %elt.head, <8 x i32> poison, <8 x i32> zeroinitializer
+  %v = call <8 x i1> @llvm.vp.icmp.v8i32(<8 x i32> %vb, <8 x i32> %va, metadata !"eq", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_ne_vv_v8i32(<8 x i32> %va, <8 x i32> %vb, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ne_vv_v8i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m2, ta, ma
+; CHECK-NEXT:    vmsne.vv v12, v8, v10, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v12
+; CHECK-NEXT:    ret
+  %v = call <8 x i1> @llvm.vp.icmp.v8i32(<8 x i32> %va, <8 x i32> %vb, metadata !"ne", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_ne_vx_v8i32(<8 x i32> %va, i32 %b, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ne_vx_v8i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a1, e32, m2, ta, ma
+; CHECK-NEXT:    vmsne.vx v10, v8, a0, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v10
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x i32> poison, i32 %b, i32 0
+  %vb = shufflevector <8 x i32> %elt.head, <8 x i32> poison, <8 x i32> zeroinitializer
+  %v = call <8 x i1> @llvm.vp.icmp.v8i32(<8 x i32> %va, <8 x i32> %vb, metadata !"ne", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_ne_vx_swap_v8i32(<8 x i32> %va, i32 %b, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ne_vx_swap_v8i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a1, e32, m2, ta, ma
+; CHECK-NEXT:    vmsne.vx v10, v8, a0, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v10
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x i32> poison, i32 %b, i32 0
+  %vb = shufflevector <8 x i32> %elt.head, <8 x i32> poison, <8 x i32> zeroinitializer
+  %v = call <8 x i1> @llvm.vp.icmp.v8i32(<8 x i32> %vb, <8 x i32> %va, metadata !"ne", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_ne_vi_v8i32(<8 x i32> %va, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ne_vi_v8i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m2, ta, ma
+; CHECK-NEXT:    vmsne.vi v10, v8, 4, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v10
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x i32> poison, i32 4, i32 0
+  %vb = shufflevector <8 x i32> %elt.head, <8 x i32> poison, <8 x i32> zeroinitializer
+  %v = call <8 x i1> @llvm.vp.icmp.v8i32(<8 x i32> %va, <8 x i32> %vb, metadata !"ne", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_ne_vi_swap_v8i32(<8 x i32> %va, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ne_vi_swap_v8i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m2, ta, ma
+; CHECK-NEXT:    vmsne.vi v10, v8, 4, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v10
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x i32> poison, i32 4, i32 0
+  %vb = shufflevector <8 x i32> %elt.head, <8 x i32> poison, <8 x i32> zeroinitializer
+  %v = call <8 x i1> @llvm.vp.icmp.v8i32(<8 x i32> %vb, <8 x i32> %va, metadata !"ne", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_ugt_vv_v8i32(<8 x i32> %va, <8 x i32> %vb, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ugt_vv_v8i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m2, ta, ma
+; CHECK-NEXT:    vmsltu.vv v12, v10, v8, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v12
+; CHECK-NEXT:    ret
+  %v = call <8 x i1> @llvm.vp.icmp.v8i32(<8 x i32> %va, <8 x i32> %vb, metadata !"ugt", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_ugt_vx_v8i32(<8 x i32> %va, i32 %b, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ugt_vx_v8i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a1, e32, m2, ta, ma
+; CHECK-NEXT:    vmsgtu.vx v10, v8, a0, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v10
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x i32> poison, i32 %b, i32 0
+  %vb = shufflevector <8 x i32> %elt.head, <8 x i32> poison, <8 x i32> zeroinitializer
+  %v = call <8 x i1> @llvm.vp.icmp.v8i32(<8 x i32> %va, <8 x i32> %vb, metadata !"ugt", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_ugt_vx_swap_v8i32(<8 x i32> %va, i32 %b, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ugt_vx_swap_v8i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a1, e32, m2, ta, ma
+; CHECK-NEXT:    vmsltu.vx v10, v8, a0, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v10
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x i32> poison, i32 %b, i32 0
+  %vb = shufflevector <8 x i32> %elt.head, <8 x i32> poison, <8 x i32> zeroinitializer
+  %v = call <8 x i1> @llvm.vp.icmp.v8i32(<8 x i32> %vb, <8 x i32> %va, metadata !"ugt", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_ugt_vi_v8i32(<8 x i32> %va, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ugt_vi_v8i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m2, ta, ma
+; CHECK-NEXT:    vmsgtu.vi v10, v8, 4, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v10
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x i32> poison, i32 4, i32 0
+  %vb = shufflevector <8 x i32> %elt.head, <8 x i32> poison, <8 x i32> zeroinitializer
+  %v = call <8 x i1> @llvm.vp.icmp.v8i32(<8 x i32> %va, <8 x i32> %vb, metadata !"ugt", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_ugt_vi_swap_v8i32(<8 x i32> %va, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ugt_vi_swap_v8i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    li a1, 4
+; CHECK-NEXT:    vsetvli zero, a0, e32, m2, ta, ma
+; CHECK-NEXT:    vmsltu.vx v10, v8, a1, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v10
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x i32> poison, i32 4, i32 0
+  %vb = shufflevector <8 x i32> %elt.head, <8 x i32> poison, <8 x i32> zeroinitializer
+  %v = call <8 x i1> @llvm.vp.icmp.v8i32(<8 x i32> %vb, <8 x i32> %va, metadata !"ugt", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_uge_vv_v8i32(<8 x i32> %va, <8 x i32> %vb, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_uge_vv_v8i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m2, ta, ma
+; CHECK-NEXT:    vmsleu.vv v12, v10, v8, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v12
+; CHECK-NEXT:    ret
+  %v = call <8 x i1> @llvm.vp.icmp.v8i32(<8 x i32> %va, <8 x i32> %vb, metadata !"uge", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_uge_vx_v8i32(<8 x i32> %va, i32 %b, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_uge_vx_v8i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli zero, 8, e32, m2, ta, mu
+; CHECK-NEXT:    vmv.v.x v12, a0
+; CHECK-NEXT:    vsetvli zero, a1, e32, m2, ta, ma
+; CHECK-NEXT:    vmsleu.vv v10, v12, v8, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v10
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x i32> poison, i32 %b, i32 0
+  %vb = shufflevector <8 x i32> %elt.head, <8 x i32> poison, <8 x i32> zeroinitializer
+  %v = call <8 x i1> @llvm.vp.icmp.v8i32(<8 x i32> %va, <8 x i32> %vb, metadata !"uge", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_uge_vx_swap_v8i32(<8 x i32> %va, i32 %b, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_uge_vx_swap_v8i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a1, e32, m2, ta, ma
+; CHECK-NEXT:    vmsleu.vx v10, v8, a0, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v10
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x i32> poison, i32 %b, i32 0
+  %vb = shufflevector <8 x i32> %elt.head, <8 x i32> poison, <8 x i32> zeroinitializer
+  %v = call <8 x i1> @llvm.vp.icmp.v8i32(<8 x i32> %vb, <8 x i32> %va, metadata !"uge", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_uge_vi_v8i32(<8 x i32> %va, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_uge_vi_v8i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m2, ta, ma
+; CHECK-NEXT:    vmsgtu.vi v10, v8, 3, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v10
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x i32> poison, i32 4, i32 0
+  %vb = shufflevector <8 x i32> %elt.head, <8 x i32> poison, <8 x i32> zeroinitializer
+  %v = call <8 x i1> @llvm.vp.icmp.v8i32(<8 x i32> %va, <8 x i32> %vb, metadata !"uge", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_uge_vi_swap_v8i32(<8 x i32> %va, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_uge_vi_swap_v8i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m2, ta, ma
+; CHECK-NEXT:    vmsleu.vi v10, v8, 4, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v10
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x i32> poison, i32 4, i32 0
+  %vb = shufflevector <8 x i32> %elt.head, <8 x i32> poison, <8 x i32> zeroinitializer
+  %v = call <8 x i1> @llvm.vp.icmp.v8i32(<8 x i32> %vb, <8 x i32> %va, metadata !"uge", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_ult_vv_v8i32(<8 x i32> %va, <8 x i32> %vb, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ult_vv_v8i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m2, ta, ma
+; CHECK-NEXT:    vmsltu.vv v12, v8, v10, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v12
+; CHECK-NEXT:    ret
+  %v = call <8 x i1> @llvm.vp.icmp.v8i32(<8 x i32> %va, <8 x i32> %vb, metadata !"ult", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_ult_vx_v8i32(<8 x i32> %va, i32 %b, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ult_vx_v8i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a1, e32, m2, ta, ma
+; CHECK-NEXT:    vmsltu.vx v10, v8, a0, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v10
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x i32> poison, i32 %b, i32 0
+  %vb = shufflevector <8 x i32> %elt.head, <8 x i32> poison, <8 x i32> zeroinitializer
+  %v = call <8 x i1> @llvm.vp.icmp.v8i32(<8 x i32> %va, <8 x i32> %vb, metadata !"ult", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_ult_vx_swap_v8i32(<8 x i32> %va, i32 %b, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ult_vx_swap_v8i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a1, e32, m2, ta, ma
+; CHECK-NEXT:    vmsgtu.vx v10, v8, a0, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v10
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x i32> poison, i32 %b, i32 0
+  %vb = shufflevector <8 x i32> %elt.head, <8 x i32> poison, <8 x i32> zeroinitializer
+  %v = call <8 x i1> @llvm.vp.icmp.v8i32(<8 x i32> %vb, <8 x i32> %va, metadata !"ult", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_ult_vi_v8i32(<8 x i32> %va, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ult_vi_v8i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m2, ta, ma
+; CHECK-NEXT:    vmsleu.vi v10, v8, 3, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v10
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x i32> poison, i32 4, i32 0
+  %vb = shufflevector <8 x i32> %elt.head, <8 x i32> poison, <8 x i32> zeroinitializer
+  %v = call <8 x i1> @llvm.vp.icmp.v8i32(<8 x i32> %va, <8 x i32> %vb, metadata !"ult", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_ult_vi_swap_v8i32(<8 x i32> %va, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ult_vi_swap_v8i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m2, ta, ma
+; CHECK-NEXT:    vmsgtu.vi v10, v8, 4, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v10
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x i32> poison, i32 4, i32 0
+  %vb = shufflevector <8 x i32> %elt.head, <8 x i32> poison, <8 x i32> zeroinitializer
+  %v = call <8 x i1> @llvm.vp.icmp.v8i32(<8 x i32> %vb, <8 x i32> %va, metadata !"ult", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_sgt_vv_v8i32(<8 x i32> %va, <8 x i32> %vb, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sgt_vv_v8i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m2, ta, ma
+; CHECK-NEXT:    vmslt.vv v12, v10, v8, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v12
+; CHECK-NEXT:    ret
+  %v = call <8 x i1> @llvm.vp.icmp.v8i32(<8 x i32> %va, <8 x i32> %vb, metadata !"sgt", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_sgt_vx_v8i32(<8 x i32> %va, i32 %b, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sgt_vx_v8i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a1, e32, m2, ta, ma
+; CHECK-NEXT:    vmsgt.vx v10, v8, a0, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v10
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x i32> poison, i32 %b, i32 0
+  %vb = shufflevector <8 x i32> %elt.head, <8 x i32> poison, <8 x i32> zeroinitializer
+  %v = call <8 x i1> @llvm.vp.icmp.v8i32(<8 x i32> %va, <8 x i32> %vb, metadata !"sgt", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_sgt_vx_swap_v8i32(<8 x i32> %va, i32 %b, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sgt_vx_swap_v8i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a1, e32, m2, ta, ma
+; CHECK-NEXT:    vmslt.vx v10, v8, a0, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v10
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x i32> poison, i32 %b, i32 0
+  %vb = shufflevector <8 x i32> %elt.head, <8 x i32> poison, <8 x i32> zeroinitializer
+  %v = call <8 x i1> @llvm.vp.icmp.v8i32(<8 x i32> %vb, <8 x i32> %va, metadata !"sgt", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_sgt_vi_v8i32(<8 x i32> %va, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sgt_vi_v8i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m2, ta, ma
+; CHECK-NEXT:    vmsgt.vi v10, v8, 4, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v10
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x i32> poison, i32 4, i32 0
+  %vb = shufflevector <8 x i32> %elt.head, <8 x i32> poison, <8 x i32> zeroinitializer
+  %v = call <8 x i1> @llvm.vp.icmp.v8i32(<8 x i32> %va, <8 x i32> %vb, metadata !"sgt", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_sgt_vi_swap_v8i32(<8 x i32> %va, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sgt_vi_swap_v8i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    li a1, 4
+; CHECK-NEXT:    vsetvli zero, a0, e32, m2, ta, ma
+; CHECK-NEXT:    vmslt.vx v10, v8, a1, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v10
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x i32> poison, i32 4, i32 0
+  %vb = shufflevector <8 x i32> %elt.head, <8 x i32> poison, <8 x i32> zeroinitializer
+  %v = call <8 x i1> @llvm.vp.icmp.v8i32(<8 x i32> %vb, <8 x i32> %va, metadata !"sgt", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_sge_vv_v8i32(<8 x i32> %va, <8 x i32> %vb, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sge_vv_v8i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m2, ta, ma
+; CHECK-NEXT:    vmsle.vv v12, v10, v8, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v12
+; CHECK-NEXT:    ret
+  %v = call <8 x i1> @llvm.vp.icmp.v8i32(<8 x i32> %va, <8 x i32> %vb, metadata !"sge", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_sge_vx_v8i32(<8 x i32> %va, i32 %b, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sge_vx_v8i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli zero, 8, e32, m2, ta, mu
+; CHECK-NEXT:    vmv.v.x v12, a0
+; CHECK-NEXT:    vsetvli zero, a1, e32, m2, ta, ma
+; CHECK-NEXT:    vmsle.vv v10, v12, v8, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v10
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x i32> poison, i32 %b, i32 0
+  %vb = shufflevector <8 x i32> %elt.head, <8 x i32> poison, <8 x i32> zeroinitializer
+  %v = call <8 x i1> @llvm.vp.icmp.v8i32(<8 x i32> %va, <8 x i32> %vb, metadata !"sge", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_sge_vx_swap_v8i32(<8 x i32> %va, i32 %b, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sge_vx_swap_v8i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a1, e32, m2, ta, ma
+; CHECK-NEXT:    vmsle.vx v10, v8, a0, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v10
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x i32> poison, i32 %b, i32 0
+  %vb = shufflevector <8 x i32> %elt.head, <8 x i32> poison, <8 x i32> zeroinitializer
+  %v = call <8 x i1> @llvm.vp.icmp.v8i32(<8 x i32> %vb, <8 x i32> %va, metadata !"sge", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_sge_vi_v8i32(<8 x i32> %va, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sge_vi_v8i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m2, ta, ma
+; CHECK-NEXT:    vmsgt.vi v10, v8, 3, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v10
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x i32> poison, i32 4, i32 0
+  %vb = shufflevector <8 x i32> %elt.head, <8 x i32> poison, <8 x i32> zeroinitializer
+  %v = call <8 x i1> @llvm.vp.icmp.v8i32(<8 x i32> %va, <8 x i32> %vb, metadata !"sge", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_sge_vi_swap_v8i32(<8 x i32> %va, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sge_vi_swap_v8i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m2, ta, ma
+; CHECK-NEXT:    vmsle.vi v10, v8, 4, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v10
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x i32> poison, i32 4, i32 0
+  %vb = shufflevector <8 x i32> %elt.head, <8 x i32> poison, <8 x i32> zeroinitializer
+  %v = call <8 x i1> @llvm.vp.icmp.v8i32(<8 x i32> %vb, <8 x i32> %va, metadata !"sge", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_slt_vv_v8i32(<8 x i32> %va, <8 x i32> %vb, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_slt_vv_v8i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m2, ta, ma
+; CHECK-NEXT:    vmslt.vv v12, v8, v10, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v12
+; CHECK-NEXT:    ret
+  %v = call <8 x i1> @llvm.vp.icmp.v8i32(<8 x i32> %va, <8 x i32> %vb, metadata !"slt", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_slt_vx_v8i32(<8 x i32> %va, i32 %b, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_slt_vx_v8i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a1, e32, m2, ta, ma
+; CHECK-NEXT:    vmslt.vx v10, v8, a0, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v10
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x i32> poison, i32 %b, i32 0
+  %vb = shufflevector <8 x i32> %elt.head, <8 x i32> poison, <8 x i32> zeroinitializer
+  %v = call <8 x i1> @llvm.vp.icmp.v8i32(<8 x i32> %va, <8 x i32> %vb, metadata !"slt", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_slt_vx_swap_v8i32(<8 x i32> %va, i32 %b, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_slt_vx_swap_v8i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a1, e32, m2, ta, ma
+; CHECK-NEXT:    vmsgt.vx v10, v8, a0, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v10
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x i32> poison, i32 %b, i32 0
+  %vb = shufflevector <8 x i32> %elt.head, <8 x i32> poison, <8 x i32> zeroinitializer
+  %v = call <8 x i1> @llvm.vp.icmp.v8i32(<8 x i32> %vb, <8 x i32> %va, metadata !"slt", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_slt_vi_v8i32(<8 x i32> %va, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_slt_vi_v8i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m2, ta, ma
+; CHECK-NEXT:    vmsle.vi v10, v8, 3, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v10
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x i32> poison, i32 4, i32 0
+  %vb = shufflevector <8 x i32> %elt.head, <8 x i32> poison, <8 x i32> zeroinitializer
+  %v = call <8 x i1> @llvm.vp.icmp.v8i32(<8 x i32> %va, <8 x i32> %vb, metadata !"slt", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_slt_vi_swap_v8i32(<8 x i32> %va, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_slt_vi_swap_v8i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m2, ta, ma
+; CHECK-NEXT:    vmsgt.vi v10, v8, 4, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v10
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x i32> poison, i32 4, i32 0
+  %vb = shufflevector <8 x i32> %elt.head, <8 x i32> poison, <8 x i32> zeroinitializer
+  %v = call <8 x i1> @llvm.vp.icmp.v8i32(<8 x i32> %vb, <8 x i32> %va, metadata !"slt", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_sle_vv_v8i32(<8 x i32> %va, <8 x i32> %vb, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sle_vv_v8i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m2, ta, ma
+; CHECK-NEXT:    vmsle.vv v12, v8, v10, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v12
+; CHECK-NEXT:    ret
+  %v = call <8 x i1> @llvm.vp.icmp.v8i32(<8 x i32> %va, <8 x i32> %vb, metadata !"sle", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_sle_vx_v8i32(<8 x i32> %va, i32 %b, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sle_vx_v8i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a1, e32, m2, ta, ma
+; CHECK-NEXT:    vmsle.vx v10, v8, a0, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v10
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x i32> poison, i32 %b, i32 0
+  %vb = shufflevector <8 x i32> %elt.head, <8 x i32> poison, <8 x i32> zeroinitializer
+  %v = call <8 x i1> @llvm.vp.icmp.v8i32(<8 x i32> %va, <8 x i32> %vb, metadata !"sle", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_sle_vx_swap_v8i32(<8 x i32> %va, i32 %b, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sle_vx_swap_v8i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli zero, 8, e32, m2, ta, mu
+; CHECK-NEXT:    vmv.v.x v12, a0
+; CHECK-NEXT:    vsetvli zero, a1, e32, m2, ta, ma
+; CHECK-NEXT:    vmsle.vv v10, v12, v8, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v10
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x i32> poison, i32 %b, i32 0
+  %vb = shufflevector <8 x i32> %elt.head, <8 x i32> poison, <8 x i32> zeroinitializer
+  %v = call <8 x i1> @llvm.vp.icmp.v8i32(<8 x i32> %vb, <8 x i32> %va, metadata !"sle", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_sle_vi_v8i32(<8 x i32> %va, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sle_vi_v8i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m2, ta, ma
+; CHECK-NEXT:    vmsle.vi v10, v8, 4, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v10
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x i32> poison, i32 4, i32 0
+  %vb = shufflevector <8 x i32> %elt.head, <8 x i32> poison, <8 x i32> zeroinitializer
+  %v = call <8 x i1> @llvm.vp.icmp.v8i32(<8 x i32> %va, <8 x i32> %vb, metadata !"sle", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_sle_vi_swap_v8i32(<8 x i32> %va, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sle_vi_swap_v8i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli zero, 8, e32, m2, ta, mu
+; CHECK-NEXT:    vmv.v.i v12, 4
+; CHECK-NEXT:    vsetvli zero, a0, e32, m2, ta, ma
+; CHECK-NEXT:    vmsle.vv v10, v12, v8, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v10
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x i32> poison, i32 4, i32 0
+  %vb = shufflevector <8 x i32> %elt.head, <8 x i32> poison, <8 x i32> zeroinitializer
+  %v = call <8 x i1> @llvm.vp.icmp.v8i32(<8 x i32> %vb, <8 x i32> %va, metadata !"sle", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+declare <8 x i1> @llvm.vp.icmp.v8i64(<8 x i64>, <8 x i64>, metadata, <8 x i1>, i32)
+
+define <8 x i1> @icmp_eq_vv_v8i64(<8 x i64> %va, <8 x i64> %vb, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_eq_vv_v8i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m4, ta, ma
+; CHECK-NEXT:    vmseq.vv v16, v8, v12, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v16
+; CHECK-NEXT:    ret
+  %v = call <8 x i1> @llvm.vp.icmp.v8i64(<8 x i64> %va, <8 x i64> %vb, metadata !"eq", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_eq_vx_v8i64(<8 x i64> %va, i64 %b, <8 x i1> %m, i32 zeroext %evl) {
+; RV32-LABEL: icmp_eq_vx_v8i64:
+; RV32:       # %bb.0:
+; RV32-NEXT:    addi sp, sp, -16
+; RV32-NEXT:    .cfi_def_cfa_offset 16
+; RV32-NEXT:    sw a1, 12(sp)
+; RV32-NEXT:    sw a0, 8(sp)
+; RV32-NEXT:    vsetivli zero, 8, e64, m4, ta, mu
+; RV32-NEXT:    addi a0, sp, 8
+; RV32-NEXT:    vlse64.v v16, (a0), zero
+; RV32-NEXT:    vsetvli zero, a2, e64, m4, ta, ma
+; RV32-NEXT:    vmseq.vv v12, v8, v16, v0.t
+; RV32-NEXT:    vmv1r.v v0, v12
+; RV32-NEXT:    addi sp, sp, 16
+; RV32-NEXT:    ret
+;
+; RV64-LABEL: icmp_eq_vx_v8i64:
+; RV64:       # %bb.0:
+; RV64-NEXT:    vsetvli zero, a1, e64, m4, ta, ma
+; RV64-NEXT:    vmseq.vx v12, v8, a0, v0.t
+; RV64-NEXT:    vmv1r.v v0, v12
+; RV64-NEXT:    ret
+  %elt.head = insertelement <8 x i64> poison, i64 %b, i32 0
+  %vb = shufflevector <8 x i64> %elt.head, <8 x i64> poison, <8 x i32> zeroinitializer
+  %v = call <8 x i1> @llvm.vp.icmp.v8i64(<8 x i64> %va, <8 x i64> %vb, metadata !"eq", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_eq_vx_swap_v8i64(<8 x i64> %va, i64 %b, <8 x i1> %m, i32 zeroext %evl) {
+; RV32-LABEL: icmp_eq_vx_swap_v8i64:
+; RV32:       # %bb.0:
+; RV32-NEXT:    addi sp, sp, -16
+; RV32-NEXT:    .cfi_def_cfa_offset 16
+; RV32-NEXT:    sw a1, 12(sp)
+; RV32-NEXT:    sw a0, 8(sp)
+; RV32-NEXT:    vsetivli zero, 8, e64, m4, ta, mu
+; RV32-NEXT:    addi a0, sp, 8
+; RV32-NEXT:    vlse64.v v16, (a0), zero
+; RV32-NEXT:    vsetvli zero, a2, e64, m4, ta, ma
+; RV32-NEXT:    vmseq.vv v12, v16, v8, v0.t
+; RV32-NEXT:    vmv1r.v v0, v12
+; RV32-NEXT:    addi sp, sp, 16
+; RV32-NEXT:    ret
+;
+; RV64-LABEL: icmp_eq_vx_swap_v8i64:
+; RV64:       # %bb.0:
+; RV64-NEXT:    vsetvli zero, a1, e64, m4, ta, ma
+; RV64-NEXT:    vmseq.vx v12, v8, a0, v0.t
+; RV64-NEXT:    vmv1r.v v0, v12
+; RV64-NEXT:    ret
+  %elt.head = insertelement <8 x i64> poison, i64 %b, i32 0
+  %vb = shufflevector <8 x i64> %elt.head, <8 x i64> poison, <8 x i32> zeroinitializer
+  %v = call <8 x i1> @llvm.vp.icmp.v8i64(<8 x i64> %vb, <8 x i64> %va, metadata !"eq", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_eq_vi_v8i64(<8 x i64> %va, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_eq_vi_v8i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m4, ta, ma
+; CHECK-NEXT:    vmseq.vi v12, v8, 4, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v12
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x i64> poison, i64 4, i32 0
+  %vb = shufflevector <8 x i64> %elt.head, <8 x i64> poison, <8 x i32> zeroinitializer
+  %v = call <8 x i1> @llvm.vp.icmp.v8i64(<8 x i64> %va, <8 x i64> %vb, metadata !"eq", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_eq_vi_swap_v8i64(<8 x i64> %va, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_eq_vi_swap_v8i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m4, ta, ma
+; CHECK-NEXT:    vmseq.vi v12, v8, 4, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v12
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x i64> poison, i64 4, i32 0
+  %vb = shufflevector <8 x i64> %elt.head, <8 x i64> poison, <8 x i32> zeroinitializer
+  %v = call <8 x i1> @llvm.vp.icmp.v8i64(<8 x i64> %vb, <8 x i64> %va, metadata !"eq", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_ne_vv_v8i64(<8 x i64> %va, <8 x i64> %vb, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ne_vv_v8i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m4, ta, ma
+; CHECK-NEXT:    vmsne.vv v16, v8, v12, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v16
+; CHECK-NEXT:    ret
+  %v = call <8 x i1> @llvm.vp.icmp.v8i64(<8 x i64> %va, <8 x i64> %vb, metadata !"ne", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_ne_vx_v8i64(<8 x i64> %va, i64 %b, <8 x i1> %m, i32 zeroext %evl) {
+; RV32-LABEL: icmp_ne_vx_v8i64:
+; RV32:       # %bb.0:
+; RV32-NEXT:    addi sp, sp, -16
+; RV32-NEXT:    .cfi_def_cfa_offset 16
+; RV32-NEXT:    sw a1, 12(sp)
+; RV32-NEXT:    sw a0, 8(sp)
+; RV32-NEXT:    vsetivli zero, 8, e64, m4, ta, mu
+; RV32-NEXT:    addi a0, sp, 8
+; RV32-NEXT:    vlse64.v v16, (a0), zero
+; RV32-NEXT:    vsetvli zero, a2, e64, m4, ta, ma
+; RV32-NEXT:    vmsne.vv v12, v8, v16, v0.t
+; RV32-NEXT:    vmv1r.v v0, v12
+; RV32-NEXT:    addi sp, sp, 16
+; RV32-NEXT:    ret
+;
+; RV64-LABEL: icmp_ne_vx_v8i64:
+; RV64:       # %bb.0:
+; RV64-NEXT:    vsetvli zero, a1, e64, m4, ta, ma
+; RV64-NEXT:    vmsne.vx v12, v8, a0, v0.t
+; RV64-NEXT:    vmv1r.v v0, v12
+; RV64-NEXT:    ret
+  %elt.head = insertelement <8 x i64> poison, i64 %b, i32 0
+  %vb = shufflevector <8 x i64> %elt.head, <8 x i64> poison, <8 x i32> zeroinitializer
+  %v = call <8 x i1> @llvm.vp.icmp.v8i64(<8 x i64> %va, <8 x i64> %vb, metadata !"ne", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_ne_vx_swap_v8i64(<8 x i64> %va, i64 %b, <8 x i1> %m, i32 zeroext %evl) {
+; RV32-LABEL: icmp_ne_vx_swap_v8i64:
+; RV32:       # %bb.0:
+; RV32-NEXT:    addi sp, sp, -16
+; RV32-NEXT:    .cfi_def_cfa_offset 16
+; RV32-NEXT:    sw a1, 12(sp)
+; RV32-NEXT:    sw a0, 8(sp)
+; RV32-NEXT:    vsetivli zero, 8, e64, m4, ta, mu
+; RV32-NEXT:    addi a0, sp, 8
+; RV32-NEXT:    vlse64.v v16, (a0), zero
+; RV32-NEXT:    vsetvli zero, a2, e64, m4, ta, ma
+; RV32-NEXT:    vmsne.vv v12, v16, v8, v0.t
+; RV32-NEXT:    vmv1r.v v0, v12
+; RV32-NEXT:    addi sp, sp, 16
+; RV32-NEXT:    ret
+;
+; RV64-LABEL: icmp_ne_vx_swap_v8i64:
+; RV64:       # %bb.0:
+; RV64-NEXT:    vsetvli zero, a1, e64, m4, ta, ma
+; RV64-NEXT:    vmsne.vx v12, v8, a0, v0.t
+; RV64-NEXT:    vmv1r.v v0, v12
+; RV64-NEXT:    ret
+  %elt.head = insertelement <8 x i64> poison, i64 %b, i32 0
+  %vb = shufflevector <8 x i64> %elt.head, <8 x i64> poison, <8 x i32> zeroinitializer
+  %v = call <8 x i1> @llvm.vp.icmp.v8i64(<8 x i64> %vb, <8 x i64> %va, metadata !"ne", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_ne_vi_v8i64(<8 x i64> %va, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ne_vi_v8i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m4, ta, ma
+; CHECK-NEXT:    vmsne.vi v12, v8, 4, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v12
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x i64> poison, i64 4, i32 0
+  %vb = shufflevector <8 x i64> %elt.head, <8 x i64> poison, <8 x i32> zeroinitializer
+  %v = call <8 x i1> @llvm.vp.icmp.v8i64(<8 x i64> %va, <8 x i64> %vb, metadata !"ne", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_ne_vi_swap_v8i64(<8 x i64> %va, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ne_vi_swap_v8i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m4, ta, ma
+; CHECK-NEXT:    vmsne.vi v12, v8, 4, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v12
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x i64> poison, i64 4, i32 0
+  %vb = shufflevector <8 x i64> %elt.head, <8 x i64> poison, <8 x i32> zeroinitializer
+  %v = call <8 x i1> @llvm.vp.icmp.v8i64(<8 x i64> %vb, <8 x i64> %va, metadata !"ne", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_ugt_vv_v8i64(<8 x i64> %va, <8 x i64> %vb, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ugt_vv_v8i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m4, ta, ma
+; CHECK-NEXT:    vmsltu.vv v16, v12, v8, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v16
+; CHECK-NEXT:    ret
+  %v = call <8 x i1> @llvm.vp.icmp.v8i64(<8 x i64> %va, <8 x i64> %vb, metadata !"ugt", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_ugt_vx_v8i64(<8 x i64> %va, i64 %b, <8 x i1> %m, i32 zeroext %evl) {
+; RV32-LABEL: icmp_ugt_vx_v8i64:
+; RV32:       # %bb.0:
+; RV32-NEXT:    addi sp, sp, -16
+; RV32-NEXT:    .cfi_def_cfa_offset 16
+; RV32-NEXT:    sw a1, 12(sp)
+; RV32-NEXT:    sw a0, 8(sp)
+; RV32-NEXT:    vsetivli zero, 8, e64, m4, ta, mu
+; RV32-NEXT:    addi a0, sp, 8
+; RV32-NEXT:    vlse64.v v16, (a0), zero
+; RV32-NEXT:    vsetvli zero, a2, e64, m4, ta, ma
+; RV32-NEXT:    vmsltu.vv v12, v16, v8, v0.t
+; RV32-NEXT:    vmv1r.v v0, v12
+; RV32-NEXT:    addi sp, sp, 16
+; RV32-NEXT:    ret
+;
+; RV64-LABEL: icmp_ugt_vx_v8i64:
+; RV64:       # %bb.0:
+; RV64-NEXT:    vsetvli zero, a1, e64, m4, ta, ma
+; RV64-NEXT:    vmsgtu.vx v12, v8, a0, v0.t
+; RV64-NEXT:    vmv1r.v v0, v12
+; RV64-NEXT:    ret
+  %elt.head = insertelement <8 x i64> poison, i64 %b, i32 0
+  %vb = shufflevector <8 x i64> %elt.head, <8 x i64> poison, <8 x i32> zeroinitializer
+  %v = call <8 x i1> @llvm.vp.icmp.v8i64(<8 x i64> %va, <8 x i64> %vb, metadata !"ugt", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_ugt_vx_swap_v8i64(<8 x i64> %va, i64 %b, <8 x i1> %m, i32 zeroext %evl) {
+; RV32-LABEL: icmp_ugt_vx_swap_v8i64:
+; RV32:       # %bb.0:
+; RV32-NEXT:    addi sp, sp, -16
+; RV32-NEXT:    .cfi_def_cfa_offset 16
+; RV32-NEXT:    sw a1, 12(sp)
+; RV32-NEXT:    sw a0, 8(sp)
+; RV32-NEXT:    vsetivli zero, 8, e64, m4, ta, mu
+; RV32-NEXT:    addi a0, sp, 8
+; RV32-NEXT:    vlse64.v v16, (a0), zero
+; RV32-NEXT:    vsetvli zero, a2, e64, m4, ta, ma
+; RV32-NEXT:    vmsltu.vv v12, v8, v16, v0.t
+; RV32-NEXT:    vmv1r.v v0, v12
+; RV32-NEXT:    addi sp, sp, 16
+; RV32-NEXT:    ret
+;
+; RV64-LABEL: icmp_ugt_vx_swap_v8i64:
+; RV64:       # %bb.0:
+; RV64-NEXT:    vsetvli zero, a1, e64, m4, ta, ma
+; RV64-NEXT:    vmsltu.vx v12, v8, a0, v0.t
+; RV64-NEXT:    vmv1r.v v0, v12
+; RV64-NEXT:    ret
+  %elt.head = insertelement <8 x i64> poison, i64 %b, i32 0
+  %vb = shufflevector <8 x i64> %elt.head, <8 x i64> poison, <8 x i32> zeroinitializer
+  %v = call <8 x i1> @llvm.vp.icmp.v8i64(<8 x i64> %vb, <8 x i64> %va, metadata !"ugt", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_ugt_vi_v8i64(<8 x i64> %va, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ugt_vi_v8i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m4, ta, ma
+; CHECK-NEXT:    vmsgtu.vi v12, v8, 4, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v12
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x i64> poison, i64 4, i32 0
+  %vb = shufflevector <8 x i64> %elt.head, <8 x i64> poison, <8 x i32> zeroinitializer
+  %v = call <8 x i1> @llvm.vp.icmp.v8i64(<8 x i64> %va, <8 x i64> %vb, metadata !"ugt", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_ugt_vi_swap_v8i64(<8 x i64> %va, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ugt_vi_swap_v8i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    li a1, 4
+; CHECK-NEXT:    vsetvli zero, a0, e64, m4, ta, ma
+; CHECK-NEXT:    vmsltu.vx v12, v8, a1, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v12
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x i64> poison, i64 4, i32 0
+  %vb = shufflevector <8 x i64> %elt.head, <8 x i64> poison, <8 x i32> zeroinitializer
+  %v = call <8 x i1> @llvm.vp.icmp.v8i64(<8 x i64> %vb, <8 x i64> %va, metadata !"ugt", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_uge_vv_v8i64(<8 x i64> %va, <8 x i64> %vb, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_uge_vv_v8i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m4, ta, ma
+; CHECK-NEXT:    vmsleu.vv v16, v12, v8, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v16
+; CHECK-NEXT:    ret
+  %v = call <8 x i1> @llvm.vp.icmp.v8i64(<8 x i64> %va, <8 x i64> %vb, metadata !"uge", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_uge_vx_v8i64(<8 x i64> %va, i64 %b, <8 x i1> %m, i32 zeroext %evl) {
+; RV32-LABEL: icmp_uge_vx_v8i64:
+; RV32:       # %bb.0:
+; RV32-NEXT:    addi sp, sp, -16
+; RV32-NEXT:    .cfi_def_cfa_offset 16
+; RV32-NEXT:    sw a1, 12(sp)
+; RV32-NEXT:    sw a0, 8(sp)
+; RV32-NEXT:    vsetivli zero, 8, e64, m4, ta, mu
+; RV32-NEXT:    addi a0, sp, 8
+; RV32-NEXT:    vlse64.v v16, (a0), zero
+; RV32-NEXT:    vsetvli zero, a2, e64, m4, ta, ma
+; RV32-NEXT:    vmsleu.vv v12, v16, v8, v0.t
+; RV32-NEXT:    vmv1r.v v0, v12
+; RV32-NEXT:    addi sp, sp, 16
+; RV32-NEXT:    ret
+;
+; RV64-LABEL: icmp_uge_vx_v8i64:
+; RV64:       # %bb.0:
+; RV64-NEXT:    vsetivli zero, 8, e64, m4, ta, mu
+; RV64-NEXT:    vmv.v.x v16, a0
+; RV64-NEXT:    vsetvli zero, a1, e64, m4, ta, ma
+; RV64-NEXT:    vmsleu.vv v12, v16, v8, v0.t
+; RV64-NEXT:    vmv1r.v v0, v12
+; RV64-NEXT:    ret
+  %elt.head = insertelement <8 x i64> poison, i64 %b, i32 0
+  %vb = shufflevector <8 x i64> %elt.head, <8 x i64> poison, <8 x i32> zeroinitializer
+  %v = call <8 x i1> @llvm.vp.icmp.v8i64(<8 x i64> %va, <8 x i64> %vb, metadata !"uge", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_uge_vx_swap_v8i64(<8 x i64> %va, i64 %b, <8 x i1> %m, i32 zeroext %evl) {
+; RV32-LABEL: icmp_uge_vx_swap_v8i64:
+; RV32:       # %bb.0:
+; RV32-NEXT:    addi sp, sp, -16
+; RV32-NEXT:    .cfi_def_cfa_offset 16
+; RV32-NEXT:    sw a1, 12(sp)
+; RV32-NEXT:    sw a0, 8(sp)
+; RV32-NEXT:    vsetivli zero, 8, e64, m4, ta, mu
+; RV32-NEXT:    addi a0, sp, 8
+; RV32-NEXT:    vlse64.v v16, (a0), zero
+; RV32-NEXT:    vsetvli zero, a2, e64, m4, ta, ma
+; RV32-NEXT:    vmsleu.vv v12, v8, v16, v0.t
+; RV32-NEXT:    vmv1r.v v0, v12
+; RV32-NEXT:    addi sp, sp, 16
+; RV32-NEXT:    ret
+;
+; RV64-LABEL: icmp_uge_vx_swap_v8i64:
+; RV64:       # %bb.0:
+; RV64-NEXT:    vsetvli zero, a1, e64, m4, ta, ma
+; RV64-NEXT:    vmsleu.vx v12, v8, a0, v0.t
+; RV64-NEXT:    vmv1r.v v0, v12
+; RV64-NEXT:    ret
+  %elt.head = insertelement <8 x i64> poison, i64 %b, i32 0
+  %vb = shufflevector <8 x i64> %elt.head, <8 x i64> poison, <8 x i32> zeroinitializer
+  %v = call <8 x i1> @llvm.vp.icmp.v8i64(<8 x i64> %vb, <8 x i64> %va, metadata !"uge", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_uge_vi_v8i64(<8 x i64> %va, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_uge_vi_v8i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m4, ta, ma
+; CHECK-NEXT:    vmsgtu.vi v12, v8, 3, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v12
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x i64> poison, i64 4, i32 0
+  %vb = shufflevector <8 x i64> %elt.head, <8 x i64> poison, <8 x i32> zeroinitializer
+  %v = call <8 x i1> @llvm.vp.icmp.v8i64(<8 x i64> %va, <8 x i64> %vb, metadata !"uge", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_uge_vi_swap_v8i64(<8 x i64> %va, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_uge_vi_swap_v8i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m4, ta, ma
+; CHECK-NEXT:    vmsleu.vi v12, v8, 4, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v12
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x i64> poison, i64 4, i32 0
+  %vb = shufflevector <8 x i64> %elt.head, <8 x i64> poison, <8 x i32> zeroinitializer
+  %v = call <8 x i1> @llvm.vp.icmp.v8i64(<8 x i64> %vb, <8 x i64> %va, metadata !"uge", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_ult_vv_v8i64(<8 x i64> %va, <8 x i64> %vb, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ult_vv_v8i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m4, ta, ma
+; CHECK-NEXT:    vmsltu.vv v16, v8, v12, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v16
+; CHECK-NEXT:    ret
+  %v = call <8 x i1> @llvm.vp.icmp.v8i64(<8 x i64> %va, <8 x i64> %vb, metadata !"ult", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_ult_vx_v8i64(<8 x i64> %va, i64 %b, <8 x i1> %m, i32 zeroext %evl) {
+; RV32-LABEL: icmp_ult_vx_v8i64:
+; RV32:       # %bb.0:
+; RV32-NEXT:    addi sp, sp, -16
+; RV32-NEXT:    .cfi_def_cfa_offset 16
+; RV32-NEXT:    sw a1, 12(sp)
+; RV32-NEXT:    sw a0, 8(sp)
+; RV32-NEXT:    vsetivli zero, 8, e64, m4, ta, mu
+; RV32-NEXT:    addi a0, sp, 8
+; RV32-NEXT:    vlse64.v v16, (a0), zero
+; RV32-NEXT:    vsetvli zero, a2, e64, m4, ta, ma
+; RV32-NEXT:    vmsltu.vv v12, v8, v16, v0.t
+; RV32-NEXT:    vmv1r.v v0, v12
+; RV32-NEXT:    addi sp, sp, 16
+; RV32-NEXT:    ret
+;
+; RV64-LABEL: icmp_ult_vx_v8i64:
+; RV64:       # %bb.0:
+; RV64-NEXT:    vsetvli zero, a1, e64, m4, ta, ma
+; RV64-NEXT:    vmsltu.vx v12, v8, a0, v0.t
+; RV64-NEXT:    vmv1r.v v0, v12
+; RV64-NEXT:    ret
+  %elt.head = insertelement <8 x i64> poison, i64 %b, i32 0
+  %vb = shufflevector <8 x i64> %elt.head, <8 x i64> poison, <8 x i32> zeroinitializer
+  %v = call <8 x i1> @llvm.vp.icmp.v8i64(<8 x i64> %va, <8 x i64> %vb, metadata !"ult", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_ult_vx_swap_v8i64(<8 x i64> %va, i64 %b, <8 x i1> %m, i32 zeroext %evl) {
+; RV32-LABEL: icmp_ult_vx_swap_v8i64:
+; RV32:       # %bb.0:
+; RV32-NEXT:    addi sp, sp, -16
+; RV32-NEXT:    .cfi_def_cfa_offset 16
+; RV32-NEXT:    sw a1, 12(sp)
+; RV32-NEXT:    sw a0, 8(sp)
+; RV32-NEXT:    vsetivli zero, 8, e64, m4, ta, mu
+; RV32-NEXT:    addi a0, sp, 8
+; RV32-NEXT:    vlse64.v v16, (a0), zero
+; RV32-NEXT:    vsetvli zero, a2, e64, m4, ta, ma
+; RV32-NEXT:    vmsltu.vv v12, v16, v8, v0.t
+; RV32-NEXT:    vmv1r.v v0, v12
+; RV32-NEXT:    addi sp, sp, 16
+; RV32-NEXT:    ret
+;
+; RV64-LABEL: icmp_ult_vx_swap_v8i64:
+; RV64:       # %bb.0:
+; RV64-NEXT:    vsetvli zero, a1, e64, m4, ta, ma
+; RV64-NEXT:    vmsgtu.vx v12, v8, a0, v0.t
+; RV64-NEXT:    vmv1r.v v0, v12
+; RV64-NEXT:    ret
+  %elt.head = insertelement <8 x i64> poison, i64 %b, i32 0
+  %vb = shufflevector <8 x i64> %elt.head, <8 x i64> poison, <8 x i32> zeroinitializer
+  %v = call <8 x i1> @llvm.vp.icmp.v8i64(<8 x i64> %vb, <8 x i64> %va, metadata !"ult", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_ult_vi_v8i64(<8 x i64> %va, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ult_vi_v8i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m4, ta, ma
+; CHECK-NEXT:    vmsleu.vi v12, v8, 3, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v12
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x i64> poison, i64 4, i32 0
+  %vb = shufflevector <8 x i64> %elt.head, <8 x i64> poison, <8 x i32> zeroinitializer
+  %v = call <8 x i1> @llvm.vp.icmp.v8i64(<8 x i64> %va, <8 x i64> %vb, metadata !"ult", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_ult_vi_swap_v8i64(<8 x i64> %va, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ult_vi_swap_v8i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m4, ta, ma
+; CHECK-NEXT:    vmsgtu.vi v12, v8, 4, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v12
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x i64> poison, i64 4, i32 0
+  %vb = shufflevector <8 x i64> %elt.head, <8 x i64> poison, <8 x i32> zeroinitializer
+  %v = call <8 x i1> @llvm.vp.icmp.v8i64(<8 x i64> %vb, <8 x i64> %va, metadata !"ult", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_sgt_vv_v8i64(<8 x i64> %va, <8 x i64> %vb, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sgt_vv_v8i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m4, ta, ma
+; CHECK-NEXT:    vmslt.vv v16, v12, v8, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v16
+; CHECK-NEXT:    ret
+  %v = call <8 x i1> @llvm.vp.icmp.v8i64(<8 x i64> %va, <8 x i64> %vb, metadata !"sgt", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_sgt_vx_v8i64(<8 x i64> %va, i64 %b, <8 x i1> %m, i32 zeroext %evl) {
+; RV32-LABEL: icmp_sgt_vx_v8i64:
+; RV32:       # %bb.0:
+; RV32-NEXT:    addi sp, sp, -16
+; RV32-NEXT:    .cfi_def_cfa_offset 16
+; RV32-NEXT:    sw a1, 12(sp)
+; RV32-NEXT:    sw a0, 8(sp)
+; RV32-NEXT:    vsetivli zero, 8, e64, m4, ta, mu
+; RV32-NEXT:    addi a0, sp, 8
+; RV32-NEXT:    vlse64.v v16, (a0), zero
+; RV32-NEXT:    vsetvli zero, a2, e64, m4, ta, ma
+; RV32-NEXT:    vmslt.vv v12, v16, v8, v0.t
+; RV32-NEXT:    vmv1r.v v0, v12
+; RV32-NEXT:    addi sp, sp, 16
+; RV32-NEXT:    ret
+;
+; RV64-LABEL: icmp_sgt_vx_v8i64:
+; RV64:       # %bb.0:
+; RV64-NEXT:    vsetvli zero, a1, e64, m4, ta, ma
+; RV64-NEXT:    vmsgt.vx v12, v8, a0, v0.t
+; RV64-NEXT:    vmv1r.v v0, v12
+; RV64-NEXT:    ret
+  %elt.head = insertelement <8 x i64> poison, i64 %b, i32 0
+  %vb = shufflevector <8 x i64> %elt.head, <8 x i64> poison, <8 x i32> zeroinitializer
+  %v = call <8 x i1> @llvm.vp.icmp.v8i64(<8 x i64> %va, <8 x i64> %vb, metadata !"sgt", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_sgt_vx_swap_v8i64(<8 x i64> %va, i64 %b, <8 x i1> %m, i32 zeroext %evl) {
+; RV32-LABEL: icmp_sgt_vx_swap_v8i64:
+; RV32:       # %bb.0:
+; RV32-NEXT:    addi sp, sp, -16
+; RV32-NEXT:    .cfi_def_cfa_offset 16
+; RV32-NEXT:    sw a1, 12(sp)
+; RV32-NEXT:    sw a0, 8(sp)
+; RV32-NEXT:    vsetivli zero, 8, e64, m4, ta, mu
+; RV32-NEXT:    addi a0, sp, 8
+; RV32-NEXT:    vlse64.v v16, (a0), zero
+; RV32-NEXT:    vsetvli zero, a2, e64, m4, ta, ma
+; RV32-NEXT:    vmslt.vv v12, v8, v16, v0.t
+; RV32-NEXT:    vmv1r.v v0, v12
+; RV32-NEXT:    addi sp, sp, 16
+; RV32-NEXT:    ret
+;
+; RV64-LABEL: icmp_sgt_vx_swap_v8i64:
+; RV64:       # %bb.0:
+; RV64-NEXT:    vsetvli zero, a1, e64, m4, ta, ma
+; RV64-NEXT:    vmslt.vx v12, v8, a0, v0.t
+; RV64-NEXT:    vmv1r.v v0, v12
+; RV64-NEXT:    ret
+  %elt.head = insertelement <8 x i64> poison, i64 %b, i32 0
+  %vb = shufflevector <8 x i64> %elt.head, <8 x i64> poison, <8 x i32> zeroinitializer
+  %v = call <8 x i1> @llvm.vp.icmp.v8i64(<8 x i64> %vb, <8 x i64> %va, metadata !"sgt", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_sgt_vi_v8i64(<8 x i64> %va, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sgt_vi_v8i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m4, ta, ma
+; CHECK-NEXT:    vmsgt.vi v12, v8, 4, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v12
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x i64> poison, i64 4, i32 0
+  %vb = shufflevector <8 x i64> %elt.head, <8 x i64> poison, <8 x i32> zeroinitializer
+  %v = call <8 x i1> @llvm.vp.icmp.v8i64(<8 x i64> %va, <8 x i64> %vb, metadata !"sgt", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_sgt_vi_swap_v8i64(<8 x i64> %va, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sgt_vi_swap_v8i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    li a1, 4
+; CHECK-NEXT:    vsetvli zero, a0, e64, m4, ta, ma
+; CHECK-NEXT:    vmslt.vx v12, v8, a1, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v12
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x i64> poison, i64 4, i32 0
+  %vb = shufflevector <8 x i64> %elt.head, <8 x i64> poison, <8 x i32> zeroinitializer
+  %v = call <8 x i1> @llvm.vp.icmp.v8i64(<8 x i64> %vb, <8 x i64> %va, metadata !"sgt", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_sge_vv_v8i64(<8 x i64> %va, <8 x i64> %vb, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sge_vv_v8i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m4, ta, ma
+; CHECK-NEXT:    vmsle.vv v16, v12, v8, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v16
+; CHECK-NEXT:    ret
+  %v = call <8 x i1> @llvm.vp.icmp.v8i64(<8 x i64> %va, <8 x i64> %vb, metadata !"sge", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_sge_vx_v8i64(<8 x i64> %va, i64 %b, <8 x i1> %m, i32 zeroext %evl) {
+; RV32-LABEL: icmp_sge_vx_v8i64:
+; RV32:       # %bb.0:
+; RV32-NEXT:    addi sp, sp, -16
+; RV32-NEXT:    .cfi_def_cfa_offset 16
+; RV32-NEXT:    sw a1, 12(sp)
+; RV32-NEXT:    sw a0, 8(sp)
+; RV32-NEXT:    vsetivli zero, 8, e64, m4, ta, mu
+; RV32-NEXT:    addi a0, sp, 8
+; RV32-NEXT:    vlse64.v v16, (a0), zero
+; RV32-NEXT:    vsetvli zero, a2, e64, m4, ta, ma
+; RV32-NEXT:    vmsle.vv v12, v16, v8, v0.t
+; RV32-NEXT:    vmv1r.v v0, v12
+; RV32-NEXT:    addi sp, sp, 16
+; RV32-NEXT:    ret
+;
+; RV64-LABEL: icmp_sge_vx_v8i64:
+; RV64:       # %bb.0:
+; RV64-NEXT:    vsetivli zero, 8, e64, m4, ta, mu
+; RV64-NEXT:    vmv.v.x v16, a0
+; RV64-NEXT:    vsetvli zero, a1, e64, m4, ta, ma
+; RV64-NEXT:    vmsle.vv v12, v16, v8, v0.t
+; RV64-NEXT:    vmv1r.v v0, v12
+; RV64-NEXT:    ret
+  %elt.head = insertelement <8 x i64> poison, i64 %b, i32 0
+  %vb = shufflevector <8 x i64> %elt.head, <8 x i64> poison, <8 x i32> zeroinitializer
+  %v = call <8 x i1> @llvm.vp.icmp.v8i64(<8 x i64> %va, <8 x i64> %vb, metadata !"sge", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_sge_vx_swap_v8i64(<8 x i64> %va, i64 %b, <8 x i1> %m, i32 zeroext %evl) {
+; RV32-LABEL: icmp_sge_vx_swap_v8i64:
+; RV32:       # %bb.0:
+; RV32-NEXT:    addi sp, sp, -16
+; RV32-NEXT:    .cfi_def_cfa_offset 16
+; RV32-NEXT:    sw a1, 12(sp)
+; RV32-NEXT:    sw a0, 8(sp)
+; RV32-NEXT:    vsetivli zero, 8, e64, m4, ta, mu
+; RV32-NEXT:    addi a0, sp, 8
+; RV32-NEXT:    vlse64.v v16, (a0), zero
+; RV32-NEXT:    vsetvli zero, a2, e64, m4, ta, ma
+; RV32-NEXT:    vmsle.vv v12, v8, v16, v0.t
+; RV32-NEXT:    vmv1r.v v0, v12
+; RV32-NEXT:    addi sp, sp, 16
+; RV32-NEXT:    ret
+;
+; RV64-LABEL: icmp_sge_vx_swap_v8i64:
+; RV64:       # %bb.0:
+; RV64-NEXT:    vsetvli zero, a1, e64, m4, ta, ma
+; RV64-NEXT:    vmsle.vx v12, v8, a0, v0.t
+; RV64-NEXT:    vmv1r.v v0, v12
+; RV64-NEXT:    ret
+  %elt.head = insertelement <8 x i64> poison, i64 %b, i32 0
+  %vb = shufflevector <8 x i64> %elt.head, <8 x i64> poison, <8 x i32> zeroinitializer
+  %v = call <8 x i1> @llvm.vp.icmp.v8i64(<8 x i64> %vb, <8 x i64> %va, metadata !"sge", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_sge_vi_v8i64(<8 x i64> %va, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sge_vi_v8i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m4, ta, ma
+; CHECK-NEXT:    vmsgt.vi v12, v8, 3, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v12
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x i64> poison, i64 4, i32 0
+  %vb = shufflevector <8 x i64> %elt.head, <8 x i64> poison, <8 x i32> zeroinitializer
+  %v = call <8 x i1> @llvm.vp.icmp.v8i64(<8 x i64> %va, <8 x i64> %vb, metadata !"sge", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_sge_vi_swap_v8i64(<8 x i64> %va, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sge_vi_swap_v8i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m4, ta, ma
+; CHECK-NEXT:    vmsle.vi v12, v8, 4, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v12
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x i64> poison, i64 4, i32 0
+  %vb = shufflevector <8 x i64> %elt.head, <8 x i64> poison, <8 x i32> zeroinitializer
+  %v = call <8 x i1> @llvm.vp.icmp.v8i64(<8 x i64> %vb, <8 x i64> %va, metadata !"sge", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_slt_vv_v8i64(<8 x i64> %va, <8 x i64> %vb, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_slt_vv_v8i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m4, ta, ma
+; CHECK-NEXT:    vmslt.vv v16, v8, v12, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v16
+; CHECK-NEXT:    ret
+  %v = call <8 x i1> @llvm.vp.icmp.v8i64(<8 x i64> %va, <8 x i64> %vb, metadata !"slt", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_slt_vx_v8i64(<8 x i64> %va, i64 %b, <8 x i1> %m, i32 zeroext %evl) {
+; RV32-LABEL: icmp_slt_vx_v8i64:
+; RV32:       # %bb.0:
+; RV32-NEXT:    addi sp, sp, -16
+; RV32-NEXT:    .cfi_def_cfa_offset 16
+; RV32-NEXT:    sw a1, 12(sp)
+; RV32-NEXT:    sw a0, 8(sp)
+; RV32-NEXT:    vsetivli zero, 8, e64, m4, ta, mu
+; RV32-NEXT:    addi a0, sp, 8
+; RV32-NEXT:    vlse64.v v16, (a0), zero
+; RV32-NEXT:    vsetvli zero, a2, e64, m4, ta, ma
+; RV32-NEXT:    vmslt.vv v12, v8, v16, v0.t
+; RV32-NEXT:    vmv1r.v v0, v12
+; RV32-NEXT:    addi sp, sp, 16
+; RV32-NEXT:    ret
+;
+; RV64-LABEL: icmp_slt_vx_v8i64:
+; RV64:       # %bb.0:
+; RV64-NEXT:    vsetvli zero, a1, e64, m4, ta, ma
+; RV64-NEXT:    vmslt.vx v12, v8, a0, v0.t
+; RV64-NEXT:    vmv1r.v v0, v12
+; RV64-NEXT:    ret
+  %elt.head = insertelement <8 x i64> poison, i64 %b, i32 0
+  %vb = shufflevector <8 x i64> %elt.head, <8 x i64> poison, <8 x i32> zeroinitializer
+  %v = call <8 x i1> @llvm.vp.icmp.v8i64(<8 x i64> %va, <8 x i64> %vb, metadata !"slt", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_slt_vx_swap_v8i64(<8 x i64> %va, i64 %b, <8 x i1> %m, i32 zeroext %evl) {
+; RV32-LABEL: icmp_slt_vx_swap_v8i64:
+; RV32:       # %bb.0:
+; RV32-NEXT:    addi sp, sp, -16
+; RV32-NEXT:    .cfi_def_cfa_offset 16
+; RV32-NEXT:    sw a1, 12(sp)
+; RV32-NEXT:    sw a0, 8(sp)
+; RV32-NEXT:    vsetivli zero, 8, e64, m4, ta, mu
+; RV32-NEXT:    addi a0, sp, 8
+; RV32-NEXT:    vlse64.v v16, (a0), zero
+; RV32-NEXT:    vsetvli zero, a2, e64, m4, ta, ma
+; RV32-NEXT:    vmslt.vv v12, v16, v8, v0.t
+; RV32-NEXT:    vmv1r.v v0, v12
+; RV32-NEXT:    addi sp, sp, 16
+; RV32-NEXT:    ret
+;
+; RV64-LABEL: icmp_slt_vx_swap_v8i64:
+; RV64:       # %bb.0:
+; RV64-NEXT:    vsetvli zero, a1, e64, m4, ta, ma
+; RV64-NEXT:    vmsgt.vx v12, v8, a0, v0.t
+; RV64-NEXT:    vmv1r.v v0, v12
+; RV64-NEXT:    ret
+  %elt.head = insertelement <8 x i64> poison, i64 %b, i32 0
+  %vb = shufflevector <8 x i64> %elt.head, <8 x i64> poison, <8 x i32> zeroinitializer
+  %v = call <8 x i1> @llvm.vp.icmp.v8i64(<8 x i64> %vb, <8 x i64> %va, metadata !"slt", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_slt_vi_v8i64(<8 x i64> %va, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_slt_vi_v8i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m4, ta, ma
+; CHECK-NEXT:    vmsle.vi v12, v8, 3, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v12
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x i64> poison, i64 4, i32 0
+  %vb = shufflevector <8 x i64> %elt.head, <8 x i64> poison, <8 x i32> zeroinitializer
+  %v = call <8 x i1> @llvm.vp.icmp.v8i64(<8 x i64> %va, <8 x i64> %vb, metadata !"slt", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_slt_vi_swap_v8i64(<8 x i64> %va, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_slt_vi_swap_v8i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m4, ta, ma
+; CHECK-NEXT:    vmsgt.vi v12, v8, 4, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v12
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x i64> poison, i64 4, i32 0
+  %vb = shufflevector <8 x i64> %elt.head, <8 x i64> poison, <8 x i32> zeroinitializer
+  %v = call <8 x i1> @llvm.vp.icmp.v8i64(<8 x i64> %vb, <8 x i64> %va, metadata !"slt", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_sle_vv_v8i64(<8 x i64> %va, <8 x i64> %vb, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sle_vv_v8i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m4, ta, ma
+; CHECK-NEXT:    vmsle.vv v16, v8, v12, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v16
+; CHECK-NEXT:    ret
+  %v = call <8 x i1> @llvm.vp.icmp.v8i64(<8 x i64> %va, <8 x i64> %vb, metadata !"sle", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_sle_vx_v8i64(<8 x i64> %va, i64 %b, <8 x i1> %m, i32 zeroext %evl) {
+; RV32-LABEL: icmp_sle_vx_v8i64:
+; RV32:       # %bb.0:
+; RV32-NEXT:    addi sp, sp, -16
+; RV32-NEXT:    .cfi_def_cfa_offset 16
+; RV32-NEXT:    sw a1, 12(sp)
+; RV32-NEXT:    sw a0, 8(sp)
+; RV32-NEXT:    vsetivli zero, 8, e64, m4, ta, mu
+; RV32-NEXT:    addi a0, sp, 8
+; RV32-NEXT:    vlse64.v v16, (a0), zero
+; RV32-NEXT:    vsetvli zero, a2, e64, m4, ta, ma
+; RV32-NEXT:    vmsle.vv v12, v8, v16, v0.t
+; RV32-NEXT:    vmv1r.v v0, v12
+; RV32-NEXT:    addi sp, sp, 16
+; RV32-NEXT:    ret
+;
+; RV64-LABEL: icmp_sle_vx_v8i64:
+; RV64:       # %bb.0:
+; RV64-NEXT:    vsetvli zero, a1, e64, m4, ta, ma
+; RV64-NEXT:    vmsle.vx v12, v8, a0, v0.t
+; RV64-NEXT:    vmv1r.v v0, v12
+; RV64-NEXT:    ret
+  %elt.head = insertelement <8 x i64> poison, i64 %b, i32 0
+  %vb = shufflevector <8 x i64> %elt.head, <8 x i64> poison, <8 x i32> zeroinitializer
+  %v = call <8 x i1> @llvm.vp.icmp.v8i64(<8 x i64> %va, <8 x i64> %vb, metadata !"sle", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_sle_vx_swap_v8i64(<8 x i64> %va, i64 %b, <8 x i1> %m, i32 zeroext %evl) {
+; RV32-LABEL: icmp_sle_vx_swap_v8i64:
+; RV32:       # %bb.0:
+; RV32-NEXT:    addi sp, sp, -16
+; RV32-NEXT:    .cfi_def_cfa_offset 16
+; RV32-NEXT:    sw a1, 12(sp)
+; RV32-NEXT:    sw a0, 8(sp)
+; RV32-NEXT:    vsetivli zero, 8, e64, m4, ta, mu
+; RV32-NEXT:    addi a0, sp, 8
+; RV32-NEXT:    vlse64.v v16, (a0), zero
+; RV32-NEXT:    vsetvli zero, a2, e64, m4, ta, ma
+; RV32-NEXT:    vmsle.vv v12, v16, v8, v0.t
+; RV32-NEXT:    vmv1r.v v0, v12
+; RV32-NEXT:    addi sp, sp, 16
+; RV32-NEXT:    ret
+;
+; RV64-LABEL: icmp_sle_vx_swap_v8i64:
+; RV64:       # %bb.0:
+; RV64-NEXT:    vsetivli zero, 8, e64, m4, ta, mu
+; RV64-NEXT:    vmv.v.x v16, a0
+; RV64-NEXT:    vsetvli zero, a1, e64, m4, ta, ma
+; RV64-NEXT:    vmsle.vv v12, v16, v8, v0.t
+; RV64-NEXT:    vmv1r.v v0, v12
+; RV64-NEXT:    ret
+  %elt.head = insertelement <8 x i64> poison, i64 %b, i32 0
+  %vb = shufflevector <8 x i64> %elt.head, <8 x i64> poison, <8 x i32> zeroinitializer
+  %v = call <8 x i1> @llvm.vp.icmp.v8i64(<8 x i64> %vb, <8 x i64> %va, metadata !"sle", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_sle_vi_v8i64(<8 x i64> %va, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sle_vi_v8i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m4, ta, ma
+; CHECK-NEXT:    vmsle.vi v12, v8, 4, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v12
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x i64> poison, i64 4, i32 0
+  %vb = shufflevector <8 x i64> %elt.head, <8 x i64> poison, <8 x i32> zeroinitializer
+  %v = call <8 x i1> @llvm.vp.icmp.v8i64(<8 x i64> %va, <8 x i64> %vb, metadata !"sle", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}
+
+define <8 x i1> @icmp_sle_vi_swap_v8i64(<8 x i64> %va, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sle_vi_swap_v8i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli zero, 8, e64, m4, ta, mu
+; CHECK-NEXT:    vmv.v.i v16, 4
+; CHECK-NEXT:    vsetvli zero, a0, e64, m4, ta, ma
+; CHECK-NEXT:    vmsle.vv v12, v16, v8, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v12
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x i64> poison, i64 4, i32 0
+  %vb = shufflevector <8 x i64> %elt.head, <8 x i64> poison, <8 x i32> zeroinitializer
+  %v = call <8 x i1> @llvm.vp.icmp.v8i64(<8 x i64> %vb, <8 x i64> %va, metadata !"sle", <8 x i1> %m, i32 %evl)
+  ret <8 x i1> %v
+}

llvm/test/CodeGen/RISCV/rvv/setcc-int-vp.ll

+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc -mtriple=riscv32 -mattr=+v -verify-machineinstrs < %s \
+; RUN:   | FileCheck %s --check-prefixes=CHECK,RV32
+; RUN: llc -mtriple=riscv64 -mattr=+v -verify-machineinstrs < %s \
+; RUN:   | FileCheck %s --check-prefixes=CHECK,RV64
+
+; FIXME: We're missing canonicalizations of ISD::VP_SETCC equivalent to those
+; for ISD::SETCC, e.g., splats aren't moved to the RHS.
+
+declare <vscale x 1 x i1> @llvm.vp.icmp.nxv1i8(<vscale x 1 x i8>, <vscale x 1 x i8>, metadata, <vscale x 1 x i1>, i32)
+
+define <vscale x 1 x i1> @icmp_eq_vv_nxv1i8(<vscale x 1 x i8> %va, <vscale x 1 x i8> %vb, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_eq_vv_nxv1i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e8, mf8, ta, ma
+; CHECK-NEXT:    vmseq.vv v0, v8, v9, v0.t
+; CHECK-NEXT:    ret
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i8(<vscale x 1 x i8> %va, <vscale x 1 x i8> %vb, metadata !"eq", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_eq_vx_nxv1i8(<vscale x 1 x i8> %va, i8 %b, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_eq_vx_nxv1i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a1, e8, mf8, ta, ma
+; CHECK-NEXT:    vmseq.vx v0, v8, a0, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x i8> poison, i8 %b, i32 0
+  %vb = shufflevector <vscale x 1 x i8> %elt.head, <vscale x 1 x i8> poison, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i8(<vscale x 1 x i8> %va, <vscale x 1 x i8> %vb, metadata !"eq", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_eq_vx_swap_nxv1i8(<vscale x 1 x i8> %va, i8 %b, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_eq_vx_swap_nxv1i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a1, e8, mf8, ta, ma
+; CHECK-NEXT:    vmseq.vx v0, v8, a0, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x i8> poison, i8 %b, i32 0
+  %vb = shufflevector <vscale x 1 x i8> %elt.head, <vscale x 1 x i8> poison, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i8(<vscale x 1 x i8> %vb, <vscale x 1 x i8> %va, metadata !"eq", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_eq_vi_nxv1i8(<vscale x 1 x i8> %va, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_eq_vi_nxv1i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e8, mf8, ta, ma
+; CHECK-NEXT:    vmseq.vi v0, v8, 4, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x i8> poison, i8 4, i32 0
+  %vb = shufflevector <vscale x 1 x i8> %elt.head, <vscale x 1 x i8> poison, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i8(<vscale x 1 x i8> %va, <vscale x 1 x i8> %vb, metadata !"eq", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_eq_vi_swap_nxv1i8(<vscale x 1 x i8> %va, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_eq_vi_swap_nxv1i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e8, mf8, ta, ma
+; CHECK-NEXT:    vmseq.vi v0, v8, 4, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x i8> poison, i8 4, i32 0
+  %vb = shufflevector <vscale x 1 x i8> %elt.head, <vscale x 1 x i8> poison, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i8(<vscale x 1 x i8> %vb, <vscale x 1 x i8> %va, metadata !"eq", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_ne_vv_nxv1i8(<vscale x 1 x i8> %va, <vscale x 1 x i8> %vb, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ne_vv_nxv1i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e8, mf8, ta, ma
+; CHECK-NEXT:    vmsne.vv v0, v8, v9, v0.t
+; CHECK-NEXT:    ret
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i8(<vscale x 1 x i8> %va, <vscale x 1 x i8> %vb, metadata !"ne", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_ne_vx_nxv1i8(<vscale x 1 x i8> %va, i8 %b, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ne_vx_nxv1i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a1, e8, mf8, ta, ma
+; CHECK-NEXT:    vmsne.vx v0, v8, a0, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x i8> poison, i8 %b, i32 0
+  %vb = shufflevector <vscale x 1 x i8> %elt.head, <vscale x 1 x i8> poison, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i8(<vscale x 1 x i8> %va, <vscale x 1 x i8> %vb, metadata !"ne", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_ne_vx_swap_nxv1i8(<vscale x 1 x i8> %va, i8 %b, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ne_vx_swap_nxv1i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a1, e8, mf8, ta, ma
+; CHECK-NEXT:    vmsne.vx v0, v8, a0, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x i8> poison, i8 %b, i32 0
+  %vb = shufflevector <vscale x 1 x i8> %elt.head, <vscale x 1 x i8> poison, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i8(<vscale x 1 x i8> %vb, <vscale x 1 x i8> %va, metadata !"ne", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_ne_vi_nxv1i8(<vscale x 1 x i8> %va, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ne_vi_nxv1i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e8, mf8, ta, ma
+; CHECK-NEXT:    vmsne.vi v0, v8, 4, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x i8> poison, i8 4, i32 0
+  %vb = shufflevector <vscale x 1 x i8> %elt.head, <vscale x 1 x i8> poison, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i8(<vscale x 1 x i8> %va, <vscale x 1 x i8> %vb, metadata !"ne", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_ne_vi_swap_nxv1i8(<vscale x 1 x i8> %va, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ne_vi_swap_nxv1i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e8, mf8, ta, ma
+; CHECK-NEXT:    vmsne.vi v0, v8, 4, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x i8> poison, i8 4, i32 0
+  %vb = shufflevector <vscale x 1 x i8> %elt.head, <vscale x 1 x i8> poison, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i8(<vscale x 1 x i8> %vb, <vscale x 1 x i8> %va, metadata !"ne", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_ugt_vv_nxv1i8(<vscale x 1 x i8> %va, <vscale x 1 x i8> %vb, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ugt_vv_nxv1i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e8, mf8, ta, ma
+; CHECK-NEXT:    vmsltu.vv v0, v9, v8, v0.t
+; CHECK-NEXT:    ret
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i8(<vscale x 1 x i8> %va, <vscale x 1 x i8> %vb, metadata !"ugt", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_ugt_vx_nxv1i8(<vscale x 1 x i8> %va, i8 %b, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ugt_vx_nxv1i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a1, e8, mf8, ta, ma
+; CHECK-NEXT:    vmsgtu.vx v0, v8, a0, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x i8> poison, i8 %b, i32 0
+  %vb = shufflevector <vscale x 1 x i8> %elt.head, <vscale x 1 x i8> poison, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i8(<vscale x 1 x i8> %va, <vscale x 1 x i8> %vb, metadata !"ugt", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_ugt_vx_swap_nxv1i8(<vscale x 1 x i8> %va, i8 %b, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ugt_vx_swap_nxv1i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a1, e8, mf8, ta, ma
+; CHECK-NEXT:    vmsltu.vx v0, v8, a0, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x i8> poison, i8 %b, i32 0
+  %vb = shufflevector <vscale x 1 x i8> %elt.head, <vscale x 1 x i8> poison, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i8(<vscale x 1 x i8> %vb, <vscale x 1 x i8> %va, metadata !"ugt", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_ugt_vi_nxv1i8(<vscale x 1 x i8> %va, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ugt_vi_nxv1i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e8, mf8, ta, ma
+; CHECK-NEXT:    vmsgtu.vi v0, v8, 4, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x i8> poison, i8 4, i32 0
+  %vb = shufflevector <vscale x 1 x i8> %elt.head, <vscale x 1 x i8> poison, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i8(<vscale x 1 x i8> %va, <vscale x 1 x i8> %vb, metadata !"ugt", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_ugt_vi_swap_nxv1i8(<vscale x 1 x i8> %va, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ugt_vi_swap_nxv1i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    li a1, 4
+; CHECK-NEXT:    vsetvli zero, a0, e8, mf8, ta, ma
+; CHECK-NEXT:    vmsltu.vx v0, v8, a1, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x i8> poison, i8 4, i32 0
+  %vb = shufflevector <vscale x 1 x i8> %elt.head, <vscale x 1 x i8> poison, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i8(<vscale x 1 x i8> %vb, <vscale x 1 x i8> %va, metadata !"ugt", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_uge_vv_nxv1i8(<vscale x 1 x i8> %va, <vscale x 1 x i8> %vb, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_uge_vv_nxv1i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e8, mf8, ta, ma
+; CHECK-NEXT:    vmsleu.vv v0, v9, v8, v0.t
+; CHECK-NEXT:    ret
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i8(<vscale x 1 x i8> %va, <vscale x 1 x i8> %vb, metadata !"uge", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_uge_vx_nxv1i8(<vscale x 1 x i8> %va, i8 %b, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_uge_vx_nxv1i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a2, zero, e8, mf8, ta, mu
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, a1, e8, mf8, ta, ma
+; CHECK-NEXT:    vmsleu.vv v0, v9, v8, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x i8> poison, i8 %b, i32 0
+  %vb = shufflevector <vscale x 1 x i8> %elt.head, <vscale x 1 x i8> poison, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i8(<vscale x 1 x i8> %va, <vscale x 1 x i8> %vb, metadata !"uge", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_uge_vx_swap_nxv1i8(<vscale x 1 x i8> %va, i8 %b, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_uge_vx_swap_nxv1i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a1, e8, mf8, ta, ma
+; CHECK-NEXT:    vmsleu.vx v0, v8, a0, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x i8> poison, i8 %b, i32 0
+  %vb = shufflevector <vscale x 1 x i8> %elt.head, <vscale x 1 x i8> poison, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i8(<vscale x 1 x i8> %vb, <vscale x 1 x i8> %va, metadata !"uge", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_uge_vi_nxv1i8(<vscale x 1 x i8> %va, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_uge_vi_nxv1i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e8, mf8, ta, ma
+; CHECK-NEXT:    vmsgtu.vi v0, v8, 3, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x i8> poison, i8 4, i32 0
+  %vb = shufflevector <vscale x 1 x i8> %elt.head, <vscale x 1 x i8> poison, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i8(<vscale x 1 x i8> %va, <vscale x 1 x i8> %vb, metadata !"uge", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_uge_vi_swap_nxv1i8(<vscale x 1 x i8> %va, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_uge_vi_swap_nxv1i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e8, mf8, ta, ma
+; CHECK-NEXT:    vmsleu.vi v0, v8, 4, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x i8> poison, i8 4, i32 0
+  %vb = shufflevector <vscale x 1 x i8> %elt.head, <vscale x 1 x i8> poison, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i8(<vscale x 1 x i8> %vb, <vscale x 1 x i8> %va, metadata !"uge", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_ult_vv_nxv1i8(<vscale x 1 x i8> %va, <vscale x 1 x i8> %vb, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ult_vv_nxv1i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e8, mf8, ta, ma
+; CHECK-NEXT:    vmsltu.vv v0, v8, v9, v0.t
+; CHECK-NEXT:    ret
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i8(<vscale x 1 x i8> %va, <vscale x 1 x i8> %vb, metadata !"ult", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_ult_vx_nxv1i8(<vscale x 1 x i8> %va, i8 %b, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ult_vx_nxv1i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a1, e8, mf8, ta, ma
+; CHECK-NEXT:    vmsltu.vx v0, v8, a0, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x i8> poison, i8 %b, i32 0
+  %vb = shufflevector <vscale x 1 x i8> %elt.head, <vscale x 1 x i8> poison, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i8(<vscale x 1 x i8> %va, <vscale x 1 x i8> %vb, metadata !"ult", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_ult_vx_swap_nxv1i8(<vscale x 1 x i8> %va, i8 %b, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ult_vx_swap_nxv1i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a1, e8, mf8, ta, ma
+; CHECK-NEXT:    vmsgtu.vx v0, v8, a0, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x i8> poison, i8 %b, i32 0
+  %vb = shufflevector <vscale x 1 x i8> %elt.head, <vscale x 1 x i8> poison, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i8(<vscale x 1 x i8> %vb, <vscale x 1 x i8> %va, metadata !"ult", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_ult_vi_nxv1i8(<vscale x 1 x i8> %va, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ult_vi_nxv1i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e8, mf8, ta, ma
+; CHECK-NEXT:    vmsleu.vi v0, v8, 3, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x i8> poison, i8 4, i32 0
+  %vb = shufflevector <vscale x 1 x i8> %elt.head, <vscale x 1 x i8> poison, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i8(<vscale x 1 x i8> %va, <vscale x 1 x i8> %vb, metadata !"ult", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_ult_vi_swap_nxv1i8(<vscale x 1 x i8> %va, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ult_vi_swap_nxv1i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e8, mf8, ta, ma
+; CHECK-NEXT:    vmsgtu.vi v0, v8, 4, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x i8> poison, i8 4, i32 0
+  %vb = shufflevector <vscale x 1 x i8> %elt.head, <vscale x 1 x i8> poison, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i8(<vscale x 1 x i8> %vb, <vscale x 1 x i8> %va, metadata !"ult", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_sgt_vv_nxv1i8(<vscale x 1 x i8> %va, <vscale x 1 x i8> %vb, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sgt_vv_nxv1i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e8, mf8, ta, ma
+; CHECK-NEXT:    vmslt.vv v0, v9, v8, v0.t
+; CHECK-NEXT:    ret
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i8(<vscale x 1 x i8> %va, <vscale x 1 x i8> %vb, metadata !"sgt", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_sgt_vx_nxv1i8(<vscale x 1 x i8> %va, i8 %b, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sgt_vx_nxv1i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a1, e8, mf8, ta, ma
+; CHECK-NEXT:    vmsgt.vx v0, v8, a0, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x i8> poison, i8 %b, i32 0
+  %vb = shufflevector <vscale x 1 x i8> %elt.head, <vscale x 1 x i8> poison, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i8(<vscale x 1 x i8> %va, <vscale x 1 x i8> %vb, metadata !"sgt", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_sgt_vx_swap_nxv1i8(<vscale x 1 x i8> %va, i8 %b, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sgt_vx_swap_nxv1i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a1, e8, mf8, ta, ma
+; CHECK-NEXT:    vmslt.vx v0, v8, a0, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x i8> poison, i8 %b, i32 0
+  %vb = shufflevector <vscale x 1 x i8> %elt.head, <vscale x 1 x i8> poison, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i8(<vscale x 1 x i8> %vb, <vscale x 1 x i8> %va, metadata !"sgt", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_sgt_vi_nxv1i8(<vscale x 1 x i8> %va, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sgt_vi_nxv1i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e8, mf8, ta, ma
+; CHECK-NEXT:    vmsgt.vi v0, v8, 4, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x i8> poison, i8 4, i32 0
+  %vb = shufflevector <vscale x 1 x i8> %elt.head, <vscale x 1 x i8> poison, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i8(<vscale x 1 x i8> %va, <vscale x 1 x i8> %vb, metadata !"sgt", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_sgt_vi_swap_nxv1i8(<vscale x 1 x i8> %va, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sgt_vi_swap_nxv1i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    li a1, 4
+; CHECK-NEXT:    vsetvli zero, a0, e8, mf8, ta, ma
+; CHECK-NEXT:    vmslt.vx v0, v8, a1, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x i8> poison, i8 4, i32 0
+  %vb = shufflevector <vscale x 1 x i8> %elt.head, <vscale x 1 x i8> poison, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i8(<vscale x 1 x i8> %vb, <vscale x 1 x i8> %va, metadata !"sgt", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_sge_vv_nxv1i8(<vscale x 1 x i8> %va, <vscale x 1 x i8> %vb, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sge_vv_nxv1i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e8, mf8, ta, ma
+; CHECK-NEXT:    vmsle.vv v0, v9, v8, v0.t
+; CHECK-NEXT:    ret
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i8(<vscale x 1 x i8> %va, <vscale x 1 x i8> %vb, metadata !"sge", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_sge_vx_nxv1i8(<vscale x 1 x i8> %va, i8 %b, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sge_vx_nxv1i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a2, zero, e8, mf8, ta, mu
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, a1, e8, mf8, ta, ma
+; CHECK-NEXT:    vmsle.vv v0, v9, v8, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x i8> poison, i8 %b, i32 0
+  %vb = shufflevector <vscale x 1 x i8> %elt.head, <vscale x 1 x i8> poison, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i8(<vscale x 1 x i8> %va, <vscale x 1 x i8> %vb, metadata !"sge", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_sge_vx_swap_nxv1i8(<vscale x 1 x i8> %va, i8 %b, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sge_vx_swap_nxv1i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a1, e8, mf8, ta, ma
+; CHECK-NEXT:    vmsle.vx v0, v8, a0, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x i8> poison, i8 %b, i32 0
+  %vb = shufflevector <vscale x 1 x i8> %elt.head, <vscale x 1 x i8> poison, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i8(<vscale x 1 x i8> %vb, <vscale x 1 x i8> %va, metadata !"sge", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_sge_vi_nxv1i8(<vscale x 1 x i8> %va, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sge_vi_nxv1i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e8, mf8, ta, ma
+; CHECK-NEXT:    vmsgt.vi v0, v8, 3, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x i8> poison, i8 4, i32 0
+  %vb = shufflevector <vscale x 1 x i8> %elt.head, <vscale x 1 x i8> poison, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i8(<vscale x 1 x i8> %va, <vscale x 1 x i8> %vb, metadata !"sge", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_sge_vi_swap_nxv1i8(<vscale x 1 x i8> %va, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sge_vi_swap_nxv1i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e8, mf8, ta, ma
+; CHECK-NEXT:    vmsle.vi v0, v8, 4, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x i8> poison, i8 4, i32 0
+  %vb = shufflevector <vscale x 1 x i8> %elt.head, <vscale x 1 x i8> poison, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i8(<vscale x 1 x i8> %vb, <vscale x 1 x i8> %va, metadata !"sge", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_slt_vv_nxv1i8(<vscale x 1 x i8> %va, <vscale x 1 x i8> %vb, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_slt_vv_nxv1i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e8, mf8, ta, ma
+; CHECK-NEXT:    vmslt.vv v0, v8, v9, v0.t
+; CHECK-NEXT:    ret
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i8(<vscale x 1 x i8> %va, <vscale x 1 x i8> %vb, metadata !"slt", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_slt_vx_nxv1i8(<vscale x 1 x i8> %va, i8 %b, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_slt_vx_nxv1i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a1, e8, mf8, ta, ma
+; CHECK-NEXT:    vmslt.vx v0, v8, a0, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x i8> poison, i8 %b, i32 0
+  %vb = shufflevector <vscale x 1 x i8> %elt.head, <vscale x 1 x i8> poison, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i8(<vscale x 1 x i8> %va, <vscale x 1 x i8> %vb, metadata !"slt", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_slt_vx_swap_nxv1i8(<vscale x 1 x i8> %va, i8 %b, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_slt_vx_swap_nxv1i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a1, e8, mf8, ta, ma
+; CHECK-NEXT:    vmsgt.vx v0, v8, a0, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x i8> poison, i8 %b, i32 0
+  %vb = shufflevector <vscale x 1 x i8> %elt.head, <vscale x 1 x i8> poison, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i8(<vscale x 1 x i8> %vb, <vscale x 1 x i8> %va, metadata !"slt", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_slt_vi_nxv1i8(<vscale x 1 x i8> %va, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_slt_vi_nxv1i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e8, mf8, ta, ma
+; CHECK-NEXT:    vmsle.vi v0, v8, 3, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x i8> poison, i8 4, i32 0
+  %vb = shufflevector <vscale x 1 x i8> %elt.head, <vscale x 1 x i8> poison, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i8(<vscale x 1 x i8> %va, <vscale x 1 x i8> %vb, metadata !"slt", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_slt_vi_swap_nxv1i8(<vscale x 1 x i8> %va, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_slt_vi_swap_nxv1i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e8, mf8, ta, ma
+; CHECK-NEXT:    vmsgt.vi v0, v8, 4, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x i8> poison, i8 4, i32 0
+  %vb = shufflevector <vscale x 1 x i8> %elt.head, <vscale x 1 x i8> poison, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i8(<vscale x 1 x i8> %vb, <vscale x 1 x i8> %va, metadata !"slt", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_sle_vv_nxv1i8(<vscale x 1 x i8> %va, <vscale x 1 x i8> %vb, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sle_vv_nxv1i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e8, mf8, ta, ma
+; CHECK-NEXT:    vmsle.vv v0, v8, v9, v0.t
+; CHECK-NEXT:    ret
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i8(<vscale x 1 x i8> %va, <vscale x 1 x i8> %vb, metadata !"sle", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_sle_vx_nxv1i8(<vscale x 1 x i8> %va, i8 %b, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sle_vx_nxv1i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a1, e8, mf8, ta, ma
+; CHECK-NEXT:    vmsle.vx v0, v8, a0, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x i8> poison, i8 %b, i32 0
+  %vb = shufflevector <vscale x 1 x i8> %elt.head, <vscale x 1 x i8> poison, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i8(<vscale x 1 x i8> %va, <vscale x 1 x i8> %vb, metadata !"sle", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_sle_vx_swap_nxv1i8(<vscale x 1 x i8> %va, i8 %b, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sle_vx_swap_nxv1i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a2, zero, e8, mf8, ta, mu
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, a1, e8, mf8, ta, ma
+; CHECK-NEXT:    vmsle.vv v0, v9, v8, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x i8> poison, i8 %b, i32 0
+  %vb = shufflevector <vscale x 1 x i8> %elt.head, <vscale x 1 x i8> poison, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i8(<vscale x 1 x i8> %vb, <vscale x 1 x i8> %va, metadata !"sle", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_sle_vi_nxv1i8(<vscale x 1 x i8> %va, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sle_vi_nxv1i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e8, mf8, ta, ma
+; CHECK-NEXT:    vmsle.vi v0, v8, 4, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x i8> poison, i8 4, i32 0
+  %vb = shufflevector <vscale x 1 x i8> %elt.head, <vscale x 1 x i8> poison, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i8(<vscale x 1 x i8> %va, <vscale x 1 x i8> %vb, metadata !"sle", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_sle_vi_swap_nxv1i8(<vscale x 1 x i8> %va, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sle_vi_swap_nxv1i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e8, mf8, ta, mu
+; CHECK-NEXT:    vmv.v.i v9, 4
+; CHECK-NEXT:    vsetvli zero, a0, e8, mf8, ta, ma
+; CHECK-NEXT:    vmsle.vv v0, v9, v8, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x i8> poison, i8 4, i32 0
+  %vb = shufflevector <vscale x 1 x i8> %elt.head, <vscale x 1 x i8> poison, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i8(<vscale x 1 x i8> %vb, <vscale x 1 x i8> %va, metadata !"sle", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+declare <vscale x 8 x i1> @llvm.vp.icmp.nxv8i8(<vscale x 8 x i8>, <vscale x 8 x i8>, metadata, <vscale x 8 x i1>, i32)
+
+define <vscale x 8 x i1> @icmp_eq_vv_nxv8i8(<vscale x 8 x i8> %va, <vscale x 8 x i8> %vb, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_eq_vv_nxv8i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e8, m1, ta, ma
+; CHECK-NEXT:    vmseq.vv v0, v8, v9, v0.t
+; CHECK-NEXT:    ret
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i8(<vscale x 8 x i8> %va, <vscale x 8 x i8> %vb, metadata !"eq", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_eq_vx_nxv8i8(<vscale x 8 x i8> %va, i8 %b, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_eq_vx_nxv8i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a1, e8, m1, ta, ma
+; CHECK-NEXT:    vmseq.vx v0, v8, a0, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x i8> poison, i8 %b, i32 0
+  %vb = shufflevector <vscale x 8 x i8> %elt.head, <vscale x 8 x i8> poison, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i8(<vscale x 8 x i8> %va, <vscale x 8 x i8> %vb, metadata !"eq", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_eq_vx_swap_nxv8i8(<vscale x 8 x i8> %va, i8 %b, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_eq_vx_swap_nxv8i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a1, e8, m1, ta, ma
+; CHECK-NEXT:    vmseq.vx v0, v8, a0, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x i8> poison, i8 %b, i32 0
+  %vb = shufflevector <vscale x 8 x i8> %elt.head, <vscale x 8 x i8> poison, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i8(<vscale x 8 x i8> %vb, <vscale x 8 x i8> %va, metadata !"eq", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_eq_vi_nxv8i8(<vscale x 8 x i8> %va, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_eq_vi_nxv8i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e8, m1, ta, ma
+; CHECK-NEXT:    vmseq.vi v0, v8, 4, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x i8> poison, i8 4, i32 0
+  %vb = shufflevector <vscale x 8 x i8> %elt.head, <vscale x 8 x i8> poison, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i8(<vscale x 8 x i8> %va, <vscale x 8 x i8> %vb, metadata !"eq", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_eq_vi_swap_nxv8i8(<vscale x 8 x i8> %va, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_eq_vi_swap_nxv8i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e8, m1, ta, ma
+; CHECK-NEXT:    vmseq.vi v0, v8, 4, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x i8> poison, i8 4, i32 0
+  %vb = shufflevector <vscale x 8 x i8> %elt.head, <vscale x 8 x i8> poison, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i8(<vscale x 8 x i8> %vb, <vscale x 8 x i8> %va, metadata !"eq", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_ne_vv_nxv8i8(<vscale x 8 x i8> %va, <vscale x 8 x i8> %vb, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ne_vv_nxv8i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e8, m1, ta, ma
+; CHECK-NEXT:    vmsne.vv v0, v8, v9, v0.t
+; CHECK-NEXT:    ret
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i8(<vscale x 8 x i8> %va, <vscale x 8 x i8> %vb, metadata !"ne", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_ne_vx_nxv8i8(<vscale x 8 x i8> %va, i8 %b, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ne_vx_nxv8i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a1, e8, m1, ta, ma
+; CHECK-NEXT:    vmsne.vx v0, v8, a0, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x i8> poison, i8 %b, i32 0
+  %vb = shufflevector <vscale x 8 x i8> %elt.head, <vscale x 8 x i8> poison, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i8(<vscale x 8 x i8> %va, <vscale x 8 x i8> %vb, metadata !"ne", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_ne_vx_swap_nxv8i8(<vscale x 8 x i8> %va, i8 %b, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ne_vx_swap_nxv8i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a1, e8, m1, ta, ma
+; CHECK-NEXT:    vmsne.vx v0, v8, a0, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x i8> poison, i8 %b, i32 0
+  %vb = shufflevector <vscale x 8 x i8> %elt.head, <vscale x 8 x i8> poison, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i8(<vscale x 8 x i8> %vb, <vscale x 8 x i8> %va, metadata !"ne", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_ne_vi_nxv8i8(<vscale x 8 x i8> %va, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ne_vi_nxv8i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e8, m1, ta, ma
+; CHECK-NEXT:    vmsne.vi v0, v8, 4, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x i8> poison, i8 4, i32 0
+  %vb = shufflevector <vscale x 8 x i8> %elt.head, <vscale x 8 x i8> poison, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i8(<vscale x 8 x i8> %va, <vscale x 8 x i8> %vb, metadata !"ne", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_ne_vi_swap_nxv8i8(<vscale x 8 x i8> %va, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ne_vi_swap_nxv8i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e8, m1, ta, ma
+; CHECK-NEXT:    vmsne.vi v0, v8, 4, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x i8> poison, i8 4, i32 0
+  %vb = shufflevector <vscale x 8 x i8> %elt.head, <vscale x 8 x i8> poison, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i8(<vscale x 8 x i8> %vb, <vscale x 8 x i8> %va, metadata !"ne", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_ugt_vv_nxv8i8(<vscale x 8 x i8> %va, <vscale x 8 x i8> %vb, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ugt_vv_nxv8i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e8, m1, ta, ma
+; CHECK-NEXT:    vmsltu.vv v0, v9, v8, v0.t
+; CHECK-NEXT:    ret
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i8(<vscale x 8 x i8> %va, <vscale x 8 x i8> %vb, metadata !"ugt", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_ugt_vx_nxv8i8(<vscale x 8 x i8> %va, i8 %b, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ugt_vx_nxv8i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a1, e8, m1, ta, ma
+; CHECK-NEXT:    vmsgtu.vx v0, v8, a0, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x i8> poison, i8 %b, i32 0
+  %vb = shufflevector <vscale x 8 x i8> %elt.head, <vscale x 8 x i8> poison, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i8(<vscale x 8 x i8> %va, <vscale x 8 x i8> %vb, metadata !"ugt", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_ugt_vx_swap_nxv8i8(<vscale x 8 x i8> %va, i8 %b, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ugt_vx_swap_nxv8i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a1, e8, m1, ta, ma
+; CHECK-NEXT:    vmsltu.vx v0, v8, a0, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x i8> poison, i8 %b, i32 0
+  %vb = shufflevector <vscale x 8 x i8> %elt.head, <vscale x 8 x i8> poison, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i8(<vscale x 8 x i8> %vb, <vscale x 8 x i8> %va, metadata !"ugt", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_ugt_vi_nxv8i8(<vscale x 8 x i8> %va, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ugt_vi_nxv8i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e8, m1, ta, ma
+; CHECK-NEXT:    vmsgtu.vi v0, v8, 4, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x i8> poison, i8 4, i32 0
+  %vb = shufflevector <vscale x 8 x i8> %elt.head, <vscale x 8 x i8> poison, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i8(<vscale x 8 x i8> %va, <vscale x 8 x i8> %vb, metadata !"ugt", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_ugt_vi_swap_nxv8i8(<vscale x 8 x i8> %va, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ugt_vi_swap_nxv8i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    li a1, 4
+; CHECK-NEXT:    vsetvli zero, a0, e8, m1, ta, ma
+; CHECK-NEXT:    vmsltu.vx v0, v8, a1, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x i8> poison, i8 4, i32 0
+  %vb = shufflevector <vscale x 8 x i8> %elt.head, <vscale x 8 x i8> poison, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i8(<vscale x 8 x i8> %vb, <vscale x 8 x i8> %va, metadata !"ugt", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_uge_vv_nxv8i8(<vscale x 8 x i8> %va, <vscale x 8 x i8> %vb, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_uge_vv_nxv8i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e8, m1, ta, ma
+; CHECK-NEXT:    vmsleu.vv v0, v9, v8, v0.t
+; CHECK-NEXT:    ret
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i8(<vscale x 8 x i8> %va, <vscale x 8 x i8> %vb, metadata !"uge", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_uge_vx_nxv8i8(<vscale x 8 x i8> %va, i8 %b, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_uge_vx_nxv8i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a2, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, a1, e8, m1, ta, ma
+; CHECK-NEXT:    vmsleu.vv v0, v9, v8, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x i8> poison, i8 %b, i32 0
+  %vb = shufflevector <vscale x 8 x i8> %elt.head, <vscale x 8 x i8> poison, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i8(<vscale x 8 x i8> %va, <vscale x 8 x i8> %vb, metadata !"uge", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_uge_vx_swap_nxv8i8(<vscale x 8 x i8> %va, i8 %b, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_uge_vx_swap_nxv8i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a1, e8, m1, ta, ma
+; CHECK-NEXT:    vmsleu.vx v0, v8, a0, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x i8> poison, i8 %b, i32 0
+  %vb = shufflevector <vscale x 8 x i8> %elt.head, <vscale x 8 x i8> poison, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i8(<vscale x 8 x i8> %vb, <vscale x 8 x i8> %va, metadata !"uge", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_uge_vi_nxv8i8(<vscale x 8 x i8> %va, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_uge_vi_nxv8i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e8, m1, ta, ma
+; CHECK-NEXT:    vmsgtu.vi v0, v8, 3, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x i8> poison, i8 4, i32 0
+  %vb = shufflevector <vscale x 8 x i8> %elt.head, <vscale x 8 x i8> poison, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i8(<vscale x 8 x i8> %va, <vscale x 8 x i8> %vb, metadata !"uge", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_uge_vi_swap_nxv8i8(<vscale x 8 x i8> %va, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_uge_vi_swap_nxv8i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e8, m1, ta, ma
+; CHECK-NEXT:    vmsleu.vi v0, v8, 4, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x i8> poison, i8 4, i32 0
+  %vb = shufflevector <vscale x 8 x i8> %elt.head, <vscale x 8 x i8> poison, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i8(<vscale x 8 x i8> %vb, <vscale x 8 x i8> %va, metadata !"uge", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_ult_vv_nxv8i8(<vscale x 8 x i8> %va, <vscale x 8 x i8> %vb, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ult_vv_nxv8i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e8, m1, ta, ma
+; CHECK-NEXT:    vmsltu.vv v0, v8, v9, v0.t
+; CHECK-NEXT:    ret
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i8(<vscale x 8 x i8> %va, <vscale x 8 x i8> %vb, metadata !"ult", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_ult_vx_nxv8i8(<vscale x 8 x i8> %va, i8 %b, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ult_vx_nxv8i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a1, e8, m1, ta, ma
+; CHECK-NEXT:    vmsltu.vx v0, v8, a0, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x i8> poison, i8 %b, i32 0
+  %vb = shufflevector <vscale x 8 x i8> %elt.head, <vscale x 8 x i8> poison, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i8(<vscale x 8 x i8> %va, <vscale x 8 x i8> %vb, metadata !"ult", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_ult_vx_swap_nxv8i8(<vscale x 8 x i8> %va, i8 %b, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ult_vx_swap_nxv8i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a1, e8, m1, ta, ma
+; CHECK-NEXT:    vmsgtu.vx v0, v8, a0, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x i8> poison, i8 %b, i32 0
+  %vb = shufflevector <vscale x 8 x i8> %elt.head, <vscale x 8 x i8> poison, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i8(<vscale x 8 x i8> %vb, <vscale x 8 x i8> %va, metadata !"ult", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_ult_vi_nxv8i8(<vscale x 8 x i8> %va, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ult_vi_nxv8i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e8, m1, ta, ma
+; CHECK-NEXT:    vmsleu.vi v0, v8, 3, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x i8> poison, i8 4, i32 0
+  %vb = shufflevector <vscale x 8 x i8> %elt.head, <vscale x 8 x i8> poison, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i8(<vscale x 8 x i8> %va, <vscale x 8 x i8> %vb, metadata !"ult", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_ult_vi_swap_nxv8i8(<vscale x 8 x i8> %va, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ult_vi_swap_nxv8i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e8, m1, ta, ma
+; CHECK-NEXT:    vmsgtu.vi v0, v8, 4, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x i8> poison, i8 4, i32 0
+  %vb = shufflevector <vscale x 8 x i8> %elt.head, <vscale x 8 x i8> poison, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i8(<vscale x 8 x i8> %vb, <vscale x 8 x i8> %va, metadata !"ult", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_sgt_vv_nxv8i8(<vscale x 8 x i8> %va, <vscale x 8 x i8> %vb, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sgt_vv_nxv8i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e8, m1, ta, ma
+; CHECK-NEXT:    vmslt.vv v0, v9, v8, v0.t
+; CHECK-NEXT:    ret
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i8(<vscale x 8 x i8> %va, <vscale x 8 x i8> %vb, metadata !"sgt", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_sgt_vx_nxv8i8(<vscale x 8 x i8> %va, i8 %b, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sgt_vx_nxv8i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a1, e8, m1, ta, ma
+; CHECK-NEXT:    vmsgt.vx v0, v8, a0, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x i8> poison, i8 %b, i32 0
+  %vb = shufflevector <vscale x 8 x i8> %elt.head, <vscale x 8 x i8> poison, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i8(<vscale x 8 x i8> %va, <vscale x 8 x i8> %vb, metadata !"sgt", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_sgt_vx_swap_nxv8i8(<vscale x 8 x i8> %va, i8 %b, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sgt_vx_swap_nxv8i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a1, e8, m1, ta, ma
+; CHECK-NEXT:    vmslt.vx v0, v8, a0, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x i8> poison, i8 %b, i32 0
+  %vb = shufflevector <vscale x 8 x i8> %elt.head, <vscale x 8 x i8> poison, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i8(<vscale x 8 x i8> %vb, <vscale x 8 x i8> %va, metadata !"sgt", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_sgt_vi_nxv8i8(<vscale x 8 x i8> %va, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sgt_vi_nxv8i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e8, m1, ta, ma
+; CHECK-NEXT:    vmsgt.vi v0, v8, 4, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x i8> poison, i8 4, i32 0
+  %vb = shufflevector <vscale x 8 x i8> %elt.head, <vscale x 8 x i8> poison, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i8(<vscale x 8 x i8> %va, <vscale x 8 x i8> %vb, metadata !"sgt", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_sgt_vi_swap_nxv8i8(<vscale x 8 x i8> %va, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sgt_vi_swap_nxv8i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    li a1, 4
+; CHECK-NEXT:    vsetvli zero, a0, e8, m1, ta, ma
+; CHECK-NEXT:    vmslt.vx v0, v8, a1, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x i8> poison, i8 4, i32 0
+  %vb = shufflevector <vscale x 8 x i8> %elt.head, <vscale x 8 x i8> poison, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i8(<vscale x 8 x i8> %vb, <vscale x 8 x i8> %va, metadata !"sgt", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_sge_vv_nxv8i8(<vscale x 8 x i8> %va, <vscale x 8 x i8> %vb, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sge_vv_nxv8i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e8, m1, ta, ma
+; CHECK-NEXT:    vmsle.vv v0, v9, v8, v0.t
+; CHECK-NEXT:    ret
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i8(<vscale x 8 x i8> %va, <vscale x 8 x i8> %vb, metadata !"sge", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_sge_vx_nxv8i8(<vscale x 8 x i8> %va, i8 %b, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sge_vx_nxv8i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a2, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, a1, e8, m1, ta, ma
+; CHECK-NEXT:    vmsle.vv v0, v9, v8, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x i8> poison, i8 %b, i32 0
+  %vb = shufflevector <vscale x 8 x i8> %elt.head, <vscale x 8 x i8> poison, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i8(<vscale x 8 x i8> %va, <vscale x 8 x i8> %vb, metadata !"sge", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_sge_vx_swap_nxv8i8(<vscale x 8 x i8> %va, i8 %b, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sge_vx_swap_nxv8i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a1, e8, m1, ta, ma
+; CHECK-NEXT:    vmsle.vx v0, v8, a0, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x i8> poison, i8 %b, i32 0
+  %vb = shufflevector <vscale x 8 x i8> %elt.head, <vscale x 8 x i8> poison, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i8(<vscale x 8 x i8> %vb, <vscale x 8 x i8> %va, metadata !"sge", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_sge_vi_nxv8i8(<vscale x 8 x i8> %va, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sge_vi_nxv8i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e8, m1, ta, ma
+; CHECK-NEXT:    vmsgt.vi v0, v8, 3, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x i8> poison, i8 4, i32 0
+  %vb = shufflevector <vscale x 8 x i8> %elt.head, <vscale x 8 x i8> poison, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i8(<vscale x 8 x i8> %va, <vscale x 8 x i8> %vb, metadata !"sge", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_sge_vi_swap_nxv8i8(<vscale x 8 x i8> %va, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sge_vi_swap_nxv8i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e8, m1, ta, ma
+; CHECK-NEXT:    vmsle.vi v0, v8, 4, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x i8> poison, i8 4, i32 0
+  %vb = shufflevector <vscale x 8 x i8> %elt.head, <vscale x 8 x i8> poison, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i8(<vscale x 8 x i8> %vb, <vscale x 8 x i8> %va, metadata !"sge", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_slt_vv_nxv8i8(<vscale x 8 x i8> %va, <vscale x 8 x i8> %vb, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_slt_vv_nxv8i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e8, m1, ta, ma
+; CHECK-NEXT:    vmslt.vv v0, v8, v9, v0.t
+; CHECK-NEXT:    ret
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i8(<vscale x 8 x i8> %va, <vscale x 8 x i8> %vb, metadata !"slt", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_slt_vx_nxv8i8(<vscale x 8 x i8> %va, i8 %b, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_slt_vx_nxv8i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a1, e8, m1, ta, ma
+; CHECK-NEXT:    vmslt.vx v0, v8, a0, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x i8> poison, i8 %b, i32 0
+  %vb = shufflevector <vscale x 8 x i8> %elt.head, <vscale x 8 x i8> poison, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i8(<vscale x 8 x i8> %va, <vscale x 8 x i8> %vb, metadata !"slt", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_slt_vx_swap_nxv8i8(<vscale x 8 x i8> %va, i8 %b, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_slt_vx_swap_nxv8i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a1, e8, m1, ta, ma
+; CHECK-NEXT:    vmsgt.vx v0, v8, a0, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x i8> poison, i8 %b, i32 0
+  %vb = shufflevector <vscale x 8 x i8> %elt.head, <vscale x 8 x i8> poison, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i8(<vscale x 8 x i8> %vb, <vscale x 8 x i8> %va, metadata !"slt", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_slt_vi_nxv8i8(<vscale x 8 x i8> %va, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_slt_vi_nxv8i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e8, m1, ta, ma
+; CHECK-NEXT:    vmsle.vi v0, v8, 3, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x i8> poison, i8 4, i32 0
+  %vb = shufflevector <vscale x 8 x i8> %elt.head, <vscale x 8 x i8> poison, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i8(<vscale x 8 x i8> %va, <vscale x 8 x i8> %vb, metadata !"slt", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_slt_vi_swap_nxv8i8(<vscale x 8 x i8> %va, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_slt_vi_swap_nxv8i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e8, m1, ta, ma
+; CHECK-NEXT:    vmsgt.vi v0, v8, 4, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x i8> poison, i8 4, i32 0
+  %vb = shufflevector <vscale x 8 x i8> %elt.head, <vscale x 8 x i8> poison, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i8(<vscale x 8 x i8> %vb, <vscale x 8 x i8> %va, metadata !"slt", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_sle_vv_nxv8i8(<vscale x 8 x i8> %va, <vscale x 8 x i8> %vb, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sle_vv_nxv8i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e8, m1, ta, ma
+; CHECK-NEXT:    vmsle.vv v0, v8, v9, v0.t
+; CHECK-NEXT:    ret
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i8(<vscale x 8 x i8> %va, <vscale x 8 x i8> %vb, metadata !"sle", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_sle_vx_nxv8i8(<vscale x 8 x i8> %va, i8 %b, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sle_vx_nxv8i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a1, e8, m1, ta, ma
+; CHECK-NEXT:    vmsle.vx v0, v8, a0, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x i8> poison, i8 %b, i32 0
+  %vb = shufflevector <vscale x 8 x i8> %elt.head, <vscale x 8 x i8> poison, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i8(<vscale x 8 x i8> %va, <vscale x 8 x i8> %vb, metadata !"sle", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_sle_vx_swap_nxv8i8(<vscale x 8 x i8> %va, i8 %b, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sle_vx_swap_nxv8i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a2, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, a1, e8, m1, ta, ma
+; CHECK-NEXT:    vmsle.vv v0, v9, v8, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x i8> poison, i8 %b, i32 0
+  %vb = shufflevector <vscale x 8 x i8> %elt.head, <vscale x 8 x i8> poison, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i8(<vscale x 8 x i8> %vb, <vscale x 8 x i8> %va, metadata !"sle", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_sle_vi_nxv8i8(<vscale x 8 x i8> %va, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sle_vi_nxv8i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e8, m1, ta, ma
+; CHECK-NEXT:    vmsle.vi v0, v8, 4, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x i8> poison, i8 4, i32 0
+  %vb = shufflevector <vscale x 8 x i8> %elt.head, <vscale x 8 x i8> poison, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i8(<vscale x 8 x i8> %va, <vscale x 8 x i8> %vb, metadata !"sle", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_sle_vi_swap_nxv8i8(<vscale x 8 x i8> %va, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sle_vi_swap_nxv8i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vmv.v.i v9, 4
+; CHECK-NEXT:    vsetvli zero, a0, e8, m1, ta, ma
+; CHECK-NEXT:    vmsle.vv v0, v9, v8, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x i8> poison, i8 4, i32 0
+  %vb = shufflevector <vscale x 8 x i8> %elt.head, <vscale x 8 x i8> poison, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i8(<vscale x 8 x i8> %vb, <vscale x 8 x i8> %va, metadata !"sle", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+declare <vscale x 1 x i1> @llvm.vp.icmp.nxv1i32(<vscale x 1 x i32>, <vscale x 1 x i32>, metadata, <vscale x 1 x i1>, i32)
+
+define <vscale x 1 x i1> @icmp_eq_vv_nxv1i32(<vscale x 1 x i32> %va, <vscale x 1 x i32> %vb, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_eq_vv_nxv1i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, mf2, ta, ma
+; CHECK-NEXT:    vmseq.vv v0, v8, v9, v0.t
+; CHECK-NEXT:    ret
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i32(<vscale x 1 x i32> %va, <vscale x 1 x i32> %vb, metadata !"eq", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_eq_vx_nxv1i32(<vscale x 1 x i32> %va, i32 %b, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_eq_vx_nxv1i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a1, e32, mf2, ta, ma
+; CHECK-NEXT:    vmseq.vx v0, v8, a0, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x i32> poison, i32 %b, i32 0
+  %vb = shufflevector <vscale x 1 x i32> %elt.head, <vscale x 1 x i32> poison, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i32(<vscale x 1 x i32> %va, <vscale x 1 x i32> %vb, metadata !"eq", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_eq_vx_swap_nxv1i32(<vscale x 1 x i32> %va, i32 %b, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_eq_vx_swap_nxv1i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a1, e32, mf2, ta, ma
+; CHECK-NEXT:    vmseq.vx v0, v8, a0, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x i32> poison, i32 %b, i32 0
+  %vb = shufflevector <vscale x 1 x i32> %elt.head, <vscale x 1 x i32> poison, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i32(<vscale x 1 x i32> %vb, <vscale x 1 x i32> %va, metadata !"eq", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_eq_vi_nxv1i32(<vscale x 1 x i32> %va, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_eq_vi_nxv1i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, mf2, ta, ma
+; CHECK-NEXT:    vmseq.vi v0, v8, 4, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x i32> poison, i32 4, i32 0
+  %vb = shufflevector <vscale x 1 x i32> %elt.head, <vscale x 1 x i32> poison, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i32(<vscale x 1 x i32> %va, <vscale x 1 x i32> %vb, metadata !"eq", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_eq_vi_swap_nxv1i32(<vscale x 1 x i32> %va, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_eq_vi_swap_nxv1i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, mf2, ta, ma
+; CHECK-NEXT:    vmseq.vi v0, v8, 4, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x i32> poison, i32 4, i32 0
+  %vb = shufflevector <vscale x 1 x i32> %elt.head, <vscale x 1 x i32> poison, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i32(<vscale x 1 x i32> %vb, <vscale x 1 x i32> %va, metadata !"eq", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_ne_vv_nxv1i32(<vscale x 1 x i32> %va, <vscale x 1 x i32> %vb, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ne_vv_nxv1i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, mf2, ta, ma
+; CHECK-NEXT:    vmsne.vv v0, v8, v9, v0.t
+; CHECK-NEXT:    ret
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i32(<vscale x 1 x i32> %va, <vscale x 1 x i32> %vb, metadata !"ne", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_ne_vx_nxv1i32(<vscale x 1 x i32> %va, i32 %b, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ne_vx_nxv1i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a1, e32, mf2, ta, ma
+; CHECK-NEXT:    vmsne.vx v0, v8, a0, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x i32> poison, i32 %b, i32 0
+  %vb = shufflevector <vscale x 1 x i32> %elt.head, <vscale x 1 x i32> poison, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i32(<vscale x 1 x i32> %va, <vscale x 1 x i32> %vb, metadata !"ne", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_ne_vx_swap_nxv1i32(<vscale x 1 x i32> %va, i32 %b, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ne_vx_swap_nxv1i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a1, e32, mf2, ta, ma
+; CHECK-NEXT:    vmsne.vx v0, v8, a0, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x i32> poison, i32 %b, i32 0
+  %vb = shufflevector <vscale x 1 x i32> %elt.head, <vscale x 1 x i32> poison, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i32(<vscale x 1 x i32> %vb, <vscale x 1 x i32> %va, metadata !"ne", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_ne_vi_nxv1i32(<vscale x 1 x i32> %va, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ne_vi_nxv1i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, mf2, ta, ma
+; CHECK-NEXT:    vmsne.vi v0, v8, 4, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x i32> poison, i32 4, i32 0
+  %vb = shufflevector <vscale x 1 x i32> %elt.head, <vscale x 1 x i32> poison, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i32(<vscale x 1 x i32> %va, <vscale x 1 x i32> %vb, metadata !"ne", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_ne_vi_swap_nxv1i32(<vscale x 1 x i32> %va, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ne_vi_swap_nxv1i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, mf2, ta, ma
+; CHECK-NEXT:    vmsne.vi v0, v8, 4, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x i32> poison, i32 4, i32 0
+  %vb = shufflevector <vscale x 1 x i32> %elt.head, <vscale x 1 x i32> poison, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i32(<vscale x 1 x i32> %vb, <vscale x 1 x i32> %va, metadata !"ne", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_ugt_vv_nxv1i32(<vscale x 1 x i32> %va, <vscale x 1 x i32> %vb, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ugt_vv_nxv1i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, mf2, ta, ma
+; CHECK-NEXT:    vmsltu.vv v0, v9, v8, v0.t
+; CHECK-NEXT:    ret
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i32(<vscale x 1 x i32> %va, <vscale x 1 x i32> %vb, metadata !"ugt", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_ugt_vx_nxv1i32(<vscale x 1 x i32> %va, i32 %b, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ugt_vx_nxv1i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a1, e32, mf2, ta, ma
+; CHECK-NEXT:    vmsgtu.vx v0, v8, a0, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x i32> poison, i32 %b, i32 0
+  %vb = shufflevector <vscale x 1 x i32> %elt.head, <vscale x 1 x i32> poison, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i32(<vscale x 1 x i32> %va, <vscale x 1 x i32> %vb, metadata !"ugt", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_ugt_vx_swap_nxv1i32(<vscale x 1 x i32> %va, i32 %b, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ugt_vx_swap_nxv1i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a1, e32, mf2, ta, ma
+; CHECK-NEXT:    vmsltu.vx v0, v8, a0, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x i32> poison, i32 %b, i32 0
+  %vb = shufflevector <vscale x 1 x i32> %elt.head, <vscale x 1 x i32> poison, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i32(<vscale x 1 x i32> %vb, <vscale x 1 x i32> %va, metadata !"ugt", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_ugt_vi_nxv1i32(<vscale x 1 x i32> %va, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ugt_vi_nxv1i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, mf2, ta, ma
+; CHECK-NEXT:    vmsgtu.vi v0, v8, 4, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x i32> poison, i32 4, i32 0
+  %vb = shufflevector <vscale x 1 x i32> %elt.head, <vscale x 1 x i32> poison, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i32(<vscale x 1 x i32> %va, <vscale x 1 x i32> %vb, metadata !"ugt", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_ugt_vi_swap_nxv1i32(<vscale x 1 x i32> %va, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ugt_vi_swap_nxv1i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    li a1, 4
+; CHECK-NEXT:    vsetvli zero, a0, e32, mf2, ta, ma
+; CHECK-NEXT:    vmsltu.vx v0, v8, a1, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x i32> poison, i32 4, i32 0
+  %vb = shufflevector <vscale x 1 x i32> %elt.head, <vscale x 1 x i32> poison, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i32(<vscale x 1 x i32> %vb, <vscale x 1 x i32> %va, metadata !"ugt", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_uge_vv_nxv1i32(<vscale x 1 x i32> %va, <vscale x 1 x i32> %vb, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_uge_vv_nxv1i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, mf2, ta, ma
+; CHECK-NEXT:    vmsleu.vv v0, v9, v8, v0.t
+; CHECK-NEXT:    ret
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i32(<vscale x 1 x i32> %va, <vscale x 1 x i32> %vb, metadata !"uge", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_uge_vx_nxv1i32(<vscale x 1 x i32> %va, i32 %b, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_uge_vx_nxv1i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a2, zero, e32, mf2, ta, mu
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, a1, e32, mf2, ta, ma
+; CHECK-NEXT:    vmsleu.vv v0, v9, v8, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x i32> poison, i32 %b, i32 0
+  %vb = shufflevector <vscale x 1 x i32> %elt.head, <vscale x 1 x i32> poison, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i32(<vscale x 1 x i32> %va, <vscale x 1 x i32> %vb, metadata !"uge", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_uge_vx_swap_nxv1i32(<vscale x 1 x i32> %va, i32 %b, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_uge_vx_swap_nxv1i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a1, e32, mf2, ta, ma
+; CHECK-NEXT:    vmsleu.vx v0, v8, a0, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x i32> poison, i32 %b, i32 0
+  %vb = shufflevector <vscale x 1 x i32> %elt.head, <vscale x 1 x i32> poison, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i32(<vscale x 1 x i32> %vb, <vscale x 1 x i32> %va, metadata !"uge", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_uge_vi_nxv1i32(<vscale x 1 x i32> %va, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_uge_vi_nxv1i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, mf2, ta, ma
+; CHECK-NEXT:    vmsgtu.vi v0, v8, 3, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x i32> poison, i32 4, i32 0
+  %vb = shufflevector <vscale x 1 x i32> %elt.head, <vscale x 1 x i32> poison, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i32(<vscale x 1 x i32> %va, <vscale x 1 x i32> %vb, metadata !"uge", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_uge_vi_swap_nxv1i32(<vscale x 1 x i32> %va, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_uge_vi_swap_nxv1i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, mf2, ta, ma
+; CHECK-NEXT:    vmsleu.vi v0, v8, 4, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x i32> poison, i32 4, i32 0
+  %vb = shufflevector <vscale x 1 x i32> %elt.head, <vscale x 1 x i32> poison, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i32(<vscale x 1 x i32> %vb, <vscale x 1 x i32> %va, metadata !"uge", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_ult_vv_nxv1i32(<vscale x 1 x i32> %va, <vscale x 1 x i32> %vb, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ult_vv_nxv1i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, mf2, ta, ma
+; CHECK-NEXT:    vmsltu.vv v0, v8, v9, v0.t
+; CHECK-NEXT:    ret
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i32(<vscale x 1 x i32> %va, <vscale x 1 x i32> %vb, metadata !"ult", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_ult_vx_nxv1i32(<vscale x 1 x i32> %va, i32 %b, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ult_vx_nxv1i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a1, e32, mf2, ta, ma
+; CHECK-NEXT:    vmsltu.vx v0, v8, a0, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x i32> poison, i32 %b, i32 0
+  %vb = shufflevector <vscale x 1 x i32> %elt.head, <vscale x 1 x i32> poison, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i32(<vscale x 1 x i32> %va, <vscale x 1 x i32> %vb, metadata !"ult", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_ult_vx_swap_nxv1i32(<vscale x 1 x i32> %va, i32 %b, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ult_vx_swap_nxv1i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a1, e32, mf2, ta, ma
+; CHECK-NEXT:    vmsgtu.vx v0, v8, a0, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x i32> poison, i32 %b, i32 0
+  %vb = shufflevector <vscale x 1 x i32> %elt.head, <vscale x 1 x i32> poison, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i32(<vscale x 1 x i32> %vb, <vscale x 1 x i32> %va, metadata !"ult", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_ult_vi_nxv1i32(<vscale x 1 x i32> %va, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ult_vi_nxv1i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, mf2, ta, ma
+; CHECK-NEXT:    vmsleu.vi v0, v8, 3, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x i32> poison, i32 4, i32 0
+  %vb = shufflevector <vscale x 1 x i32> %elt.head, <vscale x 1 x i32> poison, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i32(<vscale x 1 x i32> %va, <vscale x 1 x i32> %vb, metadata !"ult", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_ult_vi_swap_nxv1i32(<vscale x 1 x i32> %va, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ult_vi_swap_nxv1i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, mf2, ta, ma
+; CHECK-NEXT:    vmsgtu.vi v0, v8, 4, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x i32> poison, i32 4, i32 0
+  %vb = shufflevector <vscale x 1 x i32> %elt.head, <vscale x 1 x i32> poison, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i32(<vscale x 1 x i32> %vb, <vscale x 1 x i32> %va, metadata !"ult", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_sgt_vv_nxv1i32(<vscale x 1 x i32> %va, <vscale x 1 x i32> %vb, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sgt_vv_nxv1i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, mf2, ta, ma
+; CHECK-NEXT:    vmslt.vv v0, v9, v8, v0.t
+; CHECK-NEXT:    ret
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i32(<vscale x 1 x i32> %va, <vscale x 1 x i32> %vb, metadata !"sgt", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_sgt_vx_nxv1i32(<vscale x 1 x i32> %va, i32 %b, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sgt_vx_nxv1i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a1, e32, mf2, ta, ma
+; CHECK-NEXT:    vmsgt.vx v0, v8, a0, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x i32> poison, i32 %b, i32 0
+  %vb = shufflevector <vscale x 1 x i32> %elt.head, <vscale x 1 x i32> poison, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i32(<vscale x 1 x i32> %va, <vscale x 1 x i32> %vb, metadata !"sgt", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_sgt_vx_swap_nxv1i32(<vscale x 1 x i32> %va, i32 %b, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sgt_vx_swap_nxv1i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a1, e32, mf2, ta, ma
+; CHECK-NEXT:    vmslt.vx v0, v8, a0, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x i32> poison, i32 %b, i32 0
+  %vb = shufflevector <vscale x 1 x i32> %elt.head, <vscale x 1 x i32> poison, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i32(<vscale x 1 x i32> %vb, <vscale x 1 x i32> %va, metadata !"sgt", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_sgt_vi_nxv1i32(<vscale x 1 x i32> %va, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sgt_vi_nxv1i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, mf2, ta, ma
+; CHECK-NEXT:    vmsgt.vi v0, v8, 4, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x i32> poison, i32 4, i32 0
+  %vb = shufflevector <vscale x 1 x i32> %elt.head, <vscale x 1 x i32> poison, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i32(<vscale x 1 x i32> %va, <vscale x 1 x i32> %vb, metadata !"sgt", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_sgt_vi_swap_nxv1i32(<vscale x 1 x i32> %va, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sgt_vi_swap_nxv1i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    li a1, 4
+; CHECK-NEXT:    vsetvli zero, a0, e32, mf2, ta, ma
+; CHECK-NEXT:    vmslt.vx v0, v8, a1, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x i32> poison, i32 4, i32 0
+  %vb = shufflevector <vscale x 1 x i32> %elt.head, <vscale x 1 x i32> poison, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i32(<vscale x 1 x i32> %vb, <vscale x 1 x i32> %va, metadata !"sgt", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_sge_vv_nxv1i32(<vscale x 1 x i32> %va, <vscale x 1 x i32> %vb, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sge_vv_nxv1i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, mf2, ta, ma
+; CHECK-NEXT:    vmsle.vv v0, v9, v8, v0.t
+; CHECK-NEXT:    ret
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i32(<vscale x 1 x i32> %va, <vscale x 1 x i32> %vb, metadata !"sge", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_sge_vx_nxv1i32(<vscale x 1 x i32> %va, i32 %b, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sge_vx_nxv1i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a2, zero, e32, mf2, ta, mu
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, a1, e32, mf2, ta, ma
+; CHECK-NEXT:    vmsle.vv v0, v9, v8, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x i32> poison, i32 %b, i32 0
+  %vb = shufflevector <vscale x 1 x i32> %elt.head, <vscale x 1 x i32> poison, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i32(<vscale x 1 x i32> %va, <vscale x 1 x i32> %vb, metadata !"sge", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_sge_vx_swap_nxv1i32(<vscale x 1 x i32> %va, i32 %b, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sge_vx_swap_nxv1i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a1, e32, mf2, ta, ma
+; CHECK-NEXT:    vmsle.vx v0, v8, a0, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x i32> poison, i32 %b, i32 0
+  %vb = shufflevector <vscale x 1 x i32> %elt.head, <vscale x 1 x i32> poison, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i32(<vscale x 1 x i32> %vb, <vscale x 1 x i32> %va, metadata !"sge", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_sge_vi_nxv1i32(<vscale x 1 x i32> %va, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sge_vi_nxv1i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, mf2, ta, ma
+; CHECK-NEXT:    vmsgt.vi v0, v8, 3, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x i32> poison, i32 4, i32 0
+  %vb = shufflevector <vscale x 1 x i32> %elt.head, <vscale x 1 x i32> poison, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i32(<vscale x 1 x i32> %va, <vscale x 1 x i32> %vb, metadata !"sge", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_sge_vi_swap_nxv1i32(<vscale x 1 x i32> %va, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sge_vi_swap_nxv1i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, mf2, ta, ma
+; CHECK-NEXT:    vmsle.vi v0, v8, 4, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x i32> poison, i32 4, i32 0
+  %vb = shufflevector <vscale x 1 x i32> %elt.head, <vscale x 1 x i32> poison, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i32(<vscale x 1 x i32> %vb, <vscale x 1 x i32> %va, metadata !"sge", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_slt_vv_nxv1i32(<vscale x 1 x i32> %va, <vscale x 1 x i32> %vb, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_slt_vv_nxv1i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, mf2, ta, ma
+; CHECK-NEXT:    vmslt.vv v0, v8, v9, v0.t
+; CHECK-NEXT:    ret
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i32(<vscale x 1 x i32> %va, <vscale x 1 x i32> %vb, metadata !"slt", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_slt_vx_nxv1i32(<vscale x 1 x i32> %va, i32 %b, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_slt_vx_nxv1i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a1, e32, mf2, ta, ma
+; CHECK-NEXT:    vmslt.vx v0, v8, a0, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x i32> poison, i32 %b, i32 0
+  %vb = shufflevector <vscale x 1 x i32> %elt.head, <vscale x 1 x i32> poison, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i32(<vscale x 1 x i32> %va, <vscale x 1 x i32> %vb, metadata !"slt", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_slt_vx_swap_nxv1i32(<vscale x 1 x i32> %va, i32 %b, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_slt_vx_swap_nxv1i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a1, e32, mf2, ta, ma
+; CHECK-NEXT:    vmsgt.vx v0, v8, a0, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x i32> poison, i32 %b, i32 0
+  %vb = shufflevector <vscale x 1 x i32> %elt.head, <vscale x 1 x i32> poison, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i32(<vscale x 1 x i32> %vb, <vscale x 1 x i32> %va, metadata !"slt", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_slt_vi_nxv1i32(<vscale x 1 x i32> %va, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_slt_vi_nxv1i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, mf2, ta, ma
+; CHECK-NEXT:    vmsle.vi v0, v8, 3, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x i32> poison, i32 4, i32 0
+  %vb = shufflevector <vscale x 1 x i32> %elt.head, <vscale x 1 x i32> poison, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i32(<vscale x 1 x i32> %va, <vscale x 1 x i32> %vb, metadata !"slt", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_slt_vi_swap_nxv1i32(<vscale x 1 x i32> %va, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_slt_vi_swap_nxv1i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, mf2, ta, ma
+; CHECK-NEXT:    vmsgt.vi v0, v8, 4, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x i32> poison, i32 4, i32 0
+  %vb = shufflevector <vscale x 1 x i32> %elt.head, <vscale x 1 x i32> poison, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i32(<vscale x 1 x i32> %vb, <vscale x 1 x i32> %va, metadata !"slt", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_sle_vv_nxv1i32(<vscale x 1 x i32> %va, <vscale x 1 x i32> %vb, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sle_vv_nxv1i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, mf2, ta, ma
+; CHECK-NEXT:    vmsle.vv v0, v8, v9, v0.t
+; CHECK-NEXT:    ret
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i32(<vscale x 1 x i32> %va, <vscale x 1 x i32> %vb, metadata !"sle", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_sle_vx_nxv1i32(<vscale x 1 x i32> %va, i32 %b, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sle_vx_nxv1i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a1, e32, mf2, ta, ma
+; CHECK-NEXT:    vmsle.vx v0, v8, a0, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x i32> poison, i32 %b, i32 0
+  %vb = shufflevector <vscale x 1 x i32> %elt.head, <vscale x 1 x i32> poison, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i32(<vscale x 1 x i32> %va, <vscale x 1 x i32> %vb, metadata !"sle", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_sle_vx_swap_nxv1i32(<vscale x 1 x i32> %va, i32 %b, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sle_vx_swap_nxv1i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a2, zero, e32, mf2, ta, mu
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, a1, e32, mf2, ta, ma
+; CHECK-NEXT:    vmsle.vv v0, v9, v8, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x i32> poison, i32 %b, i32 0
+  %vb = shufflevector <vscale x 1 x i32> %elt.head, <vscale x 1 x i32> poison, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i32(<vscale x 1 x i32> %vb, <vscale x 1 x i32> %va, metadata !"sle", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_sle_vi_nxv1i32(<vscale x 1 x i32> %va, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sle_vi_nxv1i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, mf2, ta, ma
+; CHECK-NEXT:    vmsle.vi v0, v8, 4, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x i32> poison, i32 4, i32 0
+  %vb = shufflevector <vscale x 1 x i32> %elt.head, <vscale x 1 x i32> poison, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i32(<vscale x 1 x i32> %va, <vscale x 1 x i32> %vb, metadata !"sle", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_sle_vi_swap_nxv1i32(<vscale x 1 x i32> %va, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sle_vi_swap_nxv1i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e32, mf2, ta, mu
+; CHECK-NEXT:    vmv.v.i v9, 4
+; CHECK-NEXT:    vsetvli zero, a0, e32, mf2, ta, ma
+; CHECK-NEXT:    vmsle.vv v0, v9, v8, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x i32> poison, i32 4, i32 0
+  %vb = shufflevector <vscale x 1 x i32> %elt.head, <vscale x 1 x i32> poison, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i32(<vscale x 1 x i32> %vb, <vscale x 1 x i32> %va, metadata !"sle", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+declare <vscale x 8 x i1> @llvm.vp.icmp.nxv8i32(<vscale x 8 x i32>, <vscale x 8 x i32>, metadata, <vscale x 8 x i1>, i32)
+
+define <vscale x 8 x i1> @icmp_eq_vv_nxv8i32(<vscale x 8 x i32> %va, <vscale x 8 x i32> %vb, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_eq_vv_nxv8i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m4, ta, ma
+; CHECK-NEXT:    vmseq.vv v16, v8, v12, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v16
+; CHECK-NEXT:    ret
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i32(<vscale x 8 x i32> %va, <vscale x 8 x i32> %vb, metadata !"eq", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_eq_vx_nxv8i32(<vscale x 8 x i32> %va, i32 %b, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_eq_vx_nxv8i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a1, e32, m4, ta, ma
+; CHECK-NEXT:    vmseq.vx v12, v8, a0, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v12
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x i32> poison, i32 %b, i32 0
+  %vb = shufflevector <vscale x 8 x i32> %elt.head, <vscale x 8 x i32> poison, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i32(<vscale x 8 x i32> %va, <vscale x 8 x i32> %vb, metadata !"eq", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_eq_vx_swap_nxv8i32(<vscale x 8 x i32> %va, i32 %b, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_eq_vx_swap_nxv8i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a1, e32, m4, ta, ma
+; CHECK-NEXT:    vmseq.vx v12, v8, a0, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v12
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x i32> poison, i32 %b, i32 0
+  %vb = shufflevector <vscale x 8 x i32> %elt.head, <vscale x 8 x i32> poison, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i32(<vscale x 8 x i32> %vb, <vscale x 8 x i32> %va, metadata !"eq", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_eq_vi_nxv8i32(<vscale x 8 x i32> %va, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_eq_vi_nxv8i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m4, ta, ma
+; CHECK-NEXT:    vmseq.vi v12, v8, 4, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v12
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x i32> poison, i32 4, i32 0
+  %vb = shufflevector <vscale x 8 x i32> %elt.head, <vscale x 8 x i32> poison, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i32(<vscale x 8 x i32> %va, <vscale x 8 x i32> %vb, metadata !"eq", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_eq_vi_swap_nxv8i32(<vscale x 8 x i32> %va, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_eq_vi_swap_nxv8i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m4, ta, ma
+; CHECK-NEXT:    vmseq.vi v12, v8, 4, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v12
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x i32> poison, i32 4, i32 0
+  %vb = shufflevector <vscale x 8 x i32> %elt.head, <vscale x 8 x i32> poison, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i32(<vscale x 8 x i32> %vb, <vscale x 8 x i32> %va, metadata !"eq", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_ne_vv_nxv8i32(<vscale x 8 x i32> %va, <vscale x 8 x i32> %vb, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ne_vv_nxv8i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m4, ta, ma
+; CHECK-NEXT:    vmsne.vv v16, v8, v12, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v16
+; CHECK-NEXT:    ret
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i32(<vscale x 8 x i32> %va, <vscale x 8 x i32> %vb, metadata !"ne", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_ne_vx_nxv8i32(<vscale x 8 x i32> %va, i32 %b, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ne_vx_nxv8i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a1, e32, m4, ta, ma
+; CHECK-NEXT:    vmsne.vx v12, v8, a0, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v12
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x i32> poison, i32 %b, i32 0
+  %vb = shufflevector <vscale x 8 x i32> %elt.head, <vscale x 8 x i32> poison, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i32(<vscale x 8 x i32> %va, <vscale x 8 x i32> %vb, metadata !"ne", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_ne_vx_swap_nxv8i32(<vscale x 8 x i32> %va, i32 %b, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ne_vx_swap_nxv8i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a1, e32, m4, ta, ma
+; CHECK-NEXT:    vmsne.vx v12, v8, a0, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v12
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x i32> poison, i32 %b, i32 0
+  %vb = shufflevector <vscale x 8 x i32> %elt.head, <vscale x 8 x i32> poison, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i32(<vscale x 8 x i32> %vb, <vscale x 8 x i32> %va, metadata !"ne", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_ne_vi_nxv8i32(<vscale x 8 x i32> %va, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ne_vi_nxv8i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m4, ta, ma
+; CHECK-NEXT:    vmsne.vi v12, v8, 4, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v12
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x i32> poison, i32 4, i32 0
+  %vb = shufflevector <vscale x 8 x i32> %elt.head, <vscale x 8 x i32> poison, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i32(<vscale x 8 x i32> %va, <vscale x 8 x i32> %vb, metadata !"ne", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_ne_vi_swap_nxv8i32(<vscale x 8 x i32> %va, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ne_vi_swap_nxv8i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m4, ta, ma
+; CHECK-NEXT:    vmsne.vi v12, v8, 4, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v12
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x i32> poison, i32 4, i32 0
+  %vb = shufflevector <vscale x 8 x i32> %elt.head, <vscale x 8 x i32> poison, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i32(<vscale x 8 x i32> %vb, <vscale x 8 x i32> %va, metadata !"ne", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_ugt_vv_nxv8i32(<vscale x 8 x i32> %va, <vscale x 8 x i32> %vb, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ugt_vv_nxv8i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m4, ta, ma
+; CHECK-NEXT:    vmsltu.vv v16, v12, v8, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v16
+; CHECK-NEXT:    ret
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i32(<vscale x 8 x i32> %va, <vscale x 8 x i32> %vb, metadata !"ugt", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_ugt_vx_nxv8i32(<vscale x 8 x i32> %va, i32 %b, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ugt_vx_nxv8i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a1, e32, m4, ta, ma
+; CHECK-NEXT:    vmsgtu.vx v12, v8, a0, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v12
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x i32> poison, i32 %b, i32 0
+  %vb = shufflevector <vscale x 8 x i32> %elt.head, <vscale x 8 x i32> poison, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i32(<vscale x 8 x i32> %va, <vscale x 8 x i32> %vb, metadata !"ugt", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_ugt_vx_swap_nxv8i32(<vscale x 8 x i32> %va, i32 %b, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ugt_vx_swap_nxv8i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a1, e32, m4, ta, ma
+; CHECK-NEXT:    vmsltu.vx v12, v8, a0, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v12
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x i32> poison, i32 %b, i32 0
+  %vb = shufflevector <vscale x 8 x i32> %elt.head, <vscale x 8 x i32> poison, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i32(<vscale x 8 x i32> %vb, <vscale x 8 x i32> %va, metadata !"ugt", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_ugt_vi_nxv8i32(<vscale x 8 x i32> %va, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ugt_vi_nxv8i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m4, ta, ma
+; CHECK-NEXT:    vmsgtu.vi v12, v8, 4, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v12
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x i32> poison, i32 4, i32 0
+  %vb = shufflevector <vscale x 8 x i32> %elt.head, <vscale x 8 x i32> poison, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i32(<vscale x 8 x i32> %va, <vscale x 8 x i32> %vb, metadata !"ugt", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_ugt_vi_swap_nxv8i32(<vscale x 8 x i32> %va, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ugt_vi_swap_nxv8i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    li a1, 4
+; CHECK-NEXT:    vsetvli zero, a0, e32, m4, ta, ma
+; CHECK-NEXT:    vmsltu.vx v12, v8, a1, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v12
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x i32> poison, i32 4, i32 0
+  %vb = shufflevector <vscale x 8 x i32> %elt.head, <vscale x 8 x i32> poison, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i32(<vscale x 8 x i32> %vb, <vscale x 8 x i32> %va, metadata !"ugt", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_uge_vv_nxv8i32(<vscale x 8 x i32> %va, <vscale x 8 x i32> %vb, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_uge_vv_nxv8i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m4, ta, ma
+; CHECK-NEXT:    vmsleu.vv v16, v12, v8, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v16
+; CHECK-NEXT:    ret
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i32(<vscale x 8 x i32> %va, <vscale x 8 x i32> %vb, metadata !"uge", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_uge_vx_nxv8i32(<vscale x 8 x i32> %va, i32 %b, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_uge_vx_nxv8i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a2, zero, e32, m4, ta, mu
+; CHECK-NEXT:    vmv.v.x v16, a0
+; CHECK-NEXT:    vsetvli zero, a1, e32, m4, ta, ma
+; CHECK-NEXT:    vmsleu.vv v12, v16, v8, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v12
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x i32> poison, i32 %b, i32 0
+  %vb = shufflevector <vscale x 8 x i32> %elt.head, <vscale x 8 x i32> poison, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i32(<vscale x 8 x i32> %va, <vscale x 8 x i32> %vb, metadata !"uge", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_uge_vx_swap_nxv8i32(<vscale x 8 x i32> %va, i32 %b, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_uge_vx_swap_nxv8i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a1, e32, m4, ta, ma
+; CHECK-NEXT:    vmsleu.vx v12, v8, a0, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v12
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x i32> poison, i32 %b, i32 0
+  %vb = shufflevector <vscale x 8 x i32> %elt.head, <vscale x 8 x i32> poison, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i32(<vscale x 8 x i32> %vb, <vscale x 8 x i32> %va, metadata !"uge", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_uge_vi_nxv8i32(<vscale x 8 x i32> %va, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_uge_vi_nxv8i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m4, ta, ma
+; CHECK-NEXT:    vmsgtu.vi v12, v8, 3, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v12
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x i32> poison, i32 4, i32 0
+  %vb = shufflevector <vscale x 8 x i32> %elt.head, <vscale x 8 x i32> poison, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i32(<vscale x 8 x i32> %va, <vscale x 8 x i32> %vb, metadata !"uge", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_uge_vi_swap_nxv8i32(<vscale x 8 x i32> %va, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_uge_vi_swap_nxv8i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m4, ta, ma
+; CHECK-NEXT:    vmsleu.vi v12, v8, 4, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v12
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x i32> poison, i32 4, i32 0
+  %vb = shufflevector <vscale x 8 x i32> %elt.head, <vscale x 8 x i32> poison, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i32(<vscale x 8 x i32> %vb, <vscale x 8 x i32> %va, metadata !"uge", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_ult_vv_nxv8i32(<vscale x 8 x i32> %va, <vscale x 8 x i32> %vb, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ult_vv_nxv8i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m4, ta, ma
+; CHECK-NEXT:    vmsltu.vv v16, v8, v12, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v16
+; CHECK-NEXT:    ret
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i32(<vscale x 8 x i32> %va, <vscale x 8 x i32> %vb, metadata !"ult", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_ult_vx_nxv8i32(<vscale x 8 x i32> %va, i32 %b, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ult_vx_nxv8i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a1, e32, m4, ta, ma
+; CHECK-NEXT:    vmsltu.vx v12, v8, a0, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v12
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x i32> poison, i32 %b, i32 0
+  %vb = shufflevector <vscale x 8 x i32> %elt.head, <vscale x 8 x i32> poison, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i32(<vscale x 8 x i32> %va, <vscale x 8 x i32> %vb, metadata !"ult", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_ult_vx_swap_nxv8i32(<vscale x 8 x i32> %va, i32 %b, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ult_vx_swap_nxv8i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a1, e32, m4, ta, ma
+; CHECK-NEXT:    vmsgtu.vx v12, v8, a0, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v12
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x i32> poison, i32 %b, i32 0
+  %vb = shufflevector <vscale x 8 x i32> %elt.head, <vscale x 8 x i32> poison, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i32(<vscale x 8 x i32> %vb, <vscale x 8 x i32> %va, metadata !"ult", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_ult_vi_nxv8i32(<vscale x 8 x i32> %va, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ult_vi_nxv8i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m4, ta, ma
+; CHECK-NEXT:    vmsleu.vi v12, v8, 3, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v12
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x i32> poison, i32 4, i32 0
+  %vb = shufflevector <vscale x 8 x i32> %elt.head, <vscale x 8 x i32> poison, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i32(<vscale x 8 x i32> %va, <vscale x 8 x i32> %vb, metadata !"ult", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_ult_vi_swap_nxv8i32(<vscale x 8 x i32> %va, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ult_vi_swap_nxv8i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m4, ta, ma
+; CHECK-NEXT:    vmsgtu.vi v12, v8, 4, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v12
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x i32> poison, i32 4, i32 0
+  %vb = shufflevector <vscale x 8 x i32> %elt.head, <vscale x 8 x i32> poison, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i32(<vscale x 8 x i32> %vb, <vscale x 8 x i32> %va, metadata !"ult", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_sgt_vv_nxv8i32(<vscale x 8 x i32> %va, <vscale x 8 x i32> %vb, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sgt_vv_nxv8i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m4, ta, ma
+; CHECK-NEXT:    vmslt.vv v16, v12, v8, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v16
+; CHECK-NEXT:    ret
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i32(<vscale x 8 x i32> %va, <vscale x 8 x i32> %vb, metadata !"sgt", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_sgt_vx_nxv8i32(<vscale x 8 x i32> %va, i32 %b, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sgt_vx_nxv8i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a1, e32, m4, ta, ma
+; CHECK-NEXT:    vmsgt.vx v12, v8, a0, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v12
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x i32> poison, i32 %b, i32 0
+  %vb = shufflevector <vscale x 8 x i32> %elt.head, <vscale x 8 x i32> poison, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i32(<vscale x 8 x i32> %va, <vscale x 8 x i32> %vb, metadata !"sgt", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_sgt_vx_swap_nxv8i32(<vscale x 8 x i32> %va, i32 %b, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sgt_vx_swap_nxv8i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a1, e32, m4, ta, ma
+; CHECK-NEXT:    vmslt.vx v12, v8, a0, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v12
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x i32> poison, i32 %b, i32 0
+  %vb = shufflevector <vscale x 8 x i32> %elt.head, <vscale x 8 x i32> poison, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i32(<vscale x 8 x i32> %vb, <vscale x 8 x i32> %va, metadata !"sgt", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_sgt_vi_nxv8i32(<vscale x 8 x i32> %va, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sgt_vi_nxv8i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m4, ta, ma
+; CHECK-NEXT:    vmsgt.vi v12, v8, 4, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v12
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x i32> poison, i32 4, i32 0
+  %vb = shufflevector <vscale x 8 x i32> %elt.head, <vscale x 8 x i32> poison, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i32(<vscale x 8 x i32> %va, <vscale x 8 x i32> %vb, metadata !"sgt", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_sgt_vi_swap_nxv8i32(<vscale x 8 x i32> %va, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sgt_vi_swap_nxv8i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    li a1, 4
+; CHECK-NEXT:    vsetvli zero, a0, e32, m4, ta, ma
+; CHECK-NEXT:    vmslt.vx v12, v8, a1, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v12
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x i32> poison, i32 4, i32 0
+  %vb = shufflevector <vscale x 8 x i32> %elt.head, <vscale x 8 x i32> poison, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i32(<vscale x 8 x i32> %vb, <vscale x 8 x i32> %va, metadata !"sgt", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_sge_vv_nxv8i32(<vscale x 8 x i32> %va, <vscale x 8 x i32> %vb, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sge_vv_nxv8i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m4, ta, ma
+; CHECK-NEXT:    vmsle.vv v16, v12, v8, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v16
+; CHECK-NEXT:    ret
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i32(<vscale x 8 x i32> %va, <vscale x 8 x i32> %vb, metadata !"sge", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_sge_vx_nxv8i32(<vscale x 8 x i32> %va, i32 %b, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sge_vx_nxv8i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a2, zero, e32, m4, ta, mu
+; CHECK-NEXT:    vmv.v.x v16, a0
+; CHECK-NEXT:    vsetvli zero, a1, e32, m4, ta, ma
+; CHECK-NEXT:    vmsle.vv v12, v16, v8, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v12
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x i32> poison, i32 %b, i32 0
+  %vb = shufflevector <vscale x 8 x i32> %elt.head, <vscale x 8 x i32> poison, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i32(<vscale x 8 x i32> %va, <vscale x 8 x i32> %vb, metadata !"sge", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_sge_vx_swap_nxv8i32(<vscale x 8 x i32> %va, i32 %b, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sge_vx_swap_nxv8i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a1, e32, m4, ta, ma
+; CHECK-NEXT:    vmsle.vx v12, v8, a0, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v12
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x i32> poison, i32 %b, i32 0
+  %vb = shufflevector <vscale x 8 x i32> %elt.head, <vscale x 8 x i32> poison, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i32(<vscale x 8 x i32> %vb, <vscale x 8 x i32> %va, metadata !"sge", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_sge_vi_nxv8i32(<vscale x 8 x i32> %va, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sge_vi_nxv8i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m4, ta, ma
+; CHECK-NEXT:    vmsgt.vi v12, v8, 3, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v12
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x i32> poison, i32 4, i32 0
+  %vb = shufflevector <vscale x 8 x i32> %elt.head, <vscale x 8 x i32> poison, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i32(<vscale x 8 x i32> %va, <vscale x 8 x i32> %vb, metadata !"sge", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_sge_vi_swap_nxv8i32(<vscale x 8 x i32> %va, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sge_vi_swap_nxv8i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m4, ta, ma
+; CHECK-NEXT:    vmsle.vi v12, v8, 4, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v12
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x i32> poison, i32 4, i32 0
+  %vb = shufflevector <vscale x 8 x i32> %elt.head, <vscale x 8 x i32> poison, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i32(<vscale x 8 x i32> %vb, <vscale x 8 x i32> %va, metadata !"sge", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_slt_vv_nxv8i32(<vscale x 8 x i32> %va, <vscale x 8 x i32> %vb, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_slt_vv_nxv8i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m4, ta, ma
+; CHECK-NEXT:    vmslt.vv v16, v8, v12, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v16
+; CHECK-NEXT:    ret
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i32(<vscale x 8 x i32> %va, <vscale x 8 x i32> %vb, metadata !"slt", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_slt_vx_nxv8i32(<vscale x 8 x i32> %va, i32 %b, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_slt_vx_nxv8i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a1, e32, m4, ta, ma
+; CHECK-NEXT:    vmslt.vx v12, v8, a0, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v12
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x i32> poison, i32 %b, i32 0
+  %vb = shufflevector <vscale x 8 x i32> %elt.head, <vscale x 8 x i32> poison, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i32(<vscale x 8 x i32> %va, <vscale x 8 x i32> %vb, metadata !"slt", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_slt_vx_swap_nxv8i32(<vscale x 8 x i32> %va, i32 %b, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_slt_vx_swap_nxv8i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a1, e32, m4, ta, ma
+; CHECK-NEXT:    vmsgt.vx v12, v8, a0, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v12
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x i32> poison, i32 %b, i32 0
+  %vb = shufflevector <vscale x 8 x i32> %elt.head, <vscale x 8 x i32> poison, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i32(<vscale x 8 x i32> %vb, <vscale x 8 x i32> %va, metadata !"slt", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_slt_vi_nxv8i32(<vscale x 8 x i32> %va, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_slt_vi_nxv8i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m4, ta, ma
+; CHECK-NEXT:    vmsle.vi v12, v8, 3, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v12
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x i32> poison, i32 4, i32 0
+  %vb = shufflevector <vscale x 8 x i32> %elt.head, <vscale x 8 x i32> poison, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i32(<vscale x 8 x i32> %va, <vscale x 8 x i32> %vb, metadata !"slt", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_slt_vi_swap_nxv8i32(<vscale x 8 x i32> %va, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_slt_vi_swap_nxv8i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m4, ta, ma
+; CHECK-NEXT:    vmsgt.vi v12, v8, 4, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v12
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x i32> poison, i32 4, i32 0
+  %vb = shufflevector <vscale x 8 x i32> %elt.head, <vscale x 8 x i32> poison, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i32(<vscale x 8 x i32> %vb, <vscale x 8 x i32> %va, metadata !"slt", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_sle_vv_nxv8i32(<vscale x 8 x i32> %va, <vscale x 8 x i32> %vb, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sle_vv_nxv8i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m4, ta, ma
+; CHECK-NEXT:    vmsle.vv v16, v8, v12, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v16
+; CHECK-NEXT:    ret
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i32(<vscale x 8 x i32> %va, <vscale x 8 x i32> %vb, metadata !"sle", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_sle_vx_nxv8i32(<vscale x 8 x i32> %va, i32 %b, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sle_vx_nxv8i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a1, e32, m4, ta, ma
+; CHECK-NEXT:    vmsle.vx v12, v8, a0, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v12
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x i32> poison, i32 %b, i32 0
+  %vb = shufflevector <vscale x 8 x i32> %elt.head, <vscale x 8 x i32> poison, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i32(<vscale x 8 x i32> %va, <vscale x 8 x i32> %vb, metadata !"sle", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_sle_vx_swap_nxv8i32(<vscale x 8 x i32> %va, i32 %b, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sle_vx_swap_nxv8i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a2, zero, e32, m4, ta, mu
+; CHECK-NEXT:    vmv.v.x v16, a0
+; CHECK-NEXT:    vsetvli zero, a1, e32, m4, ta, ma
+; CHECK-NEXT:    vmsle.vv v12, v16, v8, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v12
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x i32> poison, i32 %b, i32 0
+  %vb = shufflevector <vscale x 8 x i32> %elt.head, <vscale x 8 x i32> poison, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i32(<vscale x 8 x i32> %vb, <vscale x 8 x i32> %va, metadata !"sle", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_sle_vi_nxv8i32(<vscale x 8 x i32> %va, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sle_vi_nxv8i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m4, ta, ma
+; CHECK-NEXT:    vmsle.vi v12, v8, 4, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v12
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x i32> poison, i32 4, i32 0
+  %vb = shufflevector <vscale x 8 x i32> %elt.head, <vscale x 8 x i32> poison, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i32(<vscale x 8 x i32> %va, <vscale x 8 x i32> %vb, metadata !"sle", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_sle_vi_swap_nxv8i32(<vscale x 8 x i32> %va, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sle_vi_swap_nxv8i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e32, m4, ta, mu
+; CHECK-NEXT:    vmv.v.i v16, 4
+; CHECK-NEXT:    vsetvli zero, a0, e32, m4, ta, ma
+; CHECK-NEXT:    vmsle.vv v12, v16, v8, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v12
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x i32> poison, i32 4, i32 0
+  %vb = shufflevector <vscale x 8 x i32> %elt.head, <vscale x 8 x i32> poison, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i32(<vscale x 8 x i32> %vb, <vscale x 8 x i32> %va, metadata !"sle", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+declare <vscale x 1 x i1> @llvm.vp.icmp.nxv1i64(<vscale x 1 x i64>, <vscale x 1 x i64>, metadata, <vscale x 1 x i1>, i32)
+
+define <vscale x 1 x i1> @icmp_eq_vv_nxv1i64(<vscale x 1 x i64> %va, <vscale x 1 x i64> %vb, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_eq_vv_nxv1i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m1, ta, ma
+; CHECK-NEXT:    vmseq.vv v0, v8, v9, v0.t
+; CHECK-NEXT:    ret
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i64(<vscale x 1 x i64> %va, <vscale x 1 x i64> %vb, metadata !"eq", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_eq_vx_nxv1i64(<vscale x 1 x i64> %va, i64 %b, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; RV32-LABEL: icmp_eq_vx_nxv1i64:
+; RV32:       # %bb.0:
+; RV32-NEXT:    addi sp, sp, -16
+; RV32-NEXT:    .cfi_def_cfa_offset 16
+; RV32-NEXT:    sw a1, 12(sp)
+; RV32-NEXT:    sw a0, 8(sp)
+; RV32-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; RV32-NEXT:    addi a0, sp, 8
+; RV32-NEXT:    vlse64.v v9, (a0), zero
+; RV32-NEXT:    vsetvli zero, a2, e64, m1, ta, ma
+; RV32-NEXT:    vmseq.vv v0, v8, v9, v0.t
+; RV32-NEXT:    addi sp, sp, 16
+; RV32-NEXT:    ret
+;
+; RV64-LABEL: icmp_eq_vx_nxv1i64:
+; RV64:       # %bb.0:
+; RV64-NEXT:    vsetvli zero, a1, e64, m1, ta, ma
+; RV64-NEXT:    vmseq.vx v0, v8, a0, v0.t
+; RV64-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x i64> poison, i64 %b, i32 0
+  %vb = shufflevector <vscale x 1 x i64> %elt.head, <vscale x 1 x i64> poison, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i64(<vscale x 1 x i64> %va, <vscale x 1 x i64> %vb, metadata !"eq", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_eq_vx_swap_nxv1i64(<vscale x 1 x i64> %va, i64 %b, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; RV32-LABEL: icmp_eq_vx_swap_nxv1i64:
+; RV32:       # %bb.0:
+; RV32-NEXT:    addi sp, sp, -16
+; RV32-NEXT:    .cfi_def_cfa_offset 16
+; RV32-NEXT:    sw a1, 12(sp)
+; RV32-NEXT:    sw a0, 8(sp)
+; RV32-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; RV32-NEXT:    addi a0, sp, 8
+; RV32-NEXT:    vlse64.v v9, (a0), zero
+; RV32-NEXT:    vsetvli zero, a2, e64, m1, ta, ma
+; RV32-NEXT:    vmseq.vv v0, v9, v8, v0.t
+; RV32-NEXT:    addi sp, sp, 16
+; RV32-NEXT:    ret
+;
+; RV64-LABEL: icmp_eq_vx_swap_nxv1i64:
+; RV64:       # %bb.0:
+; RV64-NEXT:    vsetvli zero, a1, e64, m1, ta, ma
+; RV64-NEXT:    vmseq.vx v0, v8, a0, v0.t
+; RV64-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x i64> poison, i64 %b, i32 0
+  %vb = shufflevector <vscale x 1 x i64> %elt.head, <vscale x 1 x i64> poison, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i64(<vscale x 1 x i64> %vb, <vscale x 1 x i64> %va, metadata !"eq", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_eq_vi_nxv1i64(<vscale x 1 x i64> %va, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_eq_vi_nxv1i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m1, ta, ma
+; CHECK-NEXT:    vmseq.vi v0, v8, 4, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x i64> poison, i64 4, i32 0
+  %vb = shufflevector <vscale x 1 x i64> %elt.head, <vscale x 1 x i64> poison, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i64(<vscale x 1 x i64> %va, <vscale x 1 x i64> %vb, metadata !"eq", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_eq_vi_swap_nxv1i64(<vscale x 1 x i64> %va, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_eq_vi_swap_nxv1i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m1, ta, ma
+; CHECK-NEXT:    vmseq.vi v0, v8, 4, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x i64> poison, i64 4, i32 0
+  %vb = shufflevector <vscale x 1 x i64> %elt.head, <vscale x 1 x i64> poison, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i64(<vscale x 1 x i64> %vb, <vscale x 1 x i64> %va, metadata !"eq", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_ne_vv_nxv1i64(<vscale x 1 x i64> %va, <vscale x 1 x i64> %vb, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ne_vv_nxv1i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m1, ta, ma
+; CHECK-NEXT:    vmsne.vv v0, v8, v9, v0.t
+; CHECK-NEXT:    ret
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i64(<vscale x 1 x i64> %va, <vscale x 1 x i64> %vb, metadata !"ne", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_ne_vx_nxv1i64(<vscale x 1 x i64> %va, i64 %b, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; RV32-LABEL: icmp_ne_vx_nxv1i64:
+; RV32:       # %bb.0:
+; RV32-NEXT:    addi sp, sp, -16
+; RV32-NEXT:    .cfi_def_cfa_offset 16
+; RV32-NEXT:    sw a1, 12(sp)
+; RV32-NEXT:    sw a0, 8(sp)
+; RV32-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; RV32-NEXT:    addi a0, sp, 8
+; RV32-NEXT:    vlse64.v v9, (a0), zero
+; RV32-NEXT:    vsetvli zero, a2, e64, m1, ta, ma
+; RV32-NEXT:    vmsne.vv v0, v8, v9, v0.t
+; RV32-NEXT:    addi sp, sp, 16
+; RV32-NEXT:    ret
+;
+; RV64-LABEL: icmp_ne_vx_nxv1i64:
+; RV64:       # %bb.0:
+; RV64-NEXT:    vsetvli zero, a1, e64, m1, ta, ma
+; RV64-NEXT:    vmsne.vx v0, v8, a0, v0.t
+; RV64-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x i64> poison, i64 %b, i32 0
+  %vb = shufflevector <vscale x 1 x i64> %elt.head, <vscale x 1 x i64> poison, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i64(<vscale x 1 x i64> %va, <vscale x 1 x i64> %vb, metadata !"ne", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_ne_vx_swap_nxv1i64(<vscale x 1 x i64> %va, i64 %b, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; RV32-LABEL: icmp_ne_vx_swap_nxv1i64:
+; RV32:       # %bb.0:
+; RV32-NEXT:    addi sp, sp, -16
+; RV32-NEXT:    .cfi_def_cfa_offset 16
+; RV32-NEXT:    sw a1, 12(sp)
+; RV32-NEXT:    sw a0, 8(sp)
+; RV32-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; RV32-NEXT:    addi a0, sp, 8
+; RV32-NEXT:    vlse64.v v9, (a0), zero
+; RV32-NEXT:    vsetvli zero, a2, e64, m1, ta, ma
+; RV32-NEXT:    vmsne.vv v0, v9, v8, v0.t
+; RV32-NEXT:    addi sp, sp, 16
+; RV32-NEXT:    ret
+;
+; RV64-LABEL: icmp_ne_vx_swap_nxv1i64:
+; RV64:       # %bb.0:
+; RV64-NEXT:    vsetvli zero, a1, e64, m1, ta, ma
+; RV64-NEXT:    vmsne.vx v0, v8, a0, v0.t
+; RV64-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x i64> poison, i64 %b, i32 0
+  %vb = shufflevector <vscale x 1 x i64> %elt.head, <vscale x 1 x i64> poison, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i64(<vscale x 1 x i64> %vb, <vscale x 1 x i64> %va, metadata !"ne", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_ne_vi_nxv1i64(<vscale x 1 x i64> %va, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ne_vi_nxv1i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m1, ta, ma
+; CHECK-NEXT:    vmsne.vi v0, v8, 4, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x i64> poison, i64 4, i32 0
+  %vb = shufflevector <vscale x 1 x i64> %elt.head, <vscale x 1 x i64> poison, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i64(<vscale x 1 x i64> %va, <vscale x 1 x i64> %vb, metadata !"ne", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_ne_vi_swap_nxv1i64(<vscale x 1 x i64> %va, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ne_vi_swap_nxv1i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m1, ta, ma
+; CHECK-NEXT:    vmsne.vi v0, v8, 4, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x i64> poison, i64 4, i32 0
+  %vb = shufflevector <vscale x 1 x i64> %elt.head, <vscale x 1 x i64> poison, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i64(<vscale x 1 x i64> %vb, <vscale x 1 x i64> %va, metadata !"ne", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_ugt_vv_nxv1i64(<vscale x 1 x i64> %va, <vscale x 1 x i64> %vb, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ugt_vv_nxv1i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m1, ta, ma
+; CHECK-NEXT:    vmsltu.vv v0, v9, v8, v0.t
+; CHECK-NEXT:    ret
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i64(<vscale x 1 x i64> %va, <vscale x 1 x i64> %vb, metadata !"ugt", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_ugt_vx_nxv1i64(<vscale x 1 x i64> %va, i64 %b, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; RV32-LABEL: icmp_ugt_vx_nxv1i64:
+; RV32:       # %bb.0:
+; RV32-NEXT:    addi sp, sp, -16
+; RV32-NEXT:    .cfi_def_cfa_offset 16
+; RV32-NEXT:    sw a1, 12(sp)
+; RV32-NEXT:    sw a0, 8(sp)
+; RV32-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; RV32-NEXT:    addi a0, sp, 8
+; RV32-NEXT:    vlse64.v v9, (a0), zero
+; RV32-NEXT:    vsetvli zero, a2, e64, m1, ta, ma
+; RV32-NEXT:    vmsltu.vv v0, v9, v8, v0.t
+; RV32-NEXT:    addi sp, sp, 16
+; RV32-NEXT:    ret
+;
+; RV64-LABEL: icmp_ugt_vx_nxv1i64:
+; RV64:       # %bb.0:
+; RV64-NEXT:    vsetvli zero, a1, e64, m1, ta, ma
+; RV64-NEXT:    vmsgtu.vx v0, v8, a0, v0.t
+; RV64-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x i64> poison, i64 %b, i32 0
+  %vb = shufflevector <vscale x 1 x i64> %elt.head, <vscale x 1 x i64> poison, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i64(<vscale x 1 x i64> %va, <vscale x 1 x i64> %vb, metadata !"ugt", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_ugt_vx_swap_nxv1i64(<vscale x 1 x i64> %va, i64 %b, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; RV32-LABEL: icmp_ugt_vx_swap_nxv1i64:
+; RV32:       # %bb.0:
+; RV32-NEXT:    addi sp, sp, -16
+; RV32-NEXT:    .cfi_def_cfa_offset 16
+; RV32-NEXT:    sw a1, 12(sp)
+; RV32-NEXT:    sw a0, 8(sp)
+; RV32-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; RV32-NEXT:    addi a0, sp, 8
+; RV32-NEXT:    vlse64.v v9, (a0), zero
+; RV32-NEXT:    vsetvli zero, a2, e64, m1, ta, ma
+; RV32-NEXT:    vmsltu.vv v0, v8, v9, v0.t
+; RV32-NEXT:    addi sp, sp, 16
+; RV32-NEXT:    ret
+;
+; RV64-LABEL: icmp_ugt_vx_swap_nxv1i64:
+; RV64:       # %bb.0:
+; RV64-NEXT:    vsetvli zero, a1, e64, m1, ta, ma
+; RV64-NEXT:    vmsltu.vx v0, v8, a0, v0.t
+; RV64-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x i64> poison, i64 %b, i32 0
+  %vb = shufflevector <vscale x 1 x i64> %elt.head, <vscale x 1 x i64> poison, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i64(<vscale x 1 x i64> %vb, <vscale x 1 x i64> %va, metadata !"ugt", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_ugt_vi_nxv1i64(<vscale x 1 x i64> %va, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ugt_vi_nxv1i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m1, ta, ma
+; CHECK-NEXT:    vmsgtu.vi v0, v8, 4, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x i64> poison, i64 4, i32 0
+  %vb = shufflevector <vscale x 1 x i64> %elt.head, <vscale x 1 x i64> poison, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i64(<vscale x 1 x i64> %va, <vscale x 1 x i64> %vb, metadata !"ugt", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_ugt_vi_swap_nxv1i64(<vscale x 1 x i64> %va, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ugt_vi_swap_nxv1i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    li a1, 4
+; CHECK-NEXT:    vsetvli zero, a0, e64, m1, ta, ma
+; CHECK-NEXT:    vmsltu.vx v0, v8, a1, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x i64> poison, i64 4, i32 0
+  %vb = shufflevector <vscale x 1 x i64> %elt.head, <vscale x 1 x i64> poison, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i64(<vscale x 1 x i64> %vb, <vscale x 1 x i64> %va, metadata !"ugt", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_uge_vv_nxv1i64(<vscale x 1 x i64> %va, <vscale x 1 x i64> %vb, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_uge_vv_nxv1i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m1, ta, ma
+; CHECK-NEXT:    vmsleu.vv v0, v9, v8, v0.t
+; CHECK-NEXT:    ret
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i64(<vscale x 1 x i64> %va, <vscale x 1 x i64> %vb, metadata !"uge", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_uge_vx_nxv1i64(<vscale x 1 x i64> %va, i64 %b, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; RV32-LABEL: icmp_uge_vx_nxv1i64:
+; RV32:       # %bb.0:
+; RV32-NEXT:    addi sp, sp, -16
+; RV32-NEXT:    .cfi_def_cfa_offset 16
+; RV32-NEXT:    sw a1, 12(sp)
+; RV32-NEXT:    sw a0, 8(sp)
+; RV32-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; RV32-NEXT:    addi a0, sp, 8
+; RV32-NEXT:    vlse64.v v9, (a0), zero
+; RV32-NEXT:    vsetvli zero, a2, e64, m1, ta, ma
+; RV32-NEXT:    vmsleu.vv v0, v9, v8, v0.t
+; RV32-NEXT:    addi sp, sp, 16
+; RV32-NEXT:    ret
+;
+; RV64-LABEL: icmp_uge_vx_nxv1i64:
+; RV64:       # %bb.0:
+; RV64-NEXT:    vsetvli a2, zero, e64, m1, ta, mu
+; RV64-NEXT:    vmv.v.x v9, a0
+; RV64-NEXT:    vsetvli zero, a1, e64, m1, ta, ma
+; RV64-NEXT:    vmsleu.vv v0, v9, v8, v0.t
+; RV64-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x i64> poison, i64 %b, i32 0
+  %vb = shufflevector <vscale x 1 x i64> %elt.head, <vscale x 1 x i64> poison, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i64(<vscale x 1 x i64> %va, <vscale x 1 x i64> %vb, metadata !"uge", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_uge_vx_swap_nxv1i64(<vscale x 1 x i64> %va, i64 %b, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; RV32-LABEL: icmp_uge_vx_swap_nxv1i64:
+; RV32:       # %bb.0:
+; RV32-NEXT:    addi sp, sp, -16
+; RV32-NEXT:    .cfi_def_cfa_offset 16
+; RV32-NEXT:    sw a1, 12(sp)
+; RV32-NEXT:    sw a0, 8(sp)
+; RV32-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; RV32-NEXT:    addi a0, sp, 8
+; RV32-NEXT:    vlse64.v v9, (a0), zero
+; RV32-NEXT:    vsetvli zero, a2, e64, m1, ta, ma
+; RV32-NEXT:    vmsleu.vv v0, v8, v9, v0.t
+; RV32-NEXT:    addi sp, sp, 16
+; RV32-NEXT:    ret
+;
+; RV64-LABEL: icmp_uge_vx_swap_nxv1i64:
+; RV64:       # %bb.0:
+; RV64-NEXT:    vsetvli zero, a1, e64, m1, ta, ma
+; RV64-NEXT:    vmsleu.vx v0, v8, a0, v0.t
+; RV64-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x i64> poison, i64 %b, i32 0
+  %vb = shufflevector <vscale x 1 x i64> %elt.head, <vscale x 1 x i64> poison, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i64(<vscale x 1 x i64> %vb, <vscale x 1 x i64> %va, metadata !"uge", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_uge_vi_nxv1i64(<vscale x 1 x i64> %va, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_uge_vi_nxv1i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m1, ta, ma
+; CHECK-NEXT:    vmsgtu.vi v0, v8, 3, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x i64> poison, i64 4, i32 0
+  %vb = shufflevector <vscale x 1 x i64> %elt.head, <vscale x 1 x i64> poison, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i64(<vscale x 1 x i64> %va, <vscale x 1 x i64> %vb, metadata !"uge", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_uge_vi_swap_nxv1i64(<vscale x 1 x i64> %va, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_uge_vi_swap_nxv1i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m1, ta, ma
+; CHECK-NEXT:    vmsleu.vi v0, v8, 4, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x i64> poison, i64 4, i32 0
+  %vb = shufflevector <vscale x 1 x i64> %elt.head, <vscale x 1 x i64> poison, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i64(<vscale x 1 x i64> %vb, <vscale x 1 x i64> %va, metadata !"uge", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_ult_vv_nxv1i64(<vscale x 1 x i64> %va, <vscale x 1 x i64> %vb, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ult_vv_nxv1i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m1, ta, ma
+; CHECK-NEXT:    vmsltu.vv v0, v8, v9, v0.t
+; CHECK-NEXT:    ret
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i64(<vscale x 1 x i64> %va, <vscale x 1 x i64> %vb, metadata !"ult", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_ult_vx_nxv1i64(<vscale x 1 x i64> %va, i64 %b, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; RV32-LABEL: icmp_ult_vx_nxv1i64:
+; RV32:       # %bb.0:
+; RV32-NEXT:    addi sp, sp, -16
+; RV32-NEXT:    .cfi_def_cfa_offset 16
+; RV32-NEXT:    sw a1, 12(sp)
+; RV32-NEXT:    sw a0, 8(sp)
+; RV32-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; RV32-NEXT:    addi a0, sp, 8
+; RV32-NEXT:    vlse64.v v9, (a0), zero
+; RV32-NEXT:    vsetvli zero, a2, e64, m1, ta, ma
+; RV32-NEXT:    vmsltu.vv v0, v8, v9, v0.t
+; RV32-NEXT:    addi sp, sp, 16
+; RV32-NEXT:    ret
+;
+; RV64-LABEL: icmp_ult_vx_nxv1i64:
+; RV64:       # %bb.0:
+; RV64-NEXT:    vsetvli zero, a1, e64, m1, ta, ma
+; RV64-NEXT:    vmsltu.vx v0, v8, a0, v0.t
+; RV64-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x i64> poison, i64 %b, i32 0
+  %vb = shufflevector <vscale x 1 x i64> %elt.head, <vscale x 1 x i64> poison, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i64(<vscale x 1 x i64> %va, <vscale x 1 x i64> %vb, metadata !"ult", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_ult_vx_swap_nxv1i64(<vscale x 1 x i64> %va, i64 %b, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; RV32-LABEL: icmp_ult_vx_swap_nxv1i64:
+; RV32:       # %bb.0:
+; RV32-NEXT:    addi sp, sp, -16
+; RV32-NEXT:    .cfi_def_cfa_offset 16
+; RV32-NEXT:    sw a1, 12(sp)
+; RV32-NEXT:    sw a0, 8(sp)
+; RV32-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; RV32-NEXT:    addi a0, sp, 8
+; RV32-NEXT:    vlse64.v v9, (a0), zero
+; RV32-NEXT:    vsetvli zero, a2, e64, m1, ta, ma
+; RV32-NEXT:    vmsltu.vv v0, v9, v8, v0.t
+; RV32-NEXT:    addi sp, sp, 16
+; RV32-NEXT:    ret
+;
+; RV64-LABEL: icmp_ult_vx_swap_nxv1i64:
+; RV64:       # %bb.0:
+; RV64-NEXT:    vsetvli zero, a1, e64, m1, ta, ma
+; RV64-NEXT:    vmsgtu.vx v0, v8, a0, v0.t
+; RV64-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x i64> poison, i64 %b, i32 0
+  %vb = shufflevector <vscale x 1 x i64> %elt.head, <vscale x 1 x i64> poison, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i64(<vscale x 1 x i64> %vb, <vscale x 1 x i64> %va, metadata !"ult", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_ult_vi_nxv1i64(<vscale x 1 x i64> %va, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ult_vi_nxv1i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m1, ta, ma
+; CHECK-NEXT:    vmsleu.vi v0, v8, 3, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x i64> poison, i64 4, i32 0
+  %vb = shufflevector <vscale x 1 x i64> %elt.head, <vscale x 1 x i64> poison, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i64(<vscale x 1 x i64> %va, <vscale x 1 x i64> %vb, metadata !"ult", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_ult_vi_swap_nxv1i64(<vscale x 1 x i64> %va, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ult_vi_swap_nxv1i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m1, ta, ma
+; CHECK-NEXT:    vmsgtu.vi v0, v8, 4, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x i64> poison, i64 4, i32 0
+  %vb = shufflevector <vscale x 1 x i64> %elt.head, <vscale x 1 x i64> poison, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i64(<vscale x 1 x i64> %vb, <vscale x 1 x i64> %va, metadata !"ult", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_sgt_vv_nxv1i64(<vscale x 1 x i64> %va, <vscale x 1 x i64> %vb, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sgt_vv_nxv1i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m1, ta, ma
+; CHECK-NEXT:    vmslt.vv v0, v9, v8, v0.t
+; CHECK-NEXT:    ret
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i64(<vscale x 1 x i64> %va, <vscale x 1 x i64> %vb, metadata !"sgt", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_sgt_vx_nxv1i64(<vscale x 1 x i64> %va, i64 %b, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; RV32-LABEL: icmp_sgt_vx_nxv1i64:
+; RV32:       # %bb.0:
+; RV32-NEXT:    addi sp, sp, -16
+; RV32-NEXT:    .cfi_def_cfa_offset 16
+; RV32-NEXT:    sw a1, 12(sp)
+; RV32-NEXT:    sw a0, 8(sp)
+; RV32-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; RV32-NEXT:    addi a0, sp, 8
+; RV32-NEXT:    vlse64.v v9, (a0), zero
+; RV32-NEXT:    vsetvli zero, a2, e64, m1, ta, ma
+; RV32-NEXT:    vmslt.vv v0, v9, v8, v0.t
+; RV32-NEXT:    addi sp, sp, 16
+; RV32-NEXT:    ret
+;
+; RV64-LABEL: icmp_sgt_vx_nxv1i64:
+; RV64:       # %bb.0:
+; RV64-NEXT:    vsetvli zero, a1, e64, m1, ta, ma
+; RV64-NEXT:    vmsgt.vx v0, v8, a0, v0.t
+; RV64-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x i64> poison, i64 %b, i32 0
+  %vb = shufflevector <vscale x 1 x i64> %elt.head, <vscale x 1 x i64> poison, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i64(<vscale x 1 x i64> %va, <vscale x 1 x i64> %vb, metadata !"sgt", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_sgt_vx_swap_nxv1i64(<vscale x 1 x i64> %va, i64 %b, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; RV32-LABEL: icmp_sgt_vx_swap_nxv1i64:
+; RV32:       # %bb.0:
+; RV32-NEXT:    addi sp, sp, -16
+; RV32-NEXT:    .cfi_def_cfa_offset 16
+; RV32-NEXT:    sw a1, 12(sp)
+; RV32-NEXT:    sw a0, 8(sp)
+; RV32-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; RV32-NEXT:    addi a0, sp, 8
+; RV32-NEXT:    vlse64.v v9, (a0), zero
+; RV32-NEXT:    vsetvli zero, a2, e64, m1, ta, ma
+; RV32-NEXT:    vmslt.vv v0, v8, v9, v0.t
+; RV32-NEXT:    addi sp, sp, 16
+; RV32-NEXT:    ret
+;
+; RV64-LABEL: icmp_sgt_vx_swap_nxv1i64:
+; RV64:       # %bb.0:
+; RV64-NEXT:    vsetvli zero, a1, e64, m1, ta, ma
+; RV64-NEXT:    vmslt.vx v0, v8, a0, v0.t
+; RV64-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x i64> poison, i64 %b, i32 0
+  %vb = shufflevector <vscale x 1 x i64> %elt.head, <vscale x 1 x i64> poison, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i64(<vscale x 1 x i64> %vb, <vscale x 1 x i64> %va, metadata !"sgt", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_sgt_vi_nxv1i64(<vscale x 1 x i64> %va, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sgt_vi_nxv1i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m1, ta, ma
+; CHECK-NEXT:    vmsgt.vi v0, v8, 4, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x i64> poison, i64 4, i32 0
+  %vb = shufflevector <vscale x 1 x i64> %elt.head, <vscale x 1 x i64> poison, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i64(<vscale x 1 x i64> %va, <vscale x 1 x i64> %vb, metadata !"sgt", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_sgt_vi_swap_nxv1i64(<vscale x 1 x i64> %va, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sgt_vi_swap_nxv1i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    li a1, 4
+; CHECK-NEXT:    vsetvli zero, a0, e64, m1, ta, ma
+; CHECK-NEXT:    vmslt.vx v0, v8, a1, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x i64> poison, i64 4, i32 0
+  %vb = shufflevector <vscale x 1 x i64> %elt.head, <vscale x 1 x i64> poison, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i64(<vscale x 1 x i64> %vb, <vscale x 1 x i64> %va, metadata !"sgt", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_sge_vv_nxv1i64(<vscale x 1 x i64> %va, <vscale x 1 x i64> %vb, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sge_vv_nxv1i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m1, ta, ma
+; CHECK-NEXT:    vmsle.vv v0, v9, v8, v0.t
+; CHECK-NEXT:    ret
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i64(<vscale x 1 x i64> %va, <vscale x 1 x i64> %vb, metadata !"sge", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_sge_vx_nxv1i64(<vscale x 1 x i64> %va, i64 %b, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; RV32-LABEL: icmp_sge_vx_nxv1i64:
+; RV32:       # %bb.0:
+; RV32-NEXT:    addi sp, sp, -16
+; RV32-NEXT:    .cfi_def_cfa_offset 16
+; RV32-NEXT:    sw a1, 12(sp)
+; RV32-NEXT:    sw a0, 8(sp)
+; RV32-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; RV32-NEXT:    addi a0, sp, 8
+; RV32-NEXT:    vlse64.v v9, (a0), zero
+; RV32-NEXT:    vsetvli zero, a2, e64, m1, ta, ma
+; RV32-NEXT:    vmsle.vv v0, v9, v8, v0.t
+; RV32-NEXT:    addi sp, sp, 16
+; RV32-NEXT:    ret
+;
+; RV64-LABEL: icmp_sge_vx_nxv1i64:
+; RV64:       # %bb.0:
+; RV64-NEXT:    vsetvli a2, zero, e64, m1, ta, mu
+; RV64-NEXT:    vmv.v.x v9, a0
+; RV64-NEXT:    vsetvli zero, a1, e64, m1, ta, ma
+; RV64-NEXT:    vmsle.vv v0, v9, v8, v0.t
+; RV64-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x i64> poison, i64 %b, i32 0
+  %vb = shufflevector <vscale x 1 x i64> %elt.head, <vscale x 1 x i64> poison, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i64(<vscale x 1 x i64> %va, <vscale x 1 x i64> %vb, metadata !"sge", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_sge_vx_swap_nxv1i64(<vscale x 1 x i64> %va, i64 %b, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; RV32-LABEL: icmp_sge_vx_swap_nxv1i64:
+; RV32:       # %bb.0:
+; RV32-NEXT:    addi sp, sp, -16
+; RV32-NEXT:    .cfi_def_cfa_offset 16
+; RV32-NEXT:    sw a1, 12(sp)
+; RV32-NEXT:    sw a0, 8(sp)
+; RV32-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; RV32-NEXT:    addi a0, sp, 8
+; RV32-NEXT:    vlse64.v v9, (a0), zero
+; RV32-NEXT:    vsetvli zero, a2, e64, m1, ta, ma
+; RV32-NEXT:    vmsle.vv v0, v8, v9, v0.t
+; RV32-NEXT:    addi sp, sp, 16
+; RV32-NEXT:    ret
+;
+; RV64-LABEL: icmp_sge_vx_swap_nxv1i64:
+; RV64:       # %bb.0:
+; RV64-NEXT:    vsetvli zero, a1, e64, m1, ta, ma
+; RV64-NEXT:    vmsle.vx v0, v8, a0, v0.t
+; RV64-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x i64> poison, i64 %b, i32 0
+  %vb = shufflevector <vscale x 1 x i64> %elt.head, <vscale x 1 x i64> poison, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i64(<vscale x 1 x i64> %vb, <vscale x 1 x i64> %va, metadata !"sge", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_sge_vi_nxv1i64(<vscale x 1 x i64> %va, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sge_vi_nxv1i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m1, ta, ma
+; CHECK-NEXT:    vmsgt.vi v0, v8, 3, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x i64> poison, i64 4, i32 0
+  %vb = shufflevector <vscale x 1 x i64> %elt.head, <vscale x 1 x i64> poison, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i64(<vscale x 1 x i64> %va, <vscale x 1 x i64> %vb, metadata !"sge", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_sge_vi_swap_nxv1i64(<vscale x 1 x i64> %va, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sge_vi_swap_nxv1i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m1, ta, ma
+; CHECK-NEXT:    vmsle.vi v0, v8, 4, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x i64> poison, i64 4, i32 0
+  %vb = shufflevector <vscale x 1 x i64> %elt.head, <vscale x 1 x i64> poison, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i64(<vscale x 1 x i64> %vb, <vscale x 1 x i64> %va, metadata !"sge", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_slt_vv_nxv1i64(<vscale x 1 x i64> %va, <vscale x 1 x i64> %vb, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_slt_vv_nxv1i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m1, ta, ma
+; CHECK-NEXT:    vmslt.vv v0, v8, v9, v0.t
+; CHECK-NEXT:    ret
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i64(<vscale x 1 x i64> %va, <vscale x 1 x i64> %vb, metadata !"slt", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_slt_vx_nxv1i64(<vscale x 1 x i64> %va, i64 %b, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; RV32-LABEL: icmp_slt_vx_nxv1i64:
+; RV32:       # %bb.0:
+; RV32-NEXT:    addi sp, sp, -16
+; RV32-NEXT:    .cfi_def_cfa_offset 16
+; RV32-NEXT:    sw a1, 12(sp)
+; RV32-NEXT:    sw a0, 8(sp)
+; RV32-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; RV32-NEXT:    addi a0, sp, 8
+; RV32-NEXT:    vlse64.v v9, (a0), zero
+; RV32-NEXT:    vsetvli zero, a2, e64, m1, ta, ma
+; RV32-NEXT:    vmslt.vv v0, v8, v9, v0.t
+; RV32-NEXT:    addi sp, sp, 16
+; RV32-NEXT:    ret
+;
+; RV64-LABEL: icmp_slt_vx_nxv1i64:
+; RV64:       # %bb.0:
+; RV64-NEXT:    vsetvli zero, a1, e64, m1, ta, ma
+; RV64-NEXT:    vmslt.vx v0, v8, a0, v0.t
+; RV64-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x i64> poison, i64 %b, i32 0
+  %vb = shufflevector <vscale x 1 x i64> %elt.head, <vscale x 1 x i64> poison, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i64(<vscale x 1 x i64> %va, <vscale x 1 x i64> %vb, metadata !"slt", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_slt_vx_swap_nxv1i64(<vscale x 1 x i64> %va, i64 %b, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; RV32-LABEL: icmp_slt_vx_swap_nxv1i64:
+; RV32:       # %bb.0:
+; RV32-NEXT:    addi sp, sp, -16
+; RV32-NEXT:    .cfi_def_cfa_offset 16
+; RV32-NEXT:    sw a1, 12(sp)
+; RV32-NEXT:    sw a0, 8(sp)
+; RV32-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; RV32-NEXT:    addi a0, sp, 8
+; RV32-NEXT:    vlse64.v v9, (a0), zero
+; RV32-NEXT:    vsetvli zero, a2, e64, m1, ta, ma
+; RV32-NEXT:    vmslt.vv v0, v9, v8, v0.t
+; RV32-NEXT:    addi sp, sp, 16
+; RV32-NEXT:    ret
+;
+; RV64-LABEL: icmp_slt_vx_swap_nxv1i64:
+; RV64:       # %bb.0:
+; RV64-NEXT:    vsetvli zero, a1, e64, m1, ta, ma
+; RV64-NEXT:    vmsgt.vx v0, v8, a0, v0.t
+; RV64-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x i64> poison, i64 %b, i32 0
+  %vb = shufflevector <vscale x 1 x i64> %elt.head, <vscale x 1 x i64> poison, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i64(<vscale x 1 x i64> %vb, <vscale x 1 x i64> %va, metadata !"slt", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_slt_vi_nxv1i64(<vscale x 1 x i64> %va, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_slt_vi_nxv1i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m1, ta, ma
+; CHECK-NEXT:    vmsle.vi v0, v8, 3, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x i64> poison, i64 4, i32 0
+  %vb = shufflevector <vscale x 1 x i64> %elt.head, <vscale x 1 x i64> poison, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i64(<vscale x 1 x i64> %va, <vscale x 1 x i64> %vb, metadata !"slt", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_slt_vi_swap_nxv1i64(<vscale x 1 x i64> %va, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_slt_vi_swap_nxv1i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m1, ta, ma
+; CHECK-NEXT:    vmsgt.vi v0, v8, 4, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x i64> poison, i64 4, i32 0
+  %vb = shufflevector <vscale x 1 x i64> %elt.head, <vscale x 1 x i64> poison, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i64(<vscale x 1 x i64> %vb, <vscale x 1 x i64> %va, metadata !"slt", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_sle_vv_nxv1i64(<vscale x 1 x i64> %va, <vscale x 1 x i64> %vb, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sle_vv_nxv1i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m1, ta, ma
+; CHECK-NEXT:    vmsle.vv v0, v8, v9, v0.t
+; CHECK-NEXT:    ret
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i64(<vscale x 1 x i64> %va, <vscale x 1 x i64> %vb, metadata !"sle", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_sle_vx_nxv1i64(<vscale x 1 x i64> %va, i64 %b, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; RV32-LABEL: icmp_sle_vx_nxv1i64:
+; RV32:       # %bb.0:
+; RV32-NEXT:    addi sp, sp, -16
+; RV32-NEXT:    .cfi_def_cfa_offset 16
+; RV32-NEXT:    sw a1, 12(sp)
+; RV32-NEXT:    sw a0, 8(sp)
+; RV32-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; RV32-NEXT:    addi a0, sp, 8
+; RV32-NEXT:    vlse64.v v9, (a0), zero
+; RV32-NEXT:    vsetvli zero, a2, e64, m1, ta, ma
+; RV32-NEXT:    vmsle.vv v0, v8, v9, v0.t
+; RV32-NEXT:    addi sp, sp, 16
+; RV32-NEXT:    ret
+;
+; RV64-LABEL: icmp_sle_vx_nxv1i64:
+; RV64:       # %bb.0:
+; RV64-NEXT:    vsetvli zero, a1, e64, m1, ta, ma
+; RV64-NEXT:    vmsle.vx v0, v8, a0, v0.t
+; RV64-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x i64> poison, i64 %b, i32 0
+  %vb = shufflevector <vscale x 1 x i64> %elt.head, <vscale x 1 x i64> poison, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i64(<vscale x 1 x i64> %va, <vscale x 1 x i64> %vb, metadata !"sle", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_sle_vx_swap_nxv1i64(<vscale x 1 x i64> %va, i64 %b, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; RV32-LABEL: icmp_sle_vx_swap_nxv1i64:
+; RV32:       # %bb.0:
+; RV32-NEXT:    addi sp, sp, -16
+; RV32-NEXT:    .cfi_def_cfa_offset 16
+; RV32-NEXT:    sw a1, 12(sp)
+; RV32-NEXT:    sw a0, 8(sp)
+; RV32-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; RV32-NEXT:    addi a0, sp, 8
+; RV32-NEXT:    vlse64.v v9, (a0), zero
+; RV32-NEXT:    vsetvli zero, a2, e64, m1, ta, ma
+; RV32-NEXT:    vmsle.vv v0, v9, v8, v0.t
+; RV32-NEXT:    addi sp, sp, 16
+; RV32-NEXT:    ret
+;
+; RV64-LABEL: icmp_sle_vx_swap_nxv1i64:
+; RV64:       # %bb.0:
+; RV64-NEXT:    vsetvli a2, zero, e64, m1, ta, mu
+; RV64-NEXT:    vmv.v.x v9, a0
+; RV64-NEXT:    vsetvli zero, a1, e64, m1, ta, ma
+; RV64-NEXT:    vmsle.vv v0, v9, v8, v0.t
+; RV64-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x i64> poison, i64 %b, i32 0
+  %vb = shufflevector <vscale x 1 x i64> %elt.head, <vscale x 1 x i64> poison, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i64(<vscale x 1 x i64> %vb, <vscale x 1 x i64> %va, metadata !"sle", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_sle_vi_nxv1i64(<vscale x 1 x i64> %va, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sle_vi_nxv1i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m1, ta, ma
+; CHECK-NEXT:    vmsle.vi v0, v8, 4, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x i64> poison, i64 4, i32 0
+  %vb = shufflevector <vscale x 1 x i64> %elt.head, <vscale x 1 x i64> poison, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i64(<vscale x 1 x i64> %va, <vscale x 1 x i64> %vb, metadata !"sle", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+define <vscale x 1 x i1> @icmp_sle_vi_swap_nxv1i64(<vscale x 1 x i64> %va, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sle_vi_swap_nxv1i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e64, m1, ta, mu
+; CHECK-NEXT:    vmv.v.i v9, 4
+; CHECK-NEXT:    vsetvli zero, a0, e64, m1, ta, ma
+; CHECK-NEXT:    vmsle.vv v0, v9, v8, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x i64> poison, i64 4, i32 0
+  %vb = shufflevector <vscale x 1 x i64> %elt.head, <vscale x 1 x i64> poison, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x i1> @llvm.vp.icmp.nxv1i64(<vscale x 1 x i64> %vb, <vscale x 1 x i64> %va, metadata !"sle", <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x i1> %v
+}
+
+declare <vscale x 8 x i1> @llvm.vp.icmp.nxv8i64(<vscale x 8 x i64>, <vscale x 8 x i64>, metadata, <vscale x 8 x i1>, i32)
+
+define <vscale x 8 x i1> @icmp_eq_vv_nxv8i64(<vscale x 8 x i64> %va, <vscale x 8 x i64> %vb, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_eq_vv_nxv8i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m8, ta, ma
+; CHECK-NEXT:    vmseq.vv v24, v8, v16, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v24
+; CHECK-NEXT:    ret
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i64(<vscale x 8 x i64> %va, <vscale x 8 x i64> %vb, metadata !"eq", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_eq_vx_nxv8i64(<vscale x 8 x i64> %va, i64 %b, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; RV32-LABEL: icmp_eq_vx_nxv8i64:
+; RV32:       # %bb.0:
+; RV32-NEXT:    addi sp, sp, -16
+; RV32-NEXT:    .cfi_def_cfa_offset 16
+; RV32-NEXT:    sw a1, 12(sp)
+; RV32-NEXT:    sw a0, 8(sp)
+; RV32-NEXT:    vsetvli a0, zero, e64, m8, ta, mu
+; RV32-NEXT:    addi a0, sp, 8
+; RV32-NEXT:    vlse64.v v24, (a0), zero
+; RV32-NEXT:    vsetvli zero, a2, e64, m8, ta, ma
+; RV32-NEXT:    vmseq.vv v16, v8, v24, v0.t
+; RV32-NEXT:    vmv1r.v v0, v16
+; RV32-NEXT:    addi sp, sp, 16
+; RV32-NEXT:    ret
+;
+; RV64-LABEL: icmp_eq_vx_nxv8i64:
+; RV64:       # %bb.0:
+; RV64-NEXT:    vsetvli zero, a1, e64, m8, ta, ma
+; RV64-NEXT:    vmseq.vx v16, v8, a0, v0.t
+; RV64-NEXT:    vmv1r.v v0, v16
+; RV64-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x i64> poison, i64 %b, i32 0
+  %vb = shufflevector <vscale x 8 x i64> %elt.head, <vscale x 8 x i64> poison, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i64(<vscale x 8 x i64> %va, <vscale x 8 x i64> %vb, metadata !"eq", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_eq_vx_swap_nxv8i64(<vscale x 8 x i64> %va, i64 %b, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; RV32-LABEL: icmp_eq_vx_swap_nxv8i64:
+; RV32:       # %bb.0:
+; RV32-NEXT:    addi sp, sp, -16
+; RV32-NEXT:    .cfi_def_cfa_offset 16
+; RV32-NEXT:    sw a1, 12(sp)
+; RV32-NEXT:    sw a0, 8(sp)
+; RV32-NEXT:    vsetvli a0, zero, e64, m8, ta, mu
+; RV32-NEXT:    addi a0, sp, 8
+; RV32-NEXT:    vlse64.v v24, (a0), zero
+; RV32-NEXT:    vsetvli zero, a2, e64, m8, ta, ma
+; RV32-NEXT:    vmseq.vv v16, v24, v8, v0.t
+; RV32-NEXT:    vmv1r.v v0, v16
+; RV32-NEXT:    addi sp, sp, 16
+; RV32-NEXT:    ret
+;
+; RV64-LABEL: icmp_eq_vx_swap_nxv8i64:
+; RV64:       # %bb.0:
+; RV64-NEXT:    vsetvli zero, a1, e64, m8, ta, ma
+; RV64-NEXT:    vmseq.vx v16, v8, a0, v0.t
+; RV64-NEXT:    vmv1r.v v0, v16
+; RV64-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x i64> poison, i64 %b, i32 0
+  %vb = shufflevector <vscale x 8 x i64> %elt.head, <vscale x 8 x i64> poison, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i64(<vscale x 8 x i64> %vb, <vscale x 8 x i64> %va, metadata !"eq", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_eq_vi_nxv8i64(<vscale x 8 x i64> %va, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_eq_vi_nxv8i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m8, ta, ma
+; CHECK-NEXT:    vmseq.vi v16, v8, 4, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v16
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x i64> poison, i64 4, i32 0
+  %vb = shufflevector <vscale x 8 x i64> %elt.head, <vscale x 8 x i64> poison, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i64(<vscale x 8 x i64> %va, <vscale x 8 x i64> %vb, metadata !"eq", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_eq_vi_swap_nxv8i64(<vscale x 8 x i64> %va, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_eq_vi_swap_nxv8i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m8, ta, ma
+; CHECK-NEXT:    vmseq.vi v16, v8, 4, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v16
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x i64> poison, i64 4, i32 0
+  %vb = shufflevector <vscale x 8 x i64> %elt.head, <vscale x 8 x i64> poison, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i64(<vscale x 8 x i64> %vb, <vscale x 8 x i64> %va, metadata !"eq", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_ne_vv_nxv8i64(<vscale x 8 x i64> %va, <vscale x 8 x i64> %vb, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ne_vv_nxv8i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m8, ta, ma
+; CHECK-NEXT:    vmsne.vv v24, v8, v16, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v24
+; CHECK-NEXT:    ret
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i64(<vscale x 8 x i64> %va, <vscale x 8 x i64> %vb, metadata !"ne", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_ne_vx_nxv8i64(<vscale x 8 x i64> %va, i64 %b, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; RV32-LABEL: icmp_ne_vx_nxv8i64:
+; RV32:       # %bb.0:
+; RV32-NEXT:    addi sp, sp, -16
+; RV32-NEXT:    .cfi_def_cfa_offset 16
+; RV32-NEXT:    sw a1, 12(sp)
+; RV32-NEXT:    sw a0, 8(sp)
+; RV32-NEXT:    vsetvli a0, zero, e64, m8, ta, mu
+; RV32-NEXT:    addi a0, sp, 8
+; RV32-NEXT:    vlse64.v v24, (a0), zero
+; RV32-NEXT:    vsetvli zero, a2, e64, m8, ta, ma
+; RV32-NEXT:    vmsne.vv v16, v8, v24, v0.t
+; RV32-NEXT:    vmv1r.v v0, v16
+; RV32-NEXT:    addi sp, sp, 16
+; RV32-NEXT:    ret
+;
+; RV64-LABEL: icmp_ne_vx_nxv8i64:
+; RV64:       # %bb.0:
+; RV64-NEXT:    vsetvli zero, a1, e64, m8, ta, ma
+; RV64-NEXT:    vmsne.vx v16, v8, a0, v0.t
+; RV64-NEXT:    vmv1r.v v0, v16
+; RV64-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x i64> poison, i64 %b, i32 0
+  %vb = shufflevector <vscale x 8 x i64> %elt.head, <vscale x 8 x i64> poison, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i64(<vscale x 8 x i64> %va, <vscale x 8 x i64> %vb, metadata !"ne", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_ne_vx_swap_nxv8i64(<vscale x 8 x i64> %va, i64 %b, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; RV32-LABEL: icmp_ne_vx_swap_nxv8i64:
+; RV32:       # %bb.0:
+; RV32-NEXT:    addi sp, sp, -16
+; RV32-NEXT:    .cfi_def_cfa_offset 16
+; RV32-NEXT:    sw a1, 12(sp)
+; RV32-NEXT:    sw a0, 8(sp)
+; RV32-NEXT:    vsetvli a0, zero, e64, m8, ta, mu
+; RV32-NEXT:    addi a0, sp, 8
+; RV32-NEXT:    vlse64.v v24, (a0), zero
+; RV32-NEXT:    vsetvli zero, a2, e64, m8, ta, ma
+; RV32-NEXT:    vmsne.vv v16, v24, v8, v0.t
+; RV32-NEXT:    vmv1r.v v0, v16
+; RV32-NEXT:    addi sp, sp, 16
+; RV32-NEXT:    ret
+;
+; RV64-LABEL: icmp_ne_vx_swap_nxv8i64:
+; RV64:       # %bb.0:
+; RV64-NEXT:    vsetvli zero, a1, e64, m8, ta, ma
+; RV64-NEXT:    vmsne.vx v16, v8, a0, v0.t
+; RV64-NEXT:    vmv1r.v v0, v16
+; RV64-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x i64> poison, i64 %b, i32 0
+  %vb = shufflevector <vscale x 8 x i64> %elt.head, <vscale x 8 x i64> poison, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i64(<vscale x 8 x i64> %vb, <vscale x 8 x i64> %va, metadata !"ne", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_ne_vi_nxv8i64(<vscale x 8 x i64> %va, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ne_vi_nxv8i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m8, ta, ma
+; CHECK-NEXT:    vmsne.vi v16, v8, 4, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v16
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x i64> poison, i64 4, i32 0
+  %vb = shufflevector <vscale x 8 x i64> %elt.head, <vscale x 8 x i64> poison, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i64(<vscale x 8 x i64> %va, <vscale x 8 x i64> %vb, metadata !"ne", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_ne_vi_swap_nxv8i64(<vscale x 8 x i64> %va, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ne_vi_swap_nxv8i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m8, ta, ma
+; CHECK-NEXT:    vmsne.vi v16, v8, 4, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v16
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x i64> poison, i64 4, i32 0
+  %vb = shufflevector <vscale x 8 x i64> %elt.head, <vscale x 8 x i64> poison, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i64(<vscale x 8 x i64> %vb, <vscale x 8 x i64> %va, metadata !"ne", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_ugt_vv_nxv8i64(<vscale x 8 x i64> %va, <vscale x 8 x i64> %vb, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ugt_vv_nxv8i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m8, ta, ma
+; CHECK-NEXT:    vmsltu.vv v24, v16, v8, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v24
+; CHECK-NEXT:    ret
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i64(<vscale x 8 x i64> %va, <vscale x 8 x i64> %vb, metadata !"ugt", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_ugt_vx_nxv8i64(<vscale x 8 x i64> %va, i64 %b, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; RV32-LABEL: icmp_ugt_vx_nxv8i64:
+; RV32:       # %bb.0:
+; RV32-NEXT:    addi sp, sp, -16
+; RV32-NEXT:    .cfi_def_cfa_offset 16
+; RV32-NEXT:    sw a1, 12(sp)
+; RV32-NEXT:    sw a0, 8(sp)
+; RV32-NEXT:    vsetvli a0, zero, e64, m8, ta, mu
+; RV32-NEXT:    addi a0, sp, 8
+; RV32-NEXT:    vlse64.v v24, (a0), zero
+; RV32-NEXT:    vsetvli zero, a2, e64, m8, ta, ma
+; RV32-NEXT:    vmsltu.vv v16, v24, v8, v0.t
+; RV32-NEXT:    vmv1r.v v0, v16
+; RV32-NEXT:    addi sp, sp, 16
+; RV32-NEXT:    ret
+;
+; RV64-LABEL: icmp_ugt_vx_nxv8i64:
+; RV64:       # %bb.0:
+; RV64-NEXT:    vsetvli zero, a1, e64, m8, ta, ma
+; RV64-NEXT:    vmsgtu.vx v16, v8, a0, v0.t
+; RV64-NEXT:    vmv1r.v v0, v16
+; RV64-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x i64> poison, i64 %b, i32 0
+  %vb = shufflevector <vscale x 8 x i64> %elt.head, <vscale x 8 x i64> poison, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i64(<vscale x 8 x i64> %va, <vscale x 8 x i64> %vb, metadata !"ugt", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_ugt_vx_swap_nxv8i64(<vscale x 8 x i64> %va, i64 %b, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; RV32-LABEL: icmp_ugt_vx_swap_nxv8i64:
+; RV32:       # %bb.0:
+; RV32-NEXT:    addi sp, sp, -16
+; RV32-NEXT:    .cfi_def_cfa_offset 16
+; RV32-NEXT:    sw a1, 12(sp)
+; RV32-NEXT:    sw a0, 8(sp)
+; RV32-NEXT:    vsetvli a0, zero, e64, m8, ta, mu
+; RV32-NEXT:    addi a0, sp, 8
+; RV32-NEXT:    vlse64.v v24, (a0), zero
+; RV32-NEXT:    vsetvli zero, a2, e64, m8, ta, ma
+; RV32-NEXT:    vmsltu.vv v16, v8, v24, v0.t
+; RV32-NEXT:    vmv1r.v v0, v16
+; RV32-NEXT:    addi sp, sp, 16
+; RV32-NEXT:    ret
+;
+; RV64-LABEL: icmp_ugt_vx_swap_nxv8i64:
+; RV64:       # %bb.0:
+; RV64-NEXT:    vsetvli zero, a1, e64, m8, ta, ma
+; RV64-NEXT:    vmsltu.vx v16, v8, a0, v0.t
+; RV64-NEXT:    vmv1r.v v0, v16
+; RV64-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x i64> poison, i64 %b, i32 0
+  %vb = shufflevector <vscale x 8 x i64> %elt.head, <vscale x 8 x i64> poison, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i64(<vscale x 8 x i64> %vb, <vscale x 8 x i64> %va, metadata !"ugt", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_ugt_vi_nxv8i64(<vscale x 8 x i64> %va, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ugt_vi_nxv8i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m8, ta, ma
+; CHECK-NEXT:    vmsgtu.vi v16, v8, 4, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v16
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x i64> poison, i64 4, i32 0
+  %vb = shufflevector <vscale x 8 x i64> %elt.head, <vscale x 8 x i64> poison, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i64(<vscale x 8 x i64> %va, <vscale x 8 x i64> %vb, metadata !"ugt", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_ugt_vi_swap_nxv8i64(<vscale x 8 x i64> %va, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ugt_vi_swap_nxv8i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    li a1, 4
+; CHECK-NEXT:    vsetvli zero, a0, e64, m8, ta, ma
+; CHECK-NEXT:    vmsltu.vx v16, v8, a1, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v16
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x i64> poison, i64 4, i32 0
+  %vb = shufflevector <vscale x 8 x i64> %elt.head, <vscale x 8 x i64> poison, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i64(<vscale x 8 x i64> %vb, <vscale x 8 x i64> %va, metadata !"ugt", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_uge_vv_nxv8i64(<vscale x 8 x i64> %va, <vscale x 8 x i64> %vb, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_uge_vv_nxv8i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m8, ta, ma
+; CHECK-NEXT:    vmsleu.vv v24, v16, v8, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v24
+; CHECK-NEXT:    ret
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i64(<vscale x 8 x i64> %va, <vscale x 8 x i64> %vb, metadata !"uge", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_uge_vx_nxv8i64(<vscale x 8 x i64> %va, i64 %b, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; RV32-LABEL: icmp_uge_vx_nxv8i64:
+; RV32:       # %bb.0:
+; RV32-NEXT:    addi sp, sp, -16
+; RV32-NEXT:    .cfi_def_cfa_offset 16
+; RV32-NEXT:    sw a1, 12(sp)
+; RV32-NEXT:    sw a0, 8(sp)
+; RV32-NEXT:    vsetvli a0, zero, e64, m8, ta, mu
+; RV32-NEXT:    addi a0, sp, 8
+; RV32-NEXT:    vlse64.v v24, (a0), zero
+; RV32-NEXT:    vsetvli zero, a2, e64, m8, ta, ma
+; RV32-NEXT:    vmsleu.vv v16, v24, v8, v0.t
+; RV32-NEXT:    vmv1r.v v0, v16
+; RV32-NEXT:    addi sp, sp, 16
+; RV32-NEXT:    ret
+;
+; RV64-LABEL: icmp_uge_vx_nxv8i64:
+; RV64:       # %bb.0:
+; RV64-NEXT:    vsetvli a2, zero, e64, m8, ta, mu
+; RV64-NEXT:    vmv.v.x v24, a0
+; RV64-NEXT:    vsetvli zero, a1, e64, m8, ta, ma
+; RV64-NEXT:    vmsleu.vv v16, v24, v8, v0.t
+; RV64-NEXT:    vmv1r.v v0, v16
+; RV64-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x i64> poison, i64 %b, i32 0
+  %vb = shufflevector <vscale x 8 x i64> %elt.head, <vscale x 8 x i64> poison, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i64(<vscale x 8 x i64> %va, <vscale x 8 x i64> %vb, metadata !"uge", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_uge_vx_swap_nxv8i64(<vscale x 8 x i64> %va, i64 %b, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; RV32-LABEL: icmp_uge_vx_swap_nxv8i64:
+; RV32:       # %bb.0:
+; RV32-NEXT:    addi sp, sp, -16
+; RV32-NEXT:    .cfi_def_cfa_offset 16
+; RV32-NEXT:    sw a1, 12(sp)
+; RV32-NEXT:    sw a0, 8(sp)
+; RV32-NEXT:    vsetvli a0, zero, e64, m8, ta, mu
+; RV32-NEXT:    addi a0, sp, 8
+; RV32-NEXT:    vlse64.v v24, (a0), zero
+; RV32-NEXT:    vsetvli zero, a2, e64, m8, ta, ma
+; RV32-NEXT:    vmsleu.vv v16, v8, v24, v0.t
+; RV32-NEXT:    vmv1r.v v0, v16
+; RV32-NEXT:    addi sp, sp, 16
+; RV32-NEXT:    ret
+;
+; RV64-LABEL: icmp_uge_vx_swap_nxv8i64:
+; RV64:       # %bb.0:
+; RV64-NEXT:    vsetvli zero, a1, e64, m8, ta, ma
+; RV64-NEXT:    vmsleu.vx v16, v8, a0, v0.t
+; RV64-NEXT:    vmv1r.v v0, v16
+; RV64-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x i64> poison, i64 %b, i32 0
+  %vb = shufflevector <vscale x 8 x i64> %elt.head, <vscale x 8 x i64> poison, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i64(<vscale x 8 x i64> %vb, <vscale x 8 x i64> %va, metadata !"uge", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_uge_vi_nxv8i64(<vscale x 8 x i64> %va, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_uge_vi_nxv8i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m8, ta, ma
+; CHECK-NEXT:    vmsgtu.vi v16, v8, 3, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v16
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x i64> poison, i64 4, i32 0
+  %vb = shufflevector <vscale x 8 x i64> %elt.head, <vscale x 8 x i64> poison, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i64(<vscale x 8 x i64> %va, <vscale x 8 x i64> %vb, metadata !"uge", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_uge_vi_swap_nxv8i64(<vscale x 8 x i64> %va, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_uge_vi_swap_nxv8i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m8, ta, ma
+; CHECK-NEXT:    vmsleu.vi v16, v8, 4, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v16
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x i64> poison, i64 4, i32 0
+  %vb = shufflevector <vscale x 8 x i64> %elt.head, <vscale x 8 x i64> poison, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i64(<vscale x 8 x i64> %vb, <vscale x 8 x i64> %va, metadata !"uge", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_ult_vv_nxv8i64(<vscale x 8 x i64> %va, <vscale x 8 x i64> %vb, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ult_vv_nxv8i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m8, ta, ma
+; CHECK-NEXT:    vmsltu.vv v24, v8, v16, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v24
+; CHECK-NEXT:    ret
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i64(<vscale x 8 x i64> %va, <vscale x 8 x i64> %vb, metadata !"ult", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_ult_vx_nxv8i64(<vscale x 8 x i64> %va, i64 %b, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; RV32-LABEL: icmp_ult_vx_nxv8i64:
+; RV32:       # %bb.0:
+; RV32-NEXT:    addi sp, sp, -16
+; RV32-NEXT:    .cfi_def_cfa_offset 16
+; RV32-NEXT:    sw a1, 12(sp)
+; RV32-NEXT:    sw a0, 8(sp)
+; RV32-NEXT:    vsetvli a0, zero, e64, m8, ta, mu
+; RV32-NEXT:    addi a0, sp, 8
+; RV32-NEXT:    vlse64.v v24, (a0), zero
+; RV32-NEXT:    vsetvli zero, a2, e64, m8, ta, ma
+; RV32-NEXT:    vmsltu.vv v16, v8, v24, v0.t
+; RV32-NEXT:    vmv1r.v v0, v16
+; RV32-NEXT:    addi sp, sp, 16
+; RV32-NEXT:    ret
+;
+; RV64-LABEL: icmp_ult_vx_nxv8i64:
+; RV64:       # %bb.0:
+; RV64-NEXT:    vsetvli zero, a1, e64, m8, ta, ma
+; RV64-NEXT:    vmsltu.vx v16, v8, a0, v0.t
+; RV64-NEXT:    vmv1r.v v0, v16
+; RV64-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x i64> poison, i64 %b, i32 0
+  %vb = shufflevector <vscale x 8 x i64> %elt.head, <vscale x 8 x i64> poison, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i64(<vscale x 8 x i64> %va, <vscale x 8 x i64> %vb, metadata !"ult", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_ult_vx_swap_nxv8i64(<vscale x 8 x i64> %va, i64 %b, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; RV32-LABEL: icmp_ult_vx_swap_nxv8i64:
+; RV32:       # %bb.0:
+; RV32-NEXT:    addi sp, sp, -16
+; RV32-NEXT:    .cfi_def_cfa_offset 16
+; RV32-NEXT:    sw a1, 12(sp)
+; RV32-NEXT:    sw a0, 8(sp)
+; RV32-NEXT:    vsetvli a0, zero, e64, m8, ta, mu
+; RV32-NEXT:    addi a0, sp, 8
+; RV32-NEXT:    vlse64.v v24, (a0), zero
+; RV32-NEXT:    vsetvli zero, a2, e64, m8, ta, ma
+; RV32-NEXT:    vmsltu.vv v16, v24, v8, v0.t
+; RV32-NEXT:    vmv1r.v v0, v16
+; RV32-NEXT:    addi sp, sp, 16
+; RV32-NEXT:    ret
+;
+; RV64-LABEL: icmp_ult_vx_swap_nxv8i64:
+; RV64:       # %bb.0:
+; RV64-NEXT:    vsetvli zero, a1, e64, m8, ta, ma
+; RV64-NEXT:    vmsgtu.vx v16, v8, a0, v0.t
+; RV64-NEXT:    vmv1r.v v0, v16
+; RV64-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x i64> poison, i64 %b, i32 0
+  %vb = shufflevector <vscale x 8 x i64> %elt.head, <vscale x 8 x i64> poison, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i64(<vscale x 8 x i64> %vb, <vscale x 8 x i64> %va, metadata !"ult", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_ult_vi_nxv8i64(<vscale x 8 x i64> %va, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ult_vi_nxv8i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m8, ta, ma
+; CHECK-NEXT:    vmsleu.vi v16, v8, 3, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v16
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x i64> poison, i64 4, i32 0
+  %vb = shufflevector <vscale x 8 x i64> %elt.head, <vscale x 8 x i64> poison, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i64(<vscale x 8 x i64> %va, <vscale x 8 x i64> %vb, metadata !"ult", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_ult_vi_swap_nxv8i64(<vscale x 8 x i64> %va, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_ult_vi_swap_nxv8i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m8, ta, ma
+; CHECK-NEXT:    vmsgtu.vi v16, v8, 4, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v16
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x i64> poison, i64 4, i32 0
+  %vb = shufflevector <vscale x 8 x i64> %elt.head, <vscale x 8 x i64> poison, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i64(<vscale x 8 x i64> %vb, <vscale x 8 x i64> %va, metadata !"ult", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_sgt_vv_nxv8i64(<vscale x 8 x i64> %va, <vscale x 8 x i64> %vb, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sgt_vv_nxv8i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m8, ta, ma
+; CHECK-NEXT:    vmslt.vv v24, v16, v8, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v24
+; CHECK-NEXT:    ret
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i64(<vscale x 8 x i64> %va, <vscale x 8 x i64> %vb, metadata !"sgt", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_sgt_vx_nxv8i64(<vscale x 8 x i64> %va, i64 %b, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; RV32-LABEL: icmp_sgt_vx_nxv8i64:
+; RV32:       # %bb.0:
+; RV32-NEXT:    addi sp, sp, -16
+; RV32-NEXT:    .cfi_def_cfa_offset 16
+; RV32-NEXT:    sw a1, 12(sp)
+; RV32-NEXT:    sw a0, 8(sp)
+; RV32-NEXT:    vsetvli a0, zero, e64, m8, ta, mu
+; RV32-NEXT:    addi a0, sp, 8
+; RV32-NEXT:    vlse64.v v24, (a0), zero
+; RV32-NEXT:    vsetvli zero, a2, e64, m8, ta, ma
+; RV32-NEXT:    vmslt.vv v16, v24, v8, v0.t
+; RV32-NEXT:    vmv1r.v v0, v16
+; RV32-NEXT:    addi sp, sp, 16
+; RV32-NEXT:    ret
+;
+; RV64-LABEL: icmp_sgt_vx_nxv8i64:
+; RV64:       # %bb.0:
+; RV64-NEXT:    vsetvli zero, a1, e64, m8, ta, ma
+; RV64-NEXT:    vmsgt.vx v16, v8, a0, v0.t
+; RV64-NEXT:    vmv1r.v v0, v16
+; RV64-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x i64> poison, i64 %b, i32 0
+  %vb = shufflevector <vscale x 8 x i64> %elt.head, <vscale x 8 x i64> poison, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i64(<vscale x 8 x i64> %va, <vscale x 8 x i64> %vb, metadata !"sgt", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_sgt_vx_swap_nxv8i64(<vscale x 8 x i64> %va, i64 %b, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; RV32-LABEL: icmp_sgt_vx_swap_nxv8i64:
+; RV32:       # %bb.0:
+; RV32-NEXT:    addi sp, sp, -16
+; RV32-NEXT:    .cfi_def_cfa_offset 16
+; RV32-NEXT:    sw a1, 12(sp)
+; RV32-NEXT:    sw a0, 8(sp)
+; RV32-NEXT:    vsetvli a0, zero, e64, m8, ta, mu
+; RV32-NEXT:    addi a0, sp, 8
+; RV32-NEXT:    vlse64.v v24, (a0), zero
+; RV32-NEXT:    vsetvli zero, a2, e64, m8, ta, ma
+; RV32-NEXT:    vmslt.vv v16, v8, v24, v0.t
+; RV32-NEXT:    vmv1r.v v0, v16
+; RV32-NEXT:    addi sp, sp, 16
+; RV32-NEXT:    ret
+;
+; RV64-LABEL: icmp_sgt_vx_swap_nxv8i64:
+; RV64:       # %bb.0:
+; RV64-NEXT:    vsetvli zero, a1, e64, m8, ta, ma
+; RV64-NEXT:    vmslt.vx v16, v8, a0, v0.t
+; RV64-NEXT:    vmv1r.v v0, v16
+; RV64-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x i64> poison, i64 %b, i32 0
+  %vb = shufflevector <vscale x 8 x i64> %elt.head, <vscale x 8 x i64> poison, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i64(<vscale x 8 x i64> %vb, <vscale x 8 x i64> %va, metadata !"sgt", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_sgt_vi_nxv8i64(<vscale x 8 x i64> %va, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sgt_vi_nxv8i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m8, ta, ma
+; CHECK-NEXT:    vmsgt.vi v16, v8, 4, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v16
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x i64> poison, i64 4, i32 0
+  %vb = shufflevector <vscale x 8 x i64> %elt.head, <vscale x 8 x i64> poison, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i64(<vscale x 8 x i64> %va, <vscale x 8 x i64> %vb, metadata !"sgt", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_sgt_vi_swap_nxv8i64(<vscale x 8 x i64> %va, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sgt_vi_swap_nxv8i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    li a1, 4
+; CHECK-NEXT:    vsetvli zero, a0, e64, m8, ta, ma
+; CHECK-NEXT:    vmslt.vx v16, v8, a1, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v16
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x i64> poison, i64 4, i32 0
+  %vb = shufflevector <vscale x 8 x i64> %elt.head, <vscale x 8 x i64> poison, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i64(<vscale x 8 x i64> %vb, <vscale x 8 x i64> %va, metadata !"sgt", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_sge_vv_nxv8i64(<vscale x 8 x i64> %va, <vscale x 8 x i64> %vb, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sge_vv_nxv8i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m8, ta, ma
+; CHECK-NEXT:    vmsle.vv v24, v16, v8, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v24
+; CHECK-NEXT:    ret
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i64(<vscale x 8 x i64> %va, <vscale x 8 x i64> %vb, metadata !"sge", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_sge_vx_nxv8i64(<vscale x 8 x i64> %va, i64 %b, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; RV32-LABEL: icmp_sge_vx_nxv8i64:
+; RV32:       # %bb.0:
+; RV32-NEXT:    addi sp, sp, -16
+; RV32-NEXT:    .cfi_def_cfa_offset 16
+; RV32-NEXT:    sw a1, 12(sp)
+; RV32-NEXT:    sw a0, 8(sp)
+; RV32-NEXT:    vsetvli a0, zero, e64, m8, ta, mu
+; RV32-NEXT:    addi a0, sp, 8
+; RV32-NEXT:    vlse64.v v24, (a0), zero
+; RV32-NEXT:    vsetvli zero, a2, e64, m8, ta, ma
+; RV32-NEXT:    vmsle.vv v16, v24, v8, v0.t
+; RV32-NEXT:    vmv1r.v v0, v16
+; RV32-NEXT:    addi sp, sp, 16
+; RV32-NEXT:    ret
+;
+; RV64-LABEL: icmp_sge_vx_nxv8i64:
+; RV64:       # %bb.0:
+; RV64-NEXT:    vsetvli a2, zero, e64, m8, ta, mu
+; RV64-NEXT:    vmv.v.x v24, a0
+; RV64-NEXT:    vsetvli zero, a1, e64, m8, ta, ma
+; RV64-NEXT:    vmsle.vv v16, v24, v8, v0.t
+; RV64-NEXT:    vmv1r.v v0, v16
+; RV64-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x i64> poison, i64 %b, i32 0
+  %vb = shufflevector <vscale x 8 x i64> %elt.head, <vscale x 8 x i64> poison, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i64(<vscale x 8 x i64> %va, <vscale x 8 x i64> %vb, metadata !"sge", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_sge_vx_swap_nxv8i64(<vscale x 8 x i64> %va, i64 %b, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; RV32-LABEL: icmp_sge_vx_swap_nxv8i64:
+; RV32:       # %bb.0:
+; RV32-NEXT:    addi sp, sp, -16
+; RV32-NEXT:    .cfi_def_cfa_offset 16
+; RV32-NEXT:    sw a1, 12(sp)
+; RV32-NEXT:    sw a0, 8(sp)
+; RV32-NEXT:    vsetvli a0, zero, e64, m8, ta, mu
+; RV32-NEXT:    addi a0, sp, 8
+; RV32-NEXT:    vlse64.v v24, (a0), zero
+; RV32-NEXT:    vsetvli zero, a2, e64, m8, ta, ma
+; RV32-NEXT:    vmsle.vv v16, v8, v24, v0.t
+; RV32-NEXT:    vmv1r.v v0, v16
+; RV32-NEXT:    addi sp, sp, 16
+; RV32-NEXT:    ret
+;
+; RV64-LABEL: icmp_sge_vx_swap_nxv8i64:
+; RV64:       # %bb.0:
+; RV64-NEXT:    vsetvli zero, a1, e64, m8, ta, ma
+; RV64-NEXT:    vmsle.vx v16, v8, a0, v0.t
+; RV64-NEXT:    vmv1r.v v0, v16
+; RV64-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x i64> poison, i64 %b, i32 0
+  %vb = shufflevector <vscale x 8 x i64> %elt.head, <vscale x 8 x i64> poison, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i64(<vscale x 8 x i64> %vb, <vscale x 8 x i64> %va, metadata !"sge", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_sge_vi_nxv8i64(<vscale x 8 x i64> %va, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sge_vi_nxv8i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m8, ta, ma
+; CHECK-NEXT:    vmsgt.vi v16, v8, 3, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v16
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x i64> poison, i64 4, i32 0
+  %vb = shufflevector <vscale x 8 x i64> %elt.head, <vscale x 8 x i64> poison, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i64(<vscale x 8 x i64> %va, <vscale x 8 x i64> %vb, metadata !"sge", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_sge_vi_swap_nxv8i64(<vscale x 8 x i64> %va, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sge_vi_swap_nxv8i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m8, ta, ma
+; CHECK-NEXT:    vmsle.vi v16, v8, 4, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v16
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x i64> poison, i64 4, i32 0
+  %vb = shufflevector <vscale x 8 x i64> %elt.head, <vscale x 8 x i64> poison, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i64(<vscale x 8 x i64> %vb, <vscale x 8 x i64> %va, metadata !"sge", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_slt_vv_nxv8i64(<vscale x 8 x i64> %va, <vscale x 8 x i64> %vb, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_slt_vv_nxv8i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m8, ta, ma
+; CHECK-NEXT:    vmslt.vv v24, v8, v16, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v24
+; CHECK-NEXT:    ret
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i64(<vscale x 8 x i64> %va, <vscale x 8 x i64> %vb, metadata !"slt", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_slt_vx_nxv8i64(<vscale x 8 x i64> %va, i64 %b, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; RV32-LABEL: icmp_slt_vx_nxv8i64:
+; RV32:       # %bb.0:
+; RV32-NEXT:    addi sp, sp, -16
+; RV32-NEXT:    .cfi_def_cfa_offset 16
+; RV32-NEXT:    sw a1, 12(sp)
+; RV32-NEXT:    sw a0, 8(sp)
+; RV32-NEXT:    vsetvli a0, zero, e64, m8, ta, mu
+; RV32-NEXT:    addi a0, sp, 8
+; RV32-NEXT:    vlse64.v v24, (a0), zero
+; RV32-NEXT:    vsetvli zero, a2, e64, m8, ta, ma
+; RV32-NEXT:    vmslt.vv v16, v8, v24, v0.t
+; RV32-NEXT:    vmv1r.v v0, v16
+; RV32-NEXT:    addi sp, sp, 16
+; RV32-NEXT:    ret
+;
+; RV64-LABEL: icmp_slt_vx_nxv8i64:
+; RV64:       # %bb.0:
+; RV64-NEXT:    vsetvli zero, a1, e64, m8, ta, ma
+; RV64-NEXT:    vmslt.vx v16, v8, a0, v0.t
+; RV64-NEXT:    vmv1r.v v0, v16
+; RV64-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x i64> poison, i64 %b, i32 0
+  %vb = shufflevector <vscale x 8 x i64> %elt.head, <vscale x 8 x i64> poison, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i64(<vscale x 8 x i64> %va, <vscale x 8 x i64> %vb, metadata !"slt", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_slt_vx_swap_nxv8i64(<vscale x 8 x i64> %va, i64 %b, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; RV32-LABEL: icmp_slt_vx_swap_nxv8i64:
+; RV32:       # %bb.0:
+; RV32-NEXT:    addi sp, sp, -16
+; RV32-NEXT:    .cfi_def_cfa_offset 16
+; RV32-NEXT:    sw a1, 12(sp)
+; RV32-NEXT:    sw a0, 8(sp)
+; RV32-NEXT:    vsetvli a0, zero, e64, m8, ta, mu
+; RV32-NEXT:    addi a0, sp, 8
+; RV32-NEXT:    vlse64.v v24, (a0), zero
+; RV32-NEXT:    vsetvli zero, a2, e64, m8, ta, ma
+; RV32-NEXT:    vmslt.vv v16, v24, v8, v0.t
+; RV32-NEXT:    vmv1r.v v0, v16
+; RV32-NEXT:    addi sp, sp, 16
+; RV32-NEXT:    ret
+;
+; RV64-LABEL: icmp_slt_vx_swap_nxv8i64:
+; RV64:       # %bb.0:
+; RV64-NEXT:    vsetvli zero, a1, e64, m8, ta, ma
+; RV64-NEXT:    vmsgt.vx v16, v8, a0, v0.t
+; RV64-NEXT:    vmv1r.v v0, v16
+; RV64-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x i64> poison, i64 %b, i32 0
+  %vb = shufflevector <vscale x 8 x i64> %elt.head, <vscale x 8 x i64> poison, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i64(<vscale x 8 x i64> %vb, <vscale x 8 x i64> %va, metadata !"slt", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_slt_vi_nxv8i64(<vscale x 8 x i64> %va, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_slt_vi_nxv8i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m8, ta, ma
+; CHECK-NEXT:    vmsle.vi v16, v8, 3, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v16
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x i64> poison, i64 4, i32 0
+  %vb = shufflevector <vscale x 8 x i64> %elt.head, <vscale x 8 x i64> poison, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i64(<vscale x 8 x i64> %va, <vscale x 8 x i64> %vb, metadata !"slt", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_slt_vi_swap_nxv8i64(<vscale x 8 x i64> %va, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_slt_vi_swap_nxv8i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m8, ta, ma
+; CHECK-NEXT:    vmsgt.vi v16, v8, 4, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v16
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x i64> poison, i64 4, i32 0
+  %vb = shufflevector <vscale x 8 x i64> %elt.head, <vscale x 8 x i64> poison, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i64(<vscale x 8 x i64> %vb, <vscale x 8 x i64> %va, metadata !"slt", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_sle_vv_nxv8i64(<vscale x 8 x i64> %va, <vscale x 8 x i64> %vb, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sle_vv_nxv8i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m8, ta, ma
+; CHECK-NEXT:    vmsle.vv v24, v8, v16, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v24
+; CHECK-NEXT:    ret
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i64(<vscale x 8 x i64> %va, <vscale x 8 x i64> %vb, metadata !"sle", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_sle_vx_nxv8i64(<vscale x 8 x i64> %va, i64 %b, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; RV32-LABEL: icmp_sle_vx_nxv8i64:
+; RV32:       # %bb.0:
+; RV32-NEXT:    addi sp, sp, -16
+; RV32-NEXT:    .cfi_def_cfa_offset 16
+; RV32-NEXT:    sw a1, 12(sp)
+; RV32-NEXT:    sw a0, 8(sp)
+; RV32-NEXT:    vsetvli a0, zero, e64, m8, ta, mu
+; RV32-NEXT:    addi a0, sp, 8
+; RV32-NEXT:    vlse64.v v24, (a0), zero
+; RV32-NEXT:    vsetvli zero, a2, e64, m8, ta, ma
+; RV32-NEXT:    vmsle.vv v16, v8, v24, v0.t
+; RV32-NEXT:    vmv1r.v v0, v16
+; RV32-NEXT:    addi sp, sp, 16
+; RV32-NEXT:    ret
+;
+; RV64-LABEL: icmp_sle_vx_nxv8i64:
+; RV64:       # %bb.0:
+; RV64-NEXT:    vsetvli zero, a1, e64, m8, ta, ma
+; RV64-NEXT:    vmsle.vx v16, v8, a0, v0.t
+; RV64-NEXT:    vmv1r.v v0, v16
+; RV64-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x i64> poison, i64 %b, i32 0
+  %vb = shufflevector <vscale x 8 x i64> %elt.head, <vscale x 8 x i64> poison, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i64(<vscale x 8 x i64> %va, <vscale x 8 x i64> %vb, metadata !"sle", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_sle_vx_swap_nxv8i64(<vscale x 8 x i64> %va, i64 %b, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; RV32-LABEL: icmp_sle_vx_swap_nxv8i64:
+; RV32:       # %bb.0:
+; RV32-NEXT:    addi sp, sp, -16
+; RV32-NEXT:    .cfi_def_cfa_offset 16
+; RV32-NEXT:    sw a1, 12(sp)
+; RV32-NEXT:    sw a0, 8(sp)
+; RV32-NEXT:    vsetvli a0, zero, e64, m8, ta, mu
+; RV32-NEXT:    addi a0, sp, 8
+; RV32-NEXT:    vlse64.v v24, (a0), zero
+; RV32-NEXT:    vsetvli zero, a2, e64, m8, ta, ma
+; RV32-NEXT:    vmsle.vv v16, v24, v8, v0.t
+; RV32-NEXT:    vmv1r.v v0, v16
+; RV32-NEXT:    addi sp, sp, 16
+; RV32-NEXT:    ret
+;
+; RV64-LABEL: icmp_sle_vx_swap_nxv8i64:
+; RV64:       # %bb.0:
+; RV64-NEXT:    vsetvli a2, zero, e64, m8, ta, mu
+; RV64-NEXT:    vmv.v.x v24, a0
+; RV64-NEXT:    vsetvli zero, a1, e64, m8, ta, ma
+; RV64-NEXT:    vmsle.vv v16, v24, v8, v0.t
+; RV64-NEXT:    vmv1r.v v0, v16
+; RV64-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x i64> poison, i64 %b, i32 0
+  %vb = shufflevector <vscale x 8 x i64> %elt.head, <vscale x 8 x i64> poison, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i64(<vscale x 8 x i64> %vb, <vscale x 8 x i64> %va, metadata !"sle", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_sle_vi_nxv8i64(<vscale x 8 x i64> %va, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sle_vi_nxv8i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m8, ta, ma
+; CHECK-NEXT:    vmsle.vi v16, v8, 4, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v16
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x i64> poison, i64 4, i32 0
+  %vb = shufflevector <vscale x 8 x i64> %elt.head, <vscale x 8 x i64> poison, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i64(<vscale x 8 x i64> %va, <vscale x 8 x i64> %vb, metadata !"sle", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}
+
+define <vscale x 8 x i1> @icmp_sle_vi_swap_nxv8i64(<vscale x 8 x i64> %va, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: icmp_sle_vi_swap_nxv8i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e64, m8, ta, mu
+; CHECK-NEXT:    vmv.v.i v24, 4
+; CHECK-NEXT:    vsetvli zero, a0, e64, m8, ta, ma
+; CHECK-NEXT:    vmsle.vv v16, v24, v8, v0.t
+; CHECK-NEXT:    vmv1r.v v0, v16
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x i64> poison, i64 4, i32 0
+  %vb = shufflevector <vscale x 8 x i64> %elt.head, <vscale x 8 x i64> poison, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x i1> @llvm.vp.icmp.nxv8i64(<vscale x 8 x i64> %vb, <vscale x 8 x i64> %va, metadata !"sle", <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x i1> %v
+}