AMDGPU/GlobalISel: cmp/select method for insert element

[rampitec]

llvm/test/CodeGen/AMDGPU/GlobalISel/insertelement.ll

299

I assume this will get better when it is moved after RegBankSelect. The issue is copies to VCC inserted by RegBankSelect. The code does not need to be like this, the same code in SelectionDAG creates here:

; %bb.0:                                ; %entry
        v_mov_b32_e32 v1, s0
        v_cmp_eq_u32_e64 vcc, s8, 0
        v_cndmask_b32_e32 v8, v1, v0, vcc
        v_mov_b32_e32 v1, s1
        v_cmp_eq_u32_e64 vcc, s8, 1
        v_cndmask_b32_e32 v1, v1, v0, vcc
        v_mov_b32_e32 v2, s2
        v_cmp_eq_u32_e64 vcc, s8, 2
        v_cndmask_b32_e32 v2, v2, v0, vcc
        v_mov_b32_e32 v3, s3
        v_cmp_eq_u32_e64 vcc, s8, 3
        v_cndmask_b32_e32 v3, v3, v0, vcc
        v_mov_b32_e32 v4, s4
        v_cmp_eq_u32_e64 vcc, s8, 4
        v_cndmask_b32_e32 v4, v4, v0, vcc
        v_mov_b32_e32 v5, s5
        v_cmp_eq_u32_e64 vcc, s8, 5
        v_cndmask_b32_e32 v5, v5, v0, vcc
        v_mov_b32_e32 v6, s6
        v_cmp_eq_u32_e64 vcc, s8, 6
        v_cndmask_b32_e32 v6, v6, v0, vcc
        v_mov_b32_e32 v7, s7
        v_cmp_eq_u32_e64 vcc, s8, 7
        v_cndmask_b32_e32 v7, v7, v0, vcc
        v_mov_b32_e32 v0, v8
        ; return to shader part epilog

[arsenm]

llvm/test/CodeGen/AMDGPU/GlobalISel/insertelement.ll

299

Boolean handling is a mess that needs cleanups, and now we get none. I recently saw a case using 5 instructions to get a constant 0.

[rampitec]

Changes similar to extractelement.

[vpykhtin]

LGTM

[arsenm]

llvm/lib/Target/AMDGPU/AMDGPUPostLegalizerCombiner.cpp

278 ↗

(On Diff #267059)

Can initialize to the size here and avoid push_back

290 ↗

(On Diff #267059)

Don't need getReg here

[rampitec]

Addressed review comments.

[rampitec]

Moved into RegBankSelect.

[rampitec]

llvm/test/CodeGen/AMDGPU/GlobalISel/insertelement.ll

32

That is an incorrect selection of G_ICMP with wave32. AMDGPUInstructionSelector::isVCC() does this:

if (RC) {
  const LLT Ty = MRI.getType(Reg);
  return RC->hasSuperClassEq(TRI.getBoolRC()) &&
         Ty.isValid() && Ty.getSizeInBits() == 1;
}

Since SGPR_32 is used for condition hasSuperClassEq() returns true, and even though we have Ty == s1 it still considers it a VCC.

[rampitec]

llvm/test/CodeGen/AMDGPU/GlobalISel/insertelement.ll

32

This can be fixed by using LLT::scalar(32) instead of LLT::scalar(1).

[rampitec]

Changed scalar condition type to S32.

[rampitec]

Updated test.

[arsenm]

llvm/lib/Target/AMDGPU/AMDGPURegisterBankInfo.cpp

1902–1903

This can ignore this? If it had a constant index the legalizer would have dealt with it already. It's also not wrong to do the rest for a constant

1936–1937

RegBankSelect should never need to consider the constant bus restriction, see the long comment at the top of the file. Any vector operation should only use VGPR operands

llvm/test/CodeGen/AMDGPU/GlobalISel/insertelement.ll

32

Yes, see the long comment at the top of the file. Scalar compares need to always produce s32, s1 is assumed VCC.

66–81

I think the register indexing looks better if the index is uniform

[rampitec]

llvm/test/CodeGen/AMDGPU/GlobalISel/insertelement.ll

66–81

It may look better, but it is not faster.

[rampitec]

Rebased.

[rampitec]

Ping.

llvm/lib/Target/AMDGPU/AMDGPURegisterBankInfo.h

public:
   const InstructionMapping &
   getInstrMapping(const MachineInstr &MI) const override;
 
 private:
 
   bool foldExtractEltToCmpSelect(MachineInstr &MI,
                                  MachineRegisterInfo &MRI,
                                  const OperandsMapper &OpdMapper) const;
+  bool foldInsertEltToCmpSelect(MachineInstr &MI,
+                                MachineRegisterInfo &MRI,
+                                const OperandsMapper &OpdMapper) const;
 };
 } // End llvm namespace.
 #endif

llvm/lib/Target/AMDGPU/AMDGPURegisterBankInfo.cpp

bool AMDGPURegisterBankInfo::foldExtractEltToCmpSelect(
   }
 
   MRI.setRegBank(MI.getOperand(0).getReg(), DstBank);
   MI.eraseFromParent();
 
   return true;
 }
 
+bool AMDGPURegisterBankInfo::foldInsertEltToCmpSelect(
+  MachineInstr &MI, MachineRegisterInfo &MRI,
+  const OperandsMapper &OpdMapper) const {
+
+  Register VecReg = MI.getOperand(1).getReg();
+  Register Idx = MI.getOperand(3).getReg();
+
+  if (getOpcodeDef(TargetOpcode::G_CONSTANT, Idx, MRI))
+    return false;
+
+  const RegisterBank &IdxBank =
+    *OpdMapper.getInstrMapping().getOperandMapping(3).BreakDown[0].RegBank;
+
+  bool IsDivergentIdx = IdxBank == AMDGPU::VGPRRegBank;
+
+  LLT VecTy = MRI.getType(VecReg);
+  unsigned EltSize = VecTy.getScalarSizeInBits();
+  unsigned NumElem = VecTy.getNumElements();
+
+  if (!SITargetLowering::shouldExpandVectorDynExt(EltSize, NumElem,
+                                                  IsDivergentIdx))
+    return false;
+
+  MachineIRBuilder B(MI);
+  LLT S32 = LLT::scalar(32);
+
+  const RegisterBank &DstBank =
+    *OpdMapper.getInstrMapping().getOperandMapping(0).BreakDown[0].RegBank;
+  const RegisterBank &SrcBank =
+    *OpdMapper.getInstrMapping().getOperandMapping(1).BreakDown[0].RegBank;
+  const RegisterBank &InsBank =
+    *OpdMapper.getInstrMapping().getOperandMapping(2).BreakDown[0].RegBank;
+
+  const RegisterBank &CCBank =
+    (DstBank == AMDGPU::SGPRRegBank &&
+     SrcBank == AMDGPU::SGPRRegBank &&
+     InsBank == AMDGPU::SGPRRegBank &&
+     IdxBank == AMDGPU::SGPRRegBank) ? AMDGPU::SGPRRegBank
+                                     : AMDGPU::VCCRegBank;
+
+  if (CCBank == AMDGPU::VCCRegBank && IdxBank == AMDGPU::SGPRRegBank &&
+      Subtarget.getConstantBusLimit(AMDGPU::V_CMP_EQ_U32_e32) < 2) {
+    // Legalize compare to avoid constant bus restriction.
+    Idx = B.buildCopy(S32, Idx)->getOperand(0).getReg();
+    MRI.setRegBank(Idx, AMDGPU::VGPRRegBank);
+  }
+
+  LLT EltTy = VecTy.getScalarType();
+  SmallVector<Register, 2> InsRegs(OpdMapper.getVRegs(2));
+  unsigned NumLanes = InsRegs.size();
+  if (!NumLanes) {
+    NumLanes = 1;
+    InsRegs.push_back(MI.getOperand(2).getReg());
+  } else {
+    EltTy = MRI.getType(InsRegs[0]);
+  }
+
+  auto UnmergeToEltTy = B.buildUnmerge(EltTy, VecReg);
+  SmallVector<Register, 16> Ops(NumElem * NumLanes);
+
+  for (unsigned I = 0; I < NumElem; ++I) {
+    auto IC = B.buildConstant(S32, I);
+    MRI.setRegBank(IC->getOperand(0).getReg(), AMDGPU::SGPRRegBank);
+    auto Cmp = B.buildICmp(CmpInst::ICMP_EQ, LLT::scalar(1), Idx, IC);
+    MRI.setRegBank(Cmp->getOperand(0).getReg(), CCBank);
+
+    for (unsigned L = 0; L < NumLanes; ++L) {
+      auto S = B.buildSelect(EltTy, Cmp, InsRegs[L],
+                             UnmergeToEltTy.getReg(I * NumLanes + L));
+
+      for (unsigned N : { 0, 2, 3 })
+        MRI.setRegBank(S->getOperand(N).getReg(), DstBank);
+
+      Ops[I * NumLanes + L] = S->getOperand(0).getReg();
+    }
+  }
+
+  LLT MergeTy = LLT::vector(Ops.size(), EltTy);
+  if (MergeTy == MRI.getType(MI.getOperand(0).getReg())) {
+    B.buildBuildVector(MI.getOperand(0), Ops);
+  } else {
+    auto Vec = B.buildBuildVector(MergeTy, Ops);
+    MRI.setRegBank(Vec->getOperand(0).getReg(), DstBank);
+    B.buildBitcast(MI.getOperand(0).getReg(), Vec);
+  }
+
+  MRI.setRegBank(MI.getOperand(0).getReg(), DstBank);
+  MI.eraseFromParent();
+
+  return true;
+}
+
 void AMDGPURegisterBankInfo::applyMappingImpl(
     const OperandsMapper &OpdMapper) const {
   MachineInstr &MI = OpdMapper.getMI();
   unsigned Opc = MI.getOpcode();
   MachineRegisterInfo &MRI = OpdMapper.getMRI();
   switch (Opc) {
   case AMDGPU::G_PHI: {
     Register DstReg = MI.getOperand(0).getReg();
case AMDGPU::G_INSERT_VECTOR_ELT: {
     SmallVector<Register, 2> InsRegs(OpdMapper.getVRegs(2));
 
     Register DstReg = MI.getOperand(0).getReg();
     LLT VecTy = MRI.getType(DstReg);
 
     assert(OpdMapper.getVRegs(0).empty());
     assert(OpdMapper.getVRegs(3).empty());
 
-    const RegisterBank *IdxBank =
-      OpdMapper.getInstrMapping().getOperandMapping(3).BreakDown[0].RegBank;
-
     if (substituteSimpleCopyRegs(OpdMapper, 1))
       MRI.setType(MI.getOperand(1).getReg(), VecTy);
 
+    if (foldInsertEltToCmpSelect(MI, MRI, OpdMapper))
+      return;
+
+    const RegisterBank *IdxBank =
+      OpdMapper.getInstrMapping().getOperandMapping(3).BreakDown[0].RegBank;
+
     Register SrcReg = MI.getOperand(1).getReg();
     Register InsReg = MI.getOperand(2).getReg();
     LLT InsTy = MRI.getType(InsReg);
     (void)InsTy;
 
     Register BaseIdxReg;
     unsigned ConstOffset;
     MachineInstr *OffsetDef;

llvm/test/CodeGen/AMDGPU/GlobalISel/insertelement.ll

 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
 ; RUN: llc -global-isel -mtriple=amdgcn-mesa-mesa3d -mcpu=gfx900 -verify-machineinstrs < %s | FileCheck -check-prefix=GPRIDX %s
 ; RUN: llc -global-isel -mtriple=amdgcn-mesa-mesa3d -mcpu=gfx1010 -verify-machineinstrs < %s | FileCheck -check-prefix=MOVREL %s
 ; RUN: not --crash llc -global-isel -mtriple=amdgcn-mesa-mesa3d -mcpu=fiji -verify-machineinstrs -o /dev/null %s 2>&1 | FileCheck -check-prefix=ERR %s
 
 ; FIXME: Need constant bus fixup pre-gfx10 for movrel
 ; ERR: Bad machine code: VOP* instruction violates constant bus restriction
 
 define amdgpu_ps <8 x i32> @dyn_insertelement_v8i32_s_s_s(<8 x i32> inreg %vec, i32 inreg %val, i32 inreg %idx) {
 ; GPRIDX-LABEL: dyn_insertelement_v8i32_s_s_s:
 ; GPRIDX:       ; %bb.0: ; %entry
-; GPRIDX-NEXT:    s_mov_b32 s0, s2
-; GPRIDX-NEXT:    s_mov_b32 s1, s3
-; GPRIDX-NEXT:    s_mov_b32 s2, s4
-; GPRIDX-NEXT:    s_mov_b32 s3, s5
-; GPRIDX-NEXT:    s_mov_b32 s4, s6
-; GPRIDX-NEXT:    s_mov_b32 s5, s7
-; GPRIDX-NEXT:    s_mov_b32 s6, s8
-; GPRIDX-NEXT:    s_mov_b32 s7, s9
-; GPRIDX-NEXT:    s_mov_b32 m0, s11
-; GPRIDX-NEXT:    s_nop 0
-; GPRIDX-NEXT:    s_movreld_b32 s0, s10
+; GPRIDX-NEXT:    s_cmp_eq_u32 s11, 0
+; GPRIDX-NEXT:    s_cselect_b32 s0, s10, s2
+; GPRIDX-NEXT:    s_cmp_eq_u32 s11, 1
+; GPRIDX-NEXT:    s_cselect_b32 s1, s10, s3
+; GPRIDX-NEXT:    s_cmp_eq_u32 s11, 2
+; GPRIDX-NEXT:    s_cselect_b32 s2, s10, s4
+; GPRIDX-NEXT:    s_cmp_eq_u32 s11, 3
+; GPRIDX-NEXT:    s_cselect_b32 s3, s10, s5
+; GPRIDX-NEXT:    s_cmp_eq_u32 s11, 4
+; GPRIDX-NEXT:    s_cselect_b32 s4, s10, s6
+; GPRIDX-NEXT:    s_cmp_eq_u32 s11, 5
+; GPRIDX-NEXT:    s_cselect_b32 s5, s10, s7
+; GPRIDX-NEXT:    s_cmp_eq_u32 s11, 6
+; GPRIDX-NEXT:    s_cselect_b32 s6, s10, s8
+; GPRIDX-NEXT:    s_cmp_eq_u32 s11, 7
+; GPRIDX-NEXT:    s_cselect_b32 s7, s10, s9
 ; GPRIDX-NEXT:    ; return to shader part epilog
 ;
 ; MOVREL-LABEL: dyn_insertelement_v8i32_s_s_s:
 ; MOVREL:       ; %bb.0: ; %entry
-; MOVREL-NEXT:    s_mov_b32 s0, s2
-; MOVREL-NEXT:    s_mov_b32 m0, s11
-; MOVREL-NEXT:    s_mov_b32 s1, s3
-; MOVREL-NEXT:    s_mov_b32 s2, s4
-; MOVREL-NEXT:    s_mov_b32 s3, s5
-; MOVREL-NEXT:    s_mov_b32 s4, s6
-; MOVREL-NEXT:    s_mov_b32 s5, s7
-; MOVREL-NEXT:    s_mov_b32 s6, s8
-; MOVREL-NEXT:    s_mov_b32 s7, s9
-; MOVREL-NEXT:    s_movreld_b32 s0, s10
-; MOVREL-NEXT:    ; implicit-def: $vcc_hi
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 s0, s11, 0
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 s1, s11, 1
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 s12, s11, 2
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 s13, s11, 3
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 s14, s11, 4
+; MOVREL-NEXT:    s_cmp_lg_u32 s0, 0
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 s15, s11, 5
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 s16, s11, 6
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 s11, s11, 7
+; MOVREL-NEXT:    ; implicit-def: $vcc_hi
+; MOVREL-NEXT:    s_cselect_b32 s0, s10, s2
+; MOVREL-NEXT:    s_cmp_lg_u32 s1, 0
+; MOVREL-NEXT:    s_cselect_b32 s1, s10, s3
+; MOVREL-NEXT:    s_cmp_lg_u32 s12, 0
+; MOVREL-NEXT:    s_cselect_b32 s2, s10, s4
+; MOVREL-NEXT:    s_cmp_lg_u32 s13, 0
+; MOVREL-NEXT:    s_cselect_b32 s3, s10, s5
+; MOVREL-NEXT:    s_cmp_lg_u32 s14, 0
+; MOVREL-NEXT:    s_cselect_b32 s4, s10, s6
+; MOVREL-NEXT:    s_cmp_lg_u32 s15, 0
+; MOVREL-NEXT:    s_cselect_b32 s5, s10, s7
+; MOVREL-NEXT:    s_cmp_lg_u32 s16, 0
+; MOVREL-NEXT:    s_cselect_b32 s6, s10, s8
+; MOVREL-NEXT:    s_cmp_lg_u32 s11, 0
+; MOVREL-NEXT:    s_cselect_b32 s7, s10, s9
 ; MOVREL-NEXT:    ; return to shader part epilog
 entry:
   %insert = insertelement <8 x i32> %vec, i32 %val, i32 %idx
   ret <8 x i32> %insert
 }
 
 define amdgpu_ps <8 x i8 addrspace(3)*> @dyn_insertelement_v8p3i8_s_s_s(<8 x i8 addrspace(3)*> inreg %vec, i8 addrspace(3)* inreg %val, i32 inreg %idx) {
 ; GPRIDX-LABEL: dyn_insertelement_v8p3i8_s_s_s:
 ; GPRIDX:       ; %bb.0: ; %entry
-; GPRIDX-NEXT:    s_mov_b32 s0, s2
-; GPRIDX-NEXT:    s_mov_b32 s1, s3
-; GPRIDX-NEXT:    s_mov_b32 s2, s4
-; GPRIDX-NEXT:    s_mov_b32 s3, s5
-; GPRIDX-NEXT:    s_mov_b32 s4, s6
-; GPRIDX-NEXT:    s_mov_b32 s5, s7
-; GPRIDX-NEXT:    s_mov_b32 s6, s8
-; GPRIDX-NEXT:    s_mov_b32 s7, s9
-; GPRIDX-NEXT:    s_mov_b32 m0, s11
-; GPRIDX-NEXT:    s_nop 0
-; GPRIDX-NEXT:    s_movreld_b32 s0, s10
+; GPRIDX-NEXT:    s_cmp_eq_u32 s11, 0
+; GPRIDX-NEXT:    s_cselect_b32 s0, s10, s2
+; GPRIDX-NEXT:    s_cmp_eq_u32 s11, 1
+; GPRIDX-NEXT:    s_cselect_b32 s1, s10, s3
+; GPRIDX-NEXT:    s_cmp_eq_u32 s11, 2
+; GPRIDX-NEXT:    s_cselect_b32 s2, s10, s4
+; GPRIDX-NEXT:    s_cmp_eq_u32 s11, 3
+; GPRIDX-NEXT:    s_cselect_b32 s3, s10, s5
+; GPRIDX-NEXT:    s_cmp_eq_u32 s11, 4
+; GPRIDX-NEXT:    s_cselect_b32 s4, s10, s6
+; GPRIDX-NEXT:    s_cmp_eq_u32 s11, 5
+; GPRIDX-NEXT:    s_cselect_b32 s5, s10, s7
+; GPRIDX-NEXT:    s_cmp_eq_u32 s11, 6
+; GPRIDX-NEXT:    s_cselect_b32 s6, s10, s8
+; GPRIDX-NEXT:    s_cmp_eq_u32 s11, 7
+; GPRIDX-NEXT:    s_cselect_b32 s7, s10, s9
 ; GPRIDX-NEXT:    ; return to shader part epilog
 ;
 ; MOVREL-LABEL: dyn_insertelement_v8p3i8_s_s_s:
 ; MOVREL:       ; %bb.0: ; %entry
-; MOVREL-NEXT:    s_mov_b32 s0, s2
-; MOVREL-NEXT:    s_mov_b32 m0, s11
-; MOVREL-NEXT:    s_mov_b32 s1, s3
-; MOVREL-NEXT:    s_mov_b32 s2, s4
-; MOVREL-NEXT:    s_mov_b32 s3, s5
-; MOVREL-NEXT:    s_mov_b32 s4, s6
-; MOVREL-NEXT:    s_mov_b32 s5, s7
-; MOVREL-NEXT:    s_mov_b32 s6, s8
-; MOVREL-NEXT:    s_mov_b32 s7, s9
-; MOVREL-NEXT:    s_movreld_b32 s0, s10
-; MOVREL-NEXT:    ; implicit-def: $vcc_hi
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 s0, s11, 0
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 s1, s11, 1
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 s12, s11, 2
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 s13, s11, 3
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 s14, s11, 4
+; MOVREL-NEXT:    s_cmp_lg_u32 s0, 0
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 s15, s11, 5
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 s16, s11, 6
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 s11, s11, 7
+; MOVREL-NEXT:    ; implicit-def: $vcc_hi
+; MOVREL-NEXT:    s_cselect_b32 s0, s10, s2
+; MOVREL-NEXT:    s_cmp_lg_u32 s1, 0
+; MOVREL-NEXT:    s_cselect_b32 s1, s10, s3
+; MOVREL-NEXT:    s_cmp_lg_u32 s12, 0
+; MOVREL-NEXT:    s_cselect_b32 s2, s10, s4
+; MOVREL-NEXT:    s_cmp_lg_u32 s13, 0
+; MOVREL-NEXT:    s_cselect_b32 s3, s10, s5
+; MOVREL-NEXT:    s_cmp_lg_u32 s14, 0
+; MOVREL-NEXT:    s_cselect_b32 s4, s10, s6
+; MOVREL-NEXT:    s_cmp_lg_u32 s15, 0
+; MOVREL-NEXT:    s_cselect_b32 s5, s10, s7
+; MOVREL-NEXT:    s_cmp_lg_u32 s16, 0
+; MOVREL-NEXT:    s_cselect_b32 s6, s10, s8
+; MOVREL-NEXT:    s_cmp_lg_u32 s11, 0
+; MOVREL-NEXT:    s_cselect_b32 s7, s10, s9
 ; MOVREL-NEXT:    ; return to shader part epilog
 entry:
   %insert = insertelement <8 x i8 addrspace(3)*> %vec, i8 addrspace(3)* %val, i32 %idx
   ret <8 x i8 addrspace(3)*> %insert
 }
 
 define <8 x float> @dyn_insertelement_v8f32_const_s_v_v(float %val, i32 %idx) {
 ; GPRIDX-LABEL: dyn_insertelement_v8f32_const_s_v_v:
 ; GPRIDX:       ; %bb.0: ; %entry
 ; GPRIDX-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
 ; GPRIDX-NEXT:    s_mov_b32 s11, 0x41000000
 ; GPRIDX-NEXT:    s_mov_b32 s10, 0x40e00000
 ; GPRIDX-NEXT:    s_mov_b32 s9, 0x40c00000
 ; GPRIDX-NEXT:    s_mov_b32 s8, 0x40a00000
 ; GPRIDX-NEXT:    s_mov_b32 s7, 4.0
 ; GPRIDX-NEXT:    s_mov_b32 s6, 0x40400000
 ; GPRIDX-NEXT:    s_mov_b32 s5, 2.0
 ; GPRIDX-NEXT:    s_mov_b32 s4, 1.0
-; GPRIDX-NEXT:    v_mov_b32_e32 v17, s11
-; GPRIDX-NEXT:    v_mov_b32_e32 v16, s10
-; GPRIDX-NEXT:    v_mov_b32_e32 v15, s9
-; GPRIDX-NEXT:    v_mov_b32_e32 v14, s8
-; GPRIDX-NEXT:    v_mov_b32_e32 v13, s7
-; GPRIDX-NEXT:    v_mov_b32_e32 v12, s6
-; GPRIDX-NEXT:    v_mov_b32_e32 v11, s5
-; GPRIDX-NEXT:    v_mov_b32_e32 v10, s4
-; GPRIDX-NEXT:    s_mov_b64 s[4:5], exec
-; GPRIDX-NEXT:  BB2_1: ; =>This Inner Loop Header: Depth=1
-; GPRIDX-NEXT:    v_readfirstlane_b32 s6, v1
-; GPRIDX-NEXT:    v_cmp_eq_u32_e32 vcc, s6, v1
-; GPRIDX-NEXT:    s_set_gpr_idx_on s6, gpr_idx(DST)
-; GPRIDX-NEXT:    v_mov_b32_e32 v2, v10
-; GPRIDX-NEXT:    v_mov_b32_e32 v3, v11
-; GPRIDX-NEXT:    v_mov_b32_e32 v4, v12
-; GPRIDX-NEXT:    v_mov_b32_e32 v5, v13
-; GPRIDX-NEXT:    v_mov_b32_e32 v6, v14
-; GPRIDX-NEXT:    v_mov_b32_e32 v7, v15
-; GPRIDX-NEXT:    v_mov_b32_e32 v8, v16
-; GPRIDX-NEXT:    v_mov_b32_e32 v9, v17
-; GPRIDX-NEXT:    v_mov_b32_e32 v2, v0
-; GPRIDX-NEXT:    s_set_gpr_idx_off
-; GPRIDX-NEXT:    s_and_saveexec_b64 vcc, vcc
-; GPRIDX-NEXT:    s_xor_b64 exec, exec, vcc
-; GPRIDX-NEXT:    s_cbranch_execnz BB2_1
-; GPRIDX-NEXT:  ; %bb.2:
-; GPRIDX-NEXT:    s_mov_b64 exec, s[4:5]
-; GPRIDX-NEXT:    v_mov_b32_e32 v0, v2
-; GPRIDX-NEXT:    v_mov_b32_e32 v1, v3
-; GPRIDX-NEXT:    v_mov_b32_e32 v2, v4
-; GPRIDX-NEXT:    v_mov_b32_e32 v3, v5
-; GPRIDX-NEXT:    v_mov_b32_e32 v4, v6
-; GPRIDX-NEXT:    v_mov_b32_e32 v5, v7
-; GPRIDX-NEXT:    v_mov_b32_e32 v6, v8
-; GPRIDX-NEXT:    v_mov_b32_e32 v7, v9
+; GPRIDX-NEXT:    v_mov_b32_e32 v15, s11
+; GPRIDX-NEXT:    v_mov_b32_e32 v8, s4
+; GPRIDX-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v1
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v8, v8, v0, vcc
+; GPRIDX-NEXT:    v_mov_b32_e32 v9, s5
+; GPRIDX-NEXT:    v_cmp_eq_u32_e32 vcc, 1, v1
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v9, v9, v0, vcc
+; GPRIDX-NEXT:    v_mov_b32_e32 v10, s6
+; GPRIDX-NEXT:    v_cmp_eq_u32_e32 vcc, 2, v1
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v2, v10, v0, vcc
+; GPRIDX-NEXT:    v_mov_b32_e32 v11, s7
+; GPRIDX-NEXT:    v_cmp_eq_u32_e32 vcc, 3, v1
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v3, v11, v0, vcc
+; GPRIDX-NEXT:    v_mov_b32_e32 v12, s8
+; GPRIDX-NEXT:    v_cmp_eq_u32_e32 vcc, 4, v1
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v4, v12, v0, vcc
+; GPRIDX-NEXT:    v_mov_b32_e32 v13, s9
+; GPRIDX-NEXT:    v_cmp_eq_u32_e32 vcc, 5, v1
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v5, v13, v0, vcc
+; GPRIDX-NEXT:    v_mov_b32_e32 v14, s10
+; GPRIDX-NEXT:    v_cmp_eq_u32_e32 vcc, 6, v1
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v6, v14, v0, vcc
+; GPRIDX-NEXT:    v_cmp_eq_u32_e32 vcc, 7, v1
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v7, v15, v0, vcc
+; GPRIDX-NEXT:    v_mov_b32_e32 v0, v8
+; GPRIDX-NEXT:    v_mov_b32_e32 v1, v9
 ; GPRIDX-NEXT:    s_setpc_b64 s[30:31]
 ;
 ; MOVREL-LABEL: dyn_insertelement_v8f32_const_s_v_v:
 ; MOVREL:       ; %bb.0: ; %entry
 ; MOVREL-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
 ; MOVREL-NEXT:    s_waitcnt_vscnt null, 0x0
 ; MOVREL-NEXT:    s_mov_b32 s11, 0x41000000
+; MOVREL-NEXT:    s_mov_b32 s4, 1.0
 ; MOVREL-NEXT:    s_mov_b32 s10, 0x40e00000
 ; MOVREL-NEXT:    s_mov_b32 s9, 0x40c00000
 ; MOVREL-NEXT:    s_mov_b32 s8, 0x40a00000
 ; MOVREL-NEXT:    s_mov_b32 s7, 4.0
 ; MOVREL-NEXT:    s_mov_b32 s6, 0x40400000
 ; MOVREL-NEXT:    s_mov_b32 s5, 2.0
-; MOVREL-NEXT:    s_mov_b32 s4, 1.0
-; MOVREL-NEXT:    v_mov_b32_e32 v17, s11
-; MOVREL-NEXT:    v_mov_b32_e32 v16, s10
-; MOVREL-NEXT:    v_mov_b32_e32 v15, s9
-; MOVREL-NEXT:    v_mov_b32_e32 v14, s8
-; MOVREL-NEXT:    v_mov_b32_e32 v13, s7
-; MOVREL-NEXT:    v_mov_b32_e32 v12, s6
-; MOVREL-NEXT:    v_mov_b32_e32 v11, s5
-; MOVREL-NEXT:    v_mov_b32_e32 v10, s4
-; MOVREL-NEXT:    s_mov_b32 s4, exec_lo
-; MOVREL-NEXT:    ; implicit-def: $vcc_hi
-; MOVREL-NEXT:  BB2_1: ; =>This Inner Loop Header: Depth=1
-; MOVREL-NEXT:    v_readfirstlane_b32 s5, v1
-; MOVREL-NEXT:    v_mov_b32_e32 v2, v10
-; MOVREL-NEXT:    v_mov_b32_e32 v3, v11
-; MOVREL-NEXT:    v_mov_b32_e32 v4, v12
-; MOVREL-NEXT:    v_mov_b32_e32 v5, v13
-; MOVREL-NEXT:    v_cmp_eq_u32_e32 vcc_lo, s5, v1
-; MOVREL-NEXT:    s_mov_b32 m0, s5
-; MOVREL-NEXT:    v_mov_b32_e32 v6, v14
-; MOVREL-NEXT:    v_mov_b32_e32 v7, v15
-; MOVREL-NEXT:    v_mov_b32_e32 v8, v16
-; MOVREL-NEXT:    v_mov_b32_e32 v9, v17
-; MOVREL-NEXT:    v_movreld_b32_e32 v2, v0
-; MOVREL-NEXT:    s_and_saveexec_b32 vcc_lo, vcc_lo
-; MOVREL-NEXT:    s_xor_b32 exec_lo, exec_lo, vcc_lo
-; MOVREL-NEXT:    s_cbranch_execnz BB2_1
-; MOVREL-NEXT:  ; %bb.2:
-; MOVREL-NEXT:    s_mov_b32 exec_lo, s4
-; MOVREL-NEXT:    v_mov_b32_e32 v0, v2
-; MOVREL-NEXT:    v_mov_b32_e32 v1, v3
-; MOVREL-NEXT:    v_mov_b32_e32 v2, v4
-; MOVREL-NEXT:    v_mov_b32_e32 v3, v5
-; MOVREL-NEXT:    v_mov_b32_e32 v4, v6
-; MOVREL-NEXT:    v_mov_b32_e32 v5, v7
-; MOVREL-NEXT:    v_mov_b32_e32 v6, v8
-; MOVREL-NEXT:    v_mov_b32_e32 v7, v9
+; MOVREL-NEXT:    v_mov_b32_e32 v15, s11
+; MOVREL-NEXT:    v_mov_b32_e32 v8, s4
+; MOVREL-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 0, v1
+; MOVREL-NEXT:    v_mov_b32_e32 v10, s6
+; MOVREL-NEXT:    v_mov_b32_e32 v9, s5
+; MOVREL-NEXT:    v_mov_b32_e32 v11, s7
+; MOVREL-NEXT:    v_mov_b32_e32 v12, s8
+; MOVREL-NEXT:    v_cndmask_b32_e32 v8, v8, v0, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 1, v1
+; MOVREL-NEXT:    v_mov_b32_e32 v13, s9
+; MOVREL-NEXT:    v_mov_b32_e32 v14, s10
+; MOVREL-NEXT:    ; implicit-def: $vcc_hi
+; MOVREL-NEXT:    v_cndmask_b32_e32 v9, v9, v0, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 2, v1
+; MOVREL-NEXT:    v_cndmask_b32_e32 v2, v10, v0, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 3, v1
+; MOVREL-NEXT:    v_cndmask_b32_e32 v3, v11, v0, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 4, v1
+; MOVREL-NEXT:    v_cndmask_b32_e32 v4, v12, v0, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 5, v1
+; MOVREL-NEXT:    v_cndmask_b32_e32 v5, v13, v0, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 6, v1
+; MOVREL-NEXT:    v_cndmask_b32_e32 v6, v14, v0, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 7, v1
+; MOVREL-NEXT:    v_mov_b32_e32 v1, v9
+; MOVREL-NEXT:    v_cndmask_b32_e32 v7, v15, v0, vcc_lo
+; MOVREL-NEXT:    v_mov_b32_e32 v0, v8
 ; MOVREL-NEXT:    s_setpc_b64 s[30:31]
 entry:
   %insert = insertelement <8 x float> <float 1.0, float 2.0, float 3.0, float 4.0, float 5.0, float 6.0, float 7.0, float 8.0>, float %val, i32 %idx
   ret <8 x float> %insert
 }
 
 define amdgpu_ps <8 x float> @dyn_insertelement_v8f32_s_s_v(<8 x float> inreg %vec, float inreg %val, i32 %idx) {
 ; GPRIDX-LABEL: dyn_insertelement_v8f32_s_s_v:
 ; GPRIDX:       ; %bb.0: ; %entry
 ; GPRIDX-NEXT:    s_mov_b32 s1, s3
 ; GPRIDX-NEXT:    s_mov_b32 s3, s5
 ; GPRIDX-NEXT:    s_mov_b32 s5, s7
 ; GPRIDX-NEXT:    s_mov_b32 s7, s9
 ; GPRIDX-NEXT:    s_mov_b32 s0, s2
 ; GPRIDX-NEXT:    s_mov_b32 s2, s4
 ; GPRIDX-NEXT:    s_mov_b32 s4, s6
 ; GPRIDX-NEXT:    s_mov_b32 s6, s8
-; GPRIDX-NEXT:    v_mov_b32_e32 v16, s7
-; GPRIDX-NEXT:    v_mov_b32_e32 v8, v0
-; GPRIDX-NEXT:    v_mov_b32_e32 v15, s6
-; GPRIDX-NEXT:    v_mov_b32_e32 v14, s5
-; GPRIDX-NEXT:    v_mov_b32_e32 v13, s4
-; GPRIDX-NEXT:    v_mov_b32_e32 v12, s3
-; GPRIDX-NEXT:    v_mov_b32_e32 v11, s2
-; GPRIDX-NEXT:    v_mov_b32_e32 v10, s1
-; GPRIDX-NEXT:    v_mov_b32_e32 v9, s0
-; GPRIDX-NEXT:    s_mov_b64 s[0:1], exec
-; GPRIDX-NEXT:  BB3_1: ; =>This Inner Loop Header: Depth=1
-; GPRIDX-NEXT:    v_readfirstlane_b32 s2, v8
-; GPRIDX-NEXT:    v_cmp_eq_u32_e32 vcc, s2, v8
-; GPRIDX-NEXT:    s_set_gpr_idx_on s2, gpr_idx(DST)
-; GPRIDX-NEXT:    v_mov_b32_e32 v0, v9
-; GPRIDX-NEXT:    v_mov_b32_e32 v1, v10
-; GPRIDX-NEXT:    v_mov_b32_e32 v2, v11
-; GPRIDX-NEXT:    v_mov_b32_e32 v3, v12
-; GPRIDX-NEXT:    v_mov_b32_e32 v4, v13
-; GPRIDX-NEXT:    v_mov_b32_e32 v5, v14
-; GPRIDX-NEXT:    v_mov_b32_e32 v6, v15
-; GPRIDX-NEXT:    v_mov_b32_e32 v7, v16
-; GPRIDX-NEXT:    v_mov_b32_e32 v0, s10
-; GPRIDX-NEXT:    s_set_gpr_idx_off
-; GPRIDX-NEXT:    s_and_saveexec_b64 vcc, vcc
-; GPRIDX-NEXT:    s_xor_b64 exec, exec, vcc
-; GPRIDX-NEXT:    s_cbranch_execnz BB3_1
-; GPRIDX-NEXT:  ; %bb.2:
-; GPRIDX-NEXT:    s_mov_b64 exec, s[0:1]
+; GPRIDX-NEXT:    v_mov_b32_e32 v15, s7
+; GPRIDX-NEXT:    v_mov_b32_e32 v7, s10
+; GPRIDX-NEXT:    v_mov_b32_e32 v8, s0
+; GPRIDX-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v8, v8, v7, vcc
+; GPRIDX-NEXT:    v_mov_b32_e32 v9, s1
+; GPRIDX-NEXT:    v_cmp_eq_u32_e32 vcc, 1, v0
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v1, v9, v7, vcc
+; GPRIDX-NEXT:    v_mov_b32_e32 v10, s2
+; GPRIDX-NEXT:    v_cmp_eq_u32_e32 vcc, 2, v0
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v2, v10, v7, vcc
+; GPRIDX-NEXT:    v_mov_b32_e32 v11, s3
+; GPRIDX-NEXT:    v_cmp_eq_u32_e32 vcc, 3, v0
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v3, v11, v7, vcc
+; GPRIDX-NEXT:    v_mov_b32_e32 v12, s4
+; GPRIDX-NEXT:    v_cmp_eq_u32_e32 vcc, 4, v0
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v4, v12, v7, vcc
+; GPRIDX-NEXT:    v_mov_b32_e32 v13, s5
+; GPRIDX-NEXT:    v_cmp_eq_u32_e32 vcc, 5, v0
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v5, v13, v7, vcc
+; GPRIDX-NEXT:    v_mov_b32_e32 v14, s6
+; GPRIDX-NEXT:    v_cmp_eq_u32_e32 vcc, 6, v0
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v6, v14, v7, vcc
+; GPRIDX-NEXT:    v_cmp_eq_u32_e32 vcc, 7, v0
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v7, v15, v7, vcc
+; GPRIDX-NEXT:    v_mov_b32_e32 v0, v8
 ; GPRIDX-NEXT:    ; return to shader part epilog
 ;
 ; MOVREL-LABEL: dyn_insertelement_v8f32_s_s_v:
 ; MOVREL:       ; %bb.0: ; %entry
 ; MOVREL-NEXT:    s_mov_b32 s1, s3
 ; MOVREL-NEXT:    s_mov_b32 s3, s5
 ; MOVREL-NEXT:    s_mov_b32 s5, s7
 ; MOVREL-NEXT:    s_mov_b32 s7, s9
 ; MOVREL-NEXT:    s_mov_b32 s0, s2
 ; MOVREL-NEXT:    s_mov_b32 s2, s4
 ; MOVREL-NEXT:    s_mov_b32 s4, s6
 ; MOVREL-NEXT:    s_mov_b32 s6, s8
-; MOVREL-NEXT:    v_mov_b32_e32 v16, s7
-; MOVREL-NEXT:    v_mov_b32_e32 v14, s5
-; MOVREL-NEXT:    v_mov_b32_e32 v12, s3
-; MOVREL-NEXT:    v_mov_b32_e32 v13, s4
-; MOVREL-NEXT:    v_mov_b32_e32 v15, s6
-; MOVREL-NEXT:    v_mov_b32_e32 v11, s2
-; MOVREL-NEXT:    v_mov_b32_e32 v10, s1
-; MOVREL-NEXT:    v_mov_b32_e32 v9, s0
-; MOVREL-NEXT:    s_mov_b32 s0, exec_lo
-; MOVREL-NEXT:    ; implicit-def: $vcc_hi
-; MOVREL-NEXT:  BB3_1: ; =>This Inner Loop Header: Depth=1
-; MOVREL-NEXT:    v_readfirstlane_b32 s1, v0
-; MOVREL-NEXT:    v_mov_b32_e32 v1, v9
-; MOVREL-NEXT:    v_mov_b32_e32 v2, v10
-; MOVREL-NEXT:    v_mov_b32_e32 v3, v11
-; MOVREL-NEXT:    v_mov_b32_e32 v4, v12
-; MOVREL-NEXT:    v_cmp_eq_u32_e32 vcc_lo, s1, v0
-; MOVREL-NEXT:    s_mov_b32 m0, s1
-; MOVREL-NEXT:    v_mov_b32_e32 v5, v13
-; MOVREL-NEXT:    v_mov_b32_e32 v6, v14
-; MOVREL-NEXT:    v_mov_b32_e32 v7, v15
-; MOVREL-NEXT:    v_mov_b32_e32 v8, v16
-; MOVREL-NEXT:    v_movreld_b32_e32 v1, s10
-; MOVREL-NEXT:    s_and_saveexec_b32 vcc_lo, vcc_lo
-; MOVREL-NEXT:    s_xor_b32 exec_lo, exec_lo, vcc_lo
-; MOVREL-NEXT:    s_cbranch_execnz BB3_1
-; MOVREL-NEXT:  ; %bb.2:
-; MOVREL-NEXT:    s_mov_b32 exec_lo, s0
-; MOVREL-NEXT:    v_mov_b32_e32 v0, v1
-; MOVREL-NEXT:    v_mov_b32_e32 v1, v2
-; MOVREL-NEXT:    v_mov_b32_e32 v2, v3
-; MOVREL-NEXT:    v_mov_b32_e32 v3, v4
-; MOVREL-NEXT:    v_mov_b32_e32 v4, v5
-; MOVREL-NEXT:    v_mov_b32_e32 v5, v6
-; MOVREL-NEXT:    v_mov_b32_e32 v6, v7
-; MOVREL-NEXT:    v_mov_b32_e32 v7, v8
+; MOVREL-NEXT:    v_mov_b32_e32 v15, s7
+; MOVREL-NEXT:    v_mov_b32_e32 v7, s10
+; MOVREL-NEXT:    v_mov_b32_e32 v8, s0
+; MOVREL-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 0, v0
+; MOVREL-NEXT:    v_mov_b32_e32 v9, s1
+; MOVREL-NEXT:    v_mov_b32_e32 v10, s2
+; MOVREL-NEXT:    v_mov_b32_e32 v11, s3
+; MOVREL-NEXT:    v_mov_b32_e32 v12, s4
+; MOVREL-NEXT:    v_cndmask_b32_e32 v8, v8, v7, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 1, v0
+; MOVREL-NEXT:    v_mov_b32_e32 v13, s5
+; MOVREL-NEXT:    v_mov_b32_e32 v14, s6
+; MOVREL-NEXT:    ; implicit-def: $vcc_hi
+; MOVREL-NEXT:    v_cndmask_b32_e32 v1, v9, v7, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 2, v0
+; MOVREL-NEXT:    v_cndmask_b32_e32 v2, v10, v7, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 3, v0
+; MOVREL-NEXT:    v_cndmask_b32_e32 v3, v11, v7, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 4, v0
+; MOVREL-NEXT:    v_cndmask_b32_e32 v4, v12, v7, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 5, v0
+; MOVREL-NEXT:    v_cndmask_b32_e32 v5, v13, v7, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 6, v0
+; MOVREL-NEXT:    v_cndmask_b32_e32 v6, v14, v7, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 7, v0
+; MOVREL-NEXT:    v_mov_b32_e32 v0, v8
+; MOVREL-NEXT:    v_cndmask_b32_e32 v7, v15, v7, vcc_lo
 ; MOVREL-NEXT:    ; return to shader part epilog
 entry:
   %insert = insertelement <8 x float> %vec, float %val, i32 %idx
   ret <8 x float> %insert
 }
 
 define amdgpu_ps <8 x float> @dyn_insertelement_v8f32_s_v_s(<8 x float> inreg %vec, float %val, i32 inreg %idx) {
 ; GPRIDX-LABEL: dyn_insertelement_v8f32_s_v_s:
 ; GPRIDX:       ; %bb.0: ; %entry
-; GPRIDX-NEXT:    s_mov_b32 s0, s2
 ; GPRIDX-NEXT:    s_mov_b32 s1, s3
-; GPRIDX-NEXT:    s_mov_b32 s2, s4
 ; GPRIDX-NEXT:    s_mov_b32 s3, s5
-; GPRIDX-NEXT:    s_mov_b32 s4, s6
 ; GPRIDX-NEXT:    s_mov_b32 s5, s7
-; GPRIDX-NEXT:    s_mov_b32 s6, s8
 ; GPRIDX-NEXT:    s_mov_b32 s7, s9
-; GPRIDX-NEXT:    v_mov_b32_e32 v8, v0
-; GPRIDX-NEXT:    v_mov_b32_e32 v0, s0
-; GPRIDX-NEXT:    v_mov_b32_e32 v1, s1
-; GPRIDX-NEXT:    v_mov_b32_e32 v2, s2
-; GPRIDX-NEXT:    v_mov_b32_e32 v3, s3
-; GPRIDX-NEXT:    v_mov_b32_e32 v4, s4
-; GPRIDX-NEXT:    v_mov_b32_e32 v5, s5
-; GPRIDX-NEXT:    v_mov_b32_e32 v6, s6
-; GPRIDX-NEXT:    v_mov_b32_e32 v7, s7
-; GPRIDX-NEXT:    s_set_gpr_idx_on s10, gpr_idx(DST)
+; GPRIDX-NEXT:    s_mov_b32 s0, s2
+; GPRIDX-NEXT:    s_mov_b32 s2, s4
+; GPRIDX-NEXT:    s_mov_b32 s4, s6
+; GPRIDX-NEXT:    s_mov_b32 s6, s8
+; GPRIDX-NEXT:    v_mov_b32_e32 v15, s7
+; GPRIDX-NEXT:    v_mov_b32_e32 v8, s0
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 vcc, s10, 0
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v8, v8, v0, vcc
+; GPRIDX-NEXT:    v_mov_b32_e32 v9, s1
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 vcc, s10, 1
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v1, v9, v0, vcc
+; GPRIDX-NEXT:    v_mov_b32_e32 v10, s2
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 vcc, s10, 2
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v2, v10, v0, vcc
+; GPRIDX-NEXT:    v_mov_b32_e32 v11, s3
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 vcc, s10, 3
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v3, v11, v0, vcc
+; GPRIDX-NEXT:    v_mov_b32_e32 v12, s4
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 vcc, s10, 4
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v4, v12, v0, vcc
+; GPRIDX-NEXT:    v_mov_b32_e32 v13, s5
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 vcc, s10, 5
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v5, v13, v0, vcc
+; GPRIDX-NEXT:    v_mov_b32_e32 v14, s6
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 vcc, s10, 6
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v6, v14, v0, vcc
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 vcc, s10, 7
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v7, v15, v0, vcc
 ; GPRIDX-NEXT:    v_mov_b32_e32 v0, v8
-; GPRIDX-NEXT:    s_set_gpr_idx_off
 ; GPRIDX-NEXT:    ; return to shader part epilog
 ;
 ; MOVREL-LABEL: dyn_insertelement_v8f32_s_v_s:
 ; MOVREL:       ; %bb.0: ; %entry
-; MOVREL-NEXT:    s_mov_b32 s0, s2
 ; MOVREL-NEXT:    s_mov_b32 s1, s3
-; MOVREL-NEXT:    s_mov_b32 s2, s4
 ; MOVREL-NEXT:    s_mov_b32 s3, s5
-; MOVREL-NEXT:    s_mov_b32 s4, s6
 ; MOVREL-NEXT:    s_mov_b32 s5, s7
-; MOVREL-NEXT:    s_mov_b32 s6, s8
 ; MOVREL-NEXT:    s_mov_b32 s7, s9
-; MOVREL-NEXT:    v_mov_b32_e32 v8, v0
-; MOVREL-NEXT:    v_mov_b32_e32 v0, s0
-; MOVREL-NEXT:    s_mov_b32 m0, s10
-; MOVREL-NEXT:    v_mov_b32_e32 v1, s1
-; MOVREL-NEXT:    v_mov_b32_e32 v2, s2
-; MOVREL-NEXT:    v_mov_b32_e32 v3, s3
-; MOVREL-NEXT:    v_mov_b32_e32 v4, s4
-; MOVREL-NEXT:    v_mov_b32_e32 v5, s5
-; MOVREL-NEXT:    v_mov_b32_e32 v6, s6
-; MOVREL-NEXT:    v_mov_b32_e32 v7, s7
-; MOVREL-NEXT:    v_movreld_b32_e32 v0, v8
-; MOVREL-NEXT:    ; implicit-def: $vcc_hi
+; MOVREL-NEXT:    s_mov_b32 s0, s2
+; MOVREL-NEXT:    s_mov_b32 s2, s4
+; MOVREL-NEXT:    s_mov_b32 s4, s6
+; MOVREL-NEXT:    s_mov_b32 s6, s8
+; MOVREL-NEXT:    v_mov_b32_e32 v15, s7
+; MOVREL-NEXT:    v_mov_b32_e32 v8, s0
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 vcc_lo, s10, 0
+; MOVREL-NEXT:    v_mov_b32_e32 v9, s1
+; MOVREL-NEXT:    v_mov_b32_e32 v10, s2
+; MOVREL-NEXT:    v_mov_b32_e32 v11, s3
+; MOVREL-NEXT:    v_mov_b32_e32 v12, s4
+; MOVREL-NEXT:    v_cndmask_b32_e32 v8, v8, v0, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 vcc_lo, s10, 1
+; MOVREL-NEXT:    v_mov_b32_e32 v13, s5
+; MOVREL-NEXT:    v_mov_b32_e32 v14, s6
+; MOVREL-NEXT:    ; implicit-def: $vcc_hi
+; MOVREL-NEXT:    v_cndmask_b32_e32 v1, v9, v0, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 vcc_lo, s10, 2
+; MOVREL-NEXT:    v_cndmask_b32_e32 v2, v10, v0, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 vcc_lo, s10, 3
+; MOVREL-NEXT:    v_cndmask_b32_e32 v3, v11, v0, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 vcc_lo, s10, 4
+; MOVREL-NEXT:    v_cndmask_b32_e32 v4, v12, v0, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 vcc_lo, s10, 5
+; MOVREL-NEXT:    v_cndmask_b32_e32 v5, v13, v0, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 vcc_lo, s10, 6
+; MOVREL-NEXT:    v_cndmask_b32_e32 v6, v14, v0, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 vcc_lo, s10, 7
+; MOVREL-NEXT:    v_cndmask_b32_e32 v7, v15, v0, vcc_lo
+; MOVREL-NEXT:    v_mov_b32_e32 v0, v8
 ; MOVREL-NEXT:    ; return to shader part epilog
 entry:
   %insert = insertelement <8 x float> %vec, float %val, i32 %idx
   ret <8 x float> %insert
 }
 
 define amdgpu_ps <8 x float> @dyn_insertelement_v8f32_v_s_s(<8 x float> %vec, float inreg %val, i32 inreg %idx) {
 ; GPRIDX-LABEL: dyn_insertelement_v8f32_v_s_s:
 ; GPRIDX:       ; %bb.0: ; %entry
-; GPRIDX-NEXT:    s_set_gpr_idx_on s3, gpr_idx(DST)
-; GPRIDX-NEXT:    v_mov_b32_e32 v0, s2
-; GPRIDX-NEXT:    s_set_gpr_idx_off
+; GPRIDX-NEXT:    v_mov_b32_e32 v8, s2
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 vcc, s3, 0
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v0, v0, v8, vcc
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 vcc, s3, 1
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v1, v1, v8, vcc
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 vcc, s3, 2
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v2, v2, v8, vcc
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 vcc, s3, 3
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v3, v3, v8, vcc
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 vcc, s3, 4
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v4, v4, v8, vcc
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 vcc, s3, 5
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v5, v5, v8, vcc
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 vcc, s3, 6
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v6, v6, v8, vcc
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 vcc, s3, 7
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v7, v7, v8, vcc
 ; GPRIDX-NEXT:    ; return to shader part epilog
 ;
 ; MOVREL-LABEL: dyn_insertelement_v8f32_v_s_s:
 ; MOVREL:       ; %bb.0: ; %entry
-; MOVREL-NEXT:    s_mov_b32 m0, s3
+; MOVREL-NEXT:    v_mov_b32_e32 v8, s2
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 vcc_lo, s3, 0
 ; MOVREL-NEXT:    ; implicit-def: $vcc_hi
-; MOVREL-NEXT:    v_movreld_b32_e32 v0, s2
+; MOVREL-NEXT:    v_cndmask_b32_e32 v0, v0, v8, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 vcc_lo, s3, 1
+; MOVREL-NEXT:    v_cndmask_b32_e32 v1, v1, v8, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 vcc_lo, s3, 2
+; MOVREL-NEXT:    v_cndmask_b32_e32 v2, v2, v8, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 vcc_lo, s3, 3
+; MOVREL-NEXT:    v_cndmask_b32_e32 v3, v3, v8, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 vcc_lo, s3, 4
+; MOVREL-NEXT:    v_cndmask_b32_e32 v4, v4, v8, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 vcc_lo, s3, 5
+; MOVREL-NEXT:    v_cndmask_b32_e32 v5, v5, v8, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 vcc_lo, s3, 6
+; MOVREL-NEXT:    v_cndmask_b32_e32 v6, v6, v8, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 vcc_lo, s3, 7
+; MOVREL-NEXT:    v_cndmask_b32_e32 v7, v7, v8, vcc_lo
 ; MOVREL-NEXT:    ; return to shader part epilog
 entry:
   %insert = insertelement <8 x float> %vec, float %val, i32 %idx
   ret <8 x float> %insert
 }
 
 define amdgpu_ps <8 x float> @dyn_insertelement_v8f32_s_v_v(<8 x float> inreg %vec, float %val, i32 %idx) {
 ; GPRIDX-LABEL: dyn_insertelement_v8f32_s_v_v:
 ; GPRIDX:       ; %bb.0: ; %entry
 ; GPRIDX-NEXT:    s_mov_b32 s1, s3
 ; GPRIDX-NEXT:    s_mov_b32 s3, s5
 ; GPRIDX-NEXT:    s_mov_b32 s5, s7
 ; GPRIDX-NEXT:    s_mov_b32 s7, s9
 ; GPRIDX-NEXT:    s_mov_b32 s0, s2
 ; GPRIDX-NEXT:    s_mov_b32 s2, s4
 ; GPRIDX-NEXT:    s_mov_b32 s4, s6
 ; GPRIDX-NEXT:    s_mov_b32 s6, s8
-; GPRIDX-NEXT:    v_mov_b32_e32 v17, s7
-; GPRIDX-NEXT:    v_mov_b32_e32 v8, v0
-; GPRIDX-NEXT:    v_mov_b32_e32 v9, v1
-; GPRIDX-NEXT:    v_mov_b32_e32 v16, s6
-; GPRIDX-NEXT:    v_mov_b32_e32 v15, s5
-; GPRIDX-NEXT:    v_mov_b32_e32 v14, s4
-; GPRIDX-NEXT:    v_mov_b32_e32 v13, s3
-; GPRIDX-NEXT:    v_mov_b32_e32 v12, s2
-; GPRIDX-NEXT:    v_mov_b32_e32 v11, s1
-; GPRIDX-NEXT:    v_mov_b32_e32 v10, s0
-; GPRIDX-NEXT:    s_mov_b64 s[0:1], exec
-; GPRIDX-NEXT:  BB6_1: ; =>This Inner Loop Header: Depth=1
-; GPRIDX-NEXT:    v_readfirstlane_b32 s2, v9
-; GPRIDX-NEXT:    v_cmp_eq_u32_e32 vcc, s2, v9
-; GPRIDX-NEXT:    s_set_gpr_idx_on s2, gpr_idx(DST)
-; GPRIDX-NEXT:    v_mov_b32_e32 v0, v10
-; GPRIDX-NEXT:    v_mov_b32_e32 v1, v11
-; GPRIDX-NEXT:    v_mov_b32_e32 v2, v12
-; GPRIDX-NEXT:    v_mov_b32_e32 v3, v13
-; GPRIDX-NEXT:    v_mov_b32_e32 v4, v14
-; GPRIDX-NEXT:    v_mov_b32_e32 v5, v15
-; GPRIDX-NEXT:    v_mov_b32_e32 v6, v16
-; GPRIDX-NEXT:    v_mov_b32_e32 v7, v17
+; GPRIDX-NEXT:    v_mov_b32_e32 v15, s7
+; GPRIDX-NEXT:    v_mov_b32_e32 v8, s0
+; GPRIDX-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v1
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v8, v8, v0, vcc
+; GPRIDX-NEXT:    v_mov_b32_e32 v9, s1
+; GPRIDX-NEXT:    v_cmp_eq_u32_e32 vcc, 1, v1
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v9, v9, v0, vcc
+; GPRIDX-NEXT:    v_mov_b32_e32 v10, s2
+; GPRIDX-NEXT:    v_cmp_eq_u32_e32 vcc, 2, v1
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v2, v10, v0, vcc
+; GPRIDX-NEXT:    v_mov_b32_e32 v11, s3
+; GPRIDX-NEXT:    v_cmp_eq_u32_e32 vcc, 3, v1
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v3, v11, v0, vcc
+; GPRIDX-NEXT:    v_mov_b32_e32 v12, s4
+; GPRIDX-NEXT:    v_cmp_eq_u32_e32 vcc, 4, v1
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v4, v12, v0, vcc
+; GPRIDX-NEXT:    v_mov_b32_e32 v13, s5
+; GPRIDX-NEXT:    v_cmp_eq_u32_e32 vcc, 5, v1
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v5, v13, v0, vcc
+; GPRIDX-NEXT:    v_mov_b32_e32 v14, s6
+; GPRIDX-NEXT:    v_cmp_eq_u32_e32 vcc, 6, v1
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v6, v14, v0, vcc
+; GPRIDX-NEXT:    v_cmp_eq_u32_e32 vcc, 7, v1
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v7, v15, v0, vcc
 ; GPRIDX-NEXT:    v_mov_b32_e32 v0, v8
-; GPRIDX-NEXT:    s_set_gpr_idx_off
-; GPRIDX-NEXT:    s_and_saveexec_b64 vcc, vcc
-; GPRIDX-NEXT:    s_xor_b64 exec, exec, vcc
-; GPRIDX-NEXT:    s_cbranch_execnz BB6_1
-; GPRIDX-NEXT:  ; %bb.2:
-; GPRIDX-NEXT:    s_mov_b64 exec, s[0:1]
+; GPRIDX-NEXT:    v_mov_b32_e32 v1, v9
 ; GPRIDX-NEXT:    ; return to shader part epilog
 ;
 ; MOVREL-LABEL: dyn_insertelement_v8f32_s_v_v:
 ; MOVREL:       ; %bb.0: ; %entry
 ; MOVREL-NEXT:    s_mov_b32 s1, s3
 ; MOVREL-NEXT:    s_mov_b32 s3, s5
 ; MOVREL-NEXT:    s_mov_b32 s5, s7
 ; MOVREL-NEXT:    s_mov_b32 s7, s9
 ; MOVREL-NEXT:    s_mov_b32 s0, s2
 ; MOVREL-NEXT:    s_mov_b32 s2, s4
 ; MOVREL-NEXT:    s_mov_b32 s4, s6
 ; MOVREL-NEXT:    s_mov_b32 s6, s8
-; MOVREL-NEXT:    v_mov_b32_e32 v17, s7
-; MOVREL-NEXT:    v_mov_b32_e32 v15, s5
-; MOVREL-NEXT:    v_mov_b32_e32 v13, s3
-; MOVREL-NEXT:    v_mov_b32_e32 v14, s4
-; MOVREL-NEXT:    v_mov_b32_e32 v16, s6
-; MOVREL-NEXT:    v_mov_b32_e32 v12, s2
-; MOVREL-NEXT:    v_mov_b32_e32 v11, s1
-; MOVREL-NEXT:    v_mov_b32_e32 v10, s0
-; MOVREL-NEXT:    s_mov_b32 s0, exec_lo
-; MOVREL-NEXT:    ; implicit-def: $vcc_hi
-; MOVREL-NEXT:  BB6_1: ; =>This Inner Loop Header: Depth=1
-; MOVREL-NEXT:    v_readfirstlane_b32 s1, v1
-; MOVREL-NEXT:    v_mov_b32_e32 v2, v10
-; MOVREL-NEXT:    v_mov_b32_e32 v3, v11
-; MOVREL-NEXT:    v_mov_b32_e32 v4, v12
-; MOVREL-NEXT:    v_mov_b32_e32 v5, v13
-; MOVREL-NEXT:    v_cmp_eq_u32_e32 vcc_lo, s1, v1
-; MOVREL-NEXT:    s_mov_b32 m0, s1
-; MOVREL-NEXT:    v_mov_b32_e32 v6, v14
-; MOVREL-NEXT:    v_mov_b32_e32 v7, v15
-; MOVREL-NEXT:    v_mov_b32_e32 v8, v16
-; MOVREL-NEXT:    v_mov_b32_e32 v9, v17
-; MOVREL-NEXT:    v_movreld_b32_e32 v2, v0
-; MOVREL-NEXT:    s_and_saveexec_b32 vcc_lo, vcc_lo
-; MOVREL-NEXT:    s_xor_b32 exec_lo, exec_lo, vcc_lo
-; MOVREL-NEXT:    s_cbranch_execnz BB6_1
-; MOVREL-NEXT:  ; %bb.2:
-; MOVREL-NEXT:    s_mov_b32 exec_lo, s0
-; MOVREL-NEXT:    v_mov_b32_e32 v0, v2
-; MOVREL-NEXT:    v_mov_b32_e32 v1, v3
-; MOVREL-NEXT:    v_mov_b32_e32 v2, v4
-; MOVREL-NEXT:    v_mov_b32_e32 v3, v5
-; MOVREL-NEXT:    v_mov_b32_e32 v4, v6
-; MOVREL-NEXT:    v_mov_b32_e32 v5, v7
-; MOVREL-NEXT:    v_mov_b32_e32 v6, v8
-; MOVREL-NEXT:    v_mov_b32_e32 v7, v9
+; MOVREL-NEXT:    v_mov_b32_e32 v15, s7
+; MOVREL-NEXT:    v_mov_b32_e32 v8, s0
+; MOVREL-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 0, v1
+; MOVREL-NEXT:    v_mov_b32_e32 v9, s1
+; MOVREL-NEXT:    v_mov_b32_e32 v10, s2
+; MOVREL-NEXT:    v_mov_b32_e32 v11, s3
+; MOVREL-NEXT:    v_mov_b32_e32 v12, s4
+; MOVREL-NEXT:    v_cndmask_b32_e32 v8, v8, v0, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 1, v1
+; MOVREL-NEXT:    v_mov_b32_e32 v13, s5
+; MOVREL-NEXT:    v_mov_b32_e32 v14, s6
+; MOVREL-NEXT:    ; implicit-def: $vcc_hi
+; MOVREL-NEXT:    v_cndmask_b32_e32 v9, v9, v0, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 2, v1
+; MOVREL-NEXT:    v_cndmask_b32_e32 v2, v10, v0, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 3, v1
+; MOVREL-NEXT:    v_cndmask_b32_e32 v3, v11, v0, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 4, v1
+; MOVREL-NEXT:    v_cndmask_b32_e32 v4, v12, v0, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 5, v1
+; MOVREL-NEXT:    v_cndmask_b32_e32 v5, v13, v0, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 6, v1
+; MOVREL-NEXT:    v_cndmask_b32_e32 v6, v14, v0, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 7, v1
+; MOVREL-NEXT:    v_mov_b32_e32 v1, v9
+; MOVREL-NEXT:    v_cndmask_b32_e32 v7, v15, v0, vcc_lo
+; MOVREL-NEXT:    v_mov_b32_e32 v0, v8
 ; MOVREL-NEXT:    ; return to shader part epilog
 entry:
   %insert = insertelement <8 x float> %vec, float %val, i32 %idx
   ret <8 x float> %insert
 }
 
 define amdgpu_ps <8 x float> @dyn_insertelement_v8f32_v_s_v(<8 x float> %vec, float inreg %val, i32 %idx) {
 ; GPRIDX-LABEL: dyn_insertelement_v8f32_v_s_v:
 ; GPRIDX:       ; %bb.0: ; %entry
-; GPRIDX-NEXT:    s_mov_b64 s[0:1], exec
-; GPRIDX-NEXT:  BB7_1: ; =>This Inner Loop Header: Depth=1
-; GPRIDX-NEXT:    v_readfirstlane_b32 s3, v8
-; GPRIDX-NEXT:    v_cmp_eq_u32_e32 vcc, s3, v8
-; GPRIDX-NEXT:    s_set_gpr_idx_on s3, gpr_idx(DST)
-; GPRIDX-NEXT:    v_mov_b32_e32 v16, v7
-; GPRIDX-NEXT:    v_mov_b32_e32 v15, v6
-; GPRIDX-NEXT:    v_mov_b32_e32 v14, v5
-; GPRIDX-NEXT:    v_mov_b32_e32 v13, v4
-; GPRIDX-NEXT:    v_mov_b32_e32 v12, v3
-; GPRIDX-NEXT:    v_mov_b32_e32 v11, v2
-; GPRIDX-NEXT:    v_mov_b32_e32 v10, v1
-; GPRIDX-NEXT:    v_mov_b32_e32 v9, v0
 ; GPRIDX-NEXT:    v_mov_b32_e32 v9, s2
-; GPRIDX-NEXT:    s_set_gpr_idx_off
-; GPRIDX-NEXT:    s_and_saveexec_b64 vcc, vcc
-; GPRIDX-NEXT:    s_xor_b64 exec, exec, vcc
-; GPRIDX-NEXT:    s_cbranch_execnz BB7_1
-; GPRIDX-NEXT:  ; %bb.2:
-; GPRIDX-NEXT:    s_mov_b64 exec, s[0:1]
-; GPRIDX-NEXT:    v_mov_b32_e32 v0, v9
-; GPRIDX-NEXT:    v_mov_b32_e32 v1, v10
-; GPRIDX-NEXT:    v_mov_b32_e32 v2, v11
-; GPRIDX-NEXT:    v_mov_b32_e32 v3, v12
-; GPRIDX-NEXT:    v_mov_b32_e32 v4, v13
-; GPRIDX-NEXT:    v_mov_b32_e32 v5, v14
-; GPRIDX-NEXT:    v_mov_b32_e32 v6, v15
-; GPRIDX-NEXT:    v_mov_b32_e32 v7, v16
+; GPRIDX-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v8
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v0, v0, v9, vcc
+; GPRIDX-NEXT:    v_cmp_eq_u32_e32 vcc, 1, v8
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v1, v1, v9, vcc
+; GPRIDX-NEXT:    v_cmp_eq_u32_e32 vcc, 2, v8
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v2, v2, v9, vcc
+; GPRIDX-NEXT:    v_cmp_eq_u32_e32 vcc, 3, v8
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v3, v3, v9, vcc
+; GPRIDX-NEXT:    v_cmp_eq_u32_e32 vcc, 4, v8
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v4, v4, v9, vcc
+; GPRIDX-NEXT:    v_cmp_eq_u32_e32 vcc, 5, v8
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v5, v5, v9, vcc
+; GPRIDX-NEXT:    v_cmp_eq_u32_e32 vcc, 6, v8
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v6, v6, v9, vcc
+; GPRIDX-NEXT:    v_cmp_eq_u32_e32 vcc, 7, v8
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v7, v7, v9, vcc
 ; GPRIDX-NEXT:    ; return to shader part epilog
 ;
 ; MOVREL-LABEL: dyn_insertelement_v8f32_v_s_v:
 ; MOVREL:       ; %bb.0: ; %entry
-; MOVREL-NEXT:    s_mov_b32 s0, exec_lo
+; MOVREL-NEXT:    v_mov_b32_e32 v9, s2
+; MOVREL-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 0, v8
 ; MOVREL-NEXT:    ; implicit-def: $vcc_hi
-; MOVREL-NEXT:  BB7_1: ; =>This Inner Loop Header: Depth=1
-; MOVREL-NEXT:    v_readfirstlane_b32 s1, v8
-; MOVREL-NEXT:    v_mov_b32_e32 v16, v7
-; MOVREL-NEXT:    v_mov_b32_e32 v9, v0
-; MOVREL-NEXT:    v_mov_b32_e32 v15, v6
-; MOVREL-NEXT:    v_mov_b32_e32 v14, v5
-; MOVREL-NEXT:    v_cmp_eq_u32_e32 vcc_lo, s1, v8
-; MOVREL-NEXT:    s_mov_b32 m0, s1
-; MOVREL-NEXT:    v_mov_b32_e32 v13, v4
-; MOVREL-NEXT:    v_mov_b32_e32 v12, v3
-; MOVREL-NEXT:    v_mov_b32_e32 v11, v2
-; MOVREL-NEXT:    v_mov_b32_e32 v10, v1
-; MOVREL-NEXT:    v_movreld_b32_e32 v9, s2
-; MOVREL-NEXT:    s_and_saveexec_b32 vcc_lo, vcc_lo
-; MOVREL-NEXT:    s_xor_b32 exec_lo, exec_lo, vcc_lo
-; MOVREL-NEXT:    s_cbranch_execnz BB7_1
-; MOVREL-NEXT:  ; %bb.2:
-; MOVREL-NEXT:    s_mov_b32 exec_lo, s0
-; MOVREL-NEXT:    v_mov_b32_e32 v0, v9
-; MOVREL-NEXT:    v_mov_b32_e32 v1, v10
-; MOVREL-NEXT:    v_mov_b32_e32 v2, v11
-; MOVREL-NEXT:    v_mov_b32_e32 v3, v12
-; MOVREL-NEXT:    v_mov_b32_e32 v4, v13
-; MOVREL-NEXT:    v_mov_b32_e32 v5, v14
-; MOVREL-NEXT:    v_mov_b32_e32 v6, v15
-; MOVREL-NEXT:    v_mov_b32_e32 v7, v16
+; MOVREL-NEXT:    v_cndmask_b32_e32 v0, v0, v9, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 1, v8
+; MOVREL-NEXT:    v_cndmask_b32_e32 v1, v1, v9, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 2, v8
+; MOVREL-NEXT:    v_cndmask_b32_e32 v2, v2, v9, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 3, v8
+; MOVREL-NEXT:    v_cndmask_b32_e32 v3, v3, v9, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 4, v8
+; MOVREL-NEXT:    v_cndmask_b32_e32 v4, v4, v9, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 5, v8
+; MOVREL-NEXT:    v_cndmask_b32_e32 v5, v5, v9, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 6, v8
+; MOVREL-NEXT:    v_cndmask_b32_e32 v6, v6, v9, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 7, v8
+; MOVREL-NEXT:    v_cndmask_b32_e32 v7, v7, v9, vcc_lo
 ; MOVREL-NEXT:    ; return to shader part epilog
 entry:
   %insert = insertelement <8 x float> %vec, float %val, i32 %idx
   ret <8 x float> %insert
 }
 
 define amdgpu_ps <8 x float> @dyn_insertelement_v8f32_v_v_s(<8 x float> %vec, float %val, i32 inreg %idx) {
 ; GPRIDX-LABEL: dyn_insertelement_v8f32_v_v_s:
 ; GPRIDX:       ; %bb.0: ; %entry
-; GPRIDX-NEXT:    s_set_gpr_idx_on s2, gpr_idx(DST)
-; GPRIDX-NEXT:    v_mov_b32_e32 v0, v8
-; GPRIDX-NEXT:    s_set_gpr_idx_off
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 vcc, s2, 0
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v0, v0, v8, vcc
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 vcc, s2, 1
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v1, v1, v8, vcc
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 vcc, s2, 2
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v2, v2, v8, vcc
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 vcc, s2, 3
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v3, v3, v8, vcc
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 vcc, s2, 4
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v4, v4, v8, vcc
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 vcc, s2, 5
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v5, v5, v8, vcc
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 vcc, s2, 6
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v6, v6, v8, vcc
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 vcc, s2, 7
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v7, v7, v8, vcc
 ; GPRIDX-NEXT:    ; return to shader part epilog
 ;
 ; MOVREL-LABEL: dyn_insertelement_v8f32_v_v_s:
 ; MOVREL:       ; %bb.0: ; %entry
-; MOVREL-NEXT:    s_mov_b32 m0, s2
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 vcc_lo, s2, 0
 ; MOVREL-NEXT:    ; implicit-def: $vcc_hi
-; MOVREL-NEXT:    v_movreld_b32_e32 v0, v8
+; MOVREL-NEXT:    v_cndmask_b32_e32 v0, v0, v8, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 vcc_lo, s2, 1
+; MOVREL-NEXT:    v_cndmask_b32_e32 v1, v1, v8, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 vcc_lo, s2, 2
+; MOVREL-NEXT:    v_cndmask_b32_e32 v2, v2, v8, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 vcc_lo, s2, 3
+; MOVREL-NEXT:    v_cndmask_b32_e32 v3, v3, v8, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 vcc_lo, s2, 4
+; MOVREL-NEXT:    v_cndmask_b32_e32 v4, v4, v8, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 vcc_lo, s2, 5
+; MOVREL-NEXT:    v_cndmask_b32_e32 v5, v5, v8, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 vcc_lo, s2, 6
+; MOVREL-NEXT:    v_cndmask_b32_e32 v6, v6, v8, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 vcc_lo, s2, 7
+; MOVREL-NEXT:    v_cndmask_b32_e32 v7, v7, v8, vcc_lo
 ; MOVREL-NEXT:    ; return to shader part epilog
 entry:
   %insert = insertelement <8 x float> %vec, float %val, i32 %idx
   ret <8 x float> %insert
 }
 
 define amdgpu_ps <8 x float> @dyn_insertelement_v8p3i8_v_v_s(<8 x i8 addrspace(3)*> %vec, i8 addrspace(3)* %val, i32 inreg %idx) {
 ; GPRIDX-LABEL: dyn_insertelement_v8p3i8_v_v_s:
 ; GPRIDX:       ; %bb.0: ; %entry
-; GPRIDX-NEXT:    s_set_gpr_idx_on s2, gpr_idx(DST)
-; GPRIDX-NEXT:    v_mov_b32_e32 v0, v8
-; GPRIDX-NEXT:    s_set_gpr_idx_off
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 vcc, s2, 0
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v0, v0, v8, vcc
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 vcc, s2, 1
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v1, v1, v8, vcc
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 vcc, s2, 2
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v2, v2, v8, vcc
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 vcc, s2, 3
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v3, v3, v8, vcc
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 vcc, s2, 4
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v4, v4, v8, vcc
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 vcc, s2, 5
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v5, v5, v8, vcc
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 vcc, s2, 6
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v6, v6, v8, vcc
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 vcc, s2, 7
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v7, v7, v8, vcc
 ; GPRIDX-NEXT:    ; return to shader part epilog
 ;
 ; MOVREL-LABEL: dyn_insertelement_v8p3i8_v_v_s:
 ; MOVREL:       ; %bb.0: ; %entry
-; MOVREL-NEXT:    s_mov_b32 m0, s2
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 vcc_lo, s2, 0
 ; MOVREL-NEXT:    ; implicit-def: $vcc_hi
-; MOVREL-NEXT:    v_movreld_b32_e32 v0, v8
+; MOVREL-NEXT:    v_cndmask_b32_e32 v0, v0, v8, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 vcc_lo, s2, 1
+; MOVREL-NEXT:    v_cndmask_b32_e32 v1, v1, v8, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 vcc_lo, s2, 2
+; MOVREL-NEXT:    v_cndmask_b32_e32 v2, v2, v8, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 vcc_lo, s2, 3
+; MOVREL-NEXT:    v_cndmask_b32_e32 v3, v3, v8, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 vcc_lo, s2, 4
+; MOVREL-NEXT:    v_cndmask_b32_e32 v4, v4, v8, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 vcc_lo, s2, 5
+; MOVREL-NEXT:    v_cndmask_b32_e32 v5, v5, v8, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 vcc_lo, s2, 6
+; MOVREL-NEXT:    v_cndmask_b32_e32 v6, v6, v8, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 vcc_lo, s2, 7
+; MOVREL-NEXT:    v_cndmask_b32_e32 v7, v7, v8, vcc_lo
 ; MOVREL-NEXT:    ; return to shader part epilog
 entry:
   %insert = insertelement <8 x i8 addrspace(3)*> %vec, i8 addrspace(3)* %val, i32 %idx
   %cast.0 = ptrtoint <8 x i8 addrspace(3)*> %insert to <8 x i32>
   %cast.1 = bitcast <8 x i32> %cast.0 to <8 x float>
   ret <8 x float> %cast.1
 }
 
 define amdgpu_ps <8 x float> @dyn_insertelement_v8f32_v_v_v(<8 x float> %vec, float %val, i32 %idx) {
 ; GPRIDX-LABEL: dyn_insertelement_v8f32_v_v_v:
 ; GPRIDX:       ; %bb.0: ; %entry
-; GPRIDX-NEXT:    s_mov_b64 s[0:1], exec
-; GPRIDX-NEXT:  BB10_1: ; =>This Inner Loop Header: Depth=1
-; GPRIDX-NEXT:    v_readfirstlane_b32 s2, v9
-; GPRIDX-NEXT:    v_cmp_eq_u32_e32 vcc, s2, v9
-; GPRIDX-NEXT:    s_set_gpr_idx_on s2, gpr_idx(DST)
-; GPRIDX-NEXT:    v_mov_b32_e32 v17, v7
-; GPRIDX-NEXT:    v_mov_b32_e32 v16, v6
-; GPRIDX-NEXT:    v_mov_b32_e32 v15, v5
-; GPRIDX-NEXT:    v_mov_b32_e32 v14, v4
-; GPRIDX-NEXT:    v_mov_b32_e32 v13, v3
-; GPRIDX-NEXT:    v_mov_b32_e32 v12, v2
-; GPRIDX-NEXT:    v_mov_b32_e32 v11, v1
-; GPRIDX-NEXT:    v_mov_b32_e32 v10, v0
-; GPRIDX-NEXT:    v_mov_b32_e32 v10, v8
-; GPRIDX-NEXT:    s_set_gpr_idx_off
-; GPRIDX-NEXT:    s_and_saveexec_b64 vcc, vcc
-; GPRIDX-NEXT:    s_xor_b64 exec, exec, vcc
-; GPRIDX-NEXT:    s_cbranch_execnz BB10_1
-; GPRIDX-NEXT:  ; %bb.2:
-; GPRIDX-NEXT:    s_mov_b64 exec, s[0:1]
-; GPRIDX-NEXT:    v_mov_b32_e32 v0, v10
-; GPRIDX-NEXT:    v_mov_b32_e32 v1, v11
-; GPRIDX-NEXT:    v_mov_b32_e32 v2, v12
-; GPRIDX-NEXT:    v_mov_b32_e32 v3, v13
-; GPRIDX-NEXT:    v_mov_b32_e32 v4, v14
-; GPRIDX-NEXT:    v_mov_b32_e32 v5, v15
-; GPRIDX-NEXT:    v_mov_b32_e32 v6, v16
-; GPRIDX-NEXT:    v_mov_b32_e32 v7, v17
+; GPRIDX-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v9
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v0, v0, v8, vcc
+; GPRIDX-NEXT:    v_cmp_eq_u32_e32 vcc, 1, v9
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v1, v1, v8, vcc
+; GPRIDX-NEXT:    v_cmp_eq_u32_e32 vcc, 2, v9
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v2, v2, v8, vcc
+; GPRIDX-NEXT:    v_cmp_eq_u32_e32 vcc, 3, v9
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v3, v3, v8, vcc
+; GPRIDX-NEXT:    v_cmp_eq_u32_e32 vcc, 4, v9
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v4, v4, v8, vcc
+; GPRIDX-NEXT:    v_cmp_eq_u32_e32 vcc, 5, v9
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v5, v5, v8, vcc
+; GPRIDX-NEXT:    v_cmp_eq_u32_e32 vcc, 6, v9
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v6, v6, v8, vcc
+; GPRIDX-NEXT:    v_cmp_eq_u32_e32 vcc, 7, v9
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v7, v7, v8, vcc
 ; GPRIDX-NEXT:    ; return to shader part epilog
 ;
 ; MOVREL-LABEL: dyn_insertelement_v8f32_v_v_v:
 ; MOVREL:       ; %bb.0: ; %entry
-; MOVREL-NEXT:    s_mov_b32 s0, exec_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 0, v9
 ; MOVREL-NEXT:    ; implicit-def: $vcc_hi
-; MOVREL-NEXT:  BB10_1: ; =>This Inner Loop Header: Depth=1
-; MOVREL-NEXT:    v_readfirstlane_b32 s1, v9
-; MOVREL-NEXT:    v_mov_b32_e32 v17, v7
-; MOVREL-NEXT:    v_mov_b32_e32 v10, v0
-; MOVREL-NEXT:    v_mov_b32_e32 v16, v6
-; MOVREL-NEXT:    v_mov_b32_e32 v15, v5
-; MOVREL-NEXT:    v_cmp_eq_u32_e32 vcc_lo, s1, v9
-; MOVREL-NEXT:    s_mov_b32 m0, s1
-; MOVREL-NEXT:    v_mov_b32_e32 v14, v4
-; MOVREL-NEXT:    v_mov_b32_e32 v13, v3
-; MOVREL-NEXT:    v_mov_b32_e32 v12, v2
-; MOVREL-NEXT:    v_mov_b32_e32 v11, v1
-; MOVREL-NEXT:    v_movreld_b32_e32 v10, v8
-; MOVREL-NEXT:    s_and_saveexec_b32 vcc_lo, vcc_lo
-; MOVREL-NEXT:    s_xor_b32 exec_lo, exec_lo, vcc_lo
-; MOVREL-NEXT:    s_cbranch_execnz BB10_1
-; MOVREL-NEXT:  ; %bb.2:
-; MOVREL-NEXT:    s_mov_b32 exec_lo, s0
-; MOVREL-NEXT:    v_mov_b32_e32 v0, v10
-; MOVREL-NEXT:    v_mov_b32_e32 v1, v11
-; MOVREL-NEXT:    v_mov_b32_e32 v2, v12
-; MOVREL-NEXT:    v_mov_b32_e32 v3, v13
-; MOVREL-NEXT:    v_mov_b32_e32 v4, v14
-; MOVREL-NEXT:    v_mov_b32_e32 v5, v15
-; MOVREL-NEXT:    v_mov_b32_e32 v6, v16
-; MOVREL-NEXT:    v_mov_b32_e32 v7, v17
+; MOVREL-NEXT:    v_cndmask_b32_e32 v0, v0, v8, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 1, v9
+; MOVREL-NEXT:    v_cndmask_b32_e32 v1, v1, v8, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 2, v9
+; MOVREL-NEXT:    v_cndmask_b32_e32 v2, v2, v8, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 3, v9
+; MOVREL-NEXT:    v_cndmask_b32_e32 v3, v3, v8, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 4, v9
+; MOVREL-NEXT:    v_cndmask_b32_e32 v4, v4, v8, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 5, v9
+; MOVREL-NEXT:    v_cndmask_b32_e32 v5, v5, v8, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 6, v9
+; MOVREL-NEXT:    v_cndmask_b32_e32 v6, v6, v8, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 7, v9
+; MOVREL-NEXT:    v_cndmask_b32_e32 v7, v7, v8, vcc_lo
 ; MOVREL-NEXT:    ; return to shader part epilog
 entry:
   %insert = insertelement <8 x float> %vec, float %val, i32 %idx
   ret <8 x float> %insert
 }
 
 define amdgpu_ps <8 x i64> @dyn_insertelement_v8i64_s_s_s(<8 x i64> inreg %vec, i64 inreg %val, i32 inreg %idx) {
 ; GPRIDX-LABEL: dyn_insertelement_v8i64_s_s_s:
entry:
   ret <8 x i8 addrspace(1)*> %insert
 }
 
 define void @dyn_insertelement_v8f64_const_s_v_v(double %val, i32 %idx) {
 ; GPRIDX-LABEL: dyn_insertelement_v8f64_const_s_v_v:
 ; GPRIDX:       ; %bb.0: ; %entry
 ; GPRIDX-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
 ; GPRIDX-NEXT:    s_mov_b32 s18, 0
+; GPRIDX-NEXT:    s_mov_b64 s[4:5], 1.0
 ; GPRIDX-NEXT:    s_mov_b32 s19, 0x40200000
 ; GPRIDX-NEXT:    s_mov_b32 s17, 0x401c0000
 ; GPRIDX-NEXT:    s_mov_b32 s16, s18
 ; GPRIDX-NEXT:    s_mov_b32 s15, 0x40180000
 ; GPRIDX-NEXT:    s_mov_b32 s14, s18
 ; GPRIDX-NEXT:    s_mov_b32 s13, 0x40140000
 ; GPRIDX-NEXT:    s_mov_b32 s12, s18
 ; GPRIDX-NEXT:    s_mov_b64 s[10:11], 4.0
 ; GPRIDX-NEXT:    s_mov_b32 s9, 0x40080000
 ; GPRIDX-NEXT:    s_mov_b32 s8, s18
 ; GPRIDX-NEXT:    s_mov_b64 s[6:7], 2.0
-; GPRIDX-NEXT:    s_mov_b64 s[4:5], 1.0
-; GPRIDX-NEXT:    v_mov_b32_e32 v34, s19
-; GPRIDX-NEXT:    v_mov_b32_e32 v33, s18
-; GPRIDX-NEXT:    v_mov_b32_e32 v32, s17
-; GPRIDX-NEXT:    v_mov_b32_e32 v31, s16
-; GPRIDX-NEXT:    v_mov_b32_e32 v30, s15
-; GPRIDX-NEXT:    v_mov_b32_e32 v29, s14
-; GPRIDX-NEXT:    v_mov_b32_e32 v28, s13
-; GPRIDX-NEXT:    v_mov_b32_e32 v27, s12
-; GPRIDX-NEXT:    v_mov_b32_e32 v26, s11
-; GPRIDX-NEXT:    v_mov_b32_e32 v25, s10
-; GPRIDX-NEXT:    v_mov_b32_e32 v24, s9
-; GPRIDX-NEXT:    v_mov_b32_e32 v23, s8
-; GPRIDX-NEXT:    v_mov_b32_e32 v22, s7
-; GPRIDX-NEXT:    v_mov_b32_e32 v21, s6
-; GPRIDX-NEXT:    v_mov_b32_e32 v20, s5
-; GPRIDX-NEXT:    v_mov_b32_e32 v19, s4
-; GPRIDX-NEXT:    s_mov_b64 s[4:5], exec
-; GPRIDX-NEXT:  BB13_1: ; =>This Inner Loop Header: Depth=1
-; GPRIDX-NEXT:    v_readfirstlane_b32 s6, v2
-; GPRIDX-NEXT:    s_lshl_b32 s7, s6, 1
-; GPRIDX-NEXT:    v_cmp_eq_u32_e32 vcc, s6, v2
-; GPRIDX-NEXT:    s_set_gpr_idx_on s7, gpr_idx(DST)
-; GPRIDX-NEXT:    v_mov_b32_e32 v3, v19
-; GPRIDX-NEXT:    v_mov_b32_e32 v4, v20
-; GPRIDX-NEXT:    v_mov_b32_e32 v5, v21
-; GPRIDX-NEXT:    v_mov_b32_e32 v6, v22
-; GPRIDX-NEXT:    v_mov_b32_e32 v7, v23
-; GPRIDX-NEXT:    v_mov_b32_e32 v8, v24
-; GPRIDX-NEXT:    v_mov_b32_e32 v9, v25
-; GPRIDX-NEXT:    v_mov_b32_e32 v10, v26
-; GPRIDX-NEXT:    v_mov_b32_e32 v11, v27
-; GPRIDX-NEXT:    v_mov_b32_e32 v12, v28
-; GPRIDX-NEXT:    v_mov_b32_e32 v13, v29
-; GPRIDX-NEXT:    v_mov_b32_e32 v14, v30
-; GPRIDX-NEXT:    v_mov_b32_e32 v15, v31
-; GPRIDX-NEXT:    v_mov_b32_e32 v16, v32
-; GPRIDX-NEXT:    v_mov_b32_e32 v17, v33
-; GPRIDX-NEXT:    v_mov_b32_e32 v18, v34
-; GPRIDX-NEXT:    v_mov_b32_e32 v3, v0
-; GPRIDX-NEXT:    s_set_gpr_idx_off
-; GPRIDX-NEXT:    s_set_gpr_idx_on s7, gpr_idx(DST)
-; GPRIDX-NEXT:    v_mov_b32_e32 v4, v1
-; GPRIDX-NEXT:    s_set_gpr_idx_off
-; GPRIDX-NEXT:    s_and_saveexec_b64 vcc, vcc
-; GPRIDX-NEXT:    s_xor_b64 exec, exec, vcc
-; GPRIDX-NEXT:    s_cbranch_execnz BB13_1
-; GPRIDX-NEXT:  ; %bb.2:
-; GPRIDX-NEXT:    s_mov_b64 exec, s[4:5]
+; GPRIDX-NEXT:    v_mov_b32_e32 v3, s4
+; GPRIDX-NEXT:    v_mov_b32_e32 v4, s5
+; GPRIDX-NEXT:    v_mov_b32_e32 v5, s6
+; GPRIDX-NEXT:    v_cmp_eq_u32_e32 vcc, 1, v2
+; GPRIDX-NEXT:    v_mov_b32_e32 v6, s7
+; GPRIDX-NEXT:    v_mov_b32_e32 v7, s8
+; GPRIDX-NEXT:    v_mov_b32_e32 v8, s9
+; GPRIDX-NEXT:    v_mov_b32_e32 v9, s10
+; GPRIDX-NEXT:    v_mov_b32_e32 v10, s11
+; GPRIDX-NEXT:    v_mov_b32_e32 v11, s12
+; GPRIDX-NEXT:    v_mov_b32_e32 v12, s13
+; GPRIDX-NEXT:    v_mov_b32_e32 v13, s14
+; GPRIDX-NEXT:    v_mov_b32_e32 v14, s15
+; GPRIDX-NEXT:    v_mov_b32_e32 v15, s16
+; GPRIDX-NEXT:    v_mov_b32_e32 v16, s17
+; GPRIDX-NEXT:    v_mov_b32_e32 v17, s18
+; GPRIDX-NEXT:    v_mov_b32_e32 v18, s19
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 s[16:17], 0, v2
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 s[4:5], 2, v2
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 s[6:7], 3, v2
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 s[8:9], 4, v2
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 s[10:11], 5, v2
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 s[12:13], 6, v2
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 s[14:15], 7, v2
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v3, v3, v0, s[16:17]
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v5, v5, v0, vcc
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v4, v4, v1, s[16:17]
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v6, v6, v1, vcc
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v7, v7, v0, s[4:5]
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v9, v9, v0, s[6:7]
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v11, v11, v0, s[8:9]
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v13, v13, v0, s[10:11]
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v15, v15, v0, s[12:13]
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v17, v17, v0, s[14:15]
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v8, v8, v1, s[4:5]
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v10, v10, v1, s[6:7]
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v12, v12, v1, s[8:9]
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v14, v14, v1, s[10:11]
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v16, v16, v1, s[12:13]
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v18, v18, v1, s[14:15]
 ; GPRIDX-NEXT:    global_store_dwordx4 v[0:1], v[3:6], off
 ; GPRIDX-NEXT:    global_store_dwordx4 v[0:1], v[7:10], off
 ; GPRIDX-NEXT:    global_store_dwordx4 v[0:1], v[11:14], off
 ; GPRIDX-NEXT:    global_store_dwordx4 v[0:1], v[15:18], off
 ; GPRIDX-NEXT:    s_waitcnt vmcnt(0)
 ; GPRIDX-NEXT:    s_setpc_b64 s[30:31]
 ;
 ; MOVREL-LABEL: dyn_insertelement_v8f64_const_s_v_v:
 ; MOVREL:       ; %bb.0: ; %entry
 ; MOVREL-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
 ; MOVREL-NEXT:    s_waitcnt_vscnt null, 0x0
 ; MOVREL-NEXT:    s_mov_b32 s18, 0
+; MOVREL-NEXT:    s_mov_b64 s[4:5], 1.0
 ; MOVREL-NEXT:    s_mov_b32 s19, 0x40200000
 ; MOVREL-NEXT:    s_mov_b32 s17, 0x401c0000
 ; MOVREL-NEXT:    s_mov_b32 s15, 0x40180000
-; MOVREL-NEXT:    s_mov_b32 s13, 0x40140000
 ; MOVREL-NEXT:    s_mov_b32 s16, s18
 ; MOVREL-NEXT:    s_mov_b32 s14, s18
+; MOVREL-NEXT:    s_mov_b32 s13, 0x40140000
 ; MOVREL-NEXT:    s_mov_b32 s12, s18
 ; MOVREL-NEXT:    s_mov_b64 s[10:11], 4.0
 ; MOVREL-NEXT:    s_mov_b32 s9, 0x40080000
 ; MOVREL-NEXT:    s_mov_b32 s8, s18
 ; MOVREL-NEXT:    s_mov_b64 s[6:7], 2.0
-; MOVREL-NEXT:    s_mov_b64 s[4:5], 1.0
-; MOVREL-NEXT:    v_mov_b32_e32 v34, s19
-; MOVREL-NEXT:    v_mov_b32_e32 v33, s18
-; MOVREL-NEXT:    v_mov_b32_e32 v32, s17
-; MOVREL-NEXT:    v_mov_b32_e32 v31, s16
-; MOVREL-NEXT:    v_mov_b32_e32 v30, s15
-; MOVREL-NEXT:    v_mov_b32_e32 v29, s14
-; MOVREL-NEXT:    v_mov_b32_e32 v28, s13
-; MOVREL-NEXT:    v_mov_b32_e32 v27, s12
-; MOVREL-NEXT:    v_mov_b32_e32 v26, s11
-; MOVREL-NEXT:    v_mov_b32_e32 v25, s10
-; MOVREL-NEXT:    v_mov_b32_e32 v24, s9
-; MOVREL-NEXT:    v_mov_b32_e32 v23, s8
-; MOVREL-NEXT:    v_mov_b32_e32 v22, s7
-; MOVREL-NEXT:    v_mov_b32_e32 v21, s6
-; MOVREL-NEXT:    v_mov_b32_e32 v20, s5
-; MOVREL-NEXT:    v_mov_b32_e32 v19, s4
-; MOVREL-NEXT:    s_mov_b32 s4, exec_lo
+; MOVREL-NEXT:    v_mov_b32_e32 v3, s4
+; MOVREL-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 0, v2
+; MOVREL-NEXT:    v_mov_b32_e32 v4, s5
+; MOVREL-NEXT:    v_mov_b32_e32 v7, s8
+; MOVREL-NEXT:    v_mov_b32_e32 v5, s6
+; MOVREL-NEXT:    v_mov_b32_e32 v6, s7
+; MOVREL-NEXT:    v_mov_b32_e32 v8, s9
+; MOVREL-NEXT:    v_mov_b32_e32 v9, s10
+; MOVREL-NEXT:    v_mov_b32_e32 v10, s11
+; MOVREL-NEXT:    v_mov_b32_e32 v11, s12
+; MOVREL-NEXT:    v_mov_b32_e32 v12, s13
+; MOVREL-NEXT:    v_mov_b32_e32 v13, s14
+; MOVREL-NEXT:    v_mov_b32_e32 v14, s15
+; MOVREL-NEXT:    v_mov_b32_e32 v15, s16
+; MOVREL-NEXT:    v_mov_b32_e32 v16, s17
+; MOVREL-NEXT:    v_mov_b32_e32 v17, s18
+; MOVREL-NEXT:    v_mov_b32_e32 v18, s19
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 s4, 1, v2
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 s6, 4, v2
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 s7, 5, v2
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 s5, 3, v2
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 s8, 6, v2
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 s9, 7, v2
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 s10, 2, v2
+; MOVREL-NEXT:    v_cndmask_b32_e32 v3, v3, v0, vcc_lo
+; MOVREL-NEXT:    v_cndmask_b32_e64 v5, v5, v0, s4
+; MOVREL-NEXT:    v_cndmask_b32_e64 v11, v11, v0, s6
+; MOVREL-NEXT:    v_cndmask_b32_e64 v13, v13, v0, s7
+; MOVREL-NEXT:    v_cndmask_b32_e32 v4, v4, v1, vcc_lo
+; MOVREL-NEXT:    v_cndmask_b32_e64 v6, v6, v1, s4
+; MOVREL-NEXT:    v_cndmask_b32_e64 v12, v12, v1, s6
+; MOVREL-NEXT:    v_cndmask_b32_e64 v14, v14, v1, s7
+; MOVREL-NEXT:    v_cndmask_b32_e64 v7, v7, v0, s10
+; MOVREL-NEXT:    v_cndmask_b32_e64 v9, v9, v0, s5
+; MOVREL-NEXT:    v_cndmask_b32_e64 v15, v15, v0, s8
+; MOVREL-NEXT:    v_cndmask_b32_e64 v17, v17, v0, s9
+; MOVREL-NEXT:    v_cndmask_b32_e64 v8, v8, v1, s10
+; MOVREL-NEXT:    v_cndmask_b32_e64 v10, v10, v1, s5
+; MOVREL-NEXT:    v_cndmask_b32_e64 v16, v16, v1, s8
+; MOVREL-NEXT:    v_cndmask_b32_e64 v18, v18, v1, s9
 ; MOVREL-NEXT:    ; implicit-def: $vcc_hi
-; MOVREL-NEXT:  BB13_1: ; =>This Inner Loop Header: Depth=1
-; MOVREL-NEXT:    v_readfirstlane_b32 s5, v2
-; MOVREL-NEXT:    v_mov_b32_e32 v3, v19
-; MOVREL-NEXT:    v_mov_b32_e32 v4, v20
-; MOVREL-NEXT:    v_mov_b32_e32 v5, v21
-; MOVREL-NEXT:    v_mov_b32_e32 v6, v22
-; MOVREL-NEXT:    s_lshl_b32 m0, s5, 1
-; MOVREL-NEXT:    v_cmp_eq_u32_e32 vcc_lo, s5, v2
-; MOVREL-NEXT:    v_mov_b32_e32 v7, v23
-; MOVREL-NEXT:    v_mov_b32_e32 v8, v24
-; MOVREL-NEXT:    v_mov_b32_e32 v9, v25
-; MOVREL-NEXT:    v_mov_b32_e32 v10, v26
-; MOVREL-NEXT:    v_mov_b32_e32 v11, v27
-; MOVREL-NEXT:    v_mov_b32_e32 v12, v28
-; MOVREL-NEXT:    v_mov_b32_e32 v13, v29
-; MOVREL-NEXT:    v_mov_b32_e32 v14, v30
-; MOVREL-NEXT:    v_mov_b32_e32 v15, v31
-; MOVREL-NEXT:    v_mov_b32_e32 v16, v32
-; MOVREL-NEXT:    v_mov_b32_e32 v17, v33
-; MOVREL-NEXT:    v_mov_b32_e32 v18, v34
-; MOVREL-NEXT:    v_movreld_b32_e32 v3, v0
-; MOVREL-NEXT:    v_movreld_b32_e32 v4, v1
-; MOVREL-NEXT:    s_and_saveexec_b32 vcc_lo, vcc_lo
-; MOVREL-NEXT:    s_xor_b32 exec_lo, exec_lo, vcc_lo
-; MOVREL-NEXT:    s_cbranch_execnz BB13_1
-; MOVREL-NEXT:  ; %bb.2:
-; MOVREL-NEXT:    s_mov_b32 exec_lo, s4
 ; MOVREL-NEXT:    global_store_dwordx4 v[0:1], v[3:6], off
 ; MOVREL-NEXT:    global_store_dwordx4 v[0:1], v[7:10], off
 ; MOVREL-NEXT:    global_store_dwordx4 v[0:1], v[11:14], off
 ; MOVREL-NEXT:    global_store_dwordx4 v[0:1], v[15:18], off
 ; MOVREL-NEXT:    s_waitcnt_vscnt null, 0x0
 ; MOVREL-NEXT:    s_setpc_b64 s[30:31]
 entry:
   %insert = insertelement <8 x double> <double 1.0, double 2.0, double 3.0, double 4.0, double 5.0, double 6.0, double 7.0, double 8.0>, double %val, i32 %idx

 ; GPRIDX-NEXT:    s_mov_b32 s0, s2
 ; GPRIDX-NEXT:    s_mov_b32 s2, s4
 ; GPRIDX-NEXT:    s_mov_b32 s4, s6
 ; GPRIDX-NEXT:    s_mov_b32 s6, s8
 ; GPRIDX-NEXT:    s_mov_b32 s8, s10
 ; GPRIDX-NEXT:    s_mov_b32 s10, s12
 ; GPRIDX-NEXT:    s_mov_b32 s12, s14
 ; GPRIDX-NEXT:    s_mov_b32 s14, s16
-; GPRIDX-NEXT:    v_mov_b32_e32 v32, s15
-; GPRIDX-NEXT:    v_mov_b32_e32 v31, s14
-; GPRIDX-NEXT:    v_mov_b32_e32 v30, s13
-; GPRIDX-NEXT:    v_mov_b32_e32 v29, s12
-; GPRIDX-NEXT:    v_mov_b32_e32 v28, s11
-; GPRIDX-NEXT:    v_mov_b32_e32 v27, s10
-; GPRIDX-NEXT:    v_mov_b32_e32 v26, s9
-; GPRIDX-NEXT:    v_mov_b32_e32 v25, s8
-; GPRIDX-NEXT:    v_mov_b32_e32 v24, s7
-; GPRIDX-NEXT:    v_mov_b32_e32 v23, s6
-; GPRIDX-NEXT:    v_mov_b32_e32 v22, s5
-; GPRIDX-NEXT:    v_mov_b32_e32 v21, s4
-; GPRIDX-NEXT:    v_mov_b32_e32 v20, s3
-; GPRIDX-NEXT:    v_mov_b32_e32 v19, s2
-; GPRIDX-NEXT:    v_mov_b32_e32 v18, s1
-; GPRIDX-NEXT:    v_mov_b32_e32 v17, s0
-; GPRIDX-NEXT:    s_mov_b64 s[0:1], exec
-; GPRIDX-NEXT:  BB14_1: ; =>This Inner Loop Header: Depth=1
-; GPRIDX-NEXT:    v_readfirstlane_b32 s2, v0
-; GPRIDX-NEXT:    s_lshl_b32 s3, s2, 1
-; GPRIDX-NEXT:    v_cmp_eq_u32_e32 vcc, s2, v0
-; GPRIDX-NEXT:    s_set_gpr_idx_on s3, gpr_idx(DST)
-; GPRIDX-NEXT:    v_mov_b32_e32 v1, v17
-; GPRIDX-NEXT:    v_mov_b32_e32 v2, v18
-; GPRIDX-NEXT:    v_mov_b32_e32 v3, v19
-; GPRIDX-NEXT:    v_mov_b32_e32 v4, v20
-; GPRIDX-NEXT:    v_mov_b32_e32 v5, v21
-; GPRIDX-NEXT:    v_mov_b32_e32 v6, v22
-; GPRIDX-NEXT:    v_mov_b32_e32 v7, v23
-; GPRIDX-NEXT:    v_mov_b32_e32 v8, v24
-; GPRIDX-NEXT:    v_mov_b32_e32 v9, v25
-; GPRIDX-NEXT:    v_mov_b32_e32 v10, v26
-; GPRIDX-NEXT:    v_mov_b32_e32 v11, v27
-; GPRIDX-NEXT:    v_mov_b32_e32 v12, v28
-; GPRIDX-NEXT:    v_mov_b32_e32 v13, v29
-; GPRIDX-NEXT:    v_mov_b32_e32 v14, v30
-; GPRIDX-NEXT:    v_mov_b32_e32 v15, v31
-; GPRIDX-NEXT:    v_mov_b32_e32 v16, v32
-; GPRIDX-NEXT:    v_mov_b32_e32 v1, s18
-; GPRIDX-NEXT:    s_set_gpr_idx_off
-; GPRIDX-NEXT:    s_set_gpr_idx_on s3, gpr_idx(DST)
-; GPRIDX-NEXT:    v_mov_b32_e32 v2, s19
-; GPRIDX-NEXT:    s_set_gpr_idx_off
-; GPRIDX-NEXT:    s_and_saveexec_b64 vcc, vcc
-; GPRIDX-NEXT:    s_xor_b64 exec, exec, vcc
-; GPRIDX-NEXT:    s_cbranch_execnz BB14_1
-; GPRIDX-NEXT:  ; %bb.2:
-; GPRIDX-NEXT:    s_mov_b64 exec, s[0:1]
+; GPRIDX-NEXT:    v_mov_b32_e32 v16, s15
+; GPRIDX-NEXT:    v_mov_b32_e32 v15, s14
+; GPRIDX-NEXT:    v_mov_b32_e32 v14, s13
+; GPRIDX-NEXT:    v_mov_b32_e32 v13, s12
+; GPRIDX-NEXT:    v_mov_b32_e32 v12, s11
+; GPRIDX-NEXT:    v_mov_b32_e32 v11, s10
+; GPRIDX-NEXT:    v_mov_b32_e32 v10, s9
+; GPRIDX-NEXT:    v_mov_b32_e32 v9, s8
+; GPRIDX-NEXT:    v_mov_b32_e32 v8, s7
+; GPRIDX-NEXT:    v_mov_b32_e32 v7, s6
+; GPRIDX-NEXT:    v_mov_b32_e32 v6, s5
+; GPRIDX-NEXT:    v_mov_b32_e32 v5, s4
+; GPRIDX-NEXT:    v_mov_b32_e32 v4, s3
+; GPRIDX-NEXT:    v_mov_b32_e32 v3, s2
+; GPRIDX-NEXT:    v_mov_b32_e32 v2, s1
+; GPRIDX-NEXT:    v_mov_b32_e32 v1, s0
+; GPRIDX-NEXT:    v_mov_b32_e32 v17, s18
+; GPRIDX-NEXT:    v_cmp_eq_u32_e32 vcc, 1, v0
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 s[0:1], 2, v0
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 s[2:3], 3, v0
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 s[4:5], 4, v0
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 s[6:7], 5, v0
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 s[8:9], 6, v0
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 s[10:11], 7, v0
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 s[12:13], 0, v0
+; GPRIDX-NEXT:    v_mov_b32_e32 v0, s19
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v1, v1, v17, s[12:13]
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v3, v3, v17, vcc
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v2, v2, v0, s[12:13]
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v4, v4, v0, vcc
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v5, v5, v17, s[0:1]
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v7, v7, v17, s[2:3]
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v9, v9, v17, s[4:5]
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v11, v11, v17, s[6:7]
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v13, v13, v17, s[8:9]
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v15, v15, v17, s[10:11]
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v6, v6, v0, s[0:1]
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v8, v8, v0, s[2:3]
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v10, v10, v0, s[4:5]
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v12, v12, v0, s[6:7]
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v14, v14, v0, s[8:9]
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v16, v16, v0, s[10:11]
 ; GPRIDX-NEXT:    global_store_dwordx4 v[0:1], v[1:4], off
 ; GPRIDX-NEXT:    global_store_dwordx4 v[0:1], v[5:8], off
 ; GPRIDX-NEXT:    global_store_dwordx4 v[0:1], v[9:12], off
 ; GPRIDX-NEXT:    global_store_dwordx4 v[0:1], v[13:16], off
 ; GPRIDX-NEXT:    s_endpgm
 ;
 ; MOVREL-LABEL: dyn_insertelement_v8f64_s_s_v:
 ; MOVREL:       ; %bb.0: ; %entry
 ; MOVREL-NEXT:    s_mov_b32 s1, s3
 ; MOVREL-NEXT:    s_mov_b32 s3, s5
 ; MOVREL-NEXT:    s_mov_b32 s5, s7
 ; MOVREL-NEXT:    s_mov_b32 s7, s9
 ; MOVREL-NEXT:    s_mov_b32 s9, s11
 ; MOVREL-NEXT:    s_mov_b32 s11, s13
 ; MOVREL-NEXT:    s_mov_b32 s13, s15
 ; MOVREL-NEXT:    s_mov_b32 s15, s17
 ; MOVREL-NEXT:    s_mov_b32 s0, s2
 ; MOVREL-NEXT:    s_mov_b32 s2, s4
 ; MOVREL-NEXT:    s_mov_b32 s4, s6
 ; MOVREL-NEXT:    s_mov_b32 s6, s8
 ; MOVREL-NEXT:    s_mov_b32 s8, s10
 ; MOVREL-NEXT:    s_mov_b32 s10, s12
 ; MOVREL-NEXT:    s_mov_b32 s12, s14
 ; MOVREL-NEXT:    s_mov_b32 s14, s16
-; MOVREL-NEXT:    v_mov_b32_e32 v32, s15
-; MOVREL-NEXT:    v_mov_b32_e32 v30, s13
-; MOVREL-NEXT:    v_mov_b32_e32 v28, s11
-; MOVREL-NEXT:    v_mov_b32_e32 v29, s12
-; MOVREL-NEXT:    v_mov_b32_e32 v31, s14
-; MOVREL-NEXT:    v_mov_b32_e32 v27, s10
-; MOVREL-NEXT:    v_mov_b32_e32 v26, s9
-; MOVREL-NEXT:    v_mov_b32_e32 v25, s8
-; MOVREL-NEXT:    v_mov_b32_e32 v24, s7
-; MOVREL-NEXT:    v_mov_b32_e32 v23, s6
-; MOVREL-NEXT:    v_mov_b32_e32 v22, s5
-; MOVREL-NEXT:    v_mov_b32_e32 v21, s4
-; MOVREL-NEXT:    v_mov_b32_e32 v20, s3
-; MOVREL-NEXT:    v_mov_b32_e32 v19, s2
-; MOVREL-NEXT:    v_mov_b32_e32 v18, s1
-; MOVREL-NEXT:    v_mov_b32_e32 v17, s0
-; MOVREL-NEXT:    s_mov_b32 s0, exec_lo
-; MOVREL-NEXT:    ; implicit-def: $vcc_hi
-; MOVREL-NEXT:  BB14_1: ; =>This Inner Loop Header: Depth=1
-; MOVREL-NEXT:    v_readfirstlane_b32 s1, v0
-; MOVREL-NEXT:    v_mov_b32_e32 v1, v17
-; MOVREL-NEXT:    v_mov_b32_e32 v2, v18
-; MOVREL-NEXT:    v_mov_b32_e32 v3, v19
-; MOVREL-NEXT:    v_mov_b32_e32 v4, v20
-; MOVREL-NEXT:    s_lshl_b32 m0, s1, 1
-; MOVREL-NEXT:    v_cmp_eq_u32_e32 vcc_lo, s1, v0
-; MOVREL-NEXT:    v_mov_b32_e32 v5, v21
-; MOVREL-NEXT:    v_mov_b32_e32 v6, v22
-; MOVREL-NEXT:    v_mov_b32_e32 v7, v23
-; MOVREL-NEXT:    v_mov_b32_e32 v8, v24
-; MOVREL-NEXT:    v_mov_b32_e32 v9, v25
-; MOVREL-NEXT:    v_mov_b32_e32 v10, v26
-; MOVREL-NEXT:    v_mov_b32_e32 v11, v27
-; MOVREL-NEXT:    v_mov_b32_e32 v12, v28
-; MOVREL-NEXT:    v_mov_b32_e32 v13, v29
-; MOVREL-NEXT:    v_mov_b32_e32 v14, v30
-; MOVREL-NEXT:    v_mov_b32_e32 v15, v31
-; MOVREL-NEXT:    v_mov_b32_e32 v16, v32
-; MOVREL-NEXT:    v_movreld_b32_e32 v1, s18
-; MOVREL-NEXT:    v_movreld_b32_e32 v2, s19
-; MOVREL-NEXT:    s_and_saveexec_b32 vcc_lo, vcc_lo
-; MOVREL-NEXT:    s_xor_b32 exec_lo, exec_lo, vcc_lo
-; MOVREL-NEXT:    s_cbranch_execnz BB14_1
-; MOVREL-NEXT:  ; %bb.2:
-; MOVREL-NEXT:    s_mov_b32 exec_lo, s0
+; MOVREL-NEXT:    v_mov_b32_e32 v16, s15
+; MOVREL-NEXT:    s_mov_b32 s30, s18
+; MOVREL-NEXT:    s_mov_b32 s31, s19
+; MOVREL-NEXT:    v_mov_b32_e32 v14, s13
+; MOVREL-NEXT:    v_mov_b32_e32 v15, s14
+; MOVREL-NEXT:    v_mov_b32_e32 v13, s12
+; MOVREL-NEXT:    v_mov_b32_e32 v12, s11
+; MOVREL-NEXT:    v_mov_b32_e32 v11, s10
+; MOVREL-NEXT:    v_mov_b32_e32 v10, s9
+; MOVREL-NEXT:    v_mov_b32_e32 v9, s8
+; MOVREL-NEXT:    v_mov_b32_e32 v8, s7
+; MOVREL-NEXT:    v_mov_b32_e32 v7, s6
+; MOVREL-NEXT:    v_mov_b32_e32 v6, s5
+; MOVREL-NEXT:    v_mov_b32_e32 v5, s4
+; MOVREL-NEXT:    v_mov_b32_e32 v4, s3
+; MOVREL-NEXT:    v_mov_b32_e32 v3, s2
+; MOVREL-NEXT:    v_mov_b32_e32 v2, s1
+; MOVREL-NEXT:    v_mov_b32_e32 v1, s0
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 s0, 1, v0
+; MOVREL-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 0, v0
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 s2, 5, v0
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 s1, 2, v0
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 s3, 6, v0
+; MOVREL-NEXT:    v_cndmask_b32_e64 v3, v3, s30, s0
+; MOVREL-NEXT:    v_cndmask_b32_e64 v4, v4, s31, s0
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 s0, 4, v0
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 s4, 7, v0
+; MOVREL-NEXT:    v_cndmask_b32_e64 v1, v1, s30, vcc_lo
+; MOVREL-NEXT:    v_cndmask_b32_e64 v2, v2, s31, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 3, v0
+; MOVREL-NEXT:    v_cndmask_b32_e64 v9, v9, s30, s0
+; MOVREL-NEXT:    v_cndmask_b32_e64 v10, v10, s31, s0
+; MOVREL-NEXT:    v_cndmask_b32_e64 v11, v11, s30, s2
+; MOVREL-NEXT:    v_cndmask_b32_e64 v12, v12, s31, s2
+; MOVREL-NEXT:    v_cndmask_b32_e64 v5, v5, s30, s1
+; MOVREL-NEXT:    v_cndmask_b32_e64 v6, v6, s31, s1
+; MOVREL-NEXT:    v_cndmask_b32_e64 v7, v7, s30, vcc_lo
+; MOVREL-NEXT:    v_cndmask_b32_e64 v8, v8, s31, vcc_lo
+; MOVREL-NEXT:    v_cndmask_b32_e64 v13, v13, s30, s3
+; MOVREL-NEXT:    v_cndmask_b32_e64 v14, v14, s31, s3
+; MOVREL-NEXT:    v_cndmask_b32_e64 v15, v15, s30, s4
+; MOVREL-NEXT:    v_cndmask_b32_e64 v16, v16, s31, s4
 ; MOVREL-NEXT:    global_store_dwordx4 v[0:1], v[1:4], off
 ; MOVREL-NEXT:    global_store_dwordx4 v[0:1], v[5:8], off
+; MOVREL-NEXT:    ; implicit-def: $vcc_hi
 ; MOVREL-NEXT:    global_store_dwordx4 v[0:1], v[9:12], off
 ; MOVREL-NEXT:    global_store_dwordx4 v[0:1], v[13:16], off
 ; MOVREL-NEXT:    s_endpgm
 entry:
   %insert = insertelement <8 x double> %vec, double %val, i32 %idx
   %vec.0 = shufflevector <8 x double> %insert, <8 x double> undef, <2 x i32> <i32 0, i32 1>
   %vec.1 = shufflevector <8 x double> %insert, <8 x double> undef, <2 x i32> <i32 2, i32 3>
   %vec.2 = shufflevector <8 x double> %insert, <8 x double> undef, <2 x i32> <i32 4, i32 5>

 ; GPRIDX-NEXT:    s_mov_b32 s0, s2
 ; GPRIDX-NEXT:    s_mov_b32 s2, s4
 ; GPRIDX-NEXT:    s_mov_b32 s4, s6
 ; GPRIDX-NEXT:    s_mov_b32 s6, s8
 ; GPRIDX-NEXT:    s_mov_b32 s8, s10
 ; GPRIDX-NEXT:    s_mov_b32 s10, s12
 ; GPRIDX-NEXT:    s_mov_b32 s12, s14
 ; GPRIDX-NEXT:    s_mov_b32 s14, s16
-; GPRIDX-NEXT:    v_mov_b32_e32 v34, s15
-; GPRIDX-NEXT:    v_mov_b32_e32 v33, s14
-; GPRIDX-NEXT:    v_mov_b32_e32 v32, s13
-; GPRIDX-NEXT:    v_mov_b32_e32 v31, s12
-; GPRIDX-NEXT:    v_mov_b32_e32 v30, s11
-; GPRIDX-NEXT:    v_mov_b32_e32 v29, s10
-; GPRIDX-NEXT:    v_mov_b32_e32 v28, s9
-; GPRIDX-NEXT:    v_mov_b32_e32 v27, s8
-; GPRIDX-NEXT:    v_mov_b32_e32 v26, s7
-; GPRIDX-NEXT:    v_mov_b32_e32 v25, s6
-; GPRIDX-NEXT:    v_mov_b32_e32 v24, s5
-; GPRIDX-NEXT:    v_mov_b32_e32 v23, s4
-; GPRIDX-NEXT:    v_mov_b32_e32 v22, s3
-; GPRIDX-NEXT:    v_mov_b32_e32 v21, s2
-; GPRIDX-NEXT:    v_mov_b32_e32 v20, s1
-; GPRIDX-NEXT:    v_mov_b32_e32 v19, s0
-; GPRIDX-NEXT:    s_mov_b64 s[0:1], exec
-; GPRIDX-NEXT:  BB17_1: ; =>This Inner Loop Header: Depth=1
-; GPRIDX-NEXT:    v_readfirstlane_b32 s2, v2
-; GPRIDX-NEXT:    s_lshl_b32 s3, s2, 1
-; GPRIDX-NEXT:    v_cmp_eq_u32_e32 vcc, s2, v2
-; GPRIDX-NEXT:    s_set_gpr_idx_on s3, gpr_idx(DST)
-; GPRIDX-NEXT:    v_mov_b32_e32 v3, v19
-; GPRIDX-NEXT:    v_mov_b32_e32 v4, v20
-; GPRIDX-NEXT:    v_mov_b32_e32 v5, v21
-; GPRIDX-NEXT:    v_mov_b32_e32 v6, v22
-; GPRIDX-NEXT:    v_mov_b32_e32 v7, v23
-; GPRIDX-NEXT:    v_mov_b32_e32 v8, v24
-; GPRIDX-NEXT:    v_mov_b32_e32 v9, v25
-; GPRIDX-NEXT:    v_mov_b32_e32 v10, v26
-; GPRIDX-NEXT:    v_mov_b32_e32 v11, v27
-; GPRIDX-NEXT:    v_mov_b32_e32 v12, v28
-; GPRIDX-NEXT:    v_mov_b32_e32 v13, v29
-; GPRIDX-NEXT:    v_mov_b32_e32 v14, v30
-; GPRIDX-NEXT:    v_mov_b32_e32 v15, v31
-; GPRIDX-NEXT:    v_mov_b32_e32 v16, v32
-; GPRIDX-NEXT:    v_mov_b32_e32 v17, v33
-; GPRIDX-NEXT:    v_mov_b32_e32 v18, v34
-; GPRIDX-NEXT:    v_mov_b32_e32 v3, v0
-; GPRIDX-NEXT:    s_set_gpr_idx_off
-; GPRIDX-NEXT:    s_set_gpr_idx_on s3, gpr_idx(DST)
-; GPRIDX-NEXT:    v_mov_b32_e32 v4, v1
-; GPRIDX-NEXT:    s_set_gpr_idx_off
-; GPRIDX-NEXT:    s_and_saveexec_b64 vcc, vcc
-; GPRIDX-NEXT:    s_xor_b64 exec, exec, vcc
-; GPRIDX-NEXT:    s_cbranch_execnz BB17_1
-; GPRIDX-NEXT:  ; %bb.2:
-; GPRIDX-NEXT:    s_mov_b64 exec, s[0:1]
+; GPRIDX-NEXT:    v_mov_b32_e32 v18, s15
+; GPRIDX-NEXT:    v_mov_b32_e32 v17, s14
+; GPRIDX-NEXT:    v_mov_b32_e32 v16, s13
+; GPRIDX-NEXT:    v_mov_b32_e32 v15, s12
+; GPRIDX-NEXT:    v_mov_b32_e32 v14, s11
+; GPRIDX-NEXT:    v_mov_b32_e32 v13, s10
+; GPRIDX-NEXT:    v_mov_b32_e32 v12, s9
+; GPRIDX-NEXT:    v_mov_b32_e32 v11, s8
+; GPRIDX-NEXT:    v_mov_b32_e32 v10, s7
+; GPRIDX-NEXT:    v_mov_b32_e32 v9, s6
+; GPRIDX-NEXT:    v_mov_b32_e32 v8, s5
+; GPRIDX-NEXT:    v_mov_b32_e32 v7, s4
+; GPRIDX-NEXT:    v_mov_b32_e32 v6, s3
+; GPRIDX-NEXT:    v_cmp_eq_u32_e32 vcc, 1, v2
+; GPRIDX-NEXT:    v_mov_b32_e32 v5, s2
+; GPRIDX-NEXT:    v_mov_b32_e32 v4, s1
+; GPRIDX-NEXT:    v_mov_b32_e32 v3, s0
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 s[12:13], 0, v2
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 s[0:1], 2, v2
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 s[2:3], 3, v2
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 s[4:5], 4, v2
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 s[6:7], 5, v2
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 s[8:9], 6, v2
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 s[10:11], 7, v2
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v3, v3, v0, s[12:13]
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v5, v5, v0, vcc
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v4, v4, v1, s[12:13]
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v6, v6, v1, vcc
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v7, v7, v0, s[0:1]
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v9, v9, v0, s[2:3]
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v11, v11, v0, s[4:5]
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v13, v13, v0, s[6:7]
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v15, v15, v0, s[8:9]
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v17, v17, v0, s[10:11]
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v8, v8, v1, s[0:1]
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v10, v10, v1, s[2:3]
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v12, v12, v1, s[4:5]
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v14, v14, v1, s[6:7]
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v16, v16, v1, s[8:9]
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v18, v18, v1, s[10:11]
 ; GPRIDX-NEXT:    global_store_dwordx4 v[0:1], v[3:6], off
 ; GPRIDX-NEXT:    global_store_dwordx4 v[0:1], v[7:10], off
 ; GPRIDX-NEXT:    global_store_dwordx4 v[0:1], v[11:14], off
 ; GPRIDX-NEXT:    global_store_dwordx4 v[0:1], v[15:18], off
 ; GPRIDX-NEXT:    s_endpgm
 ;
 ; MOVREL-LABEL: dyn_insertelement_v8f64_s_v_v:
 ; MOVREL:       ; %bb.0: ; %entry
 ; MOVREL-NEXT:    s_mov_b32 s1, s3
 ; MOVREL-NEXT:    s_mov_b32 s3, s5
 ; MOVREL-NEXT:    s_mov_b32 s5, s7
 ; MOVREL-NEXT:    s_mov_b32 s7, s9
 ; MOVREL-NEXT:    s_mov_b32 s9, s11
 ; MOVREL-NEXT:    s_mov_b32 s11, s13
 ; MOVREL-NEXT:    s_mov_b32 s13, s15
 ; MOVREL-NEXT:    s_mov_b32 s15, s17
 ; MOVREL-NEXT:    s_mov_b32 s0, s2
 ; MOVREL-NEXT:    s_mov_b32 s2, s4
 ; MOVREL-NEXT:    s_mov_b32 s4, s6
 ; MOVREL-NEXT:    s_mov_b32 s6, s8
 ; MOVREL-NEXT:    s_mov_b32 s8, s10
 ; MOVREL-NEXT:    s_mov_b32 s10, s12
 ; MOVREL-NEXT:    s_mov_b32 s12, s14
 ; MOVREL-NEXT:    s_mov_b32 s14, s16
-; MOVREL-NEXT:    v_mov_b32_e32 v34, s15
-; MOVREL-NEXT:    v_mov_b32_e32 v32, s13
-; MOVREL-NEXT:    v_mov_b32_e32 v30, s11
-; MOVREL-NEXT:    v_mov_b32_e32 v31, s12
-; MOVREL-NEXT:    v_mov_b32_e32 v33, s14
-; MOVREL-NEXT:    v_mov_b32_e32 v29, s10
-; MOVREL-NEXT:    v_mov_b32_e32 v28, s9
-; MOVREL-NEXT:    v_mov_b32_e32 v27, s8
-; MOVREL-NEXT:    v_mov_b32_e32 v26, s7
-; MOVREL-NEXT:    v_mov_b32_e32 v25, s6
-; MOVREL-NEXT:    v_mov_b32_e32 v24, s5
-; MOVREL-NEXT:    v_mov_b32_e32 v23, s4
-; MOVREL-NEXT:    v_mov_b32_e32 v22, s3
-; MOVREL-NEXT:    v_mov_b32_e32 v21, s2
-; MOVREL-NEXT:    v_mov_b32_e32 v20, s1
-; MOVREL-NEXT:    v_mov_b32_e32 v19, s0
-; MOVREL-NEXT:    s_mov_b32 s0, exec_lo
+; MOVREL-NEXT:    v_mov_b32_e32 v18, s15
+; MOVREL-NEXT:    v_mov_b32_e32 v16, s13
+; MOVREL-NEXT:    v_mov_b32_e32 v14, s11
+; MOVREL-NEXT:    v_mov_b32_e32 v15, s12
+; MOVREL-NEXT:    v_mov_b32_e32 v17, s14
+; MOVREL-NEXT:    v_mov_b32_e32 v13, s10
+; MOVREL-NEXT:    v_mov_b32_e32 v12, s9
+; MOVREL-NEXT:    v_mov_b32_e32 v11, s8
+; MOVREL-NEXT:    v_mov_b32_e32 v10, s7
+; MOVREL-NEXT:    v_mov_b32_e32 v9, s6
+; MOVREL-NEXT:    v_mov_b32_e32 v8, s5
+; MOVREL-NEXT:    v_mov_b32_e32 v7, s4
+; MOVREL-NEXT:    v_mov_b32_e32 v6, s3
+; MOVREL-NEXT:    v_mov_b32_e32 v5, s2
+; MOVREL-NEXT:    v_mov_b32_e32 v4, s1
+; MOVREL-NEXT:    v_mov_b32_e32 v3, s0
+; MOVREL-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 0, v2
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 s0, 1, v2
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 s2, 4, v2
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 s3, 5, v2
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 s1, 3, v2
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 s4, 6, v2
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 s5, 7, v2
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 s6, 2, v2
+; MOVREL-NEXT:    v_cndmask_b32_e32 v3, v3, v0, vcc_lo
+; MOVREL-NEXT:    v_cndmask_b32_e64 v5, v5, v0, s0
+; MOVREL-NEXT:    v_cndmask_b32_e64 v11, v11, v0, s2
+; MOVREL-NEXT:    v_cndmask_b32_e64 v13, v13, v0, s3
+; MOVREL-NEXT:    v_cndmask_b32_e32 v4, v4, v1, vcc_lo
+; MOVREL-NEXT:    v_cndmask_b32_e64 v6, v6, v1, s0
+; MOVREL-NEXT:    v_cndmask_b32_e64 v12, v12, v1, s2
+; MOVREL-NEXT:    v_cndmask_b32_e64 v14, v14, v1, s3
+; MOVREL-NEXT:    v_cndmask_b32_e64 v7, v7, v0, s6
+; MOVREL-NEXT:    v_cndmask_b32_e64 v9, v9, v0, s1
+; MOVREL-NEXT:    v_cndmask_b32_e64 v15, v15, v0, s4
+; MOVREL-NEXT:    v_cndmask_b32_e64 v17, v17, v0, s5
+; MOVREL-NEXT:    v_cndmask_b32_e64 v8, v8, v1, s6
+; MOVREL-NEXT:    v_cndmask_b32_e64 v10, v10, v1, s1
+; MOVREL-NEXT:    v_cndmask_b32_e64 v16, v16, v1, s4
+; MOVREL-NEXT:    v_cndmask_b32_e64 v18, v18, v1, s5
 ; MOVREL-NEXT:    ; implicit-def: $vcc_hi
-; MOVREL-NEXT:  BB17_1: ; =>This Inner Loop Header: Depth=1
-; MOVREL-NEXT:    v_readfirstlane_b32 s1, v2
-; MOVREL-NEXT:    v_mov_b32_e32 v3, v19
-; MOVREL-NEXT:    v_mov_b32_e32 v4, v20
-; MOVREL-NEXT:    v_mov_b32_e32 v5, v21
-; MOVREL-NEXT:    v_mov_b32_e32 v6, v22
-; MOVREL-NEXT:    s_lshl_b32 m0, s1, 1
-; MOVREL-NEXT:    v_cmp_eq_u32_e32 vcc_lo, s1, v2
-; MOVREL-NEXT:    v_mov_b32_e32 v7, v23
-; MOVREL-NEXT:    v_mov_b32_e32 v8, v24
-; MOVREL-NEXT:    v_mov_b32_e32 v9, v25
-; MOVREL-NEXT:    v_mov_b32_e32 v10, v26
-; MOVREL-NEXT:    v_mov_b32_e32 v11, v27
-; MOVREL-NEXT:    v_mov_b32_e32 v12, v28
-; MOVREL-NEXT:    v_mov_b32_e32 v13, v29
-; MOVREL-NEXT:    v_mov_b32_e32 v14, v30
-; MOVREL-NEXT:    v_mov_b32_e32 v15, v31
-; MOVREL-NEXT:    v_mov_b32_e32 v16, v32
-; MOVREL-NEXT:    v_mov_b32_e32 v17, v33
-; MOVREL-NEXT:    v_mov_b32_e32 v18, v34
-; MOVREL-NEXT:    v_movreld_b32_e32 v3, v0
-; MOVREL-NEXT:    v_movreld_b32_e32 v4, v1
-; MOVREL-NEXT:    s_and_saveexec_b32 vcc_lo, vcc_lo
-; MOVREL-NEXT:    s_xor_b32 exec_lo, exec_lo, vcc_lo
-; MOVREL-NEXT:    s_cbranch_execnz BB17_1
-; MOVREL-NEXT:  ; %bb.2:
-; MOVREL-NEXT:    s_mov_b32 exec_lo, s0
 ; MOVREL-NEXT:    global_store_dwordx4 v[0:1], v[3:6], off
 ; MOVREL-NEXT:    global_store_dwordx4 v[0:1], v[7:10], off
 ; MOVREL-NEXT:    global_store_dwordx4 v[0:1], v[11:14], off
 ; MOVREL-NEXT:    global_store_dwordx4 v[0:1], v[15:18], off
 ; MOVREL-NEXT:    s_endpgm
 entry:
   %insert = insertelement <8 x double> %vec, double %val, i32 %idx
   %vec.0 = shufflevector <8 x double> %insert, <8 x double> undef, <2 x i32> <i32 0, i32 1>
   %vec.1 = shufflevector <8 x double> %insert, <8 x double> undef, <2 x i32> <i32 2, i32 3>
   %vec.2 = shufflevector <8 x double> %insert, <8 x double> undef, <2 x i32> <i32 4, i32 5>
   %vec.3 = shufflevector <8 x double> %insert, <8 x double> undef, <2 x i32> <i32 6, i32 7>
   store volatile <2 x double> %vec.0, <2 x double> addrspace(1)* undef
   store volatile <2 x double> %vec.1, <2 x double> addrspace(1)* undef
   store volatile <2 x double> %vec.2, <2 x double> addrspace(1)* undef
   store volatile <2 x double> %vec.3, <2 x double> addrspace(1)* undef
   ret void
 }
 
 define amdgpu_ps void @dyn_insertelement_v8f64_v_s_v(<8 x double> %vec, double inreg %val, i32 %idx) {
 ; GPRIDX-LABEL: dyn_insertelement_v8f64_v_s_v:
 ; GPRIDX:       ; %bb.0: ; %entry
-; GPRIDX-NEXT:    s_mov_b64 s[0:1], exec
-; GPRIDX-NEXT:  BB18_1: ; =>This Inner Loop Header: Depth=1
-; GPRIDX-NEXT:    v_readfirstlane_b32 s4, v16
-; GPRIDX-NEXT:    s_lshl_b32 s5, s4, 1
-; GPRIDX-NEXT:    v_cmp_eq_u32_e32 vcc, s4, v16
-; GPRIDX-NEXT:    s_set_gpr_idx_on s5, gpr_idx(DST)
-; GPRIDX-NEXT:    v_mov_b32_e32 v32, v15
-; GPRIDX-NEXT:    v_mov_b32_e32 v31, v14
-; GPRIDX-NEXT:    v_mov_b32_e32 v30, v13
-; GPRIDX-NEXT:    v_mov_b32_e32 v29, v12
-; GPRIDX-NEXT:    v_mov_b32_e32 v28, v11
-; GPRIDX-NEXT:    v_mov_b32_e32 v27, v10
-; GPRIDX-NEXT:    v_mov_b32_e32 v26, v9
-; GPRIDX-NEXT:    v_mov_b32_e32 v25, v8
-; GPRIDX-NEXT:    v_mov_b32_e32 v24, v7
-; GPRIDX-NEXT:    v_mov_b32_e32 v23, v6
-; GPRIDX-NEXT:    v_mov_b32_e32 v22, v5
-; GPRIDX-NEXT:    v_mov_b32_e32 v21, v4
-; GPRIDX-NEXT:    v_mov_b32_e32 v20, v3
-; GPRIDX-NEXT:    v_mov_b32_e32 v19, v2
-; GPRIDX-NEXT:    v_mov_b32_e32 v18, v1
-; GPRIDX-NEXT:    v_mov_b32_e32 v17, v0
 ; GPRIDX-NEXT:    v_mov_b32_e32 v17, s2
-; GPRIDX-NEXT:    s_set_gpr_idx_off
-; GPRIDX-NEXT:    s_set_gpr_idx_on s5, gpr_idx(DST)
-; GPRIDX-NEXT:    v_mov_b32_e32 v18, s3
-; GPRIDX-NEXT:    s_set_gpr_idx_off
-; GPRIDX-NEXT:    s_and_saveexec_b64 vcc, vcc
-; GPRIDX-NEXT:    s_xor_b64 exec, exec, vcc
-; GPRIDX-NEXT:    s_cbranch_execnz BB18_1
-; GPRIDX-NEXT:  ; %bb.2:
-; GPRIDX-NEXT:    s_mov_b64 exec, s[0:1]
-; GPRIDX-NEXT:    global_store_dwordx4 v[0:1], v[17:20], off
-; GPRIDX-NEXT:    global_store_dwordx4 v[0:1], v[21:24], off
-; GPRIDX-NEXT:    global_store_dwordx4 v[0:1], v[25:28], off
-; GPRIDX-NEXT:    global_store_dwordx4 v[0:1], v[29:32], off
+; GPRIDX-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v16
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 s[0:1], 1, v16
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 s[4:5], 2, v16
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 s[6:7], 3, v16
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 s[8:9], 4, v16
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 s[10:11], 5, v16
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 s[12:13], 7, v16
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 s[14:15], 6, v16
+; GPRIDX-NEXT:    v_mov_b32_e32 v16, s3
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v0, v0, v17, vcc
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v2, v2, v17, s[0:1]
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v1, v1, v16, vcc
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v3, v3, v16, s[0:1]
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v4, v4, v17, s[4:5]
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v6, v6, v17, s[6:7]
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v8, v8, v17, s[8:9]
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v10, v10, v17, s[10:11]
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v12, v12, v17, s[14:15]
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v14, v14, v17, s[12:13]
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v5, v5, v16, s[4:5]
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v7, v7, v16, s[6:7]
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v9, v9, v16, s[8:9]
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v11, v11, v16, s[10:11]
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v13, v13, v16, s[14:15]
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v15, v15, v16, s[12:13]
+; GPRIDX-NEXT:    global_store_dwordx4 v[0:1], v[0:3], off
+; GPRIDX-NEXT:    global_store_dwordx4 v[0:1], v[4:7], off
+; GPRIDX-NEXT:    global_store_dwordx4 v[0:1], v[8:11], off
+; GPRIDX-NEXT:    global_store_dwordx4 v[0:1], v[12:15], off
 ; GPRIDX-NEXT:    s_endpgm
 ;
 ; MOVREL-LABEL: dyn_insertelement_v8f64_v_s_v:
 ; MOVREL:       ; %bb.0: ; %entry
-; MOVREL-NEXT:    s_mov_b32 s0, exec_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 0, v16
 ; MOVREL-NEXT:    ; implicit-def: $vcc_hi
-; MOVREL-NEXT:  BB18_1: ; =>This Inner Loop Header: Depth=1
-; MOVREL-NEXT:    v_readfirstlane_b32 s1, v16
-; MOVREL-NEXT:    v_mov_b32_e32 v32, v15
-; MOVREL-NEXT:    v_mov_b32_e32 v17, v0
-; MOVREL-NEXT:    v_mov_b32_e32 v31, v14
-; MOVREL-NEXT:    v_mov_b32_e32 v30, v13
-; MOVREL-NEXT:    s_lshl_b32 m0, s1, 1
-; MOVREL-NEXT:    v_cmp_eq_u32_e32 vcc_lo, s1, v16
-; MOVREL-NEXT:    v_mov_b32_e32 v29, v12
-; MOVREL-NEXT:    v_mov_b32_e32 v28, v11
-; MOVREL-NEXT:    v_mov_b32_e32 v27, v10
-; MOVREL-NEXT:    v_mov_b32_e32 v26, v9
-; MOVREL-NEXT:    v_mov_b32_e32 v25, v8
-; MOVREL-NEXT:    v_mov_b32_e32 v24, v7
-; MOVREL-NEXT:    v_mov_b32_e32 v23, v6
-; MOVREL-NEXT:    v_mov_b32_e32 v22, v5
-; MOVREL-NEXT:    v_mov_b32_e32 v21, v4
-; MOVREL-NEXT:    v_mov_b32_e32 v20, v3
-; MOVREL-NEXT:    v_mov_b32_e32 v19, v2
-; MOVREL-NEXT:    v_mov_b32_e32 v18, v1
-; MOVREL-NEXT:    v_movreld_b32_e32 v17, s2
-; MOVREL-NEXT:    v_movreld_b32_e32 v18, s3
-; MOVREL-NEXT:    s_and_saveexec_b32 vcc_lo, vcc_lo
-; MOVREL-NEXT:    s_xor_b32 exec_lo, exec_lo, vcc_lo
-; MOVREL-NEXT:    s_cbranch_execnz BB18_1
-; MOVREL-NEXT:  ; %bb.2:
-; MOVREL-NEXT:    s_mov_b32 exec_lo, s0
-; MOVREL-NEXT:    global_store_dwordx4 v[0:1], v[17:20], off
-; MOVREL-NEXT:    global_store_dwordx4 v[0:1], v[21:24], off
-; MOVREL-NEXT:    global_store_dwordx4 v[0:1], v[25:28], off
-; MOVREL-NEXT:    global_store_dwordx4 v[0:1], v[29:32], off
+; MOVREL-NEXT:    v_cndmask_b32_e64 v0, v0, s2, vcc_lo
+; MOVREL-NEXT:    v_cndmask_b32_e64 v1, v1, s3, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 1, v16
+; MOVREL-NEXT:    v_cndmask_b32_e64 v2, v2, s2, vcc_lo
+; MOVREL-NEXT:    v_cndmask_b32_e64 v3, v3, s3, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 2, v16
+; MOVREL-NEXT:    v_cndmask_b32_e64 v4, v4, s2, vcc_lo
+; MOVREL-NEXT:    v_cndmask_b32_e64 v5, v5, s3, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 3, v16
+; MOVREL-NEXT:    v_cndmask_b32_e64 v6, v6, s2, vcc_lo
+; MOVREL-NEXT:    v_cndmask_b32_e64 v7, v7, s3, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 4, v16
+; MOVREL-NEXT:    global_store_dwordx4 v[0:1], v[0:3], off
+; MOVREL-NEXT:    global_store_dwordx4 v[0:1], v[4:7], off
+; MOVREL-NEXT:    v_cndmask_b32_e64 v8, v8, s2, vcc_lo
+; MOVREL-NEXT:    v_cndmask_b32_e64 v9, v9, s3, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 5, v16
+; MOVREL-NEXT:    v_cndmask_b32_e64 v10, v10, s2, vcc_lo
+; MOVREL-NEXT:    v_cndmask_b32_e64 v11, v11, s3, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 6, v16
+; MOVREL-NEXT:    v_cndmask_b32_e64 v12, v12, s2, vcc_lo
+; MOVREL-NEXT:    v_cndmask_b32_e64 v13, v13, s3, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 7, v16
+; MOVREL-NEXT:    v_cndmask_b32_e64 v14, v14, s2, vcc_lo
+; MOVREL-NEXT:    v_cndmask_b32_e64 v15, v15, s3, vcc_lo
+; MOVREL-NEXT:    global_store_dwordx4 v[0:1], v[8:11], off
+; MOVREL-NEXT:    global_store_dwordx4 v[0:1], v[12:15], off
 ; MOVREL-NEXT:    s_endpgm
 entry:
   %insert = insertelement <8 x double> %vec, double %val, i32 %idx
   %vec.0 = shufflevector <8 x double> %insert, <8 x double> undef, <2 x i32> <i32 0, i32 1>
   %vec.1 = shufflevector <8 x double> %insert, <8 x double> undef, <2 x i32> <i32 2, i32 3>
   %vec.2 = shufflevector <8 x double> %insert, <8 x double> undef, <2 x i32> <i32 4, i32 5>
   %vec.3 = shufflevector <8 x double> %insert, <8 x double> undef, <2 x i32> <i32 6, i32 7>
   store volatile <2 x double> %vec.0, <2 x double> addrspace(1)* undef
entry:
   store volatile <2 x double> %vec.2, <2 x double> addrspace(1)* undef
   store volatile <2 x double> %vec.3, <2 x double> addrspace(1)* undef
   ret void
 }
 
 define amdgpu_ps void @dyn_insertelement_v8f64_v_v_v(<8 x double> %vec, double %val, i32 %idx) {
 ; GPRIDX-LABEL: dyn_insertelement_v8f64_v_v_v:
 ; GPRIDX:       ; %bb.0: ; %entry
-; GPRIDX-NEXT:    s_mov_b64 s[0:1], exec
-; GPRIDX-NEXT:  BB20_1: ; =>This Inner Loop Header: Depth=1
-; GPRIDX-NEXT:    v_readfirstlane_b32 s2, v18
-; GPRIDX-NEXT:    s_lshl_b32 s3, s2, 1
-; GPRIDX-NEXT:    v_cmp_eq_u32_e32 vcc, s2, v18
-; GPRIDX-NEXT:    s_set_gpr_idx_on s3, gpr_idx(DST)
-; GPRIDX-NEXT:    v_mov_b32_e32 v34, v15
-; GPRIDX-NEXT:    v_mov_b32_e32 v33, v14
-; GPRIDX-NEXT:    v_mov_b32_e32 v32, v13
-; GPRIDX-NEXT:    v_mov_b32_e32 v31, v12
-; GPRIDX-NEXT:    v_mov_b32_e32 v30, v11
-; GPRIDX-NEXT:    v_mov_b32_e32 v29, v10
-; GPRIDX-NEXT:    v_mov_b32_e32 v28, v9
-; GPRIDX-NEXT:    v_mov_b32_e32 v27, v8
-; GPRIDX-NEXT:    v_mov_b32_e32 v26, v7
-; GPRIDX-NEXT:    v_mov_b32_e32 v25, v6
-; GPRIDX-NEXT:    v_mov_b32_e32 v24, v5
-; GPRIDX-NEXT:    v_mov_b32_e32 v23, v4
-; GPRIDX-NEXT:    v_mov_b32_e32 v22, v3
-; GPRIDX-NEXT:    v_mov_b32_e32 v21, v2
-; GPRIDX-NEXT:    v_mov_b32_e32 v20, v1
-; GPRIDX-NEXT:    v_mov_b32_e32 v19, v0
-; GPRIDX-NEXT:    v_mov_b32_e32 v19, v16
-; GPRIDX-NEXT:    s_set_gpr_idx_off
-; GPRIDX-NEXT:    s_set_gpr_idx_on s3, gpr_idx(DST)
-; GPRIDX-NEXT:    v_mov_b32_e32 v20, v17
-; GPRIDX-NEXT:    s_set_gpr_idx_off
-; GPRIDX-NEXT:    s_and_saveexec_b64 vcc, vcc
-; GPRIDX-NEXT:    s_xor_b64 exec, exec, vcc
-; GPRIDX-NEXT:    s_cbranch_execnz BB20_1
-; GPRIDX-NEXT:  ; %bb.2:
-; GPRIDX-NEXT:    s_mov_b64 exec, s[0:1]
-; GPRIDX-NEXT:    global_store_dwordx4 v[0:1], v[19:22], off
-; GPRIDX-NEXT:    global_store_dwordx4 v[0:1], v[23:26], off
-; GPRIDX-NEXT:    global_store_dwordx4 v[0:1], v[27:30], off
-; GPRIDX-NEXT:    global_store_dwordx4 v[0:1], v[31:34], off
+; GPRIDX-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v18
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 s[0:1], 1, v18
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 s[2:3], 2, v18
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 s[4:5], 3, v18
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 s[6:7], 4, v18
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 s[8:9], 5, v18
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 s[10:11], 7, v18
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 s[12:13], 6, v18
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v0, v0, v16, vcc
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v2, v2, v16, s[0:1]
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v1, v1, v17, vcc
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v3, v3, v17, s[0:1]
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v4, v4, v16, s[2:3]
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v6, v6, v16, s[4:5]
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v8, v8, v16, s[6:7]
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v10, v10, v16, s[8:9]
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v12, v12, v16, s[12:13]
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v14, v14, v16, s[10:11]
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v5, v5, v17, s[2:3]
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v7, v7, v17, s[4:5]
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v9, v9, v17, s[6:7]
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v11, v11, v17, s[8:9]
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v13, v13, v17, s[12:13]
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v15, v15, v17, s[10:11]
+; GPRIDX-NEXT:    global_store_dwordx4 v[0:1], v[0:3], off
+; GPRIDX-NEXT:    global_store_dwordx4 v[0:1], v[4:7], off
+; GPRIDX-NEXT:    global_store_dwordx4 v[0:1], v[8:11], off
+; GPRIDX-NEXT:    global_store_dwordx4 v[0:1], v[12:15], off
 ; GPRIDX-NEXT:    s_endpgm
 ;
 ; MOVREL-LABEL: dyn_insertelement_v8f64_v_v_v:
 ; MOVREL:       ; %bb.0: ; %entry
-; MOVREL-NEXT:    s_mov_b32 s0, exec_lo
-; MOVREL-NEXT:    ; implicit-def: $vcc_hi
-; MOVREL-NEXT:  BB20_1: ; =>This Inner Loop Header: Depth=1
-; MOVREL-NEXT:    v_readfirstlane_b32 s1, v18
-; MOVREL-NEXT:    v_mov_b32_e32 v34, v15
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 s0, 1, v18
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 s3, 4, v18
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 s4, 5, v18
 ; MOVREL-NEXT:    v_mov_b32_e32 v19, v0
-; MOVREL-NEXT:    v_mov_b32_e32 v33, v14
-; MOVREL-NEXT:    v_mov_b32_e32 v32, v13
-; MOVREL-NEXT:    s_lshl_b32 m0, s1, 1
-; MOVREL-NEXT:    v_cmp_eq_u32_e32 vcc_lo, s1, v18
-; MOVREL-NEXT:    v_mov_b32_e32 v31, v12
-; MOVREL-NEXT:    v_mov_b32_e32 v30, v11
-; MOVREL-NEXT:    v_mov_b32_e32 v29, v10
-; MOVREL-NEXT:    v_mov_b32_e32 v28, v9
-; MOVREL-NEXT:    v_mov_b32_e32 v27, v8
-; MOVREL-NEXT:    v_mov_b32_e32 v26, v7
-; MOVREL-NEXT:    v_mov_b32_e32 v25, v6
-; MOVREL-NEXT:    v_mov_b32_e32 v24, v5
-; MOVREL-NEXT:    v_mov_b32_e32 v23, v4
-; MOVREL-NEXT:    v_mov_b32_e32 v22, v3
-; MOVREL-NEXT:    v_mov_b32_e32 v21, v2
+; MOVREL-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 0, v18
 ; MOVREL-NEXT:    v_mov_b32_e32 v20, v1
-; MOVREL-NEXT:    v_movreld_b32_e32 v19, v16
-; MOVREL-NEXT:    v_movreld_b32_e32 v20, v17
-; MOVREL-NEXT:    s_and_saveexec_b32 vcc_lo, vcc_lo
-; MOVREL-NEXT:    s_xor_b32 exec_lo, exec_lo, vcc_lo
-; MOVREL-NEXT:    s_cbranch_execnz BB20_1
-; MOVREL-NEXT:  ; %bb.2:
-; MOVREL-NEXT:    s_mov_b32 exec_lo, s0
-; MOVREL-NEXT:    global_store_dwordx4 v[0:1], v[19:22], off
-; MOVREL-NEXT:    global_store_dwordx4 v[0:1], v[23:26], off
-; MOVREL-NEXT:    global_store_dwordx4 v[0:1], v[27:30], off
-; MOVREL-NEXT:    global_store_dwordx4 v[0:1], v[31:34], off
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 s1, 2, v18
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 s2, 3, v18
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 s5, 7, v18
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 s6, 6, v18
+; MOVREL-NEXT:    v_cndmask_b32_e32 v0, v19, v16, vcc_lo
+; MOVREL-NEXT:    v_cndmask_b32_e64 v2, v2, v16, s0
+; MOVREL-NEXT:    v_cndmask_b32_e64 v8, v8, v16, s3
+; MOVREL-NEXT:    v_cndmask_b32_e64 v10, v10, v16, s4
+; MOVREL-NEXT:    v_cndmask_b32_e32 v1, v20, v17, vcc_lo
+; MOVREL-NEXT:    v_cndmask_b32_e64 v3, v3, v17, s0
+; MOVREL-NEXT:    v_cndmask_b32_e64 v9, v9, v17, s3
+; MOVREL-NEXT:    v_cndmask_b32_e64 v11, v11, v17, s4
+; MOVREL-NEXT:    v_cndmask_b32_e64 v4, v4, v16, s1
+; MOVREL-NEXT:    v_cndmask_b32_e64 v6, v6, v16, s2
+; MOVREL-NEXT:    v_cndmask_b32_e64 v12, v12, v16, s6
+; MOVREL-NEXT:    v_cndmask_b32_e64 v14, v14, v16, s5
+; MOVREL-NEXT:    v_cndmask_b32_e64 v5, v5, v17, s1
+; MOVREL-NEXT:    v_cndmask_b32_e64 v7, v7, v17, s2
+; MOVREL-NEXT:    v_cndmask_b32_e64 v13, v13, v17, s6
+; MOVREL-NEXT:    v_cndmask_b32_e64 v15, v15, v17, s5
+; MOVREL-NEXT:    ; implicit-def: $vcc_hi
+; MOVREL-NEXT:    global_store_dwordx4 v[0:1], v[0:3], off
+; MOVREL-NEXT:    global_store_dwordx4 v[0:1], v[4:7], off
+; MOVREL-NEXT:    global_store_dwordx4 v[0:1], v[8:11], off
+; MOVREL-NEXT:    global_store_dwordx4 v[0:1], v[12:15], off
 ; MOVREL-NEXT:    s_endpgm
 entry:
   %insert = insertelement <8 x double> %vec, double %val, i32 %idx
   %vec.0 = shufflevector <8 x double> %insert, <8 x double> undef, <2 x i32> <i32 0, i32 1>
   %vec.1 = shufflevector <8 x double> %insert, <8 x double> undef, <2 x i32> <i32 2, i32 3>
   %vec.2 = shufflevector <8 x double> %insert, <8 x double> undef, <2 x i32> <i32 4, i32 5>
   %vec.3 = shufflevector <8 x double> %insert, <8 x double> undef, <2 x i32> <i32 6, i32 7>
   store volatile <2 x double> %vec.0, <2 x double> addrspace(1)* undef
   store volatile <2 x double> %vec.1, <2 x double> addrspace(1)* undef
   store volatile <2 x double> %vec.2, <2 x double> addrspace(1)* undef
   store volatile <2 x double> %vec.3, <2 x double> addrspace(1)* undef
   ret void
 }
 
 define amdgpu_ps <3 x i32> @dyn_insertelement_v3i32_s_s_s(<3 x i32> inreg %vec, i32 inreg %val, i32 inreg %idx) {
 ; GPRIDX-LABEL: dyn_insertelement_v3i32_s_s_s:
 ; GPRIDX:       ; %bb.0: ; %entry
-; GPRIDX-NEXT:    s_mov_b32 s0, s2
-; GPRIDX-NEXT:    s_mov_b32 s1, s3
-; GPRIDX-NEXT:    s_mov_b32 s2, s4
-; GPRIDX-NEXT:    s_mov_b32 m0, s6
-; GPRIDX-NEXT:    s_nop 0
-; GPRIDX-NEXT:    s_movreld_b32 s0, s5
+; GPRIDX-NEXT:    s_cmp_eq_u32 s6, 0
+; GPRIDX-NEXT:    s_cselect_b32 s0, s5, s2
+; GPRIDX-NEXT:    s_cmp_eq_u32 s6, 1
+; GPRIDX-NEXT:    s_cselect_b32 s1, s5, s3
+; GPRIDX-NEXT:    s_cmp_eq_u32 s6, 2
+; GPRIDX-NEXT:    s_cselect_b32 s2, s5, s4
 ; GPRIDX-NEXT:    ; return to shader part epilog
 ;
 ; MOVREL-LABEL: dyn_insertelement_v3i32_s_s_s:
 ; MOVREL:       ; %bb.0: ; %entry
-; MOVREL-NEXT:    s_mov_b32 s0, s2
-; MOVREL-NEXT:    s_mov_b32 m0, s6
-; MOVREL-NEXT:    s_mov_b32 s1, s3
-; MOVREL-NEXT:    s_mov_b32 s2, s4
-; MOVREL-NEXT:    ; implicit-def: $vcc_hi
-; MOVREL-NEXT:    s_movreld_b32 s0, s5
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 s0, s6, 0
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 s1, s6, 1
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 s6, s6, 2
+; MOVREL-NEXT:    ; implicit-def: $vcc_hi
+; MOVREL-NEXT:    s_cmp_lg_u32 s0, 0
+; MOVREL-NEXT:    s_cselect_b32 s0, s5, s2
+; MOVREL-NEXT:    s_cmp_lg_u32 s1, 0
+; MOVREL-NEXT:    s_cselect_b32 s1, s5, s3
+; MOVREL-NEXT:    s_cmp_lg_u32 s6, 0
+; MOVREL-NEXT:    s_cselect_b32 s2, s5, s4
 ; MOVREL-NEXT:    ; return to shader part epilog
 entry:
   %insert = insertelement <3 x i32> %vec, i32 %val, i32 %idx
   ret <3 x i32> %insert
 }
 
 define amdgpu_ps <3 x float> @dyn_insertelement_v3i32_v_v_s(<3 x float> %vec, float %val, i32 inreg %idx) {
 ; GPRIDX-LABEL: dyn_insertelement_v3i32_v_v_s:
 ; GPRIDX:       ; %bb.0: ; %entry
-; GPRIDX-NEXT:    s_set_gpr_idx_on s2, gpr_idx(DST)
-; GPRIDX-NEXT:    v_mov_b32_e32 v0, v3
-; GPRIDX-NEXT:    s_set_gpr_idx_off
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 vcc, s2, 0
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v0, v0, v3, vcc
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 vcc, s2, 1
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v1, v1, v3, vcc
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 vcc, s2, 2
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v2, v2, v3, vcc
 ; GPRIDX-NEXT:    ; return to shader part epilog
 ;
 ; MOVREL-LABEL: dyn_insertelement_v3i32_v_v_s:
 ; MOVREL:       ; %bb.0: ; %entry
-; MOVREL-NEXT:    s_mov_b32 m0, s2
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 vcc_lo, s2, 0
 ; MOVREL-NEXT:    ; implicit-def: $vcc_hi
-; MOVREL-NEXT:    v_movreld_b32_e32 v0, v3
+; MOVREL-NEXT:    v_cndmask_b32_e32 v0, v0, v3, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 vcc_lo, s2, 1
+; MOVREL-NEXT:    v_cndmask_b32_e32 v1, v1, v3, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 vcc_lo, s2, 2
+; MOVREL-NEXT:    v_cndmask_b32_e32 v2, v2, v3, vcc_lo
 ; MOVREL-NEXT:    ; return to shader part epilog
 entry:
   %insert = insertelement <3 x float> %vec, float %val, i32 %idx
   ret <3 x float> %insert
 }
 
 define amdgpu_ps <5 x i32> @dyn_insertelement_v5i32_s_s_s(<5 x i32> inreg %vec, i32 inreg %val, i32 inreg %idx) {
 ; GPRIDX-LABEL: dyn_insertelement_v5i32_s_s_s:
 ; GPRIDX:       ; %bb.0: ; %entry
-; GPRIDX-NEXT:    s_mov_b32 s0, s2
-; GPRIDX-NEXT:    s_mov_b32 s1, s3
-; GPRIDX-NEXT:    s_mov_b32 s2, s4
-; GPRIDX-NEXT:    s_mov_b32 s3, s5
-; GPRIDX-NEXT:    s_mov_b32 s4, s6
-; GPRIDX-NEXT:    s_mov_b32 m0, s8
-; GPRIDX-NEXT:    s_nop 0
-; GPRIDX-NEXT:    s_movreld_b32 s0, s7
+; GPRIDX-NEXT:    s_cmp_eq_u32 s8, 0
+; GPRIDX-NEXT:    s_cselect_b32 s0, s7, s2
+; GPRIDX-NEXT:    s_cmp_eq_u32 s8, 1
+; GPRIDX-NEXT:    s_cselect_b32 s1, s7, s3
+; GPRIDX-NEXT:    s_cmp_eq_u32 s8, 2
+; GPRIDX-NEXT:    s_cselect_b32 s2, s7, s4
+; GPRIDX-NEXT:    s_cmp_eq_u32 s8, 3
+; GPRIDX-NEXT:    s_cselect_b32 s3, s7, s5
+; GPRIDX-NEXT:    s_cmp_eq_u32 s8, 4
+; GPRIDX-NEXT:    s_cselect_b32 s4, s7, s6
 ; GPRIDX-NEXT:    ; return to shader part epilog
 ;
 ; MOVREL-LABEL: dyn_insertelement_v5i32_s_s_s:
 ; MOVREL:       ; %bb.0: ; %entry
-; MOVREL-NEXT:    s_mov_b32 s0, s2
-; MOVREL-NEXT:    s_mov_b32 m0, s8
-; MOVREL-NEXT:    s_mov_b32 s1, s3
-; MOVREL-NEXT:    s_mov_b32 s2, s4
-; MOVREL-NEXT:    s_mov_b32 s3, s5
-; MOVREL-NEXT:    s_mov_b32 s4, s6
-; MOVREL-NEXT:    s_movreld_b32 s0, s7
-; MOVREL-NEXT:    ; implicit-def: $vcc_hi
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 s0, s8, 0
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 s1, s8, 1
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 s9, s8, 2
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 s10, s8, 3
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 s8, s8, 4
+; MOVREL-NEXT:    s_cmp_lg_u32 s0, 0
+; MOVREL-NEXT:    ; implicit-def: $vcc_hi
+; MOVREL-NEXT:    s_cselect_b32 s0, s7, s2
+; MOVREL-NEXT:    s_cmp_lg_u32 s1, 0
+; MOVREL-NEXT:    s_cselect_b32 s1, s7, s3
+; MOVREL-NEXT:    s_cmp_lg_u32 s9, 0
+; MOVREL-NEXT:    s_cselect_b32 s2, s7, s4
+; MOVREL-NEXT:    s_cmp_lg_u32 s10, 0
+; MOVREL-NEXT:    s_cselect_b32 s3, s7, s5
+; MOVREL-NEXT:    s_cmp_lg_u32 s8, 0
+; MOVREL-NEXT:    s_cselect_b32 s4, s7, s6
 ; MOVREL-NEXT:    ; return to shader part epilog
 entry:
   %insert = insertelement <5 x i32> %vec, i32 %val, i32 %idx
   ret <5 x i32> %insert
 }
 
 define amdgpu_ps <5 x float> @dyn_insertelement_v5i32_v_v_s(<5 x float> %vec, float %val, i32 inreg %idx) {
 ; GPRIDX-LABEL: dyn_insertelement_v5i32_v_v_s:
 ; GPRIDX:       ; %bb.0: ; %entry
-; GPRIDX-NEXT:    s_set_gpr_idx_on s2, gpr_idx(DST)
-; GPRIDX-NEXT:    v_mov_b32_e32 v0, v5
-; GPRIDX-NEXT:    s_set_gpr_idx_off
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 vcc, s2, 0
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v0, v0, v5, vcc
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 vcc, s2, 1
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v1, v1, v5, vcc
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 vcc, s2, 2
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v2, v2, v5, vcc
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 vcc, s2, 3
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v3, v3, v5, vcc
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 vcc, s2, 4
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v4, v4, v5, vcc
 ; GPRIDX-NEXT:    ; return to shader part epilog
 ;
 ; MOVREL-LABEL: dyn_insertelement_v5i32_v_v_s:
 ; MOVREL:       ; %bb.0: ; %entry
-; MOVREL-NEXT:    s_mov_b32 m0, s2
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 vcc_lo, s2, 0
 ; MOVREL-NEXT:    ; implicit-def: $vcc_hi
-; MOVREL-NEXT:    v_movreld_b32_e32 v0, v5
+; MOVREL-NEXT:    v_cndmask_b32_e32 v0, v0, v5, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 vcc_lo, s2, 1
+; MOVREL-NEXT:    v_cndmask_b32_e32 v1, v1, v5, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 vcc_lo, s2, 2
+; MOVREL-NEXT:    v_cndmask_b32_e32 v2, v2, v5, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 vcc_lo, s2, 3
+; MOVREL-NEXT:    v_cndmask_b32_e32 v3, v3, v5, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 vcc_lo, s2, 4
+; MOVREL-NEXT:    v_cndmask_b32_e32 v4, v4, v5, vcc_lo
 ; MOVREL-NEXT:    ; return to shader part epilog
 entry:
   %insert = insertelement <5 x float> %vec, float %val, i32 %idx
   ret <5 x float> %insert
 }
 
 define amdgpu_ps <32 x i32> @dyn_insertelement_v32i32_s_s_s(<32 x i32> inreg %vec, i32 inreg %val, i32 inreg %idx) {
 ; GPRIDX-LABEL: dyn_insertelement_v32i32_s_s_s:

 entry:
   %insert = insertelement <32 x float> %vec, float %val, i32 %idx
   ret <32 x float> %insert
 }
 
 define amdgpu_ps <8 x float> @dyn_insertelement_v8f32_s_s_s_add_1(<8 x float> inreg %vec, float inreg %val, i32 inreg %idx) {
 ; GPRIDX-LABEL: dyn_insertelement_v8f32_s_s_s_add_1:
 ; GPRIDX:       ; %bb.0: ; %entry
-; GPRIDX-NEXT:    s_mov_b32 s0, s2
-; GPRIDX-NEXT:    s_mov_b32 s1, s3
-; GPRIDX-NEXT:    s_mov_b32 s2, s4
-; GPRIDX-NEXT:    s_mov_b32 s3, s5
-; GPRIDX-NEXT:    s_mov_b32 s4, s6
-; GPRIDX-NEXT:    s_mov_b32 s5, s7
-; GPRIDX-NEXT:    s_mov_b32 s6, s8
-; GPRIDX-NEXT:    s_mov_b32 s7, s9
-; GPRIDX-NEXT:    s_mov_b32 m0, s11
-; GPRIDX-NEXT:    s_nop 0
-; GPRIDX-NEXT:    s_movreld_b32 s1, s10
+; GPRIDX-NEXT:    s_add_i32 s11, s11, 1
+; GPRIDX-NEXT:    s_cmp_eq_u32 s11, 0
+; GPRIDX-NEXT:    s_cselect_b32 s0, s10, s2
+; GPRIDX-NEXT:    s_cmp_eq_u32 s11, 1
+; GPRIDX-NEXT:    s_cselect_b32 s1, s10, s3
+; GPRIDX-NEXT:    s_cmp_eq_u32 s11, 2
+; GPRIDX-NEXT:    s_cselect_b32 s2, s10, s4
+; GPRIDX-NEXT:    s_cmp_eq_u32 s11, 3
+; GPRIDX-NEXT:    s_cselect_b32 s3, s10, s5
+; GPRIDX-NEXT:    s_cmp_eq_u32 s11, 4
+; GPRIDX-NEXT:    s_cselect_b32 s4, s10, s6
+; GPRIDX-NEXT:    s_cmp_eq_u32 s11, 5
+; GPRIDX-NEXT:    s_cselect_b32 s5, s10, s7
+; GPRIDX-NEXT:    s_cmp_eq_u32 s11, 6
+; GPRIDX-NEXT:    s_cselect_b32 s6, s10, s8
+; GPRIDX-NEXT:    s_cmp_eq_u32 s11, 7
+; GPRIDX-NEXT:    s_cselect_b32 s7, s10, s9
 ; GPRIDX-NEXT:    v_mov_b32_e32 v0, s0
 ; GPRIDX-NEXT:    v_mov_b32_e32 v1, s1
 ; GPRIDX-NEXT:    v_mov_b32_e32 v2, s2
 ; GPRIDX-NEXT:    v_mov_b32_e32 v3, s3
 ; GPRIDX-NEXT:    v_mov_b32_e32 v4, s4
 ; GPRIDX-NEXT:    v_mov_b32_e32 v5, s5
 ; GPRIDX-NEXT:    v_mov_b32_e32 v6, s6
 ; GPRIDX-NEXT:    v_mov_b32_e32 v7, s7
 ; GPRIDX-NEXT:    ; return to shader part epilog
 ;
 ; MOVREL-LABEL: dyn_insertelement_v8f32_s_s_s_add_1:
 ; MOVREL:       ; %bb.0: ; %entry
-; MOVREL-NEXT:    s_mov_b32 s1, s3
-; MOVREL-NEXT:    s_mov_b32 m0, s11
-; MOVREL-NEXT:    s_mov_b32 s0, s2
-; MOVREL-NEXT:    s_mov_b32 s2, s4
-; MOVREL-NEXT:    s_mov_b32 s3, s5
-; MOVREL-NEXT:    s_mov_b32 s4, s6
-; MOVREL-NEXT:    s_mov_b32 s5, s7
-; MOVREL-NEXT:    s_mov_b32 s6, s8
-; MOVREL-NEXT:    s_mov_b32 s7, s9
-; MOVREL-NEXT:    s_movreld_b32 s1, s10
+; MOVREL-NEXT:    s_add_i32 s11, s11, 1
+; MOVREL-NEXT:    ; implicit-def: $vcc_hi
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 s0, s11, 0
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 s1, s11, 1
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 s12, s11, 2
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 s13, s11, 3
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 s14, s11, 4
+; MOVREL-NEXT:    s_cmp_lg_u32 s0, 0
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 s15, s11, 5
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 s16, s11, 6
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 s11, s11, 7
+; MOVREL-NEXT:    s_cselect_b32 s0, s10, s2
+; MOVREL-NEXT:    s_cmp_lg_u32 s1, 0
 ; MOVREL-NEXT:    v_mov_b32_e32 v0, s0
+; MOVREL-NEXT:    s_cselect_b32 s1, s10, s3
+; MOVREL-NEXT:    s_cmp_lg_u32 s12, 0
 ; MOVREL-NEXT:    v_mov_b32_e32 v1, s1
+; MOVREL-NEXT:    s_cselect_b32 s2, s10, s4
+; MOVREL-NEXT:    s_cmp_lg_u32 s13, 0
 ; MOVREL-NEXT:    v_mov_b32_e32 v2, s2
+; MOVREL-NEXT:    s_cselect_b32 s3, s10, s5
+; MOVREL-NEXT:    s_cmp_lg_u32 s14, 0
 ; MOVREL-NEXT:    v_mov_b32_e32 v3, s3
+; MOVREL-NEXT:    s_cselect_b32 s4, s10, s6
+; MOVREL-NEXT:    s_cmp_lg_u32 s15, 0
 ; MOVREL-NEXT:    v_mov_b32_e32 v4, s4
+; MOVREL-NEXT:    s_cselect_b32 s5, s10, s7
+; MOVREL-NEXT:    s_cmp_lg_u32 s16, 0
 ; MOVREL-NEXT:    v_mov_b32_e32 v5, s5
+; MOVREL-NEXT:    s_cselect_b32 s6, s10, s8
+; MOVREL-NEXT:    s_cmp_lg_u32 s11, 0
 ; MOVREL-NEXT:    v_mov_b32_e32 v6, s6
+; MOVREL-NEXT:    s_cselect_b32 s7, s10, s9
 ; MOVREL-NEXT:    v_mov_b32_e32 v7, s7
-; MOVREL-NEXT:    ; implicit-def: $vcc_hi
 ; MOVREL-NEXT:    ; return to shader part epilog
 entry:
   %idx.add = add i32 %idx, 1
   %insert = insertelement <8 x float> %vec, float %val, i32 %idx.add
   ret <8 x float> %insert
 }
 
 define amdgpu_ps <8 x float> @dyn_insertelement_v8f32_s_s_s_add_7(<8 x float> inreg %vec, float inreg %val, i32 inreg %idx) {
 ; GPRIDX-LABEL: dyn_insertelement_v8f32_s_s_s_add_7:
 ; GPRIDX:       ; %bb.0: ; %entry
-; GPRIDX-NEXT:    s_mov_b32 s0, s2
-; GPRIDX-NEXT:    s_mov_b32 s1, s3
-; GPRIDX-NEXT:    s_mov_b32 s2, s4
-; GPRIDX-NEXT:    s_mov_b32 s3, s5
-; GPRIDX-NEXT:    s_mov_b32 s4, s6
-; GPRIDX-NEXT:    s_mov_b32 s5, s7
-; GPRIDX-NEXT:    s_mov_b32 s6, s8
-; GPRIDX-NEXT:    s_mov_b32 s7, s9
-; GPRIDX-NEXT:    s_mov_b32 m0, s11
-; GPRIDX-NEXT:    s_nop 0
-; GPRIDX-NEXT:    s_movreld_b32 s7, s10
+; GPRIDX-NEXT:    s_add_i32 s11, s11, 7
+; GPRIDX-NEXT:    s_cmp_eq_u32 s11, 0
+; GPRIDX-NEXT:    s_cselect_b32 s0, s10, s2
+; GPRIDX-NEXT:    s_cmp_eq_u32 s11, 1
+; GPRIDX-NEXT:    s_cselect_b32 s1, s10, s3
+; GPRIDX-NEXT:    s_cmp_eq_u32 s11, 2
+; GPRIDX-NEXT:    s_cselect_b32 s2, s10, s4
+; GPRIDX-NEXT:    s_cmp_eq_u32 s11, 3
+; GPRIDX-NEXT:    s_cselect_b32 s3, s10, s5
+; GPRIDX-NEXT:    s_cmp_eq_u32 s11, 4
+; GPRIDX-NEXT:    s_cselect_b32 s4, s10, s6
+; GPRIDX-NEXT:    s_cmp_eq_u32 s11, 5
+; GPRIDX-NEXT:    s_cselect_b32 s5, s10, s7
+; GPRIDX-NEXT:    s_cmp_eq_u32 s11, 6
+; GPRIDX-NEXT:    s_cselect_b32 s6, s10, s8
+; GPRIDX-NEXT:    s_cmp_eq_u32 s11, 7
+; GPRIDX-NEXT:    s_cselect_b32 s7, s10, s9
 ; GPRIDX-NEXT:    v_mov_b32_e32 v0, s0
 ; GPRIDX-NEXT:    v_mov_b32_e32 v1, s1
 ; GPRIDX-NEXT:    v_mov_b32_e32 v2, s2
 ; GPRIDX-NEXT:    v_mov_b32_e32 v3, s3
 ; GPRIDX-NEXT:    v_mov_b32_e32 v4, s4
 ; GPRIDX-NEXT:    v_mov_b32_e32 v5, s5
 ; GPRIDX-NEXT:    v_mov_b32_e32 v6, s6
 ; GPRIDX-NEXT:    v_mov_b32_e32 v7, s7
 ; GPRIDX-NEXT:    ; return to shader part epilog
 ;
 ; MOVREL-LABEL: dyn_insertelement_v8f32_s_s_s_add_7:
 ; MOVREL:       ; %bb.0: ; %entry
-; MOVREL-NEXT:    s_mov_b32 s1, s3
-; MOVREL-NEXT:    s_mov_b32 s3, s5
-; MOVREL-NEXT:    s_mov_b32 s5, s7
-; MOVREL-NEXT:    s_mov_b32 s7, s9
-; MOVREL-NEXT:    s_mov_b32 m0, s11
-; MOVREL-NEXT:    s_mov_b32 s0, s2
-; MOVREL-NEXT:    s_mov_b32 s2, s4
-; MOVREL-NEXT:    s_mov_b32 s4, s6
-; MOVREL-NEXT:    s_mov_b32 s6, s8
-; MOVREL-NEXT:    s_movreld_b32 s7, s10
+; MOVREL-NEXT:    s_add_i32 s11, s11, 7
+; MOVREL-NEXT:    ; implicit-def: $vcc_hi
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 s0, s11, 0
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 s1, s11, 1
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 s12, s11, 2
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 s13, s11, 3
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 s14, s11, 4
+; MOVREL-NEXT:    s_cmp_lg_u32 s0, 0
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 s15, s11, 5
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 s16, s11, 6
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 s11, s11, 7
+; MOVREL-NEXT:    s_cselect_b32 s0, s10, s2
+; MOVREL-NEXT:    s_cmp_lg_u32 s1, 0
 ; MOVREL-NEXT:    v_mov_b32_e32 v0, s0
+; MOVREL-NEXT:    s_cselect_b32 s1, s10, s3
+; MOVREL-NEXT:    s_cmp_lg_u32 s12, 0
 ; MOVREL-NEXT:    v_mov_b32_e32 v1, s1
+; MOVREL-NEXT:    s_cselect_b32 s2, s10, s4
+; MOVREL-NEXT:    s_cmp_lg_u32 s13, 0
 ; MOVREL-NEXT:    v_mov_b32_e32 v2, s2
+; MOVREL-NEXT:    s_cselect_b32 s3, s10, s5
+; MOVREL-NEXT:    s_cmp_lg_u32 s14, 0
 ; MOVREL-NEXT:    v_mov_b32_e32 v3, s3
+; MOVREL-NEXT:    s_cselect_b32 s4, s10, s6
+; MOVREL-NEXT:    s_cmp_lg_u32 s15, 0
 ; MOVREL-NEXT:    v_mov_b32_e32 v4, s4
+; MOVREL-NEXT:    s_cselect_b32 s5, s10, s7
+; MOVREL-NEXT:    s_cmp_lg_u32 s16, 0
 ; MOVREL-NEXT:    v_mov_b32_e32 v5, s5
+; MOVREL-NEXT:    s_cselect_b32 s6, s10, s8
+; MOVREL-NEXT:    s_cmp_lg_u32 s11, 0
 ; MOVREL-NEXT:    v_mov_b32_e32 v6, s6
+; MOVREL-NEXT:    s_cselect_b32 s7, s10, s9
 ; MOVREL-NEXT:    v_mov_b32_e32 v7, s7
-; MOVREL-NEXT:    ; implicit-def: $vcc_hi
 ; MOVREL-NEXT:    ; return to shader part epilog
 entry:
   %idx.add = add i32 %idx, 7
   %insert = insertelement <8 x float> %vec, float %val, i32 %idx.add
   ret <8 x float> %insert
 }
 
 define amdgpu_ps <8 x float> @dyn_insertelement_v8f32_v_v_v_add_1(<8 x float> %vec, float %val, i32 %idx) {
 ; GPRIDX-LABEL: dyn_insertelement_v8f32_v_v_v_add_1:
 ; GPRIDX:       ; %bb.0: ; %entry
-; GPRIDX-NEXT:    s_mov_b64 s[0:1], exec
-; GPRIDX-NEXT:  BB29_1: ; =>This Inner Loop Header: Depth=1
-; GPRIDX-NEXT:    v_readfirstlane_b32 s2, v9
-; GPRIDX-NEXT:    v_cmp_eq_u32_e32 vcc, s2, v9
-; GPRIDX-NEXT:    s_set_gpr_idx_on s2, gpr_idx(DST)
-; GPRIDX-NEXT:    v_mov_b32_e32 v17, v7
-; GPRIDX-NEXT:    v_mov_b32_e32 v16, v6
-; GPRIDX-NEXT:    v_mov_b32_e32 v15, v5
-; GPRIDX-NEXT:    v_mov_b32_e32 v14, v4
-; GPRIDX-NEXT:    v_mov_b32_e32 v13, v3
-; GPRIDX-NEXT:    v_mov_b32_e32 v12, v2
-; GPRIDX-NEXT:    v_mov_b32_e32 v11, v1
-; GPRIDX-NEXT:    v_mov_b32_e32 v10, v0
-; GPRIDX-NEXT:    v_mov_b32_e32 v11, v8
-; GPRIDX-NEXT:    s_set_gpr_idx_off
-; GPRIDX-NEXT:    s_and_saveexec_b64 vcc, vcc
-; GPRIDX-NEXT:    s_xor_b64 exec, exec, vcc
-; GPRIDX-NEXT:    s_cbranch_execnz BB29_1
-; GPRIDX-NEXT:  ; %bb.2:
-; GPRIDX-NEXT:    s_mov_b64 exec, s[0:1]
-; GPRIDX-NEXT:    v_mov_b32_e32 v0, v10
-; GPRIDX-NEXT:    v_mov_b32_e32 v1, v11
-; GPRIDX-NEXT:    v_mov_b32_e32 v2, v12
-; GPRIDX-NEXT:    v_mov_b32_e32 v3, v13
-; GPRIDX-NEXT:    v_mov_b32_e32 v4, v14
-; GPRIDX-NEXT:    v_mov_b32_e32 v5, v15
-; GPRIDX-NEXT:    v_mov_b32_e32 v6, v16
-; GPRIDX-NEXT:    v_mov_b32_e32 v7, v17
+; GPRIDX-NEXT:    v_add_u32_e32 v9, 1, v9
+; GPRIDX-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v9
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v0, v0, v8, vcc
+; GPRIDX-NEXT:    v_cmp_eq_u32_e32 vcc, 1, v9
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v1, v1, v8, vcc
+; GPRIDX-NEXT:    v_cmp_eq_u32_e32 vcc, 2, v9
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v2, v2, v8, vcc
+; GPRIDX-NEXT:    v_cmp_eq_u32_e32 vcc, 3, v9
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v3, v3, v8, vcc
+; GPRIDX-NEXT:    v_cmp_eq_u32_e32 vcc, 4, v9
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v4, v4, v8, vcc
+; GPRIDX-NEXT:    v_cmp_eq_u32_e32 vcc, 5, v9
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v5, v5, v8, vcc
+; GPRIDX-NEXT:    v_cmp_eq_u32_e32 vcc, 6, v9
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v6, v6, v8, vcc
+; GPRIDX-NEXT:    v_cmp_eq_u32_e32 vcc, 7, v9
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v7, v7, v8, vcc
 ; GPRIDX-NEXT:    ; return to shader part epilog
 ;
 ; MOVREL-LABEL: dyn_insertelement_v8f32_v_v_v_add_1:
 ; MOVREL:       ; %bb.0: ; %entry
-; MOVREL-NEXT:    s_mov_b32 s0, exec_lo
+; MOVREL-NEXT:    v_add_nc_u32_e32 v9, 1, v9
 ; MOVREL-NEXT:    ; implicit-def: $vcc_hi
-; MOVREL-NEXT:  BB29_1: ; =>This Inner Loop Header: Depth=1
-; MOVREL-NEXT:    v_readfirstlane_b32 s1, v9
-; MOVREL-NEXT:    v_mov_b32_e32 v17, v7
-; MOVREL-NEXT:    v_mov_b32_e32 v11, v1
-; MOVREL-NEXT:    v_mov_b32_e32 v16, v6
-; MOVREL-NEXT:    v_mov_b32_e32 v15, v5
-; MOVREL-NEXT:    v_cmp_eq_u32_e32 vcc_lo, s1, v9
-; MOVREL-NEXT:    s_mov_b32 m0, s1
-; MOVREL-NEXT:    v_mov_b32_e32 v14, v4
-; MOVREL-NEXT:    v_mov_b32_e32 v13, v3
-; MOVREL-NEXT:    v_mov_b32_e32 v12, v2
-; MOVREL-NEXT:    v_mov_b32_e32 v10, v0
-; MOVREL-NEXT:    v_movreld_b32_e32 v11, v8
-; MOVREL-NEXT:    s_and_saveexec_b32 vcc_lo, vcc_lo
-; MOVREL-NEXT:    s_xor_b32 exec_lo, exec_lo, vcc_lo
-; MOVREL-NEXT:    s_cbranch_execnz BB29_1
-; MOVREL-NEXT:  ; %bb.2:
-; MOVREL-NEXT:    s_mov_b32 exec_lo, s0
-; MOVREL-NEXT:    v_mov_b32_e32 v0, v10
-; MOVREL-NEXT:    v_mov_b32_e32 v1, v11
-; MOVREL-NEXT:    v_mov_b32_e32 v2, v12
-; MOVREL-NEXT:    v_mov_b32_e32 v3, v13
-; MOVREL-NEXT:    v_mov_b32_e32 v4, v14
-; MOVREL-NEXT:    v_mov_b32_e32 v5, v15
-; MOVREL-NEXT:    v_mov_b32_e32 v6, v16
-; MOVREL-NEXT:    v_mov_b32_e32 v7, v17
+; MOVREL-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 0, v9
+; MOVREL-NEXT:    v_cndmask_b32_e32 v0, v0, v8, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 1, v9
+; MOVREL-NEXT:    v_cndmask_b32_e32 v1, v1, v8, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 2, v9
+; MOVREL-NEXT:    v_cndmask_b32_e32 v2, v2, v8, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 3, v9
+; MOVREL-NEXT:    v_cndmask_b32_e32 v3, v3, v8, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 4, v9
+; MOVREL-NEXT:    v_cndmask_b32_e32 v4, v4, v8, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 5, v9
+; MOVREL-NEXT:    v_cndmask_b32_e32 v5, v5, v8, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 6, v9
+; MOVREL-NEXT:    v_cndmask_b32_e32 v6, v6, v8, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 7, v9
+; MOVREL-NEXT:    v_cndmask_b32_e32 v7, v7, v8, vcc_lo
 ; MOVREL-NEXT:    ; return to shader part epilog
 entry:
   %idx.add = add i32 %idx, 1
   %insert = insertelement <8 x float> %vec, float %val, i32 %idx.add
   ret <8 x float> %insert
 }
 
 define amdgpu_ps <8 x float> @dyn_insertelement_v8f32_v_v_v_add_7(<8 x float> %vec, float %val, i32 %idx) {
 ; GPRIDX-LABEL: dyn_insertelement_v8f32_v_v_v_add_7:
 ; GPRIDX:       ; %bb.0: ; %entry
-; GPRIDX-NEXT:    s_mov_b64 s[0:1], exec
-; GPRIDX-NEXT:  BB30_1: ; =>This Inner Loop Header: Depth=1
-; GPRIDX-NEXT:    v_readfirstlane_b32 s2, v9
-; GPRIDX-NEXT:    v_cmp_eq_u32_e32 vcc, s2, v9
-; GPRIDX-NEXT:    s_set_gpr_idx_on s2, gpr_idx(DST)
-; GPRIDX-NEXT:    v_mov_b32_e32 v17, v7
-; GPRIDX-NEXT:    v_mov_b32_e32 v16, v6
-; GPRIDX-NEXT:    v_mov_b32_e32 v15, v5
-; GPRIDX-NEXT:    v_mov_b32_e32 v14, v4
-; GPRIDX-NEXT:    v_mov_b32_e32 v13, v3
-; GPRIDX-NEXT:    v_mov_b32_e32 v12, v2
-; GPRIDX-NEXT:    v_mov_b32_e32 v11, v1
-; GPRIDX-NEXT:    v_mov_b32_e32 v10, v0
-; GPRIDX-NEXT:    v_mov_b32_e32 v17, v8
-; GPRIDX-NEXT:    s_set_gpr_idx_off
-; GPRIDX-NEXT:    s_and_saveexec_b64 vcc, vcc
-; GPRIDX-NEXT:    s_xor_b64 exec, exec, vcc
-; GPRIDX-NEXT:    s_cbranch_execnz BB30_1
-; GPRIDX-NEXT:  ; %bb.2:
-; GPRIDX-NEXT:    s_mov_b64 exec, s[0:1]
-; GPRIDX-NEXT:    v_mov_b32_e32 v0, v10
-; GPRIDX-NEXT:    v_mov_b32_e32 v1, v11
-; GPRIDX-NEXT:    v_mov_b32_e32 v2, v12
-; GPRIDX-NEXT:    v_mov_b32_e32 v3, v13
-; GPRIDX-NEXT:    v_mov_b32_e32 v4, v14
-; GPRIDX-NEXT:    v_mov_b32_e32 v5, v15
-; GPRIDX-NEXT:    v_mov_b32_e32 v6, v16
-; GPRIDX-NEXT:    v_mov_b32_e32 v7, v17
+; GPRIDX-NEXT:    v_add_u32_e32 v9, 7, v9
+; GPRIDX-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v9
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v0, v0, v8, vcc
+; GPRIDX-NEXT:    v_cmp_eq_u32_e32 vcc, 1, v9
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v1, v1, v8, vcc
+; GPRIDX-NEXT:    v_cmp_eq_u32_e32 vcc, 2, v9
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v2, v2, v8, vcc
+; GPRIDX-NEXT:    v_cmp_eq_u32_e32 vcc, 3, v9
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v3, v3, v8, vcc
+; GPRIDX-NEXT:    v_cmp_eq_u32_e32 vcc, 4, v9
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v4, v4, v8, vcc
+; GPRIDX-NEXT:    v_cmp_eq_u32_e32 vcc, 5, v9
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v5, v5, v8, vcc
+; GPRIDX-NEXT:    v_cmp_eq_u32_e32 vcc, 6, v9
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v6, v6, v8, vcc
+; GPRIDX-NEXT:    v_cmp_eq_u32_e32 vcc, 7, v9
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v7, v7, v8, vcc
 ; GPRIDX-NEXT:    ; return to shader part epilog
 ;
 ; MOVREL-LABEL: dyn_insertelement_v8f32_v_v_v_add_7:
 ; MOVREL:       ; %bb.0: ; %entry
-; MOVREL-NEXT:    s_mov_b32 s0, exec_lo
+; MOVREL-NEXT:    v_add_nc_u32_e32 v9, 7, v9
 ; MOVREL-NEXT:    ; implicit-def: $vcc_hi
-; MOVREL-NEXT:  BB30_1: ; =>This Inner Loop Header: Depth=1
-; MOVREL-NEXT:    v_readfirstlane_b32 s1, v9
-; MOVREL-NEXT:    v_mov_b32_e32 v17, v7
-; MOVREL-NEXT:    v_mov_b32_e32 v16, v6
-; MOVREL-NEXT:    v_mov_b32_e32 v15, v5
-; MOVREL-NEXT:    v_mov_b32_e32 v14, v4
-; MOVREL-NEXT:    v_cmp_eq_u32_e32 vcc_lo, s1, v9
-; MOVREL-NEXT:    s_mov_b32 m0, s1
-; MOVREL-NEXT:    v_mov_b32_e32 v13, v3
-; MOVREL-NEXT:    v_mov_b32_e32 v12, v2
-; MOVREL-NEXT:    v_mov_b32_e32 v11, v1
-; MOVREL-NEXT:    v_mov_b32_e32 v10, v0
-; MOVREL-NEXT:    v_movreld_b32_e32 v17, v8
-; MOVREL-NEXT:    s_and_saveexec_b32 vcc_lo, vcc_lo
-; MOVREL-NEXT:    s_xor_b32 exec_lo, exec_lo, vcc_lo
-; MOVREL-NEXT:    s_cbranch_execnz BB30_1
-; MOVREL-NEXT:  ; %bb.2:
-; MOVREL-NEXT:    s_mov_b32 exec_lo, s0
-; MOVREL-NEXT:    v_mov_b32_e32 v0, v10
-; MOVREL-NEXT:    v_mov_b32_e32 v1, v11
-; MOVREL-NEXT:    v_mov_b32_e32 v2, v12
-; MOVREL-NEXT:    v_mov_b32_e32 v3, v13
-; MOVREL-NEXT:    v_mov_b32_e32 v4, v14
-; MOVREL-NEXT:    v_mov_b32_e32 v5, v15
-; MOVREL-NEXT:    v_mov_b32_e32 v6, v16
-; MOVREL-NEXT:    v_mov_b32_e32 v7, v17
+; MOVREL-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 0, v9
+; MOVREL-NEXT:    v_cndmask_b32_e32 v0, v0, v8, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 1, v9
+; MOVREL-NEXT:    v_cndmask_b32_e32 v1, v1, v8, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 2, v9
+; MOVREL-NEXT:    v_cndmask_b32_e32 v2, v2, v8, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 3, v9
+; MOVREL-NEXT:    v_cndmask_b32_e32 v3, v3, v8, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 4, v9
+; MOVREL-NEXT:    v_cndmask_b32_e32 v4, v4, v8, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 5, v9
+; MOVREL-NEXT:    v_cndmask_b32_e32 v5, v5, v8, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 6, v9
+; MOVREL-NEXT:    v_cndmask_b32_e32 v6, v6, v8, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 7, v9
+; MOVREL-NEXT:    v_cndmask_b32_e32 v7, v7, v8, vcc_lo
 ; MOVREL-NEXT:    ; return to shader part epilog
 entry:
   %idx.add = add i32 %idx, 7
   %insert = insertelement <8 x float> %vec, float %val, i32 %idx.add
   ret <8 x float> %insert
 }
 
 define amdgpu_ps void @dyn_insertelement_v8f64_s_s_s_add_1(<8 x double> inreg %vec, double inreg %val, i32 inreg %idx) {
entry:
   store volatile <2 x double> %vec.2, <2 x double> addrspace(1)* undef
   store volatile <2 x double> %vec.3, <2 x double> addrspace(1)* undef
   ret void
 }
 
 define amdgpu_ps void @dyn_insertelement_v8f64_v_v_v_add_1(<8 x double> %vec, double %val, i32 %idx) {
 ; GPRIDX-LABEL: dyn_insertelement_v8f64_v_v_v_add_1:
 ; GPRIDX:       ; %bb.0: ; %entry
-; GPRIDX-NEXT:    s_mov_b64 s[0:1], exec
-; GPRIDX-NEXT:  BB32_1: ; =>This Inner Loop Header: Depth=1
-; GPRIDX-NEXT:    v_readfirstlane_b32 s2, v18
-; GPRIDX-NEXT:    s_add_i32 s3, s2, 1
-; GPRIDX-NEXT:    s_lshl_b32 s3, s3, 1
-; GPRIDX-NEXT:    v_cmp_eq_u32_e32 vcc, s2, v18
-; GPRIDX-NEXT:    s_set_gpr_idx_on s3, gpr_idx(DST)
-; GPRIDX-NEXT:    v_mov_b32_e32 v34, v15
-; GPRIDX-NEXT:    v_mov_b32_e32 v33, v14
-; GPRIDX-NEXT:    v_mov_b32_e32 v32, v13
-; GPRIDX-NEXT:    v_mov_b32_e32 v31, v12
-; GPRIDX-NEXT:    v_mov_b32_e32 v30, v11
-; GPRIDX-NEXT:    v_mov_b32_e32 v29, v10
-; GPRIDX-NEXT:    v_mov_b32_e32 v28, v9
-; GPRIDX-NEXT:    v_mov_b32_e32 v27, v8
-; GPRIDX-NEXT:    v_mov_b32_e32 v26, v7
-; GPRIDX-NEXT:    v_mov_b32_e32 v25, v6
-; GPRIDX-NEXT:    v_mov_b32_e32 v24, v5
-; GPRIDX-NEXT:    v_mov_b32_e32 v23, v4
-; GPRIDX-NEXT:    v_mov_b32_e32 v22, v3
-; GPRIDX-NEXT:    v_mov_b32_e32 v21, v2
-; GPRIDX-NEXT:    v_mov_b32_e32 v20, v1
-; GPRIDX-NEXT:    v_mov_b32_e32 v19, v0
-; GPRIDX-NEXT:    v_mov_b32_e32 v19, v16
-; GPRIDX-NEXT:    s_set_gpr_idx_off
-; GPRIDX-NEXT:    s_set_gpr_idx_on s3, gpr_idx(DST)
-; GPRIDX-NEXT:    v_mov_b32_e32 v20, v17
-; GPRIDX-NEXT:    s_set_gpr_idx_off
-; GPRIDX-NEXT:    s_and_saveexec_b64 vcc, vcc
-; GPRIDX-NEXT:    s_xor_b64 exec, exec, vcc
-; GPRIDX-NEXT:    s_cbranch_execnz BB32_1
-; GPRIDX-NEXT:  ; %bb.2:
-; GPRIDX-NEXT:    s_mov_b64 exec, s[0:1]
-; GPRIDX-NEXT:    global_store_dwordx4 v[0:1], v[19:22], off
-; GPRIDX-NEXT:    global_store_dwordx4 v[0:1], v[23:26], off
-; GPRIDX-NEXT:    global_store_dwordx4 v[0:1], v[27:30], off
-; GPRIDX-NEXT:    global_store_dwordx4 v[0:1], v[31:34], off
+; GPRIDX-NEXT:    v_add_u32_e32 v18, 1, v18
+; GPRIDX-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v18
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 s[0:1], 1, v18
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 s[2:3], 2, v18
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 s[4:5], 3, v18
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 s[6:7], 4, v18
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 s[8:9], 5, v18
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 s[10:11], 7, v18
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 s[12:13], 6, v18
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v0, v0, v16, vcc
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v2, v2, v16, s[0:1]
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v1, v1, v17, vcc
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v3, v3, v17, s[0:1]
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v4, v4, v16, s[2:3]
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v6, v6, v16, s[4:5]
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v8, v8, v16, s[6:7]
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v10, v10, v16, s[8:9]
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v12, v12, v16, s[12:13]
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v14, v14, v16, s[10:11]
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v5, v5, v17, s[2:3]
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v7, v7, v17, s[4:5]
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v9, v9, v17, s[6:7]
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v11, v11, v17, s[8:9]
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v13, v13, v17, s[12:13]
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v15, v15, v17, s[10:11]
+; GPRIDX-NEXT:    global_store_dwordx4 v[0:1], v[0:3], off
+; GPRIDX-NEXT:    global_store_dwordx4 v[0:1], v[4:7], off
+; GPRIDX-NEXT:    global_store_dwordx4 v[0:1], v[8:11], off
+; GPRIDX-NEXT:    global_store_dwordx4 v[0:1], v[12:15], off
 ; GPRIDX-NEXT:    s_endpgm
 ;
 ; MOVREL-LABEL: dyn_insertelement_v8f64_v_v_v_add_1:
 ; MOVREL:       ; %bb.0: ; %entry
-; MOVREL-NEXT:    s_mov_b32 s0, exec_lo
-; MOVREL-NEXT:    ; implicit-def: $vcc_hi
-; MOVREL-NEXT:  BB32_1: ; =>This Inner Loop Header: Depth=1
-; MOVREL-NEXT:    v_readfirstlane_b32 s1, v18
-; MOVREL-NEXT:    v_mov_b32_e32 v34, v15
+; MOVREL-NEXT:    v_add_nc_u32_e32 v18, 1, v18
 ; MOVREL-NEXT:    v_mov_b32_e32 v19, v0
-; MOVREL-NEXT:    v_mov_b32_e32 v33, v14
-; MOVREL-NEXT:    v_mov_b32_e32 v32, v13
-; MOVREL-NEXT:    s_add_i32 s2, s1, 1
-; MOVREL-NEXT:    v_cmp_eq_u32_e32 vcc_lo, s1, v18
-; MOVREL-NEXT:    v_mov_b32_e32 v31, v12
-; MOVREL-NEXT:    v_mov_b32_e32 v30, v11
-; MOVREL-NEXT:    v_mov_b32_e32 v29, v10
-; MOVREL-NEXT:    s_lshl_b32 m0, s2, 1
-; MOVREL-NEXT:    v_mov_b32_e32 v28, v9
-; MOVREL-NEXT:    v_mov_b32_e32 v27, v8
-; MOVREL-NEXT:    v_mov_b32_e32 v26, v7
-; MOVREL-NEXT:    v_mov_b32_e32 v25, v6
-; MOVREL-NEXT:    v_mov_b32_e32 v24, v5
-; MOVREL-NEXT:    v_mov_b32_e32 v23, v4
-; MOVREL-NEXT:    v_mov_b32_e32 v22, v3
-; MOVREL-NEXT:    v_mov_b32_e32 v21, v2
 ; MOVREL-NEXT:    v_mov_b32_e32 v20, v1
-; MOVREL-NEXT:    v_movreld_b32_e32 v19, v16
-; MOVREL-NEXT:    v_movreld_b32_e32 v20, v17
-; MOVREL-NEXT:    s_and_saveexec_b32 vcc_lo, vcc_lo
-; MOVREL-NEXT:    s_xor_b32 exec_lo, exec_lo, vcc_lo
-; MOVREL-NEXT:    s_cbranch_execnz BB32_1
-; MOVREL-NEXT:  ; %bb.2:
-; MOVREL-NEXT:    s_mov_b32 exec_lo, s0
-; MOVREL-NEXT:    global_store_dwordx4 v[0:1], v[19:22], off
-; MOVREL-NEXT:    global_store_dwordx4 v[0:1], v[23:26], off
-; MOVREL-NEXT:    global_store_dwordx4 v[0:1], v[27:30], off
-; MOVREL-NEXT:    global_store_dwordx4 v[0:1], v[31:34], off
+; MOVREL-NEXT:    ; implicit-def: $vcc_hi
+; MOVREL-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 0, v18
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 s0, 1, v18
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 s3, 4, v18
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 s4, 5, v18
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 s1, 2, v18
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 s2, 3, v18
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 s5, 7, v18
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 s6, 6, v18
+; MOVREL-NEXT:    v_cndmask_b32_e32 v0, v19, v16, vcc_lo
+; MOVREL-NEXT:    v_cndmask_b32_e64 v2, v2, v16, s0
+; MOVREL-NEXT:    v_cndmask_b32_e64 v8, v8, v16, s3
+; MOVREL-NEXT:    v_cndmask_b32_e64 v10, v10, v16, s4
+; MOVREL-NEXT:    v_cndmask_b32_e32 v1, v20, v17, vcc_lo
+; MOVREL-NEXT:    v_cndmask_b32_e64 v3, v3, v17, s0
+; MOVREL-NEXT:    v_cndmask_b32_e64 v9, v9, v17, s3
+; MOVREL-NEXT:    v_cndmask_b32_e64 v11, v11, v17, s4
+; MOVREL-NEXT:    v_cndmask_b32_e64 v4, v4, v16, s1
+; MOVREL-NEXT:    v_cndmask_b32_e64 v6, v6, v16, s2
+; MOVREL-NEXT:    v_cndmask_b32_e64 v12, v12, v16, s6
+; MOVREL-NEXT:    v_cndmask_b32_e64 v14, v14, v16, s5
+; MOVREL-NEXT:    v_cndmask_b32_e64 v5, v5, v17, s1
+; MOVREL-NEXT:    v_cndmask_b32_e64 v7, v7, v17, s2
+; MOVREL-NEXT:    v_cndmask_b32_e64 v13, v13, v17, s6
+; MOVREL-NEXT:    v_cndmask_b32_e64 v15, v15, v17, s5
+; MOVREL-NEXT:    global_store_dwordx4 v[0:1], v[0:3], off
+; MOVREL-NEXT:    global_store_dwordx4 v[0:1], v[4:7], off
+; MOVREL-NEXT:    global_store_dwordx4 v[0:1], v[8:11], off
+; MOVREL-NEXT:    global_store_dwordx4 v[0:1], v[12:15], off
 ; MOVREL-NEXT:    s_endpgm
 entry:
   %idx.add = add i32 %idx, 1
   %insert = insertelement <8 x double> %vec, double %val, i32 %idx.add
   %vec.0 = shufflevector <8 x double> %insert, <8 x double> undef, <2 x i32> <i32 0, i32 1>
   %vec.1 = shufflevector <8 x double> %insert, <8 x double> undef, <2 x i32> <i32 2, i32 3>
   %vec.2 = shufflevector <8 x double> %insert, <8 x double> undef, <2 x i32> <i32 4, i32 5>
   %vec.3 = shufflevector <8 x double> %insert, <8 x double> undef, <2 x i32> <i32 6, i32 7>

 entry:
   %insert = insertelement <16 x double> %vec, double %val, i32 %idx
   ret <16 x double> %insert
 }
 
 define amdgpu_ps <7 x i32> @dyn_insertelement_v7i32_s_s_s(<7 x i32> inreg %vec, i32 inreg %val, i32 inreg %idx) {
 ; GPRIDX-LABEL: dyn_insertelement_v7i32_s_s_s:
 ; GPRIDX:       ; %bb.0: ; %entry
-; GPRIDX-NEXT:    s_mov_b32 s0, s2
-; GPRIDX-NEXT:    s_mov_b32 s1, s3
-; GPRIDX-NEXT:    s_mov_b32 s2, s4
-; GPRIDX-NEXT:    s_mov_b32 s3, s5
-; GPRIDX-NEXT:    s_mov_b32 s4, s6
-; GPRIDX-NEXT:    s_mov_b32 s5, s7
-; GPRIDX-NEXT:    s_mov_b32 s6, s8
-; GPRIDX-NEXT:    s_mov_b32 m0, s10
-; GPRIDX-NEXT:    s_nop 0
-; GPRIDX-NEXT:    s_movreld_b32 s0, s9
+; GPRIDX-NEXT:    s_cmp_eq_u32 s10, 0
+; GPRIDX-NEXT:    s_cselect_b32 s0, s9, s2
+; GPRIDX-NEXT:    s_cmp_eq_u32 s10, 1
+; GPRIDX-NEXT:    s_cselect_b32 s1, s9, s3
+; GPRIDX-NEXT:    s_cmp_eq_u32 s10, 2
+; GPRIDX-NEXT:    s_cselect_b32 s2, s9, s4
+; GPRIDX-NEXT:    s_cmp_eq_u32 s10, 3
+; GPRIDX-NEXT:    s_cselect_b32 s3, s9, s5
+; GPRIDX-NEXT:    s_cmp_eq_u32 s10, 4
+; GPRIDX-NEXT:    s_cselect_b32 s4, s9, s6
+; GPRIDX-NEXT:    s_cmp_eq_u32 s10, 5
+; GPRIDX-NEXT:    s_cselect_b32 s5, s9, s7
+; GPRIDX-NEXT:    s_cmp_eq_u32 s10, 6
+; GPRIDX-NEXT:    s_cselect_b32 s6, s9, s8
 ; GPRIDX-NEXT:    ; return to shader part epilog
 ;
 ; MOVREL-LABEL: dyn_insertelement_v7i32_s_s_s:
 ; MOVREL:       ; %bb.0: ; %entry
-; MOVREL-NEXT:    s_mov_b32 s0, s2
-; MOVREL-NEXT:    s_mov_b32 m0, s10
-; MOVREL-NEXT:    s_mov_b32 s1, s3
-; MOVREL-NEXT:    s_mov_b32 s2, s4
-; MOVREL-NEXT:    s_mov_b32 s3, s5
-; MOVREL-NEXT:    s_mov_b32 s4, s6
-; MOVREL-NEXT:    s_mov_b32 s5, s7
-; MOVREL-NEXT:    s_mov_b32 s6, s8
-; MOVREL-NEXT:    s_movreld_b32 s0, s9
-; MOVREL-NEXT:    ; implicit-def: $vcc_hi
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 s0, s10, 0
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 s1, s10, 1
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 s11, s10, 2
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 s12, s10, 3
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 s13, s10, 4
+; MOVREL-NEXT:    s_cmp_lg_u32 s0, 0
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 s14, s10, 5
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 s10, s10, 6
+; MOVREL-NEXT:    ; implicit-def: $vcc_hi
+; MOVREL-NEXT:    s_cselect_b32 s0, s9, s2
+; MOVREL-NEXT:    s_cmp_lg_u32 s1, 0
+; MOVREL-NEXT:    s_cselect_b32 s1, s9, s3
+; MOVREL-NEXT:    s_cmp_lg_u32 s11, 0
+; MOVREL-NEXT:    s_cselect_b32 s2, s9, s4
+; MOVREL-NEXT:    s_cmp_lg_u32 s12, 0
+; MOVREL-NEXT:    s_cselect_b32 s3, s9, s5
+; MOVREL-NEXT:    s_cmp_lg_u32 s13, 0
+; MOVREL-NEXT:    s_cselect_b32 s4, s9, s6
+; MOVREL-NEXT:    s_cmp_lg_u32 s14, 0
+; MOVREL-NEXT:    s_cselect_b32 s5, s9, s7
+; MOVREL-NEXT:    s_cmp_lg_u32 s10, 0
+; MOVREL-NEXT:    s_cselect_b32 s6, s9, s8
 ; MOVREL-NEXT:    ; return to shader part epilog
 entry:
   %insert = insertelement <7 x i32> %vec, i32 %val, i32 %idx
   ret <7 x i32> %insert
 }
 
 define amdgpu_ps <7 x i8 addrspace(3)*> @dyn_insertelement_v7p3i8_s_s_s(<7 x i8 addrspace(3)*> inreg %vec, i8 addrspace(3)* inreg %val, i32 inreg %idx) {
 ; GPRIDX-LABEL: dyn_insertelement_v7p3i8_s_s_s:
 ; GPRIDX:       ; %bb.0: ; %entry
-; GPRIDX-NEXT:    s_mov_b32 s0, s2
-; GPRIDX-NEXT:    s_mov_b32 s1, s3
-; GPRIDX-NEXT:    s_mov_b32 s2, s4
-; GPRIDX-NEXT:    s_mov_b32 s3, s5
-; GPRIDX-NEXT:    s_mov_b32 s4, s6
-; GPRIDX-NEXT:    s_mov_b32 s5, s7
-; GPRIDX-NEXT:    s_mov_b32 s6, s8
-; GPRIDX-NEXT:    s_mov_b32 m0, s10
-; GPRIDX-NEXT:    s_nop 0
-; GPRIDX-NEXT:    s_movreld_b32 s0, s9
+; GPRIDX-NEXT:    s_cmp_eq_u32 s10, 0
+; GPRIDX-NEXT:    s_cselect_b32 s0, s9, s2
+; GPRIDX-NEXT:    s_cmp_eq_u32 s10, 1
+; GPRIDX-NEXT:    s_cselect_b32 s1, s9, s3
+; GPRIDX-NEXT:    s_cmp_eq_u32 s10, 2
+; GPRIDX-NEXT:    s_cselect_b32 s2, s9, s4
+; GPRIDX-NEXT:    s_cmp_eq_u32 s10, 3
+; GPRIDX-NEXT:    s_cselect_b32 s3, s9, s5
+; GPRIDX-NEXT:    s_cmp_eq_u32 s10, 4
+; GPRIDX-NEXT:    s_cselect_b32 s4, s9, s6
+; GPRIDX-NEXT:    s_cmp_eq_u32 s10, 5
+; GPRIDX-NEXT:    s_cselect_b32 s5, s9, s7
+; GPRIDX-NEXT:    s_cmp_eq_u32 s10, 6
+; GPRIDX-NEXT:    s_cselect_b32 s6, s9, s8
 ; GPRIDX-NEXT:    ; return to shader part epilog
 ;
 ; MOVREL-LABEL: dyn_insertelement_v7p3i8_s_s_s:
 ; MOVREL:       ; %bb.0: ; %entry
-; MOVREL-NEXT:    s_mov_b32 s0, s2
-; MOVREL-NEXT:    s_mov_b32 m0, s10
-; MOVREL-NEXT:    s_mov_b32 s1, s3
-; MOVREL-NEXT:    s_mov_b32 s2, s4
-; MOVREL-NEXT:    s_mov_b32 s3, s5
-; MOVREL-NEXT:    s_mov_b32 s4, s6
-; MOVREL-NEXT:    s_mov_b32 s5, s7
-; MOVREL-NEXT:    s_mov_b32 s6, s8
-; MOVREL-NEXT:    s_movreld_b32 s0, s9
-; MOVREL-NEXT:    ; implicit-def: $vcc_hi
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 s0, s10, 0
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 s1, s10, 1
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 s11, s10, 2
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 s12, s10, 3
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 s13, s10, 4
+; MOVREL-NEXT:    s_cmp_lg_u32 s0, 0
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 s14, s10, 5
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 s10, s10, 6
+; MOVREL-NEXT:    ; implicit-def: $vcc_hi
+; MOVREL-NEXT:    s_cselect_b32 s0, s9, s2
+; MOVREL-NEXT:    s_cmp_lg_u32 s1, 0
+; MOVREL-NEXT:    s_cselect_b32 s1, s9, s3
+; MOVREL-NEXT:    s_cmp_lg_u32 s11, 0
+; MOVREL-NEXT:    s_cselect_b32 s2, s9, s4
+; MOVREL-NEXT:    s_cmp_lg_u32 s12, 0
+; MOVREL-NEXT:    s_cselect_b32 s3, s9, s5
+; MOVREL-NEXT:    s_cmp_lg_u32 s13, 0
+; MOVREL-NEXT:    s_cselect_b32 s4, s9, s6
+; MOVREL-NEXT:    s_cmp_lg_u32 s14, 0
+; MOVREL-NEXT:    s_cselect_b32 s5, s9, s7
+; MOVREL-NEXT:    s_cmp_lg_u32 s10, 0
+; MOVREL-NEXT:    s_cselect_b32 s6, s9, s8
 ; MOVREL-NEXT:    ; return to shader part epilog
 entry:
   %insert = insertelement <7 x i8 addrspace(3)*> %vec, i8 addrspace(3)* %val, i32 %idx
   ret <7 x i8 addrspace(3)*> %insert
 }
 
 define amdgpu_ps <7 x float> @dyn_insertelement_v7f32_s_v_s(<7 x float> inreg %vec, float %val, i32 inreg %idx) {
 ; GPRIDX-LABEL: dyn_insertelement_v7f32_s_v_s:
 ; GPRIDX:       ; %bb.0: ; %entry
 ; GPRIDX-NEXT:    s_mov_b32 s0, s2
 ; GPRIDX-NEXT:    s_mov_b32 s1, s3
 ; GPRIDX-NEXT:    s_mov_b32 s2, s4
 ; GPRIDX-NEXT:    s_mov_b32 s3, s5
 ; GPRIDX-NEXT:    s_mov_b32 s4, s6
 ; GPRIDX-NEXT:    s_mov_b32 s5, s7
 ; GPRIDX-NEXT:    s_mov_b32 s6, s8
-; GPRIDX-NEXT:    v_mov_b32_e32 v8, v0
-; GPRIDX-NEXT:    v_mov_b32_e32 v0, s0
-; GPRIDX-NEXT:    v_mov_b32_e32 v1, s1
-; GPRIDX-NEXT:    v_mov_b32_e32 v2, s2
-; GPRIDX-NEXT:    v_mov_b32_e32 v3, s3
-; GPRIDX-NEXT:    v_mov_b32_e32 v4, s4
-; GPRIDX-NEXT:    v_mov_b32_e32 v5, s5
-; GPRIDX-NEXT:    v_mov_b32_e32 v6, s6
-; GPRIDX-NEXT:    v_mov_b32_e32 v7, s7
-; GPRIDX-NEXT:    s_set_gpr_idx_on s9, gpr_idx(DST)
-; GPRIDX-NEXT:    v_mov_b32_e32 v0, v8
-; GPRIDX-NEXT:    s_set_gpr_idx_off
+; GPRIDX-NEXT:    v_mov_b32_e32 v14, s7
+; GPRIDX-NEXT:    v_mov_b32_e32 v7, s0
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 vcc, s9, 0
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v7, v7, v0, vcc
+; GPRIDX-NEXT:    v_mov_b32_e32 v8, s1
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 vcc, s9, 1
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v1, v8, v0, vcc
+; GPRIDX-NEXT:    v_mov_b32_e32 v9, s2
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 vcc, s9, 2
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v2, v9, v0, vcc
+; GPRIDX-NEXT:    v_mov_b32_e32 v10, s3
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 vcc, s9, 3
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v3, v10, v0, vcc
+; GPRIDX-NEXT:    v_mov_b32_e32 v11, s4
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 vcc, s9, 4
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v4, v11, v0, vcc
+; GPRIDX-NEXT:    v_mov_b32_e32 v12, s5
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 vcc, s9, 5
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v5, v12, v0, vcc
+; GPRIDX-NEXT:    v_mov_b32_e32 v13, s6
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 vcc, s9, 6
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v6, v13, v0, vcc
+; GPRIDX-NEXT:    v_mov_b32_e32 v0, v7
 ; GPRIDX-NEXT:    ; return to shader part epilog
 ;
 ; MOVREL-LABEL: dyn_insertelement_v7f32_s_v_s:
 ; MOVREL:       ; %bb.0: ; %entry
 ; MOVREL-NEXT:    s_mov_b32 s0, s2
 ; MOVREL-NEXT:    s_mov_b32 s1, s3
 ; MOVREL-NEXT:    s_mov_b32 s2, s4
 ; MOVREL-NEXT:    s_mov_b32 s3, s5
 ; MOVREL-NEXT:    s_mov_b32 s4, s6
 ; MOVREL-NEXT:    s_mov_b32 s5, s7
 ; MOVREL-NEXT:    s_mov_b32 s6, s8
-; MOVREL-NEXT:    v_mov_b32_e32 v8, v0
-; MOVREL-NEXT:    v_mov_b32_e32 v0, s0
-; MOVREL-NEXT:    s_mov_b32 m0, s9
-; MOVREL-NEXT:    v_mov_b32_e32 v1, s1
-; MOVREL-NEXT:    v_mov_b32_e32 v2, s2
-; MOVREL-NEXT:    v_mov_b32_e32 v3, s3
-; MOVREL-NEXT:    v_mov_b32_e32 v4, s4
-; MOVREL-NEXT:    v_mov_b32_e32 v5, s5
-; MOVREL-NEXT:    v_mov_b32_e32 v6, s6
-; MOVREL-NEXT:    v_mov_b32_e32 v7, s7
-; MOVREL-NEXT:    v_movreld_b32_e32 v0, v8
+; MOVREL-NEXT:    v_mov_b32_e32 v16, s7
+; MOVREL-NEXT:    v_mov_b32_e32 v9, s0
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 vcc_lo, s9, 0
+; MOVREL-NEXT:    v_mov_b32_e32 v10, s1
+; MOVREL-NEXT:    v_mov_b32_e32 v11, s2
+; MOVREL-NEXT:    v_mov_b32_e32 v12, s3
+; MOVREL-NEXT:    v_mov_b32_e32 v13, s4
+; MOVREL-NEXT:    v_cndmask_b32_e32 v7, v9, v0, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 vcc_lo, s9, 1
+; MOVREL-NEXT:    v_mov_b32_e32 v14, s5
+; MOVREL-NEXT:    v_mov_b32_e32 v15, s6
 ; MOVREL-NEXT:    ; implicit-def: $vcc_hi
+; MOVREL-NEXT:    v_cndmask_b32_e32 v1, v10, v0, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 vcc_lo, s9, 2
+; MOVREL-NEXT:    v_cndmask_b32_e32 v2, v11, v0, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 vcc_lo, s9, 3
+; MOVREL-NEXT:    v_cndmask_b32_e32 v3, v12, v0, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 vcc_lo, s9, 4
+; MOVREL-NEXT:    v_cndmask_b32_e32 v4, v13, v0, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 vcc_lo, s9, 5
+; MOVREL-NEXT:    v_cndmask_b32_e32 v5, v14, v0, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 vcc_lo, s9, 6
+; MOVREL-NEXT:    v_cndmask_b32_e32 v6, v15, v0, vcc_lo
+; MOVREL-NEXT:    v_mov_b32_e32 v0, v7
 ; MOVREL-NEXT:    ; return to shader part epilog
 entry:
   %insert = insertelement <7 x float> %vec, float %val, i32 %idx
   ret <7 x float> %insert
 }
 
 define amdgpu_ps <7 x float> @dyn_insertelement_v7f32_s_v_v(<7 x float> inreg %vec, float %val, i32 %idx) {
 ; GPRIDX-LABEL: dyn_insertelement_v7f32_s_v_v:
 ; GPRIDX:       ; %bb.0: ; %entry
 ; GPRIDX-NEXT:    s_mov_b32 s0, s2
 ; GPRIDX-NEXT:    s_mov_b32 s1, s3
 ; GPRIDX-NEXT:    s_mov_b32 s2, s4
 ; GPRIDX-NEXT:    s_mov_b32 s3, s5
 ; GPRIDX-NEXT:    s_mov_b32 s4, s6
 ; GPRIDX-NEXT:    s_mov_b32 s5, s7
 ; GPRIDX-NEXT:    s_mov_b32 s6, s8
-; GPRIDX-NEXT:    v_mov_b32_e32 v17, s7
-; GPRIDX-NEXT:    v_mov_b32_e32 v8, v0
-; GPRIDX-NEXT:    v_mov_b32_e32 v9, v1
-; GPRIDX-NEXT:    v_mov_b32_e32 v16, s6
-; GPRIDX-NEXT:    v_mov_b32_e32 v15, s5
-; GPRIDX-NEXT:    v_mov_b32_e32 v14, s4
-; GPRIDX-NEXT:    v_mov_b32_e32 v13, s3
-; GPRIDX-NEXT:    v_mov_b32_e32 v12, s2
-; GPRIDX-NEXT:    v_mov_b32_e32 v11, s1
-; GPRIDX-NEXT:    v_mov_b32_e32 v10, s0
-; GPRIDX-NEXT:    s_mov_b64 s[0:1], exec
-; GPRIDX-NEXT:  BB46_1: ; =>This Inner Loop Header: Depth=1
-; GPRIDX-NEXT:    v_readfirstlane_b32 s2, v9
-; GPRIDX-NEXT:    v_cmp_eq_u32_e32 vcc, s2, v9
-; GPRIDX-NEXT:    s_set_gpr_idx_on s2, gpr_idx(DST)
-; GPRIDX-NEXT:    v_mov_b32_e32 v0, v10
-; GPRIDX-NEXT:    v_mov_b32_e32 v1, v11
-; GPRIDX-NEXT:    v_mov_b32_e32 v2, v12
-; GPRIDX-NEXT:    v_mov_b32_e32 v3, v13
-; GPRIDX-NEXT:    v_mov_b32_e32 v4, v14
-; GPRIDX-NEXT:    v_mov_b32_e32 v5, v15
-; GPRIDX-NEXT:    v_mov_b32_e32 v6, v16
-; GPRIDX-NEXT:    v_mov_b32_e32 v7, v17
+; GPRIDX-NEXT:    v_mov_b32_e32 v15, s7
+; GPRIDX-NEXT:    v_mov_b32_e32 v8, s0
+; GPRIDX-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v1
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v8, v8, v0, vcc
+; GPRIDX-NEXT:    v_mov_b32_e32 v9, s1
+; GPRIDX-NEXT:    v_cmp_eq_u32_e32 vcc, 1, v1
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v7, v9, v0, vcc
+; GPRIDX-NEXT:    v_mov_b32_e32 v10, s2
+; GPRIDX-NEXT:    v_cmp_eq_u32_e32 vcc, 2, v1
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v2, v10, v0, vcc
+; GPRIDX-NEXT:    v_mov_b32_e32 v11, s3
+; GPRIDX-NEXT:    v_cmp_eq_u32_e32 vcc, 3, v1
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v3, v11, v0, vcc
+; GPRIDX-NEXT:    v_mov_b32_e32 v12, s4
+; GPRIDX-NEXT:    v_cmp_eq_u32_e32 vcc, 4, v1
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v4, v12, v0, vcc
+; GPRIDX-NEXT:    v_mov_b32_e32 v13, s5
+; GPRIDX-NEXT:    v_cmp_eq_u32_e32 vcc, 5, v1
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v5, v13, v0, vcc
+; GPRIDX-NEXT:    v_cmp_eq_u32_e32 vcc, 6, v1
+; GPRIDX-NEXT:    v_mov_b32_e32 v14, s6
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v6, v14, v0, vcc
 ; GPRIDX-NEXT:    v_mov_b32_e32 v0, v8
-; GPRIDX-NEXT:    s_set_gpr_idx_off
-; GPRIDX-NEXT:    s_and_saveexec_b64 vcc, vcc
-; GPRIDX-NEXT:    s_xor_b64 exec, exec, vcc
-; GPRIDX-NEXT:    s_cbranch_execnz BB46_1
-; GPRIDX-NEXT:  ; %bb.2:
-; GPRIDX-NEXT:    s_mov_b64 exec, s[0:1]
+; GPRIDX-NEXT:    v_mov_b32_e32 v1, v7
 ; GPRIDX-NEXT:    ; return to shader part epilog
 ;
 ; MOVREL-LABEL: dyn_insertelement_v7f32_s_v_v:
 ; MOVREL:       ; %bb.0: ; %entry
 ; MOVREL-NEXT:    s_mov_b32 s0, s2
 ; MOVREL-NEXT:    s_mov_b32 s1, s3
 ; MOVREL-NEXT:    s_mov_b32 s2, s4
 ; MOVREL-NEXT:    s_mov_b32 s3, s5
 ; MOVREL-NEXT:    s_mov_b32 s4, s6
 ; MOVREL-NEXT:    s_mov_b32 s5, s7
 ; MOVREL-NEXT:    s_mov_b32 s6, s8
-; MOVREL-NEXT:    v_mov_b32_e32 v17, s7
-; MOVREL-NEXT:    v_mov_b32_e32 v13, s3
-; MOVREL-NEXT:    v_mov_b32_e32 v14, s4
-; MOVREL-NEXT:    v_mov_b32_e32 v15, s5
-; MOVREL-NEXT:    v_mov_b32_e32 v16, s6
-; MOVREL-NEXT:    v_mov_b32_e32 v12, s2
-; MOVREL-NEXT:    v_mov_b32_e32 v11, s1
-; MOVREL-NEXT:    v_mov_b32_e32 v10, s0
-; MOVREL-NEXT:    s_mov_b32 s0, exec_lo
+; MOVREL-NEXT:    v_mov_b32_e32 v16, s7
+; MOVREL-NEXT:    v_mov_b32_e32 v9, s0
+; MOVREL-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 0, v1
+; MOVREL-NEXT:    v_mov_b32_e32 v10, s1
+; MOVREL-NEXT:    v_mov_b32_e32 v11, s2
+; MOVREL-NEXT:    v_mov_b32_e32 v12, s3
+; MOVREL-NEXT:    v_mov_b32_e32 v13, s4
+; MOVREL-NEXT:    v_cndmask_b32_e32 v8, v9, v0, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 1, v1
+; MOVREL-NEXT:    v_mov_b32_e32 v14, s5
+; MOVREL-NEXT:    v_mov_b32_e32 v15, s6
 ; MOVREL-NEXT:    ; implicit-def: $vcc_hi
-; MOVREL-NEXT:  BB46_1: ; =>This Inner Loop Header: Depth=1
-; MOVREL-NEXT:    v_readfirstlane_b32 s1, v1
-; MOVREL-NEXT:    v_mov_b32_e32 v2, v10
-; MOVREL-NEXT:    v_mov_b32_e32 v3, v11
-; MOVREL-NEXT:    v_mov_b32_e32 v4, v12
-; MOVREL-NEXT:    v_mov_b32_e32 v5, v13
-; MOVREL-NEXT:    v_cmp_eq_u32_e32 vcc_lo, s1, v1
-; MOVREL-NEXT:    s_mov_b32 m0, s1
-; MOVREL-NEXT:    v_mov_b32_e32 v6, v14
-; MOVREL-NEXT:    v_mov_b32_e32 v7, v15
-; MOVREL-NEXT:    v_mov_b32_e32 v8, v16
-; MOVREL-NEXT:    v_mov_b32_e32 v9, v17
-; MOVREL-NEXT:    v_movreld_b32_e32 v2, v0
-; MOVREL-NEXT:    s_and_saveexec_b32 vcc_lo, vcc_lo
-; MOVREL-NEXT:    s_xor_b32 exec_lo, exec_lo, vcc_lo
-; MOVREL-NEXT:    s_cbranch_execnz BB46_1
-; MOVREL-NEXT:  ; %bb.2:
-; MOVREL-NEXT:    s_mov_b32 exec_lo, s0
-; MOVREL-NEXT:    v_mov_b32_e32 v0, v2
-; MOVREL-NEXT:    v_mov_b32_e32 v1, v3
-; MOVREL-NEXT:    v_mov_b32_e32 v2, v4
-; MOVREL-NEXT:    v_mov_b32_e32 v3, v5
-; MOVREL-NEXT:    v_mov_b32_e32 v4, v6
-; MOVREL-NEXT:    v_mov_b32_e32 v5, v7
-; MOVREL-NEXT:    v_mov_b32_e32 v6, v8
+; MOVREL-NEXT:    v_cndmask_b32_e32 v7, v10, v0, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 2, v1
+; MOVREL-NEXT:    v_cndmask_b32_e32 v2, v11, v0, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 3, v1
+; MOVREL-NEXT:    v_cndmask_b32_e32 v3, v12, v0, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 4, v1
+; MOVREL-NEXT:    v_cndmask_b32_e32 v4, v13, v0, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 5, v1
+; MOVREL-NEXT:    v_cndmask_b32_e32 v5, v14, v0, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 6, v1
+; MOVREL-NEXT:    v_mov_b32_e32 v1, v7
+; MOVREL-NEXT:    v_cndmask_b32_e32 v6, v15, v0, vcc_lo
+; MOVREL-NEXT:    v_mov_b32_e32 v0, v8
 ; MOVREL-NEXT:    ; return to shader part epilog
 entry:
   %insert = insertelement <7 x float> %vec, float %val, i32 %idx
   ret <7 x float> %insert
 }
 
 define amdgpu_ps <7 x float> @dyn_insertelement_v7f32_v_v_s(<7 x float> %vec, float %val, i32 inreg %idx) {
 ; GPRIDX-LABEL: dyn_insertelement_v7f32_v_v_s:
 ; GPRIDX:       ; %bb.0: ; %entry
-; GPRIDX-NEXT:    s_set_gpr_idx_on s2, gpr_idx(DST)
-; GPRIDX-NEXT:    v_mov_b32_e32 v0, v7
-; GPRIDX-NEXT:    s_set_gpr_idx_off
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 vcc, s2, 0
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v0, v0, v7, vcc
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 vcc, s2, 1
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v1, v1, v7, vcc
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 vcc, s2, 2
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v2, v2, v7, vcc
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 vcc, s2, 3
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v3, v3, v7, vcc
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 vcc, s2, 4
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v4, v4, v7, vcc
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 vcc, s2, 5
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v5, v5, v7, vcc
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 vcc, s2, 6
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v6, v6, v7, vcc
 ; GPRIDX-NEXT:    ; return to shader part epilog
 ;
 ; MOVREL-LABEL: dyn_insertelement_v7f32_v_v_s:
 ; MOVREL:       ; %bb.0: ; %entry
-; MOVREL-NEXT:    s_mov_b32 m0, s2
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 vcc_lo, s2, 0
 ; MOVREL-NEXT:    ; implicit-def: $vcc_hi
-; MOVREL-NEXT:    v_movreld_b32_e32 v0, v7
+; MOVREL-NEXT:    v_cndmask_b32_e32 v0, v0, v7, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 vcc_lo, s2, 1
+; MOVREL-NEXT:    v_cndmask_b32_e32 v1, v1, v7, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 vcc_lo, s2, 2
+; MOVREL-NEXT:    v_cndmask_b32_e32 v2, v2, v7, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 vcc_lo, s2, 3
+; MOVREL-NEXT:    v_cndmask_b32_e32 v3, v3, v7, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 vcc_lo, s2, 4
+; MOVREL-NEXT:    v_cndmask_b32_e32 v4, v4, v7, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 vcc_lo, s2, 5
+; MOVREL-NEXT:    v_cndmask_b32_e32 v5, v5, v7, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 vcc_lo, s2, 6
+; MOVREL-NEXT:    v_cndmask_b32_e32 v6, v6, v7, vcc_lo
 ; MOVREL-NEXT:    ; return to shader part epilog
 entry:
   %insert = insertelement <7 x float> %vec, float %val, i32 %idx
   ret <7 x float> %insert
 }
 
 define amdgpu_ps <7 x float> @dyn_insertelement_v7f32_v_v_v(<7 x float> %vec, float %val, i32 %idx) {
 ; GPRIDX-LABEL: dyn_insertelement_v7f32_v_v_v:
 ; GPRIDX:       ; %bb.0: ; %entry
-; GPRIDX-NEXT:    s_mov_b64 s[0:1], exec
-; GPRIDX-NEXT:  BB48_1: ; =>This Inner Loop Header: Depth=1
-; GPRIDX-NEXT:    v_readfirstlane_b32 s2, v8
-; GPRIDX-NEXT:    v_cmp_eq_u32_e32 vcc, s2, v8
-; GPRIDX-NEXT:    s_set_gpr_idx_on s2, gpr_idx(DST)
-; GPRIDX-NEXT:    v_mov_b32_e32 v16, v7
-; GPRIDX-NEXT:    v_mov_b32_e32 v15, v6
-; GPRIDX-NEXT:    v_mov_b32_e32 v14, v5
-; GPRIDX-NEXT:    v_mov_b32_e32 v13, v4
-; GPRIDX-NEXT:    v_mov_b32_e32 v12, v3
-; GPRIDX-NEXT:    v_mov_b32_e32 v11, v2
-; GPRIDX-NEXT:    v_mov_b32_e32 v10, v1
-; GPRIDX-NEXT:    v_mov_b32_e32 v9, v0
-; GPRIDX-NEXT:    v_mov_b32_e32 v9, v7
-; GPRIDX-NEXT:    s_set_gpr_idx_off
-; GPRIDX-NEXT:    s_and_saveexec_b64 vcc, vcc
-; GPRIDX-NEXT:    s_xor_b64 exec, exec, vcc
-; GPRIDX-NEXT:    s_cbranch_execnz BB48_1
-; GPRIDX-NEXT:  ; %bb.2:
-; GPRIDX-NEXT:    s_mov_b64 exec, s[0:1]
-; GPRIDX-NEXT:    v_mov_b32_e32 v0, v9
-; GPRIDX-NEXT:    v_mov_b32_e32 v1, v10
-; GPRIDX-NEXT:    v_mov_b32_e32 v2, v11
-; GPRIDX-NEXT:    v_mov_b32_e32 v3, v12
-; GPRIDX-NEXT:    v_mov_b32_e32 v4, v13
-; GPRIDX-NEXT:    v_mov_b32_e32 v5, v14
-; GPRIDX-NEXT:    v_mov_b32_e32 v6, v15
+; GPRIDX-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v8
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v0, v0, v7, vcc
+; GPRIDX-NEXT:    v_cmp_eq_u32_e32 vcc, 1, v8
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v1, v1, v7, vcc
+; GPRIDX-NEXT:    v_cmp_eq_u32_e32 vcc, 2, v8
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v2, v2, v7, vcc
+; GPRIDX-NEXT:    v_cmp_eq_u32_e32 vcc, 3, v8
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v3, v3, v7, vcc
+; GPRIDX-NEXT:    v_cmp_eq_u32_e32 vcc, 4, v8
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v4, v4, v7, vcc
+; GPRIDX-NEXT:    v_cmp_eq_u32_e32 vcc, 5, v8
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v5, v5, v7, vcc
+; GPRIDX-NEXT:    v_cmp_eq_u32_e32 vcc, 6, v8
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v6, v6, v7, vcc
 ; GPRIDX-NEXT:    ; return to shader part epilog
 ;
 ; MOVREL-LABEL: dyn_insertelement_v7f32_v_v_v:
 ; MOVREL:       ; %bb.0: ; %entry
-; MOVREL-NEXT:    s_mov_b32 s0, exec_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 0, v8
 ; MOVREL-NEXT:    ; implicit-def: $vcc_hi
-; MOVREL-NEXT:  BB48_1: ; =>This Inner Loop Header: Depth=1
-; MOVREL-NEXT:    v_readfirstlane_b32 s1, v8
-; MOVREL-NEXT:    v_mov_b32_e32 v16, v7
-; MOVREL-NEXT:    v_mov_b32_e32 v9, v0
-; MOVREL-NEXT:    v_mov_b32_e32 v15, v6
-; MOVREL-NEXT:    v_mov_b32_e32 v14, v5
-; MOVREL-NEXT:    v_cmp_eq_u32_e32 vcc_lo, s1, v8
-; MOVREL-NEXT:    s_mov_b32 m0, s1
-; MOVREL-NEXT:    v_mov_b32_e32 v13, v4
-; MOVREL-NEXT:    v_mov_b32_e32 v12, v3
-; MOVREL-NEXT:    v_mov_b32_e32 v11, v2
-; MOVREL-NEXT:    v_mov_b32_e32 v10, v1
-; MOVREL-NEXT:    v_movreld_b32_e32 v9, v7
-; MOVREL-NEXT:    s_and_saveexec_b32 vcc_lo, vcc_lo
-; MOVREL-NEXT:    s_xor_b32 exec_lo, exec_lo, vcc_lo
-; MOVREL-NEXT:    s_cbranch_execnz BB48_1
-; MOVREL-NEXT:  ; %bb.2:
-; MOVREL-NEXT:    s_mov_b32 exec_lo, s0
-; MOVREL-NEXT:    v_mov_b32_e32 v0, v9
-; MOVREL-NEXT:    v_mov_b32_e32 v1, v10
-; MOVREL-NEXT:    v_mov_b32_e32 v2, v11
-; MOVREL-NEXT:    v_mov_b32_e32 v3, v12
-; MOVREL-NEXT:    v_mov_b32_e32 v4, v13
-; MOVREL-NEXT:    v_mov_b32_e32 v5, v14
-; MOVREL-NEXT:    v_mov_b32_e32 v6, v15
+; MOVREL-NEXT:    v_cndmask_b32_e32 v0, v0, v7, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 1, v8
+; MOVREL-NEXT:    v_cndmask_b32_e32 v1, v1, v7, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 2, v8
+; MOVREL-NEXT:    v_cndmask_b32_e32 v2, v2, v7, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 3, v8
+; MOVREL-NEXT:    v_cndmask_b32_e32 v3, v3, v7, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 4, v8
+; MOVREL-NEXT:    v_cndmask_b32_e32 v4, v4, v7, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 5, v8
+; MOVREL-NEXT:    v_cndmask_b32_e32 v5, v5, v7, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 6, v8
+; MOVREL-NEXT:    v_cndmask_b32_e32 v6, v6, v7, vcc_lo
 ; MOVREL-NEXT:    ; return to shader part epilog
 entry:
   %insert = insertelement <7 x float> %vec, float %val, i32 %idx
   ret <7 x float> %insert
 }
 
 define amdgpu_ps <7 x double> @dyn_insertelement_v7f64_s_s_s(<7 x double> inreg %vec, double inreg %val, i32 inreg %idx) {
 ; GPRIDX-LABEL: dyn_insertelement_v7f64_s_s_s:

 ; GPRIDX-NEXT:    s_mov_b32 s6, s8
 ; GPRIDX-NEXT:    s_mov_b32 s7, s9
 ; GPRIDX-NEXT:    s_mov_b32 s8, s10
 ; GPRIDX-NEXT:    s_mov_b32 s9, s11
 ; GPRIDX-NEXT:    s_mov_b32 s10, s12
 ; GPRIDX-NEXT:    s_mov_b32 s11, s13
 ; GPRIDX-NEXT:    s_mov_b32 s12, s14
 ; GPRIDX-NEXT:    s_mov_b32 s13, s15
-; GPRIDX-NEXT:    v_mov_b32_e32 v34, s15
-; GPRIDX-NEXT:    v_mov_b32_e32 v33, s14
-; GPRIDX-NEXT:    v_mov_b32_e32 v32, s13
-; GPRIDX-NEXT:    v_mov_b32_e32 v31, s12
-; GPRIDX-NEXT:    v_mov_b32_e32 v30, s11
-; GPRIDX-NEXT:    v_mov_b32_e32 v29, s10
-; GPRIDX-NEXT:    v_mov_b32_e32 v28, s9
-; GPRIDX-NEXT:    v_mov_b32_e32 v27, s8
-; GPRIDX-NEXT:    v_mov_b32_e32 v26, s7
-; GPRIDX-NEXT:    v_mov_b32_e32 v25, s6
-; GPRIDX-NEXT:    v_mov_b32_e32 v24, s5
-; GPRIDX-NEXT:    v_mov_b32_e32 v23, s4
-; GPRIDX-NEXT:    v_mov_b32_e32 v22, s3
-; GPRIDX-NEXT:    v_mov_b32_e32 v21, s2
-; GPRIDX-NEXT:    v_mov_b32_e32 v20, s1
-; GPRIDX-NEXT:    v_mov_b32_e32 v19, s0
-; GPRIDX-NEXT:    s_mov_b64 s[0:1], exec
-; GPRIDX-NEXT:  BB51_1: ; =>This Inner Loop Header: Depth=1
-; GPRIDX-NEXT:    v_readfirstlane_b32 s2, v2
-; GPRIDX-NEXT:    s_lshl_b32 s3, s2, 1
-; GPRIDX-NEXT:    v_cmp_eq_u32_e32 vcc, s2, v2
-; GPRIDX-NEXT:    s_set_gpr_idx_on s3, gpr_idx(DST)
-; GPRIDX-NEXT:    v_mov_b32_e32 v3, v19
-; GPRIDX-NEXT:    v_mov_b32_e32 v4, v20
-; GPRIDX-NEXT:    v_mov_b32_e32 v5, v21
-; GPRIDX-NEXT:    v_mov_b32_e32 v6, v22
-; GPRIDX-NEXT:    v_mov_b32_e32 v7, v23
-; GPRIDX-NEXT:    v_mov_b32_e32 v8, v24
-; GPRIDX-NEXT:    v_mov_b32_e32 v9, v25
-; GPRIDX-NEXT:    v_mov_b32_e32 v10, v26
-; GPRIDX-NEXT:    v_mov_b32_e32 v11, v27
-; GPRIDX-NEXT:    v_mov_b32_e32 v12, v28
-; GPRIDX-NEXT:    v_mov_b32_e32 v13, v29
-; GPRIDX-NEXT:    v_mov_b32_e32 v14, v30
-; GPRIDX-NEXT:    v_mov_b32_e32 v15, v31
-; GPRIDX-NEXT:    v_mov_b32_e32 v16, v32
-; GPRIDX-NEXT:    v_mov_b32_e32 v17, v33
-; GPRIDX-NEXT:    v_mov_b32_e32 v18, v34
-; GPRIDX-NEXT:    v_mov_b32_e32 v3, v0
-; GPRIDX-NEXT:    s_set_gpr_idx_off
-; GPRIDX-NEXT:    s_set_gpr_idx_on s3, gpr_idx(DST)
-; GPRIDX-NEXT:    v_mov_b32_e32 v4, v1
-; GPRIDX-NEXT:    s_set_gpr_idx_off
-; GPRIDX-NEXT:    s_and_saveexec_b64 vcc, vcc
-; GPRIDX-NEXT:    s_xor_b64 exec, exec, vcc
-; GPRIDX-NEXT:    s_cbranch_execnz BB51_1
-; GPRIDX-NEXT:  ; %bb.2:
-; GPRIDX-NEXT:    s_mov_b64 exec, s[0:1]
+; GPRIDX-NEXT:    v_mov_b32_e32 v18, s15
+; GPRIDX-NEXT:    v_mov_b32_e32 v17, s14
+; GPRIDX-NEXT:    v_mov_b32_e32 v16, s13
+; GPRIDX-NEXT:    v_mov_b32_e32 v15, s12
+; GPRIDX-NEXT:    v_mov_b32_e32 v14, s11
+; GPRIDX-NEXT:    v_mov_b32_e32 v13, s10
+; GPRIDX-NEXT:    v_mov_b32_e32 v12, s9
+; GPRIDX-NEXT:    v_mov_b32_e32 v11, s8
+; GPRIDX-NEXT:    v_mov_b32_e32 v10, s7
+; GPRIDX-NEXT:    v_mov_b32_e32 v9, s6
+; GPRIDX-NEXT:    v_mov_b32_e32 v8, s5
+; GPRIDX-NEXT:    v_mov_b32_e32 v7, s4
+; GPRIDX-NEXT:    v_mov_b32_e32 v6, s3
+; GPRIDX-NEXT:    v_mov_b32_e32 v5, s2
+; GPRIDX-NEXT:    v_mov_b32_e32 v4, s1
+; GPRIDX-NEXT:    v_mov_b32_e32 v3, s0
+; GPRIDX-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v2
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 s[0:1], 2, v2
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 s[2:3], 3, v2
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 s[4:5], 4, v2
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 s[10:11], 1, v2
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 s[6:7], 5, v2
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 s[8:9], 6, v2
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v2, v5, v0, s[10:11]
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v5, v7, v0, s[0:1]
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v7, v9, v0, s[2:3]
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v9, v11, v0, s[4:5]
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v11, v13, v0, s[6:7]
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v3, v3, v0, vcc
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v0, v15, v0, s[8:9]
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v6, v6, v1, s[10:11]
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v10, v10, v1, s[2:3]
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v13, v14, v1, s[6:7]
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v8, v8, v1, s[0:1]
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v12, v12, v1, s[4:5]
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v4, v4, v1, vcc
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v1, v16, v1, s[8:9]
 ; GPRIDX-NEXT:    v_readfirstlane_b32 s0, v3
 ; GPRIDX-NEXT:    v_readfirstlane_b32 s1, v4
-; GPRIDX-NEXT:    v_readfirstlane_b32 s2, v5
+; GPRIDX-NEXT:    v_readfirstlane_b32 s2, v2
 ; GPRIDX-NEXT:    v_readfirstlane_b32 s3, v6
-; GPRIDX-NEXT:    v_readfirstlane_b32 s4, v7
+; GPRIDX-NEXT:    v_readfirstlane_b32 s4, v5
 ; GPRIDX-NEXT:    v_readfirstlane_b32 s5, v8
-; GPRIDX-NEXT:    v_readfirstlane_b32 s6, v9
+; GPRIDX-NEXT:    v_readfirstlane_b32 s6, v7
 ; GPRIDX-NEXT:    v_readfirstlane_b32 s7, v10
-; GPRIDX-NEXT:    v_readfirstlane_b32 s8, v11
+; GPRIDX-NEXT:    v_readfirstlane_b32 s8, v9
 ; GPRIDX-NEXT:    v_readfirstlane_b32 s9, v12
-; GPRIDX-NEXT:    v_readfirstlane_b32 s10, v13
-; GPRIDX-NEXT:    v_readfirstlane_b32 s11, v14
-; GPRIDX-NEXT:    v_readfirstlane_b32 s12, v15
-; GPRIDX-NEXT:    v_readfirstlane_b32 s13, v16
+; GPRIDX-NEXT:    v_readfirstlane_b32 s10, v11
+; GPRIDX-NEXT:    v_readfirstlane_b32 s11, v13
+; GPRIDX-NEXT:    v_readfirstlane_b32 s12, v0
+; GPRIDX-NEXT:    v_readfirstlane_b32 s13, v1
 ; GPRIDX-NEXT:    ; return to shader part epilog
 ;
 ; MOVREL-LABEL: dyn_insertelement_v7f64_s_v_v:
 ; MOVREL:       ; %bb.0: ; %entry
 ; MOVREL-NEXT:    s_mov_b32 s0, s2
 ; MOVREL-NEXT:    s_mov_b32 s1, s3
 ; MOVREL-NEXT:    s_mov_b32 s2, s4
 ; MOVREL-NEXT:    s_mov_b32 s3, s5
 ; MOVREL-NEXT:    s_mov_b32 s4, s6
 ; MOVREL-NEXT:    s_mov_b32 s5, s7
 ; MOVREL-NEXT:    s_mov_b32 s6, s8
 ; MOVREL-NEXT:    s_mov_b32 s7, s9
 ; MOVREL-NEXT:    s_mov_b32 s8, s10
 ; MOVREL-NEXT:    s_mov_b32 s9, s11
 ; MOVREL-NEXT:    s_mov_b32 s10, s12
 ; MOVREL-NEXT:    s_mov_b32 s11, s13
 ; MOVREL-NEXT:    s_mov_b32 s12, s14
 ; MOVREL-NEXT:    s_mov_b32 s13, s15
-; MOVREL-NEXT:    v_mov_b32_e32 v34, s15
-; MOVREL-NEXT:    v_mov_b32_e32 v33, s14
-; MOVREL-NEXT:    v_mov_b32_e32 v30, s11
-; MOVREL-NEXT:    v_mov_b32_e32 v31, s12
-; MOVREL-NEXT:    v_mov_b32_e32 v32, s13
-; MOVREL-NEXT:    v_mov_b32_e32 v29, s10
-; MOVREL-NEXT:    v_mov_b32_e32 v28, s9
-; MOVREL-NEXT:    v_mov_b32_e32 v27, s8
-; MOVREL-NEXT:    v_mov_b32_e32 v26, s7
-; MOVREL-NEXT:    v_mov_b32_e32 v25, s6
-; MOVREL-NEXT:    v_mov_b32_e32 v24, s5
-; MOVREL-NEXT:    v_mov_b32_e32 v23, s4
-; MOVREL-NEXT:    v_mov_b32_e32 v22, s3
-; MOVREL-NEXT:    v_mov_b32_e32 v21, s2
-; MOVREL-NEXT:    v_mov_b32_e32 v20, s1
-; MOVREL-NEXT:    v_mov_b32_e32 v19, s0
-; MOVREL-NEXT:    s_mov_b32 s0, exec_lo
-; MOVREL-NEXT:    ; implicit-def: $vcc_hi
-; MOVREL-NEXT:  BB51_1: ; =>This Inner Loop Header: Depth=1
-; MOVREL-NEXT:    v_readfirstlane_b32 s1, v2
-; MOVREL-NEXT:    v_mov_b32_e32 v3, v19
-; MOVREL-NEXT:    v_mov_b32_e32 v4, v20
-; MOVREL-NEXT:    v_mov_b32_e32 v5, v21
-; MOVREL-NEXT:    v_mov_b32_e32 v6, v22
-; MOVREL-NEXT:    s_lshl_b32 m0, s1, 1
-; MOVREL-NEXT:    v_cmp_eq_u32_e32 vcc_lo, s1, v2
-; MOVREL-NEXT:    v_mov_b32_e32 v7, v23
-; MOVREL-NEXT:    v_mov_b32_e32 v8, v24
-; MOVREL-NEXT:    v_mov_b32_e32 v9, v25
-; MOVREL-NEXT:    v_mov_b32_e32 v10, v26
-; MOVREL-NEXT:    v_mov_b32_e32 v11, v27
-; MOVREL-NEXT:    v_mov_b32_e32 v12, v28
-; MOVREL-NEXT:    v_mov_b32_e32 v13, v29
-; MOVREL-NEXT:    v_mov_b32_e32 v14, v30
-; MOVREL-NEXT:    v_mov_b32_e32 v15, v31
-; MOVREL-NEXT:    v_mov_b32_e32 v16, v32
-; MOVREL-NEXT:    v_mov_b32_e32 v17, v33
-; MOVREL-NEXT:    v_mov_b32_e32 v18, v34
-; MOVREL-NEXT:    v_movreld_b32_e32 v3, v0
-; MOVREL-NEXT:    v_movreld_b32_e32 v4, v1
-; MOVREL-NEXT:    s_and_saveexec_b32 vcc_lo, vcc_lo
-; MOVREL-NEXT:    s_xor_b32 exec_lo, exec_lo, vcc_lo
-; MOVREL-NEXT:    s_cbranch_execnz BB51_1
-; MOVREL-NEXT:  ; %bb.2:
-; MOVREL-NEXT:    s_mov_b32 exec_lo, s0
-; MOVREL-NEXT:    v_readfirstlane_b32 s0, v3
-; MOVREL-NEXT:    v_readfirstlane_b32 s1, v4
+; MOVREL-NEXT:    v_mov_b32_e32 v18, s15
+; MOVREL-NEXT:    v_mov_b32_e32 v17, s14
+; MOVREL-NEXT:    v_mov_b32_e32 v14, s11
+; MOVREL-NEXT:    v_mov_b32_e32 v15, s12
+; MOVREL-NEXT:    v_mov_b32_e32 v16, s13
+; MOVREL-NEXT:    v_mov_b32_e32 v13, s10
+; MOVREL-NEXT:    v_mov_b32_e32 v12, s9
+; MOVREL-NEXT:    v_mov_b32_e32 v11, s8
+; MOVREL-NEXT:    v_mov_b32_e32 v10, s7
+; MOVREL-NEXT:    v_mov_b32_e32 v9, s6
+; MOVREL-NEXT:    v_mov_b32_e32 v8, s5
+; MOVREL-NEXT:    v_mov_b32_e32 v7, s4
+; MOVREL-NEXT:    v_mov_b32_e32 v6, s3
+; MOVREL-NEXT:    v_mov_b32_e32 v5, s2
+; MOVREL-NEXT:    v_mov_b32_e32 v4, s1
+; MOVREL-NEXT:    v_mov_b32_e32 v3, s0
+; MOVREL-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 0, v2
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 s0, 1, v2
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 s1, 6, v2
+; MOVREL-NEXT:    ; implicit-def: $vcc_hi
+; MOVREL-NEXT:    v_cndmask_b32_e32 v3, v3, v0, vcc_lo
+; MOVREL-NEXT:    v_cndmask_b32_e32 v4, v4, v1, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 2, v2
+; MOVREL-NEXT:    v_cndmask_b32_e64 v5, v5, v0, s0
+; MOVREL-NEXT:    v_cndmask_b32_e64 v6, v6, v1, s0
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 s0, 3, v2
+; MOVREL-NEXT:    v_cndmask_b32_e32 v7, v7, v0, vcc_lo
+; MOVREL-NEXT:    v_cndmask_b32_e32 v8, v8, v1, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 4, v2
+; MOVREL-NEXT:    v_cndmask_b32_e64 v9, v9, v0, s0
+; MOVREL-NEXT:    v_cndmask_b32_e64 v10, v10, v1, s0
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 s0, 5, v2
 ; MOVREL-NEXT:    v_readfirstlane_b32 s2, v5
+; MOVREL-NEXT:    v_cndmask_b32_e32 v2, v12, v1, vcc_lo
+; MOVREL-NEXT:    v_cndmask_b32_e32 v11, v11, v0, vcc_lo
 ; MOVREL-NEXT:    v_readfirstlane_b32 s3, v6
+; MOVREL-NEXT:    v_cndmask_b32_e64 v12, v13, v0, s0
+; MOVREL-NEXT:    v_cndmask_b32_e64 v13, v14, v1, s0
+; MOVREL-NEXT:    v_cndmask_b32_e64 v0, v15, v0, s1
+; MOVREL-NEXT:    v_cndmask_b32_e64 v1, v16, v1, s1
+; MOVREL-NEXT:    v_readfirstlane_b32 s0, v3
+; MOVREL-NEXT:    v_readfirstlane_b32 s1, v4
 ; MOVREL-NEXT:    v_readfirstlane_b32 s4, v7
 ; MOVREL-NEXT:    v_readfirstlane_b32 s5, v8
 ; MOVREL-NEXT:    v_readfirstlane_b32 s6, v9
 ; MOVREL-NEXT:    v_readfirstlane_b32 s7, v10
 ; MOVREL-NEXT:    v_readfirstlane_b32 s8, v11
-; MOVREL-NEXT:    v_readfirstlane_b32 s9, v12
-; MOVREL-NEXT:    v_readfirstlane_b32 s10, v13
-; MOVREL-NEXT:    v_readfirstlane_b32 s11, v14
-; MOVREL-NEXT:    v_readfirstlane_b32 s12, v15
-; MOVREL-NEXT:    v_readfirstlane_b32 s13, v16
+; MOVREL-NEXT:    v_readfirstlane_b32 s9, v2
+; MOVREL-NEXT:    v_readfirstlane_b32 s10, v12
+; MOVREL-NEXT:    v_readfirstlane_b32 s11, v13
+; MOVREL-NEXT:    v_readfirstlane_b32 s12, v0
+; MOVREL-NEXT:    v_readfirstlane_b32 s13, v1
 ; MOVREL-NEXT:    ; return to shader part epilog
 entry:
   %insert = insertelement <7 x double> %vec, double %val, i32 %idx
   ret <7 x double> %insert
 }
 
 define amdgpu_ps <7 x double> @dyn_insertelement_v7f64_v_v_s(<7 x double> %vec, double %val, i32 inreg %idx) {
 ; GPRIDX-LABEL: dyn_insertelement_v7f64_v_v_s:

 entry:
   %insert = insertelement <7 x double> %vec, double %val, i32 %idx
   ret <7 x double> %insert
 }
 
 define amdgpu_ps <7 x double> @dyn_insertelement_v7f64_v_v_v(<7 x double> %vec, double %val, i32 %idx) {
 ; GPRIDX-LABEL: dyn_insertelement_v7f64_v_v_v:
 ; GPRIDX:       ; %bb.0: ; %entry
-; GPRIDX-NEXT:    s_mov_b64 s[0:1], exec
-; GPRIDX-NEXT:  BB53_1: ; =>This Inner Loop Header: Depth=1
-; GPRIDX-NEXT:    v_readfirstlane_b32 s2, v16
-; GPRIDX-NEXT:    s_lshl_b32 s3, s2, 1
-; GPRIDX-NEXT:    v_cmp_eq_u32_e32 vcc, s2, v16
-; GPRIDX-NEXT:    s_set_gpr_idx_on s3, gpr_idx(DST)
-; GPRIDX-NEXT:    v_mov_b32_e32 v32, v15
-; GPRIDX-NEXT:    v_mov_b32_e32 v31, v14
-; GPRIDX-NEXT:    v_mov_b32_e32 v30, v13
-; GPRIDX-NEXT:    v_mov_b32_e32 v29, v12
-; GPRIDX-NEXT:    v_mov_b32_e32 v28, v11
-; GPRIDX-NEXT:    v_mov_b32_e32 v27, v10
-; GPRIDX-NEXT:    v_mov_b32_e32 v26, v9
-; GPRIDX-NEXT:    v_mov_b32_e32 v25, v8
-; GPRIDX-NEXT:    v_mov_b32_e32 v24, v7
-; GPRIDX-NEXT:    v_mov_b32_e32 v23, v6
-; GPRIDX-NEXT:    v_mov_b32_e32 v22, v5
-; GPRIDX-NEXT:    v_mov_b32_e32 v21, v4
-; GPRIDX-NEXT:    v_mov_b32_e32 v20, v3
-; GPRIDX-NEXT:    v_mov_b32_e32 v19, v2
-; GPRIDX-NEXT:    v_mov_b32_e32 v18, v1
-; GPRIDX-NEXT:    v_mov_b32_e32 v17, v0
-; GPRIDX-NEXT:    v_mov_b32_e32 v17, v14
-; GPRIDX-NEXT:    s_set_gpr_idx_off
-; GPRIDX-NEXT:    s_set_gpr_idx_on s3, gpr_idx(DST)
-; GPRIDX-NEXT:    v_mov_b32_e32 v18, v15
-; GPRIDX-NEXT:    s_set_gpr_idx_off
-; GPRIDX-NEXT:    s_and_saveexec_b64 vcc, vcc
-; GPRIDX-NEXT:    s_xor_b64 exec, exec, vcc
-; GPRIDX-NEXT:    s_cbranch_execnz BB53_1
-; GPRIDX-NEXT:  ; %bb.2:
-; GPRIDX-NEXT:    s_mov_b64 exec, s[0:1]
-; GPRIDX-NEXT:    v_readfirstlane_b32 s0, v17
-; GPRIDX-NEXT:    v_readfirstlane_b32 s1, v18
-; GPRIDX-NEXT:    v_readfirstlane_b32 s2, v19
-; GPRIDX-NEXT:    v_readfirstlane_b32 s3, v20
-; GPRIDX-NEXT:    v_readfirstlane_b32 s4, v21
-; GPRIDX-NEXT:    v_readfirstlane_b32 s5, v22
-; GPRIDX-NEXT:    v_readfirstlane_b32 s6, v23
-; GPRIDX-NEXT:    v_readfirstlane_b32 s7, v24
-; GPRIDX-NEXT:    v_readfirstlane_b32 s8, v25
-; GPRIDX-NEXT:    v_readfirstlane_b32 s9, v26
-; GPRIDX-NEXT:    v_readfirstlane_b32 s10, v27
-; GPRIDX-NEXT:    v_readfirstlane_b32 s11, v28
-; GPRIDX-NEXT:    v_readfirstlane_b32 s12, v29
-; GPRIDX-NEXT:    v_readfirstlane_b32 s13, v30
+; GPRIDX-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v16
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 s[0:1], 1, v16
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 s[2:3], 2, v16
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 s[4:5], 3, v16
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 s[6:7], 4, v16
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 s[8:9], 5, v16
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 s[10:11], 6, v16
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v12, v12, v14, s[10:11]
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v13, v13, v15, s[10:11]
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v10, v10, v14, s[8:9]
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v11, v11, v15, s[8:9]
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v8, v8, v14, s[6:7]
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v9, v9, v15, s[6:7]
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v6, v6, v14, s[4:5]
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v7, v7, v15, s[4:5]
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v4, v4, v14, s[2:3]
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v5, v5, v15, s[2:3]
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v2, v2, v14, s[0:1]
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v3, v3, v15, s[0:1]
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v0, v0, v14, vcc
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v1, v1, v15, vcc
+; GPRIDX-NEXT:    v_readfirstlane_b32 s0, v0
+; GPRIDX-NEXT:    v_readfirstlane_b32 s1, v1
+; GPRIDX-NEXT:    v_readfirstlane_b32 s2, v2
+; GPRIDX-NEXT:    v_readfirstlane_b32 s3, v3
+; GPRIDX-NEXT:    v_readfirstlane_b32 s4, v4
+; GPRIDX-NEXT:    v_readfirstlane_b32 s5, v5
+; GPRIDX-NEXT:    v_readfirstlane_b32 s6, v6
+; GPRIDX-NEXT:    v_readfirstlane_b32 s7, v7
+; GPRIDX-NEXT:    v_readfirstlane_b32 s8, v8
+; GPRIDX-NEXT:    v_readfirstlane_b32 s9, v9
+; GPRIDX-NEXT:    v_readfirstlane_b32 s10, v10
+; GPRIDX-NEXT:    v_readfirstlane_b32 s11, v11
+; GPRIDX-NEXT:    v_readfirstlane_b32 s12, v12
+; GPRIDX-NEXT:    v_readfirstlane_b32 s13, v13
 ; GPRIDX-NEXT:    ; return to shader part epilog
 ;
 ; MOVREL-LABEL: dyn_insertelement_v7f64_v_v_v:
 ; MOVREL:       ; %bb.0: ; %entry
-; MOVREL-NEXT:    s_mov_b32 s0, exec_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 0, v16
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 s1, 2, v16
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 s2, 3, v16
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 s3, 4, v16
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 s4, 5, v16
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 s5, 6, v16
+; MOVREL-NEXT:    v_mov_b32_e32 v17, v2
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 s0, 1, v16
+; MOVREL-NEXT:    v_mov_b32_e32 v18, v3
+; MOVREL-NEXT:    v_cndmask_b32_e64 v6, v6, v14, s2
+; MOVREL-NEXT:    v_cndmask_b32_e64 v8, v8, v14, s3
+; MOVREL-NEXT:    v_cndmask_b32_e64 v10, v10, v14, s4
+; MOVREL-NEXT:    v_cndmask_b32_e64 v12, v12, v14, s5
+; MOVREL-NEXT:    v_cndmask_b32_e64 v7, v7, v15, s2
+; MOVREL-NEXT:    v_cndmask_b32_e64 v2, v17, v14, s0
+; MOVREL-NEXT:    v_cndmask_b32_e64 v9, v9, v15, s3
+; MOVREL-NEXT:    v_cndmask_b32_e64 v3, v18, v15, s0
+; MOVREL-NEXT:    v_cndmask_b32_e32 v0, v0, v14, vcc_lo
+; MOVREL-NEXT:    v_cndmask_b32_e64 v11, v11, v15, s4
+; MOVREL-NEXT:    v_cndmask_b32_e64 v4, v4, v14, s1
+; MOVREL-NEXT:    v_cndmask_b32_e64 v13, v13, v15, s5
+; MOVREL-NEXT:    v_cndmask_b32_e64 v5, v5, v15, s1
+; MOVREL-NEXT:    v_cndmask_b32_e32 v1, v1, v15, vcc_lo
+; MOVREL-NEXT:    v_readfirstlane_b32 s0, v0
+; MOVREL-NEXT:    v_readfirstlane_b32 s2, v2
+; MOVREL-NEXT:    v_readfirstlane_b32 s3, v3
+; MOVREL-NEXT:    v_readfirstlane_b32 s4, v4
+; MOVREL-NEXT:    v_readfirstlane_b32 s1, v1
+; MOVREL-NEXT:    v_readfirstlane_b32 s5, v5
+; MOVREL-NEXT:    v_readfirstlane_b32 s6, v6
+; MOVREL-NEXT:    v_readfirstlane_b32 s7, v7
+; MOVREL-NEXT:    v_readfirstlane_b32 s8, v8
+; MOVREL-NEXT:    v_readfirstlane_b32 s9, v9
+; MOVREL-NEXT:    v_readfirstlane_b32 s10, v10
+; MOVREL-NEXT:    v_readfirstlane_b32 s11, v11
+; MOVREL-NEXT:    v_readfirstlane_b32 s12, v12
+; MOVREL-NEXT:    v_readfirstlane_b32 s13, v13
 ; MOVREL-NEXT:    ; implicit-def: $vcc_hi
-; MOVREL-NEXT:  BB53_1: ; =>This Inner Loop Header: Depth=1
-; MOVREL-NEXT:    v_readfirstlane_b32 s1, v16
-; MOVREL-NEXT:    v_mov_b32_e32 v32, v15
-; MOVREL-NEXT:    v_mov_b32_e32 v17, v0
-; MOVREL-NEXT:    v_mov_b32_e32 v31, v14
-; MOVREL-NEXT:    v_mov_b32_e32 v30, v13
-; MOVREL-NEXT:    s_lshl_b32 m0, s1, 1
-; MOVREL-NEXT:    v_cmp_eq_u32_e32 vcc_lo, s1, v16
-; MOVREL-NEXT:    v_mov_b32_e32 v29, v12
-; MOVREL-NEXT:    v_mov_b32_e32 v28, v11
-; MOVREL-NEXT:    v_mov_b32_e32 v27, v10
-; MOVREL-NEXT:    v_mov_b32_e32 v26, v9
-; MOVREL-NEXT:    v_mov_b32_e32 v25, v8
-; MOVREL-NEXT:    v_mov_b32_e32 v24, v7
-; MOVREL-NEXT:    v_mov_b32_e32 v23, v6
-; MOVREL-NEXT:    v_mov_b32_e32 v22, v5
-; MOVREL-NEXT:    v_mov_b32_e32 v21, v4
-; MOVREL-NEXT:    v_mov_b32_e32 v20, v3
-; MOVREL-NEXT:    v_mov_b32_e32 v19, v2
-; MOVREL-NEXT:    v_mov_b32_e32 v18, v1
-; MOVREL-NEXT:    v_movreld_b32_e32 v17, v14
-; MOVREL-NEXT:    v_movreld_b32_e32 v18, v15
-; MOVREL-NEXT:    s_and_saveexec_b32 vcc_lo, vcc_lo
-; MOVREL-NEXT:    s_xor_b32 exec_lo, exec_lo, vcc_lo
-; MOVREL-NEXT:    s_cbranch_execnz BB53_1
-; MOVREL-NEXT:  ; %bb.2:
-; MOVREL-NEXT:    s_mov_b32 exec_lo, s0
-; MOVREL-NEXT:    v_readfirstlane_b32 s0, v17
-; MOVREL-NEXT:    v_readfirstlane_b32 s1, v18
-; MOVREL-NEXT:    v_readfirstlane_b32 s2, v19
-; MOVREL-NEXT:    v_readfirstlane_b32 s3, v20
-; MOVREL-NEXT:    v_readfirstlane_b32 s4, v21
-; MOVREL-NEXT:    v_readfirstlane_b32 s5, v22
-; MOVREL-NEXT:    v_readfirstlane_b32 s6, v23
-; MOVREL-NEXT:    v_readfirstlane_b32 s7, v24
-; MOVREL-NEXT:    v_readfirstlane_b32 s8, v25
-; MOVREL-NEXT:    v_readfirstlane_b32 s9, v26
-; MOVREL-NEXT:    v_readfirstlane_b32 s10, v27
-; MOVREL-NEXT:    v_readfirstlane_b32 s11, v28
-; MOVREL-NEXT:    v_readfirstlane_b32 s12, v29
-; MOVREL-NEXT:    v_readfirstlane_b32 s13, v30
 ; MOVREL-NEXT:    ; return to shader part epilog
 entry:
   %insert = insertelement <7 x double> %vec, double %val, i32 %idx
   ret <7 x double> %insert
 }
 
 define amdgpu_ps <5 x double> @dyn_insertelement_v5f64_s_s_s(<5 x double> inreg %vec, double inreg %val, i32 inreg %idx) {
 ; GPRIDX-LABEL: dyn_insertelement_v5f64_s_s_s:
 ; GPRIDX:       ; %bb.0: ; %entry
-; GPRIDX-NEXT:    s_mov_b32 s0, s2
-; GPRIDX-NEXT:    s_mov_b32 s1, s3
-; GPRIDX-NEXT:    s_mov_b32 s2, s4
-; GPRIDX-NEXT:    s_mov_b32 s3, s5
-; GPRIDX-NEXT:    s_mov_b32 s4, s6
-; GPRIDX-NEXT:    s_mov_b32 s5, s7
-; GPRIDX-NEXT:    s_mov_b32 s6, s8
-; GPRIDX-NEXT:    s_mov_b32 s7, s9
-; GPRIDX-NEXT:    s_mov_b32 s8, s10
-; GPRIDX-NEXT:    s_mov_b32 s9, s11
-; GPRIDX-NEXT:    s_mov_b32 m0, s14
-; GPRIDX-NEXT:    s_nop 0
-; GPRIDX-NEXT:    s_movreld_b64 s[0:1], s[12:13]
+; GPRIDX-NEXT:    s_cmp_eq_u32 s14, 0
+; GPRIDX-NEXT:    s_cselect_b64 s[0:1], s[12:13], s[2:3]
+; GPRIDX-NEXT:    s_cmp_eq_u32 s14, 1
+; GPRIDX-NEXT:    s_cselect_b64 s[2:3], s[12:13], s[4:5]
+; GPRIDX-NEXT:    s_cmp_eq_u32 s14, 2
+; GPRIDX-NEXT:    s_cselect_b64 s[4:5], s[12:13], s[6:7]
+; GPRIDX-NEXT:    s_cmp_eq_u32 s14, 3
+; GPRIDX-NEXT:    s_cselect_b64 s[6:7], s[12:13], s[8:9]
+; GPRIDX-NEXT:    s_cmp_eq_u32 s14, 4
+; GPRIDX-NEXT:    s_cselect_b64 s[8:9], s[12:13], s[10:11]
 ; GPRIDX-NEXT:    ; return to shader part epilog
 ;
 ; MOVREL-LABEL: dyn_insertelement_v5f64_s_s_s:
 ; MOVREL:       ; %bb.0: ; %entry
-; MOVREL-NEXT:    s_mov_b32 s0, s2
-; MOVREL-NEXT:    s_mov_b32 s1, s3
-; MOVREL-NEXT:    s_mov_b32 m0, s14
-; MOVREL-NEXT:    s_mov_b32 s2, s4
-; MOVREL-NEXT:    s_mov_b32 s3, s5
-; MOVREL-NEXT:    s_mov_b32 s4, s6
-; MOVREL-NEXT:    s_mov_b32 s5, s7
-; MOVREL-NEXT:    s_mov_b32 s6, s8
-; MOVREL-NEXT:    s_mov_b32 s7, s9
-; MOVREL-NEXT:    s_mov_b32 s8, s10
-; MOVREL-NEXT:    s_mov_b32 s9, s11
-; MOVREL-NEXT:    s_movreld_b64 s[0:1], s[12:13]
-; MOVREL-NEXT:    ; implicit-def: $vcc_hi
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 s0, s14, 0
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 s15, s14, 1
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 s16, s14, 2
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 s17, s14, 3
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 s14, s14, 4
+; MOVREL-NEXT:    s_cmp_lg_u32 s0, 0
+; MOVREL-NEXT:    ; implicit-def: $vcc_hi
+; MOVREL-NEXT:    s_cselect_b64 s[0:1], s[12:13], s[2:3]
+; MOVREL-NEXT:    s_cmp_lg_u32 s15, 0
+; MOVREL-NEXT:    s_cselect_b64 s[2:3], s[12:13], s[4:5]
+; MOVREL-NEXT:    s_cmp_lg_u32 s16, 0
+; MOVREL-NEXT:    s_cselect_b64 s[4:5], s[12:13], s[6:7]
+; MOVREL-NEXT:    s_cmp_lg_u32 s17, 0
+; MOVREL-NEXT:    s_cselect_b64 s[6:7], s[12:13], s[8:9]
+; MOVREL-NEXT:    s_cmp_lg_u32 s14, 0
+; MOVREL-NEXT:    s_cselect_b64 s[8:9], s[12:13], s[10:11]
 ; MOVREL-NEXT:    ; return to shader part epilog
 entry:
   %insert = insertelement <5 x double> %vec, double %val, i32 %idx
   ret <5 x double> %insert
 }
 
 define amdgpu_ps <5 x double> @dyn_insertelement_v5f64_s_v_s(<5 x double> inreg %vec, double %val, i32 inreg %idx) {
 ; GPRIDX-LABEL: dyn_insertelement_v5f64_s_v_s:

 ; GPRIDX-NEXT:    v_mov_b32_e32 v9, s7
 ; GPRIDX-NEXT:    v_mov_b32_e32 v8, s6
 ; GPRIDX-NEXT:    v_mov_b32_e32 v7, s5
 ; GPRIDX-NEXT:    v_mov_b32_e32 v6, s4
 ; GPRIDX-NEXT:    v_mov_b32_e32 v5, s3
 ; GPRIDX-NEXT:    v_mov_b32_e32 v4, s2
 ; GPRIDX-NEXT:    v_mov_b32_e32 v3, s1
 ; GPRIDX-NEXT:    v_mov_b32_e32 v2, s0
-; GPRIDX-NEXT:    s_lshl_b32 s0, s12, 1
-; GPRIDX-NEXT:    s_set_gpr_idx_on s0, gpr_idx(DST)
-; GPRIDX-NEXT:    v_mov_b32_e32 v2, v0
-; GPRIDX-NEXT:    v_mov_b32_e32 v3, v1
-; GPRIDX-NEXT:    s_set_gpr_idx_off
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 vcc, s12, 0
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 s[0:1], s12, 1
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 s[2:3], s12, 3
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 s[6:7], s12, 2
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 s[4:5], s12, 4
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v2, v2, v0, vcc
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v4, v4, v0, s[0:1]
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v5, v5, v1, s[0:1]
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v3, v3, v1, vcc
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v6, v6, v0, s[6:7]
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v8, v8, v0, s[2:3]
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v9, v9, v1, s[2:3]
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v7, v7, v1, s[6:7]
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v0, v10, v0, s[4:5]
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v1, v11, v1, s[4:5]
 ; GPRIDX-NEXT:    v_readfirstlane_b32 s0, v2
 ; GPRIDX-NEXT:    v_readfirstlane_b32 s1, v3
 ; GPRIDX-NEXT:    v_readfirstlane_b32 s2, v4
 ; GPRIDX-NEXT:    v_readfirstlane_b32 s3, v5
 ; GPRIDX-NEXT:    v_readfirstlane_b32 s4, v6
 ; GPRIDX-NEXT:    v_readfirstlane_b32 s5, v7
 ; GPRIDX-NEXT:    v_readfirstlane_b32 s6, v8
 ; GPRIDX-NEXT:    v_readfirstlane_b32 s7, v9
-; GPRIDX-NEXT:    v_readfirstlane_b32 s8, v10
-; GPRIDX-NEXT:    v_readfirstlane_b32 s9, v11
+; GPRIDX-NEXT:    v_readfirstlane_b32 s8, v0
+; GPRIDX-NEXT:    v_readfirstlane_b32 s9, v1
 ; GPRIDX-NEXT:    ; return to shader part epilog
 ;
 ; MOVREL-LABEL: dyn_insertelement_v5f64_s_v_s:
 ; MOVREL:       ; %bb.0: ; %entry
 ; MOVREL-NEXT:    s_mov_b32 s0, s2
 ; MOVREL-NEXT:    s_mov_b32 s1, s3
 ; MOVREL-NEXT:    s_mov_b32 s2, s4
 ; MOVREL-NEXT:    s_mov_b32 s3, s5
 ; MOVREL-NEXT:    s_mov_b32 s4, s6
 ; MOVREL-NEXT:    s_mov_b32 s5, s7
 ; MOVREL-NEXT:    s_mov_b32 s6, s8
 ; MOVREL-NEXT:    s_mov_b32 s7, s9
 ; MOVREL-NEXT:    s_mov_b32 s8, s10
 ; MOVREL-NEXT:    s_mov_b32 s9, s11
-; MOVREL-NEXT:    v_mov_b32_e32 v17, s15
-; MOVREL-NEXT:    v_mov_b32_e32 v2, s0
-; MOVREL-NEXT:    s_lshl_b32 m0, s12, 1
-; MOVREL-NEXT:    v_mov_b32_e32 v16, s14
-; MOVREL-NEXT:    v_mov_b32_e32 v15, s13
-; MOVREL-NEXT:    v_mov_b32_e32 v14, s12
-; MOVREL-NEXT:    v_mov_b32_e32 v13, s11
-; MOVREL-NEXT:    v_mov_b32_e32 v12, s10
-; MOVREL-NEXT:    v_mov_b32_e32 v11, s9
-; MOVREL-NEXT:    v_mov_b32_e32 v10, s8
-; MOVREL-NEXT:    v_mov_b32_e32 v9, s7
-; MOVREL-NEXT:    v_mov_b32_e32 v8, s6
-; MOVREL-NEXT:    v_mov_b32_e32 v7, s5
-; MOVREL-NEXT:    v_mov_b32_e32 v6, s4
-; MOVREL-NEXT:    v_mov_b32_e32 v5, s3
-; MOVREL-NEXT:    v_mov_b32_e32 v4, s2
-; MOVREL-NEXT:    v_mov_b32_e32 v3, s1
-; MOVREL-NEXT:    v_movreld_b32_e32 v2, v0
-; MOVREL-NEXT:    v_movreld_b32_e32 v3, v1
-; MOVREL-NEXT:    v_readfirstlane_b32 s0, v2
-; MOVREL-NEXT:    v_readfirstlane_b32 s1, v3
+; MOVREL-NEXT:    v_mov_b32_e32 v18, s15
+; MOVREL-NEXT:    v_mov_b32_e32 v17, s14
+; MOVREL-NEXT:    v_mov_b32_e32 v16, s13
+; MOVREL-NEXT:    v_mov_b32_e32 v15, s12
+; MOVREL-NEXT:    v_mov_b32_e32 v14, s11
+; MOVREL-NEXT:    v_mov_b32_e32 v13, s10
+; MOVREL-NEXT:    v_mov_b32_e32 v12, s9
+; MOVREL-NEXT:    v_mov_b32_e32 v11, s8
+; MOVREL-NEXT:    v_mov_b32_e32 v10, s7
+; MOVREL-NEXT:    v_mov_b32_e32 v9, s6
+; MOVREL-NEXT:    v_mov_b32_e32 v8, s5
+; MOVREL-NEXT:    v_mov_b32_e32 v7, s4
+; MOVREL-NEXT:    v_mov_b32_e32 v6, s3
+; MOVREL-NEXT:    v_mov_b32_e32 v5, s2
+; MOVREL-NEXT:    v_mov_b32_e32 v4, s1
+; MOVREL-NEXT:    v_mov_b32_e32 v3, s0
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 vcc_lo, s12, 0
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 s0, s12, 1
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 s1, s12, 4
+; MOVREL-NEXT:    ; implicit-def: $vcc_hi
+; MOVREL-NEXT:    v_cndmask_b32_e32 v2, v3, v0, vcc_lo
+; MOVREL-NEXT:    v_cndmask_b32_e32 v3, v4, v1, vcc_lo
+; MOVREL-NEXT:    v_cndmask_b32_e64 v4, v5, v0, s0
+; MOVREL-NEXT:    v_cndmask_b32_e64 v5, v6, v1, s0
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 vcc_lo, s12, 2
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 s0, s12, 3
 ; MOVREL-NEXT:    v_readfirstlane_b32 s2, v4
 ; MOVREL-NEXT:    v_readfirstlane_b32 s3, v5
+; MOVREL-NEXT:    v_cndmask_b32_e32 v6, v7, v0, vcc_lo
+; MOVREL-NEXT:    v_cndmask_b32_e32 v7, v8, v1, vcc_lo
+; MOVREL-NEXT:    v_cndmask_b32_e64 v8, v9, v0, s0
+; MOVREL-NEXT:    v_cndmask_b32_e64 v9, v10, v1, s0
+; MOVREL-NEXT:    v_cndmask_b32_e64 v0, v11, v0, s1
+; MOVREL-NEXT:    v_cndmask_b32_e64 v1, v12, v1, s1
+; MOVREL-NEXT:    v_readfirstlane_b32 s0, v2
+; MOVREL-NEXT:    v_readfirstlane_b32 s1, v3
 ; MOVREL-NEXT:    v_readfirstlane_b32 s4, v6
 ; MOVREL-NEXT:    v_readfirstlane_b32 s5, v7
 ; MOVREL-NEXT:    v_readfirstlane_b32 s6, v8
 ; MOVREL-NEXT:    v_readfirstlane_b32 s7, v9
-; MOVREL-NEXT:    v_readfirstlane_b32 s8, v10
-; MOVREL-NEXT:    v_readfirstlane_b32 s9, v11
-; MOVREL-NEXT:    ; implicit-def: $vcc_hi
+; MOVREL-NEXT:    v_readfirstlane_b32 s8, v0
+; MOVREL-NEXT:    v_readfirstlane_b32 s9, v1
 ; MOVREL-NEXT:    ; return to shader part epilog
 entry:
   %insert = insertelement <5 x double> %vec, double %val, i32 %idx
   ret <5 x double> %insert
 }
 
 define amdgpu_ps <5 x double> @dyn_insertelement_v5f64_s_v_v(<5 x double> inreg %vec, double %val, i32 %idx) {
 ; GPRIDX-LABEL: dyn_insertelement_v5f64_s_v_v:
 ; GPRIDX:       ; %bb.0: ; %entry
 ; GPRIDX-NEXT:    s_mov_b32 s0, s2
 ; GPRIDX-NEXT:    s_mov_b32 s1, s3
 ; GPRIDX-NEXT:    s_mov_b32 s2, s4
 ; GPRIDX-NEXT:    s_mov_b32 s3, s5
 ; GPRIDX-NEXT:    s_mov_b32 s4, s6
 ; GPRIDX-NEXT:    s_mov_b32 s5, s7
 ; GPRIDX-NEXT:    s_mov_b32 s6, s8
 ; GPRIDX-NEXT:    s_mov_b32 s7, s9
 ; GPRIDX-NEXT:    s_mov_b32 s8, s10
 ; GPRIDX-NEXT:    s_mov_b32 s9, s11
-; GPRIDX-NEXT:    v_mov_b32_e32 v34, s15
-; GPRIDX-NEXT:    v_mov_b32_e32 v33, s14
-; GPRIDX-NEXT:    v_mov_b32_e32 v32, s13
-; GPRIDX-NEXT:    v_mov_b32_e32 v31, s12
-; GPRIDX-NEXT:    v_mov_b32_e32 v30, s11
-; GPRIDX-NEXT:    v_mov_b32_e32 v29, s10
-; GPRIDX-NEXT:    v_mov_b32_e32 v28, s9
-; GPRIDX-NEXT:    v_mov_b32_e32 v27, s8
-; GPRIDX-NEXT:    v_mov_b32_e32 v26, s7
-; GPRIDX-NEXT:    v_mov_b32_e32 v25, s6
-; GPRIDX-NEXT:    v_mov_b32_e32 v24, s5
-; GPRIDX-NEXT:    v_mov_b32_e32 v23, s4
-; GPRIDX-NEXT:    v_mov_b32_e32 v22, s3
-; GPRIDX-NEXT:    v_mov_b32_e32 v21, s2
-; GPRIDX-NEXT:    v_mov_b32_e32 v20, s1
-; GPRIDX-NEXT:    v_mov_b32_e32 v19, s0
-; GPRIDX-NEXT:    s_mov_b64 s[0:1], exec
-; GPRIDX-NEXT:  BB56_1: ; =>This Inner Loop Header: Depth=1
-; GPRIDX-NEXT:    v_readfirstlane_b32 s2, v2
-; GPRIDX-NEXT:    s_lshl_b32 s3, s2, 1
-; GPRIDX-NEXT:    v_cmp_eq_u32_e32 vcc, s2, v2
-; GPRIDX-NEXT:    s_set_gpr_idx_on s3, gpr_idx(DST)
-; GPRIDX-NEXT:    v_mov_b32_e32 v3, v19
-; GPRIDX-NEXT:    v_mov_b32_e32 v4, v20
-; GPRIDX-NEXT:    v_mov_b32_e32 v5, v21
-; GPRIDX-NEXT:    v_mov_b32_e32 v6, v22
-; GPRIDX-NEXT:    v_mov_b32_e32 v7, v23
-; GPRIDX-NEXT:    v_mov_b32_e32 v8, v24
-; GPRIDX-NEXT:    v_mov_b32_e32 v9, v25
-; GPRIDX-NEXT:    v_mov_b32_e32 v10, v26
-; GPRIDX-NEXT:    v_mov_b32_e32 v11, v27
-; GPRIDX-NEXT:    v_mov_b32_e32 v12, v28
-; GPRIDX-NEXT:    v_mov_b32_e32 v13, v29
-; GPRIDX-NEXT:    v_mov_b32_e32 v14, v30
-; GPRIDX-NEXT:    v_mov_b32_e32 v15, v31
-; GPRIDX-NEXT:    v_mov_b32_e32 v16, v32
-; GPRIDX-NEXT:    v_mov_b32_e32 v17, v33
-; GPRIDX-NEXT:    v_mov_b32_e32 v18, v34
-; GPRIDX-NEXT:    v_mov_b32_e32 v3, v0
-; GPRIDX-NEXT:    s_set_gpr_idx_off
-; GPRIDX-NEXT:    s_set_gpr_idx_on s3, gpr_idx(DST)
-; GPRIDX-NEXT:    v_mov_b32_e32 v4, v1
-; GPRIDX-NEXT:    s_set_gpr_idx_off
-; GPRIDX-NEXT:    s_and_saveexec_b64 vcc, vcc
-; GPRIDX-NEXT:    s_xor_b64 exec, exec, vcc
-; GPRIDX-NEXT:    s_cbranch_execnz BB56_1
-; GPRIDX-NEXT:  ; %bb.2:
-; GPRIDX-NEXT:    s_mov_b64 exec, s[0:1]
+; GPRIDX-NEXT:    v_mov_b32_e32 v18, s15
+; GPRIDX-NEXT:    v_mov_b32_e32 v17, s14
+; GPRIDX-NEXT:    v_mov_b32_e32 v16, s13
+; GPRIDX-NEXT:    v_mov_b32_e32 v15, s12
+; GPRIDX-NEXT:    v_mov_b32_e32 v14, s11
+; GPRIDX-NEXT:    v_mov_b32_e32 v13, s10
+; GPRIDX-NEXT:    v_mov_b32_e32 v12, s9
+; GPRIDX-NEXT:    v_mov_b32_e32 v11, s8
+; GPRIDX-NEXT:    v_mov_b32_e32 v10, s7
+; GPRIDX-NEXT:    v_mov_b32_e32 v9, s6
+; GPRIDX-NEXT:    v_mov_b32_e32 v8, s5
+; GPRIDX-NEXT:    v_mov_b32_e32 v7, s4
+; GPRIDX-NEXT:    v_mov_b32_e32 v6, s3
+; GPRIDX-NEXT:    v_mov_b32_e32 v5, s2
+; GPRIDX-NEXT:    v_mov_b32_e32 v4, s1
+; GPRIDX-NEXT:    v_mov_b32_e32 v3, s0
+; GPRIDX-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v2
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 s[0:1], 2, v2
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 s[6:7], 1, v2
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 s[2:3], 3, v2
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 s[4:5], 4, v2
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v2, v5, v0, s[6:7]
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v5, v7, v0, s[0:1]
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v7, v9, v0, s[2:3]
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v3, v3, v0, vcc
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v0, v11, v0, s[4:5]
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v6, v6, v1, s[6:7]
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v9, v10, v1, s[2:3]
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v8, v8, v1, s[0:1]
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v4, v4, v1, vcc
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v1, v12, v1, s[4:5]
 ; GPRIDX-NEXT:    v_readfirstlane_b32 s0, v3
 ; GPRIDX-NEXT:    v_readfirstlane_b32 s1, v4
-; GPRIDX-NEXT:    v_readfirstlane_b32 s2, v5
+; GPRIDX-NEXT:    v_readfirstlane_b32 s2, v2
 ; GPRIDX-NEXT:    v_readfirstlane_b32 s3, v6
-; GPRIDX-NEXT:    v_readfirstlane_b32 s4, v7
+; GPRIDX-NEXT:    v_readfirstlane_b32 s4, v5
 ; GPRIDX-NEXT:    v_readfirstlane_b32 s5, v8
-; GPRIDX-NEXT:    v_readfirstlane_b32 s6, v9
-; GPRIDX-NEXT:    v_readfirstlane_b32 s7, v10
-; GPRIDX-NEXT:    v_readfirstlane_b32 s8, v11
-; GPRIDX-NEXT:    v_readfirstlane_b32 s9, v12
+; GPRIDX-NEXT:    v_readfirstlane_b32 s6, v7
+; GPRIDX-NEXT:    v_readfirstlane_b32 s7, v9
+; GPRIDX-NEXT:    v_readfirstlane_b32 s8, v0
+; GPRIDX-NEXT:    v_readfirstlane_b32 s9, v1
 ; GPRIDX-NEXT:    ; return to shader part epilog
 ;
 ; MOVREL-LABEL: dyn_insertelement_v5f64_s_v_v:
 ; MOVREL:       ; %bb.0: ; %entry
 ; MOVREL-NEXT:    s_mov_b32 s0, s2
 ; MOVREL-NEXT:    s_mov_b32 s1, s3
 ; MOVREL-NEXT:    s_mov_b32 s2, s4
 ; MOVREL-NEXT:    s_mov_b32 s3, s5
 ; MOVREL-NEXT:    s_mov_b32 s4, s6
 ; MOVREL-NEXT:    s_mov_b32 s5, s7
 ; MOVREL-NEXT:    s_mov_b32 s6, s8
 ; MOVREL-NEXT:    s_mov_b32 s7, s9
 ; MOVREL-NEXT:    s_mov_b32 s8, s10
 ; MOVREL-NEXT:    s_mov_b32 s9, s11
-; MOVREL-NEXT:    v_mov_b32_e32 v34, s15
-; MOVREL-NEXT:    v_mov_b32_e32 v33, s14
-; MOVREL-NEXT:    v_mov_b32_e32 v32, s13
-; MOVREL-NEXT:    v_mov_b32_e32 v31, s12
-; MOVREL-NEXT:    v_mov_b32_e32 v30, s11
-; MOVREL-NEXT:    v_mov_b32_e32 v29, s10
-; MOVREL-NEXT:    v_mov_b32_e32 v28, s9
-; MOVREL-NEXT:    v_mov_b32_e32 v27, s8
-; MOVREL-NEXT:    v_mov_b32_e32 v26, s7
-; MOVREL-NEXT:    v_mov_b32_e32 v25, s6
-; MOVREL-NEXT:    v_mov_b32_e32 v24, s5
-; MOVREL-NEXT:    v_mov_b32_e32 v23, s4
-; MOVREL-NEXT:    v_mov_b32_e32 v22, s3
-; MOVREL-NEXT:    v_mov_b32_e32 v21, s2
-; MOVREL-NEXT:    v_mov_b32_e32 v20, s1
-; MOVREL-NEXT:    v_mov_b32_e32 v19, s0
-; MOVREL-NEXT:    s_mov_b32 s0, exec_lo
-; MOVREL-NEXT:    ; implicit-def: $vcc_hi
-; MOVREL-NEXT:  BB56_1: ; =>This Inner Loop Header: Depth=1
-; MOVREL-NEXT:    v_readfirstlane_b32 s1, v2
-; MOVREL-NEXT:    v_mov_b32_e32 v3, v19
-; MOVREL-NEXT:    v_mov_b32_e32 v4, v20
-; MOVREL-NEXT:    v_mov_b32_e32 v5, v21
-; MOVREL-NEXT:    v_mov_b32_e32 v6, v22
-; MOVREL-NEXT:    s_lshl_b32 m0, s1, 1
-; MOVREL-NEXT:    v_cmp_eq_u32_e32 vcc_lo, s1, v2
-; MOVREL-NEXT:    v_mov_b32_e32 v7, v23
-; MOVREL-NEXT:    v_mov_b32_e32 v8, v24
-; MOVREL-NEXT:    v_mov_b32_e32 v9, v25
-; MOVREL-NEXT:    v_mov_b32_e32 v10, v26
-; MOVREL-NEXT:    v_mov_b32_e32 v11, v27
-; MOVREL-NEXT:    v_mov_b32_e32 v12, v28
-; MOVREL-NEXT:    v_mov_b32_e32 v13, v29
-; MOVREL-NEXT:    v_mov_b32_e32 v14, v30
-; MOVREL-NEXT:    v_mov_b32_e32 v15, v31
-; MOVREL-NEXT:    v_mov_b32_e32 v16, v32
-; MOVREL-NEXT:    v_mov_b32_e32 v17, v33
-; MOVREL-NEXT:    v_mov_b32_e32 v18, v34
-; MOVREL-NEXT:    v_movreld_b32_e32 v3, v0
-; MOVREL-NEXT:    v_movreld_b32_e32 v4, v1
-; MOVREL-NEXT:    s_and_saveexec_b32 vcc_lo, vcc_lo
-; MOVREL-NEXT:    s_xor_b32 exec_lo, exec_lo, vcc_lo
-; MOVREL-NEXT:    s_cbranch_execnz BB56_1
-; MOVREL-NEXT:  ; %bb.2:
-; MOVREL-NEXT:    s_mov_b32 exec_lo, s0
-; MOVREL-NEXT:    v_readfirstlane_b32 s0, v3
-; MOVREL-NEXT:    v_readfirstlane_b32 s1, v4
+; MOVREL-NEXT:    v_mov_b32_e32 v18, s15
+; MOVREL-NEXT:    v_mov_b32_e32 v17, s14
+; MOVREL-NEXT:    v_mov_b32_e32 v16, s13
+; MOVREL-NEXT:    v_mov_b32_e32 v15, s12
+; MOVREL-NEXT:    v_mov_b32_e32 v14, s11
+; MOVREL-NEXT:    v_mov_b32_e32 v13, s10
+; MOVREL-NEXT:    v_mov_b32_e32 v12, s9
+; MOVREL-NEXT:    v_mov_b32_e32 v11, s8
+; MOVREL-NEXT:    v_mov_b32_e32 v10, s7
+; MOVREL-NEXT:    v_mov_b32_e32 v9, s6
+; MOVREL-NEXT:    v_mov_b32_e32 v8, s5
+; MOVREL-NEXT:    v_mov_b32_e32 v7, s4
+; MOVREL-NEXT:    v_mov_b32_e32 v6, s3
+; MOVREL-NEXT:    v_mov_b32_e32 v5, s2
+; MOVREL-NEXT:    v_mov_b32_e32 v4, s1
+; MOVREL-NEXT:    v_mov_b32_e32 v3, s0
+; MOVREL-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 0, v2
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 s0, 1, v2
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 s1, 4, v2
+; MOVREL-NEXT:    ; implicit-def: $vcc_hi
+; MOVREL-NEXT:    v_cndmask_b32_e32 v3, v3, v0, vcc_lo
+; MOVREL-NEXT:    v_cndmask_b32_e32 v4, v4, v1, vcc_lo
+; MOVREL-NEXT:    v_cndmask_b32_e64 v5, v5, v0, s0
+; MOVREL-NEXT:    v_cndmask_b32_e64 v6, v6, v1, s0
+; MOVREL-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 2, v2
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 s0, 3, v2
 ; MOVREL-NEXT:    v_readfirstlane_b32 s2, v5
 ; MOVREL-NEXT:    v_readfirstlane_b32 s3, v6
+; MOVREL-NEXT:    v_cndmask_b32_e32 v2, v8, v1, vcc_lo
+; MOVREL-NEXT:    v_cndmask_b32_e64 v8, v9, v0, s0
+; MOVREL-NEXT:    v_cndmask_b32_e64 v9, v10, v1, s0
+; MOVREL-NEXT:    v_cndmask_b32_e32 v7, v7, v0, vcc_lo
+; MOVREL-NEXT:    v_cndmask_b32_e64 v0, v11, v0, s1
+; MOVREL-NEXT:    v_cndmask_b32_e64 v1, v12, v1, s1
+; MOVREL-NEXT:    v_readfirstlane_b32 s0, v3
+; MOVREL-NEXT:    v_readfirstlane_b32 s1, v4
 ; MOVREL-NEXT:    v_readfirstlane_b32 s4, v7
-; MOVREL-NEXT:    v_readfirstlane_b32 s5, v8
-; MOVREL-NEXT:    v_readfirstlane_b32 s6, v9
-; MOVREL-NEXT:    v_readfirstlane_b32 s7, v10
-; MOVREL-NEXT:    v_readfirstlane_b32 s8, v11
-; MOVREL-NEXT:    v_readfirstlane_b32 s9, v12
+; MOVREL-NEXT:    v_readfirstlane_b32 s5, v2
+; MOVREL-NEXT:    v_readfirstlane_b32 s6, v8
+; MOVREL-NEXT:    v_readfirstlane_b32 s7, v9
+; MOVREL-NEXT:    v_readfirstlane_b32 s8, v0
+; MOVREL-NEXT:    v_readfirstlane_b32 s9, v1
 ; MOVREL-NEXT:    ; return to shader part epilog
 entry:
   %insert = insertelement <5 x double> %vec, double %val, i32 %idx
   ret <5 x double> %insert
 }
 
 define amdgpu_ps <5 x double> @dyn_insertelement_v5f64_v_v_s(<5 x double> %vec, double %val, i32 inreg %idx) {
 ; GPRIDX-LABEL: dyn_insertelement_v5f64_v_v_s:
 ; GPRIDX:       ; %bb.0: ; %entry
-; GPRIDX-NEXT:    s_lshl_b32 s0, s2, 1
-; GPRIDX-NEXT:    v_mov_b32_e32 v16, v11
-; GPRIDX-NEXT:    s_set_gpr_idx_on s0, gpr_idx(DST)
-; GPRIDX-NEXT:    v_mov_b32_e32 v0, v10
-; GPRIDX-NEXT:    v_mov_b32_e32 v1, v16
-; GPRIDX-NEXT:    s_set_gpr_idx_off
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 vcc, s2, 0
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 s[0:1], s2, 1
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 s[8:9], s2, 2
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 s[4:5], s2, 3
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 s[6:7], s2, 4
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v2, v2, v10, s[0:1]
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v3, v3, v11, s[0:1]
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v0, v0, v10, vcc
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v1, v1, v11, vcc
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v4, v4, v10, s[8:9]
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v6, v6, v10, s[4:5]
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v7, v7, v11, s[4:5]
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v5, v5, v11, s[8:9]
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v8, v8, v10, s[6:7]
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v9, v9, v11, s[6:7]
 ; GPRIDX-NEXT:    v_readfirstlane_b32 s0, v0
 ; GPRIDX-NEXT:    v_readfirstlane_b32 s1, v1
 ; GPRIDX-NEXT:    v_readfirstlane_b32 s2, v2
 ; GPRIDX-NEXT:    v_readfirstlane_b32 s3, v3
 ; GPRIDX-NEXT:    v_readfirstlane_b32 s4, v4
 ; GPRIDX-NEXT:    v_readfirstlane_b32 s5, v5
 ; GPRIDX-NEXT:    v_readfirstlane_b32 s6, v6
 ; GPRIDX-NEXT:    v_readfirstlane_b32 s7, v7
 ; GPRIDX-NEXT:    v_readfirstlane_b32 s8, v8
 ; GPRIDX-NEXT:    v_readfirstlane_b32 s9, v9
 ; GPRIDX-NEXT:    ; return to shader part epilog
 ;
 ; MOVREL-LABEL: dyn_insertelement_v5f64_v_v_s:
 ; MOVREL:       ; %bb.0: ; %entry
-; MOVREL-NEXT:    s_lshl_b32 m0, s2, 1
-; MOVREL-NEXT:    v_mov_b32_e32 v16, v11
-; MOVREL-NEXT:    ; implicit-def: $vcc_hi
-; MOVREL-NEXT:    v_movreld_b32_e32 v0, v10
-; MOVREL-NEXT:    v_movreld_b32_e32 v1, v16
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 vcc_lo, s2, 0
+; MOVREL-NEXT:    v_mov_b32_e32 v13, v2
+; MOVREL-NEXT:    v_mov_b32_e32 v14, v3
+; MOVREL-NEXT:    ; implicit-def: $vcc_hi
+; MOVREL-NEXT:    v_cndmask_b32_e32 v0, v0, v10, vcc_lo
+; MOVREL-NEXT:    v_cndmask_b32_e32 v1, v1, v11, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 vcc_lo, s2, 1
 ; MOVREL-NEXT:    v_readfirstlane_b32 s0, v0
 ; MOVREL-NEXT:    v_readfirstlane_b32 s1, v1
-; MOVREL-NEXT:    v_readfirstlane_b32 s2, v2
+; MOVREL-NEXT:    v_cndmask_b32_e32 v3, v14, v11, vcc_lo
+; MOVREL-NEXT:    v_cndmask_b32_e32 v2, v13, v10, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 vcc_lo, s2, 2
 ; MOVREL-NEXT:    v_readfirstlane_b32 s3, v3
+; MOVREL-NEXT:    v_cndmask_b32_e32 v4, v4, v10, vcc_lo
+; MOVREL-NEXT:    v_cndmask_b32_e32 v5, v5, v11, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 vcc_lo, s2, 3
 ; MOVREL-NEXT:    v_readfirstlane_b32 s4, v4
 ; MOVREL-NEXT:    v_readfirstlane_b32 s5, v5
+; MOVREL-NEXT:    v_cndmask_b32_e32 v6, v6, v10, vcc_lo
+; MOVREL-NEXT:    v_cndmask_b32_e32 v7, v7, v11, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e64 vcc_lo, s2, 4
+; MOVREL-NEXT:    v_readfirstlane_b32 s2, v2
 ; MOVREL-NEXT:    v_readfirstlane_b32 s6, v6
 ; MOVREL-NEXT:    v_readfirstlane_b32 s7, v7
+; MOVREL-NEXT:    v_cndmask_b32_e32 v8, v8, v10, vcc_lo
+; MOVREL-NEXT:    v_cndmask_b32_e32 v9, v9, v11, vcc_lo
 ; MOVREL-NEXT:    v_readfirstlane_b32 s8, v8
 ; MOVREL-NEXT:    v_readfirstlane_b32 s9, v9
 ; MOVREL-NEXT:    ; return to shader part epilog
 entry:
   %insert = insertelement <5 x double> %vec, double %val, i32 %idx
   ret <5 x double> %insert
 }
 
 define amdgpu_ps <5 x double> @dyn_insertelement_v5f64_v_v_v(<5 x double> %vec, double %val, i32 %idx) {
 ; GPRIDX-LABEL: dyn_insertelement_v5f64_v_v_v:
 ; GPRIDX:       ; %bb.0: ; %entry
-; GPRIDX-NEXT:    s_mov_b64 s[0:1], exec
-; GPRIDX-NEXT:  BB58_1: ; =>This Inner Loop Header: Depth=1
-; GPRIDX-NEXT:    v_readfirstlane_b32 s2, v12
-; GPRIDX-NEXT:    s_lshl_b32 s3, s2, 1
-; GPRIDX-NEXT:    v_cmp_eq_u32_e32 vcc, s2, v12
-; GPRIDX-NEXT:    s_set_gpr_idx_on s3, gpr_idx(DST)
-; GPRIDX-NEXT:    v_mov_b32_e32 v28, v15
-; GPRIDX-NEXT:    v_mov_b32_e32 v27, v14
-; GPRIDX-NEXT:    v_mov_b32_e32 v26, v13
-; GPRIDX-NEXT:    v_mov_b32_e32 v25, v12
-; GPRIDX-NEXT:    v_mov_b32_e32 v24, v11
-; GPRIDX-NEXT:    v_mov_b32_e32 v23, v10
-; GPRIDX-NEXT:    v_mov_b32_e32 v22, v9
-; GPRIDX-NEXT:    v_mov_b32_e32 v21, v8
-; GPRIDX-NEXT:    v_mov_b32_e32 v20, v7
-; GPRIDX-NEXT:    v_mov_b32_e32 v19, v6
-; GPRIDX-NEXT:    v_mov_b32_e32 v18, v5
-; GPRIDX-NEXT:    v_mov_b32_e32 v17, v4
-; GPRIDX-NEXT:    v_mov_b32_e32 v16, v3
-; GPRIDX-NEXT:    v_mov_b32_e32 v15, v2
-; GPRIDX-NEXT:    v_mov_b32_e32 v14, v1
-; GPRIDX-NEXT:    v_mov_b32_e32 v13, v0
-; GPRIDX-NEXT:    v_mov_b32_e32 v13, v10
-; GPRIDX-NEXT:    s_set_gpr_idx_off
-; GPRIDX-NEXT:    s_set_gpr_idx_on s3, gpr_idx(DST)
-; GPRIDX-NEXT:    v_mov_b32_e32 v14, v11
-; GPRIDX-NEXT:    s_set_gpr_idx_off
-; GPRIDX-NEXT:    s_and_saveexec_b64 vcc, vcc
-; GPRIDX-NEXT:    s_xor_b64 exec, exec, vcc
-; GPRIDX-NEXT:    s_cbranch_execnz BB58_1
-; GPRIDX-NEXT:  ; %bb.2:
-; GPRIDX-NEXT:    s_mov_b64 exec, s[0:1]
-; GPRIDX-NEXT:    v_readfirstlane_b32 s0, v13
-; GPRIDX-NEXT:    v_readfirstlane_b32 s1, v14
-; GPRIDX-NEXT:    v_readfirstlane_b32 s2, v15
-; GPRIDX-NEXT:    v_readfirstlane_b32 s3, v16
-; GPRIDX-NEXT:    v_readfirstlane_b32 s4, v17
-; GPRIDX-NEXT:    v_readfirstlane_b32 s5, v18
-; GPRIDX-NEXT:    v_readfirstlane_b32 s6, v19
-; GPRIDX-NEXT:    v_readfirstlane_b32 s7, v20
-; GPRIDX-NEXT:    v_readfirstlane_b32 s8, v21
-; GPRIDX-NEXT:    v_readfirstlane_b32 s9, v22
+; GPRIDX-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v12
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 s[0:1], 1, v12
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 s[2:3], 2, v12
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 s[4:5], 3, v12
+; GPRIDX-NEXT:    v_cmp_eq_u32_e64 s[6:7], 4, v12
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v8, v8, v10, s[6:7]
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v9, v9, v11, s[6:7]
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v6, v6, v10, s[4:5]
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v7, v7, v11, s[4:5]
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v4, v4, v10, s[2:3]
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v5, v5, v11, s[2:3]
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v2, v2, v10, s[0:1]
+; GPRIDX-NEXT:    v_cndmask_b32_e64 v3, v3, v11, s[0:1]
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v0, v0, v10, vcc
+; GPRIDX-NEXT:    v_cndmask_b32_e32 v1, v1, v11, vcc
+; GPRIDX-NEXT:    v_readfirstlane_b32 s0, v0
+; GPRIDX-NEXT:    v_readfirstlane_b32 s1, v1
+; GPRIDX-NEXT:    v_readfirstlane_b32 s2, v2
+; GPRIDX-NEXT:    v_readfirstlane_b32 s3, v3
+; GPRIDX-NEXT:    v_readfirstlane_b32 s4, v4
+; GPRIDX-NEXT:    v_readfirstlane_b32 s5, v5
+; GPRIDX-NEXT:    v_readfirstlane_b32 s6, v6
+; GPRIDX-NEXT:    v_readfirstlane_b32 s7, v7
+; GPRIDX-NEXT:    v_readfirstlane_b32 s8, v8
+; GPRIDX-NEXT:    v_readfirstlane_b32 s9, v9
 ; GPRIDX-NEXT:    ; return to shader part epilog
 ;
 ; MOVREL-LABEL: dyn_insertelement_v5f64_v_v_v:
 ; MOVREL:       ; %bb.0: ; %entry
-; MOVREL-NEXT:    s_mov_b32 s0, exec_lo
-; MOVREL-NEXT:    ; implicit-def: $vcc_hi
-; MOVREL-NEXT:  BB58_1: ; =>This Inner Loop Header: Depth=1
-; MOVREL-NEXT:    v_readfirstlane_b32 s1, v12
-; MOVREL-NEXT:    v_mov_b32_e32 v28, v15
-; MOVREL-NEXT:    v_mov_b32_e32 v27, v14
-; MOVREL-NEXT:    v_mov_b32_e32 v26, v13
-; MOVREL-NEXT:    v_mov_b32_e32 v25, v12
-; MOVREL-NEXT:    v_mov_b32_e32 v24, v11
-; MOVREL-NEXT:    v_mov_b32_e32 v23, v10
-; MOVREL-NEXT:    v_mov_b32_e32 v22, v9
-; MOVREL-NEXT:    v_mov_b32_e32 v21, v8
-; MOVREL-NEXT:    v_mov_b32_e32 v20, v7
-; MOVREL-NEXT:    v_mov_b32_e32 v19, v6
-; MOVREL-NEXT:    v_mov_b32_e32 v18, v5
-; MOVREL-NEXT:    v_mov_b32_e32 v17, v4
-; MOVREL-NEXT:    v_mov_b32_e32 v16, v3
-; MOVREL-NEXT:    v_mov_b32_e32 v15, v2
-; MOVREL-NEXT:    v_mov_b32_e32 v14, v1
-; MOVREL-NEXT:    v_mov_b32_e32 v13, v0
-; MOVREL-NEXT:    s_lshl_b32 m0, s1, 1
-; MOVREL-NEXT:    v_cmp_eq_u32_e32 vcc_lo, s1, v12
-; MOVREL-NEXT:    v_movreld_b32_e32 v13, v10
-; MOVREL-NEXT:    v_movreld_b32_e32 v14, v11
-; MOVREL-NEXT:    s_and_saveexec_b32 vcc_lo, vcc_lo
-; MOVREL-NEXT:    s_xor_b32 exec_lo, exec_lo, vcc_lo
-; MOVREL-NEXT:    s_cbranch_execnz BB58_1
-; MOVREL-NEXT:  ; %bb.2:
-; MOVREL-NEXT:    s_mov_b32 exec_lo, s0
-; MOVREL-NEXT:    v_readfirstlane_b32 s0, v13
-; MOVREL-NEXT:    v_readfirstlane_b32 s1, v14
-; MOVREL-NEXT:    v_readfirstlane_b32 s2, v15
-; MOVREL-NEXT:    v_readfirstlane_b32 s3, v16
-; MOVREL-NEXT:    v_readfirstlane_b32 s4, v17
-; MOVREL-NEXT:    v_readfirstlane_b32 s5, v18
-; MOVREL-NEXT:    v_readfirstlane_b32 s6, v19
-; MOVREL-NEXT:    v_readfirstlane_b32 s7, v20
-; MOVREL-NEXT:    v_readfirstlane_b32 s8, v21
-; MOVREL-NEXT:    v_readfirstlane_b32 s9, v22
+; MOVREL-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 0, v12
+; MOVREL-NEXT:    v_mov_b32_e32 v13, v2
+; MOVREL-NEXT:    v_mov_b32_e32 v14, v3
+; MOVREL-NEXT:    ; implicit-def: $vcc_hi
+; MOVREL-NEXT:    v_cndmask_b32_e32 v0, v0, v10, vcc_lo
+; MOVREL-NEXT:    v_cndmask_b32_e32 v1, v1, v11, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 1, v12
+; MOVREL-NEXT:    v_readfirstlane_b32 s0, v0
+; MOVREL-NEXT:    v_readfirstlane_b32 s1, v1
+; MOVREL-NEXT:    v_cndmask_b32_e32 v2, v13, v10, vcc_lo
+; MOVREL-NEXT:    v_cndmask_b32_e32 v3, v14, v11, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 2, v12
+; MOVREL-NEXT:    v_readfirstlane_b32 s2, v2
+; MOVREL-NEXT:    v_readfirstlane_b32 s3, v3
+; MOVREL-NEXT:    v_cndmask_b32_e32 v4, v4, v10, vcc_lo
+; MOVREL-NEXT:    v_cndmask_b32_e32 v5, v5, v11, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 3, v12
+; MOVREL-NEXT:    v_readfirstlane_b32 s4, v4
+; MOVREL-NEXT:    v_readfirstlane_b32 s5, v5
+; MOVREL-NEXT:    v_cndmask_b32_e32 v6, v6, v10, vcc_lo
+; MOVREL-NEXT:    v_cndmask_b32_e32 v7, v7, v11, vcc_lo
+; MOVREL-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 4, v12
+; MOVREL-NEXT:    v_readfirstlane_b32 s6, v6
+; MOVREL-NEXT:    v_readfirstlane_b32 s7, v7
+; MOVREL-NEXT:    v_cndmask_b32_e32 v8, v8, v10, vcc_lo
+; MOVREL-NEXT:    v_cndmask_b32_e32 v9, v9, v11, vcc_lo
+; MOVREL-NEXT:    v_readfirstlane_b32 s8, v8
+; MOVREL-NEXT:    v_readfirstlane_b32 s9, v9
 ; MOVREL-NEXT:    ; return to shader part epilog
 entry:
   %insert = insertelement <5 x double> %vec, double %val, i32 %idx
   ret <5 x double> %insert
 }

llvm/test/CodeGen/AMDGPU/GlobalISel/regbankselect-insert-vector-elt.mir

 # NOTE: Assertions have been autogenerated by utils/update_mir_test_checks.py
-# RUN: llc -march=amdgcn -mcpu=fiji -run-pass=regbankselect %s -verify-machineinstrs -o - -regbankselect-fast | FileCheck %s
-# RUN: llc -march=amdgcn -mcpu=fiji -run-pass=regbankselect %s -verify-machineinstrs -o - -regbankselect-greedy | FileCheck %s
+# RUN: llc -march=amdgcn -mcpu=fiji -run-pass=regbankselect %s -verify-machineinstrs -o - -regbankselect-fast | FileCheck -check-prefixes=GCN,FAST %s
+# RUN: llc -march=amdgcn -mcpu=fiji -run-pass=regbankselect %s -verify-machineinstrs -o - -regbankselect-greedy | FileCheck -check-prefixes=GCN,GREEDY %s
 
 ---
 name: insert_vector_elt_v4i32_s_s_s
 legalized: true
 
 body: |
   bb.0:
     liveins: $sgpr0_sgpr1_sgpr2_sgpr3, $sgpr4, $sgpr5
 
-    ; CHECK-LABEL: name: insert_vector_elt_v4i32_s_s_s
-    ; CHECK: [[COPY:%[0-9]+]]:sgpr(<4 x s32>) = COPY $sgpr0_sgpr1_sgpr2_sgpr3
-    ; CHECK: [[COPY1:%[0-9]+]]:sgpr(s32) = COPY $sgpr4
-    ; CHECK: [[COPY2:%[0-9]+]]:sgpr(s32) = COPY $sgpr5
-    ; CHECK: [[IVEC:%[0-9]+]]:sgpr(<4 x s32>) = G_INSERT_VECTOR_ELT [[COPY]], [[COPY1]](s32), [[COPY2]](s32)
-    ; CHECK: $sgpr0_sgpr1_sgpr2_sgpr3 = COPY [[IVEC]](<4 x s32>)
+    ; FAST-LABEL: name: insert_vector_elt_v4i32_s_s_s
+    ; FAST: [[COPY:%[0-9]+]]:sgpr(<4 x s32>) = COPY $sgpr0_sgpr1_sgpr2_sgpr3
+    ; FAST: [[COPY1:%[0-9]+]]:sgpr(s32) = COPY $sgpr4
+    ; FAST: [[COPY2:%[0-9]+]]:sgpr(s32) = COPY $sgpr5
+    ; FAST: [[UV:%[0-9]+]]:sgpr(s32), [[UV1:%[0-9]+]]:sgpr(s32), [[UV2:%[0-9]+]]:sgpr(s32), [[UV3:%[0-9]+]]:sgpr(s32) = G_UNMERGE_VALUES [[COPY]](<4 x s32>)
+    ; FAST: [[C:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 0
+    ; FAST: [[ICMP:%[0-9]+]]:sgpr(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C]]
+    ; FAST: [[SELECT:%[0-9]+]]:sgpr(s32) = G_SELECT [[ICMP]](s1), [[COPY1]], [[UV]]
+    ; FAST: [[C1:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 1
+    ; FAST: [[ICMP1:%[0-9]+]]:sgpr(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C1]]
+    ; FAST: [[SELECT1:%[0-9]+]]:sgpr(s32) = G_SELECT [[ICMP1]](s1), [[COPY1]], [[UV1]]
+    ; FAST: [[C2:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 2
+    ; FAST: [[ICMP2:%[0-9]+]]:sgpr(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C2]]
+    ; FAST: [[SELECT2:%[0-9]+]]:sgpr(s32) = G_SELECT [[ICMP2]](s1), [[COPY1]], [[UV2]]
+    ; FAST: [[C3:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 3
+    ; FAST: [[ICMP3:%[0-9]+]]:sgpr(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C3]]
+    ; FAST: [[SELECT3:%[0-9]+]]:sgpr(s32) = G_SELECT [[ICMP3]](s1), [[COPY1]], [[UV3]]
+    ; FAST: [[BUILD_VECTOR:%[0-9]+]]:sgpr(<4 x s32>) = G_BUILD_VECTOR [[SELECT]](s32), [[SELECT1]](s32), [[SELECT2]](s32), [[SELECT3]](s32)
+    ; FAST: $sgpr0_sgpr1_sgpr2_sgpr3 = COPY [[BUILD_VECTOR]](<4 x s32>)
+    ; GREEDY-LABEL: name: insert_vector_elt_v4i32_s_s_s
+    ; GREEDY: [[COPY:%[0-9]+]]:sgpr(<4 x s32>) = COPY $sgpr0_sgpr1_sgpr2_sgpr3
+    ; GREEDY: [[COPY1:%[0-9]+]]:sgpr(s32) = COPY $sgpr4
+    ; GREEDY: [[COPY2:%[0-9]+]]:sgpr(s32) = COPY $sgpr5
+    ; GREEDY: [[UV:%[0-9]+]]:sgpr(s32), [[UV1:%[0-9]+]]:sgpr(s32), [[UV2:%[0-9]+]]:sgpr(s32), [[UV3:%[0-9]+]]:sgpr(s32) = G_UNMERGE_VALUES [[COPY]](<4 x s32>)
+    ; GREEDY: [[C:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 0
+    ; GREEDY: [[ICMP:%[0-9]+]]:sgpr(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C]]
+    ; GREEDY: [[SELECT:%[0-9]+]]:sgpr(s32) = G_SELECT [[ICMP]](s1), [[COPY1]], [[UV]]
+    ; GREEDY: [[C1:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 1
+    ; GREEDY: [[ICMP1:%[0-9]+]]:sgpr(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C1]]
+    ; GREEDY: [[SELECT1:%[0-9]+]]:sgpr(s32) = G_SELECT [[ICMP1]](s1), [[COPY1]], [[UV1]]
+    ; GREEDY: [[C2:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 2
+    ; GREEDY: [[ICMP2:%[0-9]+]]:sgpr(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C2]]
+    ; GREEDY: [[SELECT2:%[0-9]+]]:sgpr(s32) = G_SELECT [[ICMP2]](s1), [[COPY1]], [[UV2]]
+    ; GREEDY: [[C3:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 3
+    ; GREEDY: [[ICMP3:%[0-9]+]]:sgpr(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C3]]
+    ; GREEDY: [[SELECT3:%[0-9]+]]:sgpr(s32) = G_SELECT [[ICMP3]](s1), [[COPY1]], [[UV3]]
+    ; GREEDY: [[BUILD_VECTOR:%[0-9]+]]:sgpr(<4 x s32>) = G_BUILD_VECTOR [[SELECT]](s32), [[SELECT1]](s32), [[SELECT2]](s32), [[SELECT3]](s32)
+    ; GREEDY: $sgpr0_sgpr1_sgpr2_sgpr3 = COPY [[BUILD_VECTOR]](<4 x s32>)
     %0:_(<4 x s32>) = COPY $sgpr0_sgpr1_sgpr2_sgpr3
     %1:_(s32) = COPY $sgpr4
     %2:_(s32) = COPY $sgpr5
     %3:_(<4 x s32>) = G_INSERT_VECTOR_ELT %0, %1, %2
     $sgpr0_sgpr1_sgpr2_sgpr3 = COPY %3
 ...
 
 ---
 name: insert_vector_elt_v4i32_v_s_s
 legalized: true
 
 body: |
   bb.0:
     liveins: $vgpr0_vgpr1_vgpr2_vgpr3, $sgpr0, $sgpr1
 
-    ; CHECK-LABEL: name: insert_vector_elt_v4i32_v_s_s
-    ; CHECK: [[COPY:%[0-9]+]]:vgpr(<4 x s32>) = COPY $vgpr0_vgpr1_vgpr2_vgpr3
-    ; CHECK: [[COPY1:%[0-9]+]]:sgpr(s32) = COPY $sgpr0
-    ; CHECK: [[COPY2:%[0-9]+]]:sgpr(s32) = COPY $sgpr1
-    ; CHECK: [[IVEC:%[0-9]+]]:vgpr(<4 x s32>) = G_INSERT_VECTOR_ELT [[COPY]], [[COPY1]](s32), [[COPY2]](s32)
-    ; CHECK: $vgpr0_vgpr1_vgpr2_vgpr3 = COPY [[IVEC]](<4 x s32>)
+    ; FAST-LABEL: name: insert_vector_elt_v4i32_v_s_s
+    ; FAST: [[COPY:%[0-9]+]]:vgpr(<4 x s32>) = COPY $vgpr0_vgpr1_vgpr2_vgpr3
+    ; FAST: [[COPY1:%[0-9]+]]:vgpr(s32) = COPY $sgpr0
+    ; FAST: [[COPY2:%[0-9]+]]:sgpr(s32) = COPY $sgpr1
+    ; FAST: [[COPY3:%[0-9]+]]:vgpr(s32) = COPY [[COPY2]](s32)
+    ; FAST: [[UV:%[0-9]+]]:vgpr(s32), [[UV1:%[0-9]+]]:vgpr(s32), [[UV2:%[0-9]+]]:vgpr(s32), [[UV3:%[0-9]+]]:vgpr(s32) = G_UNMERGE_VALUES [[COPY]](<4 x s32>)
+    ; FAST: [[C:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 0
+    ; FAST: [[ICMP:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY3]](s32), [[C]]
+    ; FAST: [[SELECT:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP]](s1), [[COPY1]], [[UV]]
+    ; FAST: [[C1:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 1
+    ; FAST: [[ICMP1:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY3]](s32), [[C1]]
+    ; FAST: [[SELECT1:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP1]](s1), [[COPY1]], [[UV1]]
+    ; FAST: [[C2:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 2
+    ; FAST: [[ICMP2:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY3]](s32), [[C2]]
+    ; FAST: [[SELECT2:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP2]](s1), [[COPY1]], [[UV2]]
+    ; FAST: [[C3:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 3
+    ; FAST: [[ICMP3:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY3]](s32), [[C3]]
+    ; FAST: [[SELECT3:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP3]](s1), [[COPY1]], [[UV3]]
+    ; FAST: [[BUILD_VECTOR:%[0-9]+]]:vgpr(<4 x s32>) = G_BUILD_VECTOR [[SELECT]](s32), [[SELECT1]](s32), [[SELECT2]](s32), [[SELECT3]](s32)
+    ; FAST: $vgpr0_vgpr1_vgpr2_vgpr3 = COPY [[BUILD_VECTOR]](<4 x s32>)
+    ; GREEDY-LABEL: name: insert_vector_elt_v4i32_v_s_s
+    ; GREEDY: [[COPY:%[0-9]+]]:vgpr(<4 x s32>) = COPY $vgpr0_vgpr1_vgpr2_vgpr3
+    ; GREEDY: [[COPY1:%[0-9]+]]:vgpr(s32) = COPY $sgpr0
+    ; GREEDY: [[COPY2:%[0-9]+]]:sgpr(s32) = COPY $sgpr1
+    ; GREEDY: [[COPY3:%[0-9]+]]:vgpr(s32) = COPY [[COPY2]](s32)
+    ; GREEDY: [[UV:%[0-9]+]]:vgpr(s32), [[UV1:%[0-9]+]]:vgpr(s32), [[UV2:%[0-9]+]]:vgpr(s32), [[UV3:%[0-9]+]]:vgpr(s32) = G_UNMERGE_VALUES [[COPY]](<4 x s32>)
+    ; GREEDY: [[C:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 0
+    ; GREEDY: [[ICMP:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY3]](s32), [[C]]
+    ; GREEDY: [[SELECT:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP]](s1), [[COPY1]], [[UV]]
+    ; GREEDY: [[C1:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 1
+    ; GREEDY: [[ICMP1:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY3]](s32), [[C1]]
+    ; GREEDY: [[SELECT1:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP1]](s1), [[COPY1]], [[UV1]]
+    ; GREEDY: [[C2:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 2
+    ; GREEDY: [[ICMP2:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY3]](s32), [[C2]]
+    ; GREEDY: [[SELECT2:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP2]](s1), [[COPY1]], [[UV2]]
+    ; GREEDY: [[C3:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 3
+    ; GREEDY: [[ICMP3:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY3]](s32), [[C3]]
+    ; GREEDY: [[SELECT3:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP3]](s1), [[COPY1]], [[UV3]]
+    ; GREEDY: [[BUILD_VECTOR:%[0-9]+]]:vgpr(<4 x s32>) = G_BUILD_VECTOR [[SELECT]](s32), [[SELECT1]](s32), [[SELECT2]](s32), [[SELECT3]](s32)
+    ; GREEDY: $vgpr0_vgpr1_vgpr2_vgpr3 = COPY [[BUILD_VECTOR]](<4 x s32>)
     %0:_(<4 x s32>) = COPY $vgpr0_vgpr1_vgpr2_vgpr3
     %1:_(s32) = COPY $sgpr0
     %2:_(s32) = COPY $sgpr1
     %3:_(<4 x s32>) = G_INSERT_VECTOR_ELT %0, %1, %2
     $vgpr0_vgpr1_vgpr2_vgpr3 = COPY %3
 ...
 
 ---
 name: insert_vector_elt_v4i32_s_v_s
 legalized: true
 
 body: |
   bb.0:
     liveins: $sgpr0_sgpr1_sgpr2_sgpr3, $vgpr0, $sgpr4
 
-    ; CHECK-LABEL: name: insert_vector_elt_v4i32_s_v_s
-    ; CHECK: [[COPY:%[0-9]+]]:sgpr(<4 x s32>) = COPY $sgpr0_sgpr1_sgpr2_sgpr3
-    ; CHECK: [[COPY1:%[0-9]+]]:vgpr(s32) = COPY $vgpr0
-    ; CHECK: [[COPY2:%[0-9]+]]:sgpr(s32) = COPY $sgpr4
-    ; CHECK: [[COPY3:%[0-9]+]]:vgpr(<4 x s32>) = COPY [[COPY]](<4 x s32>)
-    ; CHECK: [[IVEC:%[0-9]+]]:vgpr(<4 x s32>) = G_INSERT_VECTOR_ELT [[COPY3]], [[COPY1]](s32), [[COPY2]](s32)
-    ; CHECK: $vgpr0_vgpr1_vgpr2_vgpr3 = COPY [[IVEC]](<4 x s32>)
+    ; FAST-LABEL: name: insert_vector_elt_v4i32_s_v_s
+    ; FAST: [[COPY:%[0-9]+]]:sgpr(<4 x s32>) = COPY $sgpr0_sgpr1_sgpr2_sgpr3
+    ; FAST: [[COPY1:%[0-9]+]]:vgpr(s32) = COPY $vgpr0
+    ; FAST: [[COPY2:%[0-9]+]]:sgpr(s32) = COPY $sgpr4
+    ; FAST: [[COPY3:%[0-9]+]]:vgpr(<4 x s32>) = COPY [[COPY]](<4 x s32>)
+    ; FAST: [[COPY4:%[0-9]+]]:vgpr(s32) = COPY [[COPY2]](s32)
+    ; FAST: [[UV:%[0-9]+]]:vgpr(s32), [[UV1:%[0-9]+]]:vgpr(s32), [[UV2:%[0-9]+]]:vgpr(s32), [[UV3:%[0-9]+]]:vgpr(s32) = G_UNMERGE_VALUES [[COPY3]](<4 x s32>)
+    ; FAST: [[C:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 0
+    ; FAST: [[ICMP:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY4]](s32), [[C]]
+    ; FAST: [[SELECT:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP]](s1), [[COPY1]], [[UV]]
+    ; FAST: [[C1:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 1
+    ; FAST: [[ICMP1:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY4]](s32), [[C1]]
+    ; FAST: [[SELECT1:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP1]](s1), [[COPY1]], [[UV1]]
+    ; FAST: [[C2:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 2
+    ; FAST: [[ICMP2:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY4]](s32), [[C2]]
+    ; FAST: [[SELECT2:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP2]](s1), [[COPY1]], [[UV2]]
+    ; FAST: [[C3:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 3
+    ; FAST: [[ICMP3:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY4]](s32), [[C3]]
+    ; FAST: [[SELECT3:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP3]](s1), [[COPY1]], [[UV3]]
+    ; FAST: [[BUILD_VECTOR:%[0-9]+]]:vgpr(<4 x s32>) = G_BUILD_VECTOR [[SELECT]](s32), [[SELECT1]](s32), [[SELECT2]](s32), [[SELECT3]](s32)
+    ; FAST: $vgpr0_vgpr1_vgpr2_vgpr3 = COPY [[BUILD_VECTOR]](<4 x s32>)
+    ; GREEDY-LABEL: name: insert_vector_elt_v4i32_s_v_s
+    ; GREEDY: [[COPY:%[0-9]+]]:sgpr(<4 x s32>) = COPY $sgpr0_sgpr1_sgpr2_sgpr3
+    ; GREEDY: [[COPY1:%[0-9]+]]:vgpr(s32) = COPY $vgpr0
+    ; GREEDY: [[COPY2:%[0-9]+]]:sgpr(s32) = COPY $sgpr4
+    ; GREEDY: [[COPY3:%[0-9]+]]:vgpr(<4 x s32>) = COPY [[COPY]](<4 x s32>)
+    ; GREEDY: [[COPY4:%[0-9]+]]:vgpr(s32) = COPY [[COPY2]](s32)
+    ; GREEDY: [[UV:%[0-9]+]]:vgpr(s32), [[UV1:%[0-9]+]]:vgpr(s32), [[UV2:%[0-9]+]]:vgpr(s32), [[UV3:%[0-9]+]]:vgpr(s32) = G_UNMERGE_VALUES [[COPY3]](<4 x s32>)
+    ; GREEDY: [[C:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 0
+    ; GREEDY: [[ICMP:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY4]](s32), [[C]]
+    ; GREEDY: [[SELECT:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP]](s1), [[COPY1]], [[UV]]
+    ; GREEDY: [[C1:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 1
+    ; GREEDY: [[ICMP1:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY4]](s32), [[C1]]
+    ; GREEDY: [[SELECT1:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP1]](s1), [[COPY1]], [[UV1]]
+    ; GREEDY: [[C2:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 2
+    ; GREEDY: [[ICMP2:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY4]](s32), [[C2]]
+    ; GREEDY: [[SELECT2:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP2]](s1), [[COPY1]], [[UV2]]
+    ; GREEDY: [[C3:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 3
+    ; GREEDY: [[ICMP3:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY4]](s32), [[C3]]
+    ; GREEDY: [[SELECT3:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP3]](s1), [[COPY1]], [[UV3]]
+    ; GREEDY: [[BUILD_VECTOR:%[0-9]+]]:vgpr(<4 x s32>) = G_BUILD_VECTOR [[SELECT]](s32), [[SELECT1]](s32), [[SELECT2]](s32), [[SELECT3]](s32)
+    ; GREEDY: $vgpr0_vgpr1_vgpr2_vgpr3 = COPY [[BUILD_VECTOR]](<4 x s32>)
     %0:_(<4 x s32>) = COPY $sgpr0_sgpr1_sgpr2_sgpr3
     %1:_(s32) = COPY $vgpr0
     %2:_(s32) = COPY $sgpr4
     %3:_(<4 x s32>) = G_INSERT_VECTOR_ELT %0, %1, %2
     $vgpr0_vgpr1_vgpr2_vgpr3 = COPY %3
 ...
 
 ---
 name: insert_vector_elt_v4i32_s_s_v
 legalized: true
 tracksRegLiveness: true
 
 body: |
   bb.0:
     liveins: $sgpr0_sgpr1_sgpr2_sgpr3, $sgpr4, $vgpr0
 
-    ; CHECK-LABEL: name: insert_vector_elt_v4i32_s_s_v
-    ; CHECK: successors: %bb.1(0x80000000)
-    ; CHECK: liveins: $sgpr0_sgpr1_sgpr2_sgpr3, $sgpr4, $vgpr0
-    ; CHECK: [[COPY:%[0-9]+]]:sgpr(<4 x s32>) = COPY $sgpr0_sgpr1_sgpr2_sgpr3
-    ; CHECK: [[COPY1:%[0-9]+]]:sgpr(s32) = COPY $sgpr4
-    ; CHECK: [[COPY2:%[0-9]+]]:vgpr_32(s32) = COPY $vgpr0
-    ; CHECK: [[COPY3:%[0-9]+]]:vgpr(<4 x s32>) = COPY [[COPY]](<4 x s32>)
-    ; CHECK: [[DEF:%[0-9]+]]:vgpr(<4 x s32>) = G_IMPLICIT_DEF
-    ; CHECK: [[DEF1:%[0-9]+]]:sreg_64_xexec = IMPLICIT_DEF
-    ; CHECK: [[S_MOV_B64_term:%[0-9]+]]:sreg_64_xexec = S_MOV_B64_term $exec
-    ; CHECK: .1:
-    ; CHECK: successors: %bb.2(0x40000000), %bb.1(0x40000000)
-    ; CHECK: [[PHI:%[0-9]+]]:sreg_64_xexec = PHI [[DEF1]], %bb.0, %10, %bb.1
-    ; CHECK: [[PHI1:%[0-9]+]]:vgpr(<4 x s32>) = G_PHI [[DEF]](<4 x s32>), %bb.0, %3(<4 x s32>), %bb.1
-    ; CHECK: [[V_READFIRSTLANE_B32_:%[0-9]+]]:sreg_32_xm0(s32) = V_READFIRSTLANE_B32 [[COPY2]](s32), implicit $exec
-    ; CHECK: [[V_CMP_EQ_U32_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U32_e64 [[V_READFIRSTLANE_B32_]](s32), [[COPY2]](s32), implicit $exec
-    ; CHECK: [[IVEC:%[0-9]+]]:vgpr(<4 x s32>) = G_INSERT_VECTOR_ELT [[COPY3]], [[COPY1]](s32), [[V_READFIRSTLANE_B32_]](s32)
-    ; CHECK: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[V_CMP_EQ_U32_e64_]], implicit-def $exec, implicit-def $scc, implicit $exec
-    ; CHECK: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc
-    ; CHECK: S_CBRANCH_EXECNZ %bb.1, implicit $exec
-    ; CHECK: .2:
-    ; CHECK: successors: %bb.3(0x80000000)
-    ; CHECK: $exec = S_MOV_B64_term [[S_MOV_B64_term]]
-    ; CHECK: .3:
-    ; CHECK: $vgpr0_vgpr1_vgpr2_vgpr3 = COPY [[IVEC]](<4 x s32>)
+    ; FAST-LABEL: name: insert_vector_elt_v4i32_s_s_v
+    ; FAST: liveins: $sgpr0_sgpr1_sgpr2_sgpr3, $sgpr4, $vgpr0
+    ; FAST: [[COPY:%[0-9]+]]:sgpr(<4 x s32>) = COPY $sgpr0_sgpr1_sgpr2_sgpr3
+    ; FAST: [[COPY1:%[0-9]+]]:vgpr(s32) = COPY $sgpr4
+    ; FAST: [[COPY2:%[0-9]+]]:vgpr(s32) = COPY $vgpr0
+    ; FAST: [[COPY3:%[0-9]+]]:vgpr(<4 x s32>) = COPY [[COPY]](<4 x s32>)
+    ; FAST: [[UV:%[0-9]+]]:vgpr(s32), [[UV1:%[0-9]+]]:vgpr(s32), [[UV2:%[0-9]+]]:vgpr(s32), [[UV3:%[0-9]+]]:vgpr(s32) = G_UNMERGE_VALUES [[COPY3]](<4 x s32>)
+    ; FAST: [[C:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 0
+    ; FAST: [[ICMP:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C]]
+    ; FAST: [[SELECT:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP]](s1), [[COPY1]], [[UV]]
+    ; FAST: [[C1:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 1
+    ; FAST: [[ICMP1:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C1]]
+    ; FAST: [[SELECT1:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP1]](s1), [[COPY1]], [[UV1]]
+    ; FAST: [[C2:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 2
+    ; FAST: [[ICMP2:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C2]]
+    ; FAST: [[SELECT2:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP2]](s1), [[COPY1]], [[UV2]]
+    ; FAST: [[C3:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 3
+    ; FAST: [[ICMP3:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C3]]
+    ; FAST: [[SELECT3:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP3]](s1), [[COPY1]], [[UV3]]
+    ; FAST: [[BUILD_VECTOR:%[0-9]+]]:vgpr(<4 x s32>) = G_BUILD_VECTOR [[SELECT]](s32), [[SELECT1]](s32), [[SELECT2]](s32), [[SELECT3]](s32)
+    ; FAST: $vgpr0_vgpr1_vgpr2_vgpr3 = COPY [[BUILD_VECTOR]](<4 x s32>)
+    ; GREEDY-LABEL: name: insert_vector_elt_v4i32_s_s_v
+    ; GREEDY: liveins: $sgpr0_sgpr1_sgpr2_sgpr3, $sgpr4, $vgpr0
+    ; GREEDY: [[COPY:%[0-9]+]]:sgpr(<4 x s32>) = COPY $sgpr0_sgpr1_sgpr2_sgpr3
+    ; GREEDY: [[COPY1:%[0-9]+]]:vgpr(s32) = COPY $sgpr4
+    ; GREEDY: [[COPY2:%[0-9]+]]:vgpr(s32) = COPY $vgpr0
+    ; GREEDY: [[COPY3:%[0-9]+]]:vgpr(<4 x s32>) = COPY [[COPY]](<4 x s32>)
+    ; GREEDY: [[UV:%[0-9]+]]:vgpr(s32), [[UV1:%[0-9]+]]:vgpr(s32), [[UV2:%[0-9]+]]:vgpr(s32), [[UV3:%[0-9]+]]:vgpr(s32) = G_UNMERGE_VALUES [[COPY3]](<4 x s32>)
+    ; GREEDY: [[C:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 0
+    ; GREEDY: [[ICMP:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C]]
+    ; GREEDY: [[SELECT:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP]](s1), [[COPY1]], [[UV]]
+    ; GREEDY: [[C1:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 1
+    ; GREEDY: [[ICMP1:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C1]]
+    ; GREEDY: [[SELECT1:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP1]](s1), [[COPY1]], [[UV1]]
+    ; GREEDY: [[C2:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 2
+    ; GREEDY: [[ICMP2:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C2]]
+    ; GREEDY: [[SELECT2:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP2]](s1), [[COPY1]], [[UV2]]
+    ; GREEDY: [[C3:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 3
+    ; GREEDY: [[ICMP3:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C3]]
+    ; GREEDY: [[SELECT3:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP3]](s1), [[COPY1]], [[UV3]]
+    ; GREEDY: [[BUILD_VECTOR:%[0-9]+]]:vgpr(<4 x s32>) = G_BUILD_VECTOR [[SELECT]](s32), [[SELECT1]](s32), [[SELECT2]](s32), [[SELECT3]](s32)
+    ; GREEDY: $vgpr0_vgpr1_vgpr2_vgpr3 = COPY [[BUILD_VECTOR]](<4 x s32>)
     %0:_(<4 x s32>) = COPY $sgpr0_sgpr1_sgpr2_sgpr3
     %1:_(s32) = COPY $sgpr4
     %2:_(s32) = COPY $vgpr0
     %3:_(<4 x s32>) = G_INSERT_VECTOR_ELT %0, %1, %2
     $vgpr0_vgpr1_vgpr2_vgpr3 = COPY %3
 ...
 
 ---
 name: insert_vector_elt_v4i32_s_v_v
 legalized: true
 tracksRegLiveness: true
 
 body: |
   bb.0:
     liveins: $sgpr0_sgpr1_sgpr2_sgpr3, $vgpr0, $vgpr1
 
-    ; CHECK-LABEL: name: insert_vector_elt_v4i32_s_v_v
-    ; CHECK: successors: %bb.1(0x80000000)
-    ; CHECK: liveins: $sgpr0_sgpr1_sgpr2_sgpr3, $vgpr0, $vgpr1
-    ; CHECK: [[COPY:%[0-9]+]]:sgpr(<4 x s32>) = COPY $sgpr0_sgpr1_sgpr2_sgpr3
-    ; CHECK: [[COPY1:%[0-9]+]]:vgpr(s32) = COPY $vgpr0
-    ; CHECK: [[COPY2:%[0-9]+]]:vgpr_32(s32) = COPY $vgpr1
-    ; CHECK: [[COPY3:%[0-9]+]]:vgpr(<4 x s32>) = COPY [[COPY]](<4 x s32>)
-    ; CHECK: [[DEF:%[0-9]+]]:vgpr(<4 x s32>) = G_IMPLICIT_DEF
-    ; CHECK: [[DEF1:%[0-9]+]]:sreg_64_xexec = IMPLICIT_DEF
-    ; CHECK: [[S_MOV_B64_term:%[0-9]+]]:sreg_64_xexec = S_MOV_B64_term $exec
-    ; CHECK: .1:
-    ; CHECK: successors: %bb.2(0x40000000), %bb.1(0x40000000)
-    ; CHECK: [[PHI:%[0-9]+]]:sreg_64_xexec = PHI [[DEF1]], %bb.0, %10, %bb.1
-    ; CHECK: [[PHI1:%[0-9]+]]:vgpr(<4 x s32>) = G_PHI [[DEF]](<4 x s32>), %bb.0, %3(<4 x s32>), %bb.1
-    ; CHECK: [[V_READFIRSTLANE_B32_:%[0-9]+]]:sreg_32_xm0(s32) = V_READFIRSTLANE_B32 [[COPY2]](s32), implicit $exec
-    ; CHECK: [[V_CMP_EQ_U32_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U32_e64 [[V_READFIRSTLANE_B32_]](s32), [[COPY2]](s32), implicit $exec
-    ; CHECK: [[IVEC:%[0-9]+]]:vgpr(<4 x s32>) = G_INSERT_VECTOR_ELT [[COPY3]], [[COPY1]](s32), [[V_READFIRSTLANE_B32_]](s32)
-    ; CHECK: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[V_CMP_EQ_U32_e64_]], implicit-def $exec, implicit-def $scc, implicit $exec
-    ; CHECK: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc
-    ; CHECK: S_CBRANCH_EXECNZ %bb.1, implicit $exec
-    ; CHECK: .2:
-    ; CHECK: successors: %bb.3(0x80000000)
-    ; CHECK: $exec = S_MOV_B64_term [[S_MOV_B64_term]]
-    ; CHECK: .3:
-    ; CHECK: $vgpr0_vgpr1_vgpr2_vgpr3 = COPY [[IVEC]](<4 x s32>)
+    ; FAST-LABEL: name: insert_vector_elt_v4i32_s_v_v
+    ; FAST: liveins: $sgpr0_sgpr1_sgpr2_sgpr3, $vgpr0, $vgpr1
+    ; FAST: [[COPY:%[0-9]+]]:sgpr(<4 x s32>) = COPY $sgpr0_sgpr1_sgpr2_sgpr3
+    ; FAST: [[COPY1:%[0-9]+]]:vgpr(s32) = COPY $vgpr0
+    ; FAST: [[COPY2:%[0-9]+]]:vgpr(s32) = COPY $vgpr1
+    ; FAST: [[COPY3:%[0-9]+]]:vgpr(<4 x s32>) = COPY [[COPY]](<4 x s32>)
+    ; FAST: [[UV:%[0-9]+]]:vgpr(s32), [[UV1:%[0-9]+]]:vgpr(s32), [[UV2:%[0-9]+]]:vgpr(s32), [[UV3:%[0-9]+]]:vgpr(s32) = G_UNMERGE_VALUES [[COPY3]](<4 x s32>)
+    ; FAST: [[C:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 0
+    ; FAST: [[ICMP:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C]]
+    ; FAST: [[SELECT:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP]](s1), [[COPY1]], [[UV]]
+    ; FAST: [[C1:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 1
+    ; FAST: [[ICMP1:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C1]]
+    ; FAST: [[SELECT1:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP1]](s1), [[COPY1]], [[UV1]]
+    ; FAST: [[C2:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 2
+    ; FAST: [[ICMP2:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C2]]
+    ; FAST: [[SELECT2:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP2]](s1), [[COPY1]], [[UV2]]
+    ; FAST: [[C3:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 3
+    ; FAST: [[ICMP3:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C3]]
+    ; FAST: [[SELECT3:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP3]](s1), [[COPY1]], [[UV3]]
+    ; FAST: [[BUILD_VECTOR:%[0-9]+]]:vgpr(<4 x s32>) = G_BUILD_VECTOR [[SELECT]](s32), [[SELECT1]](s32), [[SELECT2]](s32), [[SELECT3]](s32)
+    ; FAST: $vgpr0_vgpr1_vgpr2_vgpr3 = COPY [[BUILD_VECTOR]](<4 x s32>)
+    ; GREEDY-LABEL: name: insert_vector_elt_v4i32_s_v_v
+    ; GREEDY: liveins: $sgpr0_sgpr1_sgpr2_sgpr3, $vgpr0, $vgpr1
+    ; GREEDY: [[COPY:%[0-9]+]]:sgpr(<4 x s32>) = COPY $sgpr0_sgpr1_sgpr2_sgpr3
+    ; GREEDY: [[COPY1:%[0-9]+]]:vgpr(s32) = COPY $vgpr0
+    ; GREEDY: [[COPY2:%[0-9]+]]:vgpr(s32) = COPY $vgpr1
+    ; GREEDY: [[COPY3:%[0-9]+]]:vgpr(<4 x s32>) = COPY [[COPY]](<4 x s32>)
+    ; GREEDY: [[UV:%[0-9]+]]:vgpr(s32), [[UV1:%[0-9]+]]:vgpr(s32), [[UV2:%[0-9]+]]:vgpr(s32), [[UV3:%[0-9]+]]:vgpr(s32) = G_UNMERGE_VALUES [[COPY3]](<4 x s32>)
+    ; GREEDY: [[C:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 0
+    ; GREEDY: [[ICMP:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C]]
+    ; GREEDY: [[SELECT:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP]](s1), [[COPY1]], [[UV]]
+    ; GREEDY: [[C1:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 1
+    ; GREEDY: [[ICMP1:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C1]]
+    ; GREEDY: [[SELECT1:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP1]](s1), [[COPY1]], [[UV1]]
+    ; GREEDY: [[C2:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 2
+    ; GREEDY: [[ICMP2:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C2]]
+    ; GREEDY: [[SELECT2:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP2]](s1), [[COPY1]], [[UV2]]
+    ; GREEDY: [[C3:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 3
+    ; GREEDY: [[ICMP3:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C3]]
+    ; GREEDY: [[SELECT3:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP3]](s1), [[COPY1]], [[UV3]]
+    ; GREEDY: [[BUILD_VECTOR:%[0-9]+]]:vgpr(<4 x s32>) = G_BUILD_VECTOR [[SELECT]](s32), [[SELECT1]](s32), [[SELECT2]](s32), [[SELECT3]](s32)
+    ; GREEDY: $vgpr0_vgpr1_vgpr2_vgpr3 = COPY [[BUILD_VECTOR]](<4 x s32>)
     %0:_(<4 x s32>) = COPY $sgpr0_sgpr1_sgpr2_sgpr3
     %1:_(s32) = COPY $vgpr0
     %2:_(s32) = COPY $vgpr1
     %3:_(<4 x s32>) = G_INSERT_VECTOR_ELT %0, %1, %2
     $vgpr0_vgpr1_vgpr2_vgpr3 = COPY %3
 ...
 
 ---
 name: insert_vector_elt_var_v4i32_v_s_v
 legalized: true
 tracksRegLiveness: true
 
 body: |
   bb.0:
     liveins: $vgpr0_vgpr1_vgpr2_vgpr3, $sgpr4, $vgpr0
 
-    ; CHECK-LABEL: name: insert_vector_elt_var_v4i32_v_s_v
-    ; CHECK: successors: %bb.1(0x80000000)
-    ; CHECK: liveins: $vgpr0_vgpr1_vgpr2_vgpr3, $sgpr4, $vgpr0
-    ; CHECK: [[COPY:%[0-9]+]]:vgpr(<4 x s32>) = COPY $vgpr0_vgpr1_vgpr2_vgpr3
-    ; CHECK: [[COPY1:%[0-9]+]]:sgpr(s32) = COPY $sgpr4
-    ; CHECK: [[COPY2:%[0-9]+]]:vgpr_32(s32) = COPY $vgpr0
-    ; CHECK: [[DEF:%[0-9]+]]:vgpr(<4 x s32>) = G_IMPLICIT_DEF
-    ; CHECK: [[DEF1:%[0-9]+]]:sreg_64_xexec = IMPLICIT_DEF
-    ; CHECK: [[S_MOV_B64_term:%[0-9]+]]:sreg_64_xexec = S_MOV_B64_term $exec
-    ; CHECK: .1:
-    ; CHECK: successors: %bb.2(0x40000000), %bb.1(0x40000000)
-    ; CHECK: [[PHI:%[0-9]+]]:sreg_64_xexec = PHI [[DEF1]], %bb.0, %9, %bb.1
-    ; CHECK: [[PHI1:%[0-9]+]]:vgpr(<4 x s32>) = G_PHI [[DEF]](<4 x s32>), %bb.0, %3(<4 x s32>), %bb.1
-    ; CHECK: [[V_READFIRSTLANE_B32_:%[0-9]+]]:sreg_32_xm0(s32) = V_READFIRSTLANE_B32 [[COPY2]](s32), implicit $exec
-    ; CHECK: [[V_CMP_EQ_U32_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U32_e64 [[V_READFIRSTLANE_B32_]](s32), [[COPY2]](s32), implicit $exec
-    ; CHECK: [[IVEC:%[0-9]+]]:vgpr(<4 x s32>) = G_INSERT_VECTOR_ELT [[COPY]], [[COPY1]](s32), [[V_READFIRSTLANE_B32_]](s32)
-    ; CHECK: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[V_CMP_EQ_U32_e64_]], implicit-def $exec, implicit-def $scc, implicit $exec
-    ; CHECK: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc
-    ; CHECK: S_CBRANCH_EXECNZ %bb.1, implicit $exec
-    ; CHECK: .2:
-    ; CHECK: successors: %bb.3(0x80000000)
-    ; CHECK: $exec = S_MOV_B64_term [[S_MOV_B64_term]]
-    ; CHECK: .3:
-    ; CHECK: $vgpr0_vgpr1_vgpr2_vgpr3 = COPY [[IVEC]](<4 x s32>)
+    ; FAST-LABEL: name: insert_vector_elt_var_v4i32_v_s_v
+    ; FAST: liveins: $vgpr0_vgpr1_vgpr2_vgpr3, $sgpr4, $vgpr0
+    ; FAST: [[COPY:%[0-9]+]]:vgpr(<4 x s32>) = COPY $vgpr0_vgpr1_vgpr2_vgpr3
+    ; FAST: [[COPY1:%[0-9]+]]:vgpr(s32) = COPY $sgpr4
+    ; FAST: [[COPY2:%[0-9]+]]:vgpr(s32) = COPY $vgpr0
+    ; FAST: [[UV:%[0-9]+]]:vgpr(s32), [[UV1:%[0-9]+]]:vgpr(s32), [[UV2:%[0-9]+]]:vgpr(s32), [[UV3:%[0-9]+]]:vgpr(s32) = G_UNMERGE_VALUES [[COPY]](<4 x s32>)
+    ; FAST: [[C:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 0
+    ; FAST: [[ICMP:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C]]
+    ; FAST: [[SELECT:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP]](s1), [[COPY1]], [[UV]]
+    ; FAST: [[C1:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 1
+    ; FAST: [[ICMP1:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C1]]
+    ; FAST: [[SELECT1:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP1]](s1), [[COPY1]], [[UV1]]
+    ; FAST: [[C2:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 2
+    ; FAST: [[ICMP2:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C2]]
+    ; FAST: [[SELECT2:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP2]](s1), [[COPY1]], [[UV2]]
+    ; FAST: [[C3:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 3
+    ; FAST: [[ICMP3:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C3]]
+    ; FAST: [[SELECT3:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP3]](s1), [[COPY1]], [[UV3]]
+    ; FAST: [[BUILD_VECTOR:%[0-9]+]]:vgpr(<4 x s32>) = G_BUILD_VECTOR [[SELECT]](s32), [[SELECT1]](s32), [[SELECT2]](s32), [[SELECT3]](s32)
+    ; FAST: $vgpr0_vgpr1_vgpr2_vgpr3 = COPY [[BUILD_VECTOR]](<4 x s32>)
+    ; GREEDY-LABEL: name: insert_vector_elt_var_v4i32_v_s_v
+    ; GREEDY: liveins: $vgpr0_vgpr1_vgpr2_vgpr3, $sgpr4, $vgpr0
+    ; GREEDY: [[COPY:%[0-9]+]]:vgpr(<4 x s32>) = COPY $vgpr0_vgpr1_vgpr2_vgpr3
+    ; GREEDY: [[COPY1:%[0-9]+]]:vgpr(s32) = COPY $sgpr4
+    ; GREEDY: [[COPY2:%[0-9]+]]:vgpr(s32) = COPY $vgpr0
+    ; GREEDY: [[UV:%[0-9]+]]:vgpr(s32), [[UV1:%[0-9]+]]:vgpr(s32), [[UV2:%[0-9]+]]:vgpr(s32), [[UV3:%[0-9]+]]:vgpr(s32) = G_UNMERGE_VALUES [[COPY]](<4 x s32>)
+    ; GREEDY: [[C:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 0
+    ; GREEDY: [[ICMP:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C]]
+    ; GREEDY: [[SELECT:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP]](s1), [[COPY1]], [[UV]]
+    ; GREEDY: [[C1:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 1
+    ; GREEDY: [[ICMP1:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C1]]
+    ; GREEDY: [[SELECT1:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP1]](s1), [[COPY1]], [[UV1]]
+    ; GREEDY: [[C2:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 2
+    ; GREEDY: [[ICMP2:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C2]]
+    ; GREEDY: [[SELECT2:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP2]](s1), [[COPY1]], [[UV2]]
+    ; GREEDY: [[C3:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 3
+    ; GREEDY: [[ICMP3:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C3]]
+    ; GREEDY: [[SELECT3:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP3]](s1), [[COPY1]], [[UV3]]
+    ; GREEDY: [[BUILD_VECTOR:%[0-9]+]]:vgpr(<4 x s32>) = G_BUILD_VECTOR [[SELECT]](s32), [[SELECT1]](s32), [[SELECT2]](s32), [[SELECT3]](s32)
+    ; GREEDY: $vgpr0_vgpr1_vgpr2_vgpr3 = COPY [[BUILD_VECTOR]](<4 x s32>)
     %0:_(<4 x s32>) = COPY $vgpr0_vgpr1_vgpr2_vgpr3
     %1:_(s32) = COPY $sgpr4
     %2:_(s32) = COPY $vgpr0
     %3:_(<4 x s32>) = G_INSERT_VECTOR_ELT %0, %1, %2
     $vgpr0_vgpr1_vgpr2_vgpr3 = COPY %3
 ...
 
 ---
 name: insert_vector_elt_var_v4i32_v_v_s
 legalized: true
 tracksRegLiveness: true
 
 body: |
   bb.0:
     liveins: $vgpr0_vgpr1_vgpr2_vgpr3, $vgpr4, $sgpr0
 
-    ; CHECK-LABEL: name: insert_vector_elt_var_v4i32_v_v_s
-    ; CHECK: liveins: $vgpr0_vgpr1_vgpr2_vgpr3, $vgpr4, $sgpr0
-    ; CHECK: [[COPY:%[0-9]+]]:vgpr(<4 x s32>) = COPY $vgpr0_vgpr1_vgpr2_vgpr3
-    ; CHECK: [[COPY1:%[0-9]+]]:vgpr(s32) = COPY $vgpr0
-    ; CHECK: [[COPY2:%[0-9]+]]:sgpr(s32) = COPY $sgpr0
-    ; CHECK: [[IVEC:%[0-9]+]]:vgpr(<4 x s32>) = G_INSERT_VECTOR_ELT [[COPY]], [[COPY1]](s32), [[COPY2]](s32)
-    ; CHECK: $vgpr0_vgpr1_vgpr2_vgpr3 = COPY [[IVEC]](<4 x s32>)
+    ; FAST-LABEL: name: insert_vector_elt_var_v4i32_v_v_s
+    ; FAST: liveins: $vgpr0_vgpr1_vgpr2_vgpr3, $vgpr4, $sgpr0
+    ; FAST: [[COPY:%[0-9]+]]:vgpr(<4 x s32>) = COPY $vgpr0_vgpr1_vgpr2_vgpr3
+    ; FAST: [[COPY1:%[0-9]+]]:vgpr(s32) = COPY $vgpr0
+    ; FAST: [[COPY2:%[0-9]+]]:sgpr(s32) = COPY $sgpr0
+    ; FAST: [[COPY3:%[0-9]+]]:vgpr(s32) = COPY [[COPY2]](s32)
+    ; FAST: [[UV:%[0-9]+]]:vgpr(s32), [[UV1:%[0-9]+]]:vgpr(s32), [[UV2:%[0-9]+]]:vgpr(s32), [[UV3:%[0-9]+]]:vgpr(s32) = G_UNMERGE_VALUES [[COPY]](<4 x s32>)
+    ; FAST: [[C:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 0
+    ; FAST: [[ICMP:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY3]](s32), [[C]]
+    ; FAST: [[SELECT:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP]](s1), [[COPY1]], [[UV]]
+    ; FAST: [[C1:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 1
+    ; FAST: [[ICMP1:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY3]](s32), [[C1]]
+    ; FAST: [[SELECT1:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP1]](s1), [[COPY1]], [[UV1]]
+    ; FAST: [[C2:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 2
+    ; FAST: [[ICMP2:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY3]](s32), [[C2]]
+    ; FAST: [[SELECT2:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP2]](s1), [[COPY1]], [[UV2]]
+    ; FAST: [[C3:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 3
+    ; FAST: [[ICMP3:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY3]](s32), [[C3]]
+    ; FAST: [[SELECT3:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP3]](s1), [[COPY1]], [[UV3]]
+    ; FAST: [[BUILD_VECTOR:%[0-9]+]]:vgpr(<4 x s32>) = G_BUILD_VECTOR [[SELECT]](s32), [[SELECT1]](s32), [[SELECT2]](s32), [[SELECT3]](s32)
+    ; FAST: $vgpr0_vgpr1_vgpr2_vgpr3 = COPY [[BUILD_VECTOR]](<4 x s32>)
+    ; GREEDY-LABEL: name: insert_vector_elt_var_v4i32_v_v_s
+    ; GREEDY: liveins: $vgpr0_vgpr1_vgpr2_vgpr3, $vgpr4, $sgpr0
+    ; GREEDY: [[COPY:%[0-9]+]]:vgpr(<4 x s32>) = COPY $vgpr0_vgpr1_vgpr2_vgpr3
+    ; GREEDY: [[COPY1:%[0-9]+]]:vgpr(s32) = COPY $vgpr0
+    ; GREEDY: [[COPY2:%[0-9]+]]:sgpr(s32) = COPY $sgpr0
+    ; GREEDY: [[COPY3:%[0-9]+]]:vgpr(s32) = COPY [[COPY2]](s32)
+    ; GREEDY: [[UV:%[0-9]+]]:vgpr(s32), [[UV1:%[0-9]+]]:vgpr(s32), [[UV2:%[0-9]+]]:vgpr(s32), [[UV3:%[0-9]+]]:vgpr(s32) = G_UNMERGE_VALUES [[COPY]](<4 x s32>)
+    ; GREEDY: [[C:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 0
+    ; GREEDY: [[ICMP:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY3]](s32), [[C]]
+    ; GREEDY: [[SELECT:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP]](s1), [[COPY1]], [[UV]]
+    ; GREEDY: [[C1:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 1
+    ; GREEDY: [[ICMP1:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY3]](s32), [[C1]]
+    ; GREEDY: [[SELECT1:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP1]](s1), [[COPY1]], [[UV1]]
+    ; GREEDY: [[C2:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 2
+    ; GREEDY: [[ICMP2:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY3]](s32), [[C2]]
+    ; GREEDY: [[SELECT2:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP2]](s1), [[COPY1]], [[UV2]]
+    ; GREEDY: [[C3:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 3
+    ; GREEDY: [[ICMP3:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY3]](s32), [[C3]]
+    ; GREEDY: [[SELECT3:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP3]](s1), [[COPY1]], [[UV3]]
+    ; GREEDY: [[BUILD_VECTOR:%[0-9]+]]:vgpr(<4 x s32>) = G_BUILD_VECTOR [[SELECT]](s32), [[SELECT1]](s32), [[SELECT2]](s32), [[SELECT3]](s32)
+    ; GREEDY: $vgpr0_vgpr1_vgpr2_vgpr3 = COPY [[BUILD_VECTOR]](<4 x s32>)
     %0:_(<4 x s32>) = COPY $vgpr0_vgpr1_vgpr2_vgpr3
     %1:_(s32) = COPY $vgpr0
     %2:_(s32) = COPY $sgpr0
     %3:_(<4 x s32>) = G_INSERT_VECTOR_ELT %0, %1, %2
     $vgpr0_vgpr1_vgpr2_vgpr3 = COPY %3
 ...
 
 ---
 name: insert_vector_elt_var_v4i32_v_v_v
 legalized: true
 tracksRegLiveness: true
 
 body: |
   bb.0:
     liveins: $vgpr0_vgpr1_vgpr2_vgpr3, $vgpr4, $vgpr5
 
-    ; CHECK-LABEL: name: insert_vector_elt_var_v4i32_v_v_v
-    ; CHECK: successors: %bb.1(0x80000000)
-    ; CHECK: liveins: $vgpr0_vgpr1_vgpr2_vgpr3, $vgpr4, $vgpr5
-    ; CHECK: [[COPY:%[0-9]+]]:vgpr(<4 x s32>) = COPY $vgpr0_vgpr1_vgpr2_vgpr3
-    ; CHECK: [[COPY1:%[0-9]+]]:vgpr(s32) = COPY $vgpr4
-    ; CHECK: [[COPY2:%[0-9]+]]:vgpr_32(s32) = COPY $vgpr5
-    ; CHECK: [[DEF:%[0-9]+]]:vgpr(<4 x s32>) = G_IMPLICIT_DEF
-    ; CHECK: [[DEF1:%[0-9]+]]:sreg_64_xexec = IMPLICIT_DEF
-    ; CHECK: [[S_MOV_B64_term:%[0-9]+]]:sreg_64_xexec = S_MOV_B64_term $exec
-    ; CHECK: .1:
-    ; CHECK: successors: %bb.2(0x40000000), %bb.1(0x40000000)
-    ; CHECK: [[PHI:%[0-9]+]]:sreg_64_xexec = PHI [[DEF1]], %bb.0, %9, %bb.1
-    ; CHECK: [[PHI1:%[0-9]+]]:vgpr(<4 x s32>) = G_PHI [[DEF]](<4 x s32>), %bb.0, %3(<4 x s32>), %bb.1
-    ; CHECK: [[V_READFIRSTLANE_B32_:%[0-9]+]]:sreg_32_xm0(s32) = V_READFIRSTLANE_B32 [[COPY2]](s32), implicit $exec
-    ; CHECK: [[V_CMP_EQ_U32_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U32_e64 [[V_READFIRSTLANE_B32_]](s32), [[COPY2]](s32), implicit $exec
-    ; CHECK: [[IVEC:%[0-9]+]]:vgpr(<4 x s32>) = G_INSERT_VECTOR_ELT [[COPY]], [[COPY1]](s32), [[V_READFIRSTLANE_B32_]](s32)
-    ; CHECK: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[V_CMP_EQ_U32_e64_]], implicit-def $exec, implicit-def $scc, implicit $exec
-    ; CHECK: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc
-    ; CHECK: S_CBRANCH_EXECNZ %bb.1, implicit $exec
-    ; CHECK: .2:
-    ; CHECK: successors: %bb.3(0x80000000)
-    ; CHECK: $exec = S_MOV_B64_term [[S_MOV_B64_term]]
-    ; CHECK: .3:
-    ; CHECK: $vgpr0_vgpr1_vgpr2_vgpr3 = COPY [[IVEC]](<4 x s32>)
+    ; FAST-LABEL: name: insert_vector_elt_var_v4i32_v_v_v
+    ; FAST: liveins: $vgpr0_vgpr1_vgpr2_vgpr3, $vgpr4, $vgpr5
+    ; FAST: [[COPY:%[0-9]+]]:vgpr(<4 x s32>) = COPY $vgpr0_vgpr1_vgpr2_vgpr3
+    ; FAST: [[COPY1:%[0-9]+]]:vgpr(s32) = COPY $vgpr4
+    ; FAST: [[COPY2:%[0-9]+]]:vgpr(s32) = COPY $vgpr5
+    ; FAST: [[UV:%[0-9]+]]:vgpr(s32), [[UV1:%[0-9]+]]:vgpr(s32), [[UV2:%[0-9]+]]:vgpr(s32), [[UV3:%[0-9]+]]:vgpr(s32) = G_UNMERGE_VALUES [[COPY]](<4 x s32>)
+    ; FAST: [[C:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 0
+    ; FAST: [[ICMP:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C]]
+    ; FAST: [[SELECT:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP]](s1), [[COPY1]], [[UV]]
+    ; FAST: [[C1:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 1
+    ; FAST: [[ICMP1:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C1]]
+    ; FAST: [[SELECT1:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP1]](s1), [[COPY1]], [[UV1]]
+    ; FAST: [[C2:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 2
+    ; FAST: [[ICMP2:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C2]]
+    ; FAST: [[SELECT2:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP2]](s1), [[COPY1]], [[UV2]]
+    ; FAST: [[C3:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 3
+    ; FAST: [[ICMP3:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C3]]
+    ; FAST: [[SELECT3:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP3]](s1), [[COPY1]], [[UV3]]
+    ; FAST: [[BUILD_VECTOR:%[0-9]+]]:vgpr(<4 x s32>) = G_BUILD_VECTOR [[SELECT]](s32), [[SELECT1]](s32), [[SELECT2]](s32), [[SELECT3]](s32)
+    ; FAST: $vgpr0_vgpr1_vgpr2_vgpr3 = COPY [[BUILD_VECTOR]](<4 x s32>)
+    ; GREEDY-LABEL: name: insert_vector_elt_var_v4i32_v_v_v
+    ; GREEDY: liveins: $vgpr0_vgpr1_vgpr2_vgpr3, $vgpr4, $vgpr5
+    ; GREEDY: [[COPY:%[0-9]+]]:vgpr(<4 x s32>) = COPY $vgpr0_vgpr1_vgpr2_vgpr3
+    ; GREEDY: [[COPY1:%[0-9]+]]:vgpr(s32) = COPY $vgpr4
+    ; GREEDY: [[COPY2:%[0-9]+]]:vgpr(s32) = COPY $vgpr5
+    ; GREEDY: [[UV:%[0-9]+]]:vgpr(s32), [[UV1:%[0-9]+]]:vgpr(s32), [[UV2:%[0-9]+]]:vgpr(s32), [[UV3:%[0-9]+]]:vgpr(s32) = G_UNMERGE_VALUES [[COPY]](<4 x s32>)
+    ; GREEDY: [[C:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 0
+    ; GREEDY: [[ICMP:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C]]
+    ; GREEDY: [[SELECT:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP]](s1), [[COPY1]], [[UV]]
+    ; GREEDY: [[C1:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 1
+    ; GREEDY: [[ICMP1:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C1]]
+    ; GREEDY: [[SELECT1:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP1]](s1), [[COPY1]], [[UV1]]
+    ; GREEDY: [[C2:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 2
+    ; GREEDY: [[ICMP2:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C2]]
+    ; GREEDY: [[SELECT2:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP2]](s1), [[COPY1]], [[UV2]]
+    ; GREEDY: [[C3:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 3
+    ; GREEDY: [[ICMP3:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C3]]
+    ; GREEDY: [[SELECT3:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP3]](s1), [[COPY1]], [[UV3]]
+    ; GREEDY: [[BUILD_VECTOR:%[0-9]+]]:vgpr(<4 x s32>) = G_BUILD_VECTOR [[SELECT]](s32), [[SELECT1]](s32), [[SELECT2]](s32), [[SELECT3]](s32)
+    ; GREEDY: $vgpr0_vgpr1_vgpr2_vgpr3 = COPY [[BUILD_VECTOR]](<4 x s32>)
     %0:_(<4 x s32>) = COPY $vgpr0_vgpr1_vgpr2_vgpr3
     %1:_(s32) = COPY $vgpr4
     %2:_(s32) = COPY $vgpr5
     %3:_(<4 x s32>) = G_INSERT_VECTOR_ELT %0, %1, %2
     $vgpr0_vgpr1_vgpr2_vgpr3 = COPY %3
 ...
 
 ---
 name: insert_vector_elt_v8s64_s_s_s
 legalized: true
 tracksRegLiveness: true
 
 body: |
   bb.0:
     liveins: $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7_sgpr8_sgpr9_sgpr10_sgpr11_sgpr12_sgpr13_sgpr14_sgpr15, $sgpr16_sgpr17, $sgpr18
 
-    ; CHECK-LABEL: name: insert_vector_elt_v8s64_s_s_s
-    ; CHECK: liveins: $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7_sgpr8_sgpr9_sgpr10_sgpr11_sgpr12_sgpr13_sgpr14_sgpr15, $sgpr16_sgpr17, $sgpr18
-    ; CHECK: [[COPY:%[0-9]+]]:sgpr(<8 x s64>) = COPY $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7_sgpr8_sgpr9_sgpr10_sgpr11_sgpr12_sgpr13_sgpr14_sgpr15
-    ; CHECK: [[COPY1:%[0-9]+]]:sgpr(s64) = COPY $sgpr16_sgpr17
-    ; CHECK: [[COPY2:%[0-9]+]]:sgpr(s32) = COPY $sgpr18
-    ; CHECK: [[IVEC:%[0-9]+]]:sgpr(<8 x s64>) = G_INSERT_VECTOR_ELT [[COPY]], [[COPY1]](s64), [[COPY2]](s32)
-    ; CHECK: $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7_sgpr8_sgpr9_sgpr10_sgpr11_sgpr12_sgpr13_sgpr14_sgpr15 = COPY [[IVEC]](<8 x s64>)
+    ; FAST-LABEL: name: insert_vector_elt_v8s64_s_s_s
+    ; FAST: liveins: $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7_sgpr8_sgpr9_sgpr10_sgpr11_sgpr12_sgpr13_sgpr14_sgpr15, $sgpr16_sgpr17, $sgpr18
+    ; FAST: [[COPY:%[0-9]+]]:sgpr(<8 x s64>) = COPY $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7_sgpr8_sgpr9_sgpr10_sgpr11_sgpr12_sgpr13_sgpr14_sgpr15
+    ; FAST: [[COPY1:%[0-9]+]]:sgpr(s64) = COPY $sgpr16_sgpr17
+    ; FAST: [[COPY2:%[0-9]+]]:sgpr(s32) = COPY $sgpr18
+    ; FAST: [[IVEC:%[0-9]+]]:sgpr(<8 x s64>) = G_INSERT_VECTOR_ELT [[COPY]], [[COPY1]](s64), [[COPY2]](s32)
+    ; FAST: $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7_sgpr8_sgpr9_sgpr10_sgpr11_sgpr12_sgpr13_sgpr14_sgpr15 = COPY [[IVEC]](<8 x s64>)
+    ; GREEDY-LABEL: name: insert_vector_elt_v8s64_s_s_s
+    ; GREEDY: liveins: $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7_sgpr8_sgpr9_sgpr10_sgpr11_sgpr12_sgpr13_sgpr14_sgpr15, $sgpr16_sgpr17, $sgpr18
+    ; GREEDY: [[COPY:%[0-9]+]]:sgpr(<8 x s64>) = COPY $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7_sgpr8_sgpr9_sgpr10_sgpr11_sgpr12_sgpr13_sgpr14_sgpr15
+    ; GREEDY: [[COPY1:%[0-9]+]]:sgpr(s64) = COPY $sgpr16_sgpr17
+    ; GREEDY: [[COPY2:%[0-9]+]]:sgpr(s32) = COPY $sgpr18
+    ; GREEDY: [[IVEC:%[0-9]+]]:sgpr(<8 x s64>) = G_INSERT_VECTOR_ELT [[COPY]], [[COPY1]](s64), [[COPY2]](s32)
+    ; GREEDY: $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7_sgpr8_sgpr9_sgpr10_sgpr11_sgpr12_sgpr13_sgpr14_sgpr15 = COPY [[IVEC]](<8 x s64>)
     %0:_(<8 x s64>) = COPY $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7_sgpr8_sgpr9_sgpr10_sgpr11_sgpr12_sgpr13_sgpr14_sgpr15
     %1:_(s64) = COPY $sgpr16_sgpr17
     %2:_(s32) = COPY $sgpr18
     %3:_(<8 x s64>) = G_INSERT_VECTOR_ELT %0, %1, %2
     $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7_sgpr8_sgpr9_sgpr10_sgpr11_sgpr12_sgpr13_sgpr14_sgpr15 = COPY %3
 ...
 
 ---
 name: insert_vector_elt_v8s64_v_s_s
 legalized: true
 tracksRegLiveness: true
 
 body: |
   bb.0:
     liveins: $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15, $sgpr16_sgpr17, $sgpr18
 
-    ; CHECK-LABEL: name: insert_vector_elt_v8s64_v_s_s
-    ; CHECK: liveins: $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15, $sgpr16_sgpr17, $sgpr18
-    ; CHECK: [[COPY:%[0-9]+]]:vgpr(<8 x s64>) = COPY $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15
-    ; CHECK: [[COPY1:%[0-9]+]]:sgpr(s64) = COPY $sgpr16_sgpr17
-    ; CHECK: [[COPY2:%[0-9]+]]:sgpr(s32) = COPY $sgpr18
-    ; CHECK: [[UV:%[0-9]+]]:sgpr(s32), [[UV1:%[0-9]+]]:sgpr(s32) = G_UNMERGE_VALUES [[COPY1]](s64)
-    ; CHECK: [[BITCAST:%[0-9]+]]:vgpr(<16 x s32>) = G_BITCAST [[COPY]](<8 x s64>)
-    ; CHECK: [[C:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 1
-    ; CHECK: [[SHL:%[0-9]+]]:sgpr(s32) = G_SHL [[COPY2]], [[C]](s32)
-    ; CHECK: [[ADD:%[0-9]+]]:sgpr(s32) = G_ADD [[SHL]], [[C]]
-    ; CHECK: [[IVEC:%[0-9]+]]:vgpr(<16 x s32>) = G_INSERT_VECTOR_ELT [[BITCAST]], [[UV]](s32), [[SHL]](s32)
-    ; CHECK: [[IVEC1:%[0-9]+]]:vgpr(<16 x s32>) = G_INSERT_VECTOR_ELT [[IVEC]], [[UV1]](s32), [[ADD]](s32)
-    ; CHECK: [[BITCAST1:%[0-9]+]]:vgpr(<8 x s64>) = G_BITCAST [[IVEC1]](<16 x s32>)
-    ; CHECK: $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15 = COPY [[BITCAST1]](<8 x s64>)
+    ; FAST-LABEL: name: insert_vector_elt_v8s64_v_s_s
+    ; FAST: liveins: $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15, $sgpr16_sgpr17, $sgpr18
+    ; FAST: [[COPY:%[0-9]+]]:vgpr(<8 x s64>) = COPY $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15
+    ; FAST: [[COPY1:%[0-9]+]]:sgpr(s64) = COPY $sgpr16_sgpr17
+    ; FAST: [[COPY2:%[0-9]+]]:sgpr(s32) = COPY $sgpr18
+    ; FAST: [[UV:%[0-9]+]]:sgpr(s32), [[UV1:%[0-9]+]]:sgpr(s32) = G_UNMERGE_VALUES [[COPY1]](s64)
+    ; FAST: [[BITCAST:%[0-9]+]]:vgpr(<16 x s32>) = G_BITCAST [[COPY]](<8 x s64>)
+    ; FAST: [[C:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 1
+    ; FAST: [[SHL:%[0-9]+]]:sgpr(s32) = G_SHL [[COPY2]], [[C]](s32)
+    ; FAST: [[ADD:%[0-9]+]]:sgpr(s32) = G_ADD [[SHL]], [[C]]
+    ; FAST: [[IVEC:%[0-9]+]]:vgpr(<16 x s32>) = G_INSERT_VECTOR_ELT [[BITCAST]], [[UV]](s32), [[SHL]](s32)
+    ; FAST: [[IVEC1:%[0-9]+]]:vgpr(<16 x s32>) = G_INSERT_VECTOR_ELT [[IVEC]], [[UV1]](s32), [[ADD]](s32)
+    ; FAST: [[BITCAST1:%[0-9]+]]:vgpr(<8 x s64>) = G_BITCAST [[IVEC1]](<16 x s32>)
+    ; FAST: $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15 = COPY [[BITCAST1]](<8 x s64>)
+    ; GREEDY-LABEL: name: insert_vector_elt_v8s64_v_s_s
+    ; GREEDY: liveins: $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15, $sgpr16_sgpr17, $sgpr18
+    ; GREEDY: [[COPY:%[0-9]+]]:vgpr(<8 x s64>) = COPY $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15
+    ; GREEDY: [[COPY1:%[0-9]+]]:sgpr(s64) = COPY $sgpr16_sgpr17
+    ; GREEDY: [[COPY2:%[0-9]+]]:sgpr(s32) = COPY $sgpr18
+    ; GREEDY: [[UV:%[0-9]+]]:sgpr(s32), [[UV1:%[0-9]+]]:sgpr(s32) = G_UNMERGE_VALUES [[COPY1]](s64)
+    ; GREEDY: [[BITCAST:%[0-9]+]]:vgpr(<16 x s32>) = G_BITCAST [[COPY]](<8 x s64>)
+    ; GREEDY: [[C:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 1
+    ; GREEDY: [[SHL:%[0-9]+]]:sgpr(s32) = G_SHL [[COPY2]], [[C]](s32)
+    ; GREEDY: [[ADD:%[0-9]+]]:sgpr(s32) = G_ADD [[SHL]], [[C]]
+    ; GREEDY: [[IVEC:%[0-9]+]]:vgpr(<16 x s32>) = G_INSERT_VECTOR_ELT [[BITCAST]], [[UV]](s32), [[SHL]](s32)
+    ; GREEDY: [[IVEC1:%[0-9]+]]:vgpr(<16 x s32>) = G_INSERT_VECTOR_ELT [[IVEC]], [[UV1]](s32), [[ADD]](s32)
+    ; GREEDY: [[BITCAST1:%[0-9]+]]:vgpr(<8 x s64>) = G_BITCAST [[IVEC1]](<16 x s32>)
+    ; GREEDY: $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15 = COPY [[BITCAST1]](<8 x s64>)
     %0:_(<8 x s64>) = COPY $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15
     %1:_(s64) = COPY $sgpr16_sgpr17
     %2:_(s32) = COPY $sgpr18
     %3:_(<8 x s64>) = G_INSERT_VECTOR_ELT %0, %1, %2
     $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15 = COPY %3
 ...
 
 ---
 name: insert_vector_elt_v8s64_s_v_s
 legalized: true
 tracksRegLiveness: true
 
 body: |
   bb.0:
     liveins: $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7_sgpr8_sgpr9_sgpr10_sgpr11_sgpr12_sgpr13_sgpr14_sgpr15, $vgpr0_vgpr1, $sgpr16
 
-    ; CHECK-LABEL: name: insert_vector_elt_v8s64_s_v_s
-    ; CHECK: liveins: $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7_sgpr8_sgpr9_sgpr10_sgpr11_sgpr12_sgpr13_sgpr14_sgpr15, $vgpr0_vgpr1, $sgpr16
-    ; CHECK: [[COPY:%[0-9]+]]:sgpr(<8 x s64>) = COPY $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7_sgpr8_sgpr9_sgpr10_sgpr11_sgpr12_sgpr13_sgpr14_sgpr15
-    ; CHECK: [[COPY1:%[0-9]+]]:vgpr(s64) = COPY $vgpr0_vgpr1
-    ; CHECK: [[COPY2:%[0-9]+]]:sgpr(s32) = COPY $sgpr16
-    ; CHECK: [[COPY3:%[0-9]+]]:vgpr(<8 x s64>) = COPY [[COPY]](<8 x s64>)
-    ; CHECK: [[UV:%[0-9]+]]:vgpr(s32), [[UV1:%[0-9]+]]:vgpr(s32) = G_UNMERGE_VALUES [[COPY1]](s64)
-    ; CHECK: [[BITCAST:%[0-9]+]]:vgpr(<16 x s32>) = G_BITCAST [[COPY3]](<8 x s64>)
-    ; CHECK: [[C:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 1
-    ; CHECK: [[SHL:%[0-9]+]]:sgpr(s32) = G_SHL [[COPY2]], [[C]](s32)
-    ; CHECK: [[ADD:%[0-9]+]]:sgpr(s32) = G_ADD [[SHL]], [[C]]
-    ; CHECK: [[IVEC:%[0-9]+]]:vgpr(<16 x s32>) = G_INSERT_VECTOR_ELT [[BITCAST]], [[UV]](s32), [[SHL]](s32)
-    ; CHECK: [[IVEC1:%[0-9]+]]:vgpr(<16 x s32>) = G_INSERT_VECTOR_ELT [[IVEC]], [[UV1]](s32), [[ADD]](s32)
-    ; CHECK: [[BITCAST1:%[0-9]+]]:vgpr(<8 x s64>) = G_BITCAST [[IVEC1]](<16 x s32>)
-    ; CHECK: $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15 = COPY [[BITCAST1]](<8 x s64>)
+    ; FAST-LABEL: name: insert_vector_elt_v8s64_s_v_s
+    ; FAST: liveins: $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7_sgpr8_sgpr9_sgpr10_sgpr11_sgpr12_sgpr13_sgpr14_sgpr15, $vgpr0_vgpr1, $sgpr16
+    ; FAST: [[COPY:%[0-9]+]]:sgpr(<8 x s64>) = COPY $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7_sgpr8_sgpr9_sgpr10_sgpr11_sgpr12_sgpr13_sgpr14_sgpr15
+    ; FAST: [[COPY1:%[0-9]+]]:vgpr(s64) = COPY $vgpr0_vgpr1
+    ; FAST: [[COPY2:%[0-9]+]]:sgpr(s32) = COPY $sgpr16
+    ; FAST: [[COPY3:%[0-9]+]]:vgpr(<8 x s64>) = COPY [[COPY]](<8 x s64>)
+    ; FAST: [[UV:%[0-9]+]]:vgpr(s32), [[UV1:%[0-9]+]]:vgpr(s32) = G_UNMERGE_VALUES [[COPY1]](s64)
+    ; FAST: [[BITCAST:%[0-9]+]]:vgpr(<16 x s32>) = G_BITCAST [[COPY3]](<8 x s64>)
+    ; FAST: [[C:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 1
+    ; FAST: [[SHL:%[0-9]+]]:sgpr(s32) = G_SHL [[COPY2]], [[C]](s32)
+    ; FAST: [[ADD:%[0-9]+]]:sgpr(s32) = G_ADD [[SHL]], [[C]]
+    ; FAST: [[IVEC:%[0-9]+]]:vgpr(<16 x s32>) = G_INSERT_VECTOR_ELT [[BITCAST]], [[UV]](s32), [[SHL]](s32)
+    ; FAST: [[IVEC1:%[0-9]+]]:vgpr(<16 x s32>) = G_INSERT_VECTOR_ELT [[IVEC]], [[UV1]](s32), [[ADD]](s32)
+    ; FAST: [[BITCAST1:%[0-9]+]]:vgpr(<8 x s64>) = G_BITCAST [[IVEC1]](<16 x s32>)
+    ; FAST: $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15 = COPY [[BITCAST1]](<8 x s64>)
+    ; GREEDY-LABEL: name: insert_vector_elt_v8s64_s_v_s
+    ; GREEDY: liveins: $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7_sgpr8_sgpr9_sgpr10_sgpr11_sgpr12_sgpr13_sgpr14_sgpr15, $vgpr0_vgpr1, $sgpr16
+    ; GREEDY: [[COPY:%[0-9]+]]:sgpr(<8 x s64>) = COPY $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7_sgpr8_sgpr9_sgpr10_sgpr11_sgpr12_sgpr13_sgpr14_sgpr15
+    ; GREEDY: [[COPY1:%[0-9]+]]:vgpr(s64) = COPY $vgpr0_vgpr1
+    ; GREEDY: [[COPY2:%[0-9]+]]:sgpr(s32) = COPY $sgpr16
+    ; GREEDY: [[COPY3:%[0-9]+]]:vgpr(<8 x s64>) = COPY [[COPY]](<8 x s64>)
+    ; GREEDY: [[UV:%[0-9]+]]:vgpr(s32), [[UV1:%[0-9]+]]:vgpr(s32) = G_UNMERGE_VALUES [[COPY1]](s64)
+    ; GREEDY: [[BITCAST:%[0-9]+]]:vgpr(<16 x s32>) = G_BITCAST [[COPY3]](<8 x s64>)
+    ; GREEDY: [[C:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 1
+    ; GREEDY: [[SHL:%[0-9]+]]:sgpr(s32) = G_SHL [[COPY2]], [[C]](s32)
+    ; GREEDY: [[ADD:%[0-9]+]]:sgpr(s32) = G_ADD [[SHL]], [[C]]
+    ; GREEDY: [[IVEC:%[0-9]+]]:vgpr(<16 x s32>) = G_INSERT_VECTOR_ELT [[BITCAST]], [[UV]](s32), [[SHL]](s32)
+    ; GREEDY: [[IVEC1:%[0-9]+]]:vgpr(<16 x s32>) = G_INSERT_VECTOR_ELT [[IVEC]], [[UV1]](s32), [[ADD]](s32)
+    ; GREEDY: [[BITCAST1:%[0-9]+]]:vgpr(<8 x s64>) = G_BITCAST [[IVEC1]](<16 x s32>)
+    ; GREEDY: $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15 = COPY [[BITCAST1]](<8 x s64>)
     %0:_(<8 x s64>) = COPY $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7_sgpr8_sgpr9_sgpr10_sgpr11_sgpr12_sgpr13_sgpr14_sgpr15
     %1:_(s64) = COPY $vgpr0_vgpr1
     %2:_(s32) = COPY $sgpr16
     %3:_(<8 x s64>) = G_INSERT_VECTOR_ELT %0, %1, %2
     $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15 = COPY %3
 ...
 
 ---
 name: insert_vector_elt_v8s64_s_s_v
 legalized: true
 tracksRegLiveness: true
 
 body: |
   bb.0:
     liveins: $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7_sgpr8_sgpr9_sgpr10_sgpr11_sgpr12_sgpr13_sgpr14_sgpr15, $sgpr16_sgpr17, $vgpr0
 
-    ; CHECK-LABEL: name: insert_vector_elt_v8s64_s_s_v
-    ; CHECK: successors: %bb.1(0x80000000)
-    ; CHECK: liveins: $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7_sgpr8_sgpr9_sgpr10_sgpr11_sgpr12_sgpr13_sgpr14_sgpr15, $sgpr16_sgpr17, $vgpr0
-    ; CHECK: [[COPY:%[0-9]+]]:sgpr(<8 x s64>) = COPY $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7_sgpr8_sgpr9_sgpr10_sgpr11_sgpr12_sgpr13_sgpr14_sgpr15
-    ; CHECK: [[COPY1:%[0-9]+]]:sgpr(s64) = COPY $sgpr16_sgpr17
-    ; CHECK: [[COPY2:%[0-9]+]]:vgpr_32(s32) = COPY $vgpr0
-    ; CHECK: [[COPY3:%[0-9]+]]:vgpr(<8 x s64>) = COPY [[COPY]](<8 x s64>)
-    ; CHECK: [[UV:%[0-9]+]]:sgpr(s32), [[UV1:%[0-9]+]]:sgpr(s32) = G_UNMERGE_VALUES [[COPY1]](s64)
-    ; CHECK: [[BITCAST:%[0-9]+]]:vgpr(<16 x s32>) = G_BITCAST [[COPY3]](<8 x s64>)
-    ; CHECK: [[C:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 1
-    ; CHECK: [[DEF:%[0-9]+]]:sgpr(s32) = G_IMPLICIT_DEF
-    ; CHECK: [[DEF1:%[0-9]+]]:sgpr(s32) = G_IMPLICIT_DEF
-    ; CHECK: [[DEF2:%[0-9]+]]:vgpr(<16 x s32>) = G_IMPLICIT_DEF
-    ; CHECK: [[DEF3:%[0-9]+]]:vgpr(<16 x s32>) = G_IMPLICIT_DEF
-    ; CHECK: [[DEF4:%[0-9]+]]:sreg_64_xexec = IMPLICIT_DEF
-    ; CHECK: [[S_MOV_B64_term:%[0-9]+]]:sreg_64_xexec = S_MOV_B64_term $exec
-    ; CHECK: .1:
-    ; CHECK: successors: %bb.2(0x40000000), %bb.1(0x40000000)
-    ; CHECK: [[PHI:%[0-9]+]]:sreg_64_xexec = PHI [[DEF4]], %bb.0, %24, %bb.1
-    ; CHECK: [[PHI1:%[0-9]+]]:sgpr(s32) = G_PHI [[DEF]](s32), %bb.0, %9(s32), %bb.1
-    ; CHECK: [[PHI2:%[0-9]+]]:sgpr(s32) = G_PHI [[DEF1]](s32), %bb.0, %10(s32), %bb.1
-    ; CHECK: [[PHI3:%[0-9]+]]:vgpr(<16 x s32>) = G_PHI [[DEF2]](<16 x s32>), %bb.0, %11(<16 x s32>), %bb.1
-    ; CHECK: [[PHI4:%[0-9]+]]:vgpr(<16 x s32>) = G_PHI [[DEF3]](<16 x s32>), %bb.0, %12(<16 x s32>), %bb.1
-    ; CHECK: [[V_READFIRSTLANE_B32_:%[0-9]+]]:sreg_32_xm0(s32) = V_READFIRSTLANE_B32 [[COPY2]](s32), implicit $exec
-    ; CHECK: [[V_CMP_EQ_U32_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U32_e64 [[V_READFIRSTLANE_B32_]](s32), [[COPY2]](s32), implicit $exec
-    ; CHECK: [[SHL:%[0-9]+]]:sgpr(s32) = G_SHL [[V_READFIRSTLANE_B32_]], [[C]](s32)
-    ; CHECK: [[ADD:%[0-9]+]]:sgpr(s32) = G_ADD [[SHL]], [[C]]
-    ; CHECK: [[IVEC:%[0-9]+]]:vgpr(<16 x s32>) = G_INSERT_VECTOR_ELT [[BITCAST]], [[UV]](s32), [[SHL]](s32)
-    ; CHECK: [[IVEC1:%[0-9]+]]:vgpr(<16 x s32>) = G_INSERT_VECTOR_ELT [[IVEC]], [[UV1]](s32), [[ADD]](s32)
-    ; CHECK: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[V_CMP_EQ_U32_e64_]], implicit-def $exec, implicit-def $scc, implicit $exec
-    ; CHECK: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc
-    ; CHECK: S_CBRANCH_EXECNZ %bb.1, implicit $exec
-    ; CHECK: .2:
-    ; CHECK: successors: %bb.3(0x80000000)
-    ; CHECK: $exec = S_MOV_B64_term [[S_MOV_B64_term]]
-    ; CHECK: .3:
-    ; CHECK: [[BITCAST1:%[0-9]+]]:vgpr(<8 x s64>) = G_BITCAST [[IVEC1]](<16 x s32>)
-    ; CHECK: $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15 = COPY [[BITCAST1]](<8 x s64>)
+    ; FAST-LABEL: name: insert_vector_elt_v8s64_s_s_v
+    ; FAST: liveins: $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7_sgpr8_sgpr9_sgpr10_sgpr11_sgpr12_sgpr13_sgpr14_sgpr15, $sgpr16_sgpr17, $vgpr0
+    ; FAST: [[COPY:%[0-9]+]]:sgpr(<8 x s64>) = COPY $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7_sgpr8_sgpr9_sgpr10_sgpr11_sgpr12_sgpr13_sgpr14_sgpr15
+    ; FAST: [[COPY1:%[0-9]+]]:sgpr(s64) = COPY $sgpr16_sgpr17
+    ; FAST: [[COPY2:%[0-9]+]]:vgpr(s32) = COPY $vgpr0
+    ; FAST: [[COPY3:%[0-9]+]]:vgpr(<8 x s64>) = COPY [[COPY]](<8 x s64>)
+    ; FAST: [[UV:%[0-9]+]]:vgpr(s32), [[UV1:%[0-9]+]]:vgpr(s32) = G_UNMERGE_VALUES [[COPY1]](s64)
+    ; FAST: [[UV2:%[0-9]+]]:vgpr(s32), [[UV3:%[0-9]+]]:vgpr(s32), [[UV4:%[0-9]+]]:vgpr(s32), [[UV5:%[0-9]+]]:vgpr(s32), [[UV6:%[0-9]+]]:vgpr(s32), [[UV7:%[0-9]+]]:vgpr(s32), [[UV8:%[0-9]+]]:vgpr(s32), [[UV9:%[0-9]+]]:vgpr(s32), [[UV10:%[0-9]+]]:vgpr(s32), [[UV11:%[0-9]+]]:vgpr(s32), [[UV12:%[0-9]+]]:vgpr(s32), [[UV13:%[0-9]+]]:vgpr(s32), [[UV14:%[0-9]+]]:vgpr(s32), [[UV15:%[0-9]+]]:vgpr(s32), [[UV16:%[0-9]+]]:vgpr(s32), [[UV17:%[0-9]+]]:vgpr(s32) = G_UNMERGE_VALUES [[COPY3]](<8 x s64>)
+    ; FAST: [[C:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 0
+    ; FAST: [[ICMP:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C]]
+    ; FAST: [[SELECT:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP]](s1), [[UV]], [[UV2]]
+    ; FAST: [[SELECT1:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP]](s1), [[UV1]], [[UV3]]
+    ; FAST: [[C1:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 1
+    ; FAST: [[ICMP1:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C1]]
+    ; FAST: [[SELECT2:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP1]](s1), [[UV]], [[UV4]]
+    ; FAST: [[SELECT3:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP1]](s1), [[UV1]], [[UV5]]
+    ; FAST: [[C2:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 2
+    ; FAST: [[ICMP2:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C2]]
+    ; FAST: [[SELECT4:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP2]](s1), [[UV]], [[UV6]]
+    ; FAST: [[SELECT5:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP2]](s1), [[UV1]], [[UV7]]
+    ; FAST: [[C3:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 3
+    ; FAST: [[ICMP3:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C3]]
+    ; FAST: [[SELECT6:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP3]](s1), [[UV]], [[UV8]]
+    ; FAST: [[SELECT7:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP3]](s1), [[UV1]], [[UV9]]
+    ; FAST: [[C4:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 4
+    ; FAST: [[ICMP4:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C4]]
+    ; FAST: [[SELECT8:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP4]](s1), [[UV]], [[UV10]]
+    ; FAST: [[SELECT9:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP4]](s1), [[UV1]], [[UV11]]
+    ; FAST: [[C5:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 5
+    ; FAST: [[ICMP5:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C5]]
+    ; FAST: [[SELECT10:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP5]](s1), [[UV]], [[UV12]]
+    ; FAST: [[SELECT11:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP5]](s1), [[UV1]], [[UV13]]
+    ; FAST: [[C6:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 6
+    ; FAST: [[ICMP6:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C6]]
+    ; FAST: [[SELECT12:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP6]](s1), [[UV]], [[UV14]]
+    ; FAST: [[SELECT13:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP6]](s1), [[UV1]], [[UV15]]
+    ; FAST: [[C7:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 7
+    ; FAST: [[ICMP7:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C7]]
+    ; FAST: [[SELECT14:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP7]](s1), [[UV]], [[UV16]]
+    ; FAST: [[SELECT15:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP7]](s1), [[UV1]], [[UV17]]
+    ; FAST: [[BUILD_VECTOR:%[0-9]+]]:vgpr(<16 x s32>) = G_BUILD_VECTOR [[SELECT]](s32), [[SELECT1]](s32), [[SELECT2]](s32), [[SELECT3]](s32), [[SELECT4]](s32), [[SELECT5]](s32), [[SELECT6]](s32), [[SELECT7]](s32), [[SELECT8]](s32), [[SELECT9]](s32), [[SELECT10]](s32), [[SELECT11]](s32), [[SELECT12]](s32), [[SELECT13]](s32), [[SELECT14]](s32), [[SELECT15]](s32)
+    ; FAST: [[BITCAST:%[0-9]+]]:vgpr(<8 x s64>) = G_BITCAST [[BUILD_VECTOR]](<16 x s32>)
+    ; FAST: $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15 = COPY [[BITCAST]](<8 x s64>)
+    ; GREEDY-LABEL: name: insert_vector_elt_v8s64_s_s_v
+    ; GREEDY: liveins: $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7_sgpr8_sgpr9_sgpr10_sgpr11_sgpr12_sgpr13_sgpr14_sgpr15, $sgpr16_sgpr17, $vgpr0
+    ; GREEDY: [[COPY:%[0-9]+]]:sgpr(<8 x s64>) = COPY $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7_sgpr8_sgpr9_sgpr10_sgpr11_sgpr12_sgpr13_sgpr14_sgpr15
+    ; GREEDY: [[COPY1:%[0-9]+]]:sgpr(s64) = COPY $sgpr16_sgpr17
+    ; GREEDY: [[COPY2:%[0-9]+]]:vgpr(s32) = COPY $vgpr0
+    ; GREEDY: [[COPY3:%[0-9]+]]:vgpr(<8 x s64>) = COPY [[COPY]](<8 x s64>)
+    ; GREEDY: [[UV:%[0-9]+]]:vgpr(s32), [[UV1:%[0-9]+]]:vgpr(s32) = G_UNMERGE_VALUES [[COPY1]](s64)
+    ; GREEDY: [[UV2:%[0-9]+]]:vgpr(s32), [[UV3:%[0-9]+]]:vgpr(s32), [[UV4:%[0-9]+]]:vgpr(s32), [[UV5:%[0-9]+]]:vgpr(s32), [[UV6:%[0-9]+]]:vgpr(s32), [[UV7:%[0-9]+]]:vgpr(s32), [[UV8:%[0-9]+]]:vgpr(s32), [[UV9:%[0-9]+]]:vgpr(s32), [[UV10:%[0-9]+]]:vgpr(s32), [[UV11:%[0-9]+]]:vgpr(s32), [[UV12:%[0-9]+]]:vgpr(s32), [[UV13:%[0-9]+]]:vgpr(s32), [[UV14:%[0-9]+]]:vgpr(s32), [[UV15:%[0-9]+]]:vgpr(s32), [[UV16:%[0-9]+]]:vgpr(s32), [[UV17:%[0-9]+]]:vgpr(s32) = G_UNMERGE_VALUES [[COPY3]](<8 x s64>)
+    ; GREEDY: [[C:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 0
+    ; GREEDY: [[ICMP:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C]]
+    ; GREEDY: [[SELECT:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP]](s1), [[UV]], [[UV2]]
+    ; GREEDY: [[SELECT1:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP]](s1), [[UV1]], [[UV3]]
+    ; GREEDY: [[C1:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 1
+    ; GREEDY: [[ICMP1:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C1]]
+    ; GREEDY: [[SELECT2:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP1]](s1), [[UV]], [[UV4]]
+    ; GREEDY: [[SELECT3:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP1]](s1), [[UV1]], [[UV5]]
+    ; GREEDY: [[C2:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 2
+    ; GREEDY: [[ICMP2:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C2]]
+    ; GREEDY: [[SELECT4:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP2]](s1), [[UV]], [[UV6]]
+    ; GREEDY: [[SELECT5:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP2]](s1), [[UV1]], [[UV7]]
+    ; GREEDY: [[C3:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 3
+    ; GREEDY: [[ICMP3:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C3]]
+    ; GREEDY: [[SELECT6:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP3]](s1), [[UV]], [[UV8]]
+    ; GREEDY: [[SELECT7:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP3]](s1), [[UV1]], [[UV9]]
+    ; GREEDY: [[C4:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 4
+    ; GREEDY: [[ICMP4:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C4]]
+    ; GREEDY: [[SELECT8:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP4]](s1), [[UV]], [[UV10]]
+    ; GREEDY: [[SELECT9:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP4]](s1), [[UV1]], [[UV11]]
+    ; GREEDY: [[C5:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 5
+    ; GREEDY: [[ICMP5:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C5]]
+    ; GREEDY: [[SELECT10:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP5]](s1), [[UV]], [[UV12]]
+    ; GREEDY: [[SELECT11:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP5]](s1), [[UV1]], [[UV13]]
+    ; GREEDY: [[C6:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 6
+    ; GREEDY: [[ICMP6:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C6]]
+    ; GREEDY: [[SELECT12:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP6]](s1), [[UV]], [[UV14]]
+    ; GREEDY: [[SELECT13:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP6]](s1), [[UV1]], [[UV15]]
+    ; GREEDY: [[C7:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 7
+    ; GREEDY: [[ICMP7:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C7]]
+    ; GREEDY: [[SELECT14:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP7]](s1), [[UV]], [[UV16]]
+    ; GREEDY: [[SELECT15:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP7]](s1), [[UV1]], [[UV17]]
+    ; GREEDY: [[BUILD_VECTOR:%[0-9]+]]:vgpr(<16 x s32>) = G_BUILD_VECTOR [[SELECT]](s32), [[SELECT1]](s32), [[SELECT2]](s32), [[SELECT3]](s32), [[SELECT4]](s32), [[SELECT5]](s32), [[SELECT6]](s32), [[SELECT7]](s32), [[SELECT8]](s32), [[SELECT9]](s32), [[SELECT10]](s32), [[SELECT11]](s32), [[SELECT12]](s32), [[SELECT13]](s32), [[SELECT14]](s32), [[SELECT15]](s32)
+    ; GREEDY: [[BITCAST:%[0-9]+]]:vgpr(<8 x s64>) = G_BITCAST [[BUILD_VECTOR]](<16 x s32>)
+    ; GREEDY: $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15 = COPY [[BITCAST]](<8 x s64>)
     %0:_(<8 x s64>) = COPY $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7_sgpr8_sgpr9_sgpr10_sgpr11_sgpr12_sgpr13_sgpr14_sgpr15
     %1:_(s64) = COPY $sgpr16_sgpr17
     %2:_(s32) = COPY $vgpr0
     %3:_(<8 x s64>) = G_INSERT_VECTOR_ELT %0, %1, %2
     $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15 = COPY %3
 ...
 
 ---
 name: insert_vector_elt_v8s64_s_v_v
 legalized: true
 tracksRegLiveness: true
 
 body: |
   bb.0:
     liveins: $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7_sgpr8_sgpr9_sgpr10_sgpr11_sgpr12_sgpr13_sgpr14_sgpr15, $vgpr0_vgpr1, $vgpr2
 
-    ; CHECK-LABEL: name: insert_vector_elt_v8s64_s_v_v
-    ; CHECK: successors: %bb.1(0x80000000)
-    ; CHECK: liveins: $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7_sgpr8_sgpr9_sgpr10_sgpr11_sgpr12_sgpr13_sgpr14_sgpr15, $vgpr0_vgpr1, $vgpr2
-    ; CHECK: [[COPY:%[0-9]+]]:sgpr(<8 x s64>) = COPY $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7_sgpr8_sgpr9_sgpr10_sgpr11_sgpr12_sgpr13_sgpr14_sgpr15
-    ; CHECK: [[COPY1:%[0-9]+]]:vgpr(s64) = COPY $vgpr0_vgpr1
-    ; CHECK: [[COPY2:%[0-9]+]]:vgpr_32(s32) = COPY $vgpr2
-    ; CHECK: [[COPY3:%[0-9]+]]:vgpr(<8 x s64>) = COPY [[COPY]](<8 x s64>)
-    ; CHECK: [[UV:%[0-9]+]]:vgpr(s32), [[UV1:%[0-9]+]]:vgpr(s32) = G_UNMERGE_VALUES [[COPY1]](s64)
-    ; CHECK: [[BITCAST:%[0-9]+]]:vgpr(<16 x s32>) = G_BITCAST [[COPY3]](<8 x s64>)
-    ; CHECK: [[C:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 1
-    ; CHECK: [[DEF:%[0-9]+]]:sgpr(s32) = G_IMPLICIT_DEF
-    ; CHECK: [[DEF1:%[0-9]+]]:sgpr(s32) = G_IMPLICIT_DEF
-    ; CHECK: [[DEF2:%[0-9]+]]:vgpr(<16 x s32>) = G_IMPLICIT_DEF
-    ; CHECK: [[DEF3:%[0-9]+]]:vgpr(<16 x s32>) = G_IMPLICIT_DEF
-    ; CHECK: [[DEF4:%[0-9]+]]:sreg_64_xexec = IMPLICIT_DEF
-    ; CHECK: [[S_MOV_B64_term:%[0-9]+]]:sreg_64_xexec = S_MOV_B64_term $exec
-    ; CHECK: .1:
-    ; CHECK: successors: %bb.2(0x40000000), %bb.1(0x40000000)
-    ; CHECK: [[PHI:%[0-9]+]]:sreg_64_xexec = PHI [[DEF4]], %bb.0, %24, %bb.1
-    ; CHECK: [[PHI1:%[0-9]+]]:sgpr(s32) = G_PHI [[DEF]](s32), %bb.0, %9(s32), %bb.1
-    ; CHECK: [[PHI2:%[0-9]+]]:sgpr(s32) = G_PHI [[DEF1]](s32), %bb.0, %10(s32), %bb.1
-    ; CHECK: [[PHI3:%[0-9]+]]:vgpr(<16 x s32>) = G_PHI [[DEF2]](<16 x s32>), %bb.0, %11(<16 x s32>), %bb.1
-    ; CHECK: [[PHI4:%[0-9]+]]:vgpr(<16 x s32>) = G_PHI [[DEF3]](<16 x s32>), %bb.0, %12(<16 x s32>), %bb.1
-    ; CHECK: [[V_READFIRSTLANE_B32_:%[0-9]+]]:sreg_32_xm0(s32) = V_READFIRSTLANE_B32 [[COPY2]](s32), implicit $exec
-    ; CHECK: [[V_CMP_EQ_U32_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U32_e64 [[V_READFIRSTLANE_B32_]](s32), [[COPY2]](s32), implicit $exec
-    ; CHECK: [[SHL:%[0-9]+]]:sgpr(s32) = G_SHL [[V_READFIRSTLANE_B32_]], [[C]](s32)
-    ; CHECK: [[ADD:%[0-9]+]]:sgpr(s32) = G_ADD [[SHL]], [[C]]
-    ; CHECK: [[IVEC:%[0-9]+]]:vgpr(<16 x s32>) = G_INSERT_VECTOR_ELT [[BITCAST]], [[UV]](s32), [[SHL]](s32)
-    ; CHECK: [[IVEC1:%[0-9]+]]:vgpr(<16 x s32>) = G_INSERT_VECTOR_ELT [[IVEC]], [[UV1]](s32), [[ADD]](s32)
-    ; CHECK: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[V_CMP_EQ_U32_e64_]], implicit-def $exec, implicit-def $scc, implicit $exec
-    ; CHECK: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc
-    ; CHECK: S_CBRANCH_EXECNZ %bb.1, implicit $exec
-    ; CHECK: .2:
-    ; CHECK: successors: %bb.3(0x80000000)
-    ; CHECK: $exec = S_MOV_B64_term [[S_MOV_B64_term]]
-    ; CHECK: .3:
-    ; CHECK: [[BITCAST1:%[0-9]+]]:vgpr(<8 x s64>) = G_BITCAST [[IVEC1]](<16 x s32>)
-    ; CHECK: $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15 = COPY [[BITCAST1]](<8 x s64>)
+    ; FAST-LABEL: name: insert_vector_elt_v8s64_s_v_v
+    ; FAST: liveins: $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7_sgpr8_sgpr9_sgpr10_sgpr11_sgpr12_sgpr13_sgpr14_sgpr15, $vgpr0_vgpr1, $vgpr2
+    ; FAST: [[COPY:%[0-9]+]]:sgpr(<8 x s64>) = COPY $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7_sgpr8_sgpr9_sgpr10_sgpr11_sgpr12_sgpr13_sgpr14_sgpr15
+    ; FAST: [[COPY1:%[0-9]+]]:vgpr(s64) = COPY $vgpr0_vgpr1
+    ; FAST: [[COPY2:%[0-9]+]]:vgpr(s32) = COPY $vgpr2
+    ; FAST: [[COPY3:%[0-9]+]]:vgpr(<8 x s64>) = COPY [[COPY]](<8 x s64>)
+    ; FAST: [[UV:%[0-9]+]]:vgpr(s32), [[UV1:%[0-9]+]]:vgpr(s32) = G_UNMERGE_VALUES [[COPY1]](s64)
+    ; FAST: [[UV2:%[0-9]+]]:vgpr(s32), [[UV3:%[0-9]+]]:vgpr(s32), [[UV4:%[0-9]+]]:vgpr(s32), [[UV5:%[0-9]+]]:vgpr(s32), [[UV6:%[0-9]+]]:vgpr(s32), [[UV7:%[0-9]+]]:vgpr(s32), [[UV8:%[0-9]+]]:vgpr(s32), [[UV9:%[0-9]+]]:vgpr(s32), [[UV10:%[0-9]+]]:vgpr(s32), [[UV11:%[0-9]+]]:vgpr(s32), [[UV12:%[0-9]+]]:vgpr(s32), [[UV13:%[0-9]+]]:vgpr(s32), [[UV14:%[0-9]+]]:vgpr(s32), [[UV15:%[0-9]+]]:vgpr(s32), [[UV16:%[0-9]+]]:vgpr(s32), [[UV17:%[0-9]+]]:vgpr(s32) = G_UNMERGE_VALUES [[COPY3]](<8 x s64>)
+    ; FAST: [[C:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 0
+    ; FAST: [[ICMP:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C]]
+    ; FAST: [[SELECT:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP]](s1), [[UV]], [[UV2]]
+    ; FAST: [[SELECT1:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP]](s1), [[UV1]], [[UV3]]
+    ; FAST: [[C1:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 1
+    ; FAST: [[ICMP1:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C1]]
+    ; FAST: [[SELECT2:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP1]](s1), [[UV]], [[UV4]]
+    ; FAST: [[SELECT3:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP1]](s1), [[UV1]], [[UV5]]
+    ; FAST: [[C2:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 2
+    ; FAST: [[ICMP2:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C2]]
+    ; FAST: [[SELECT4:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP2]](s1), [[UV]], [[UV6]]
+    ; FAST: [[SELECT5:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP2]](s1), [[UV1]], [[UV7]]
+    ; FAST: [[C3:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 3
+    ; FAST: [[ICMP3:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C3]]
+    ; FAST: [[SELECT6:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP3]](s1), [[UV]], [[UV8]]
+    ; FAST: [[SELECT7:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP3]](s1), [[UV1]], [[UV9]]
+    ; FAST: [[C4:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 4
+    ; FAST: [[ICMP4:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C4]]
+    ; FAST: [[SELECT8:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP4]](s1), [[UV]], [[UV10]]
+    ; FAST: [[SELECT9:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP4]](s1), [[UV1]], [[UV11]]
+    ; FAST: [[C5:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 5
+    ; FAST: [[ICMP5:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C5]]
+    ; FAST: [[SELECT10:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP5]](s1), [[UV]], [[UV12]]
+    ; FAST: [[SELECT11:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP5]](s1), [[UV1]], [[UV13]]
+    ; FAST: [[C6:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 6
+    ; FAST: [[ICMP6:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C6]]
+    ; FAST: [[SELECT12:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP6]](s1), [[UV]], [[UV14]]
+    ; FAST: [[SELECT13:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP6]](s1), [[UV1]], [[UV15]]
+    ; FAST: [[C7:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 7
+    ; FAST: [[ICMP7:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C7]]
+    ; FAST: [[SELECT14:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP7]](s1), [[UV]], [[UV16]]
+    ; FAST: [[SELECT15:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP7]](s1), [[UV1]], [[UV17]]
+    ; FAST: [[BUILD_VECTOR:%[0-9]+]]:vgpr(<16 x s32>) = G_BUILD_VECTOR [[SELECT]](s32), [[SELECT1]](s32), [[SELECT2]](s32), [[SELECT3]](s32), [[SELECT4]](s32), [[SELECT5]](s32), [[SELECT6]](s32), [[SELECT7]](s32), [[SELECT8]](s32), [[SELECT9]](s32), [[SELECT10]](s32), [[SELECT11]](s32), [[SELECT12]](s32), [[SELECT13]](s32), [[SELECT14]](s32), [[SELECT15]](s32)
+    ; FAST: [[BITCAST:%[0-9]+]]:vgpr(<8 x s64>) = G_BITCAST [[BUILD_VECTOR]](<16 x s32>)
+    ; FAST: $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15 = COPY [[BITCAST]](<8 x s64>)
+    ; GREEDY-LABEL: name: insert_vector_elt_v8s64_s_v_v
+    ; GREEDY: liveins: $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7_sgpr8_sgpr9_sgpr10_sgpr11_sgpr12_sgpr13_sgpr14_sgpr15, $vgpr0_vgpr1, $vgpr2
+    ; GREEDY: [[COPY:%[0-9]+]]:sgpr(<8 x s64>) = COPY $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7_sgpr8_sgpr9_sgpr10_sgpr11_sgpr12_sgpr13_sgpr14_sgpr15
+    ; GREEDY: [[COPY1:%[0-9]+]]:vgpr(s64) = COPY $vgpr0_vgpr1
+    ; GREEDY: [[COPY2:%[0-9]+]]:vgpr(s32) = COPY $vgpr2
+    ; GREEDY: [[COPY3:%[0-9]+]]:vgpr(<8 x s64>) = COPY [[COPY]](<8 x s64>)
+    ; GREEDY: [[UV:%[0-9]+]]:vgpr(s32), [[UV1:%[0-9]+]]:vgpr(s32) = G_UNMERGE_VALUES [[COPY1]](s64)
+    ; GREEDY: [[UV2:%[0-9]+]]:vgpr(s32), [[UV3:%[0-9]+]]:vgpr(s32), [[UV4:%[0-9]+]]:vgpr(s32), [[UV5:%[0-9]+]]:vgpr(s32), [[UV6:%[0-9]+]]:vgpr(s32), [[UV7:%[0-9]+]]:vgpr(s32), [[UV8:%[0-9]+]]:vgpr(s32), [[UV9:%[0-9]+]]:vgpr(s32), [[UV10:%[0-9]+]]:vgpr(s32), [[UV11:%[0-9]+]]:vgpr(s32), [[UV12:%[0-9]+]]:vgpr(s32), [[UV13:%[0-9]+]]:vgpr(s32), [[UV14:%[0-9]+]]:vgpr(s32), [[UV15:%[0-9]+]]:vgpr(s32), [[UV16:%[0-9]+]]:vgpr(s32), [[UV17:%[0-9]+]]:vgpr(s32) = G_UNMERGE_VALUES [[COPY3]](<8 x s64>)
+    ; GREEDY: [[C:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 0
+    ; GREEDY: [[ICMP:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C]]
+    ; GREEDY: [[SELECT:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP]](s1), [[UV]], [[UV2]]
+    ; GREEDY: [[SELECT1:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP]](s1), [[UV1]], [[UV3]]
+    ; GREEDY: [[C1:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 1
+    ; GREEDY: [[ICMP1:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C1]]
+    ; GREEDY: [[SELECT2:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP1]](s1), [[UV]], [[UV4]]
+    ; GREEDY: [[SELECT3:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP1]](s1), [[UV1]], [[UV5]]
+    ; GREEDY: [[C2:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 2
+    ; GREEDY: [[ICMP2:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C2]]
+    ; GREEDY: [[SELECT4:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP2]](s1), [[UV]], [[UV6]]
+    ; GREEDY: [[SELECT5:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP2]](s1), [[UV1]], [[UV7]]
+    ; GREEDY: [[C3:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 3
+    ; GREEDY: [[ICMP3:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C3]]
+    ; GREEDY: [[SELECT6:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP3]](s1), [[UV]], [[UV8]]
+    ; GREEDY: [[SELECT7:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP3]](s1), [[UV1]], [[UV9]]
+    ; GREEDY: [[C4:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 4
+    ; GREEDY: [[ICMP4:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C4]]
+    ; GREEDY: [[SELECT8:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP4]](s1), [[UV]], [[UV10]]
+    ; GREEDY: [[SELECT9:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP4]](s1), [[UV1]], [[UV11]]
+    ; GREEDY: [[C5:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 5
+    ; GREEDY: [[ICMP5:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C5]]
+    ; GREEDY: [[SELECT10:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP5]](s1), [[UV]], [[UV12]]
+    ; GREEDY: [[SELECT11:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP5]](s1), [[UV1]], [[UV13]]
+    ; GREEDY: [[C6:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 6
+    ; GREEDY: [[ICMP6:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C6]]
+    ; GREEDY: [[SELECT12:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP6]](s1), [[UV]], [[UV14]]
+    ; GREEDY: [[SELECT13:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP6]](s1), [[UV1]], [[UV15]]
+    ; GREEDY: [[C7:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 7
+    ; GREEDY: [[ICMP7:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C7]]
+    ; GREEDY: [[SELECT14:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP7]](s1), [[UV]], [[UV16]]
+    ; GREEDY: [[SELECT15:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP7]](s1), [[UV1]], [[UV17]]
+    ; GREEDY: [[BUILD_VECTOR:%[0-9]+]]:vgpr(<16 x s32>) = G_BUILD_VECTOR [[SELECT]](s32), [[SELECT1]](s32), [[SELECT2]](s32), [[SELECT3]](s32), [[SELECT4]](s32), [[SELECT5]](s32), [[SELECT6]](s32), [[SELECT7]](s32), [[SELECT8]](s32), [[SELECT9]](s32), [[SELECT10]](s32), [[SELECT11]](s32), [[SELECT12]](s32), [[SELECT13]](s32), [[SELECT14]](s32), [[SELECT15]](s32)
+    ; GREEDY: [[BITCAST:%[0-9]+]]:vgpr(<8 x s64>) = G_BITCAST [[BUILD_VECTOR]](<16 x s32>)
+    ; GREEDY: $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15 = COPY [[BITCAST]](<8 x s64>)
     %0:_(<8 x s64>) = COPY $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7_sgpr8_sgpr9_sgpr10_sgpr11_sgpr12_sgpr13_sgpr14_sgpr15
     %1:_(s64) = COPY $vgpr0_vgpr1
     %2:_(s32) = COPY $vgpr2
     %3:_(<8 x s64>) = G_INSERT_VECTOR_ELT %0, %1, %2
     $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15 = COPY %3
 ...
 
 ---
 name: insert_vector_elt_v8s64_v_v_s
 legalized: true
 tracksRegLiveness: true
 
 body: |
   bb.0:
     liveins: $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15, $vgpr16_vgpr17, $sgpr18
 
-    ; CHECK-LABEL: name: insert_vector_elt_v8s64_v_v_s
-    ; CHECK: liveins: $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15, $vgpr16_vgpr17, $sgpr18
-    ; CHECK: [[COPY:%[0-9]+]]:vgpr(<8 x s64>) = COPY $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15
-    ; CHECK: [[COPY1:%[0-9]+]]:vgpr(s64) = COPY $vgpr16_vgpr17
-    ; CHECK: [[COPY2:%[0-9]+]]:sgpr(s32) = COPY $sgpr18
-    ; CHECK: [[UV:%[0-9]+]]:vgpr(s32), [[UV1:%[0-9]+]]:vgpr(s32) = G_UNMERGE_VALUES [[COPY1]](s64)
-    ; CHECK: [[BITCAST:%[0-9]+]]:vgpr(<16 x s32>) = G_BITCAST [[COPY]](<8 x s64>)
-    ; CHECK: [[C:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 1
-    ; CHECK: [[SHL:%[0-9]+]]:sgpr(s32) = G_SHL [[COPY2]], [[C]](s32)
-    ; CHECK: [[ADD:%[0-9]+]]:sgpr(s32) = G_ADD [[SHL]], [[C]]
-    ; CHECK: [[IVEC:%[0-9]+]]:vgpr(<16 x s32>) = G_INSERT_VECTOR_ELT [[BITCAST]], [[UV]](s32), [[SHL]](s32)
-    ; CHECK: [[IVEC1:%[0-9]+]]:vgpr(<16 x s32>) = G_INSERT_VECTOR_ELT [[IVEC]], [[UV1]](s32), [[ADD]](s32)
-    ; CHECK: [[BITCAST1:%[0-9]+]]:vgpr(<8 x s64>) = G_BITCAST [[IVEC1]](<16 x s32>)
-    ; CHECK: $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15 = COPY [[BITCAST1]](<8 x s64>)
+    ; FAST-LABEL: name: insert_vector_elt_v8s64_v_v_s
+    ; FAST: liveins: $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15, $vgpr16_vgpr17, $sgpr18
+    ; FAST: [[COPY:%[0-9]+]]:vgpr(<8 x s64>) = COPY $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15
+    ; FAST: [[COPY1:%[0-9]+]]:vgpr(s64) = COPY $vgpr16_vgpr17
+    ; FAST: [[COPY2:%[0-9]+]]:sgpr(s32) = COPY $sgpr18
+    ; FAST: [[UV:%[0-9]+]]:vgpr(s32), [[UV1:%[0-9]+]]:vgpr(s32) = G_UNMERGE_VALUES [[COPY1]](s64)
+    ; FAST: [[BITCAST:%[0-9]+]]:vgpr(<16 x s32>) = G_BITCAST [[COPY]](<8 x s64>)
+    ; FAST: [[C:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 1
+    ; FAST: [[SHL:%[0-9]+]]:sgpr(s32) = G_SHL [[COPY2]], [[C]](s32)
+    ; FAST: [[ADD:%[0-9]+]]:sgpr(s32) = G_ADD [[SHL]], [[C]]
+    ; FAST: [[IVEC:%[0-9]+]]:vgpr(<16 x s32>) = G_INSERT_VECTOR_ELT [[BITCAST]], [[UV]](s32), [[SHL]](s32)
+    ; FAST: [[IVEC1:%[0-9]+]]:vgpr(<16 x s32>) = G_INSERT_VECTOR_ELT [[IVEC]], [[UV1]](s32), [[ADD]](s32)
+    ; FAST: [[BITCAST1:%[0-9]+]]:vgpr(<8 x s64>) = G_BITCAST [[IVEC1]](<16 x s32>)
+    ; FAST: $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15 = COPY [[BITCAST1]](<8 x s64>)
+    ; GREEDY-LABEL: name: insert_vector_elt_v8s64_v_v_s
+    ; GREEDY: liveins: $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15, $vgpr16_vgpr17, $sgpr18
+    ; GREEDY: [[COPY:%[0-9]+]]:vgpr(<8 x s64>) = COPY $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15
+    ; GREEDY: [[COPY1:%[0-9]+]]:vgpr(s64) = COPY $vgpr16_vgpr17
+    ; GREEDY: [[COPY2:%[0-9]+]]:sgpr(s32) = COPY $sgpr18
+    ; GREEDY: [[UV:%[0-9]+]]:vgpr(s32), [[UV1:%[0-9]+]]:vgpr(s32) = G_UNMERGE_VALUES [[COPY1]](s64)
+    ; GREEDY: [[BITCAST:%[0-9]+]]:vgpr(<16 x s32>) = G_BITCAST [[COPY]](<8 x s64>)
+    ; GREEDY: [[C:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 1
+    ; GREEDY: [[SHL:%[0-9]+]]:sgpr(s32) = G_SHL [[COPY2]], [[C]](s32)
+    ; GREEDY: [[ADD:%[0-9]+]]:sgpr(s32) = G_ADD [[SHL]], [[C]]
+    ; GREEDY: [[IVEC:%[0-9]+]]:vgpr(<16 x s32>) = G_INSERT_VECTOR_ELT [[BITCAST]], [[UV]](s32), [[SHL]](s32)
+    ; GREEDY: [[IVEC1:%[0-9]+]]:vgpr(<16 x s32>) = G_INSERT_VECTOR_ELT [[IVEC]], [[UV1]](s32), [[ADD]](s32)
+    ; GREEDY: [[BITCAST1:%[0-9]+]]:vgpr(<8 x s64>) = G_BITCAST [[IVEC1]](<16 x s32>)
+    ; GREEDY: $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15 = COPY [[BITCAST1]](<8 x s64>)
     %0:_(<8 x s64>) = COPY $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15
     %1:_(s64) = COPY $vgpr16_vgpr17
     %2:_(s32) = COPY $sgpr18
     %3:_(<8 x s64>) = G_INSERT_VECTOR_ELT %0, %1, %2
     $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15 = COPY %3
 ...
 
 ---
 name: insert_vector_elt_v8s64_v_s_v
 legalized: true
 tracksRegLiveness: true
 
 body: |
   bb.0:
     liveins: $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15, $sgpr0_sgpr1, $vgpr16
 
-    ; CHECK-LABEL: name: insert_vector_elt_v8s64_v_s_v
-    ; CHECK: successors: %bb.1(0x80000000)
-    ; CHECK: liveins: $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15, $sgpr0_sgpr1, $vgpr16
-    ; CHECK: [[COPY:%[0-9]+]]:vgpr(<8 x s64>) = COPY $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15
-    ; CHECK: [[COPY1:%[0-9]+]]:sgpr(s64) = COPY $sgpr0_sgpr1
-    ; CHECK: [[COPY2:%[0-9]+]]:vgpr_32(s32) = COPY $vgpr16
-    ; CHECK: [[UV:%[0-9]+]]:sgpr(s32), [[UV1:%[0-9]+]]:sgpr(s32) = G_UNMERGE_VALUES [[COPY1]](s64)
-    ; CHECK: [[BITCAST:%[0-9]+]]:vgpr(<16 x s32>) = G_BITCAST [[COPY]](<8 x s64>)
-    ; CHECK: [[C:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 1
-    ; CHECK: [[DEF:%[0-9]+]]:sgpr(s32) = G_IMPLICIT_DEF
-    ; CHECK: [[DEF1:%[0-9]+]]:sgpr(s32) = G_IMPLICIT_DEF
-    ; CHECK: [[DEF2:%[0-9]+]]:vgpr(<16 x s32>) = G_IMPLICIT_DEF
-    ; CHECK: [[DEF3:%[0-9]+]]:vgpr(<16 x s32>) = G_IMPLICIT_DEF
-    ; CHECK: [[DEF4:%[0-9]+]]:sreg_64_xexec = IMPLICIT_DEF
-    ; CHECK: [[S_MOV_B64_term:%[0-9]+]]:sreg_64_xexec = S_MOV_B64_term $exec
-    ; CHECK: .1:
-    ; CHECK: successors: %bb.2(0x40000000), %bb.1(0x40000000)
-    ; CHECK: [[PHI:%[0-9]+]]:sreg_64_xexec = PHI [[DEF4]], %bb.0, %23, %bb.1
-    ; CHECK: [[PHI1:%[0-9]+]]:sgpr(s32) = G_PHI [[DEF]](s32), %bb.0, %8(s32), %bb.1
-    ; CHECK: [[PHI2:%[0-9]+]]:sgpr(s32) = G_PHI [[DEF1]](s32), %bb.0, %9(s32), %bb.1
-    ; CHECK: [[PHI3:%[0-9]+]]:vgpr(<16 x s32>) = G_PHI [[DEF2]](<16 x s32>), %bb.0, %10(<16 x s32>), %bb.1
-    ; CHECK: [[PHI4:%[0-9]+]]:vgpr(<16 x s32>) = G_PHI [[DEF3]](<16 x s32>), %bb.0, %11(<16 x s32>), %bb.1
-    ; CHECK: [[V_READFIRSTLANE_B32_:%[0-9]+]]:sreg_32_xm0(s32) = V_READFIRSTLANE_B32 [[COPY2]](s32), implicit $exec
-    ; CHECK: [[V_CMP_EQ_U32_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U32_e64 [[V_READFIRSTLANE_B32_]](s32), [[COPY2]](s32), implicit $exec
-    ; CHECK: [[SHL:%[0-9]+]]:sgpr(s32) = G_SHL [[V_READFIRSTLANE_B32_]], [[C]](s32)
-    ; CHECK: [[ADD:%[0-9]+]]:sgpr(s32) = G_ADD [[SHL]], [[C]]
-    ; CHECK: [[IVEC:%[0-9]+]]:vgpr(<16 x s32>) = G_INSERT_VECTOR_ELT [[BITCAST]], [[UV]](s32), [[SHL]](s32)
-    ; CHECK: [[IVEC1:%[0-9]+]]:vgpr(<16 x s32>) = G_INSERT_VECTOR_ELT [[IVEC]], [[UV1]](s32), [[ADD]](s32)
-    ; CHECK: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[V_CMP_EQ_U32_e64_]], implicit-def $exec, implicit-def $scc, implicit $exec
-    ; CHECK: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc
-    ; CHECK: S_CBRANCH_EXECNZ %bb.1, implicit $exec
-    ; CHECK: .2:
-    ; CHECK: successors: %bb.3(0x80000000)
-    ; CHECK: $exec = S_MOV_B64_term [[S_MOV_B64_term]]
-    ; CHECK: .3:
-    ; CHECK: [[BITCAST1:%[0-9]+]]:vgpr(<8 x s64>) = G_BITCAST [[IVEC1]](<16 x s32>)
-    ; CHECK: $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15 = COPY [[BITCAST1]](<8 x s64>)
+    ; FAST-LABEL: name: insert_vector_elt_v8s64_v_s_v
+    ; FAST: liveins: $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15, $sgpr0_sgpr1, $vgpr16
+    ; FAST: [[COPY:%[0-9]+]]:vgpr(<8 x s64>) = COPY $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15
+    ; FAST: [[COPY1:%[0-9]+]]:sgpr(s64) = COPY $sgpr0_sgpr1
+    ; FAST: [[COPY2:%[0-9]+]]:vgpr(s32) = COPY $vgpr16
+    ; FAST: [[UV:%[0-9]+]]:vgpr(s32), [[UV1:%[0-9]+]]:vgpr(s32) = G_UNMERGE_VALUES [[COPY1]](s64)
+    ; FAST: [[UV2:%[0-9]+]]:vgpr(s32), [[UV3:%[0-9]+]]:vgpr(s32), [[UV4:%[0-9]+]]:vgpr(s32), [[UV5:%[0-9]+]]:vgpr(s32), [[UV6:%[0-9]+]]:vgpr(s32), [[UV7:%[0-9]+]]:vgpr(s32), [[UV8:%[0-9]+]]:vgpr(s32), [[UV9:%[0-9]+]]:vgpr(s32), [[UV10:%[0-9]+]]:vgpr(s32), [[UV11:%[0-9]+]]:vgpr(s32), [[UV12:%[0-9]+]]:vgpr(s32), [[UV13:%[0-9]+]]:vgpr(s32), [[UV14:%[0-9]+]]:vgpr(s32), [[UV15:%[0-9]+]]:vgpr(s32), [[UV16:%[0-9]+]]:vgpr(s32), [[UV17:%[0-9]+]]:vgpr(s32) = G_UNMERGE_VALUES [[COPY]](<8 x s64>)
+    ; FAST: [[C:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 0
+    ; FAST: [[ICMP:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C]]
+    ; FAST: [[SELECT:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP]](s1), [[UV]], [[UV2]]
+    ; FAST: [[SELECT1:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP]](s1), [[UV1]], [[UV3]]
+    ; FAST: [[C1:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 1
+    ; FAST: [[ICMP1:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C1]]
+    ; FAST: [[SELECT2:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP1]](s1), [[UV]], [[UV4]]
+    ; FAST: [[SELECT3:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP1]](s1), [[UV1]], [[UV5]]
+    ; FAST: [[C2:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 2
+    ; FAST: [[ICMP2:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C2]]
+    ; FAST: [[SELECT4:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP2]](s1), [[UV]], [[UV6]]
+    ; FAST: [[SELECT5:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP2]](s1), [[UV1]], [[UV7]]
+    ; FAST: [[C3:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 3
+    ; FAST: [[ICMP3:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C3]]
+    ; FAST: [[SELECT6:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP3]](s1), [[UV]], [[UV8]]
+    ; FAST: [[SELECT7:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP3]](s1), [[UV1]], [[UV9]]
+    ; FAST: [[C4:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 4
+    ; FAST: [[ICMP4:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C4]]
+    ; FAST: [[SELECT8:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP4]](s1), [[UV]], [[UV10]]
+    ; FAST: [[SELECT9:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP4]](s1), [[UV1]], [[UV11]]
+    ; FAST: [[C5:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 5
+    ; FAST: [[ICMP5:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C5]]
+    ; FAST: [[SELECT10:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP5]](s1), [[UV]], [[UV12]]
+    ; FAST: [[SELECT11:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP5]](s1), [[UV1]], [[UV13]]
+    ; FAST: [[C6:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 6
+    ; FAST: [[ICMP6:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C6]]
+    ; FAST: [[SELECT12:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP6]](s1), [[UV]], [[UV14]]
+    ; FAST: [[SELECT13:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP6]](s1), [[UV1]], [[UV15]]
+    ; FAST: [[C7:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 7
+    ; FAST: [[ICMP7:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C7]]
+    ; FAST: [[SELECT14:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP7]](s1), [[UV]], [[UV16]]
+    ; FAST: [[SELECT15:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP7]](s1), [[UV1]], [[UV17]]
+    ; FAST: [[BUILD_VECTOR:%[0-9]+]]:vgpr(<16 x s32>) = G_BUILD_VECTOR [[SELECT]](s32), [[SELECT1]](s32), [[SELECT2]](s32), [[SELECT3]](s32), [[SELECT4]](s32), [[SELECT5]](s32), [[SELECT6]](s32), [[SELECT7]](s32), [[SELECT8]](s32), [[SELECT9]](s32), [[SELECT10]](s32), [[SELECT11]](s32), [[SELECT12]](s32), [[SELECT13]](s32), [[SELECT14]](s32), [[SELECT15]](s32)
+    ; FAST: [[BITCAST:%[0-9]+]]:vgpr(<8 x s64>) = G_BITCAST [[BUILD_VECTOR]](<16 x s32>)
+    ; FAST: $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15 = COPY [[BITCAST]](<8 x s64>)
+    ; GREEDY-LABEL: name: insert_vector_elt_v8s64_v_s_v
+    ; GREEDY: liveins: $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15, $sgpr0_sgpr1, $vgpr16
+    ; GREEDY: [[COPY:%[0-9]+]]:vgpr(<8 x s64>) = COPY $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15
+    ; GREEDY: [[COPY1:%[0-9]+]]:sgpr(s64) = COPY $sgpr0_sgpr1
+    ; GREEDY: [[COPY2:%[0-9]+]]:vgpr(s32) = COPY $vgpr16
+    ; GREEDY: [[UV:%[0-9]+]]:vgpr(s32), [[UV1:%[0-9]+]]:vgpr(s32) = G_UNMERGE_VALUES [[COPY1]](s64)
+    ; GREEDY: [[UV2:%[0-9]+]]:vgpr(s32), [[UV3:%[0-9]+]]:vgpr(s32), [[UV4:%[0-9]+]]:vgpr(s32), [[UV5:%[0-9]+]]:vgpr(s32), [[UV6:%[0-9]+]]:vgpr(s32), [[UV7:%[0-9]+]]:vgpr(s32), [[UV8:%[0-9]+]]:vgpr(s32), [[UV9:%[0-9]+]]:vgpr(s32), [[UV10:%[0-9]+]]:vgpr(s32), [[UV11:%[0-9]+]]:vgpr(s32), [[UV12:%[0-9]+]]:vgpr(s32), [[UV13:%[0-9]+]]:vgpr(s32), [[UV14:%[0-9]+]]:vgpr(s32), [[UV15:%[0-9]+]]:vgpr(s32), [[UV16:%[0-9]+]]:vgpr(s32), [[UV17:%[0-9]+]]:vgpr(s32) = G_UNMERGE_VALUES [[COPY]](<8 x s64>)
+    ; GREEDY: [[C:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 0
+    ; GREEDY: [[ICMP:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C]]
+    ; GREEDY: [[SELECT:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP]](s1), [[UV]], [[UV2]]
+    ; GREEDY: [[SELECT1:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP]](s1), [[UV1]], [[UV3]]
+    ; GREEDY: [[C1:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 1
+    ; GREEDY: [[ICMP1:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C1]]
+    ; GREEDY: [[SELECT2:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP1]](s1), [[UV]], [[UV4]]
+    ; GREEDY: [[SELECT3:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP1]](s1), [[UV1]], [[UV5]]
+    ; GREEDY: [[C2:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 2
+    ; GREEDY: [[ICMP2:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C2]]
+    ; GREEDY: [[SELECT4:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP2]](s1), [[UV]], [[UV6]]
+    ; GREEDY: [[SELECT5:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP2]](s1), [[UV1]], [[UV7]]
+    ; GREEDY: [[C3:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 3
+    ; GREEDY: [[ICMP3:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C3]]
+    ; GREEDY: [[SELECT6:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP3]](s1), [[UV]], [[UV8]]
+    ; GREEDY: [[SELECT7:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP3]](s1), [[UV1]], [[UV9]]
+    ; GREEDY: [[C4:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 4
+    ; GREEDY: [[ICMP4:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C4]]
+    ; GREEDY: [[SELECT8:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP4]](s1), [[UV]], [[UV10]]
+    ; GREEDY: [[SELECT9:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP4]](s1), [[UV1]], [[UV11]]
+    ; GREEDY: [[C5:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 5
+    ; GREEDY: [[ICMP5:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C5]]
+    ; GREEDY: [[SELECT10:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP5]](s1), [[UV]], [[UV12]]
+    ; GREEDY: [[SELECT11:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP5]](s1), [[UV1]], [[UV13]]
+    ; GREEDY: [[C6:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 6
+    ; GREEDY: [[ICMP6:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C6]]
+    ; GREEDY: [[SELECT12:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP6]](s1), [[UV]], [[UV14]]
+    ; GREEDY: [[SELECT13:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP6]](s1), [[UV1]], [[UV15]]
+    ; GREEDY: [[C7:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 7
+    ; GREEDY: [[ICMP7:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C7]]
+    ; GREEDY: [[SELECT14:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP7]](s1), [[UV]], [[UV16]]
+    ; GREEDY: [[SELECT15:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP7]](s1), [[UV1]], [[UV17]]
+    ; GREEDY: [[BUILD_VECTOR:%[0-9]+]]:vgpr(<16 x s32>) = G_BUILD_VECTOR [[SELECT]](s32), [[SELECT1]](s32), [[SELECT2]](s32), [[SELECT3]](s32), [[SELECT4]](s32), [[SELECT5]](s32), [[SELECT6]](s32), [[SELECT7]](s32), [[SELECT8]](s32), [[SELECT9]](s32), [[SELECT10]](s32), [[SELECT11]](s32), [[SELECT12]](s32), [[SELECT13]](s32), [[SELECT14]](s32), [[SELECT15]](s32)
+    ; GREEDY: [[BITCAST:%[0-9]+]]:vgpr(<8 x s64>) = G_BITCAST [[BUILD_VECTOR]](<16 x s32>)
+    ; GREEDY: $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15 = COPY [[BITCAST]](<8 x s64>)
     %0:_(<8 x s64>) = COPY $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15
     %1:_(s64) = COPY $sgpr0_sgpr1
     %2:_(s32) = COPY $vgpr16
     %3:_(<8 x s64>) = G_INSERT_VECTOR_ELT %0, %1, %2
     $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15 = COPY %3
 ...
 
 ---
 name: insert_vector_elt_v8s64_v_v_v
 legalized: true
 tracksRegLiveness: true
 
 body: |
   bb.0:
     liveins: $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15, $vgpr16_vgpr17, $vgpr18
 
-    ; CHECK-LABEL: name: insert_vector_elt_v8s64_v_v_v
-    ; CHECK: successors: %bb.1(0x80000000)
-    ; CHECK: liveins: $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15, $vgpr16_vgpr17, $vgpr18
-    ; CHECK: [[COPY:%[0-9]+]]:vgpr(<8 x s64>) = COPY $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15
-    ; CHECK: [[COPY1:%[0-9]+]]:vgpr(s64) = COPY $vgpr16_vgpr17
-    ; CHECK: [[COPY2:%[0-9]+]]:vgpr_32(s32) = COPY $vgpr18
-    ; CHECK: [[UV:%[0-9]+]]:vgpr(s32), [[UV1:%[0-9]+]]:vgpr(s32) = G_UNMERGE_VALUES [[COPY1]](s64)
-    ; CHECK: [[BITCAST:%[0-9]+]]:vgpr(<16 x s32>) = G_BITCAST [[COPY]](<8 x s64>)
-    ; CHECK: [[C:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 1
-    ; CHECK: [[DEF:%[0-9]+]]:sgpr(s32) = G_IMPLICIT_DEF
-    ; CHECK: [[DEF1:%[0-9]+]]:sgpr(s32) = G_IMPLICIT_DEF
-    ; CHECK: [[DEF2:%[0-9]+]]:vgpr(<16 x s32>) = G_IMPLICIT_DEF
-    ; CHECK: [[DEF3:%[0-9]+]]:vgpr(<16 x s32>) = G_IMPLICIT_DEF
-    ; CHECK: [[DEF4:%[0-9]+]]:sreg_64_xexec = IMPLICIT_DEF
-    ; CHECK: [[S_MOV_B64_term:%[0-9]+]]:sreg_64_xexec = S_MOV_B64_term $exec
-    ; CHECK: .1:
-    ; CHECK: successors: %bb.2(0x40000000), %bb.1(0x40000000)
-    ; CHECK: [[PHI:%[0-9]+]]:sreg_64_xexec = PHI [[DEF4]], %bb.0, %23, %bb.1
-    ; CHECK: [[PHI1:%[0-9]+]]:sgpr(s32) = G_PHI [[DEF]](s32), %bb.0, %8(s32), %bb.1
-    ; CHECK: [[PHI2:%[0-9]+]]:sgpr(s32) = G_PHI [[DEF1]](s32), %bb.0, %9(s32), %bb.1
-    ; CHECK: [[PHI3:%[0-9]+]]:vgpr(<16 x s32>) = G_PHI [[DEF2]](<16 x s32>), %bb.0, %10(<16 x s32>), %bb.1
-    ; CHECK: [[PHI4:%[0-9]+]]:vgpr(<16 x s32>) = G_PHI [[DEF3]](<16 x s32>), %bb.0, %11(<16 x s32>), %bb.1
-    ; CHECK: [[V_READFIRSTLANE_B32_:%[0-9]+]]:sreg_32_xm0(s32) = V_READFIRSTLANE_B32 [[COPY2]](s32), implicit $exec
-    ; CHECK: [[V_CMP_EQ_U32_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U32_e64 [[V_READFIRSTLANE_B32_]](s32), [[COPY2]](s32), implicit $exec
-    ; CHECK: [[SHL:%[0-9]+]]:sgpr(s32) = G_SHL [[V_READFIRSTLANE_B32_]], [[C]](s32)
-    ; CHECK: [[ADD:%[0-9]+]]:sgpr(s32) = G_ADD [[SHL]], [[C]]
-    ; CHECK: [[IVEC:%[0-9]+]]:vgpr(<16 x s32>) = G_INSERT_VECTOR_ELT [[BITCAST]], [[UV]](s32), [[SHL]](s32)
-    ; CHECK: [[IVEC1:%[0-9]+]]:vgpr(<16 x s32>) = G_INSERT_VECTOR_ELT [[IVEC]], [[UV1]](s32), [[ADD]](s32)
-    ; CHECK: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[V_CMP_EQ_U32_e64_]], implicit-def $exec, implicit-def $scc, implicit $exec
-    ; CHECK: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc
-    ; CHECK: S_CBRANCH_EXECNZ %bb.1, implicit $exec
-    ; CHECK: .2:
-    ; CHECK: successors: %bb.3(0x80000000)
-    ; CHECK: $exec = S_MOV_B64_term [[S_MOV_B64_term]]
-    ; CHECK: .3:
-    ; CHECK: [[BITCAST1:%[0-9]+]]:vgpr(<8 x s64>) = G_BITCAST [[IVEC1]](<16 x s32>)
-    ; CHECK: $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15 = COPY [[BITCAST1]](<8 x s64>)
+    ; FAST-LABEL: name: insert_vector_elt_v8s64_v_v_v
+    ; FAST: liveins: $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15, $vgpr16_vgpr17, $vgpr18
+    ; FAST: [[COPY:%[0-9]+]]:vgpr(<8 x s64>) = COPY $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15
+    ; FAST: [[COPY1:%[0-9]+]]:vgpr(s64) = COPY $vgpr16_vgpr17
+    ; FAST: [[COPY2:%[0-9]+]]:vgpr(s32) = COPY $vgpr18
+    ; FAST: [[UV:%[0-9]+]]:vgpr(s32), [[UV1:%[0-9]+]]:vgpr(s32) = G_UNMERGE_VALUES [[COPY1]](s64)
+    ; FAST: [[UV2:%[0-9]+]]:vgpr(s32), [[UV3:%[0-9]+]]:vgpr(s32), [[UV4:%[0-9]+]]:vgpr(s32), [[UV5:%[0-9]+]]:vgpr(s32), [[UV6:%[0-9]+]]:vgpr(s32), [[UV7:%[0-9]+]]:vgpr(s32), [[UV8:%[0-9]+]]:vgpr(s32), [[UV9:%[0-9]+]]:vgpr(s32), [[UV10:%[0-9]+]]:vgpr(s32), [[UV11:%[0-9]+]]:vgpr(s32), [[UV12:%[0-9]+]]:vgpr(s32), [[UV13:%[0-9]+]]:vgpr(s32), [[UV14:%[0-9]+]]:vgpr(s32), [[UV15:%[0-9]+]]:vgpr(s32), [[UV16:%[0-9]+]]:vgpr(s32), [[UV17:%[0-9]+]]:vgpr(s32) = G_UNMERGE_VALUES [[COPY]](<8 x s64>)
+    ; FAST: [[C:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 0
+    ; FAST: [[ICMP:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C]]
+    ; FAST: [[SELECT:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP]](s1), [[UV]], [[UV2]]
+    ; FAST: [[SELECT1:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP]](s1), [[UV1]], [[UV3]]
+    ; FAST: [[C1:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 1
+    ; FAST: [[ICMP1:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C1]]
+    ; FAST: [[SELECT2:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP1]](s1), [[UV]], [[UV4]]
+    ; FAST: [[SELECT3:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP1]](s1), [[UV1]], [[UV5]]
+    ; FAST: [[C2:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 2
+    ; FAST: [[ICMP2:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C2]]
+    ; FAST: [[SELECT4:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP2]](s1), [[UV]], [[UV6]]
+    ; FAST: [[SELECT5:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP2]](s1), [[UV1]], [[UV7]]
+    ; FAST: [[C3:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 3
+    ; FAST: [[ICMP3:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C3]]
+    ; FAST: [[SELECT6:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP3]](s1), [[UV]], [[UV8]]
+    ; FAST: [[SELECT7:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP3]](s1), [[UV1]], [[UV9]]
+    ; FAST: [[C4:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 4
+    ; FAST: [[ICMP4:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C4]]
+    ; FAST: [[SELECT8:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP4]](s1), [[UV]], [[UV10]]
+    ; FAST: [[SELECT9:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP4]](s1), [[UV1]], [[UV11]]
+    ; FAST: [[C5:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 5
+    ; FAST: [[ICMP5:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C5]]
+    ; FAST: [[SELECT10:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP5]](s1), [[UV]], [[UV12]]
+    ; FAST: [[SELECT11:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP5]](s1), [[UV1]], [[UV13]]
+    ; FAST: [[C6:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 6
+    ; FAST: [[ICMP6:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C6]]
+    ; FAST: [[SELECT12:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP6]](s1), [[UV]], [[UV14]]
+    ; FAST: [[SELECT13:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP6]](s1), [[UV1]], [[UV15]]
+    ; FAST: [[C7:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 7
+    ; FAST: [[ICMP7:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C7]]
+    ; FAST: [[SELECT14:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP7]](s1), [[UV]], [[UV16]]
+    ; FAST: [[SELECT15:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP7]](s1), [[UV1]], [[UV17]]
+    ; FAST: [[BUILD_VECTOR:%[0-9]+]]:vgpr(<16 x s32>) = G_BUILD_VECTOR [[SELECT]](s32), [[SELECT1]](s32), [[SELECT2]](s32), [[SELECT3]](s32), [[SELECT4]](s32), [[SELECT5]](s32), [[SELECT6]](s32), [[SELECT7]](s32), [[SELECT8]](s32), [[SELECT9]](s32), [[SELECT10]](s32), [[SELECT11]](s32), [[SELECT12]](s32), [[SELECT13]](s32), [[SELECT14]](s32), [[SELECT15]](s32)
+    ; FAST: [[BITCAST:%[0-9]+]]:vgpr(<8 x s64>) = G_BITCAST [[BUILD_VECTOR]](<16 x s32>)
+    ; FAST: $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15 = COPY [[BITCAST]](<8 x s64>)
+    ; GREEDY-LABEL: name: insert_vector_elt_v8s64_v_v_v
+    ; GREEDY: liveins: $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15, $vgpr16_vgpr17, $vgpr18
+    ; GREEDY: [[COPY:%[0-9]+]]:vgpr(<8 x s64>) = COPY $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15
+    ; GREEDY: [[COPY1:%[0-9]+]]:vgpr(s64) = COPY $vgpr16_vgpr17
+    ; GREEDY: [[COPY2:%[0-9]+]]:vgpr(s32) = COPY $vgpr18
+    ; GREEDY: [[UV:%[0-9]+]]:vgpr(s32), [[UV1:%[0-9]+]]:vgpr(s32) = G_UNMERGE_VALUES [[COPY1]](s64)
+    ; GREEDY: [[UV2:%[0-9]+]]:vgpr(s32), [[UV3:%[0-9]+]]:vgpr(s32), [[UV4:%[0-9]+]]:vgpr(s32), [[UV5:%[0-9]+]]:vgpr(s32), [[UV6:%[0-9]+]]:vgpr(s32), [[UV7:%[0-9]+]]:vgpr(s32), [[UV8:%[0-9]+]]:vgpr(s32), [[UV9:%[0-9]+]]:vgpr(s32), [[UV10:%[0-9]+]]:vgpr(s32), [[UV11:%[0-9]+]]:vgpr(s32), [[UV12:%[0-9]+]]:vgpr(s32), [[UV13:%[0-9]+]]:vgpr(s32), [[UV14:%[0-9]+]]:vgpr(s32), [[UV15:%[0-9]+]]:vgpr(s32), [[UV16:%[0-9]+]]:vgpr(s32), [[UV17:%[0-9]+]]:vgpr(s32) = G_UNMERGE_VALUES [[COPY]](<8 x s64>)
+    ; GREEDY: [[C:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 0
+    ; GREEDY: [[ICMP:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C]]
+    ; GREEDY: [[SELECT:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP]](s1), [[UV]], [[UV2]]
+    ; GREEDY: [[SELECT1:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP]](s1), [[UV1]], [[UV3]]
+    ; GREEDY: [[C1:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 1
+    ; GREEDY: [[ICMP1:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C1]]
+    ; GREEDY: [[SELECT2:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP1]](s1), [[UV]], [[UV4]]
+    ; GREEDY: [[SELECT3:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP1]](s1), [[UV1]], [[UV5]]
+    ; GREEDY: [[C2:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 2
+    ; GREEDY: [[ICMP2:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C2]]
+    ; GREEDY: [[SELECT4:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP2]](s1), [[UV]], [[UV6]]
+    ; GREEDY: [[SELECT5:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP2]](s1), [[UV1]], [[UV7]]
+    ; GREEDY: [[C3:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 3
+    ; GREEDY: [[ICMP3:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C3]]
+    ; GREEDY: [[SELECT6:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP3]](s1), [[UV]], [[UV8]]
+    ; GREEDY: [[SELECT7:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP3]](s1), [[UV1]], [[UV9]]
+    ; GREEDY: [[C4:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 4
+    ; GREEDY: [[ICMP4:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C4]]
+    ; GREEDY: [[SELECT8:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP4]](s1), [[UV]], [[UV10]]
+    ; GREEDY: [[SELECT9:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP4]](s1), [[UV1]], [[UV11]]
+    ; GREEDY: [[C5:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 5
+    ; GREEDY: [[ICMP5:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C5]]
+    ; GREEDY: [[SELECT10:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP5]](s1), [[UV]], [[UV12]]
+    ; GREEDY: [[SELECT11:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP5]](s1), [[UV1]], [[UV13]]
+    ; GREEDY: [[C6:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 6
+    ; GREEDY: [[ICMP6:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C6]]
+    ; GREEDY: [[SELECT12:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP6]](s1), [[UV]], [[UV14]]
+    ; GREEDY: [[SELECT13:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP6]](s1), [[UV1]], [[UV15]]
+    ; GREEDY: [[C7:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 7
+    ; GREEDY: [[ICMP7:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C7]]
+    ; GREEDY: [[SELECT14:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP7]](s1), [[UV]], [[UV16]]
+    ; GREEDY: [[SELECT15:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP7]](s1), [[UV1]], [[UV17]]
+    ; GREEDY: [[BUILD_VECTOR:%[0-9]+]]:vgpr(<16 x s32>) = G_BUILD_VECTOR [[SELECT]](s32), [[SELECT1]](s32), [[SELECT2]](s32), [[SELECT3]](s32), [[SELECT4]](s32), [[SELECT5]](s32), [[SELECT6]](s32), [[SELECT7]](s32), [[SELECT8]](s32), [[SELECT9]](s32), [[SELECT10]](s32), [[SELECT11]](s32), [[SELECT12]](s32), [[SELECT13]](s32), [[SELECT14]](s32), [[SELECT15]](s32)
+    ; GREEDY: [[BITCAST:%[0-9]+]]:vgpr(<8 x s64>) = G_BITCAST [[BUILD_VECTOR]](<16 x s32>)
+    ; GREEDY: $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15 = COPY [[BITCAST]](<8 x s64>)
     %0:_(<8 x s64>) = COPY $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15
     %1:_(s64) = COPY $vgpr16_vgpr17
     %2:_(s32) = COPY $vgpr18
     %3:_(<8 x s64>) = G_INSERT_VECTOR_ELT %0, %1, %2
     $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15 = COPY %3
 ...
 
 # Make sure nothing goes wrong if the G_INSERT_VECTOR_ELT is the last
 # instruction in a block
 ---
 name: insert_vector_elt_v8s64_v_v_v_last_in_block
 legalized: true
 tracksRegLiveness: true
 
 body: |
-  ; CHECK-LABEL: name: insert_vector_elt_v8s64_v_v_v_last_in_block
-  ; CHECK: bb.0:
-  ; CHECK:   successors: %bb.2(0x80000000)
-  ; CHECK:   liveins: $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15, $vgpr16_vgpr17, $vgpr18
-  ; CHECK:   [[COPY:%[0-9]+]]:vgpr(<8 x s64>) = COPY $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15
-  ; CHECK:   [[COPY1:%[0-9]+]]:vgpr(s64) = COPY $vgpr16_vgpr17
-  ; CHECK:   [[COPY2:%[0-9]+]]:vgpr_32(s32) = COPY $vgpr18
-  ; CHECK:   [[UV:%[0-9]+]]:vgpr(s32), [[UV1:%[0-9]+]]:vgpr(s32) = G_UNMERGE_VALUES [[COPY1]](s64)
-  ; CHECK:   [[BITCAST:%[0-9]+]]:vgpr(<16 x s32>) = G_BITCAST [[COPY]](<8 x s64>)
-  ; CHECK:   [[C:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 1
-  ; CHECK:   [[DEF:%[0-9]+]]:sgpr(s32) = G_IMPLICIT_DEF
-  ; CHECK:   [[DEF1:%[0-9]+]]:sgpr(s32) = G_IMPLICIT_DEF
-  ; CHECK:   [[DEF2:%[0-9]+]]:vgpr(<16 x s32>) = G_IMPLICIT_DEF
-  ; CHECK:   [[DEF3:%[0-9]+]]:vgpr(<16 x s32>) = G_IMPLICIT_DEF
-  ; CHECK:   [[DEF4:%[0-9]+]]:sreg_64_xexec = IMPLICIT_DEF
-  ; CHECK:   [[S_MOV_B64_term:%[0-9]+]]:sreg_64_xexec = S_MOV_B64_term $exec
-  ; CHECK: bb.2:
-  ; CHECK:   successors: %bb.3(0x40000000), %bb.2(0x40000000)
-  ; CHECK:   [[PHI:%[0-9]+]]:sreg_64_xexec = PHI [[DEF4]], %bb.0, %23, %bb.2
-  ; CHECK:   [[PHI1:%[0-9]+]]:sgpr(s32) = G_PHI [[DEF]](s32), %bb.0, %8(s32), %bb.2
-  ; CHECK:   [[PHI2:%[0-9]+]]:sgpr(s32) = G_PHI [[DEF1]](s32), %bb.0, %9(s32), %bb.2
-  ; CHECK:   [[PHI3:%[0-9]+]]:vgpr(<16 x s32>) = G_PHI [[DEF2]](<16 x s32>), %bb.0, %10(<16 x s32>), %bb.2
-  ; CHECK:   [[PHI4:%[0-9]+]]:vgpr(<16 x s32>) = G_PHI [[DEF3]](<16 x s32>), %bb.0, %11(<16 x s32>), %bb.2
-  ; CHECK:   [[V_READFIRSTLANE_B32_:%[0-9]+]]:sreg_32_xm0(s32) = V_READFIRSTLANE_B32 [[COPY2]](s32), implicit $exec
-  ; CHECK:   [[V_CMP_EQ_U32_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U32_e64 [[V_READFIRSTLANE_B32_]](s32), [[COPY2]](s32), implicit $exec
-  ; CHECK:   [[SHL:%[0-9]+]]:sgpr(s32) = G_SHL [[V_READFIRSTLANE_B32_]], [[C]](s32)
-  ; CHECK:   [[ADD:%[0-9]+]]:sgpr(s32) = G_ADD [[SHL]], [[C]]
-  ; CHECK:   [[IVEC:%[0-9]+]]:vgpr(<16 x s32>) = G_INSERT_VECTOR_ELT [[BITCAST]], [[UV]](s32), [[SHL]](s32)
-  ; CHECK:   [[IVEC1:%[0-9]+]]:vgpr(<16 x s32>) = G_INSERT_VECTOR_ELT [[IVEC]], [[UV1]](s32), [[ADD]](s32)
-  ; CHECK:   [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[V_CMP_EQ_U32_e64_]], implicit-def $exec, implicit-def $scc, implicit $exec
-  ; CHECK:   $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc
-  ; CHECK:   S_CBRANCH_EXECNZ %bb.2, implicit $exec
-  ; CHECK: bb.3:
-  ; CHECK:   successors: %bb.4(0x80000000)
-  ; CHECK:   $exec = S_MOV_B64_term [[S_MOV_B64_term]]
-  ; CHECK: bb.4:
-  ; CHECK:   successors: %bb.1(0x80000000)
-  ; CHECK:   [[BITCAST1:%[0-9]+]]:vgpr(<8 x s64>) = G_BITCAST [[IVEC1]](<16 x s32>)
-  ; CHECK: bb.1:
-  ; CHECK:   $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15 = COPY [[BITCAST1]](<8 x s64>)
+  ; FAST-LABEL: name: insert_vector_elt_v8s64_v_v_v_last_in_block
+  ; FAST: bb.0:
+  ; FAST:   successors: %bb.1(0x80000000)
+  ; FAST:   liveins: $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15, $vgpr16_vgpr17, $vgpr18
+  ; FAST:   [[COPY:%[0-9]+]]:vgpr(<8 x s64>) = COPY $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15
+  ; FAST:   [[COPY1:%[0-9]+]]:vgpr(s64) = COPY $vgpr16_vgpr17
+  ; FAST:   [[COPY2:%[0-9]+]]:vgpr(s32) = COPY $vgpr18
+  ; FAST:   [[UV:%[0-9]+]]:vgpr(s32), [[UV1:%[0-9]+]]:vgpr(s32) = G_UNMERGE_VALUES [[COPY1]](s64)
+  ; FAST:   [[UV2:%[0-9]+]]:vgpr(s32), [[UV3:%[0-9]+]]:vgpr(s32), [[UV4:%[0-9]+]]:vgpr(s32), [[UV5:%[0-9]+]]:vgpr(s32), [[UV6:%[0-9]+]]:vgpr(s32), [[UV7:%[0-9]+]]:vgpr(s32), [[UV8:%[0-9]+]]:vgpr(s32), [[UV9:%[0-9]+]]:vgpr(s32), [[UV10:%[0-9]+]]:vgpr(s32), [[UV11:%[0-9]+]]:vgpr(s32), [[UV12:%[0-9]+]]:vgpr(s32), [[UV13:%[0-9]+]]:vgpr(s32), [[UV14:%[0-9]+]]:vgpr(s32), [[UV15:%[0-9]+]]:vgpr(s32), [[UV16:%[0-9]+]]:vgpr(s32), [[UV17:%[0-9]+]]:vgpr(s32) = G_UNMERGE_VALUES [[COPY]](<8 x s64>)
+  ; FAST:   [[C:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 0
+  ; FAST:   [[ICMP:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C]]
+  ; FAST:   [[SELECT:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP]](s1), [[UV]], [[UV2]]
+  ; FAST:   [[SELECT1:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP]](s1), [[UV1]], [[UV3]]
+  ; FAST:   [[C1:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 1
+  ; FAST:   [[ICMP1:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C1]]
+  ; FAST:   [[SELECT2:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP1]](s1), [[UV]], [[UV4]]
+  ; FAST:   [[SELECT3:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP1]](s1), [[UV1]], [[UV5]]
+  ; FAST:   [[C2:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 2
+  ; FAST:   [[ICMP2:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C2]]
+  ; FAST:   [[SELECT4:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP2]](s1), [[UV]], [[UV6]]
+  ; FAST:   [[SELECT5:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP2]](s1), [[UV1]], [[UV7]]
+  ; FAST:   [[C3:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 3
+  ; FAST:   [[ICMP3:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C3]]
+  ; FAST:   [[SELECT6:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP3]](s1), [[UV]], [[UV8]]
+  ; FAST:   [[SELECT7:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP3]](s1), [[UV1]], [[UV9]]
+  ; FAST:   [[C4:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 4
+  ; FAST:   [[ICMP4:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C4]]
+  ; FAST:   [[SELECT8:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP4]](s1), [[UV]], [[UV10]]
+  ; FAST:   [[SELECT9:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP4]](s1), [[UV1]], [[UV11]]
+  ; FAST:   [[C5:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 5
+  ; FAST:   [[ICMP5:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C5]]
+  ; FAST:   [[SELECT10:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP5]](s1), [[UV]], [[UV12]]
+  ; FAST:   [[SELECT11:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP5]](s1), [[UV1]], [[UV13]]
+  ; FAST:   [[C6:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 6
+  ; FAST:   [[ICMP6:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C6]]
+  ; FAST:   [[SELECT12:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP6]](s1), [[UV]], [[UV14]]
+  ; FAST:   [[SELECT13:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP6]](s1), [[UV1]], [[UV15]]
+  ; FAST:   [[C7:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 7
+  ; FAST:   [[ICMP7:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C7]]
+  ; FAST:   [[SELECT14:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP7]](s1), [[UV]], [[UV16]]
+  ; FAST:   [[SELECT15:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP7]](s1), [[UV1]], [[UV17]]
+  ; FAST:   [[BUILD_VECTOR:%[0-9]+]]:vgpr(<16 x s32>) = G_BUILD_VECTOR [[SELECT]](s32), [[SELECT1]](s32), [[SELECT2]](s32), [[SELECT3]](s32), [[SELECT4]](s32), [[SELECT5]](s32), [[SELECT6]](s32), [[SELECT7]](s32), [[SELECT8]](s32), [[SELECT9]](s32), [[SELECT10]](s32), [[SELECT11]](s32), [[SELECT12]](s32), [[SELECT13]](s32), [[SELECT14]](s32), [[SELECT15]](s32)
+  ; FAST:   [[BITCAST:%[0-9]+]]:vgpr(<8 x s64>) = G_BITCAST [[BUILD_VECTOR]](<16 x s32>)
+  ; FAST: bb.1:
+  ; FAST:   $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15 = COPY [[BITCAST]](<8 x s64>)
+  ; GREEDY-LABEL: name: insert_vector_elt_v8s64_v_v_v_last_in_block
+  ; GREEDY: bb.0:
+  ; GREEDY:   successors: %bb.1(0x80000000)
+  ; GREEDY:   liveins: $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15, $vgpr16_vgpr17, $vgpr18
+  ; GREEDY:   [[COPY:%[0-9]+]]:vgpr(<8 x s64>) = COPY $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15
+  ; GREEDY:   [[COPY1:%[0-9]+]]:vgpr(s64) = COPY $vgpr16_vgpr17
+  ; GREEDY:   [[COPY2:%[0-9]+]]:vgpr(s32) = COPY $vgpr18
+  ; GREEDY:   [[UV:%[0-9]+]]:vgpr(s32), [[UV1:%[0-9]+]]:vgpr(s32) = G_UNMERGE_VALUES [[COPY1]](s64)
+  ; GREEDY:   [[UV2:%[0-9]+]]:vgpr(s32), [[UV3:%[0-9]+]]:vgpr(s32), [[UV4:%[0-9]+]]:vgpr(s32), [[UV5:%[0-9]+]]:vgpr(s32), [[UV6:%[0-9]+]]:vgpr(s32), [[UV7:%[0-9]+]]:vgpr(s32), [[UV8:%[0-9]+]]:vgpr(s32), [[UV9:%[0-9]+]]:vgpr(s32), [[UV10:%[0-9]+]]:vgpr(s32), [[UV11:%[0-9]+]]:vgpr(s32), [[UV12:%[0-9]+]]:vgpr(s32), [[UV13:%[0-9]+]]:vgpr(s32), [[UV14:%[0-9]+]]:vgpr(s32), [[UV15:%[0-9]+]]:vgpr(s32), [[UV16:%[0-9]+]]:vgpr(s32), [[UV17:%[0-9]+]]:vgpr(s32) = G_UNMERGE_VALUES [[COPY]](<8 x s64>)
+  ; GREEDY:   [[C:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 0
+  ; GREEDY:   [[ICMP:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C]]
+  ; GREEDY:   [[SELECT:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP]](s1), [[UV]], [[UV2]]
+  ; GREEDY:   [[SELECT1:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP]](s1), [[UV1]], [[UV3]]
+  ; GREEDY:   [[C1:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 1
+  ; GREEDY:   [[ICMP1:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C1]]
+  ; GREEDY:   [[SELECT2:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP1]](s1), [[UV]], [[UV4]]
+  ; GREEDY:   [[SELECT3:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP1]](s1), [[UV1]], [[UV5]]
+  ; GREEDY:   [[C2:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 2
+  ; GREEDY:   [[ICMP2:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C2]]
+  ; GREEDY:   [[SELECT4:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP2]](s1), [[UV]], [[UV6]]
+  ; GREEDY:   [[SELECT5:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP2]](s1), [[UV1]], [[UV7]]
+  ; GREEDY:   [[C3:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 3
+  ; GREEDY:   [[ICMP3:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C3]]
+  ; GREEDY:   [[SELECT6:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP3]](s1), [[UV]], [[UV8]]
+  ; GREEDY:   [[SELECT7:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP3]](s1), [[UV1]], [[UV9]]
+  ; GREEDY:   [[C4:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 4
+  ; GREEDY:   [[ICMP4:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C4]]
+  ; GREEDY:   [[SELECT8:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP4]](s1), [[UV]], [[UV10]]
+  ; GREEDY:   [[SELECT9:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP4]](s1), [[UV1]], [[UV11]]
+  ; GREEDY:   [[C5:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 5
+  ; GREEDY:   [[ICMP5:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C5]]
+  ; GREEDY:   [[SELECT10:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP5]](s1), [[UV]], [[UV12]]
+  ; GREEDY:   [[SELECT11:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP5]](s1), [[UV1]], [[UV13]]
+  ; GREEDY:   [[C6:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 6
+  ; GREEDY:   [[ICMP6:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C6]]
+  ; GREEDY:   [[SELECT12:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP6]](s1), [[UV]], [[UV14]]
+  ; GREEDY:   [[SELECT13:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP6]](s1), [[UV1]], [[UV15]]
+  ; GREEDY:   [[C7:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 7
+  ; GREEDY:   [[ICMP7:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[COPY2]](s32), [[C7]]
+  ; GREEDY:   [[SELECT14:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP7]](s1), [[UV]], [[UV16]]
+  ; GREEDY:   [[SELECT15:%[0-9]+]]:vgpr(s32) = G_SELECT [[ICMP7]](s1), [[UV1]], [[UV17]]
+  ; GREEDY:   [[BUILD_VECTOR:%[0-9]+]]:vgpr(<16 x s32>) = G_BUILD_VECTOR [[SELECT]](s32), [[SELECT1]](s32), [[SELECT2]](s32), [[SELECT3]](s32), [[SELECT4]](s32), [[SELECT5]](s32), [[SELECT6]](s32), [[SELECT7]](s32), [[SELECT8]](s32), [[SELECT9]](s32), [[SELECT10]](s32), [[SELECT11]](s32), [[SELECT12]](s32), [[SELECT13]](s32), [[SELECT14]](s32), [[SELECT15]](s32)
+  ; GREEDY:   [[BITCAST:%[0-9]+]]:vgpr(<8 x s64>) = G_BITCAST [[BUILD_VECTOR]](<16 x s32>)
+  ; GREEDY: bb.1:
+  ; GREEDY:   $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15 = COPY [[BITCAST]](<8 x s64>)
   bb.0:
     liveins: $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15, $vgpr16_vgpr17, $vgpr18
 
     %0:_(<8 x s64>) = COPY $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15
     %1:_(s64) = COPY $vgpr16_vgpr17
     %2:_(s32) = COPY $vgpr18
     %3:_(<8 x s64>) = G_INSERT_VECTOR_ELT %0, %1, %2
 
   bb.1:
     $vgpr0_vgpr1_vgpr2_vgpr3_vgpr4_vgpr5_vgpr6_vgpr7_vgpr8_vgpr9_vgpr10_vgpr11_vgpr12_vgpr13_vgpr14_vgpr15 = COPY %3
 ...