[RISCV] Add initial support for 128-bit fixed vectors with RVV.

[craig.topper]

Summary

This adds basic support for 128-bit fixed vectors using a single vector
register as the 1.0 V spec has a minimum VLEN of 128 bits. This
first patch includes support for VV versions of the integer and FP
binary operations and vector loads and stores. This is all off by
default and requires a command line to enable.

Portions of this are inspired by the D91638 proof of concept from
Fraser Cormack.

[craig.topper]

Drop some things that aren't needed yet.

[craig.topper]

Add rv32 command lines

[craig.topper]

Support fneg based on D95568

[HsiangKai]

LGTM. Wait for Fraser Cormack's review.

[frasercrmck]

LGTM in general. I've been wondering if/how we're going to interact with the proposed P extension. Because that will presumably be adding patterns for v2i16, v4i8, etc. Do we need an additional Tablegen predicate to control our fixed-len V patterns?

[craig.topper]

In D95563#2527473, @frasercrmck wrote:

LGTM in general. I've been wondering if/how we're going to interact with the proposed P extension. Because that will presumably be adding patterns for v2i16, v4i8, etc. Do we need an additional Tablegen predicate to control our fixed-len V patterns?

Maybe. Is it likely that both would be used at the same time? And if they are which one should have priority?

[frasercrmck]

In D95563#2527547, @craig.topper wrote:

Maybe. Is it likely that both would be used at the same time? And if they are which one should have priority?

That's what I've been wondering. If it's a possibility we should handle it somewhat gracefully. But I struggle to see us selectively handling certain types with V and others with P, so I reckon we'd just make a top-level all-or-nothing choice, possibly based on a user option. I guess I intuitively think P takes priority since it's specialised for those types?

This isn't something I've solved for myself, but it was making me hesitant to update my fixed-length patches.

[craig.topper]

In D95563#2527569, @frasercrmck wrote:

In D95563#2527547, @craig.topper wrote:

Maybe. Is it likely that both would be used at the same time? And if they are which one should have priority?

That's what I've been wondering. If it's a possibility we should handle it somewhat gracefully. But I struggle to see us selectively handling certain types with V and others with P, so I reckon we'd just make a top-level all-or-nothing choice, possibly based on a user option. I guess I intuitively think P takes priority since it's specialised for those types?

This isn't something I've solved for myself, but it was making me hesitant to update my fixed-length patches.

At least in the td file patterns we shouldn’t have a problem with types overlapping if we use 128 bit for fixed vectors with V. I think we may eventually end up needing to lower fixed vectors to a user controlled vector width. Which likely means we’ll eventually need to lower fixed vectors to scalable vectors with a VL argument earlier than isel patterns. I think AArch64 has some support for something similar.

[khchen]

Had you consider SVE's minimum vector length approach?
https://reviews.llvm.org/D80385
https://github.com/llvm/llvm-project/blob/main/llvm/lib/Target/AArch64/AArch64Subtarget.cpp#L50-L54
maybe giving the variable minimum vector length is more flexible?

[craig.topper]

Abandoning in favor of D95705

llvm/lib/Target/RISCV/RISCVISelLowering.cpp

 //===-- RISCVISelLowering.cpp - RISCV DAG Lowering Implementation  --------===//
 //
 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
 // See https://llvm.org/LICENSE.txt for license information.
 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 //
 //===----------------------------------------------------------------------===//
 //

 #include "llvm/Support/raw_ostream.h"
 
 using namespace llvm;
 
 #define DEBUG_TYPE "riscv-lower"
 
 STATISTIC(NumTailCalls, "Number of tail calls");
 
+cl::opt<bool> UseRVVForFixedVectors(
+    "riscv-use-rvv-for-fixed-vectors", cl::Hidden,
+    cl::desc("Allow RVV instructions to be used for fixed vector types"),
+    cl::init(false));
+
 RISCVTargetLowering::RISCVTargetLowering(const TargetMachine &TM,
                                          const RISCVSubtarget &STI)
     : TargetLowering(TM), Subtarget(STI) {
 
   if (Subtarget.isRV32E())
     report_fatal_error("Codegen not yet implemented for RV32E");
 
   RISCVABI::ABI ABI = Subtarget.getTargetABI();
if (Subtarget.hasStdExtV()) {
 
     if (Subtarget.hasStdExtF())
       for (MVT VT : F32VecVTs)
         addRegClassForRVV(VT);
 
     if (Subtarget.hasStdExtD())
       for (MVT VT : F64VecVTs)
         addRegClassForRVV(VT);
+
+    if (UseRVVForFixedVectors) {
+      addRegisterClass(MVT::v16i8, &RISCV::VRRegClass);
+      addRegisterClass(MVT::v8i16, &RISCV::VRRegClass);
+      addRegisterClass(MVT::v4i32, &RISCV::VRRegClass);
+      addRegisterClass(MVT::v2i64, &RISCV::VRRegClass);
+
+      if (Subtarget.hasStdExtZfh())
+        addRegisterClass(MVT::v8f16, &RISCV::VRRegClass);
+
+      if (Subtarget.hasStdExtF())
+        addRegisterClass(MVT::v4f32, &RISCV::VRRegClass);
+
+      if (Subtarget.hasStdExtD())
+        addRegisterClass(MVT::v2f64, &RISCV::VRRegClass);
+    }
   }
 
   // Compute derived properties from the register classes.
   computeRegisterProperties(STI.getRegisterInfo());
 
   setStackPointerRegisterToSaveRestore(RISCV::X2);
 
   for (auto N : {ISD::EXTLOAD, ISD::SEXTLOAD, ISD::ZEXTLOAD})

llvm/lib/Target/RISCV/RISCVInstrInfoVPseudos.td

VReg RC = !cast<VReg>("VRN" # nf # !cond(!eq(m.value, V_MF8.value): V_M1.MX,
                                            true: m.MX));
 }
 
 //===----------------------------------------------------------------------===//
 // Vector register and vector group type information.
 //===----------------------------------------------------------------------===//
 
 class VTypeInfo<ValueType Vec, ValueType Mas, int Sew, VReg Reg, LMULInfo M,
-                ValueType Scal = XLenVT, RegisterClass ScalarReg = GPR>
+                ValueType Scal = XLenVT, RegisterClass ScalarReg = GPR,
+                int NElts = -1>
 {
   ValueType Vector = Vec;
   ValueType Mask = Mas;
   int SEW = Sew;
   VReg RegClass = Reg;
   LMULInfo LMul = M;
   ValueType Scalar = Scal;
   RegisterClass ScalarRegClass = ScalarReg;
+  int NumElts = NElts;
   // The pattern fragment which produces the AVL operand, representing the
   // "natural" vector length for this type. For scalable vectors this is VLMax.
-  OutPatFrag AVL = VLMax;
+  OutPatFrag AVL = !if(!eq(NumElts, -1),
+                       VLMax, OutPatFrag<(ops), (ADDI X0, NumElts)>);
 
   string ScalarSuffix = !cond(!eq(Scal, XLenVT) : "X",
                               !eq(Scal, f16) : "F16",
                               !eq(Scal, f32) : "F32",
                               !eq(Scal, f64) : "F64");
 }
 
 class GroupVTypeInfo<ValueType Vec, ValueType VecM1, ValueType Mas, int Sew,
defset list<GroupVTypeInfo> GroupFloatVectors = {
       def VF64M4: GroupVTypeInfo<vfloat64m4_t, vfloat64m1_t, vbool16_t, 64,
                                   VRM4, V_M4, f64, FPR64>;
       def VF64M8: GroupVTypeInfo<vfloat64m8_t, vfloat64m1_t, vbool8_t,  64,
                                   VRM8, V_M8, f64, FPR64>;
     }
   }
 }
 
+// Non-scalable vector types which assume VLEN >= 128.
+defset list<VTypeInfo> AllFixedVectors = {
+  defset list<VTypeInfo> FixedIntegerVectors = {
+    def V16I8 : VTypeInfo<v16i8, v16i1, 8, VR, V_M1, XLenVT, GPR, 16>;
+    def V8I16 : VTypeInfo<v8i16, v8i1, 16, VR, V_M1, XLenVT, GPR, 8>;
+    def V4I32 : VTypeInfo<v4i32, v4i1, 32, VR, V_M1, XLenVT, GPR, 4>;
+    def V2I64 : VTypeInfo<v2i64, v2i1, 64, VR, V_M1, XLenVT, GPR, 2>;
+  }
+
+  defset list<VTypeInfo> FixedFloatVectors = {
+    def V8F16 : VTypeInfo<v8f16, v8i1, 16, VR, V_M1, f16, FPR16, 8>;
+    def V4F32 : VTypeInfo<v4f32, v4i1, 32, VR, V_M1, f32, FPR32, 4>;
+    def V2F64 : VTypeInfo<v2f64, v2i1, 64, VR, V_M1, f64, FPR64, 2>;
+  }
+} // AllFixedVectors
+
 // This functor is used to obtain the int vector type that has the same SEW and
 // multiplier as the input parameter type
 class GetIntVTypeInfo<VTypeInfo vti>
 {
   // Equivalent integer vector type. Eg.
   //   VI8M1 → VI8M1 (identity)
   //   VF64M4 → VI64M4
   VTypeInfo Vti = !cast<VTypeInfo>(!subst("VF", "VI", !cast<string>(vti)));

llvm/lib/Target/RISCV/RISCVInstrInfoVSDPatterns.td

multiclass VPatUSLoadStoreSDNode<LLVMType type,
   def : Pat<(type (load reg_rs1:$rs1)),
             (load_instr reg_rs1:$rs1, avl, sew)>;
   // Store
   def : Pat<(store type:$rs2, reg_rs1:$rs1),
             (store_instr reg_class:$rs2, reg_rs1:$rs1, avl, sew)>;
 }
 
 multiclass VPatUSLoadStoreSDNodes<RegisterClass reg_rs1> {
-  foreach vti = AllVectors in
+  foreach vti = !listconcat(AllVectors, AllFixedVectors) in
     defm "" : VPatUSLoadStoreSDNode<vti.Vector, vti.Mask, vti.SEW, vti.LMul,
                                     vti.AVL, reg_rs1, vti.RegClass>;
 }
 
 class VPatBinarySDNode_VV<SDNode vop,
                           string instruction_name,
                           ValueType result_type,
                           ValueType op_type,
Pat<(result_type (vop
                      (vop_type (SplatPatKind xop_kind:$rs2)))),
         (!cast<Instruction>(instruction_name#_#suffix#_# vlmul.MX)
                      vop_reg_class:$rs1,
                      xop_kind:$rs2,
                      avl, sew)>;
 
 multiclass VPatBinarySDNode_VV_VX<SDNode vop, string instruction_name>
 {
-  foreach vti = AllIntegerVectors in {
+  foreach vti = !listconcat(AllIntegerVectors, FixedIntegerVectors) in {
     def : VPatBinarySDNode_VV<vop, instruction_name,
                               vti.Vector, vti.Vector, vti.Mask, vti.SEW,
                               vti.LMul, vti.AVL, vti.RegClass, vti.RegClass>;
+  }
+  foreach vti = AllIntegerVectors in {
     def : VPatBinarySDNode_XI<vop, instruction_name, "VX",
                               vti.Vector, vti.Vector, XLenVT, vti.Mask, vti.SEW,
                               vti.LMul, vti.AVL, vti.RegClass, vti.RegClass,
                               SplatPat, GPR>;
   }
 }
 
 multiclass VPatBinarySDNode_VV_VX_VI<SDNode vop, string instruction_name,
                                      Operand ImmType = simm5>
 {
-  foreach vti = AllIntegerVectors in {
+  foreach vti = !listconcat(AllIntegerVectors, FixedIntegerVectors) in {
     def : VPatBinarySDNode_VV<vop, instruction_name,
                               vti.Vector, vti.Vector, vti.Mask, vti.SEW,
                               vti.LMul, vti.AVL, vti.RegClass, vti.RegClass>;
+  }
+  foreach vti = AllIntegerVectors in {
     def : VPatBinarySDNode_XI<vop, instruction_name, "VX",
                               vti.Vector, vti.Vector, XLenVT, vti.Mask, vti.SEW,
                               vti.LMul, vti.AVL, vti.RegClass, vti.RegClass,
                               SplatPat, GPR>;
     def : VPatBinarySDNode_XI<vop, instruction_name, "VI",
                               vti.Vector, vti.Vector, XLenVT, vti.Mask, vti.SEW,
                               vti.LMul, vti.AVL, vti.RegClass, vti.RegClass,
                               !cast<ComplexPattern>(SplatPat#_#ImmType),
class VPatBinarySDNode_VF<SDNode vop,
     Pat<(result_type (vop (vop_type vop_reg_class:$rs1),
                           (vop_type (splat_vector xop_kind:$rs2)))),
         (!cast<Instruction>(instruction_name#"_"#vlmul.MX)
                      vop_reg_class:$rs1,
                      (xop_type xop_kind:$rs2),
                      avl, sew)>;
 
 multiclass VPatBinaryFPSDNode_VV_VF<SDNode vop, string instruction_name> {
-  foreach vti = AllFloatVectors in {
+  foreach vti = !listconcat(AllFloatVectors, FixedFloatVectors) in {
     def : VPatBinarySDNode_VV<vop, instruction_name,
                               vti.Vector, vti.Vector, vti.Mask, vti.SEW,
                               vti.LMul, vti.AVL, vti.RegClass, vti.RegClass>;
+  }
+  foreach vti = AllFloatVectors in {
     def : VPatBinarySDNode_VF<vop, instruction_name#"_V"#vti.ScalarSuffix,
                               vti.Vector, vti.Vector, vti.Scalar, vti.Mask,
                               vti.SEW, vti.LMul, vti.AVL, vti.RegClass, vti.RegClass,
                               vti.ScalarRegClass>;
   }
 }
 
 multiclass VPatBinaryFPSDNode_R_VF<SDNode vop, string instruction_name> {

 
 // 14.4. Vector Single-Width Floating-Point Multiply/Divide Instructions
 defm "" : VPatBinaryFPSDNode_VV_VF<fmul, "PseudoVFMUL">;
 defm "" : VPatBinaryFPSDNode_VV_VF<fdiv, "PseudoVFDIV">;
 defm "" : VPatBinaryFPSDNode_R_VF<fdiv, "PseudoVFRDIV">;
 
 // 14.10. Vector Floating-Point Sign-Injection Instructions
 // Handle fneg with VFSGNJN using the same input for both operands.
-foreach vti = AllFloatVectors in {
+foreach vti = !listconcat(AllFloatVectors, FixedFloatVectors) in {
   def : Pat<(fneg (vti.Vector vti.RegClass:$rs)),
             (!cast<Instruction>("PseudoVFSGNJN_VV_"# vti.LMul.MX)
                  vti.RegClass:$rs, vti.RegClass:$rs, vti.AVL, vti.SEW)>;
 }
 
 // 14.11. Vector Floating-Point Compare Instructions
 defm "" : VPatFPSetCCSDNode_VV_VF_FV<SETEQ,  "PseudoVMFEQ", "PseudoVMFEQ">;
 defm "" : VPatFPSetCCSDNode_VV_VF_FV<SETOEQ, "PseudoVMFEQ", "PseudoVMFEQ">;

llvm/lib/Target/RISCV/RISCVRegisterInfo.td

 }
 
 def VR : VReg<[vint8mf2_t, vint8mf4_t, vint8mf8_t,
                vint16mf2_t, vint16mf4_t, vint32mf2_t,
                vint8m1_t, vint16m1_t, vint32m1_t, vint64m1_t,
                vfloat16mf4_t, vfloat16mf2_t, vfloat16m1_t,
                vfloat32mf2_t, vfloat32m1_t, vfloat64m1_t,
                vbool64_t, vbool32_t, vbool16_t, vbool8_t, vbool4_t,
-               vbool2_t, vbool1_t],
+               vbool2_t, vbool1_t, v16i8, v8i16, v4i32, v2i64, v8f16, v4f32,
+               v2f64, v1i1, v2i1, v4i1, v8i1, v16i1],
            (add (sequence "V%u", 25, 31),
                 (sequence "V%u", 8, 24),
                 (sequence "V%u", 0, 7)), 1>;
 
 def VRNoV0 : VReg<[vint8mf2_t, vint8mf4_t, vint8mf8_t,
                    vint16mf2_t, vint16mf4_t, vint32mf2_t,
                    vint8m1_t, vint16m1_t, vint32m1_t, vint64m1_t,
                    vfloat16mf4_t, vfloat16mf2_t, vfloat16m1_t,
def VRM8 : VReg<[vint8m8_t, vint16m8_t, vint32m8_t, vint64m8_t,
                  vfloat16m8_t, vfloat32m8_t, vfloat64m8_t],
              (add V8M8, V16M8, V24M8, V0M8), 8>;
 
 def VRM8NoV0 : VReg<[vint8m8_t, vint16m8_t, vint32m8_t, vint64m8_t,
                      vfloat16m8_t, vfloat32m8_t, vfloat64m8_t],
                  (add V8M8, V16M8, V24M8), 8>;
 
 defvar VMaskVTs = [vbool64_t, vbool32_t, vbool16_t, vbool8_t,
-                   vbool4_t, vbool2_t, vbool1_t];
+                   vbool4_t, vbool2_t, vbool1_t, v1i1, v2i1, v4i1, v8i1, v16i1];
 
 def VMV0 : RegisterClass<"RISCV", VMaskVTs, 64, (add V0)> {
   let Size = 64;
 }
 
 foreach m = LMULList.m in {
   foreach nf = NFList<m>.L in {
     def "VRN" # nf # "M" # m : VReg<[untyped],
                                (add !cast<RegisterTuples>("VN" # nf # "M" # m)),
                                     !mul(nf, m)>;
   }
 }

llvm/test/CodeGen/RISCV/rvv/fixed-vectors-fp.ll

+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc -mtriple=riscv32 -mattr=+experimental-v,+experimental-Zfh,+f,+d -verify-machineinstrs -riscv-use-rvv-for-fixed-vectors < %s | FileCheck %s
+; RUN: llc -mtriple=riscv64 -mattr=+experimental-v,+experimental-Zfh,+f,+d -verify-machineinstrs -riscv-use-rvv-for-fixed-vectors < %s | FileCheck %s
+
+define void @fadd_v8f16(<8 x half>* %x, <8 x half>* %y) {
+; CHECK-LABEL: fadd_v8f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    addi a2, zero, 8
+; CHECK-NEXT:    vsetvli a2, a2, e16,m1,ta,mu
+; CHECK-NEXT:    vle16.v v25, (a0)
+; CHECK-NEXT:    vle16.v v26, (a1)
+; CHECK-NEXT:    vfadd.vv v25, v25, v26
+; CHECK-NEXT:    vse16.v v25, (a0)
+; CHECK-NEXT:    ret
+  %a = load <8 x half>, <8 x half>* %x
+  %b = load <8 x half>, <8 x half>* %y
+  %c = fadd <8 x half> %a, %b
+  store <8 x half> %c, <8 x half>* %x
+  ret void
+}
+
+define void @fadd_v4f32(<4 x float>* %x, <4 x float>* %y) {
+; CHECK-LABEL: fadd_v4f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    addi a2, zero, 4
+; CHECK-NEXT:    vsetvli a2, a2, e32,m1,ta,mu
+; CHECK-NEXT:    vle32.v v25, (a0)
+; CHECK-NEXT:    vle32.v v26, (a1)
+; CHECK-NEXT:    vfadd.vv v25, v25, v26
+; CHECK-NEXT:    vse32.v v25, (a0)
+; CHECK-NEXT:    ret
+  %a = load <4 x float>, <4 x float>* %x
+  %b = load <4 x float>, <4 x float>* %y
+  %c = fadd <4 x float> %a, %b
+  store <4 x float> %c, <4 x float>* %x
+  ret void
+}
+
+define void @fadd_v2f64(<2 x double>* %x, <2 x double>* %y) {
+; CHECK-LABEL: fadd_v2f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    addi a2, zero, 2
+; CHECK-NEXT:    vsetvli a2, a2, e64,m1,ta,mu
+; CHECK-NEXT:    vle64.v v25, (a0)
+; CHECK-NEXT:    vle64.v v26, (a1)
+; CHECK-NEXT:    vfadd.vv v25, v25, v26
+; CHECK-NEXT:    vse64.v v25, (a0)
+; CHECK-NEXT:    ret
+  %a = load <2 x double>, <2 x double>* %x
+  %b = load <2 x double>, <2 x double>* %y
+  %c = fadd <2 x double> %a, %b
+  store <2 x double> %c, <2 x double>* %x
+  ret void
+}
+
+define void @fsub_v8f16(<8 x half>* %x, <8 x half>* %y) {
+; CHECK-LABEL: fsub_v8f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    addi a2, zero, 8
+; CHECK-NEXT:    vsetvli a2, a2, e16,m1,ta,mu
+; CHECK-NEXT:    vle16.v v25, (a0)
+; CHECK-NEXT:    vle16.v v26, (a1)
+; CHECK-NEXT:    vfsub.vv v25, v25, v26
+; CHECK-NEXT:    vse16.v v25, (a0)
+; CHECK-NEXT:    ret
+  %a = load <8 x half>, <8 x half>* %x
+  %b = load <8 x half>, <8 x half>* %y
+  %c = fsub <8 x half> %a, %b
+  store <8 x half> %c, <8 x half>* %x
+  ret void
+}
+
+define void @fsub_v4f32(<4 x float>* %x, <4 x float>* %y) {
+; CHECK-LABEL: fsub_v4f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    addi a2, zero, 4
+; CHECK-NEXT:    vsetvli a2, a2, e32,m1,ta,mu
+; CHECK-NEXT:    vle32.v v25, (a0)
+; CHECK-NEXT:    vle32.v v26, (a1)
+; CHECK-NEXT:    vfsub.vv v25, v25, v26
+; CHECK-NEXT:    vse32.v v25, (a0)
+; CHECK-NEXT:    ret
+  %a = load <4 x float>, <4 x float>* %x
+  %b = load <4 x float>, <4 x float>* %y
+  %c = fsub <4 x float> %a, %b
+  store <4 x float> %c, <4 x float>* %x
+  ret void
+}
+
+define void @fsub_v2f64(<2 x double>* %x, <2 x double>* %y) {
+; CHECK-LABEL: fsub_v2f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    addi a2, zero, 2
+; CHECK-NEXT:    vsetvli a2, a2, e64,m1,ta,mu
+; CHECK-NEXT:    vle64.v v25, (a0)
+; CHECK-NEXT:    vle64.v v26, (a1)
+; CHECK-NEXT:    vfsub.vv v25, v25, v26
+; CHECK-NEXT:    vse64.v v25, (a0)
+; CHECK-NEXT:    ret
+  %a = load <2 x double>, <2 x double>* %x
+  %b = load <2 x double>, <2 x double>* %y
+  %c = fsub <2 x double> %a, %b
+  store <2 x double> %c, <2 x double>* %x
+  ret void
+}
+
+define void @fmul_v8f16(<8 x half>* %x, <8 x half>* %y) {
+; CHECK-LABEL: fmul_v8f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    addi a2, zero, 8
+; CHECK-NEXT:    vsetvli a2, a2, e16,m1,ta,mu
+; CHECK-NEXT:    vle16.v v25, (a0)
+; CHECK-NEXT:    vle16.v v26, (a1)
+; CHECK-NEXT:    vfmul.vv v25, v25, v26
+; CHECK-NEXT:    vse16.v v25, (a0)
+; CHECK-NEXT:    ret
+  %a = load <8 x half>, <8 x half>* %x
+  %b = load <8 x half>, <8 x half>* %y
+  %c = fmul <8 x half> %a, %b
+  store <8 x half> %c, <8 x half>* %x
+  ret void
+}
+
+define void @fmul_v4f32(<4 x float>* %x, <4 x float>* %y) {
+; CHECK-LABEL: fmul_v4f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    addi a2, zero, 4
+; CHECK-NEXT:    vsetvli a2, a2, e32,m1,ta,mu
+; CHECK-NEXT:    vle32.v v25, (a0)
+; CHECK-NEXT:    vle32.v v26, (a1)
+; CHECK-NEXT:    vfmul.vv v25, v25, v26
+; CHECK-NEXT:    vse32.v v25, (a0)
+; CHECK-NEXT:    ret
+  %a = load <4 x float>, <4 x float>* %x
+  %b = load <4 x float>, <4 x float>* %y
+  %c = fmul <4 x float> %a, %b
+  store <4 x float> %c, <4 x float>* %x
+  ret void
+}
+
+define void @fmul_v2f64(<2 x double>* %x, <2 x double>* %y) {
+; CHECK-LABEL: fmul_v2f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    addi a2, zero, 2
+; CHECK-NEXT:    vsetvli a2, a2, e64,m1,ta,mu
+; CHECK-NEXT:    vle64.v v25, (a0)
+; CHECK-NEXT:    vle64.v v26, (a1)
+; CHECK-NEXT:    vfmul.vv v25, v25, v26
+; CHECK-NEXT:    vse64.v v25, (a0)
+; CHECK-NEXT:    ret
+  %a = load <2 x double>, <2 x double>* %x
+  %b = load <2 x double>, <2 x double>* %y
+  %c = fmul <2 x double> %a, %b
+  store <2 x double> %c, <2 x double>* %x
+  ret void
+}
+
+define void @fdiv_v8f16(<8 x half>* %x, <8 x half>* %y) {
+; CHECK-LABEL: fdiv_v8f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    addi a2, zero, 8
+; CHECK-NEXT:    vsetvli a2, a2, e16,m1,ta,mu
+; CHECK-NEXT:    vle16.v v25, (a0)
+; CHECK-NEXT:    vle16.v v26, (a1)
+; CHECK-NEXT:    vfdiv.vv v25, v25, v26
+; CHECK-NEXT:    vse16.v v25, (a0)
+; CHECK-NEXT:    ret
+  %a = load <8 x half>, <8 x half>* %x
+  %b = load <8 x half>, <8 x half>* %y
+  %c = fdiv <8 x half> %a, %b
+  store <8 x half> %c, <8 x half>* %x
+  ret void
+}
+
+define void @fdiv_v4f32(<4 x float>* %x, <4 x float>* %y) {
+; CHECK-LABEL: fdiv_v4f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    addi a2, zero, 4
+; CHECK-NEXT:    vsetvli a2, a2, e32,m1,ta,mu
+; CHECK-NEXT:    vle32.v v25, (a0)
+; CHECK-NEXT:    vle32.v v26, (a1)
+; CHECK-NEXT:    vfdiv.vv v25, v25, v26
+; CHECK-NEXT:    vse32.v v25, (a0)
+; CHECK-NEXT:    ret
+  %a = load <4 x float>, <4 x float>* %x
+  %b = load <4 x float>, <4 x float>* %y
+  %c = fdiv <4 x float> %a, %b
+  store <4 x float> %c, <4 x float>* %x
+  ret void
+}
+
+define void @fdiv_v2f64(<2 x double>* %x, <2 x double>* %y) {
+; CHECK-LABEL: fdiv_v2f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    addi a2, zero, 2
+; CHECK-NEXT:    vsetvli a2, a2, e64,m1,ta,mu
+; CHECK-NEXT:    vle64.v v25, (a0)
+; CHECK-NEXT:    vle64.v v26, (a1)
+; CHECK-NEXT:    vfdiv.vv v25, v25, v26
+; CHECK-NEXT:    vse64.v v25, (a0)
+; CHECK-NEXT:    ret
+  %a = load <2 x double>, <2 x double>* %x
+  %b = load <2 x double>, <2 x double>* %y
+  %c = fdiv <2 x double> %a, %b
+  store <2 x double> %c, <2 x double>* %x
+  ret void
+}
+
+define void @fneg_v8f16(<8 x half>* %x) {
+; CHECK-LABEL: fneg_v8f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    addi a1, zero, 8
+; CHECK-NEXT:    vsetvli a1, a1, e16,m1,ta,mu
+; CHECK-NEXT:    vle16.v v25, (a0)
+; CHECK-NEXT:    vfsgnjn.vv v25, v25, v25
+; CHECK-NEXT:    vse16.v v25, (a0)
+; CHECK-NEXT:    ret
+  %a = load <8 x half>, <8 x half>* %x
+  %b = fneg <8 x half> %a
+  store <8 x half> %b, <8 x half>* %x
+  ret void
+}
+
+define void @fneg_v4f32(<4 x float>* %x) {
+; CHECK-LABEL: fneg_v4f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    addi a1, zero, 4
+; CHECK-NEXT:    vsetvli a1, a1, e32,m1,ta,mu
+; CHECK-NEXT:    vle32.v v25, (a0)
+; CHECK-NEXT:    vfsgnjn.vv v25, v25, v25
+; CHECK-NEXT:    vse32.v v25, (a0)
+; CHECK-NEXT:    ret
+  %a = load <4 x float>, <4 x float>* %x
+  %b = fneg <4 x float> %a
+  store <4 x float> %b, <4 x float>* %x
+  ret void
+}
+
+define void @fneg_v2f64(<2 x double>* %x) {
+; CHECK-LABEL: fneg_v2f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    addi a1, zero, 2
+; CHECK-NEXT:    vsetvli a1, a1, e64,m1,ta,mu
+; CHECK-NEXT:    vle64.v v25, (a0)
+; CHECK-NEXT:    vfsgnjn.vv v25, v25, v25
+; CHECK-NEXT:    vse64.v v25, (a0)
+; CHECK-NEXT:    ret
+  %a = load <2 x double>, <2 x double>* %x
+  %b = fneg <2 x double> %a
+  store <2 x double> %b, <2 x double>* %x
+  ret void
+}

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

+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc -mtriple=riscv32 -mattr=+experimental-v -verify-machineinstrs -riscv-use-rvv-for-fixed-vectors < %s | FileCheck %s
+; RUN: llc -mtriple=riscv64 -mattr=+experimental-v -verify-machineinstrs -riscv-use-rvv-for-fixed-vectors < %s | FileCheck %s
+
+define void @add_v16i8(<16 x i8>* %x, <16 x i8>* %y) {
+; CHECK-LABEL: add_v16i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    addi a2, zero, 16
+; CHECK-NEXT:    vsetvli a2, a2, e8,m1,ta,mu
+; CHECK-NEXT:    vle8.v v25, (a0)
+; CHECK-NEXT:    vle8.v v26, (a1)
+; CHECK-NEXT:    vadd.vv v25, v25, v26
+; CHECK-NEXT:    vse8.v v25, (a0)
+; CHECK-NEXT:    ret
+  %a = load <16 x i8>, <16 x i8>* %x
+  %b = load <16 x i8>, <16 x i8>* %y
+  %c = add <16 x i8> %a, %b
+  store <16 x i8> %c, <16 x i8>* %x
+  ret void
+}
+
+define void @add_v8i16(<8 x i16>* %x, <8 x i16>* %y) {
+; CHECK-LABEL: add_v8i16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    addi a2, zero, 8
+; CHECK-NEXT:    vsetvli a2, a2, e16,m1,ta,mu
+; CHECK-NEXT:    vle16.v v25, (a0)
+; CHECK-NEXT:    vle16.v v26, (a1)
+; CHECK-NEXT:    vadd.vv v25, v25, v26
+; CHECK-NEXT:    vse16.v v25, (a0)
+; CHECK-NEXT:    ret
+  %a = load <8 x i16>, <8 x i16>* %x
+  %b = load <8 x i16>, <8 x i16>* %y
+  %c = add <8 x i16> %a, %b
+  store <8 x i16> %c, <8 x i16>* %x
+  ret void
+}
+
+define void @add_v4i32(<4 x i32>* %x, <4 x i32>* %y) {
+; CHECK-LABEL: add_v4i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    addi a2, zero, 4
+; CHECK-NEXT:    vsetvli a2, a2, e32,m1,ta,mu
+; CHECK-NEXT:    vle32.v v25, (a0)
+; CHECK-NEXT:    vle32.v v26, (a1)
+; CHECK-NEXT:    vadd.vv v25, v25, v26
+; CHECK-NEXT:    vse32.v v25, (a0)
+; CHECK-NEXT:    ret
+  %a = load <4 x i32>, <4 x i32>* %x
+  %b = load <4 x i32>, <4 x i32>* %y
+  %c = add <4 x i32> %a, %b
+  store <4 x i32> %c, <4 x i32>* %x
+  ret void
+}
+
+define void @add_v2i64(<2 x i64>* %x, <2 x i64>* %y) {
+; CHECK-LABEL: add_v2i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    addi a2, zero, 2
+; CHECK-NEXT:    vsetvli a2, a2, e64,m1,ta,mu
+; CHECK-NEXT:    vle64.v v25, (a0)
+; CHECK-NEXT:    vle64.v v26, (a1)
+; CHECK-NEXT:    vadd.vv v25, v25, v26
+; CHECK-NEXT:    vse64.v v25, (a0)
+; CHECK-NEXT:    ret
+  %a = load <2 x i64>, <2 x i64>* %x
+  %b = load <2 x i64>, <2 x i64>* %y
+  %c = add <2 x i64> %a, %b
+  store <2 x i64> %c, <2 x i64>* %x
+  ret void
+}
+
+define void @sub_v16i8(<16 x i8>* %x, <16 x i8>* %y) {
+; CHECK-LABEL: sub_v16i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    addi a2, zero, 16
+; CHECK-NEXT:    vsetvli a2, a2, e8,m1,ta,mu
+; CHECK-NEXT:    vle8.v v25, (a0)
+; CHECK-NEXT:    vle8.v v26, (a1)
+; CHECK-NEXT:    vsub.vv v25, v25, v26
+; CHECK-NEXT:    vse8.v v25, (a0)
+; CHECK-NEXT:    ret
+  %a = load <16 x i8>, <16 x i8>* %x
+  %b = load <16 x i8>, <16 x i8>* %y
+  %c = sub <16 x i8> %a, %b
+  store <16 x i8> %c, <16 x i8>* %x
+  ret void
+}
+
+define void @sub_v8i16(<8 x i16>* %x, <8 x i16>* %y) {
+; CHECK-LABEL: sub_v8i16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    addi a2, zero, 8
+; CHECK-NEXT:    vsetvli a2, a2, e16,m1,ta,mu
+; CHECK-NEXT:    vle16.v v25, (a0)
+; CHECK-NEXT:    vle16.v v26, (a1)
+; CHECK-NEXT:    vsub.vv v25, v25, v26
+; CHECK-NEXT:    vse16.v v25, (a0)
+; CHECK-NEXT:    ret
+  %a = load <8 x i16>, <8 x i16>* %x
+  %b = load <8 x i16>, <8 x i16>* %y
+  %c = sub <8 x i16> %a, %b
+  store <8 x i16> %c, <8 x i16>* %x
+  ret void
+}
+
+define void @sub_v4i32(<4 x i32>* %x, <4 x i32>* %y) {
+; CHECK-LABEL: sub_v4i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    addi a2, zero, 4
+; CHECK-NEXT:    vsetvli a2, a2, e32,m1,ta,mu
+; CHECK-NEXT:    vle32.v v25, (a0)
+; CHECK-NEXT:    vle32.v v26, (a1)
+; CHECK-NEXT:    vsub.vv v25, v25, v26
+; CHECK-NEXT:    vse32.v v25, (a0)
+; CHECK-NEXT:    ret
+  %a = load <4 x i32>, <4 x i32>* %x
+  %b = load <4 x i32>, <4 x i32>* %y
+  %c = sub <4 x i32> %a, %b
+  store <4 x i32> %c, <4 x i32>* %x
+  ret void
+}
+
+define void @sub_v2i64(<2 x i64>* %x, <2 x i64>* %y) {
+; CHECK-LABEL: sub_v2i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    addi a2, zero, 2
+; CHECK-NEXT:    vsetvli a2, a2, e64,m1,ta,mu
+; CHECK-NEXT:    vle64.v v25, (a0)
+; CHECK-NEXT:    vle64.v v26, (a1)
+; CHECK-NEXT:    vsub.vv v25, v25, v26
+; CHECK-NEXT:    vse64.v v25, (a0)
+; CHECK-NEXT:    ret
+  %a = load <2 x i64>, <2 x i64>* %x
+  %b = load <2 x i64>, <2 x i64>* %y
+  %c = sub <2 x i64> %a, %b
+  store <2 x i64> %c, <2 x i64>* %x
+  ret void
+}
+
+define void @mul_v16i8(<16 x i8>* %x, <16 x i8>* %y) {
+; CHECK-LABEL: mul_v16i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    addi a2, zero, 16
+; CHECK-NEXT:    vsetvli a2, a2, e8,m1,ta,mu
+; CHECK-NEXT:    vle8.v v25, (a0)
+; CHECK-NEXT:    vle8.v v26, (a1)
+; CHECK-NEXT:    vmul.vv v25, v25, v26
+; CHECK-NEXT:    vse8.v v25, (a0)
+; CHECK-NEXT:    ret
+  %a = load <16 x i8>, <16 x i8>* %x
+  %b = load <16 x i8>, <16 x i8>* %y
+  %c = mul <16 x i8> %a, %b
+  store <16 x i8> %c, <16 x i8>* %x
+  ret void
+}
+
+define void @mul_v8i16(<8 x i16>* %x, <8 x i16>* %y) {
+; CHECK-LABEL: mul_v8i16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    addi a2, zero, 8
+; CHECK-NEXT:    vsetvli a2, a2, e16,m1,ta,mu
+; CHECK-NEXT:    vle16.v v25, (a0)
+; CHECK-NEXT:    vle16.v v26, (a1)
+; CHECK-NEXT:    vmul.vv v25, v25, v26
+; CHECK-NEXT:    vse16.v v25, (a0)
+; CHECK-NEXT:    ret
+  %a = load <8 x i16>, <8 x i16>* %x
+  %b = load <8 x i16>, <8 x i16>* %y
+  %c = mul <8 x i16> %a, %b
+  store <8 x i16> %c, <8 x i16>* %x
+  ret void
+}
+
+define void @mul_v4i32(<4 x i32>* %x, <4 x i32>* %y) {
+; CHECK-LABEL: mul_v4i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    addi a2, zero, 4
+; CHECK-NEXT:    vsetvli a2, a2, e32,m1,ta,mu
+; CHECK-NEXT:    vle32.v v25, (a0)
+; CHECK-NEXT:    vle32.v v26, (a1)
+; CHECK-NEXT:    vmul.vv v25, v25, v26
+; CHECK-NEXT:    vse32.v v25, (a0)
+; CHECK-NEXT:    ret
+  %a = load <4 x i32>, <4 x i32>* %x
+  %b = load <4 x i32>, <4 x i32>* %y
+  %c = mul <4 x i32> %a, %b
+  store <4 x i32> %c, <4 x i32>* %x
+  ret void
+}
+
+define void @mul_v2i64(<2 x i64>* %x, <2 x i64>* %y) {
+; CHECK-LABEL: mul_v2i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    addi a2, zero, 2
+; CHECK-NEXT:    vsetvli a2, a2, e64,m1,ta,mu
+; CHECK-NEXT:    vle64.v v25, (a0)
+; CHECK-NEXT:    vle64.v v26, (a1)
+; CHECK-NEXT:    vmul.vv v25, v25, v26
+; CHECK-NEXT:    vse64.v v25, (a0)
+; CHECK-NEXT:    ret
+  %a = load <2 x i64>, <2 x i64>* %x
+  %b = load <2 x i64>, <2 x i64>* %y
+  %c = mul <2 x i64> %a, %b
+  store <2 x i64> %c, <2 x i64>* %x
+  ret void
+}
+
+define void @and_v16i8(<16 x i8>* %x, <16 x i8>* %y) {
+; CHECK-LABEL: and_v16i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    addi a2, zero, 16
+; CHECK-NEXT:    vsetvli a2, a2, e8,m1,ta,mu
+; CHECK-NEXT:    vle8.v v25, (a0)
+; CHECK-NEXT:    vle8.v v26, (a1)
+; CHECK-NEXT:    vand.vv v25, v25, v26
+; CHECK-NEXT:    vse8.v v25, (a0)
+; CHECK-NEXT:    ret
+  %a = load <16 x i8>, <16 x i8>* %x
+  %b = load <16 x i8>, <16 x i8>* %y
+  %c = and <16 x i8> %a, %b
+  store <16 x i8> %c, <16 x i8>* %x
+  ret void
+}
+
+define void @and_v8i16(<8 x i16>* %x, <8 x i16>* %y) {
+; CHECK-LABEL: and_v8i16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    addi a2, zero, 8
+; CHECK-NEXT:    vsetvli a2, a2, e16,m1,ta,mu
+; CHECK-NEXT:    vle16.v v25, (a0)
+; CHECK-NEXT:    vle16.v v26, (a1)
+; CHECK-NEXT:    vand.vv v25, v25, v26
+; CHECK-NEXT:    vse16.v v25, (a0)
+; CHECK-NEXT:    ret
+  %a = load <8 x i16>, <8 x i16>* %x
+  %b = load <8 x i16>, <8 x i16>* %y
+  %c = and <8 x i16> %a, %b
+  store <8 x i16> %c, <8 x i16>* %x
+  ret void
+}
+
+define void @and_v4i32(<4 x i32>* %x, <4 x i32>* %y) {
+; CHECK-LABEL: and_v4i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    addi a2, zero, 4
+; CHECK-NEXT:    vsetvli a2, a2, e32,m1,ta,mu
+; CHECK-NEXT:    vle32.v v25, (a0)
+; CHECK-NEXT:    vle32.v v26, (a1)
+; CHECK-NEXT:    vand.vv v25, v25, v26
+; CHECK-NEXT:    vse32.v v25, (a0)
+; CHECK-NEXT:    ret
+  %a = load <4 x i32>, <4 x i32>* %x
+  %b = load <4 x i32>, <4 x i32>* %y
+  %c = and <4 x i32> %a, %b
+  store <4 x i32> %c, <4 x i32>* %x
+  ret void
+}
+
+define void @and_v2i64(<2 x i64>* %x, <2 x i64>* %y) {
+; CHECK-LABEL: and_v2i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    addi a2, zero, 2
+; CHECK-NEXT:    vsetvli a2, a2, e64,m1,ta,mu
+; CHECK-NEXT:    vle64.v v25, (a0)
+; CHECK-NEXT:    vle64.v v26, (a1)
+; CHECK-NEXT:    vand.vv v25, v25, v26
+; CHECK-NEXT:    vse64.v v25, (a0)
+; CHECK-NEXT:    ret
+  %a = load <2 x i64>, <2 x i64>* %x
+  %b = load <2 x i64>, <2 x i64>* %y
+  %c = and <2 x i64> %a, %b
+  store <2 x i64> %c, <2 x i64>* %x
+  ret void
+}
+
+define void @or_v16i8(<16 x i8>* %x, <16 x i8>* %y) {
+; CHECK-LABEL: or_v16i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    addi a2, zero, 16
+; CHECK-NEXT:    vsetvli a2, a2, e8,m1,ta,mu
+; CHECK-NEXT:    vle8.v v25, (a0)
+; CHECK-NEXT:    vle8.v v26, (a1)
+; CHECK-NEXT:    vor.vv v25, v25, v26
+; CHECK-NEXT:    vse8.v v25, (a0)
+; CHECK-NEXT:    ret
+  %a = load <16 x i8>, <16 x i8>* %x
+  %b = load <16 x i8>, <16 x i8>* %y
+  %c = or <16 x i8> %a, %b
+  store <16 x i8> %c, <16 x i8>* %x
+  ret void
+}
+
+define void @or_v8i16(<8 x i16>* %x, <8 x i16>* %y) {
+; CHECK-LABEL: or_v8i16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    addi a2, zero, 8
+; CHECK-NEXT:    vsetvli a2, a2, e16,m1,ta,mu
+; CHECK-NEXT:    vle16.v v25, (a0)
+; CHECK-NEXT:    vle16.v v26, (a1)
+; CHECK-NEXT:    vor.vv v25, v25, v26
+; CHECK-NEXT:    vse16.v v25, (a0)
+; CHECK-NEXT:    ret
+  %a = load <8 x i16>, <8 x i16>* %x
+  %b = load <8 x i16>, <8 x i16>* %y
+  %c = or <8 x i16> %a, %b
+  store <8 x i16> %c, <8 x i16>* %x
+  ret void
+}
+
+define void @or_v4i32(<4 x i32>* %x, <4 x i32>* %y) {
+; CHECK-LABEL: or_v4i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    addi a2, zero, 4
+; CHECK-NEXT:    vsetvli a2, a2, e32,m1,ta,mu
+; CHECK-NEXT:    vle32.v v25, (a0)
+; CHECK-NEXT:    vle32.v v26, (a1)
+; CHECK-NEXT:    vor.vv v25, v25, v26
+; CHECK-NEXT:    vse32.v v25, (a0)
+; CHECK-NEXT:    ret
+  %a = load <4 x i32>, <4 x i32>* %x
+  %b = load <4 x i32>, <4 x i32>* %y
+  %c = or <4 x i32> %a, %b
+  store <4 x i32> %c, <4 x i32>* %x
+  ret void
+}
+
+define void @or_v2i64(<2 x i64>* %x, <2 x i64>* %y) {
+; CHECK-LABEL: or_v2i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    addi a2, zero, 2
+; CHECK-NEXT:    vsetvli a2, a2, e64,m1,ta,mu
+; CHECK-NEXT:    vle64.v v25, (a0)
+; CHECK-NEXT:    vle64.v v26, (a1)
+; CHECK-NEXT:    vor.vv v25, v25, v26
+; CHECK-NEXT:    vse64.v v25, (a0)
+; CHECK-NEXT:    ret
+  %a = load <2 x i64>, <2 x i64>* %x
+  %b = load <2 x i64>, <2 x i64>* %y
+  %c = or <2 x i64> %a, %b
+  store <2 x i64> %c, <2 x i64>* %x
+  ret void
+}
+
+define void @xor_v16i8(<16 x i8>* %x, <16 x i8>* %y) {
+; CHECK-LABEL: xor_v16i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    addi a2, zero, 16
+; CHECK-NEXT:    vsetvli a2, a2, e8,m1,ta,mu
+; CHECK-NEXT:    vle8.v v25, (a0)
+; CHECK-NEXT:    vle8.v v26, (a1)
+; CHECK-NEXT:    vxor.vv v25, v25, v26
+; CHECK-NEXT:    vse8.v v25, (a0)
+; CHECK-NEXT:    ret
+  %a = load <16 x i8>, <16 x i8>* %x
+  %b = load <16 x i8>, <16 x i8>* %y
+  %c = xor <16 x i8> %a, %b
+  store <16 x i8> %c, <16 x i8>* %x
+  ret void
+}
+
+define void @xor_v8i16(<8 x i16>* %x, <8 x i16>* %y) {
+; CHECK-LABEL: xor_v8i16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    addi a2, zero, 8
+; CHECK-NEXT:    vsetvli a2, a2, e16,m1,ta,mu
+; CHECK-NEXT:    vle16.v v25, (a0)
+; CHECK-NEXT:    vle16.v v26, (a1)
+; CHECK-NEXT:    vxor.vv v25, v25, v26
+; CHECK-NEXT:    vse16.v v25, (a0)
+; CHECK-NEXT:    ret
+  %a = load <8 x i16>, <8 x i16>* %x
+  %b = load <8 x i16>, <8 x i16>* %y
+  %c = xor <8 x i16> %a, %b
+  store <8 x i16> %c, <8 x i16>* %x
+  ret void
+}
+
+define void @xor_v4i32(<4 x i32>* %x, <4 x i32>* %y) {
+; CHECK-LABEL: xor_v4i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    addi a2, zero, 4
+; CHECK-NEXT:    vsetvli a2, a2, e32,m1,ta,mu
+; CHECK-NEXT:    vle32.v v25, (a0)
+; CHECK-NEXT:    vle32.v v26, (a1)
+; CHECK-NEXT:    vxor.vv v25, v25, v26
+; CHECK-NEXT:    vse32.v v25, (a0)
+; CHECK-NEXT:    ret
+  %a = load <4 x i32>, <4 x i32>* %x
+  %b = load <4 x i32>, <4 x i32>* %y
+  %c = xor <4 x i32> %a, %b
+  store <4 x i32> %c, <4 x i32>* %x
+  ret void
+}
+
+define void @xor_v2i64(<2 x i64>* %x, <2 x i64>* %y) {
+; CHECK-LABEL: xor_v2i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    addi a2, zero, 2
+; CHECK-NEXT:    vsetvli a2, a2, e64,m1,ta,mu
+; CHECK-NEXT:    vle64.v v25, (a0)
+; CHECK-NEXT:    vle64.v v26, (a1)
+; CHECK-NEXT:    vxor.vv v25, v25, v26
+; CHECK-NEXT:    vse64.v v25, (a0)
+; CHECK-NEXT:    ret
+  %a = load <2 x i64>, <2 x i64>* %x
+  %b = load <2 x i64>, <2 x i64>* %y
+  %c = xor <2 x i64> %a, %b
+  store <2 x i64> %c, <2 x i64>* %x
+  ret void
+}
+
+define void @lshr_v16i8(<16 x i8>* %x, <16 x i8>* %y) {
+; CHECK-LABEL: lshr_v16i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    addi a2, zero, 16
+; CHECK-NEXT:    vsetvli a2, a2, e8,m1,ta,mu
+; CHECK-NEXT:    vle8.v v25, (a0)
+; CHECK-NEXT:    vle8.v v26, (a1)
+; CHECK-NEXT:    vsrl.vv v25, v25, v26
+; CHECK-NEXT:    vse8.v v25, (a0)
+; CHECK-NEXT:    ret
+  %a = load <16 x i8>, <16 x i8>* %x
+  %b = load <16 x i8>, <16 x i8>* %y
+  %c = lshr <16 x i8> %a, %b
+  store <16 x i8> %c, <16 x i8>* %x
+  ret void
+}
+
+define void @lshr_v8i16(<8 x i16>* %x, <8 x i16>* %y) {
+; CHECK-LABEL: lshr_v8i16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    addi a2, zero, 8
+; CHECK-NEXT:    vsetvli a2, a2, e16,m1,ta,mu
+; CHECK-NEXT:    vle16.v v25, (a0)
+; CHECK-NEXT:    vle16.v v26, (a1)
+; CHECK-NEXT:    vsrl.vv v25, v25, v26
+; CHECK-NEXT:    vse16.v v25, (a0)
+; CHECK-NEXT:    ret
+  %a = load <8 x i16>, <8 x i16>* %x
+  %b = load <8 x i16>, <8 x i16>* %y
+  %c = lshr <8 x i16> %a, %b
+  store <8 x i16> %c, <8 x i16>* %x
+  ret void
+}
+
+define void @lshr_v4i32(<4 x i32>* %x, <4 x i32>* %y) {
+; CHECK-LABEL: lshr_v4i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    addi a2, zero, 4
+; CHECK-NEXT:    vsetvli a2, a2, e32,m1,ta,mu
+; CHECK-NEXT:    vle32.v v25, (a0)
+; CHECK-NEXT:    vle32.v v26, (a1)
+; CHECK-NEXT:    vsrl.vv v25, v25, v26
+; CHECK-NEXT:    vse32.v v25, (a0)
+; CHECK-NEXT:    ret
+  %a = load <4 x i32>, <4 x i32>* %x
+  %b = load <4 x i32>, <4 x i32>* %y
+  %c = lshr <4 x i32> %a, %b
+  store <4 x i32> %c, <4 x i32>* %x
+  ret void
+}
+
+define void @lshr_v2i64(<2 x i64>* %x, <2 x i64>* %y) {
+; CHECK-LABEL: lshr_v2i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    addi a2, zero, 2
+; CHECK-NEXT:    vsetvli a2, a2, e64,m1,ta,mu
+; CHECK-NEXT:    vle64.v v25, (a0)
+; CHECK-NEXT:    vle64.v v26, (a1)
+; CHECK-NEXT:    vsrl.vv v25, v25, v26
+; CHECK-NEXT:    vse64.v v25, (a0)
+; CHECK-NEXT:    ret
+  %a = load <2 x i64>, <2 x i64>* %x
+  %b = load <2 x i64>, <2 x i64>* %y
+  %c = lshr <2 x i64> %a, %b
+  store <2 x i64> %c, <2 x i64>* %x
+  ret void
+}
+
+define void @ashr_v16i8(<16 x i8>* %x, <16 x i8>* %y) {
+; CHECK-LABEL: ashr_v16i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    addi a2, zero, 16
+; CHECK-NEXT:    vsetvli a2, a2, e8,m1,ta,mu
+; CHECK-NEXT:    vle8.v v25, (a0)
+; CHECK-NEXT:    vle8.v v26, (a1)
+; CHECK-NEXT:    vsra.vv v25, v25, v26
+; CHECK-NEXT:    vse8.v v25, (a0)
+; CHECK-NEXT:    ret
+  %a = load <16 x i8>, <16 x i8>* %x
+  %b = load <16 x i8>, <16 x i8>* %y
+  %c = ashr <16 x i8> %a, %b
+  store <16 x i8> %c, <16 x i8>* %x
+  ret void
+}
+
+define void @ashr_v8i16(<8 x i16>* %x, <8 x i16>* %y) {
+; CHECK-LABEL: ashr_v8i16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    addi a2, zero, 8
+; CHECK-NEXT:    vsetvli a2, a2, e16,m1,ta,mu
+; CHECK-NEXT:    vle16.v v25, (a0)
+; CHECK-NEXT:    vle16.v v26, (a1)
+; CHECK-NEXT:    vsra.vv v25, v25, v26
+; CHECK-NEXT:    vse16.v v25, (a0)
+; CHECK-NEXT:    ret
+  %a = load <8 x i16>, <8 x i16>* %x
+  %b = load <8 x i16>, <8 x i16>* %y
+  %c = ashr <8 x i16> %a, %b
+  store <8 x i16> %c, <8 x i16>* %x
+  ret void
+}
+
+define void @ashr_v4i32(<4 x i32>* %x, <4 x i32>* %y) {
+; CHECK-LABEL: ashr_v4i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    addi a2, zero, 4
+; CHECK-NEXT:    vsetvli a2, a2, e32,m1,ta,mu
+; CHECK-NEXT:    vle32.v v25, (a0)
+; CHECK-NEXT:    vle32.v v26, (a1)
+; CHECK-NEXT:    vsra.vv v25, v25, v26
+; CHECK-NEXT:    vse32.v v25, (a0)
+; CHECK-NEXT:    ret
+  %a = load <4 x i32>, <4 x i32>* %x
+  %b = load <4 x i32>, <4 x i32>* %y
+  %c = ashr <4 x i32> %a, %b
+  store <4 x i32> %c, <4 x i32>* %x
+  ret void
+}
+
+define void @ashr_v2i64(<2 x i64>* %x, <2 x i64>* %y) {
+; CHECK-LABEL: ashr_v2i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    addi a2, zero, 2
+; CHECK-NEXT:    vsetvli a2, a2, e64,m1,ta,mu
+; CHECK-NEXT:    vle64.v v25, (a0)
+; CHECK-NEXT:    vle64.v v26, (a1)
+; CHECK-NEXT:    vsra.vv v25, v25, v26
+; CHECK-NEXT:    vse64.v v25, (a0)
+; CHECK-NEXT:    ret
+  %a = load <2 x i64>, <2 x i64>* %x
+  %b = load <2 x i64>, <2 x i64>* %y
+  %c = ashr <2 x i64> %a, %b
+  store <2 x i64> %c, <2 x i64>* %x
+  ret void
+}
+
+define void @shl_v16i8(<16 x i8>* %x, <16 x i8>* %y) {
+; CHECK-LABEL: shl_v16i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    addi a2, zero, 16
+; CHECK-NEXT:    vsetvli a2, a2, e8,m1,ta,mu
+; CHECK-NEXT:    vle8.v v25, (a0)
+; CHECK-NEXT:    vle8.v v26, (a1)
+; CHECK-NEXT:    vsll.vv v25, v25, v26
+; CHECK-NEXT:    vse8.v v25, (a0)
+; CHECK-NEXT:    ret
+  %a = load <16 x i8>, <16 x i8>* %x
+  %b = load <16 x i8>, <16 x i8>* %y
+  %c = shl <16 x i8> %a, %b
+  store <16 x i8> %c, <16 x i8>* %x
+  ret void
+}
+
+define void @shl_v8i16(<8 x i16>* %x, <8 x i16>* %y) {
+; CHECK-LABEL: shl_v8i16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    addi a2, zero, 8
+; CHECK-NEXT:    vsetvli a2, a2, e16,m1,ta,mu
+; CHECK-NEXT:    vle16.v v25, (a0)
+; CHECK-NEXT:    vle16.v v26, (a1)
+; CHECK-NEXT:    vsll.vv v25, v25, v26
+; CHECK-NEXT:    vse16.v v25, (a0)
+; CHECK-NEXT:    ret
+  %a = load <8 x i16>, <8 x i16>* %x
+  %b = load <8 x i16>, <8 x i16>* %y
+  %c = shl <8 x i16> %a, %b
+  store <8 x i16> %c, <8 x i16>* %x
+  ret void
+}
+
+define void @shl_v4i32(<4 x i32>* %x, <4 x i32>* %y) {
+; CHECK-LABEL: shl_v4i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    addi a2, zero, 4
+; CHECK-NEXT:    vsetvli a2, a2, e32,m1,ta,mu
+; CHECK-NEXT:    vle32.v v25, (a0)
+; CHECK-NEXT:    vle32.v v26, (a1)
+; CHECK-NEXT:    vsll.vv v25, v25, v26
+; CHECK-NEXT:    vse32.v v25, (a0)
+; CHECK-NEXT:    ret
+  %a = load <4 x i32>, <4 x i32>* %x
+  %b = load <4 x i32>, <4 x i32>* %y
+  %c = shl <4 x i32> %a, %b
+  store <4 x i32> %c, <4 x i32>* %x
+  ret void
+}
+
+define void @shl_v2i64(<2 x i64>* %x, <2 x i64>* %y) {
+; CHECK-LABEL: shl_v2i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    addi a2, zero, 2
+; CHECK-NEXT:    vsetvli a2, a2, e64,m1,ta,mu
+; CHECK-NEXT:    vle64.v v25, (a0)
+; CHECK-NEXT:    vle64.v v26, (a1)
+; CHECK-NEXT:    vsll.vv v25, v25, v26
+; CHECK-NEXT:    vse64.v v25, (a0)
+; CHECK-NEXT:    ret
+  %a = load <2 x i64>, <2 x i64>* %x
+  %b = load <2 x i64>, <2 x i64>* %y
+  %c = shl <2 x i64> %a, %b
+  store <2 x i64> %c, <2 x i64>* %x
+  ret void
+}
+
+define void @sdiv_v16i8(<16 x i8>* %x, <16 x i8>* %y) {
+; CHECK-LABEL: sdiv_v16i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    addi a2, zero, 16
+; CHECK-NEXT:    vsetvli a2, a2, e8,m1,ta,mu
+; CHECK-NEXT:    vle8.v v25, (a0)
+; CHECK-NEXT:    vle8.v v26, (a1)
+; CHECK-NEXT:    vdivu.vv v25, v25, v26
+; CHECK-NEXT:    vse8.v v25, (a0)
+; CHECK-NEXT:    ret
+  %a = load <16 x i8>, <16 x i8>* %x
+  %b = load <16 x i8>, <16 x i8>* %y
+  %c = sdiv <16 x i8> %a, %b
+  store <16 x i8> %c, <16 x i8>* %x
+  ret void
+}
+
+define void @sdiv_v8i16(<8 x i16>* %x, <8 x i16>* %y) {
+; CHECK-LABEL: sdiv_v8i16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    addi a2, zero, 8
+; CHECK-NEXT:    vsetvli a2, a2, e16,m1,ta,mu
+; CHECK-NEXT:    vle16.v v25, (a0)
+; CHECK-NEXT:    vle16.v v26, (a1)
+; CHECK-NEXT:    vdivu.vv v25, v25, v26
+; CHECK-NEXT:    vse16.v v25, (a0)
+; CHECK-NEXT:    ret
+  %a = load <8 x i16>, <8 x i16>* %x
+  %b = load <8 x i16>, <8 x i16>* %y
+  %c = sdiv <8 x i16> %a, %b
+  store <8 x i16> %c, <8 x i16>* %x
+  ret void
+}
+
+define void @sdiv_v4i32(<4 x i32>* %x, <4 x i32>* %y) {
+; CHECK-LABEL: sdiv_v4i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    addi a2, zero, 4
+; CHECK-NEXT:    vsetvli a2, a2, e32,m1,ta,mu
+; CHECK-NEXT:    vle32.v v25, (a0)
+; CHECK-NEXT:    vle32.v v26, (a1)
+; CHECK-NEXT:    vdivu.vv v25, v25, v26
+; CHECK-NEXT:    vse32.v v25, (a0)
+; CHECK-NEXT:    ret
+  %a = load <4 x i32>, <4 x i32>* %x
+  %b = load <4 x i32>, <4 x i32>* %y
+  %c = sdiv <4 x i32> %a, %b
+  store <4 x i32> %c, <4 x i32>* %x
+  ret void
+}
+
+define void @sdiv_v2i64(<2 x i64>* %x, <2 x i64>* %y) {
+; CHECK-LABEL: sdiv_v2i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    addi a2, zero, 2
+; CHECK-NEXT:    vsetvli a2, a2, e64,m1,ta,mu
+; CHECK-NEXT:    vle64.v v25, (a0)
+; CHECK-NEXT:    vle64.v v26, (a1)
+; CHECK-NEXT:    vdivu.vv v25, v25, v26
+; CHECK-NEXT:    vse64.v v25, (a0)
+; CHECK-NEXT:    ret
+  %a = load <2 x i64>, <2 x i64>* %x
+  %b = load <2 x i64>, <2 x i64>* %y
+  %c = sdiv <2 x i64> %a, %b
+  store <2 x i64> %c, <2 x i64>* %x
+  ret void
+}
+
+define void @srem_v16i8(<16 x i8>* %x, <16 x i8>* %y) {
+; CHECK-LABEL: srem_v16i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    addi a2, zero, 16
+; CHECK-NEXT:    vsetvli a2, a2, e8,m1,ta,mu
+; CHECK-NEXT:    vle8.v v25, (a0)
+; CHECK-NEXT:    vle8.v v26, (a1)
+; CHECK-NEXT:    vrem.vv v25, v25, v26
+; CHECK-NEXT:    vse8.v v25, (a0)
+; CHECK-NEXT:    ret
+  %a = load <16 x i8>, <16 x i8>* %x
+  %b = load <16 x i8>, <16 x i8>* %y
+  %c = srem <16 x i8> %a, %b
+  store <16 x i8> %c, <16 x i8>* %x
+  ret void
+}
+
+define void @srem_v8i16(<8 x i16>* %x, <8 x i16>* %y) {
+; CHECK-LABEL: srem_v8i16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    addi a2, zero, 8
+; CHECK-NEXT:    vsetvli a2, a2, e16,m1,ta,mu
+; CHECK-NEXT:    vle16.v v25, (a0)
+; CHECK-NEXT:    vle16.v v26, (a1)
+; CHECK-NEXT:    vrem.vv v25, v25, v26
+; CHECK-NEXT:    vse16.v v25, (a0)
+; CHECK-NEXT:    ret
+  %a = load <8 x i16>, <8 x i16>* %x
+  %b = load <8 x i16>, <8 x i16>* %y
+  %c = srem <8 x i16> %a, %b
+  store <8 x i16> %c, <8 x i16>* %x
+  ret void
+}
+
+define void @srem_v4i32(<4 x i32>* %x, <4 x i32>* %y) {
+; CHECK-LABEL: srem_v4i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    addi a2, zero, 4
+; CHECK-NEXT:    vsetvli a2, a2, e32,m1,ta,mu
+; CHECK-NEXT:    vle32.v v25, (a0)
+; CHECK-NEXT:    vle32.v v26, (a1)
+; CHECK-NEXT:    vrem.vv v25, v25, v26
+; CHECK-NEXT:    vse32.v v25, (a0)
+; CHECK-NEXT:    ret
+  %a = load <4 x i32>, <4 x i32>* %x
+  %b = load <4 x i32>, <4 x i32>* %y
+  %c = srem <4 x i32> %a, %b
+  store <4 x i32> %c, <4 x i32>* %x
+  ret void
+}
+
+define void @srem_v2i64(<2 x i64>* %x, <2 x i64>* %y) {
+; CHECK-LABEL: srem_v2i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    addi a2, zero, 2
+; CHECK-NEXT:    vsetvli a2, a2, e64,m1,ta,mu
+; CHECK-NEXT:    vle64.v v25, (a0)
+; CHECK-NEXT:    vle64.v v26, (a1)
+; CHECK-NEXT:    vrem.vv v25, v25, v26
+; CHECK-NEXT:    vse64.v v25, (a0)
+; CHECK-NEXT:    ret
+  %a = load <2 x i64>, <2 x i64>* %x
+  %b = load <2 x i64>, <2 x i64>* %y
+  %c = srem <2 x i64> %a, %b
+  store <2 x i64> %c, <2 x i64>* %x
+  ret void
+}
+
+define void @udiv_v16i8(<16 x i8>* %x, <16 x i8>* %y) {
+; CHECK-LABEL: udiv_v16i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    addi a2, zero, 16
+; CHECK-NEXT:    vsetvli a2, a2, e8,m1,ta,mu
+; CHECK-NEXT:    vle8.v v25, (a0)
+; CHECK-NEXT:    vle8.v v26, (a1)
+; CHECK-NEXT:    vdiv.vv v25, v25, v26
+; CHECK-NEXT:    vse8.v v25, (a0)
+; CHECK-NEXT:    ret
+  %a = load <16 x i8>, <16 x i8>* %x
+  %b = load <16 x i8>, <16 x i8>* %y
+  %c = udiv <16 x i8> %a, %b
+  store <16 x i8> %c, <16 x i8>* %x
+  ret void
+}
+
+define void @udiv_v8i16(<8 x i16>* %x, <8 x i16>* %y) {
+; CHECK-LABEL: udiv_v8i16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    addi a2, zero, 8
+; CHECK-NEXT:    vsetvli a2, a2, e16,m1,ta,mu
+; CHECK-NEXT:    vle16.v v25, (a0)
+; CHECK-NEXT:    vle16.v v26, (a1)
+; CHECK-NEXT:    vdiv.vv v25, v25, v26
+; CHECK-NEXT:    vse16.v v25, (a0)
+; CHECK-NEXT:    ret
+  %a = load <8 x i16>, <8 x i16>* %x
+  %b = load <8 x i16>, <8 x i16>* %y
+  %c = udiv <8 x i16> %a, %b
+  store <8 x i16> %c, <8 x i16>* %x
+  ret void
+}
+
+define void @udiv_v4i32(<4 x i32>* %x, <4 x i32>* %y) {
+; CHECK-LABEL: udiv_v4i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    addi a2, zero, 4
+; CHECK-NEXT:    vsetvli a2, a2, e32,m1,ta,mu
+; CHECK-NEXT:    vle32.v v25, (a0)
+; CHECK-NEXT:    vle32.v v26, (a1)
+; CHECK-NEXT:    vdiv.vv v25, v25, v26
+; CHECK-NEXT:    vse32.v v25, (a0)
+; CHECK-NEXT:    ret
+  %a = load <4 x i32>, <4 x i32>* %x
+  %b = load <4 x i32>, <4 x i32>* %y
+  %c = udiv <4 x i32> %a, %b
+  store <4 x i32> %c, <4 x i32>* %x
+  ret void
+}
+
+define void @udiv_v2i64(<2 x i64>* %x, <2 x i64>* %y) {
+; CHECK-LABEL: udiv_v2i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    addi a2, zero, 2
+; CHECK-NEXT:    vsetvli a2, a2, e64,m1,ta,mu
+; CHECK-NEXT:    vle64.v v25, (a0)
+; CHECK-NEXT:    vle64.v v26, (a1)
+; CHECK-NEXT:    vdiv.vv v25, v25, v26
+; CHECK-NEXT:    vse64.v v25, (a0)
+; CHECK-NEXT:    ret
+  %a = load <2 x i64>, <2 x i64>* %x
+  %b = load <2 x i64>, <2 x i64>* %y
+  %c = udiv <2 x i64> %a, %b
+  store <2 x i64> %c, <2 x i64>* %x
+  ret void
+}
+
+define void @urem_v16i8(<16 x i8>* %x, <16 x i8>* %y) {
+; CHECK-LABEL: urem_v16i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    addi a2, zero, 16
+; CHECK-NEXT:    vsetvli a2, a2, e8,m1,ta,mu
+; CHECK-NEXT:    vle8.v v25, (a0)
+; CHECK-NEXT:    vle8.v v26, (a1)
+; CHECK-NEXT:    vremu.vv v25, v25, v26
+; CHECK-NEXT:    vse8.v v25, (a0)
+; CHECK-NEXT:    ret
+  %a = load <16 x i8>, <16 x i8>* %x
+  %b = load <16 x i8>, <16 x i8>* %y
+  %c = urem <16 x i8> %a, %b
+  store <16 x i8> %c, <16 x i8>* %x
+  ret void
+}
+
+define void @urem_v8i16(<8 x i16>* %x, <8 x i16>* %y) {
+; CHECK-LABEL: urem_v8i16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    addi a2, zero, 8
+; CHECK-NEXT:    vsetvli a2, a2, e16,m1,ta,mu
+; CHECK-NEXT:    vle16.v v25, (a0)
+; CHECK-NEXT:    vle16.v v26, (a1)
+; CHECK-NEXT:    vremu.vv v25, v25, v26
+; CHECK-NEXT:    vse16.v v25, (a0)
+; CHECK-NEXT:    ret
+  %a = load <8 x i16>, <8 x i16>* %x
+  %b = load <8 x i16>, <8 x i16>* %y
+  %c = urem <8 x i16> %a, %b
+  store <8 x i16> %c, <8 x i16>* %x
+  ret void
+}
+
+define void @urem_v4i32(<4 x i32>* %x, <4 x i32>* %y) {
+; CHECK-LABEL: urem_v4i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    addi a2, zero, 4
+; CHECK-NEXT:    vsetvli a2, a2, e32,m1,ta,mu
+; CHECK-NEXT:    vle32.v v25, (a0)
+; CHECK-NEXT:    vle32.v v26, (a1)
+; CHECK-NEXT:    vremu.vv v25, v25, v26
+; CHECK-NEXT:    vse32.v v25, (a0)
+; CHECK-NEXT:    ret
+  %a = load <4 x i32>, <4 x i32>* %x
+  %b = load <4 x i32>, <4 x i32>* %y
+  %c = urem <4 x i32> %a, %b
+  store <4 x i32> %c, <4 x i32>* %x
+  ret void
+}
+
+define void @urem_v2i64(<2 x i64>* %x, <2 x i64>* %y) {
+; CHECK-LABEL: urem_v2i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    addi a2, zero, 2
+; CHECK-NEXT:    vsetvli a2, a2, e64,m1,ta,mu
+; CHECK-NEXT:    vle64.v v25, (a0)
+; CHECK-NEXT:    vle64.v v26, (a1)
+; CHECK-NEXT:    vremu.vv v25, v25, v26
+; CHECK-NEXT:    vse64.v v25, (a0)
+; CHECK-NEXT:    ret
+  %a = load <2 x i64>, <2 x i64>* %x
+  %b = load <2 x i64>, <2 x i64>* %y
+  %c = urem <2 x i64> %a, %b
+  store <2 x i64> %c, <2 x i64>* %x
+  ret void
+}