[AMDGPU][RFC] New llvm.amdgcn.ballot intrinsic

[Flakebi]

Summary

Add a new llvm.amdgcn.ballot intrinsic modeled on the ballot function
in GLSL and other shader languages. It returns a bitfield containing the
result of its boolean argument in all active lanes, and zero in all
inactive lanes.

This is intended to replace the existing llvm.amdgcn.icmp and
llvm.amdgcn.fcmp intrinsics after a suitable transition period.

Use the new intrinsic in the atomic optimizer pass.

I'm not going to commit this as-is because tests are failing due to
poor code generation, e.g. test2 in ballot.ll generates:

v_cmp_eq_u32_e32 vcc, v0, v1
v_cndmask_b32_e64 v0, 0, 1, vcc
v_cmp_ne_u32_e64 s[4:5], 0, v0
v_mov_b32_e32 v0, s4
v_mov_b32_e32 v1, s5

instead of:

v_cmp_eq_u32_e32 s[4:5], v0, v1
v_mov_b32_e32 v0, s4
v_mov_b32_e32 v1, s5

I'd appreciate feedback on (a) the idea, (b) the implementation and
(c) how best to improve the code generation.

[arsenm]

llvm/lib/Target/AMDGPU/SIISelLowering.cpp

4276–4277

Should also do this in instcombine, as icmp/fcmp already do

4291

This is already the right type, so this should be unnecessary

llvm/test/CodeGen/AMDGPU/ballot.ll

4 ↗

(On Diff #211078)

This needs the .i64 for the mangling, I'm surprised this works

49 ↗

(On Diff #211078)

Can you add some different imp and fcmp sources?

Also need a separate wave32 test

[foad]

Address review comments.

[foad]

llvm/lib/Target/AMDGPU/SIISelLowering.cpp

4276–4277

I've had a go, but is it really necessary to implement the same optimization cases in two places (especially the one that just replaces a ballot intrinsic with a read_register intrinsic), and how do I test them independently?

4291

Does anything guarantee that it's the right type? Is it OK for the compiler to crash if someone writes a .ll file that uses it with the wrong type?

[nhaehnle]

Thanks for doing this. For the codegen quality question, I wonder if something like the following could be done:

1. Remove what's currently in SITargetLowering.
2. Add a custom DAG combine which combines (ballot (ISD::SETCC ...)) to (AMDGPUISD::SETCC ...)
3. Do code generation for the remaining ballot cases either in AMDGPUISelDAGToDAG or via a TableGen pattern (the most generic case would have to map to some combination of COPY_TO_REGCLASS and S_AND_B32/B64).

llvm/lib/Target/AMDGPU/SIISelLowering.cpp

4276–4277

Independent tests as opt tests should go into test/Transforms/InstCombine. You can find existing tests for the icmp/fcmp intrinsics there.

Nitpick: please add braces for the if-body (multiple lines guarded by if).

[arsenm]

llvm/lib/Target/AMDGPU/SIISelLowering.cpp

4276–4277

Yes. If I were to pick one, it would probably be instcombine. I think it’s unlikely this would come up in a context only visible after lowering. Reducing intrinsics is useful when reducing testcases, and any code in the DAG is just going to need to be redone in globalisel eventually

4291

Ideally we would verify wave32 IR somewhere and error, but I don’t think that exists today

[arsenm]

Any update on this?

[foad]

In D65088#1817656, @arsenm wrote:

Any update on this?

No, I haven't done any more work on it. I guess I could pick it up again some time soon, unless anyone else wants to?

[arsenm]

In D65088#1817743, @foad wrote:

In D65088#1817656, @arsenm wrote:

Any update on this?

No, I haven't done any more work on it. I guess I could pick it up again some time soon, unless anyone else wants to?

So far I've put off handling amdgcn.icmp/fcmp in globalisel, hoping this would supersede that so I'd like to see this move along

[nhaehnle]

Agreed, it would be good to see progress on this.

[Flakebi]

Address Nicolai’s comments and implement this as DAG combines and TableGen patterns.

The code generation for test2 is currently not optimal:

%trunc = trunc i32 %x to i1
%ballot = call i64 @llvm.amdgcn.ballot.i64(i1 %trunc)

generates

v_and_b32_e32 v0, 1, v0
v_cmp_eq_u32_e32 vcc, 1, v0
s_and_b64 s[4:5], vcc, exec

where the first compare stems from the truncate.
We could handle the case of (ballot (truncate x)) in the combining. Any opinions on this?

[foad]

In D65088#1916351, @Flakebi wrote:

The code generation for test2 is currently not optimal:

%trunc = trunc i32 %x to i1
%ballot = call i64 @llvm.amdgcn.ballot.i64(i1 %trunc)

generates

v_and_b32_e32 v0, 1, v0
v_cmp_eq_u32_e32 vcc, 1, v0
s_and_b64 s[4:5], vcc, exec

where the first compare stems from the truncate.

I'm confused by this. What is the optimal code generation?

[foad]

llvm/lib/Target/AMDGPU/SIInstructions.td

428 ↗

(On Diff #249563)

If $src is the result of a v_cmp instruction then this s_and will be redundant, won't it? Because v_cmp is defined to return a zero in vcc for all inactive lanes.

llvm/lib/Transforms/InstCombine/InstCombineCalls.cpp

3968

Are you sure it's safe to use read_register(exec) for this? I'm not an expert but I have been told it's not safe, because the compiler doesn't understand that it can't reorder the read_register call past other instructions that modify exec.

[Flakebi]

llvm/lib/Target/AMDGPU/SIInstructions.td

428 ↗

(On Diff #249563)

Yes, I agree that this is redundant.
As this happens in the lowering phase, we can only match for (i64 (int_amdgcn_ballot (i1 (trunc $src)))) in this case. I think we would need to combine machine instructions to get the v_cmp.

llvm/lib/Transforms/InstCombine/InstCombineCalls.cpp

3968

Is there a safe way to get the exec register? I’m fine with removing this optimization here and doing it only in the SDag.

[Flakebi]

In D65088#1916416, @foad wrote:

In D65088#1916351, @Flakebi wrote:

The code generation for test2 is currently not optimal:

%trunc = trunc i32 %x to i1
%ballot = call i64 @llvm.amdgcn.ballot.i64(i1 %trunc)

generates

v_and_b32_e32 v0, 1, v0
v_cmp_eq_u32_e32 vcc, 1, v0
s_and_b64 s[4:5], vcc, exec

where the first compare stems from the truncate.

I'm confused by this. What is the optimal code generation?

The more optimal version would be to merge the compare and s_and exec like you said in your comment?

v_and_b32_e32 v0, 1, v0
v_cmp_eq_u32_e32 s[4:5], 1, v0

[Flakebi]

llvm/lib/Transforms/InstCombine/InstCombineCalls.cpp

3968

Actually, icmp and fcmp already create read_register(exec) in InstCombine. So, having this combination and replacing icmp with ballot won’t change behavior.

[Flakebi]

Use getCopyFromReg(exec) and rebase on fixed whole-wave-mode.

The remaining exec copy can get fixed in multiple ways:

• Prevent sinking the ctpop into the next basic block. Unfortunately marking the instruction as convergent does not work because the Machine Sinking pass moves it instead.
• Reuse the copy of exec that gets created by s_and_saveexec. This might work with GlobalISel, so it’s probably worth to wait until this gets used.

[arsenm]

In D65088#1928542, @Flakebi wrote:

Use getCopyFromReg(exec) and rebase on fixed whole-wave-mode.

The remaining exec copy can get fixed in multiple ways:

• Prevent sinking the ctpop into the next basic block. Unfortunately marking the instruction as convergent does not work because the Machine Sinking pass moves it instead.
• Reuse the copy of exec that gets created by s_and_saveexec. This might work with GlobalISel, so it’s probably worth to wait until this gets used.

I was afraid this would be a problem. We probably need a version of COPY with convergent set. If MachineSinking is moving a convergent instruction, that's also a bug

llvm/lib/Target/AMDGPU/SIISelLowering.cpp

8762

I didn't think of it before, but reading exec this way could potentially be dangerous due to the fact that we have the terrible operations that modify exec in the middle of IR blocks, and we split them later. We might have to do this fold later

llvm/lib/Target/AMDGPU/SIInstructions.td

428 ↗

(On Diff #251027)

Is the COPY_TO_REGCLASS just a tablegen workaround?

433 ↗

(On Diff #251027)

Probably need an explicit Src_b32:$src to make this work with GlobalISel

llvm/lib/Transforms/InstCombine/InstCombineCalls.cpp

3969

What happens for wave32 here?

llvm/test/Transforms/InstCombine/AMDGPU/amdgcn-intrinsics.ll

2412

Wave32 tests also

[Flakebi]

llvm/lib/Target/AMDGPU/SIISelLowering.cpp

8762

What are these operations and how can I fix them?
And shouldn’t this work as the instruction reads exec and thus should not be touched?

llvm/lib/Target/AMDGPU/SIInstructions.td

428 ↗

(On Diff #251027)

Yes, we get an i1 as input though we know that it is stored an sgpr (pair), so we "cast" it into one. The same happens in line 803 to optimize icmp:

def : Pat <
  (i64 (int_amdgcn_icmp i1:$src, (i1 0), (i32 33))),
  (COPY $src) // Return the SGPRs representing i1 src
>;

[Flakebi]

Add missing wave32 instcombining

[arsenm]

Also should add GlobalISel tests

llvm/lib/Target/AMDGPU/SIISelLowering.cpp

8762

You can't really fix them in SelectionDAG. Things like the wwm/wqm intrinsics, and kills can exist in the middle of blocks (plus inline asm)

[Flakebi]

This comment was removed by Flakebi.

[Flakebi]

I removed the COPY_TO_REGCLASS, it looked flaky and does not work with GlobalISel.
Instead, the ballot intrinsic is morphed into an AMDGPUISD::SETCC. Compares are the only reasonable way to get a boolean value into the wavefront form as an i32/i64 and use it in LLVM.

Matt: How do I add these combines for GlobalISel?
Should they go into AMDGPUCombine.td? And do I have to write functions for this or does it work purely with pattern matching?

[arsenm]

In D65088#1933161, @Flakebi wrote:

I removed the COPY_TO_REGCLASS, it looked flaky and does not work with GlobalISel.
Instead, the ballot intrinsic is morphed into an AMDGPUISD::SETCC. Compares are the only reasonable way to get a boolean value into the wavefront form as an i32/i64 and use it in LLVM.

Matt: How do I add these combines for GlobalISel?
Should they go into AMDGPUCombine.td? And do I have to write functions for this or does it work purely with pattern matching?

If you're not going to have a fallback pattern in tablegen, then the GlobalISel support is probably a separate patch. This would not go in AMDGPUCombines.td as this is not an optimization combine. By having everything go through AMDGPUISD::SETCC, this develops the same problem the existing icmp/fcmp intrinsics have for GlobalISel support. As a separate patch, we need to replace how AMDGPUISD::SETCC works to avoid relying on the CondCode fiel

llvm/lib/Target/AMDGPU/SIISelLowering.cpp

8771–8773

Since you're now treating this as the full lowering path, and not an optimization for the special case with a compare, this should go in LowerINTRINSIC_WO_CHAIN

llvm/lib/Transforms/InstCombine/InstCombineCalls.cpp

3967–3968

Should move the getDeclaration below to where you actually use it. If you happened to hit the other size break case, you would have introduced a dead declaration

[Flakebi]

Move the code to lowering again, I’m back were Jay started.
Report a fatal error if the size is neither i32 nor i64.

[Flakebi]

return instead of report_fatal_error

[arsenm]

llvm/lib/Target/AMDGPU/AMDGPUTargetTransformInfo.cpp

807

Can you add a test for this in test/Analysis/DivergenceAnalysis/AMDGPU

llvm/lib/Target/AMDGPU/SIISelLowering.cpp

4299–4301

DAG.getSetCC?

[Flakebi]

Add uniformity test

llvm/lib/Target/AMDGPU/SIISelLowering.cpp

4299–4301

This creates AMDGPUISD::SETCC not setcc :)

llvm/include/llvm/IR/IntrinsicsAMDGPU.td

 def int_amdgcn_icmp :
   Intrinsic<[llvm_anyint_ty], [llvm_anyint_ty, LLVMMatchType<1>, llvm_i32_ty],
             [IntrNoMem, IntrConvergent, ImmArg<2>]>;
 
 def int_amdgcn_fcmp :
   Intrinsic<[llvm_anyint_ty], [llvm_anyfloat_ty, LLVMMatchType<1>, llvm_i32_ty],
             [IntrNoMem, IntrConvergent, ImmArg<2>]>;
 
+def int_amdgcn_ballot :
+  Intrinsic<[llvm_anyint_ty], [llvm_i1_ty], [IntrNoMem, IntrConvergent]>;
+
 def int_amdgcn_readfirstlane :
   GCCBuiltin<"__builtin_amdgcn_readfirstlane">,
   Intrinsic<[llvm_i32_ty], [llvm_i32_ty], [IntrNoMem, IntrConvergent]>;
 
 // The lane argument must be uniform across the currently active threads of the
 // current wave. Otherwise, the result is undefined.
 def int_amdgcn_readlane :
   GCCBuiltin<"__builtin_amdgcn_readlane">,

llvm/lib/Target/AMDGPU/AMDGPUAtomicOptimizer.cpp

void AMDGPUAtomicOptimizer::optimizeAtomic(Instruction &I,
 
   // This is the value in the atomic operation we need to combine in order to
   // reduce the number of atomic operations.
   Value *const V = I.getOperand(ValIdx);
 
   // We need to know how many lanes are active within the wavefront, and we do
   // this by doing a ballot of active lanes.
   Type *const WaveTy = B.getIntNTy(ST->getWavefrontSize());
-  CallInst *const Ballot = B.CreateIntrinsic(
-      Intrinsic::amdgcn_icmp, {WaveTy, B.getInt32Ty()},
-      {B.getInt32(1), B.getInt32(0), B.getInt32(CmpInst::ICMP_NE)});
+  CallInst *const Ballot =
+      B.CreateIntrinsic(Intrinsic::amdgcn_ballot, WaveTy, B.getTrue());
 
   // We need to know how many lanes are active within the wavefront that are
   // below us. If we counted each lane linearly starting from 0, a lane is
   // below us only if its associated index was less than ours. We do this by
   // using the mbcnt intrinsic.
   Value *Mbcnt;
   if (ST->isWave32()) {
     Mbcnt = B.CreateIntrinsic(Intrinsic::amdgcn_mbcnt_lo, {},

llvm/lib/Target/AMDGPU/AMDGPUTargetTransformInfo.cpp

bool GCNTTIImpl::isAlwaysUniform(const Value *V) const {
   if (const IntrinsicInst *Intrinsic = dyn_cast<IntrinsicInst>(V)) {
     switch (Intrinsic->getIntrinsicID()) {
     default:
       return false;
     case Intrinsic::amdgcn_readfirstlane:
     case Intrinsic::amdgcn_readlane:
     case Intrinsic::amdgcn_icmp:
     case Intrinsic::amdgcn_fcmp:
+    case Intrinsic::amdgcn_ballot:
     case Intrinsic::amdgcn_if_break:
       return true;
     }
   }
 
   if (const CallInst *CI = dyn_cast<CallInst>(V)) {
     if (isa<InlineAsm>(CI->getCalledValue()))
       return !isInlineAsmSourceOfDivergence(CI);

llvm/lib/Target/AMDGPU/SIISelLowering.cpp

static SDValue lowerFCMPIntrinsic(const SITargetLowering &TLI,
   EVT CCVT = EVT::getIntegerVT(*DAG.getContext(), WavefrontSize);
   SDValue SetCC = DAG.getNode(AMDGPUISD::SETCC, SL, CCVT, Src0,
                               Src1, DAG.getCondCode(CCOpcode));
   if (VT.bitsEq(CCVT))
     return SetCC;
   return DAG.getZExtOrTrunc(SetCC, SL, VT);
 }
 
+static SDValue lowerBALLOTIntrinsic(const SITargetLowering &TLI, SDNode *N,
+                                    SelectionDAG &DAG) {
+  EVT VT = N->getValueType(0);
+  SDValue Src = N->getOperand(1);
+  SDLoc SL(N);
+
+  if (Src.getOpcode() == ISD::SETCC) {
+    // (ballot (ISD::SETCC ...)) -> (AMDGPUISD::SETCC ...)
+    return DAG.getNode(AMDGPUISD::SETCC, SL, VT, Src.getOperand(0),
+                       Src.getOperand(1), Src.getOperand(2));
+  }
+  if (const ConstantSDNode *Arg = dyn_cast<ConstantSDNode>(Src)) {
+    // (ballot 0) -> 0
+    if (Arg->isNullValue())
+      return DAG.getConstant(0, SL, VT);
+
+    // (ballot 1) -> EXEC/EXEC_LO
+    if (Arg->isOne()) {
+      Register Exec;
+      if (VT.getScalarSizeInBits() == 32)
+        Exec = AMDGPU::EXEC_LO;
+      else if (VT.getScalarSizeInBits() == 64)
+        Exec = AMDGPU::EXEC;
+      else
+        return SDValue();
+
+      return DAG.getCopyFromReg(DAG.getEntryNode(), SL, Exec, VT);
+    }
+  }
+
+  // (ballot (i1 $src)) -> (AMDGPUISD::SETCC (i32 (zext $src)) (i32 0)
+  // ISD::SETNE)
+  return DAG.getNode(
+      AMDGPUISD::SETCC, SL, VT, DAG.getZExtOrTrunc(Src, SL, MVT::i32),
+      DAG.getConstant(0, SL, MVT::i32), DAG.getCondCode(ISD::SETNE));
+}
+
 void SITargetLowering::ReplaceNodeResults(SDNode *N,
                                           SmallVectorImpl<SDValue> &Results,
                                           SelectionDAG &DAG) const {
   switch (N->getOpcode()) {
   case ISD::INSERT_VECTOR_ELT: {
     if (SDValue Res = lowerINSERT_VECTOR_ELT(SDValue(N, 0), DAG))
       Results.push_back(Res);
     return;
if (Op.getOperand(1).getValueType() == MVT::i1 &&
         Op.getConstantOperandVal(2) == 0 &&
         Op.getConstantOperandVal(3) == ICmpInst::Predicate::ICMP_NE)
       return Op;
     return lowerICMPIntrinsic(*this, Op.getNode(), DAG);
   }
   case Intrinsic::amdgcn_fcmp: {
     return lowerFCMPIntrinsic(*this, Op.getNode(), DAG);
   }
+  case Intrinsic::amdgcn_ballot:
+    return lowerBALLOTIntrinsic(*this, Op.getNode(), DAG);
   case Intrinsic::amdgcn_fmed3:
     return DAG.getNode(AMDGPUISD::FMED3, DL, VT,
                        Op.getOperand(1), Op.getOperand(2), Op.getOperand(3));
   case Intrinsic::amdgcn_fdot2:
     return DAG.getNode(AMDGPUISD::FDOT2, DL, VT,
                        Op.getOperand(1), Op.getOperand(2), Op.getOperand(3),
                        Op.getOperand(4));
   case Intrinsic::amdgcn_fmul_legacy:
bool SITargetLowering::isCanonicalized(SelectionDAG &DAG, SDValue Op,
 
   switch (Opcode) {
   // These will flush denorms if required.
   case ISD::FADD:
   case ISD::FSUB:
   case ISD::FMUL:
   case ISD::FCEIL:
   case ISD::FFLOOR:
   case ISD::FMA:
   case ISD::FMAD:
   case ISD::FSQRT:
   case ISD::FDIV:
   case ISD::FREM:
   case ISD::FP_ROUND:
   case ISD::FP_EXTEND:
   case AMDGPUISD::FMUL_LEGACY:
   case AMDGPUISD::FMAD_FTZ:
   case AMDGPUISD::RCP:
   case AMDGPUISD::RSQ:
   case AMDGPUISD::RSQ_CLAMP:
   case AMDGPUISD::RCP_LEGACY:
   case AMDGPUISD::RSQ_LEGACY:
   case AMDGPUISD::RCP_IFLAG:
   case AMDGPUISD::TRIG_PREOP:
   case AMDGPUISD::DIV_SCALE:
   case AMDGPUISD::DIV_FMAS:
   case AMDGPUISD::DIV_FIXUP:
   case AMDGPUISD::FRACT:

llvm/lib/Transforms/InstCombine/InstCombineCalls.cpp

if (match(Src1, m_Zero()) &&
                         ConstantInt::get(CC->getType(), SrcPred) };
       CallInst *NewCall = Builder.CreateCall(NewF, Args);
       NewCall->takeName(II);
       return replaceInstUsesWith(*II, NewCall);
     }
 
     break;
   }
+  case Intrinsic::amdgcn_ballot: {
+    if (auto *Src = dyn_cast<ConstantInt>(II->getArgOperand(0))) {
+      if (Src->isZero()) {
+        // amdgcn.ballot(i1 0) is zero.
+        return replaceInstUsesWith(*II, Constant::getNullValue(II->getType()));
+      }
+
+      if (Src->isOne()) {
+        // amdgcn.ballot(i1 1) is exec.
+        const char *RegName = "exec";
+        if (II->getType()->isIntegerTy(32))
+          RegName = "exec_lo";
+        else if (!II->getType()->isIntegerTy(64))
+          break;
+
+        Function *NewF = Intrinsic::getDeclaration(
+            II->getModule(), Intrinsic::read_register, II->getType());
+        Metadata *MDArgs[] = {MDString::get(II->getContext(), RegName)};
+        MDNode *MD = MDNode::get(II->getContext(), MDArgs);
+        Value *Args[] = {MetadataAsValue::get(II->getContext(), MD)};
+        CallInst *NewCall = Builder.CreateCall(NewF, Args);
+        NewCall->addAttribute(AttributeList::FunctionIndex,
+                              Attribute::Convergent);
+        NewCall->takeName(II);
+        return replaceInstUsesWith(*II, NewCall);
+      }
+    }
+    break;
+  }
   case Intrinsic::amdgcn_wqm_vote: {
     // wqm_vote is identity when the argument is constant.
     if (!isa<Constant>(II->getArgOperand(0)))
       break;
 
     return replaceInstUsesWith(*II, II->getArgOperand(0));
   }
   case Intrinsic::amdgcn_kill: {
case Intrinsic::experimental_gc_relocate: {
     // If we have two copies of the same pointer in the statepoint argument
     // list, canonicalize to one.  This may let us common gc.relocates.
     if (GCR.getBasePtr() == GCR.getDerivedPtr() &&
         GCR.getBasePtrIndex() != GCR.getDerivedPtrIndex()) {
       auto *OpIntTy = GCR.getOperand(2)->getType();
       return replaceOperand(*II, 2,
           ConstantInt::get(OpIntTy, GCR.getBasePtrIndex()));
     }
-    
     // Translate facts known about a pointer before relocating into
     // facts about the relocate value, while being careful to
     // preserve relocation semantics.
     Value *DerivedPtr = GCR.getDerivedPtr();
 
     // Remove the relocation if unused, note that this check is required
     // to prevent the cases below from looping forever.
     if (II->use_empty())

llvm/test/Analysis/DivergenceAnalysis/AMDGPU/always_uniform.ll

 
 ; CHECK-LABEL: for function 'fcmp':
 define amdgpu_kernel void @fcmp(float inreg %x, float inreg %y) {
 ; CHECK-NOT: DIVERGENT:  %fcmp = call i64 @llvm.amdgcn.fcmp.i32
   %fcmp = call i64 @llvm.amdgcn.fcmp.i32(float %x, float %y, i32 33)
   ret void
 }
 
+; CHECK-LABEL: for function 'ballot':
+define amdgpu_kernel void @ballot(i1 inreg %x) {
+; CHECK-NOT: DIVERGENT:  %ballot = call i64 @llvm.amdgcn.ballot.i32
+  %ballot = call i64 @llvm.amdgcn.ballot.i32(i1 %x)
+  ret void
+}
+
 ; SGPR asm outputs are uniform regardless of the input operands.
 ; CHECK-LABEL: for function 'asm_sgpr':
 ; CHECK: DIVERGENT: i32 %divergent
 ; CHECK-NOT: DIVERGENT
 define i32 @asm_sgpr(i32 %divergent) {
   %sgpr = call i32 asm "; def $0, $1","=s,v"(i32 %divergent)
   ret i32 %sgpr
 }
define void @asm_mixed_sgpr_vgpr(i32 %divergent) {
   store i32 %vgpr, i32 addrspace(1)* undef
   ret void
 }
 
 declare i32 @llvm.amdgcn.workitem.id.x() #0
 declare i32 @llvm.amdgcn.readfirstlane(i32) #0
 declare i64 @llvm.amdgcn.icmp.i32(i32, i32, i32) #1
 declare i64 @llvm.amdgcn.fcmp.i32(float, float, i32) #1
+declare i64 @llvm.amdgcn.ballot.i32(i1) #1
 
 attributes #0 = { nounwind readnone }
 attributes #1 = { nounwind readnone convergent }

llvm/test/CodeGen/AMDGPU/atomic_optimizations_buffer.ll

 ; RUN: llc -march=amdgcn -mtriple=amdgcn---amdgiz -amdgpu-atomic-optimizations=true -verify-machineinstrs < %s | FileCheck -enable-var-scope -check-prefixes=GCN,GCN64,GFX7LESS %s
 ; RUN: llc -march=amdgcn -mtriple=amdgcn---amdgiz -mcpu=tonga -mattr=-flat-for-global -amdgpu-atomic-optimizations=true -amdgpu-dpp-combine=false -verify-machineinstrs < %s | FileCheck -enable-var-scope -check-prefixes=GCN,GCN64,GFX8MORE,GFX8MORE64,GFX89 %s
 ; RUN: llc -march=amdgcn -mtriple=amdgcn---amdgiz -mcpu=gfx900 -mattr=-flat-for-global -amdgpu-atomic-optimizations=true -amdgpu-dpp-combine=false -verify-machineinstrs < %s | FileCheck -enable-var-scope -check-prefixes=GCN,GCN64,GFX8MORE,GFX8MORE64,GFX89 %s
 ; RUN: llc -march=amdgcn -mtriple=amdgcn---amdgiz -mcpu=gfx1010 -mattr=-wavefrontsize32,+wavefrontsize64 -mattr=-flat-for-global -amdgpu-atomic-optimizations=true -amdgpu-dpp-combine=false -verify-machineinstrs < %s | FileCheck -enable-var-scope -check-prefixes=GCN,GCN64,GFX8MORE,GFX8MORE64 %s
 ; RUN: llc -march=amdgcn -mtriple=amdgcn---amdgiz -mcpu=gfx1010 -mattr=+wavefrontsize32,-wavefrontsize64 -mattr=-flat-for-global -amdgpu-atomic-optimizations=true -amdgpu-dpp-combine=false -verify-machineinstrs < %s | FileCheck -enable-var-scope -check-prefixes=GCN,GCN32,GFX8MORE,GFX8MORE32 %s
 
 declare i32 @llvm.amdgcn.workitem.id.x()
 declare i32 @llvm.amdgcn.raw.buffer.atomic.add(i32, <4 x i32>, i32, i32, i32 immarg)
 declare i32 @llvm.amdgcn.struct.buffer.atomic.add(i32, <4 x i32>, i32, i32, i32, i32 immarg)
 declare i32 @llvm.amdgcn.raw.buffer.atomic.sub(i32, <4 x i32>, i32, i32, i32 immarg)
 
-; Show that what the atomic optimization pass will do for raw buffers.
+; Show what the atomic optimization pass will do for raw buffers.
 
 ; GCN-LABEL: add_i32_constant:
-; GCN32: v_cmp_ne_u32_e64 s[[exec_lo:[0-9]+]], 1, 0
-; GCN64: v_cmp_ne_u32_e64 s{{\[}}[[exec_lo:[0-9]+]]:[[exec_hi:[0-9]+]]{{\]}}, 1, 0
+; GCN32: s_mov_b32 s[[exec_lo:[0-9]+]], exec_lo
+; GCN64: s_mov_b64 s{{\[}}[[exec_lo:[0-9]+]]:[[exec_hi:[0-9]+]]{{\]}}, exec
 ; GCN: v_mbcnt_lo_u32_b32{{(_e[0-9]+)?}} v[[mbcnt:[0-9]+]], s[[exec_lo]], 0
 ; GCN64: v_mbcnt_hi_u32_b32{{(_e[0-9]+)?}} v[[mbcnt]], s[[exec_hi]], v[[mbcnt]]
 ; GCN: v_cmp_eq_u32{{(_e[0-9]+)?}} vcc{{(_lo)?}}, 0, v[[mbcnt]]
+; GCN: s_and_saveexec_b{{32|64}} s[[exec:\[?[0-9:]+\]?]], vcc
 ; GCN32: s_bcnt1_i32_b32 s[[popcount:[0-9]+]], s[[exec_lo]]
 ; GCN64: s_bcnt1_i32_b64 s[[popcount:[0-9]+]], s{{\[}}[[exec_lo]]:[[exec_hi]]{{\]}}
 ; GCN: v_mul_u32_u24{{(_e[0-9]+)?}} v[[value:[0-9]+]], s[[popcount]], 5
 ; GCN: buffer_atomic_add v[[value]]
 define amdgpu_kernel void @add_i32_constant(i32 addrspace(1)* %out, <4 x i32> %inout) {
 entry:
   %old = call i32 @llvm.amdgcn.raw.buffer.atomic.add(i32 5, <4 x i32> %inout, i32 0, i32 0, i32 0)
   store i32 %old, i32 addrspace(1)* %out
   ret void
 }
 
 ; GCN-LABEL: add_i32_uniform:
-; GCN32: v_cmp_ne_u32_e64 s[[exec_lo:[0-9]+]], 1, 0
-; GCN64: v_cmp_ne_u32_e64 s{{\[}}[[exec_lo:[0-9]+]]:[[exec_hi:[0-9]+]]{{\]}}, 1, 0
+; GCN32: s_mov_b32 s[[exec_lo:[0-9]+]], exec_lo
+; GCN64: s_mov_b64 s{{\[}}[[exec_lo:[0-9]+]]:[[exec_hi:[0-9]+]]{{\]}}, exec
 ; GCN: v_mbcnt_lo_u32_b32{{(_e[0-9]+)?}} v[[mbcnt:[0-9]+]], s[[exec_lo]], 0
 ; GCN64: v_mbcnt_hi_u32_b32{{(_e[0-9]+)?}} v[[mbcnt]], s[[exec_hi]], v[[mbcnt]]
 ; GCN: v_cmp_eq_u32{{(_e[0-9]+)?}} vcc{{(_lo)?}}, 0, v[[mbcnt]]
+; GCN: s_and_saveexec_b{{32|64}} s[[exec:\[?[0-9:]+\]?]], vcc
 ; GCN32: s_bcnt1_i32_b32 s[[popcount:[0-9]+]], s[[exec_lo]]
 ; GCN64: s_bcnt1_i32_b64 s[[popcount:[0-9]+]], s{{\[}}[[exec_lo]]:[[exec_hi]]{{\]}}
 ; GCN: s_mul_i32 s[[scalar_value:[0-9]+]], s{{[0-9]+}}, s[[popcount]]
 ; GCN: v_mov_b32{{(_e[0-9]+)?}} v[[value:[0-9]+]], s[[scalar_value]]
 ; GCN: buffer_atomic_add v[[value]]
 define amdgpu_kernel void @add_i32_uniform(i32 addrspace(1)* %out, <4 x i32> %inout, i32 %additive) {
 entry:
   %old = call i32 @llvm.amdgcn.raw.buffer.atomic.add(i32 %additive, <4 x i32> %inout, i32 0, i32 0, i32 0)

 entry:
   %lane = call i32 @llvm.amdgcn.workitem.id.x()
   %old = call i32 @llvm.amdgcn.raw.buffer.atomic.add(i32 1, <4 x i32> %inout, i32 %lane, i32 0, i32 0)
   store i32 %old, i32 addrspace(1)* %out
   ret void
 }
 
 ; GCN-LABEL: sub_i32_constant:
-; GCN32: v_cmp_ne_u32_e64 s[[exec_lo:[0-9]+]], 1, 0
-; GCN64: v_cmp_ne_u32_e64 s{{\[}}[[exec_lo:[0-9]+]]:[[exec_hi:[0-9]+]]{{\]}}, 1, 0
+; GCN32: s_mov_b32 s[[exec_lo:[0-9]+]], exec_lo
+; GCN64: s_mov_b64 s{{\[}}[[exec_lo:[0-9]+]]:[[exec_hi:[0-9]+]]{{\]}}, exec
 ; GCN: v_mbcnt_lo_u32_b32{{(_e[0-9]+)?}} v[[mbcnt:[0-9]+]], s[[exec_lo]], 0
 ; GCN64: v_mbcnt_hi_u32_b32{{(_e[0-9]+)?}} v[[mbcnt]], s[[exec_hi]], v[[mbcnt]]
 ; GCN: v_cmp_eq_u32{{(_e[0-9]+)?}} vcc{{(_lo)?}}, 0, v[[mbcnt]]
+; GCN: s_and_saveexec_b{{32|64}} s[[exec:\[?[0-9:]+\]?]], vcc
 ; GCN32: s_bcnt1_i32_b32 s[[popcount:[0-9]+]], s[[exec_lo]]
 ; GCN64: s_bcnt1_i32_b64 s[[popcount:[0-9]+]], s{{\[}}[[exec_lo]]:[[exec_hi]]{{\]}}
 ; GCN: v_mul_u32_u24{{(_e[0-9]+)?}} v[[value:[0-9]+]], s[[popcount]], 5
 ; GCN: buffer_atomic_sub v[[value]]
 define amdgpu_kernel void @sub_i32_constant(i32 addrspace(1)* %out, <4 x i32> %inout) {
 entry:
   %old = call i32 @llvm.amdgcn.raw.buffer.atomic.sub(i32 5, <4 x i32> %inout, i32 0, i32 0, i32 0)
   store i32 %old, i32 addrspace(1)* %out
   ret void
 }
 
 ; GCN-LABEL: sub_i32_uniform:
-; GCN32: v_cmp_ne_u32_e64 s[[exec_lo:[0-9]+]], 1, 0
-; GCN64: v_cmp_ne_u32_e64 s{{\[}}[[exec_lo:[0-9]+]]:[[exec_hi:[0-9]+]]{{\]}}, 1, 0
+; GCN32: s_mov_b32 s[[exec_lo:[0-9]+]], exec_lo
+; GCN64: s_mov_b64 s{{\[}}[[exec_lo:[0-9]+]]:[[exec_hi:[0-9]+]]{{\]}}, exec
 ; GCN: v_mbcnt_lo_u32_b32{{(_e[0-9]+)?}} v[[mbcnt:[0-9]+]], s[[exec_lo]], 0
 ; GCN64: v_mbcnt_hi_u32_b32{{(_e[0-9]+)?}} v[[mbcnt]], s[[exec_hi]], v[[mbcnt]]
 ; GCN: v_cmp_eq_u32{{(_e[0-9]+)?}} vcc{{(_lo)?}}, 0, v[[mbcnt]]
+; GCN: s_and_saveexec_b{{32|64}} s[[exec:\[?[0-9:]+\]?]], vcc
 ; GCN32: s_bcnt1_i32_b32 s[[popcount:[0-9]+]], s[[exec_lo]]
 ; GCN64: s_bcnt1_i32_b64 s[[popcount:[0-9]+]], s{{\[}}[[exec_lo]]:[[exec_hi]]{{\]}}
 ; GCN: s_mul_i32 s[[scalar_value:[0-9]+]], s{{[0-9]+}}, s[[popcount]]
 ; GCN: v_mov_b32{{(_e[0-9]+)?}} v[[value:[0-9]+]], s[[scalar_value]]
 ; GCN: buffer_atomic_sub v[[value]]
 define amdgpu_kernel void @sub_i32_uniform(i32 addrspace(1)* %out, <4 x i32> %inout, i32 %subitive) {
 entry:
   %old = call i32 @llvm.amdgcn.raw.buffer.atomic.sub(i32 %subitive, <4 x i32> %inout, i32 0, i32 0, i32 0)

llvm/test/CodeGen/AMDGPU/atomic_optimizations_global_pointer.ll

 ; RUN: llc -march=amdgcn -mtriple=amdgcn---amdgiz -amdgpu-atomic-optimizations=true -verify-machineinstrs < %s | FileCheck -enable-var-scope -check-prefixes=GCN,GCN64,GFX7LESS %s
 ; RUN: llc -march=amdgcn -mtriple=amdgcn---amdgiz -mcpu=tonga -mattr=-flat-for-global -amdgpu-atomic-optimizations=true -verify-machineinstrs < %s | FileCheck -enable-var-scope -check-prefixes=GCN,GCN64,GFX8MORE,GFX8MORE64,DPPCOMB %s
 ; RUN: llc -march=amdgcn -mtriple=amdgcn---amdgiz -mcpu=gfx900 -mattr=-flat-for-global -amdgpu-atomic-optimizations=true -verify-machineinstrs < %s | FileCheck -enable-var-scope -check-prefixes=GCN,GCN64,GFX8MORE,GFX8MORE64,DPPCOMB %s
 ; RUN: llc -march=amdgcn -mtriple=amdgcn---amdgiz -mcpu=gfx1010 -mattr=-wavefrontsize32,+wavefrontsize64 -mattr=-flat-for-global -amdgpu-atomic-optimizations=true -verify-machineinstrs < %s | FileCheck -enable-var-scope -check-prefixes=GCN,GCN64,GFX8MORE,GFX8MORE64 %s
 ; RUN: llc -march=amdgcn -mtriple=amdgcn---amdgiz -mcpu=gfx1010 -mattr=+wavefrontsize32,-wavefrontsize64 -mattr=-flat-for-global -amdgpu-atomic-optimizations=true -verify-machineinstrs < %s | FileCheck -enable-var-scope -check-prefixes=GCN,GCN32,GFX8MORE,GFX8MORE32 %s
 
 declare i32 @llvm.amdgcn.workitem.id.x()
 
-; Show that what the atomic optimization pass will do for global pointers.
+; Show what the atomic optimization pass will do for global pointers.
 
 ; GCN-LABEL: add_i32_constant:
-; GCN32: v_cmp_ne_u32_e64 s[[exec_lo:[0-9]+]], 1, 0
-; GCN64: v_cmp_ne_u32_e64 s{{\[}}[[exec_lo:[0-9]+]]:[[exec_hi:[0-9]+]]{{\]}}, 1, 0
+; GCN32: s_mov_b32 s[[exec_lo:[0-9]+]], exec_lo
+; GCN64: s_mov_b64 s{{\[}}[[exec_lo:[0-9]+]]:[[exec_hi:[0-9]+]]{{\]}}, exec
 ; GCN: v_mbcnt_lo_u32_b32{{(_e[0-9]+)?}} v[[mbcnt:[0-9]+]], s[[exec_lo]], 0
 ; GCN64: v_mbcnt_hi_u32_b32{{(_e[0-9]+)?}} v[[mbcnt]], s[[exec_hi]], v[[mbcnt]]
 ; GCN: v_cmp_eq_u32{{(_e[0-9]+)?}} vcc{{(_lo)?}}, 0, v[[mbcnt]]
+; GCN: s_and_saveexec_b{{32|64}} s[[exec:\[?[0-9:]+\]?]], vcc
 ; GCN32: s_bcnt1_i32_b32 s[[popcount:[0-9]+]], s[[exec_lo]]
 ; GCN64: s_bcnt1_i32_b64 s[[popcount:[0-9]+]], s{{\[}}[[exec_lo]]:[[exec_hi]]{{\]}}
 ; GCN: v_mul_u32_u24{{(_e[0-9]+)?}} v[[value:[0-9]+]], s[[popcount]], 5
 ; GCN: {{flat|buffer|global}}_atomic_add v[[value]]
 define amdgpu_kernel void @add_i32_constant(i32 addrspace(1)* %out, i32 addrspace(1)* %inout) {
 entry:
   %old = atomicrmw add i32 addrspace(1)* %inout, i32 5 acq_rel
   store i32 %old, i32 addrspace(1)* %out
   ret void
 }
 
 ; GCN-LABEL: add_i32_uniform:
-; GCN32: v_cmp_ne_u32_e64 s[[exec_lo:[0-9]+]], 1, 0
-; GCN64: v_cmp_ne_u32_e64 s{{\[}}[[exec_lo:[0-9]+]]:[[exec_hi:[0-9]+]]{{\]}}, 1, 0
+; GCN32: s_mov_b32 s[[exec_lo:[0-9]+]], exec_lo
+; GCN64: s_mov_b64 s{{\[}}[[exec_lo:[0-9]+]]:[[exec_hi:[0-9]+]]{{\]}}, exec
 ; GCN: v_mbcnt_lo_u32_b32{{(_e[0-9]+)?}} v[[mbcnt:[0-9]+]], s[[exec_lo]], 0
 ; GCN64: v_mbcnt_hi_u32_b32{{(_e[0-9]+)?}} v[[mbcnt]], s[[exec_hi]], v[[mbcnt]]
 ; GCN: v_cmp_eq_u32{{(_e[0-9]+)?}} vcc{{(_lo)?}}, 0, v[[mbcnt]]
+; GCN: s_and_saveexec_b{{32|64}} s[[exec:\[?[0-9:]+\]?]], vcc
 ; GCN32: s_bcnt1_i32_b32 s[[popcount:[0-9]+]], s[[exec_lo]]
 ; GCN64: s_bcnt1_i32_b64 s[[popcount:[0-9]+]], s{{\[}}[[exec_lo]]:[[exec_hi]]{{\]}}
 ; GCN: s_mul_i32 s[[scalar_value:[0-9]+]], s{{[0-9]+}}, s[[popcount]]
 ; GCN: v_mov_b32{{(_e[0-9]+)?}} v[[value:[0-9]+]], s[[scalar_value]]
 ; GCN: {{flat|buffer|global}}_atomic_add v[[value]]
 define amdgpu_kernel void @add_i32_uniform(i32 addrspace(1)* %out, i32 addrspace(1)* %inout, i32 %additive) {
 entry:
   %old = atomicrmw add i32 addrspace(1)* %inout, i32 %additive acq_rel

 entry:
   %lane = call i32 @llvm.amdgcn.workitem.id.x()
   %old = atomicrmw add i32 addrspace(1)* %inout, i32 %lane acq_rel
   store i32 %old, i32 addrspace(1)* %out
   ret void
 }
 
 ; GCN-LABEL: add_i64_constant:
-; GCN32: v_cmp_ne_u32_e64 s[[exec_lo:[0-9]+]], 1, 0
-; GCN64: v_cmp_ne_u32_e64 s{{\[}}[[exec_lo:[0-9]+]]:[[exec_hi:[0-9]+]]{{\]}}, 1, 0
+; GCN32: s_mov_b32 s[[exec_lo:[0-9]+]], exec_lo
+; GCN64: s_mov_b64 s{{\[}}[[exec_lo:[0-9]+]]:[[exec_hi:[0-9]+]]{{\]}}, exec
 ; GCN: v_mbcnt_lo_u32_b32{{(_e[0-9]+)?}} v[[mbcnt:[0-9]+]], s[[exec_lo]], 0
 ; GCN64: v_mbcnt_hi_u32_b32{{(_e[0-9]+)?}} v[[mbcnt]], s[[exec_hi]], v[[mbcnt]]
 ; GCN: v_cmp_eq_u32{{(_e[0-9]+)?}} vcc{{(_lo)?}}, 0, v[[mbcnt]]
+; GCN: s_and_saveexec_b{{32|64}} s[[exec:\[?[0-9:]+\]?]], vcc
 ; GCN32: s_bcnt1_i32_b32 s[[popcount:[0-9]+]], s[[exec_lo]]
 ; GCN64: s_bcnt1_i32_b64 s[[popcount:[0-9]+]], s{{\[}}[[exec_lo]]:[[exec_hi]]{{\]}}
 ; GCN: v_mul_hi_u32_u24{{(_e[0-9]+)?}} v[[value_hi:[0-9]+]], s[[popcount]], 5
 ; GCN: v_mul_u32_u24{{(_e[0-9]+)?}} v[[value_lo:[0-9]+]], s[[popcount]], 5
 ; GCN: {{flat|buffer|global}}_atomic_add_x2 v{{\[}}[[value_lo]]:[[value_hi]]{{\]}}
 define amdgpu_kernel void @add_i64_constant(i64 addrspace(1)* %out, i64 addrspace(1)* %inout) {
 entry:
   %old = atomicrmw add i64 addrspace(1)* %inout, i64 5 acq_rel
   store i64 %old, i64 addrspace(1)* %out
   ret void
 }
 
 ; GCN-LABEL: add_i64_uniform:
-; GCN32: v_cmp_ne_u32_e64 s[[exec_lo:[0-9]+]], 1, 0
-; GCN64: v_cmp_ne_u32_e64 s{{\[}}[[exec_lo:[0-9]+]]:[[exec_hi:[0-9]+]]{{\]}}, 1, 0
+; GCN32: s_mov_b32 s[[exec_lo:[0-9]+]], exec_lo
+; GCN64: s_mov_b64 s{{\[}}[[exec_lo:[0-9]+]]:[[exec_hi:[0-9]+]]{{\]}}, exec
 ; GCN: v_mbcnt_lo_u32_b32{{(_e[0-9]+)?}} v[[mbcnt:[0-9]+]], s[[exec_lo]], 0
 ; GCN64: v_mbcnt_hi_u32_b32{{(_e[0-9]+)?}} v[[mbcnt]], s[[exec_hi]], v[[mbcnt]]
 ; GCN: v_cmp_eq_u32{{(_e[0-9]+)?}} vcc{{(_lo)?}}, 0, v[[mbcnt]]
-; GCN32: s_bcnt1_i32_b32 s{{[0-9]+}}, s[[exec_lo]]
-; GCN64: s_bcnt1_i32_b64 s{{[0-9]+}}, s{{\[}}[[exec_lo]]:[[exec_hi]]{{\]}}
+; GCN: s_and_saveexec_b{{32|64}} s[[exec:\[?[0-9:]+\]?]], vcc
+; GCN32: s_bcnt1_i32_b32 s[[popcount:[0-9]+]], s[[exec_lo]]
+; GCN64: s_bcnt1_i32_b64 s[[popcount:[0-9]+]], s{{\[}}[[exec_lo]]:[[exec_hi]]{{\]}}
 ; GCN: {{flat|buffer|global}}_atomic_add_x2 v{{\[}}{{[0-9]+}}:{{[0-9]+}}{{\]}}
 define amdgpu_kernel void @add_i64_uniform(i64 addrspace(1)* %out, i64 addrspace(1)* %inout, i64 %additive) {
 entry:
   %old = atomicrmw add i64 addrspace(1)* %inout, i64 %additive acq_rel
   store i64 %old, i64 addrspace(1)* %out
   ret void
 }
 
 ; GCN-LABEL: add_i64_varying:
 ; GCN-NOT: v_mbcnt_lo_u32_b32
 ; GCN-NOT: v_mbcnt_hi_u32_b32
 ; GCN-NOT: s_bcnt1_i32_b64
 ; GCN: {{flat|buffer|global}}_atomic_add_x2 v{{\[}}{{[0-9]+}}:{{[0-9]+}}{{\]}}
 define amdgpu_kernel void @add_i64_varying(i64 addrspace(1)* %out, i64 addrspace(1)* %inout) {
 entry:
   %lane = call i32 @llvm.amdgcn.workitem.id.x()
   %zext = zext i32 %lane to i64
   %old = atomicrmw add i64 addrspace(1)* %inout, i64 %zext acq_rel
   store i64 %old, i64 addrspace(1)* %out
   ret void
 }
 
 ; GCN-LABEL: sub_i32_constant:
-; GCN32: v_cmp_ne_u32_e64 s[[exec_lo:[0-9]+]], 1, 0
-; GCN64: v_cmp_ne_u32_e64 s{{\[}}[[exec_lo:[0-9]+]]:[[exec_hi:[0-9]+]]{{\]}}, 1, 0
+; GCN32: s_mov_b32 s[[exec_lo:[0-9]+]], exec_lo
+; GCN64: s_mov_b64 s{{\[}}[[exec_lo:[0-9]+]]:[[exec_hi:[0-9]+]]{{\]}}, exec
 ; GCN: v_mbcnt_lo_u32_b32{{(_e[0-9]+)?}} v[[mbcnt:[0-9]+]], s[[exec_lo]], 0
 ; GCN64: v_mbcnt_hi_u32_b32{{(_e[0-9]+)?}} v[[mbcnt]], s[[exec_hi]], v[[mbcnt]]
 ; GCN: v_cmp_eq_u32{{(_e[0-9]+)?}} vcc{{(_lo)?}}, 0, v[[mbcnt]]
+; GCN: s_and_saveexec_b{{32|64}} s[[exec:\[?[0-9:]+\]?]], vcc
 ; GCN32: s_bcnt1_i32_b32 s[[popcount:[0-9]+]], s[[exec_lo]]
 ; GCN64: s_bcnt1_i32_b64 s[[popcount:[0-9]+]], s{{\[}}[[exec_lo]]:[[exec_hi]]{{\]}}
 ; GCN: v_mul_u32_u24{{(_e[0-9]+)?}} v[[value:[0-9]+]], s[[popcount]], 5
 ; GCN: {{flat|buffer|global}}_atomic_sub v[[value]]
 define amdgpu_kernel void @sub_i32_constant(i32 addrspace(1)* %out, i32 addrspace(1)* %inout) {
 entry:
   %old = atomicrmw sub i32 addrspace(1)* %inout, i32 5 acq_rel
   store i32 %old, i32 addrspace(1)* %out
   ret void
 }
 
 ; GCN-LABEL: sub_i32_uniform:
-; GCN32: v_cmp_ne_u32_e64 s[[exec_lo:[0-9]+]], 1, 0
-; GCN64: v_cmp_ne_u32_e64 s{{\[}}[[exec_lo:[0-9]+]]:[[exec_hi:[0-9]+]]{{\]}}, 1, 0
+; GCN32: s_mov_b32 s[[exec_lo:[0-9]+]], exec_lo
+; GCN64: s_mov_b64 s{{\[}}[[exec_lo:[0-9]+]]:[[exec_hi:[0-9]+]]{{\]}}, exec
 ; GCN: v_mbcnt_lo_u32_b32{{(_e[0-9]+)?}} v[[mbcnt:[0-9]+]], s[[exec_lo]], 0
 ; GCN64: v_mbcnt_hi_u32_b32{{(_e[0-9]+)?}} v[[mbcnt]], s[[exec_hi]], v[[mbcnt]]
 ; GCN: v_cmp_eq_u32{{(_e[0-9]+)?}} vcc{{(_lo)?}}, 0, v[[mbcnt]]
+; GCN: s_and_saveexec_b{{32|64}} s[[exec:\[?[0-9:]+\]?]], vcc
 ; GCN32: s_bcnt1_i32_b32 s[[popcount:[0-9]+]], s[[exec_lo]]
 ; GCN64: s_bcnt1_i32_b64 s[[popcount:[0-9]+]], s{{\[}}[[exec_lo]]:[[exec_hi]]{{\]}}
 ; GCN: s_mul_i32 s[[scalar_value:[0-9]+]], s{{[0-9]+}}, s[[popcount]]
 ; GCN: v_mov_b32{{(_e[0-9]+)?}} v[[value:[0-9]+]], s[[scalar_value]]
 ; GCN: {{flat|buffer|global}}_atomic_sub v[[value]]
 define amdgpu_kernel void @sub_i32_uniform(i32 addrspace(1)* %out, i32 addrspace(1)* %inout, i32 %subitive) {
 entry:
   %old = atomicrmw sub i32 addrspace(1)* %inout, i32 %subitive acq_rel

 entry:
   %lane = call i32 @llvm.amdgcn.workitem.id.x()
   %old = atomicrmw sub i32 addrspace(1)* %inout, i32 %lane acq_rel
   store i32 %old, i32 addrspace(1)* %out
   ret void
 }
 
 ; GCN-LABEL: sub_i64_constant:
-; GCN32: v_cmp_ne_u32_e64 s[[exec_lo:[0-9]+]], 1, 0
-; GCN64: v_cmp_ne_u32_e64 s{{\[}}[[exec_lo:[0-9]+]]:[[exec_hi:[0-9]+]]{{\]}}, 1, 0
+; GCN32: s_mov_b32 s[[exec_lo:[0-9]+]], exec_lo
+; GCN64: s_mov_b64 s{{\[}}[[exec_lo:[0-9]+]]:[[exec_hi:[0-9]+]]{{\]}}, exec
 ; GCN: v_mbcnt_lo_u32_b32{{(_e[0-9]+)?}} v[[mbcnt:[0-9]+]], s[[exec_lo]], 0
 ; GCN64: v_mbcnt_hi_u32_b32{{(_e[0-9]+)?}} v[[mbcnt]], s[[exec_hi]], v[[mbcnt]]
 ; GCN: v_cmp_eq_u32{{(_e[0-9]+)?}} vcc{{(_lo)?}}, 0, v[[mbcnt]]
+; GCN: s_and_saveexec_b{{32|64}} s[[exec:\[?[0-9:]+\]?]], vcc
 ; GCN32: s_bcnt1_i32_b32 s[[popcount:[0-9]+]], s[[exec_lo]]
 ; GCN64: s_bcnt1_i32_b64 s[[popcount:[0-9]+]], s{{\[}}[[exec_lo]]:[[exec_hi]]{{\]}}
 ; GCN: v_mul_hi_u32_u24{{(_e[0-9]+)?}} v[[value_hi:[0-9]+]], s[[popcount]], 5
 ; GCN: v_mul_u32_u24{{(_e[0-9]+)?}} v[[value_lo:[0-9]+]], s[[popcount]], 5
 ; GCN: {{flat|buffer|global}}_atomic_sub_x2 v{{\[}}[[value_lo]]:[[value_hi]]{{\]}}
 define amdgpu_kernel void @sub_i64_constant(i64 addrspace(1)* %out, i64 addrspace(1)* %inout) {
 entry:
   %old = atomicrmw sub i64 addrspace(1)* %inout, i64 5 acq_rel
   store i64 %old, i64 addrspace(1)* %out
   ret void
 }
 
 ; GCN-LABEL: sub_i64_uniform:
-; GCN32: v_cmp_ne_u32_e64 s[[exec_lo:[0-9]+]], 1, 0
-; GCN64: v_cmp_ne_u32_e64 s{{\[}}[[exec_lo:[0-9]+]]:[[exec_hi:[0-9]+]]{{\]}}, 1, 0
+; GCN32: s_mov_b32 s[[exec_lo:[0-9]+]], exec_lo
+; GCN64: s_mov_b64 s{{\[}}[[exec_lo:[0-9]+]]:[[exec_hi:[0-9]+]]{{\]}}, exec
 ; GCN: v_mbcnt_lo_u32_b32{{(_e[0-9]+)?}} v[[mbcnt:[0-9]+]], s[[exec_lo]], 0
 ; GCN64: v_mbcnt_hi_u32_b32{{(_e[0-9]+)?}} v[[mbcnt]], s[[exec_hi]], v[[mbcnt]]
 ; GCN: v_cmp_eq_u32{{(_e[0-9]+)?}} vcc{{(_lo)?}}, 0, v[[mbcnt]]
-; GCN32: s_bcnt1_i32_b32 s{{[0-9]+}}, s[[exec_lo]]
-; GCN64: s_bcnt1_i32_b64 s{{[0-9]+}}, s{{\[}}[[exec_lo]]:[[exec_hi]]{{\]}}
+; GCN: s_and_saveexec_b{{32|64}} s[[exec:\[?[0-9:]+\]?]], vcc
+; GCN32: s_bcnt1_i32_b32 s[[popcount:[0-9]+]], s[[exec_lo]]
+; GCN64: s_bcnt1_i32_b64 s[[popcount:[0-9]+]], s{{\[}}[[exec_lo]]:[[exec_hi]]{{\]}}
 ; GCN: {{flat|buffer|global}}_atomic_sub_x2 v{{\[}}{{[0-9]+}}:{{[0-9]+}}{{\]}}
 define amdgpu_kernel void @sub_i64_uniform(i64 addrspace(1)* %out, i64 addrspace(1)* %inout, i64 %subitive) {
 entry:
   %old = atomicrmw sub i64 addrspace(1)* %inout, i64 %subitive acq_rel
   store i64 %old, i64 addrspace(1)* %out
   ret void
 }
 

llvm/test/CodeGen/AMDGPU/atomic_optimizations_local_pointer.ll

 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
 ; RUN: llc -march=amdgcn -mtriple=amdgcn---amdgiz -amdgpu-atomic-optimizations=true -verify-machineinstrs < %s | FileCheck -enable-var-scope -check-prefixes=GFX7LESS %s
 ; RUN: llc -march=amdgcn -mtriple=amdgcn---amdgiz -mcpu=tonga -mattr=-flat-for-global -amdgpu-atomic-optimizations=true -verify-machineinstrs < %s | FileCheck -enable-var-scope -check-prefixes=GFX8 %s
 ; RUN: llc -march=amdgcn -mtriple=amdgcn---amdgiz -mcpu=gfx900 -mattr=-flat-for-global -amdgpu-atomic-optimizations=true -verify-machineinstrs < %s | FileCheck -enable-var-scope -check-prefixes=GFX9 %s
 ; RUN: llc -march=amdgcn -mtriple=amdgcn---amdgiz -mcpu=gfx1010 -mattr=-wavefrontsize32,+wavefrontsize64 -mattr=-flat-for-global -amdgpu-atomic-optimizations=true -verify-machineinstrs < %s | FileCheck -enable-var-scope -check-prefixes=GFX1064 %s
 ; RUN: llc -march=amdgcn -mtriple=amdgcn---amdgiz -mcpu=gfx1010 -mattr=+wavefrontsize32,-wavefrontsize64 -mattr=-flat-for-global -amdgpu-atomic-optimizations=true -verify-machineinstrs < %s | FileCheck -enable-var-scope -check-prefixes=GFX1032 %s
 
 declare i32 @llvm.amdgcn.workitem.id.x()
 
 @local_var32 = addrspace(3) global i32 undef, align 4
 @local_var64 = addrspace(3) global i64 undef, align 8
 
-; Show that what the atomic optimization pass will do for local pointers.
+; Show what the atomic optimization pass will do for local pointers.
 
 define amdgpu_kernel void @add_i32_constant(i32 addrspace(1)* %out) {
 ;
 ;
 ; GFX7LESS-LABEL: add_i32_constant:
 ; GFX7LESS:       ; %bb.0: ; %entry
+; GFX7LESS-NEXT:    s_mov_b64 s[2:3], exec
 ; GFX7LESS-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x9
-; GFX7LESS-NEXT:    v_cmp_ne_u32_e64 s[4:5], 1, 0
-; GFX7LESS-NEXT:    v_mbcnt_lo_u32_b32_e64 v0, s4, 0
-; GFX7LESS-NEXT:    v_mbcnt_hi_u32_b32_e32 v0, s5, v0
+; GFX7LESS-NEXT:    v_mbcnt_lo_u32_b32_e64 v0, s2, 0
+; GFX7LESS-NEXT:    v_mbcnt_hi_u32_b32_e32 v0, s3, v0
 ; GFX7LESS-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX7LESS-NEXT:    ; implicit-def: $vgpr1
-; GFX7LESS-NEXT:    s_and_saveexec_b64 s[2:3], vcc
+; GFX7LESS-NEXT:    s_and_saveexec_b64 s[4:5], vcc
 ; GFX7LESS-NEXT:    s_cbranch_execz BB0_2
 ; GFX7LESS-NEXT:  ; %bb.1:
-; GFX7LESS-NEXT:    s_bcnt1_i32_b64 s4, s[4:5]
+; GFX7LESS-NEXT:    s_bcnt1_i32_b64 s2, s[2:3]
 ; GFX7LESS-NEXT:    v_mov_b32_e32 v1, local_var32@abs32@lo
-; GFX7LESS-NEXT:    v_mul_u32_u24_e64 v2, s4, 5
+; GFX7LESS-NEXT:    v_mul_u32_u24_e64 v2, s2, 5
 ; GFX7LESS-NEXT:    s_mov_b32 m0, -1
 ; GFX7LESS-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX7LESS-NEXT:    ds_add_rtn_u32 v1, v1, v2
 ; GFX7LESS-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX7LESS-NEXT:    buffer_wbinvl1
 ; GFX7LESS-NEXT:  BB0_2:
-; GFX7LESS-NEXT:    s_or_b64 exec, exec, s[2:3]
+; GFX7LESS-NEXT:    s_or_b64 exec, exec, s[4:5]
 ; GFX7LESS-NEXT:    v_readfirstlane_b32 s2, v1
 ; GFX7LESS-NEXT:    s_mov_b32 s3, 0xf000
 ; GFX7LESS-NEXT:    v_mad_u32_u24 v0, v0, 5, s2
 ; GFX7LESS-NEXT:    s_mov_b32 s2, -1
 ; GFX7LESS-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX7LESS-NEXT:    buffer_store_dword v0, off, s[0:3], 0
 ; GFX7LESS-NEXT:    s_endpgm
 ;
 ; GFX8-LABEL: add_i32_constant:
 ; GFX8:       ; %bb.0: ; %entry
 ; GFX8-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
-; GFX8-NEXT:    v_cmp_ne_u32_e64 s[4:5], 1, 0
-; GFX8-NEXT:    v_mbcnt_lo_u32_b32 v0, s4, 0
-; GFX8-NEXT:    v_mbcnt_hi_u32_b32 v0, s5, v0
+; GFX8-NEXT:    s_mov_b64 s[2:3], exec
+; GFX8-NEXT:    v_mbcnt_lo_u32_b32 v0, s2, 0
+; GFX8-NEXT:    v_mbcnt_hi_u32_b32 v0, s3, v0
 ; GFX8-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX8-NEXT:    ; implicit-def: $vgpr1
-; GFX8-NEXT:    s_and_saveexec_b64 s[2:3], vcc
+; GFX8-NEXT:    s_and_saveexec_b64 s[4:5], vcc
 ; GFX8-NEXT:    s_cbranch_execz BB0_2
 ; GFX8-NEXT:  ; %bb.1:
-; GFX8-NEXT:    s_bcnt1_i32_b64 s4, s[4:5]
-; GFX8-NEXT:    v_mul_u32_u24_e64 v1, s4, 5
+; GFX8-NEXT:    s_bcnt1_i32_b64 s2, s[2:3]
+; GFX8-NEXT:    v_mul_u32_u24_e64 v1, s2, 5
 ; GFX8-NEXT:    v_mov_b32_e32 v2, local_var32@abs32@lo
 ; GFX8-NEXT:    s_mov_b32 m0, -1
 ; GFX8-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX8-NEXT:    ds_add_rtn_u32 v1, v2, v1
 ; GFX8-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX8-NEXT:    buffer_wbinvl1_vol
 ; GFX8-NEXT:  BB0_2:
-; GFX8-NEXT:    s_or_b64 exec, exec, s[2:3]
+; GFX8-NEXT:    s_or_b64 exec, exec, s[4:5]
 ; GFX8-NEXT:    v_readfirstlane_b32 s2, v1
 ; GFX8-NEXT:    v_mad_u32_u24 v0, v0, 5, s2
 ; GFX8-NEXT:    s_mov_b32 s3, 0xf000
 ; GFX8-NEXT:    s_mov_b32 s2, -1
 ; GFX8-NEXT:    s_nop 1
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX8-NEXT:    buffer_store_dword v0, off, s[0:3], 0
 ; GFX8-NEXT:    s_endpgm
 ;
 ; GFX9-LABEL: add_i32_constant:
 ; GFX9:       ; %bb.0: ; %entry
 ; GFX9-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
-; GFX9-NEXT:    v_cmp_ne_u32_e64 s[4:5], 1, 0
-; GFX9-NEXT:    v_mbcnt_lo_u32_b32 v0, s4, 0
-; GFX9-NEXT:    v_mbcnt_hi_u32_b32 v0, s5, v0
+; GFX9-NEXT:    s_mov_b64 s[2:3], exec
+; GFX9-NEXT:    v_mbcnt_lo_u32_b32 v0, s2, 0
+; GFX9-NEXT:    v_mbcnt_hi_u32_b32 v0, s3, v0
 ; GFX9-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX9-NEXT:    ; implicit-def: $vgpr1
-; GFX9-NEXT:    s_and_saveexec_b64 s[2:3], vcc
+; GFX9-NEXT:    s_and_saveexec_b64 s[4:5], vcc
 ; GFX9-NEXT:    s_cbranch_execz BB0_2
 ; GFX9-NEXT:  ; %bb.1:
-; GFX9-NEXT:    s_bcnt1_i32_b64 s4, s[4:5]
-; GFX9-NEXT:    v_mul_u32_u24_e64 v1, s4, 5
+; GFX9-NEXT:    s_bcnt1_i32_b64 s2, s[2:3]
+; GFX9-NEXT:    v_mul_u32_u24_e64 v1, s2, 5
 ; GFX9-NEXT:    v_mov_b32_e32 v2, local_var32@abs32@lo
 ; GFX9-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX9-NEXT:    ds_add_rtn_u32 v1, v2, v1
 ; GFX9-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX9-NEXT:    buffer_wbinvl1_vol
 ; GFX9-NEXT:  BB0_2:
-; GFX9-NEXT:    s_or_b64 exec, exec, s[2:3]
+; GFX9-NEXT:    s_or_b64 exec, exec, s[4:5]
 ; GFX9-NEXT:    v_readfirstlane_b32 s2, v1
 ; GFX9-NEXT:    v_mad_u32_u24 v0, v0, 5, s2
 ; GFX9-NEXT:    s_mov_b32 s3, 0xf000
 ; GFX9-NEXT:    s_mov_b32 s2, -1
 ; GFX9-NEXT:    s_nop 1
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:    buffer_store_dword v0, off, s[0:3], 0
 ; GFX9-NEXT:    s_endpgm
 ;
 ; GFX1064-LABEL: add_i32_constant:
 ; GFX1064:       ; %bb.0: ; %entry
-; GFX1064-NEXT:    v_cmp_ne_u32_e64 s[4:5], 1, 0
+; GFX1064-NEXT:    s_mov_b64 s[2:3], exec
 ; GFX1064-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
 ; GFX1064-NEXT:    ; implicit-def: $vgpr1
-; GFX1064-NEXT:    v_mbcnt_lo_u32_b32_e64 v0, s4, 0
-; GFX1064-NEXT:    v_mbcnt_hi_u32_b32_e64 v0, s5, v0
+; GFX1064-NEXT:    v_mbcnt_lo_u32_b32_e64 v0, s2, 0
+; GFX1064-NEXT:    v_mbcnt_hi_u32_b32_e64 v0, s3, v0
 ; GFX1064-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
-; GFX1064-NEXT:    s_and_saveexec_b64 s[2:3], vcc
+; GFX1064-NEXT:    s_and_saveexec_b64 s[4:5], vcc
 ; GFX1064-NEXT:    s_cbranch_execz BB0_2
 ; GFX1064-NEXT:  ; %bb.1:
-; GFX1064-NEXT:    s_bcnt1_i32_b64 s4, s[4:5]
+; GFX1064-NEXT:    s_bcnt1_i32_b64 s2, s[2:3]
 ; GFX1064-NEXT:    v_mov_b32_e32 v2, local_var32@abs32@lo
-; GFX1064-NEXT:    v_mul_u32_u24_e64 v1, s4, 5
+; GFX1064-NEXT:    v_mul_u32_u24_e64 v1, s2, 5
 ; GFX1064-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX1064-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX1064-NEXT:    ds_add_rtn_u32 v1, v2, v1
 ; GFX1064-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX1064-NEXT:    buffer_gl0_inv
 ; GFX1064-NEXT:    buffer_gl1_inv
 ; GFX1064-NEXT:  BB0_2:
 ; GFX1064-NEXT:    v_nop
-; GFX1064-NEXT:    s_or_b64 exec, exec, s[2:3]
+; GFX1064-NEXT:    s_or_b64 exec, exec, s[4:5]
 ; GFX1064-NEXT:    v_readfirstlane_b32 s2, v1
 ; GFX1064-NEXT:    s_mov_b32 s3, 0x31016000
 ; GFX1064-NEXT:    v_mad_u32_u24 v0, v0, 5, s2
 ; GFX1064-NEXT:    s_mov_b32 s2, -1
 ; GFX1064-NEXT:    s_nop 1
 ; GFX1064-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX1064-NEXT:    buffer_store_dword v0, off, s[0:3], 0
 ; GFX1064-NEXT:    s_endpgm
 ;
 ; GFX1032-LABEL: add_i32_constant:
 ; GFX1032:       ; %bb.0: ; %entry
 ; GFX1032-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
-; GFX1032-NEXT:    v_cmp_ne_u32_e64 s3, 1, 0
+; GFX1032-NEXT:    s_mov_b32 s2, exec_lo
 ; GFX1032-NEXT:    ; implicit-def: $vcc_hi
 ; GFX1032-NEXT:    ; implicit-def: $vgpr1
-; GFX1032-NEXT:    v_mbcnt_lo_u32_b32_e64 v0, s3, 0
+; GFX1032-NEXT:    v_mbcnt_lo_u32_b32_e64 v0, s2, 0
 ; GFX1032-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 0, v0
-; GFX1032-NEXT:    s_and_saveexec_b32 s2, vcc_lo
+; GFX1032-NEXT:    s_and_saveexec_b32 s3, vcc_lo
 ; GFX1032-NEXT:    s_cbranch_execz BB0_2
 ; GFX1032-NEXT:  ; %bb.1:
-; GFX1032-NEXT:    s_bcnt1_i32_b32 s3, s3
+; GFX1032-NEXT:    s_bcnt1_i32_b32 s2, s2
 ; GFX1032-NEXT:    v_mov_b32_e32 v2, local_var32@abs32@lo
-; GFX1032-NEXT:    v_mul_u32_u24_e64 v1, s3, 5
+; GFX1032-NEXT:    v_mul_u32_u24_e64 v1, s2, 5
 ; GFX1032-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX1032-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX1032-NEXT:    ds_add_rtn_u32 v1, v2, v1
 ; GFX1032-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX1032-NEXT:    buffer_gl0_inv
 ; GFX1032-NEXT:    buffer_gl1_inv
 ; GFX1032-NEXT:  BB0_2:
 ; GFX1032-NEXT:    v_nop
-; GFX1032-NEXT:    s_or_b32 exec_lo, exec_lo, s2
+; GFX1032-NEXT:    s_or_b32 exec_lo, exec_lo, s3
 ; GFX1032-NEXT:    v_readfirstlane_b32 s2, v1
 ; GFX1032-NEXT:    s_mov_b32 s3, 0x31016000
 ; GFX1032-NEXT:    v_mad_u32_u24 v0, v0, 5, s2
 ; GFX1032-NEXT:    s_mov_b32 s2, -1
 ; GFX1032-NEXT:    s_nop 1
 ; GFX1032-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX1032-NEXT:    buffer_store_dword v0, off, s[0:3], 0
 ; GFX1032-NEXT:    s_endpgm
 entry:
   %old = atomicrmw add i32 addrspace(3)* @local_var32, i32 5 acq_rel
   store i32 %old, i32 addrspace(1)* %out
   ret void
 }
 
 define amdgpu_kernel void @add_i32_uniform(i32 addrspace(1)* %out, i32 %additive) {
 ;
 ;
 ; GFX7LESS-LABEL: add_i32_uniform:
 ; GFX7LESS:       ; %bb.0: ; %entry
+; GFX7LESS-NEXT:    s_mov_b64 s[6:7], exec
 ; GFX7LESS-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x9
 ; GFX7LESS-NEXT:    s_load_dword s2, s[0:1], 0xb
-; GFX7LESS-NEXT:    v_cmp_ne_u32_e64 s[6:7], 1, 0
 ; GFX7LESS-NEXT:    v_mbcnt_lo_u32_b32_e64 v0, s6, 0
 ; GFX7LESS-NEXT:    v_mbcnt_hi_u32_b32_e32 v0, s7, v0
 ; GFX7LESS-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX7LESS-NEXT:    ; implicit-def: $vgpr1
 ; GFX7LESS-NEXT:    s_and_saveexec_b64 s[0:1], vcc
 ; GFX7LESS-NEXT:    s_cbranch_execz BB1_2
 ; GFX7LESS-NEXT:  ; %bb.1:
 ; GFX7LESS-NEXT:    s_bcnt1_i32_b64 s3, s[6:7]

 ; GFX7LESS-NEXT:    s_mov_b32 s6, -1
 ; GFX7LESS-NEXT:    buffer_store_dword v0, off, s[4:7], 0
 ; GFX7LESS-NEXT:    s_endpgm
 ;
 ; GFX8-LABEL: add_i32_uniform:
 ; GFX8:       ; %bb.0: ; %entry
 ; GFX8-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x24
 ; GFX8-NEXT:    s_load_dword s0, s[0:1], 0x2c
-; GFX8-NEXT:    v_cmp_ne_u32_e64 s[6:7], 1, 0
-; GFX8-NEXT:    v_mbcnt_lo_u32_b32 v0, s6, 0
-; GFX8-NEXT:    v_mbcnt_hi_u32_b32 v0, s7, v0
+; GFX8-NEXT:    s_mov_b64 s[2:3], exec
+; GFX8-NEXT:    v_mbcnt_lo_u32_b32 v0, s2, 0
+; GFX8-NEXT:    v_mbcnt_hi_u32_b32 v0, s3, v0
 ; GFX8-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX8-NEXT:    ; implicit-def: $vgpr1
-; GFX8-NEXT:    s_and_saveexec_b64 s[2:3], vcc
+; GFX8-NEXT:    s_and_saveexec_b64 s[6:7], vcc
 ; GFX8-NEXT:    s_cbranch_execz BB1_2
 ; GFX8-NEXT:  ; %bb.1:
-; GFX8-NEXT:    s_bcnt1_i32_b64 s1, s[6:7]
+; GFX8-NEXT:    s_bcnt1_i32_b64 s1, s[2:3]
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX8-NEXT:    s_mul_i32 s1, s0, s1
 ; GFX8-NEXT:    v_mov_b32_e32 v1, local_var32@abs32@lo
 ; GFX8-NEXT:    v_mov_b32_e32 v2, s1
 ; GFX8-NEXT:    s_mov_b32 m0, -1
 ; GFX8-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX8-NEXT:    ds_add_rtn_u32 v1, v1, v2
 ; GFX8-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX8-NEXT:    buffer_wbinvl1_vol
 ; GFX8-NEXT:  BB1_2:
-; GFX8-NEXT:    s_or_b64 exec, exec, s[2:3]
+; GFX8-NEXT:    s_or_b64 exec, exec, s[6:7]
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX8-NEXT:    v_mul_lo_u32 v0, s0, v0
 ; GFX8-NEXT:    v_readfirstlane_b32 s0, v1
 ; GFX8-NEXT:    s_mov_b32 s7, 0xf000
 ; GFX8-NEXT:    s_mov_b32 s6, -1
 ; GFX8-NEXT:    v_add_u32_e32 v0, vcc, s0, v0
 ; GFX8-NEXT:    buffer_store_dword v0, off, s[4:7], 0
 ; GFX8-NEXT:    s_endpgm
 ;
 ; GFX9-LABEL: add_i32_uniform:
 ; GFX9:       ; %bb.0: ; %entry
 ; GFX9-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x24
 ; GFX9-NEXT:    s_load_dword s0, s[0:1], 0x2c
-; GFX9-NEXT:    v_cmp_ne_u32_e64 s[6:7], 1, 0
-; GFX9-NEXT:    v_mbcnt_lo_u32_b32 v0, s6, 0
-; GFX9-NEXT:    v_mbcnt_hi_u32_b32 v0, s7, v0
+; GFX9-NEXT:    s_mov_b64 s[2:3], exec
+; GFX9-NEXT:    v_mbcnt_lo_u32_b32 v0, s2, 0
+; GFX9-NEXT:    v_mbcnt_hi_u32_b32 v0, s3, v0
 ; GFX9-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX9-NEXT:    ; implicit-def: $vgpr1
-; GFX9-NEXT:    s_and_saveexec_b64 s[2:3], vcc
+; GFX9-NEXT:    s_and_saveexec_b64 s[6:7], vcc
 ; GFX9-NEXT:    s_cbranch_execz BB1_2
 ; GFX9-NEXT:  ; %bb.1:
-; GFX9-NEXT:    s_bcnt1_i32_b64 s1, s[6:7]
+; GFX9-NEXT:    s_bcnt1_i32_b64 s1, s[2:3]
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:    s_mul_i32 s1, s0, s1
 ; GFX9-NEXT:    v_mov_b32_e32 v1, local_var32@abs32@lo
 ; GFX9-NEXT:    v_mov_b32_e32 v2, s1
 ; GFX9-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX9-NEXT:    ds_add_rtn_u32 v1, v1, v2
 ; GFX9-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX9-NEXT:    buffer_wbinvl1_vol
 ; GFX9-NEXT:  BB1_2:
-; GFX9-NEXT:    s_or_b64 exec, exec, s[2:3]
+; GFX9-NEXT:    s_or_b64 exec, exec, s[6:7]
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:    v_mul_lo_u32 v0, s0, v0
 ; GFX9-NEXT:    v_readfirstlane_b32 s0, v1
 ; GFX9-NEXT:    s_mov_b32 s7, 0xf000
 ; GFX9-NEXT:    s_mov_b32 s6, -1
 ; GFX9-NEXT:    v_add_u32_e32 v0, s0, v0
 ; GFX9-NEXT:    buffer_store_dword v0, off, s[4:7], 0
 ; GFX9-NEXT:    s_endpgm
 ;
 ; GFX1064-LABEL: add_i32_uniform:
 ; GFX1064:       ; %bb.0: ; %entry
-; GFX1064-NEXT:    v_cmp_ne_u32_e64 s[6:7], 1, 0
+; GFX1064-NEXT:    s_mov_b64 s[2:3], exec
 ; GFX1064-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x24
 ; GFX1064-NEXT:    s_load_dword s0, s[0:1], 0x2c
 ; GFX1064-NEXT:    ; implicit-def: $vgpr1
-; GFX1064-NEXT:    v_mbcnt_lo_u32_b32_e64 v0, s6, 0
-; GFX1064-NEXT:    v_mbcnt_hi_u32_b32_e64 v0, s7, v0
+; GFX1064-NEXT:    v_mbcnt_lo_u32_b32_e64 v0, s2, 0
+; GFX1064-NEXT:    v_mbcnt_hi_u32_b32_e64 v0, s3, v0
 ; GFX1064-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
-; GFX1064-NEXT:    s_and_saveexec_b64 s[2:3], vcc
+; GFX1064-NEXT:    s_and_saveexec_b64 s[6:7], vcc
 ; GFX1064-NEXT:    s_cbranch_execz BB1_2
 ; GFX1064-NEXT:  ; %bb.1:
-; GFX1064-NEXT:    s_bcnt1_i32_b64 s1, s[6:7]
+; GFX1064-NEXT:    s_bcnt1_i32_b64 s1, s[2:3]
 ; GFX1064-NEXT:    v_mov_b32_e32 v1, local_var32@abs32@lo
 ; GFX1064-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX1064-NEXT:    s_mul_i32 s1, s0, s1
 ; GFX1064-NEXT:    v_mov_b32_e32 v2, s1
 ; GFX1064-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX1064-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX1064-NEXT:    ds_add_rtn_u32 v1, v1, v2
 ; GFX1064-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX1064-NEXT:    buffer_gl0_inv
 ; GFX1064-NEXT:    buffer_gl1_inv
 ; GFX1064-NEXT:  BB1_2:
 ; GFX1064-NEXT:    v_nop
-; GFX1064-NEXT:    s_or_b64 exec, exec, s[2:3]
+; GFX1064-NEXT:    s_or_b64 exec, exec, s[6:7]
 ; GFX1064-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX1064-NEXT:    v_mul_lo_u32 v0, s0, v0
 ; GFX1064-NEXT:    v_readfirstlane_b32 s0, v1
 ; GFX1064-NEXT:    s_mov_b32 s7, 0x31016000
 ; GFX1064-NEXT:    s_mov_b32 s6, -1
 ; GFX1064-NEXT:    v_add_nc_u32_e32 v0, s0, v0
 ; GFX1064-NEXT:    buffer_store_dword v0, off, s[4:7], 0
 ; GFX1064-NEXT:    s_endpgm
 ;
 ; GFX1032-LABEL: add_i32_uniform:
 ; GFX1032:       ; %bb.0: ; %entry
 ; GFX1032-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x24
 ; GFX1032-NEXT:    s_load_dword s0, s[0:1], 0x2c
-; GFX1032-NEXT:    v_cmp_ne_u32_e64 s2, 1, 0
+; GFX1032-NEXT:    s_mov_b32 s2, exec_lo
 ; GFX1032-NEXT:    ; implicit-def: $vcc_hi
 ; GFX1032-NEXT:    ; implicit-def: $vgpr1
 ; GFX1032-NEXT:    v_mbcnt_lo_u32_b32_e64 v0, s2, 0
 ; GFX1032-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 0, v0
 ; GFX1032-NEXT:    s_and_saveexec_b32 s1, vcc_lo
 ; GFX1032-NEXT:    s_cbranch_execz BB1_2
 ; GFX1032-NEXT:  ; %bb.1:
 ; GFX1032-NEXT:    s_bcnt1_i32_b32 s2, s2

 ; GFX1032-NEXT:    buffer_store_dword v0, off, s[4:7], 0
 ; GFX1032-NEXT:    s_endpgm
 entry:
   %old = atomicrmw add i32 addrspace(3)* @local_var32, i32 %additive acq_rel
   store i32 %old, i32 addrspace(1)* %out
   ret void
 }
 
-; GFX7LESS-NOT: v_mbcnt_lo_u32_b32
-; GFX7LESS-NOT: v_mbcnt_hi_u32_b32
-; GFX7LESS-NOT: s_bcnt1_i32_b64
-; DPPCOMB: v_add_u32_dpp
-; DPPCOMB: v_add_u32_dpp
-; GFX8MORE32: v_readlane_b32 s[[scalar_value:[0-9]+]], v{{[0-9]+}}, 31
-; GFX8MORE: v_mov_b32{{(_e[0-9]+)?}} v[[value:[0-9]+]], s[[scalar_value]]
-; GFX8MORE: ds_add_rtn_u32 v{{[0-9]+}}, v{{[0-9]+}}, v[[value]]
 define amdgpu_kernel void @add_i32_varying(i32 addrspace(1)* %out) {
 ;
 ;
 ; GFX7LESS-LABEL: add_i32_varying:
 ; GFX7LESS:       ; %bb.0: ; %entry
 ; GFX7LESS-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x9
 ; GFX7LESS-NEXT:    v_mov_b32_e32 v1, local_var32@abs32@lo
 ; GFX7LESS-NEXT:    s_mov_b32 m0, -1
 ; GFX7LESS-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX7LESS-NEXT:    ds_add_rtn_u32 v0, v1, v0
 ; GFX7LESS-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX7LESS-NEXT:    buffer_wbinvl1
 ; GFX7LESS-NEXT:    s_mov_b32 s3, 0xf000
 ; GFX7LESS-NEXT:    s_mov_b32 s2, -1
 ; GFX7LESS-NEXT:    buffer_store_dword v0, off, s[0:3], 0
 ; GFX7LESS-NEXT:    s_endpgm
 ;
 ; GFX8-LABEL: add_i32_varying:
 ; GFX8:       ; %bb.0: ; %entry
 ; GFX8-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
+; GFX8-NEXT:    s_mov_b64 s[2:3], exec
 ; GFX8-NEXT:    v_mov_b32_e32 v2, v0
-; GFX8-NEXT:    s_or_saveexec_b64 s[2:3], -1
+; GFX8-NEXT:    s_or_saveexec_b64 s[4:5], -1
 ; GFX8-NEXT:    v_mov_b32_e32 v1, 0
-; GFX8-NEXT:    s_mov_b64 exec, s[2:3]
-; GFX8-NEXT:    v_cmp_ne_u32_e64 s[2:3], 1, 0
+; GFX8-NEXT:    s_mov_b64 exec, s[4:5]
 ; GFX8-NEXT:    v_mbcnt_lo_u32_b32 v0, s2, 0
 ; GFX8-NEXT:    v_mbcnt_hi_u32_b32 v0, s3, v0
 ; GFX8-NEXT:    s_not_b64 exec, exec
 ; GFX8-NEXT:    v_mov_b32_e32 v2, 0
 ; GFX8-NEXT:    s_not_b64 exec, exec
 ; GFX8-NEXT:    s_or_saveexec_b64 s[4:5], -1
 ; GFX8-NEXT:    v_add_u32_dpp v2, vcc, v2, v2 row_shr:1 row_mask:0xf bank_mask:0xf bound_ctrl:0
 ; GFX8-NEXT:    s_nop 1

 ; GFX8-NEXT:    s_nop 0
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX8-NEXT:    buffer_store_dword v0, off, s[0:3], 0
 ; GFX8-NEXT:    s_endpgm
 ;
 ; GFX9-LABEL: add_i32_varying:
 ; GFX9:       ; %bb.0: ; %entry
 ; GFX9-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
+; GFX9-NEXT:    s_mov_b64 s[2:3], exec
 ; GFX9-NEXT:    v_mov_b32_e32 v2, v0
-; GFX9-NEXT:    s_or_saveexec_b64 s[2:3], -1
+; GFX9-NEXT:    s_or_saveexec_b64 s[4:5], -1
 ; GFX9-NEXT:    v_mov_b32_e32 v1, 0
-; GFX9-NEXT:    s_mov_b64 exec, s[2:3]
-; GFX9-NEXT:    v_cmp_ne_u32_e64 s[2:3], 1, 0
+; GFX9-NEXT:    s_mov_b64 exec, s[4:5]
 ; GFX9-NEXT:    v_mbcnt_lo_u32_b32 v0, s2, 0
 ; GFX9-NEXT:    v_mbcnt_hi_u32_b32 v0, s3, v0
 ; GFX9-NEXT:    s_not_b64 exec, exec
 ; GFX9-NEXT:    v_mov_b32_e32 v2, 0
 ; GFX9-NEXT:    s_not_b64 exec, exec
 ; GFX9-NEXT:    s_or_saveexec_b64 s[4:5], -1
 ; GFX9-NEXT:    v_add_u32_dpp v2, v2, v2 row_shr:1 row_mask:0xf bank_mask:0xf bound_ctrl:0
 ; GFX9-NEXT:    s_nop 1

 ; GFX9-NEXT:    s_nop 0
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:    buffer_store_dword v0, off, s[0:3], 0
 ; GFX9-NEXT:    s_endpgm
 ;
 ; GFX1064-LABEL: add_i32_varying:
 ; GFX1064:       ; %bb.0: ; %entry
 ; GFX1064-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
+; GFX1064-NEXT:    s_mov_b64 s[2:3], exec
 ; GFX1064-NEXT:    v_mov_b32_e32 v2, v0
-; GFX1064-NEXT:    s_or_saveexec_b64 s[2:3], -1
+; GFX1064-NEXT:    s_or_saveexec_b64 s[4:5], -1
 ; GFX1064-NEXT:    v_mov_b32_e32 v1, 0
-; GFX1064-NEXT:    s_mov_b64 exec, s[2:3]
-; GFX1064-NEXT:    v_cmp_ne_u32_e64 s[2:3], 1, 0
+; GFX1064-NEXT:    s_mov_b64 exec, s[4:5]
 ; GFX1064-NEXT:    v_mbcnt_lo_u32_b32_e64 v0, s2, 0
 ; GFX1064-NEXT:    v_mbcnt_hi_u32_b32_e64 v0, s3, v0
 ; GFX1064-NEXT:    s_not_b64 exec, exec
 ; GFX1064-NEXT:    v_mov_b32_e32 v2, 0
 ; GFX1064-NEXT:    s_not_b64 exec, exec
 ; GFX1064-NEXT:    s_or_saveexec_b64 s[4:5], -1
 ; GFX1064-NEXT:    v_add_nc_u32_dpp v2, v2, v2 row_shr:1 row_mask:0xf bank_mask:0xf bound_ctrl:0
 ; GFX1064-NEXT:    v_add_nc_u32_dpp v2, v2, v2 row_shr:2 row_mask:0xf bank_mask:0xf bound_ctrl:0

 ; GFX1064-NEXT:    s_nop 1
 ; GFX1064-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX1064-NEXT:    buffer_store_dword v0, off, s[0:3], 0
 ; GFX1064-NEXT:    s_endpgm
 ;
 ; GFX1032-LABEL: add_i32_varying:
 ; GFX1032:       ; %bb.0: ; %entry
 ; GFX1032-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
+; GFX1032-NEXT:    s_mov_b32 s2, exec_lo
 ; GFX1032-NEXT:    ; implicit-def: $vcc_hi
 ; GFX1032-NEXT:    v_mov_b32_e32 v2, v0
-; GFX1032-NEXT:    s_or_saveexec_b32 s2, -1
+; GFX1032-NEXT:    s_or_saveexec_b32 s3, -1
 ; GFX1032-NEXT:    v_mov_b32_e32 v1, 0
-; GFX1032-NEXT:    s_mov_b32 exec_lo, s2
-; GFX1032-NEXT:    v_cmp_ne_u32_e64 s2, 1, 0
+; GFX1032-NEXT:    s_mov_b32 exec_lo, s3
 ; GFX1032-NEXT:    v_mbcnt_lo_u32_b32_e64 v0, s2, 0
 ; GFX1032-NEXT:    s_not_b32 exec_lo, exec_lo
 ; GFX1032-NEXT:    v_mov_b32_e32 v2, 0
 ; GFX1032-NEXT:    s_not_b32 exec_lo, exec_lo
 ; GFX1032-NEXT:    s_or_saveexec_b32 s4, -1
 ; GFX1032-NEXT:    v_add_nc_u32_dpp v2, v2, v2 row_shr:1 row_mask:0xf bank_mask:0xf bound_ctrl:0
 ; GFX1032-NEXT:    s_mov_b32 s2, -1
 ; GFX1032-NEXT:    v_add_nc_u32_dpp v2, v2, v2 row_shr:2 row_mask:0xf bank_mask:0xf bound_ctrl:0

 ; GFX7LESS-NEXT:    s_mov_b32 s3, 0xf000
 ; GFX7LESS-NEXT:    s_mov_b32 s2, -1
 ; GFX7LESS-NEXT:    buffer_store_dword v0, off, s[0:3], 0
 ; GFX7LESS-NEXT:    s_endpgm
 ;
 ; GFX8-LABEL: add_i32_varying_gfx1032:
 ; GFX8:       ; %bb.0: ; %entry
 ; GFX8-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
+; GFX8-NEXT:    s_mov_b64 s[2:3], exec
 ; GFX8-NEXT:    v_mov_b32_e32 v2, v0
-; GFX8-NEXT:    s_or_saveexec_b64 s[2:3], -1
+; GFX8-NEXT:    s_or_saveexec_b64 s[4:5], -1
 ; GFX8-NEXT:    v_mov_b32_e32 v1, 0
-; GFX8-NEXT:    s_mov_b64 exec, s[2:3]
-; GFX8-NEXT:    v_cmp_ne_u32_e64 s[2:3], 1, 0
+; GFX8-NEXT:    s_mov_b64 exec, s[4:5]
 ; GFX8-NEXT:    v_mbcnt_lo_u32_b32 v0, s2, 0
 ; GFX8-NEXT:    v_mbcnt_hi_u32_b32 v0, s3, v0
 ; GFX8-NEXT:    s_not_b64 exec, exec
 ; GFX8-NEXT:    v_mov_b32_e32 v2, 0
 ; GFX8-NEXT:    s_not_b64 exec, exec
 ; GFX8-NEXT:    s_or_saveexec_b64 s[4:5], -1
 ; GFX8-NEXT:    v_add_u32_dpp v2, vcc, v2, v2 row_shr:1 row_mask:0xf bank_mask:0xf bound_ctrl:0
 ; GFX8-NEXT:    s_nop 1

 ; GFX8-NEXT:    s_nop 0
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX8-NEXT:    buffer_store_dword v0, off, s[0:3], 0
 ; GFX8-NEXT:    s_endpgm
 ;
 ; GFX9-LABEL: add_i32_varying_gfx1032:
 ; GFX9:       ; %bb.0: ; %entry
 ; GFX9-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
+; GFX9-NEXT:    s_mov_b64 s[2:3], exec
 ; GFX9-NEXT:    v_mov_b32_e32 v2, v0
-; GFX9-NEXT:    s_or_saveexec_b64 s[2:3], -1
+; GFX9-NEXT:    s_or_saveexec_b64 s[4:5], -1
 ; GFX9-NEXT:    v_mov_b32_e32 v1, 0
-; GFX9-NEXT:    s_mov_b64 exec, s[2:3]
-; GFX9-NEXT:    v_cmp_ne_u32_e64 s[2:3], 1, 0
+; GFX9-NEXT:    s_mov_b64 exec, s[4:5]
 ; GFX9-NEXT:    v_mbcnt_lo_u32_b32 v0, s2, 0
 ; GFX9-NEXT:    v_mbcnt_hi_u32_b32 v0, s3, v0
 ; GFX9-NEXT:    s_not_b64 exec, exec
 ; GFX9-NEXT:    v_mov_b32_e32 v2, 0
 ; GFX9-NEXT:    s_not_b64 exec, exec
 ; GFX9-NEXT:    s_or_saveexec_b64 s[4:5], -1
 ; GFX9-NEXT:    v_add_u32_dpp v2, v2, v2 row_shr:1 row_mask:0xf bank_mask:0xf bound_ctrl:0
 ; GFX9-NEXT:    s_nop 1

 ; GFX9-NEXT:    s_nop 0
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:    buffer_store_dword v0, off, s[0:3], 0
 ; GFX9-NEXT:    s_endpgm
 ;
 ; GFX1064-LABEL: add_i32_varying_gfx1032:
 ; GFX1064:       ; %bb.0: ; %entry
 ; GFX1064-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
+; GFX1064-NEXT:    s_mov_b64 s[2:3], exec
 ; GFX1064-NEXT:    v_mov_b32_e32 v2, v0
-; GFX1064-NEXT:    s_or_saveexec_b64 s[2:3], -1
+; GFX1064-NEXT:    s_or_saveexec_b64 s[4:5], -1
 ; GFX1064-NEXT:    v_mov_b32_e32 v1, 0
-; GFX1064-NEXT:    s_mov_b64 exec, s[2:3]
-; GFX1064-NEXT:    v_cmp_ne_u32_e64 s[2:3], 1, 0
+; GFX1064-NEXT:    s_mov_b64 exec, s[4:5]
 ; GFX1064-NEXT:    v_mbcnt_lo_u32_b32_e64 v0, s2, 0
 ; GFX1064-NEXT:    v_mbcnt_hi_u32_b32_e64 v0, s3, v0
 ; GFX1064-NEXT:    s_not_b64 exec, exec
 ; GFX1064-NEXT:    v_mov_b32_e32 v2, 0
 ; GFX1064-NEXT:    s_not_b64 exec, exec
 ; GFX1064-NEXT:    s_or_saveexec_b64 s[4:5], -1
 ; GFX1064-NEXT:    v_add_nc_u32_dpp v2, v2, v2 row_shr:1 row_mask:0xf bank_mask:0xf bound_ctrl:0
 ; GFX1064-NEXT:    v_add_nc_u32_dpp v2, v2, v2 row_shr:2 row_mask:0xf bank_mask:0xf bound_ctrl:0

 ; GFX1064-NEXT:    s_nop 1
 ; GFX1064-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX1064-NEXT:    buffer_store_dword v0, off, s[0:3], 0
 ; GFX1064-NEXT:    s_endpgm
 ;
 ; GFX1032-LABEL: add_i32_varying_gfx1032:
 ; GFX1032:       ; %bb.0: ; %entry
 ; GFX1032-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
+; GFX1032-NEXT:    s_mov_b32 s2, exec_lo
 ; GFX1032-NEXT:    ; implicit-def: $vcc_hi
 ; GFX1032-NEXT:    v_mov_b32_e32 v2, v0
-; GFX1032-NEXT:    s_or_saveexec_b32 s2, -1
+; GFX1032-NEXT:    s_or_saveexec_b32 s3, -1
 ; GFX1032-NEXT:    v_mov_b32_e32 v1, 0
-; GFX1032-NEXT:    s_mov_b32 exec_lo, s2
-; GFX1032-NEXT:    v_cmp_ne_u32_e64 s2, 1, 0
+; GFX1032-NEXT:    s_mov_b32 exec_lo, s3
 ; GFX1032-NEXT:    v_mbcnt_lo_u32_b32_e64 v0, s2, 0
 ; GFX1032-NEXT:    s_not_b32 exec_lo, exec_lo
 ; GFX1032-NEXT:    v_mov_b32_e32 v2, 0
 ; GFX1032-NEXT:    s_not_b32 exec_lo, exec_lo
 ; GFX1032-NEXT:    s_or_saveexec_b32 s4, -1
 ; GFX1032-NEXT:    v_add_nc_u32_dpp v2, v2, v2 row_shr:1 row_mask:0xf bank_mask:0xf bound_ctrl:0
 ; GFX1032-NEXT:    s_mov_b32 s2, -1
 ; GFX1032-NEXT:    v_add_nc_u32_dpp v2, v2, v2 row_shr:2 row_mask:0xf bank_mask:0xf bound_ctrl:0

 ; GFX7LESS-NEXT:    s_mov_b32 s3, 0xf000
 ; GFX7LESS-NEXT:    s_mov_b32 s2, -1
 ; GFX7LESS-NEXT:    buffer_store_dword v0, off, s[0:3], 0
 ; GFX7LESS-NEXT:    s_endpgm
 ;
 ; GFX8-LABEL: add_i32_varying_gfx1064:
 ; GFX8:       ; %bb.0: ; %entry
 ; GFX8-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
+; GFX8-NEXT:    s_mov_b64 s[2:3], exec
 ; GFX8-NEXT:    v_mov_b32_e32 v2, v0
-; GFX8-NEXT:    s_or_saveexec_b64 s[2:3], -1
+; GFX8-NEXT:    s_or_saveexec_b64 s[4:5], -1
 ; GFX8-NEXT:    v_mov_b32_e32 v1, 0
-; GFX8-NEXT:    s_mov_b64 exec, s[2:3]
-; GFX8-NEXT:    v_cmp_ne_u32_e64 s[2:3], 1, 0
+; GFX8-NEXT:    s_mov_b64 exec, s[4:5]
 ; GFX8-NEXT:    v_mbcnt_lo_u32_b32 v0, s2, 0
 ; GFX8-NEXT:    v_mbcnt_hi_u32_b32 v0, s3, v0
 ; GFX8-NEXT:    s_not_b64 exec, exec
 ; GFX8-NEXT:    v_mov_b32_e32 v2, 0
 ; GFX8-NEXT:    s_not_b64 exec, exec
 ; GFX8-NEXT:    s_or_saveexec_b64 s[4:5], -1
 ; GFX8-NEXT:    v_add_u32_dpp v2, vcc, v2, v2 row_shr:1 row_mask:0xf bank_mask:0xf bound_ctrl:0
 ; GFX8-NEXT:    s_nop 1

 ; GFX8-NEXT:    s_nop 0
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX8-NEXT:    buffer_store_dword v0, off, s[0:3], 0
 ; GFX8-NEXT:    s_endpgm
 ;
 ; GFX9-LABEL: add_i32_varying_gfx1064:
 ; GFX9:       ; %bb.0: ; %entry
 ; GFX9-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
+; GFX9-NEXT:    s_mov_b64 s[2:3], exec
 ; GFX9-NEXT:    v_mov_b32_e32 v2, v0
-; GFX9-NEXT:    s_or_saveexec_b64 s[2:3], -1
+; GFX9-NEXT:    s_or_saveexec_b64 s[4:5], -1
 ; GFX9-NEXT:    v_mov_b32_e32 v1, 0
-; GFX9-NEXT:    s_mov_b64 exec, s[2:3]
-; GFX9-NEXT:    v_cmp_ne_u32_e64 s[2:3], 1, 0
+; GFX9-NEXT:    s_mov_b64 exec, s[4:5]
 ; GFX9-NEXT:    v_mbcnt_lo_u32_b32 v0, s2, 0
 ; GFX9-NEXT:    v_mbcnt_hi_u32_b32 v0, s3, v0
 ; GFX9-NEXT:    s_not_b64 exec, exec
 ; GFX9-NEXT:    v_mov_b32_e32 v2, 0
 ; GFX9-NEXT:    s_not_b64 exec, exec
 ; GFX9-NEXT:    s_or_saveexec_b64 s[4:5], -1
 ; GFX9-NEXT:    v_add_u32_dpp v2, v2, v2 row_shr:1 row_mask:0xf bank_mask:0xf bound_ctrl:0
 ; GFX9-NEXT:    s_nop 1

 ; GFX9-NEXT:    s_nop 0
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:    buffer_store_dword v0, off, s[0:3], 0
 ; GFX9-NEXT:    s_endpgm
 ;
 ; GFX1064-LABEL: add_i32_varying_gfx1064:
 ; GFX1064:       ; %bb.0: ; %entry
 ; GFX1064-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
+; GFX1064-NEXT:    s_mov_b64 s[2:3], exec
 ; GFX1064-NEXT:    v_mov_b32_e32 v2, v0
-; GFX1064-NEXT:    s_or_saveexec_b64 s[2:3], -1
+; GFX1064-NEXT:    s_or_saveexec_b64 s[4:5], -1
 ; GFX1064-NEXT:    v_mov_b32_e32 v1, 0
-; GFX1064-NEXT:    s_mov_b64 exec, s[2:3]
-; GFX1064-NEXT:    v_cmp_ne_u32_e64 s[2:3], 1, 0
+; GFX1064-NEXT:    s_mov_b64 exec, s[4:5]
 ; GFX1064-NEXT:    v_mbcnt_lo_u32_b32_e64 v0, s2, 0
 ; GFX1064-NEXT:    v_mbcnt_hi_u32_b32_e64 v0, s3, v0
 ; GFX1064-NEXT:    s_not_b64 exec, exec
 ; GFX1064-NEXT:    v_mov_b32_e32 v2, 0
 ; GFX1064-NEXT:    s_not_b64 exec, exec
 ; GFX1064-NEXT:    s_or_saveexec_b64 s[4:5], -1
 ; GFX1064-NEXT:    v_add_nc_u32_dpp v2, v2, v2 row_shr:1 row_mask:0xf bank_mask:0xf bound_ctrl:0
 ; GFX1064-NEXT:    v_add_nc_u32_dpp v2, v2, v2 row_shr:2 row_mask:0xf bank_mask:0xf bound_ctrl:0

 ; GFX1064-NEXT:    s_nop 1
 ; GFX1064-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX1064-NEXT:    buffer_store_dword v0, off, s[0:3], 0
 ; GFX1064-NEXT:    s_endpgm
 ;
 ; GFX1032-LABEL: add_i32_varying_gfx1064:
 ; GFX1032:       ; %bb.0: ; %entry
 ; GFX1032-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
+; GFX1032-NEXT:    s_mov_b32 s2, exec_lo
 ; GFX1032-NEXT:    ; implicit-def: $vcc_hi
 ; GFX1032-NEXT:    v_mov_b32_e32 v2, v0
-; GFX1032-NEXT:    s_or_saveexec_b32 s2, -1
+; GFX1032-NEXT:    s_or_saveexec_b32 s3, -1
 ; GFX1032-NEXT:    v_mov_b32_e32 v1, 0
-; GFX1032-NEXT:    s_mov_b32 exec_lo, s2
-; GFX1032-NEXT:    v_cmp_ne_u32_e64 s2, 1, 0
+; GFX1032-NEXT:    s_mov_b32 exec_lo, s3
 ; GFX1032-NEXT:    v_mbcnt_lo_u32_b32_e64 v0, s2, 0
 ; GFX1032-NEXT:    s_not_b32 exec_lo, exec_lo
 ; GFX1032-NEXT:    v_mov_b32_e32 v2, 0
 ; GFX1032-NEXT:    s_not_b32 exec_lo, exec_lo
 ; GFX1032-NEXT:    s_or_saveexec_b32 s4, -1
 ; GFX1032-NEXT:    v_add_nc_u32_dpp v2, v2, v2 row_shr:1 row_mask:0xf bank_mask:0xf bound_ctrl:0
 ; GFX1032-NEXT:    s_mov_b32 s2, -1
 ; GFX1032-NEXT:    v_add_nc_u32_dpp v2, v2, v2 row_shr:2 row_mask:0xf bank_mask:0xf bound_ctrl:0
entry:
   ret void
 }
 
 define amdgpu_kernel void @add_i64_constant(i64 addrspace(1)* %out) {
 ;
 ;
 ; GFX7LESS-LABEL: add_i64_constant:
 ; GFX7LESS:       ; %bb.0: ; %entry
+; GFX7LESS-NEXT:    s_mov_b64 s[4:5], exec
 ; GFX7LESS-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x9
-; GFX7LESS-NEXT:    v_cmp_ne_u32_e64 s[4:5], 1, 0
 ; GFX7LESS-NEXT:    v_mbcnt_lo_u32_b32_e64 v0, s4, 0
 ; GFX7LESS-NEXT:    v_mbcnt_hi_u32_b32_e32 v0, s5, v0
 ; GFX7LESS-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX7LESS-NEXT:    ; implicit-def: $vgpr1_vgpr2
 ; GFX7LESS-NEXT:    s_and_saveexec_b64 s[2:3], vcc
 ; GFX7LESS-NEXT:    s_cbranch_execz BB5_2
 ; GFX7LESS-NEXT:  ; %bb.1:
 ; GFX7LESS-NEXT:    s_bcnt1_i32_b64 s4, s[4:5]

 ; GFX7LESS-NEXT:    s_mov_b32 s2, -1
 ; GFX7LESS-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX7LESS-NEXT:    buffer_store_dwordx2 v[0:1], off, s[0:3], 0
 ; GFX7LESS-NEXT:    s_endpgm
 ;
 ; GFX8-LABEL: add_i64_constant:
 ; GFX8:       ; %bb.0: ; %entry
 ; GFX8-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
-; GFX8-NEXT:    v_cmp_ne_u32_e64 s[4:5], 1, 0
+; GFX8-NEXT:    s_mov_b64 s[4:5], exec
 ; GFX8-NEXT:    v_mbcnt_lo_u32_b32 v0, s4, 0
 ; GFX8-NEXT:    v_mbcnt_hi_u32_b32 v0, s5, v0
 ; GFX8-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX8-NEXT:    ; implicit-def: $vgpr1_vgpr2
 ; GFX8-NEXT:    s_and_saveexec_b64 s[2:3], vcc
 ; GFX8-NEXT:    s_cbranch_execz BB5_2
 ; GFX8-NEXT:  ; %bb.1:
 ; GFX8-NEXT:    s_bcnt1_i32_b64 s4, s[4:5]

 ; GFX8-NEXT:    s_nop 2
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX8-NEXT:    buffer_store_dwordx2 v[0:1], off, s[0:3], 0
 ; GFX8-NEXT:    s_endpgm
 ;
 ; GFX9-LABEL: add_i64_constant:
 ; GFX9:       ; %bb.0: ; %entry
 ; GFX9-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
-; GFX9-NEXT:    v_cmp_ne_u32_e64 s[4:5], 1, 0
+; GFX9-NEXT:    s_mov_b64 s[4:5], exec
 ; GFX9-NEXT:    v_mbcnt_lo_u32_b32 v0, s4, 0
 ; GFX9-NEXT:    v_mbcnt_hi_u32_b32 v0, s5, v0
 ; GFX9-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX9-NEXT:    ; implicit-def: $vgpr1_vgpr2
 ; GFX9-NEXT:    s_and_saveexec_b64 s[2:3], vcc
 ; GFX9-NEXT:    s_cbranch_execz BB5_2
 ; GFX9-NEXT:  ; %bb.1:
 ; GFX9-NEXT:    s_bcnt1_i32_b64 s4, s[4:5]

 ; GFX9-NEXT:    s_mov_b32 s2, -1
 ; GFX9-NEXT:    s_nop 2
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:    buffer_store_dwordx2 v[0:1], off, s[0:3], 0
 ; GFX9-NEXT:    s_endpgm
 ;
 ; GFX1064-LABEL: add_i64_constant:
 ; GFX1064:       ; %bb.0: ; %entry
-; GFX1064-NEXT:    v_cmp_ne_u32_e64 s[4:5], 1, 0
+; GFX1064-NEXT:    s_mov_b64 s[4:5], exec
 ; GFX1064-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
 ; GFX1064-NEXT:    ; implicit-def: $vgpr1_vgpr2
 ; GFX1064-NEXT:    v_mbcnt_lo_u32_b32_e64 v0, s4, 0
 ; GFX1064-NEXT:    v_mbcnt_hi_u32_b32_e64 v0, s5, v0
 ; GFX1064-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX1064-NEXT:    s_and_saveexec_b64 s[2:3], vcc
 ; GFX1064-NEXT:    s_cbranch_execz BB5_2
 ; GFX1064-NEXT:  ; %bb.1:

 ; GFX1064-NEXT:    s_nop 2
 ; GFX1064-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX1064-NEXT:    buffer_store_dwordx2 v[0:1], off, s[0:3], 0
 ; GFX1064-NEXT:    s_endpgm
 ;
 ; GFX1032-LABEL: add_i64_constant:
 ; GFX1032:       ; %bb.0: ; %entry
 ; GFX1032-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
-; GFX1032-NEXT:    v_cmp_ne_u32_e64 s3, 1, 0
+; GFX1032-NEXT:    s_mov_b32 s3, exec_lo
 ; GFX1032-NEXT:    ; implicit-def: $vcc_hi
 ; GFX1032-NEXT:    ; implicit-def: $vgpr1_vgpr2
 ; GFX1032-NEXT:    v_mbcnt_lo_u32_b32_e64 v0, s3, 0
 ; GFX1032-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 0, v0
 ; GFX1032-NEXT:    s_and_saveexec_b32 s2, vcc_lo
 ; GFX1032-NEXT:    s_cbranch_execz BB5_2
 ; GFX1032-NEXT:  ; %bb.1:
 ; GFX1032-NEXT:    s_bcnt1_i32_b32 s3, s3
entry:
   ret void
 }
 
 define amdgpu_kernel void @add_i64_uniform(i64 addrspace(1)* %out, i64 %additive) {
 ;
 ;
 ; GFX7LESS-LABEL: add_i64_uniform:
 ; GFX7LESS:       ; %bb.0: ; %entry
+; GFX7LESS-NEXT:    s_mov_b64 s[6:7], exec
 ; GFX7LESS-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x9
-; GFX7LESS-NEXT:    v_cmp_ne_u32_e64 s[6:7], 1, 0
 ; GFX7LESS-NEXT:    v_mbcnt_lo_u32_b32_e64 v0, s6, 0
 ; GFX7LESS-NEXT:    v_mbcnt_hi_u32_b32_e32 v0, s7, v0
 ; GFX7LESS-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX7LESS-NEXT:    ; implicit-def: $vgpr1_vgpr2
 ; GFX7LESS-NEXT:    s_and_saveexec_b64 s[4:5], vcc
 ; GFX7LESS-NEXT:    s_cbranch_execz BB6_2
 ; GFX7LESS-NEXT:  ; %bb.1:
 ; GFX7LESS-NEXT:    s_bcnt1_i32_b64 s6, s[6:7]

 ; GFX7LESS-NEXT:    v_add_i32_e32 v0, vcc, s0, v0
 ; GFX7LESS-NEXT:    v_addc_u32_e32 v1, vcc, v2, v1, vcc
 ; GFX7LESS-NEXT:    buffer_store_dwordx2 v[0:1], off, s[4:7], 0
 ; GFX7LESS-NEXT:    s_endpgm
 ;
 ; GFX8-LABEL: add_i64_uniform:
 ; GFX8:       ; %bb.0: ; %entry
 ; GFX8-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x24
-; GFX8-NEXT:    v_cmp_ne_u32_e64 s[6:7], 1, 0
+; GFX8-NEXT:    s_mov_b64 s[6:7], exec
 ; GFX8-NEXT:    v_mbcnt_lo_u32_b32 v0, s6, 0
 ; GFX8-NEXT:    v_mbcnt_hi_u32_b32 v0, s7, v0
 ; GFX8-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX8-NEXT:    ; implicit-def: $vgpr1_vgpr2
 ; GFX8-NEXT:    s_and_saveexec_b64 s[4:5], vcc
 ; GFX8-NEXT:    s_cbranch_execz BB6_2
 ; GFX8-NEXT:  ; %bb.1:
 ; GFX8-NEXT:    s_bcnt1_i32_b64 s6, s[6:7]

 ; GFX8-NEXT:    s_mov_b32 s6, -1
 ; GFX8-NEXT:    v_addc_u32_e32 v1, vcc, v2, v1, vcc
 ; GFX8-NEXT:    buffer_store_dwordx2 v[0:1], off, s[4:7], 0
 ; GFX8-NEXT:    s_endpgm
 ;
 ; GFX9-LABEL: add_i64_uniform:
 ; GFX9:       ; %bb.0: ; %entry
 ; GFX9-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x24
-; GFX9-NEXT:    v_cmp_ne_u32_e64 s[6:7], 1, 0
+; GFX9-NEXT:    s_mov_b64 s[6:7], exec
 ; GFX9-NEXT:    v_mbcnt_lo_u32_b32 v0, s6, 0
 ; GFX9-NEXT:    v_mbcnt_hi_u32_b32 v0, s7, v0
 ; GFX9-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX9-NEXT:    ; implicit-def: $vgpr1_vgpr2
 ; GFX9-NEXT:    s_and_saveexec_b64 s[4:5], vcc
 ; GFX9-NEXT:    s_cbranch_execz BB6_2
 ; GFX9-NEXT:  ; %bb.1:
 ; GFX9-NEXT:    s_bcnt1_i32_b64 s6, s[6:7]

 ; GFX9-NEXT:    s_mov_b32 s7, 0xf000
 ; GFX9-NEXT:    s_mov_b32 s6, -1
 ; GFX9-NEXT:    v_addc_co_u32_e32 v1, vcc, v2, v1, vcc
 ; GFX9-NEXT:    buffer_store_dwordx2 v[0:1], off, s[4:7], 0
 ; GFX9-NEXT:    s_endpgm
 ;
 ; GFX1064-LABEL: add_i64_uniform:
 ; GFX1064:       ; %bb.0: ; %entry
-; GFX1064-NEXT:    v_cmp_ne_u32_e64 s[6:7], 1, 0
+; GFX1064-NEXT:    s_mov_b64 s[6:7], exec
 ; GFX1064-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x24
 ; GFX1064-NEXT:    ; implicit-def: $vgpr1_vgpr2
 ; GFX1064-NEXT:    v_mbcnt_lo_u32_b32_e64 v0, s6, 0
 ; GFX1064-NEXT:    v_mbcnt_hi_u32_b32_e64 v0, s7, v0
 ; GFX1064-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX1064-NEXT:    s_and_saveexec_b64 s[4:5], vcc
 ; GFX1064-NEXT:    s_cbranch_execz BB6_2
 ; GFX1064-NEXT:  ; %bb.1:

 ; GFX1064-NEXT:    v_add_co_u32_e64 v0, vcc, s4, v0
 ; GFX1064-NEXT:    v_add_co_ci_u32_e32 v1, vcc, s5, v1, vcc
 ; GFX1064-NEXT:    buffer_store_dwordx2 v[0:1], off, s[0:3], 0
 ; GFX1064-NEXT:    s_endpgm
 ;
 ; GFX1032-LABEL: add_i64_uniform:
 ; GFX1032:       ; %bb.0: ; %entry
 ; GFX1032-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x24
-; GFX1032-NEXT:    v_cmp_ne_u32_e64 s5, 1, 0
+; GFX1032-NEXT:    s_mov_b32 s5, exec_lo
 ; GFX1032-NEXT:    ; implicit-def: $vcc_hi
 ; GFX1032-NEXT:    ; implicit-def: $vgpr1_vgpr2
 ; GFX1032-NEXT:    v_mbcnt_lo_u32_b32_e64 v0, s5, 0
 ; GFX1032-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 0, v0
 ; GFX1032-NEXT:    s_and_saveexec_b32 s4, vcc_lo
 ; GFX1032-NEXT:    s_cbranch_execz BB6_2
 ; GFX1032-NEXT:  ; %bb.1:
 ; GFX1032-NEXT:    s_bcnt1_i32_b32 s5, s5
entry:
   ret void
 }
 
 define amdgpu_kernel void @sub_i32_constant(i32 addrspace(1)* %out) {
 ;
 ;
 ; GFX7LESS-LABEL: sub_i32_constant:
 ; GFX7LESS:       ; %bb.0: ; %entry
+; GFX7LESS-NEXT:    s_mov_b64 s[2:3], exec
 ; GFX7LESS-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x9
-; GFX7LESS-NEXT:    v_cmp_ne_u32_e64 s[4:5], 1, 0
-; GFX7LESS-NEXT:    v_mbcnt_lo_u32_b32_e64 v0, s4, 0
-; GFX7LESS-NEXT:    v_mbcnt_hi_u32_b32_e32 v0, s5, v0
+; GFX7LESS-NEXT:    v_mbcnt_lo_u32_b32_e64 v0, s2, 0
+; GFX7LESS-NEXT:    v_mbcnt_hi_u32_b32_e32 v0, s3, v0
 ; GFX7LESS-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX7LESS-NEXT:    ; implicit-def: $vgpr1
-; GFX7LESS-NEXT:    s_and_saveexec_b64 s[2:3], vcc
+; GFX7LESS-NEXT:    s_and_saveexec_b64 s[4:5], vcc
 ; GFX7LESS-NEXT:    s_cbranch_execz BB8_2
 ; GFX7LESS-NEXT:  ; %bb.1:
-; GFX7LESS-NEXT:    s_bcnt1_i32_b64 s4, s[4:5]
+; GFX7LESS-NEXT:    s_bcnt1_i32_b64 s2, s[2:3]
 ; GFX7LESS-NEXT:    v_mov_b32_e32 v1, local_var32@abs32@lo
-; GFX7LESS-NEXT:    v_mul_u32_u24_e64 v2, s4, 5
+; GFX7LESS-NEXT:    v_mul_u32_u24_e64 v2, s2, 5
 ; GFX7LESS-NEXT:    s_mov_b32 m0, -1
 ; GFX7LESS-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX7LESS-NEXT:    ds_sub_rtn_u32 v1, v1, v2
 ; GFX7LESS-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX7LESS-NEXT:    buffer_wbinvl1
 ; GFX7LESS-NEXT:  BB8_2:
-; GFX7LESS-NEXT:    s_or_b64 exec, exec, s[2:3]
+; GFX7LESS-NEXT:    s_or_b64 exec, exec, s[4:5]
 ; GFX7LESS-NEXT:    v_readfirstlane_b32 s2, v1
 ; GFX7LESS-NEXT:    v_mul_u32_u24_e32 v0, 5, v0
 ; GFX7LESS-NEXT:    s_mov_b32 s3, 0xf000
 ; GFX7LESS-NEXT:    v_sub_i32_e32 v0, vcc, s2, v0
 ; GFX7LESS-NEXT:    s_mov_b32 s2, -1
 ; GFX7LESS-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX7LESS-NEXT:    buffer_store_dword v0, off, s[0:3], 0
 ; GFX7LESS-NEXT:    s_endpgm
 ;
 ; GFX8-LABEL: sub_i32_constant:
 ; GFX8:       ; %bb.0: ; %entry
 ; GFX8-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
-; GFX8-NEXT:    v_cmp_ne_u32_e64 s[4:5], 1, 0
-; GFX8-NEXT:    v_mbcnt_lo_u32_b32 v0, s4, 0
-; GFX8-NEXT:    v_mbcnt_hi_u32_b32 v0, s5, v0
+; GFX8-NEXT:    s_mov_b64 s[2:3], exec
+; GFX8-NEXT:    v_mbcnt_lo_u32_b32 v0, s2, 0
+; GFX8-NEXT:    v_mbcnt_hi_u32_b32 v0, s3, v0
 ; GFX8-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX8-NEXT:    ; implicit-def: $vgpr1
-; GFX8-NEXT:    s_and_saveexec_b64 s[2:3], vcc
+; GFX8-NEXT:    s_and_saveexec_b64 s[4:5], vcc
 ; GFX8-NEXT:    s_cbranch_execz BB8_2
 ; GFX8-NEXT:  ; %bb.1:
-; GFX8-NEXT:    s_bcnt1_i32_b64 s4, s[4:5]
-; GFX8-NEXT:    v_mul_u32_u24_e64 v1, s4, 5
+; GFX8-NEXT:    s_bcnt1_i32_b64 s2, s[2:3]
+; GFX8-NEXT:    v_mul_u32_u24_e64 v1, s2, 5
 ; GFX8-NEXT:    v_mov_b32_e32 v2, local_var32@abs32@lo
 ; GFX8-NEXT:    s_mov_b32 m0, -1
 ; GFX8-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX8-NEXT:    ds_sub_rtn_u32 v1, v2, v1
 ; GFX8-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX8-NEXT:    buffer_wbinvl1_vol
 ; GFX8-NEXT:  BB8_2:
-; GFX8-NEXT:    s_or_b64 exec, exec, s[2:3]
+; GFX8-NEXT:    s_or_b64 exec, exec, s[4:5]
 ; GFX8-NEXT:    v_readfirstlane_b32 s2, v1
 ; GFX8-NEXT:    v_mul_u32_u24_e32 v0, 5, v0
 ; GFX8-NEXT:    v_sub_u32_e32 v0, vcc, s2, v0
 ; GFX8-NEXT:    s_mov_b32 s3, 0xf000
 ; GFX8-NEXT:    s_mov_b32 s2, -1
 ; GFX8-NEXT:    s_nop 0
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX8-NEXT:    buffer_store_dword v0, off, s[0:3], 0
 ; GFX8-NEXT:    s_endpgm
 ;
 ; GFX9-LABEL: sub_i32_constant:
 ; GFX9:       ; %bb.0: ; %entry
 ; GFX9-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
-; GFX9-NEXT:    v_cmp_ne_u32_e64 s[4:5], 1, 0
-; GFX9-NEXT:    v_mbcnt_lo_u32_b32 v0, s4, 0
-; GFX9-NEXT:    v_mbcnt_hi_u32_b32 v0, s5, v0
+; GFX9-NEXT:    s_mov_b64 s[2:3], exec
+; GFX9-NEXT:    v_mbcnt_lo_u32_b32 v0, s2, 0
+; GFX9-NEXT:    v_mbcnt_hi_u32_b32 v0, s3, v0
 ; GFX9-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX9-NEXT:    ; implicit-def: $vgpr1
-; GFX9-NEXT:    s_and_saveexec_b64 s[2:3], vcc
+; GFX9-NEXT:    s_and_saveexec_b64 s[4:5], vcc
 ; GFX9-NEXT:    s_cbranch_execz BB8_2
 ; GFX9-NEXT:  ; %bb.1:
-; GFX9-NEXT:    s_bcnt1_i32_b64 s4, s[4:5]
-; GFX9-NEXT:    v_mul_u32_u24_e64 v1, s4, 5
+; GFX9-NEXT:    s_bcnt1_i32_b64 s2, s[2:3]
+; GFX9-NEXT:    v_mul_u32_u24_e64 v1, s2, 5
 ; GFX9-NEXT:    v_mov_b32_e32 v2, local_var32@abs32@lo
 ; GFX9-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX9-NEXT:    ds_sub_rtn_u32 v1, v2, v1
 ; GFX9-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX9-NEXT:    buffer_wbinvl1_vol
 ; GFX9-NEXT:  BB8_2:
-; GFX9-NEXT:    s_or_b64 exec, exec, s[2:3]
+; GFX9-NEXT:    s_or_b64 exec, exec, s[4:5]
 ; GFX9-NEXT:    v_readfirstlane_b32 s2, v1
 ; GFX9-NEXT:    v_mul_u32_u24_e32 v0, 5, v0
 ; GFX9-NEXT:    v_sub_u32_e32 v0, s2, v0
 ; GFX9-NEXT:    s_mov_b32 s3, 0xf000
 ; GFX9-NEXT:    s_mov_b32 s2, -1
 ; GFX9-NEXT:    s_nop 0
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:    buffer_store_dword v0, off, s[0:3], 0
 ; GFX9-NEXT:    s_endpgm
 ;
 ; GFX1064-LABEL: sub_i32_constant:
 ; GFX1064:       ; %bb.0: ; %entry
-; GFX1064-NEXT:    v_cmp_ne_u32_e64 s[4:5], 1, 0
+; GFX1064-NEXT:    s_mov_b64 s[2:3], exec
 ; GFX1064-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
 ; GFX1064-NEXT:    ; implicit-def: $vgpr1
-; GFX1064-NEXT:    v_mbcnt_lo_u32_b32_e64 v0, s4, 0
-; GFX1064-NEXT:    v_mbcnt_hi_u32_b32_e64 v0, s5, v0
+; GFX1064-NEXT:    v_mbcnt_lo_u32_b32_e64 v0, s2, 0
+; GFX1064-NEXT:    v_mbcnt_hi_u32_b32_e64 v0, s3, v0
 ; GFX1064-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
-; GFX1064-NEXT:    s_and_saveexec_b64 s[2:3], vcc
+; GFX1064-NEXT:    s_and_saveexec_b64 s[4:5], vcc
 ; GFX1064-NEXT:    s_cbranch_execz BB8_2
 ; GFX1064-NEXT:  ; %bb.1:
-; GFX1064-NEXT:    s_bcnt1_i32_b64 s4, s[4:5]
+; GFX1064-NEXT:    s_bcnt1_i32_b64 s2, s[2:3]
 ; GFX1064-NEXT:    v_mov_b32_e32 v2, local_var32@abs32@lo
-; GFX1064-NEXT:    v_mul_u32_u24_e64 v1, s4, 5
+; GFX1064-NEXT:    v_mul_u32_u24_e64 v1, s2, 5
 ; GFX1064-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX1064-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX1064-NEXT:    ds_sub_rtn_u32 v1, v2, v1
 ; GFX1064-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX1064-NEXT:    buffer_gl0_inv
 ; GFX1064-NEXT:    buffer_gl1_inv
 ; GFX1064-NEXT:  BB8_2:
 ; GFX1064-NEXT:    v_nop
-; GFX1064-NEXT:    s_or_b64 exec, exec, s[2:3]
+; GFX1064-NEXT:    s_or_b64 exec, exec, s[4:5]
 ; GFX1064-NEXT:    v_readfirstlane_b32 s2, v1
 ; GFX1064-NEXT:    v_mul_u32_u24_e32 v0, 5, v0
 ; GFX1064-NEXT:    s_mov_b32 s3, 0x31016000
 ; GFX1064-NEXT:    v_sub_nc_u32_e32 v0, s2, v0
 ; GFX1064-NEXT:    s_mov_b32 s2, -1
 ; GFX1064-NEXT:    s_nop 0
 ; GFX1064-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX1064-NEXT:    buffer_store_dword v0, off, s[0:3], 0
 ; GFX1064-NEXT:    s_endpgm
 ;
 ; GFX1032-LABEL: sub_i32_constant:
 ; GFX1032:       ; %bb.0: ; %entry
 ; GFX1032-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
-; GFX1032-NEXT:    v_cmp_ne_u32_e64 s3, 1, 0
+; GFX1032-NEXT:    s_mov_b32 s2, exec_lo
 ; GFX1032-NEXT:    ; implicit-def: $vcc_hi
 ; GFX1032-NEXT:    ; implicit-def: $vgpr1
-; GFX1032-NEXT:    v_mbcnt_lo_u32_b32_e64 v0, s3, 0
+; GFX1032-NEXT:    v_mbcnt_lo_u32_b32_e64 v0, s2, 0
 ; GFX1032-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 0, v0
-; GFX1032-NEXT:    s_and_saveexec_b32 s2, vcc_lo
+; GFX1032-NEXT:    s_and_saveexec_b32 s3, vcc_lo
 ; GFX1032-NEXT:    s_cbranch_execz BB8_2
 ; GFX1032-NEXT:  ; %bb.1:
-; GFX1032-NEXT:    s_bcnt1_i32_b32 s3, s3
+; GFX1032-NEXT:    s_bcnt1_i32_b32 s2, s2
 ; GFX1032-NEXT:    v_mov_b32_e32 v2, local_var32@abs32@lo
-; GFX1032-NEXT:    v_mul_u32_u24_e64 v1, s3, 5
+; GFX1032-NEXT:    v_mul_u32_u24_e64 v1, s2, 5
 ; GFX1032-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX1032-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX1032-NEXT:    ds_sub_rtn_u32 v1, v2, v1
 ; GFX1032-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX1032-NEXT:    buffer_gl0_inv
 ; GFX1032-NEXT:    buffer_gl1_inv
 ; GFX1032-NEXT:  BB8_2:
 ; GFX1032-NEXT:    v_nop
-; GFX1032-NEXT:    s_or_b32 exec_lo, exec_lo, s2
+; GFX1032-NEXT:    s_or_b32 exec_lo, exec_lo, s3
 ; GFX1032-NEXT:    v_readfirstlane_b32 s2, v1
 ; GFX1032-NEXT:    v_mul_u32_u24_e32 v0, 5, v0
 ; GFX1032-NEXT:    s_mov_b32 s3, 0x31016000
 ; GFX1032-NEXT:    v_sub_nc_u32_e32 v0, s2, v0
 ; GFX1032-NEXT:    s_mov_b32 s2, -1
 ; GFX1032-NEXT:    s_nop 0
 ; GFX1032-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX1032-NEXT:    buffer_store_dword v0, off, s[0:3], 0
 ; GFX1032-NEXT:    s_endpgm
 entry:
   %old = atomicrmw sub i32 addrspace(3)* @local_var32, i32 5 acq_rel
   store i32 %old, i32 addrspace(1)* %out
   ret void
 }
 
 define amdgpu_kernel void @sub_i32_uniform(i32 addrspace(1)* %out, i32 %subitive) {
 ;
 ;
 ; GFX7LESS-LABEL: sub_i32_uniform:
 ; GFX7LESS:       ; %bb.0: ; %entry
+; GFX7LESS-NEXT:    s_mov_b64 s[6:7], exec
 ; GFX7LESS-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x9
 ; GFX7LESS-NEXT:    s_load_dword s2, s[0:1], 0xb
-; GFX7LESS-NEXT:    v_cmp_ne_u32_e64 s[6:7], 1, 0
 ; GFX7LESS-NEXT:    v_mbcnt_lo_u32_b32_e64 v0, s6, 0
 ; GFX7LESS-NEXT:    v_mbcnt_hi_u32_b32_e32 v0, s7, v0
 ; GFX7LESS-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX7LESS-NEXT:    ; implicit-def: $vgpr1
 ; GFX7LESS-NEXT:    s_and_saveexec_b64 s[0:1], vcc
 ; GFX7LESS-NEXT:    s_cbranch_execz BB9_2
 ; GFX7LESS-NEXT:  ; %bb.1:
 ; GFX7LESS-NEXT:    s_bcnt1_i32_b64 s3, s[6:7]

 ; GFX7LESS-NEXT:    s_mov_b32 s6, -1
 ; GFX7LESS-NEXT:    buffer_store_dword v0, off, s[4:7], 0
 ; GFX7LESS-NEXT:    s_endpgm
 ;
 ; GFX8-LABEL: sub_i32_uniform:
 ; GFX8:       ; %bb.0: ; %entry
 ; GFX8-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x24
 ; GFX8-NEXT:    s_load_dword s0, s[0:1], 0x2c
-; GFX8-NEXT:    v_cmp_ne_u32_e64 s[6:7], 1, 0
-; GFX8-NEXT:    v_mbcnt_lo_u32_b32 v0, s6, 0
-; GFX8-NEXT:    v_mbcnt_hi_u32_b32 v0, s7, v0
+; GFX8-NEXT:    s_mov_b64 s[2:3], exec
+; GFX8-NEXT:    v_mbcnt_lo_u32_b32 v0, s2, 0
+; GFX8-NEXT:    v_mbcnt_hi_u32_b32 v0, s3, v0
 ; GFX8-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX8-NEXT:    ; implicit-def: $vgpr1
-; GFX8-NEXT:    s_and_saveexec_b64 s[2:3], vcc
+; GFX8-NEXT:    s_and_saveexec_b64 s[6:7], vcc
 ; GFX8-NEXT:    s_cbranch_execz BB9_2
 ; GFX8-NEXT:  ; %bb.1:
-; GFX8-NEXT:    s_bcnt1_i32_b64 s1, s[6:7]
+; GFX8-NEXT:    s_bcnt1_i32_b64 s1, s[2:3]
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX8-NEXT:    s_mul_i32 s1, s0, s1
 ; GFX8-NEXT:    v_mov_b32_e32 v1, local_var32@abs32@lo
 ; GFX8-NEXT:    v_mov_b32_e32 v2, s1
 ; GFX8-NEXT:    s_mov_b32 m0, -1
 ; GFX8-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX8-NEXT:    ds_sub_rtn_u32 v1, v1, v2
 ; GFX8-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX8-NEXT:    buffer_wbinvl1_vol
 ; GFX8-NEXT:  BB9_2:
-; GFX8-NEXT:    s_or_b64 exec, exec, s[2:3]
+; GFX8-NEXT:    s_or_b64 exec, exec, s[6:7]
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX8-NEXT:    v_mul_lo_u32 v0, s0, v0
 ; GFX8-NEXT:    v_readfirstlane_b32 s0, v1
 ; GFX8-NEXT:    s_mov_b32 s7, 0xf000
 ; GFX8-NEXT:    s_mov_b32 s6, -1
 ; GFX8-NEXT:    v_sub_u32_e32 v0, vcc, s0, v0
 ; GFX8-NEXT:    buffer_store_dword v0, off, s[4:7], 0
 ; GFX8-NEXT:    s_endpgm
 ;
 ; GFX9-LABEL: sub_i32_uniform:
 ; GFX9:       ; %bb.0: ; %entry
 ; GFX9-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x24
 ; GFX9-NEXT:    s_load_dword s0, s[0:1], 0x2c
-; GFX9-NEXT:    v_cmp_ne_u32_e64 s[6:7], 1, 0
-; GFX9-NEXT:    v_mbcnt_lo_u32_b32 v0, s6, 0
-; GFX9-NEXT:    v_mbcnt_hi_u32_b32 v0, s7, v0
+; GFX9-NEXT:    s_mov_b64 s[2:3], exec
+; GFX9-NEXT:    v_mbcnt_lo_u32_b32 v0, s2, 0
+; GFX9-NEXT:    v_mbcnt_hi_u32_b32 v0, s3, v0
 ; GFX9-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX9-NEXT:    ; implicit-def: $vgpr1
-; GFX9-NEXT:    s_and_saveexec_b64 s[2:3], vcc
+; GFX9-NEXT:    s_and_saveexec_b64 s[6:7], vcc
 ; GFX9-NEXT:    s_cbranch_execz BB9_2
 ; GFX9-NEXT:  ; %bb.1:
-; GFX9-NEXT:    s_bcnt1_i32_b64 s1, s[6:7]
+; GFX9-NEXT:    s_bcnt1_i32_b64 s1, s[2:3]
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:    s_mul_i32 s1, s0, s1
 ; GFX9-NEXT:    v_mov_b32_e32 v1, local_var32@abs32@lo
 ; GFX9-NEXT:    v_mov_b32_e32 v2, s1
 ; GFX9-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX9-NEXT:    ds_sub_rtn_u32 v1, v1, v2
 ; GFX9-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX9-NEXT:    buffer_wbinvl1_vol
 ; GFX9-NEXT:  BB9_2:
-; GFX9-NEXT:    s_or_b64 exec, exec, s[2:3]
+; GFX9-NEXT:    s_or_b64 exec, exec, s[6:7]
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:    v_mul_lo_u32 v0, s0, v0
 ; GFX9-NEXT:    v_readfirstlane_b32 s0, v1
 ; GFX9-NEXT:    s_mov_b32 s7, 0xf000
 ; GFX9-NEXT:    s_mov_b32 s6, -1
 ; GFX9-NEXT:    v_sub_u32_e32 v0, s0, v0
 ; GFX9-NEXT:    buffer_store_dword v0, off, s[4:7], 0
 ; GFX9-NEXT:    s_endpgm
 ;
 ; GFX1064-LABEL: sub_i32_uniform:
 ; GFX1064:       ; %bb.0: ; %entry
-; GFX1064-NEXT:    v_cmp_ne_u32_e64 s[6:7], 1, 0
+; GFX1064-NEXT:    s_mov_b64 s[2:3], exec
 ; GFX1064-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x24
 ; GFX1064-NEXT:    s_load_dword s0, s[0:1], 0x2c
 ; GFX1064-NEXT:    ; implicit-def: $vgpr1
-; GFX1064-NEXT:    v_mbcnt_lo_u32_b32_e64 v0, s6, 0
-; GFX1064-NEXT:    v_mbcnt_hi_u32_b32_e64 v0, s7, v0
+; GFX1064-NEXT:    v_mbcnt_lo_u32_b32_e64 v0, s2, 0
+; GFX1064-NEXT:    v_mbcnt_hi_u32_b32_e64 v0, s3, v0
 ; GFX1064-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
-; GFX1064-NEXT:    s_and_saveexec_b64 s[2:3], vcc
+; GFX1064-NEXT:    s_and_saveexec_b64 s[6:7], vcc
 ; GFX1064-NEXT:    s_cbranch_execz BB9_2
 ; GFX1064-NEXT:  ; %bb.1:
-; GFX1064-NEXT:    s_bcnt1_i32_b64 s1, s[6:7]
+; GFX1064-NEXT:    s_bcnt1_i32_b64 s1, s[2:3]
 ; GFX1064-NEXT:    v_mov_b32_e32 v1, local_var32@abs32@lo
 ; GFX1064-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX1064-NEXT:    s_mul_i32 s1, s0, s1
 ; GFX1064-NEXT:    v_mov_b32_e32 v2, s1
 ; GFX1064-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX1064-NEXT:    s_waitcnt_vscnt null, 0x0
 ; GFX1064-NEXT:    ds_sub_rtn_u32 v1, v1, v2
 ; GFX1064-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX1064-NEXT:    buffer_gl0_inv
 ; GFX1064-NEXT:    buffer_gl1_inv
 ; GFX1064-NEXT:  BB9_2:
 ; GFX1064-NEXT:    v_nop
-; GFX1064-NEXT:    s_or_b64 exec, exec, s[2:3]
+; GFX1064-NEXT:    s_or_b64 exec, exec, s[6:7]
 ; GFX1064-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX1064-NEXT:    v_mul_lo_u32 v0, s0, v0
 ; GFX1064-NEXT:    v_readfirstlane_b32 s0, v1
 ; GFX1064-NEXT:    s_mov_b32 s7, 0x31016000
 ; GFX1064-NEXT:    s_mov_b32 s6, -1
 ; GFX1064-NEXT:    v_sub_nc_u32_e32 v0, s0, v0
 ; GFX1064-NEXT:    buffer_store_dword v0, off, s[4:7], 0
 ; GFX1064-NEXT:    s_endpgm
 ;
 ; GFX1032-LABEL: sub_i32_uniform:
 ; GFX1032:       ; %bb.0: ; %entry
 ; GFX1032-NEXT:    s_load_dwordx2 s[4:5], s[0:1], 0x24
 ; GFX1032-NEXT:    s_load_dword s0, s[0:1], 0x2c
-; GFX1032-NEXT:    v_cmp_ne_u32_e64 s2, 1, 0
+; GFX1032-NEXT:    s_mov_b32 s2, exec_lo
 ; GFX1032-NEXT:    ; implicit-def: $vcc_hi
 ; GFX1032-NEXT:    ; implicit-def: $vgpr1
 ; GFX1032-NEXT:    v_mbcnt_lo_u32_b32_e64 v0, s2, 0
 ; GFX1032-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 0, v0
 ; GFX1032-NEXT:    s_and_saveexec_b32 s1, vcc_lo
 ; GFX1032-NEXT:    s_cbranch_execz BB9_2
 ; GFX1032-NEXT:  ; %bb.1:
 ; GFX1032-NEXT:    s_bcnt1_i32_b32 s2, s2

 ; GFX1032-NEXT:    buffer_store_dword v0, off, s[4:7], 0
 ; GFX1032-NEXT:    s_endpgm
 entry:
   %old = atomicrmw sub i32 addrspace(3)* @local_var32, i32 %subitive acq_rel
   store i32 %old, i32 addrspace(1)* %out
   ret void
 }
 
-; GFX7LESS-NOT: v_mbcnt_lo_u32_b32
-; GFX7LESS-NOT: v_mbcnt_hi_u32_b32
-; GFX7LESS-NOT: s_bcnt1_i32_b64
-; DPPCOMB: v_add_u32_dpp
-; DPPCOMB: v_add_u32_dpp
-; GFX8MORE32: v_readlane_b32 s[[scalar_value:[0-9]+]], v{{[0-9]+}}, 31
-; GFX8MORE: v_mov_b32{{(_e[0-9]+)?}} v[[value:[0-9]+]], s[[scalar_value]]
-; GFX8MORE: ds_sub_rtn_u32 v{{[0-9]+}}, v{{[0-9]+}}, v[[value]]
 define amdgpu_kernel void @sub_i32_varying(i32 addrspace(1)* %out) {
 ;
 ;
 ; GFX7LESS-LABEL: sub_i32_varying:
 ; GFX7LESS:       ; %bb.0: ; %entry
 ; GFX7LESS-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x9
 ; GFX7LESS-NEXT:    v_mov_b32_e32 v1, local_var32@abs32@lo
 ; GFX7LESS-NEXT:    s_mov_b32 m0, -1
 ; GFX7LESS-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX7LESS-NEXT:    ds_sub_rtn_u32 v0, v1, v0
 ; GFX7LESS-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX7LESS-NEXT:    buffer_wbinvl1
 ; GFX7LESS-NEXT:    s_mov_b32 s3, 0xf000
 ; GFX7LESS-NEXT:    s_mov_b32 s2, -1
 ; GFX7LESS-NEXT:    buffer_store_dword v0, off, s[0:3], 0
 ; GFX7LESS-NEXT:    s_endpgm
 ;
 ; GFX8-LABEL: sub_i32_varying:
 ; GFX8:       ; %bb.0: ; %entry
 ; GFX8-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
+; GFX8-NEXT:    s_mov_b64 s[2:3], exec
 ; GFX8-NEXT:    v_mov_b32_e32 v2, v0
-; GFX8-NEXT:    s_or_saveexec_b64 s[2:3], -1
+; GFX8-NEXT:    s_or_saveexec_b64 s[4:5], -1
 ; GFX8-NEXT:    v_mov_b32_e32 v1, 0
-; GFX8-NEXT:    s_mov_b64 exec, s[2:3]
-; GFX8-NEXT:    v_cmp_ne_u32_e64 s[2:3], 1, 0
+; GFX8-NEXT:    s_mov_b64 exec, s[4:5]
 ; GFX8-NEXT:    v_mbcnt_lo_u32_b32 v0, s2, 0
 ; GFX8-NEXT:    v_mbcnt_hi_u32_b32 v0, s3, v0
 ; GFX8-NEXT:    s_not_b64 exec, exec
 ; GFX8-NEXT:    v_mov_b32_e32 v2, 0
 ; GFX8-NEXT:    s_not_b64 exec, exec
 ; GFX8-NEXT:    s_or_saveexec_b64 s[4:5], -1
 ; GFX8-NEXT:    v_add_u32_dpp v2, vcc, v2, v2 row_shr:1 row_mask:0xf bank_mask:0xf bound_ctrl:0
 ; GFX8-NEXT:    s_nop 1

 ; GFX8-NEXT:    s_nop 0
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX8-NEXT:    buffer_store_dword v0, off, s[0:3], 0
 ; GFX8-NEXT:    s_endpgm
 ;
 ; GFX9-LABEL: sub_i32_varying:
 ; GFX9:       ; %bb.0: ; %entry
 ; GFX9-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
+; GFX9-NEXT:    s_mov_b64 s[2:3], exec
 ; GFX9-NEXT:    v_mov_b32_e32 v2, v0
-; GFX9-NEXT:    s_or_saveexec_b64 s[2:3], -1
+; GFX9-NEXT:    s_or_saveexec_b64 s[4:5], -1
 ; GFX9-NEXT:    v_mov_b32_e32 v1, 0
-; GFX9-NEXT:    s_mov_b64 exec, s[2:3]
-; GFX9-NEXT:    v_cmp_ne_u32_e64 s[2:3], 1, 0
+; GFX9-NEXT:    s_mov_b64 exec, s[4:5]
 ; GFX9-NEXT:    v_mbcnt_lo_u32_b32 v0, s2, 0
 ; GFX9-NEXT:    v_mbcnt_hi_u32_b32 v0, s3, v0
 ; GFX9-NEXT:    s_not_b64 exec, exec
 ; GFX9-NEXT:    v_mov_b32_e32 v2, 0
 ; GFX9-NEXT:    s_not_b64 exec, exec
 ; GFX9-NEXT:    s_or_saveexec_b64 s[4:5], -1
 ; GFX9-NEXT:    v_add_u32_dpp v2, v2, v2 row_shr:1 row_mask:0xf bank_mask:0xf bound_ctrl:0
 ; GFX9-NEXT:    s_nop 1

 ; GFX9-NEXT:    s_nop 0
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:    buffer_store_dword v0, off, s[0:3], 0
 ; GFX9-NEXT:    s_endpgm
 ;
 ; GFX1064-LABEL: sub_i32_varying:
 ; GFX1064:       ; %bb.0: ; %entry
 ; GFX1064-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
+; GFX1064-NEXT:    s_mov_b64 s[2:3], exec
 ; GFX1064-NEXT:    v_mov_b32_e32 v2, v0
-; GFX1064-NEXT:    s_or_saveexec_b64 s[2:3], -1
+; GFX1064-NEXT:    s_or_saveexec_b64 s[4:5], -1
 ; GFX1064-NEXT:    v_mov_b32_e32 v1, 0
-; GFX1064-NEXT:    s_mov_b64 exec, s[2:3]
-; GFX1064-NEXT:    v_cmp_ne_u32_e64 s[2:3], 1, 0
+; GFX1064-NEXT:    s_mov_b64 exec, s[4:5]
 ; GFX1064-NEXT:    v_mbcnt_lo_u32_b32_e64 v0, s2, 0
 ; GFX1064-NEXT:    v_mbcnt_hi_u32_b32_e64 v0, s3, v0
 ; GFX1064-NEXT:    s_not_b64 exec, exec
 ; GFX1064-NEXT:    v_mov_b32_e32 v2, 0
 ; GFX1064-NEXT:    s_not_b64 exec, exec
 ; GFX1064-NEXT:    s_or_saveexec_b64 s[4:5], -1
 ; GFX1064-NEXT:    v_add_nc_u32_dpp v2, v2, v2 row_shr:1 row_mask:0xf bank_mask:0xf bound_ctrl:0
 ; GFX1064-NEXT:    v_add_nc_u32_dpp v2, v2, v2 row_shr:2 row_mask:0xf bank_mask:0xf bound_ctrl:0

 ; GFX1064-NEXT:    s_nop 1
 ; GFX1064-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX1064-NEXT:    buffer_store_dword v0, off, s[0:3], 0
 ; GFX1064-NEXT:    s_endpgm
 ;
 ; GFX1032-LABEL: sub_i32_varying:
 ; GFX1032:       ; %bb.0: ; %entry
 ; GFX1032-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
+; GFX1032-NEXT:    s_mov_b32 s2, exec_lo
 ; GFX1032-NEXT:    ; implicit-def: $vcc_hi
 ; GFX1032-NEXT:    v_mov_b32_e32 v2, v0
-; GFX1032-NEXT:    s_or_saveexec_b32 s2, -1
+; GFX1032-NEXT:    s_or_saveexec_b32 s3, -1
 ; GFX1032-NEXT:    v_mov_b32_e32 v1, 0
-; GFX1032-NEXT:    s_mov_b32 exec_lo, s2
-; GFX1032-NEXT:    v_cmp_ne_u32_e64 s2, 1, 0
+; GFX1032-NEXT:    s_mov_b32 exec_lo, s3
 ; GFX1032-NEXT:    v_mbcnt_lo_u32_b32_e64 v0, s2, 0
 ; GFX1032-NEXT:    s_not_b32 exec_lo, exec_lo
 ; GFX1032-NEXT:    v_mov_b32_e32 v2, 0
 ; GFX1032-NEXT:    s_not_b32 exec_lo, exec_lo
 ; GFX1032-NEXT:    s_or_saveexec_b32 s4, -1
 ; GFX1032-NEXT:    v_add_nc_u32_dpp v2, v2, v2 row_shr:1 row_mask:0xf bank_mask:0xf bound_ctrl:0
 ; GFX1032-NEXT:    s_mov_b32 s2, -1
 ; GFX1032-NEXT:    v_add_nc_u32_dpp v2, v2, v2 row_shr:2 row_mask:0xf bank_mask:0xf bound_ctrl:0
entry:
   ret void
 }
 
 define amdgpu_kernel void @sub_i64_constant(i64 addrspace(1)* %out) {
 ;
 ;
 ; GFX7LESS-LABEL: sub_i64_constant:
 ; GFX7LESS:       ; %bb.0: ; %entry
+; GFX7LESS-NEXT:    s_mov_b64 s[4:5], exec
 ; GFX7LESS-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x9
-; GFX7LESS-NEXT:    v_cmp_ne_u32_e64 s[4:5], 1, 0
 ; GFX7LESS-NEXT:    v_mbcnt_lo_u32_b32_e64 v0, s4, 0
 ; GFX7LESS-NEXT:    v_mbcnt_hi_u32_b32_e32 v0, s5, v0
 ; GFX7LESS-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX7LESS-NEXT:    ; implicit-def: $vgpr1_vgpr2
 ; GFX7LESS-NEXT:    s_and_saveexec_b64 s[2:3], vcc
 ; GFX7LESS-NEXT:    s_cbranch_execz BB11_2
 ; GFX7LESS-NEXT:  ; %bb.1:
 ; GFX7LESS-NEXT:    s_bcnt1_i32_b64 s4, s[4:5]

 ; GFX7LESS-NEXT:    s_mov_b32 s2, -1
 ; GFX7LESS-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX7LESS-NEXT:    buffer_store_dwordx2 v[0:1], off, s[0:3], 0
 ; GFX7LESS-NEXT:    s_endpgm
 ;
 ; GFX8-LABEL: sub_i64_constant:
 ; GFX8:       ; %bb.0: ; %entry
 ; GFX8-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
-; GFX8-NEXT:    v_cmp_ne_u32_e64 s[4:5], 1, 0
+; GFX8-NEXT:    s_mov_b64 s[4:5], exec
 ; GFX8-NEXT:    v_mbcnt_lo_u32_b32 v0, s4, 0
 ; GFX8-NEXT:    v_mbcnt_hi_u32_b32 v0, s5, v0
 ; GFX8-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX8-NEXT:    ; implicit-def: $vgpr1_vgpr2
 ; GFX8-NEXT:    s_and_saveexec_b64 s[2:3], vcc
 ; GFX8-NEXT:    s_cbranch_execz BB11_2
 ; GFX8-NEXT:  ; %bb.1:
 ; GFX8-NEXT:    s_bcnt1_i32_b64 s4, s[4:5]

 ; GFX8-NEXT:    s_mov_b32 s2, -1
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX8-NEXT:    buffer_store_dwordx2 v[0:1], off, s[0:3], 0
 ; GFX8-NEXT:    s_endpgm
 ;
 ; GFX9-LABEL: sub_i64_constant:
 ; GFX9:       ; %bb.0: ; %entry
 ; GFX9-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
-; GFX9-NEXT:    v_cmp_ne_u32_e64 s[4:5], 1, 0
+; GFX9-NEXT:    s_mov_b64 s[4:5], exec
 ; GFX9-NEXT:    v_mbcnt_lo_u32_b32 v0, s4, 0
 ; GFX9-NEXT:    v_mbcnt_hi_u32_b32 v0, s5, v0
 ; GFX9-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX9-NEXT:    ; implicit-def: $vgpr1_vgpr2
 ; GFX9-NEXT:    s_and_saveexec_b64 s[2:3], vcc
 ; GFX9-NEXT:    s_cbranch_execz BB11_2
 ; GFX9-NEXT:  ; %bb.1:
 ; GFX9-NEXT:    s_bcnt1_i32_b64 s4, s[4:5]

 ; GFX9-NEXT:    s_mov_b32 s3, 0xf000
 ; GFX9-NEXT:    s_mov_b32 s2, -1
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:    buffer_store_dwordx2 v[0:1], off, s[0:3], 0
 ; GFX9-NEXT:    s_endpgm
 ;
 ; GFX1064-LABEL: sub_i64_constant:
 ; GFX1064:       ; %bb.0: ; %entry
-; GFX1064-NEXT:    v_cmp_ne_u32_e64 s[4:5], 1, 0
+; GFX1064-NEXT:    s_mov_b64 s[4:5], exec
 ; GFX1064-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
 ; GFX1064-NEXT:    ; implicit-def: $vgpr1_vgpr2
 ; GFX1064-NEXT:    v_mbcnt_lo_u32_b32_e64 v0, s4, 0
 ; GFX1064-NEXT:    v_mbcnt_hi_u32_b32_e64 v0, s5, v0
 ; GFX1064-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX1064-NEXT:    s_and_saveexec_b64 s[2:3], vcc
 ; GFX1064-NEXT:    s_cbranch_execz BB11_2
 ; GFX1064-NEXT:  ; %bb.1:

 ; GFX1064-NEXT:    s_mov_b32 s3, 0x31016000
 ; GFX1064-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX1064-NEXT:    buffer_store_dwordx2 v[0:1], off, s[0:3], 0
 ; GFX1064-NEXT:    s_endpgm
 ;
 ; GFX1032-LABEL: sub_i64_constant:
 ; GFX1032:       ; %bb.0: ; %entry
 ; GFX1032-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
-; GFX1032-NEXT:    v_cmp_ne_u32_e64 s3, 1, 0
+; GFX1032-NEXT:    s_mov_b32 s3, exec_lo
 ; GFX1032-NEXT:    ; implicit-def: $vcc_hi
 ; GFX1032-NEXT:    ; implicit-def: $vgpr1_vgpr2
 ; GFX1032-NEXT:    v_mbcnt_lo_u32_b32_e64 v0, s3, 0
 ; GFX1032-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 0, v0
 ; GFX1032-NEXT:    s_and_saveexec_b32 s2, vcc_lo
 ; GFX1032-NEXT:    s_cbranch_execz BB11_2
 ; GFX1032-NEXT:  ; %bb.1:
 ; GFX1032-NEXT:    s_bcnt1_i32_b32 s3, s3
entry:
   ret void
 }
 
 define amdgpu_kernel void @sub_i64_uniform(i64 addrspace(1)* %out, i64 %subitive) {
 ;
 ;
 ; GFX7LESS-LABEL: sub_i64_uniform:
 ; GFX7LESS:       ; %bb.0: ; %entry
+; GFX7LESS-NEXT:    s_mov_b64 s[6:7], exec
 ; GFX7LESS-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x9
-; GFX7LESS-NEXT:    v_cmp_ne_u32_e64 s[6:7], 1, 0
 ; GFX7LESS-NEXT:    v_mbcnt_lo_u32_b32_e64 v0, s6, 0
 ; GFX7LESS-NEXT:    v_mbcnt_hi_u32_b32_e32 v0, s7, v0
 ; GFX7LESS-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX7LESS-NEXT:    ; implicit-def: $vgpr1_vgpr2
 ; GFX7LESS-NEXT:    s_and_saveexec_b64 s[4:5], vcc
 ; GFX7LESS-NEXT:    s_cbranch_execz BB12_2
 ; GFX7LESS-NEXT:  ; %bb.1:
 ; GFX7LESS-NEXT:    s_bcnt1_i32_b64 s6, s[6:7]

 ; GFX7LESS-NEXT:    v_sub_i32_e32 v0, vcc, s0, v0
 ; GFX7LESS-NEXT:    v_subb_u32_e32 v1, vcc, v2, v1, vcc
 ; GFX7LESS-NEXT:    buffer_store_dwordx2 v[0:1], off, s[4:7], 0
 ; GFX7LESS-NEXT:    s_endpgm
 ;
 ; GFX8-LABEL: sub_i64_uniform:
 ; GFX8:       ; %bb.0: ; %entry
 ; GFX8-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x24
-; GFX8-NEXT:    v_cmp_ne_u32_e64 s[6:7], 1, 0
+; GFX8-NEXT:    s_mov_b64 s[6:7], exec
 ; GFX8-NEXT:    v_mbcnt_lo_u32_b32 v0, s6, 0
 ; GFX8-NEXT:    v_mbcnt_hi_u32_b32 v0, s7, v0
 ; GFX8-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX8-NEXT:    ; implicit-def: $vgpr1_vgpr2
 ; GFX8-NEXT:    s_and_saveexec_b64 s[4:5], vcc
 ; GFX8-NEXT:    s_cbranch_execz BB12_2
 ; GFX8-NEXT:  ; %bb.1:
 ; GFX8-NEXT:    s_bcnt1_i32_b64 s6, s[6:7]

 ; GFX8-NEXT:    s_mov_b32 s6, -1
 ; GFX8-NEXT:    v_subb_u32_e32 v1, vcc, v2, v1, vcc
 ; GFX8-NEXT:    buffer_store_dwordx2 v[0:1], off, s[4:7], 0
 ; GFX8-NEXT:    s_endpgm
 ;
 ; GFX9-LABEL: sub_i64_uniform:
 ; GFX9:       ; %bb.0: ; %entry
 ; GFX9-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x24
-; GFX9-NEXT:    v_cmp_ne_u32_e64 s[6:7], 1, 0
+; GFX9-NEXT:    s_mov_b64 s[6:7], exec
 ; GFX9-NEXT:    v_mbcnt_lo_u32_b32 v0, s6, 0
 ; GFX9-NEXT:    v_mbcnt_hi_u32_b32 v0, s7, v0
 ; GFX9-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX9-NEXT:    ; implicit-def: $vgpr1_vgpr2
 ; GFX9-NEXT:    s_and_saveexec_b64 s[4:5], vcc
 ; GFX9-NEXT:    s_cbranch_execz BB12_2
 ; GFX9-NEXT:  ; %bb.1:
 ; GFX9-NEXT:    s_bcnt1_i32_b64 s6, s[6:7]

 ; GFX9-NEXT:    s_mov_b32 s7, 0xf000
 ; GFX9-NEXT:    s_mov_b32 s6, -1
 ; GFX9-NEXT:    v_subb_co_u32_e32 v1, vcc, v2, v1, vcc
 ; GFX9-NEXT:    buffer_store_dwordx2 v[0:1], off, s[4:7], 0
 ; GFX9-NEXT:    s_endpgm
 ;
 ; GFX1064-LABEL: sub_i64_uniform:
 ; GFX1064:       ; %bb.0: ; %entry
-; GFX1064-NEXT:    v_cmp_ne_u32_e64 s[6:7], 1, 0
+; GFX1064-NEXT:    s_mov_b64 s[6:7], exec
 ; GFX1064-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x24
 ; GFX1064-NEXT:    ; implicit-def: $vgpr1_vgpr2
 ; GFX1064-NEXT:    v_mbcnt_lo_u32_b32_e64 v0, s6, 0
 ; GFX1064-NEXT:    v_mbcnt_hi_u32_b32_e64 v0, s7, v0
 ; GFX1064-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX1064-NEXT:    s_and_saveexec_b64 s[4:5], vcc
 ; GFX1064-NEXT:    s_cbranch_execz BB12_2
 ; GFX1064-NEXT:  ; %bb.1:

 ; GFX1064-NEXT:    v_sub_co_u32_e64 v0, vcc, s4, v0
 ; GFX1064-NEXT:    v_sub_co_ci_u32_e32 v1, vcc, s5, v1, vcc
 ; GFX1064-NEXT:    buffer_store_dwordx2 v[0:1], off, s[0:3], 0
 ; GFX1064-NEXT:    s_endpgm
 ;
 ; GFX1032-LABEL: sub_i64_uniform:
 ; GFX1032:       ; %bb.0: ; %entry
 ; GFX1032-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x24
-; GFX1032-NEXT:    v_cmp_ne_u32_e64 s5, 1, 0
+; GFX1032-NEXT:    s_mov_b32 s5, exec_lo
 ; GFX1032-NEXT:    ; implicit-def: $vcc_hi
 ; GFX1032-NEXT:    ; implicit-def: $vgpr1_vgpr2
 ; GFX1032-NEXT:    v_mbcnt_lo_u32_b32_e64 v0, s5, 0
 ; GFX1032-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 0, v0
 ; GFX1032-NEXT:    s_and_saveexec_b32 s4, vcc_lo
 ; GFX1032-NEXT:    s_cbranch_execz BB12_2
 ; GFX1032-NEXT:  ; %bb.1:
 ; GFX1032-NEXT:    s_bcnt1_i32_b32 s5, s5

 entry:
   %lane = call i32 @llvm.amdgcn.workitem.id.x()
   %zext = zext i32 %lane to i64
   %old = atomicrmw sub i64 addrspace(3)* @local_var64, i64 %zext acq_rel
   store i64 %old, i64 addrspace(1)* %out
   ret void
 }
 
-; GFX8MORE32: v_readlane_b32 s[[scalar_value:[0-9]+]], v{{[0-9]+}}, 31
-; GFX8MORE: v_mov_b32{{(_e[0-9]+)?}} v[[value:[0-9]+]], s[[scalar_value]]
-; GFX8MORE: ds_and_rtn_b32 v{{[0-9]+}}, v{{[0-9]+}}, v[[value]]
 define amdgpu_kernel void @and_i32_varying(i32 addrspace(1)* %out) {
 ;
 ;
 ; GFX7LESS-LABEL: and_i32_varying:
 ; GFX7LESS:       ; %bb.0: ; %entry
 ; GFX7LESS-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x9
 ; GFX7LESS-NEXT:    v_mov_b32_e32 v1, local_var32@abs32@lo
 ; GFX7LESS-NEXT:    s_mov_b32 m0, -1
 ; GFX7LESS-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX7LESS-NEXT:    ds_and_rtn_b32 v0, v1, v0
 ; GFX7LESS-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX7LESS-NEXT:    buffer_wbinvl1
 ; GFX7LESS-NEXT:    s_mov_b32 s3, 0xf000
 ; GFX7LESS-NEXT:    s_mov_b32 s2, -1
 ; GFX7LESS-NEXT:    buffer_store_dword v0, off, s[0:3], 0
 ; GFX7LESS-NEXT:    s_endpgm
 ;
 ; GFX8-LABEL: and_i32_varying:
 ; GFX8:       ; %bb.0: ; %entry
 ; GFX8-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
-; GFX8-NEXT:    v_cmp_ne_u32_e64 s[2:3], 1, 0
-; GFX8-NEXT:    v_mbcnt_lo_u32_b32 v3, s2, 0
-; GFX8-NEXT:    v_mbcnt_hi_u32_b32 v3, s3, v3
+; GFX8-NEXT:    v_mbcnt_lo_u32_b32 v3, exec_lo, 0
+; GFX8-NEXT:    v_mbcnt_hi_u32_b32 v3, exec_hi, v3
 ; GFX8-NEXT:    v_mov_b32_e32 v2, v0
 ; GFX8-NEXT:    s_or_saveexec_b64 s[2:3], -1
 ; GFX8-NEXT:    v_mov_b32_e32 v1, -1
 ; GFX8-NEXT:    s_mov_b64 exec, s[2:3]
 ; GFX8-NEXT:    s_not_b64 exec, exec
 ; GFX8-NEXT:    v_mov_b32_e32 v2, -1
 ; GFX8-NEXT:    s_not_b64 exec, exec
 ; GFX8-NEXT:    s_or_saveexec_b64 s[4:5], -1

 ; GFX8-NEXT:    s_nop 0
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX8-NEXT:    buffer_store_dword v0, off, s[0:3], 0
 ; GFX8-NEXT:    s_endpgm
 ;
 ; GFX9-LABEL: and_i32_varying:
 ; GFX9:       ; %bb.0: ; %entry
 ; GFX9-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
-; GFX9-NEXT:    v_cmp_ne_u32_e64 s[2:3], 1, 0
-; GFX9-NEXT:    v_mbcnt_lo_u32_b32 v3, s2, 0
-; GFX9-NEXT:    v_mbcnt_hi_u32_b32 v3, s3, v3
+; GFX9-NEXT:    v_mbcnt_lo_u32_b32 v3, exec_lo, 0
+; GFX9-NEXT:    v_mbcnt_hi_u32_b32 v3, exec_hi, v3
 ; GFX9-NEXT:    v_mov_b32_e32 v2, v0
 ; GFX9-NEXT:    s_or_saveexec_b64 s[2:3], -1
 ; GFX9-NEXT:    v_mov_b32_e32 v1, -1
 ; GFX9-NEXT:    s_mov_b64 exec, s[2:3]
 ; GFX9-NEXT:    s_not_b64 exec, exec
 ; GFX9-NEXT:    v_mov_b32_e32 v2, -1
 ; GFX9-NEXT:    s_not_b64 exec, exec
 ; GFX9-NEXT:    s_or_saveexec_b64 s[4:5], -1

 ; GFX9-NEXT:    s_nop 0
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:    buffer_store_dword v0, off, s[0:3], 0
 ; GFX9-NEXT:    s_endpgm
 ;
 ; GFX1064-LABEL: and_i32_varying:
 ; GFX1064:       ; %bb.0: ; %entry
 ; GFX1064-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
-; GFX1064-NEXT:    v_cmp_ne_u32_e64 s[2:3], 1, 0
+; GFX1064-NEXT:    v_mbcnt_lo_u32_b32_e64 v4, exec_lo, 0
 ; GFX1064-NEXT:    v_mov_b32_e32 v2, v0
-; GFX1064-NEXT:    v_mbcnt_lo_u32_b32_e64 v4, s2, 0
-; GFX1064-NEXT:    v_mbcnt_hi_u32_b32_e64 v4, s3, v4
+; GFX1064-NEXT:    v_mbcnt_hi_u32_b32_e64 v4, exec_hi, v4
 ; GFX1064-NEXT:    s_or_saveexec_b64 s[2:3], -1
 ; GFX1064-NEXT:    v_mov_b32_e32 v1, -1
 ; GFX1064-NEXT:    s_mov_b64 exec, s[2:3]
 ; GFX1064-NEXT:    s_not_b64 exec, exec
 ; GFX1064-NEXT:    v_mov_b32_e32 v2, -1
 ; GFX1064-NEXT:    s_not_b64 exec, exec
 ; GFX1064-NEXT:    s_or_saveexec_b64 s[4:5], -1
 ; GFX1064-NEXT:    v_and_b32_dpp v2, v2, v2 row_shr:1 row_mask:0xf bank_mask:0xf

 ; GFX1064-NEXT:    s_nop 1
 ; GFX1064-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX1064-NEXT:    buffer_store_dword v0, off, s[0:3], 0
 ; GFX1064-NEXT:    s_endpgm
 ;
 ; GFX1032-LABEL: and_i32_varying:
 ; GFX1032:       ; %bb.0: ; %entry
 ; GFX1032-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
-; GFX1032-NEXT:    v_cmp_ne_u32_e64 s2, 1, 0
+; GFX1032-NEXT:    v_mbcnt_lo_u32_b32_e64 v4, exec_lo, 0
 ; GFX1032-NEXT:    ; implicit-def: $vcc_hi
 ; GFX1032-NEXT:    v_mov_b32_e32 v2, v0
-; GFX1032-NEXT:    v_mbcnt_lo_u32_b32_e64 v4, s2, 0
 ; GFX1032-NEXT:    s_or_saveexec_b32 s2, -1
 ; GFX1032-NEXT:    v_mov_b32_e32 v1, -1
 ; GFX1032-NEXT:    s_mov_b32 exec_lo, s2
 ; GFX1032-NEXT:    s_not_b32 exec_lo, exec_lo
 ; GFX1032-NEXT:    v_mov_b32_e32 v2, -1
 ; GFX1032-NEXT:    s_not_b32 exec_lo, exec_lo
 ; GFX1032-NEXT:    s_or_saveexec_b32 s4, -1
 ; GFX1032-NEXT:    v_and_b32_dpp v2, v2, v2 row_shr:1 row_mask:0xf bank_mask:0xf

 ; GFX1032-NEXT:    s_endpgm
 entry:
   %lane = call i32 @llvm.amdgcn.workitem.id.x()
   %old = atomicrmw and i32 addrspace(3)* @local_var32, i32 %lane acq_rel
   store i32 %old, i32 addrspace(1)* %out
   ret void
 }
 
-; GFX8MORE32: v_readlane_b32 s[[scalar_value:[0-9]+]], v{{[0-9]+}}, 31
-; GFX8MORE: v_mov_b32{{(_e[0-9]+)?}} v[[value:[0-9]+]], s[[scalar_value]]
-; GFX8MORE: ds_or_rtn_b32 v{{[0-9]+}}, v{{[0-9]+}}, v[[value]]
 define amdgpu_kernel void @or_i32_varying(i32 addrspace(1)* %out) {
 ;
 ;
 ; GFX7LESS-LABEL: or_i32_varying:
 ; GFX7LESS:       ; %bb.0: ; %entry
 ; GFX7LESS-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x9
 ; GFX7LESS-NEXT:    v_mov_b32_e32 v1, local_var32@abs32@lo
 ; GFX7LESS-NEXT:    s_mov_b32 m0, -1
 ; GFX7LESS-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX7LESS-NEXT:    ds_or_rtn_b32 v0, v1, v0
 ; GFX7LESS-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX7LESS-NEXT:    buffer_wbinvl1
 ; GFX7LESS-NEXT:    s_mov_b32 s3, 0xf000
 ; GFX7LESS-NEXT:    s_mov_b32 s2, -1
 ; GFX7LESS-NEXT:    buffer_store_dword v0, off, s[0:3], 0
 ; GFX7LESS-NEXT:    s_endpgm
 ;
 ; GFX8-LABEL: or_i32_varying:
 ; GFX8:       ; %bb.0: ; %entry
 ; GFX8-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
+; GFX8-NEXT:    s_mov_b64 s[2:3], exec
 ; GFX8-NEXT:    v_mov_b32_e32 v2, v0
-; GFX8-NEXT:    s_or_saveexec_b64 s[2:3], -1
+; GFX8-NEXT:    s_or_saveexec_b64 s[4:5], -1
 ; GFX8-NEXT:    v_mov_b32_e32 v1, 0
-; GFX8-NEXT:    s_mov_b64 exec, s[2:3]
-; GFX8-NEXT:    v_cmp_ne_u32_e64 s[2:3], 1, 0
+; GFX8-NEXT:    s_mov_b64 exec, s[4:5]
 ; GFX8-NEXT:    v_mbcnt_lo_u32_b32 v0, s2, 0
 ; GFX8-NEXT:    v_mbcnt_hi_u32_b32 v0, s3, v0
 ; GFX8-NEXT:    s_not_b64 exec, exec
 ; GFX8-NEXT:    v_mov_b32_e32 v2, 0
 ; GFX8-NEXT:    s_not_b64 exec, exec
 ; GFX8-NEXT:    s_or_saveexec_b64 s[4:5], -1
 ; GFX8-NEXT:    v_or_b32_dpp v2, v2, v2 row_shr:1 row_mask:0xf bank_mask:0xf bound_ctrl:0
 ; GFX8-NEXT:    s_nop 1

 ; GFX8-NEXT:    s_nop 0
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX8-NEXT:    buffer_store_dword v0, off, s[0:3], 0
 ; GFX8-NEXT:    s_endpgm
 ;
 ; GFX9-LABEL: or_i32_varying:
 ; GFX9:       ; %bb.0: ; %entry
 ; GFX9-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
+; GFX9-NEXT:    s_mov_b64 s[2:3], exec
 ; GFX9-NEXT:    v_mov_b32_e32 v2, v0
-; GFX9-NEXT:    s_or_saveexec_b64 s[2:3], -1
+; GFX9-NEXT:    s_or_saveexec_b64 s[4:5], -1
 ; GFX9-NEXT:    v_mov_b32_e32 v1, 0
-; GFX9-NEXT:    s_mov_b64 exec, s[2:3]
-; GFX9-NEXT:    v_cmp_ne_u32_e64 s[2:3], 1, 0
+; GFX9-NEXT:    s_mov_b64 exec, s[4:5]
 ; GFX9-NEXT:    v_mbcnt_lo_u32_b32 v0, s2, 0
 ; GFX9-NEXT:    v_mbcnt_hi_u32_b32 v0, s3, v0
 ; GFX9-NEXT:    s_not_b64 exec, exec
 ; GFX9-NEXT:    v_mov_b32_e32 v2, 0
 ; GFX9-NEXT:    s_not_b64 exec, exec
 ; GFX9-NEXT:    s_or_saveexec_b64 s[4:5], -1
 ; GFX9-NEXT:    v_or_b32_dpp v2, v2, v2 row_shr:1 row_mask:0xf bank_mask:0xf bound_ctrl:0
 ; GFX9-NEXT:    s_nop 1

 ; GFX9-NEXT:    s_nop 0
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:    buffer_store_dword v0, off, s[0:3], 0
 ; GFX9-NEXT:    s_endpgm
 ;
 ; GFX1064-LABEL: or_i32_varying:
 ; GFX1064:       ; %bb.0: ; %entry
 ; GFX1064-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
+; GFX1064-NEXT:    s_mov_b64 s[2:3], exec
 ; GFX1064-NEXT:    v_mov_b32_e32 v2, v0
-; GFX1064-NEXT:    s_or_saveexec_b64 s[2:3], -1
+; GFX1064-NEXT:    s_or_saveexec_b64 s[4:5], -1
 ; GFX1064-NEXT:    v_mov_b32_e32 v1, 0
-; GFX1064-NEXT:    s_mov_b64 exec, s[2:3]
-; GFX1064-NEXT:    v_cmp_ne_u32_e64 s[2:3], 1, 0
+; GFX1064-NEXT:    s_mov_b64 exec, s[4:5]
 ; GFX1064-NEXT:    v_mbcnt_lo_u32_b32_e64 v0, s2, 0
 ; GFX1064-NEXT:    v_mbcnt_hi_u32_b32_e64 v0, s3, v0
 ; GFX1064-NEXT:    s_not_b64 exec, exec
 ; GFX1064-NEXT:    v_mov_b32_e32 v2, 0
 ; GFX1064-NEXT:    s_not_b64 exec, exec
 ; GFX1064-NEXT:    s_or_saveexec_b64 s[4:5], -1
 ; GFX1064-NEXT:    v_or_b32_dpp v2, v2, v2 row_shr:1 row_mask:0xf bank_mask:0xf bound_ctrl:0
 ; GFX1064-NEXT:    v_or_b32_dpp v2, v2, v2 row_shr:2 row_mask:0xf bank_mask:0xf bound_ctrl:0

 ; GFX1064-NEXT:    s_nop 1
 ; GFX1064-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX1064-NEXT:    buffer_store_dword v0, off, s[0:3], 0
 ; GFX1064-NEXT:    s_endpgm
 ;
 ; GFX1032-LABEL: or_i32_varying:
 ; GFX1032:       ; %bb.0: ; %entry
 ; GFX1032-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
+; GFX1032-NEXT:    s_mov_b32 s2, exec_lo
 ; GFX1032-NEXT:    ; implicit-def: $vcc_hi
 ; GFX1032-NEXT:    v_mov_b32_e32 v2, v0
-; GFX1032-NEXT:    s_or_saveexec_b32 s2, -1
+; GFX1032-NEXT:    s_or_saveexec_b32 s3, -1
 ; GFX1032-NEXT:    v_mov_b32_e32 v1, 0
-; GFX1032-NEXT:    s_mov_b32 exec_lo, s2
-; GFX1032-NEXT:    v_cmp_ne_u32_e64 s2, 1, 0
+; GFX1032-NEXT:    s_mov_b32 exec_lo, s3
 ; GFX1032-NEXT:    v_mbcnt_lo_u32_b32_e64 v0, s2, 0
 ; GFX1032-NEXT:    s_not_b32 exec_lo, exec_lo
 ; GFX1032-NEXT:    v_mov_b32_e32 v2, 0
 ; GFX1032-NEXT:    s_not_b32 exec_lo, exec_lo
 ; GFX1032-NEXT:    s_or_saveexec_b32 s4, -1
 ; GFX1032-NEXT:    v_or_b32_dpp v2, v2, v2 row_shr:1 row_mask:0xf bank_mask:0xf bound_ctrl:0
 ; GFX1032-NEXT:    s_mov_b32 s2, -1
 ; GFX1032-NEXT:    v_or_b32_dpp v2, v2, v2 row_shr:2 row_mask:0xf bank_mask:0xf bound_ctrl:0

 ; GFX1032-NEXT:    s_endpgm
 entry:
   %lane = call i32 @llvm.amdgcn.workitem.id.x()
   %old = atomicrmw or i32 addrspace(3)* @local_var32, i32 %lane acq_rel
   store i32 %old, i32 addrspace(1)* %out
   ret void
 }
 
-; GFX8MORE32: v_readlane_b32 s[[scalar_value:[0-9]+]], v{{[0-9]+}}, 31
-; GFX8MORE: v_mov_b32{{(_e[0-9]+)?}} v[[value:[0-9]+]], s[[scalar_value]]
-; GFX8MORE: ds_xor_rtn_b32 v{{[0-9]+}}, v{{[0-9]+}}, v[[value]]
 define amdgpu_kernel void @xor_i32_varying(i32 addrspace(1)* %out) {
 ;
 ;
 ; GFX7LESS-LABEL: xor_i32_varying:
 ; GFX7LESS:       ; %bb.0: ; %entry
 ; GFX7LESS-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x9
 ; GFX7LESS-NEXT:    v_mov_b32_e32 v1, local_var32@abs32@lo
 ; GFX7LESS-NEXT:    s_mov_b32 m0, -1
 ; GFX7LESS-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX7LESS-NEXT:    ds_xor_rtn_b32 v0, v1, v0
 ; GFX7LESS-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX7LESS-NEXT:    buffer_wbinvl1
 ; GFX7LESS-NEXT:    s_mov_b32 s3, 0xf000
 ; GFX7LESS-NEXT:    s_mov_b32 s2, -1
 ; GFX7LESS-NEXT:    buffer_store_dword v0, off, s[0:3], 0
 ; GFX7LESS-NEXT:    s_endpgm
 ;
 ; GFX8-LABEL: xor_i32_varying:
 ; GFX8:       ; %bb.0: ; %entry
 ; GFX8-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
+; GFX8-NEXT:    s_mov_b64 s[2:3], exec
 ; GFX8-NEXT:    v_mov_b32_e32 v2, v0
-; GFX8-NEXT:    s_or_saveexec_b64 s[2:3], -1
+; GFX8-NEXT:    s_or_saveexec_b64 s[4:5], -1
 ; GFX8-NEXT:    v_mov_b32_e32 v1, 0
-; GFX8-NEXT:    s_mov_b64 exec, s[2:3]
-; GFX8-NEXT:    v_cmp_ne_u32_e64 s[2:3], 1, 0
+; GFX8-NEXT:    s_mov_b64 exec, s[4:5]
 ; GFX8-NEXT:    v_mbcnt_lo_u32_b32 v0, s2, 0
 ; GFX8-NEXT:    v_mbcnt_hi_u32_b32 v0, s3, v0
 ; GFX8-NEXT:    s_not_b64 exec, exec
 ; GFX8-NEXT:    v_mov_b32_e32 v2, 0
 ; GFX8-NEXT:    s_not_b64 exec, exec
 ; GFX8-NEXT:    s_or_saveexec_b64 s[4:5], -1
 ; GFX8-NEXT:    v_xor_b32_dpp v2, v2, v2 row_shr:1 row_mask:0xf bank_mask:0xf bound_ctrl:0
 ; GFX8-NEXT:    s_nop 1

 ; GFX8-NEXT:    s_nop 0
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX8-NEXT:    buffer_store_dword v0, off, s[0:3], 0
 ; GFX8-NEXT:    s_endpgm
 ;
 ; GFX9-LABEL: xor_i32_varying:
 ; GFX9:       ; %bb.0: ; %entry
 ; GFX9-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
+; GFX9-NEXT:    s_mov_b64 s[2:3], exec
 ; GFX9-NEXT:    v_mov_b32_e32 v2, v0
-; GFX9-NEXT:    s_or_saveexec_b64 s[2:3], -1
+; GFX9-NEXT:    s_or_saveexec_b64 s[4:5], -1
 ; GFX9-NEXT:    v_mov_b32_e32 v1, 0
-; GFX9-NEXT:    s_mov_b64 exec, s[2:3]
-; GFX9-NEXT:    v_cmp_ne_u32_e64 s[2:3], 1, 0
+; GFX9-NEXT:    s_mov_b64 exec, s[4:5]
 ; GFX9-NEXT:    v_mbcnt_lo_u32_b32 v0, s2, 0
 ; GFX9-NEXT:    v_mbcnt_hi_u32_b32 v0, s3, v0
 ; GFX9-NEXT:    s_not_b64 exec, exec
 ; GFX9-NEXT:    v_mov_b32_e32 v2, 0
 ; GFX9-NEXT:    s_not_b64 exec, exec
 ; GFX9-NEXT:    s_or_saveexec_b64 s[4:5], -1
 ; GFX9-NEXT:    v_xor_b32_dpp v2, v2, v2 row_shr:1 row_mask:0xf bank_mask:0xf bound_ctrl:0
 ; GFX9-NEXT:    s_nop 1

 ; GFX9-NEXT:    s_nop 0
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:    buffer_store_dword v0, off, s[0:3], 0
 ; GFX9-NEXT:    s_endpgm
 ;
 ; GFX1064-LABEL: xor_i32_varying:
 ; GFX1064:       ; %bb.0: ; %entry
 ; GFX1064-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
+; GFX1064-NEXT:    s_mov_b64 s[2:3], exec
 ; GFX1064-NEXT:    v_mov_b32_e32 v2, v0
-; GFX1064-NEXT:    s_or_saveexec_b64 s[2:3], -1
+; GFX1064-NEXT:    s_or_saveexec_b64 s[4:5], -1
 ; GFX1064-NEXT:    v_mov_b32_e32 v1, 0
-; GFX1064-NEXT:    s_mov_b64 exec, s[2:3]
-; GFX1064-NEXT:    v_cmp_ne_u32_e64 s[2:3], 1, 0
+; GFX1064-NEXT:    s_mov_b64 exec, s[4:5]
 ; GFX1064-NEXT:    v_mbcnt_lo_u32_b32_e64 v0, s2, 0
 ; GFX1064-NEXT:    v_mbcnt_hi_u32_b32_e64 v0, s3, v0
 ; GFX1064-NEXT:    s_not_b64 exec, exec
 ; GFX1064-NEXT:    v_mov_b32_e32 v2, 0
 ; GFX1064-NEXT:    s_not_b64 exec, exec
 ; GFX1064-NEXT:    s_or_saveexec_b64 s[4:5], -1
 ; GFX1064-NEXT:    v_xor_b32_dpp v2, v2, v2 row_shr:1 row_mask:0xf bank_mask:0xf bound_ctrl:0
 ; GFX1064-NEXT:    v_xor_b32_dpp v2, v2, v2 row_shr:2 row_mask:0xf bank_mask:0xf bound_ctrl:0

 ; GFX1064-NEXT:    s_nop 1
 ; GFX1064-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX1064-NEXT:    buffer_store_dword v0, off, s[0:3], 0
 ; GFX1064-NEXT:    s_endpgm
 ;
 ; GFX1032-LABEL: xor_i32_varying:
 ; GFX1032:       ; %bb.0: ; %entry
 ; GFX1032-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
+; GFX1032-NEXT:    s_mov_b32 s2, exec_lo
 ; GFX1032-NEXT:    ; implicit-def: $vcc_hi
 ; GFX1032-NEXT:    v_mov_b32_e32 v2, v0
-; GFX1032-NEXT:    s_or_saveexec_b32 s2, -1
+; GFX1032-NEXT:    s_or_saveexec_b32 s3, -1
 ; GFX1032-NEXT:    v_mov_b32_e32 v1, 0
-; GFX1032-NEXT:    s_mov_b32 exec_lo, s2
-; GFX1032-NEXT:    v_cmp_ne_u32_e64 s2, 1, 0
+; GFX1032-NEXT:    s_mov_b32 exec_lo, s3
 ; GFX1032-NEXT:    v_mbcnt_lo_u32_b32_e64 v0, s2, 0
 ; GFX1032-NEXT:    s_not_b32 exec_lo, exec_lo
 ; GFX1032-NEXT:    v_mov_b32_e32 v2, 0
 ; GFX1032-NEXT:    s_not_b32 exec_lo, exec_lo
 ; GFX1032-NEXT:    s_or_saveexec_b32 s4, -1
 ; GFX1032-NEXT:    v_xor_b32_dpp v2, v2, v2 row_shr:1 row_mask:0xf bank_mask:0xf bound_ctrl:0
 ; GFX1032-NEXT:    s_mov_b32 s2, -1
 ; GFX1032-NEXT:    v_xor_b32_dpp v2, v2, v2 row_shr:2 row_mask:0xf bank_mask:0xf bound_ctrl:0

 ; GFX1032-NEXT:    s_endpgm
 entry:
   %lane = call i32 @llvm.amdgcn.workitem.id.x()
   %old = atomicrmw xor i32 addrspace(3)* @local_var32, i32 %lane acq_rel
   store i32 %old, i32 addrspace(1)* %out
   ret void
 }
 
-; GFX8MORE32: v_readlane_b32 s[[scalar_value:[0-9]+]], v{{[0-9]+}}, 31
-; GFX8MORE: v_mov_b32{{(_e[0-9]+)?}} v[[value:[0-9]+]], s[[scalar_value]]
-; GFX8MORE: ds_max_rtn_i32 v{{[0-9]+}}, v{{[0-9]+}}, v[[value]]
 define amdgpu_kernel void @max_i32_varying(i32 addrspace(1)* %out) {
 ;
 ;
 ; GFX7LESS-LABEL: max_i32_varying:
 ; GFX7LESS:       ; %bb.0: ; %entry
 ; GFX7LESS-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x9
 ; GFX7LESS-NEXT:    v_mov_b32_e32 v1, local_var32@abs32@lo
 ; GFX7LESS-NEXT:    s_mov_b32 m0, -1
 ; GFX7LESS-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX7LESS-NEXT:    ds_max_rtn_i32 v0, v1, v0
 ; GFX7LESS-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX7LESS-NEXT:    buffer_wbinvl1
 ; GFX7LESS-NEXT:    s_mov_b32 s3, 0xf000
 ; GFX7LESS-NEXT:    s_mov_b32 s2, -1
 ; GFX7LESS-NEXT:    buffer_store_dword v0, off, s[0:3], 0
 ; GFX7LESS-NEXT:    s_endpgm
 ;
 ; GFX8-LABEL: max_i32_varying:
 ; GFX8:       ; %bb.0: ; %entry
 ; GFX8-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
-; GFX8-NEXT:    v_cmp_ne_u32_e64 s[2:3], 1, 0
-; GFX8-NEXT:    v_mbcnt_lo_u32_b32 v3, s2, 0
-; GFX8-NEXT:    v_mbcnt_hi_u32_b32 v3, s3, v3
+; GFX8-NEXT:    v_mbcnt_lo_u32_b32 v3, exec_lo, 0
+; GFX8-NEXT:    v_mbcnt_hi_u32_b32 v3, exec_hi, v3
 ; GFX8-NEXT:    v_mov_b32_e32 v2, v0
 ; GFX8-NEXT:    s_or_saveexec_b64 s[2:3], -1
 ; GFX8-NEXT:    v_bfrev_b32_e32 v1, 1
 ; GFX8-NEXT:    s_mov_b64 exec, s[2:3]
 ; GFX8-NEXT:    s_not_b64 exec, exec
 ; GFX8-NEXT:    v_mov_b32_e32 v2, v1
 ; GFX8-NEXT:    s_not_b64 exec, exec
 ; GFX8-NEXT:    s_or_saveexec_b64 s[4:5], -1

 ; GFX8-NEXT:    s_nop 0
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX8-NEXT:    buffer_store_dword v0, off, s[0:3], 0
 ; GFX8-NEXT:    s_endpgm
 ;
 ; GFX9-LABEL: max_i32_varying:
 ; GFX9:       ; %bb.0: ; %entry
 ; GFX9-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
-; GFX9-NEXT:    v_cmp_ne_u32_e64 s[2:3], 1, 0
-; GFX9-NEXT:    v_mbcnt_lo_u32_b32 v3, s2, 0
-; GFX9-NEXT:    v_mbcnt_hi_u32_b32 v3, s3, v3
+; GFX9-NEXT:    v_mbcnt_lo_u32_b32 v3, exec_lo, 0
+; GFX9-NEXT:    v_mbcnt_hi_u32_b32 v3, exec_hi, v3
 ; GFX9-NEXT:    v_mov_b32_e32 v2, v0
 ; GFX9-NEXT:    s_or_saveexec_b64 s[2:3], -1
 ; GFX9-NEXT:    v_bfrev_b32_e32 v1, 1
 ; GFX9-NEXT:    s_mov_b64 exec, s[2:3]
 ; GFX9-NEXT:    s_not_b64 exec, exec
 ; GFX9-NEXT:    v_mov_b32_e32 v2, v1
 ; GFX9-NEXT:    s_not_b64 exec, exec
 ; GFX9-NEXT:    s_or_saveexec_b64 s[4:5], -1

 ; GFX9-NEXT:    s_nop 0
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:    buffer_store_dword v0, off, s[0:3], 0
 ; GFX9-NEXT:    s_endpgm
 ;
 ; GFX1064-LABEL: max_i32_varying:
 ; GFX1064:       ; %bb.0: ; %entry
 ; GFX1064-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
-; GFX1064-NEXT:    v_cmp_ne_u32_e64 s[2:3], 1, 0
+; GFX1064-NEXT:    v_mbcnt_lo_u32_b32_e64 v4, exec_lo, 0
 ; GFX1064-NEXT:    v_mov_b32_e32 v2, v0
-; GFX1064-NEXT:    v_mbcnt_lo_u32_b32_e64 v4, s2, 0
-; GFX1064-NEXT:    v_mbcnt_hi_u32_b32_e64 v4, s3, v4
+; GFX1064-NEXT:    v_mbcnt_hi_u32_b32_e64 v4, exec_hi, v4
 ; GFX1064-NEXT:    s_or_saveexec_b64 s[2:3], -1
 ; GFX1064-NEXT:    v_bfrev_b32_e32 v1, 1
 ; GFX1064-NEXT:    s_mov_b64 exec, s[2:3]
 ; GFX1064-NEXT:    s_not_b64 exec, exec
 ; GFX1064-NEXT:    v_mov_b32_e32 v2, v1
 ; GFX1064-NEXT:    s_not_b64 exec, exec
 ; GFX1064-NEXT:    s_or_saveexec_b64 s[4:5], -1
 ; GFX1064-NEXT:    v_max_i32_dpp v2, v2, v2 row_shr:1 row_mask:0xf bank_mask:0xf

 ; GFX1064-NEXT:    s_nop 1
 ; GFX1064-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX1064-NEXT:    buffer_store_dword v0, off, s[0:3], 0
 ; GFX1064-NEXT:    s_endpgm
 ;
 ; GFX1032-LABEL: max_i32_varying:
 ; GFX1032:       ; %bb.0: ; %entry
 ; GFX1032-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
-; GFX1032-NEXT:    v_cmp_ne_u32_e64 s2, 1, 0
+; GFX1032-NEXT:    v_mbcnt_lo_u32_b32_e64 v4, exec_lo, 0
 ; GFX1032-NEXT:    ; implicit-def: $vcc_hi
 ; GFX1032-NEXT:    v_mov_b32_e32 v2, v0
-; GFX1032-NEXT:    v_mbcnt_lo_u32_b32_e64 v4, s2, 0
 ; GFX1032-NEXT:    s_or_saveexec_b32 s2, -1
 ; GFX1032-NEXT:    v_bfrev_b32_e32 v1, 1
 ; GFX1032-NEXT:    s_mov_b32 exec_lo, s2
 ; GFX1032-NEXT:    s_not_b32 exec_lo, exec_lo
 ; GFX1032-NEXT:    v_mov_b32_e32 v2, v1
 ; GFX1032-NEXT:    s_not_b32 exec_lo, exec_lo
 ; GFX1032-NEXT:    s_or_saveexec_b32 s4, -1
 ; GFX1032-NEXT:    v_max_i32_dpp v2, v2, v2 row_shr:1 row_mask:0xf bank_mask:0xf

 }
 
 define amdgpu_kernel void @max_i64_constant(i64 addrspace(1)* %out) {
 ;
 ;
 ; GFX7LESS-LABEL: max_i64_constant:
 ; GFX7LESS:       ; %bb.0: ; %entry
 ; GFX7LESS-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x9
-; GFX7LESS-NEXT:    v_cmp_ne_u32_e64 s[2:3], 1, 0
-; GFX7LESS-NEXT:    v_mbcnt_lo_u32_b32_e64 v0, s2, 0
-; GFX7LESS-NEXT:    v_mbcnt_hi_u32_b32_e32 v0, s3, v0
+; GFX7LESS-NEXT:    v_mbcnt_lo_u32_b32_e64 v0, exec_lo, 0
+; GFX7LESS-NEXT:    v_mbcnt_hi_u32_b32_e32 v0, exec_hi, v0
 ; GFX7LESS-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX7LESS-NEXT:    ; implicit-def: $vgpr0_vgpr1
 ; GFX7LESS-NEXT:    s_and_saveexec_b64 s[2:3], vcc
 ; GFX7LESS-NEXT:    s_cbranch_execz BB18_2
 ; GFX7LESS-NEXT:  ; %bb.1:
 ; GFX7LESS-NEXT:    v_mov_b32_e32 v2, local_var64@abs32@lo
 ; GFX7LESS-NEXT:    v_mov_b32_e32 v0, 5
 ; GFX7LESS-NEXT:    v_mov_b32_e32 v1, 0

 ; GFX7LESS-NEXT:    s_mov_b32 s2, -1
 ; GFX7LESS-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX7LESS-NEXT:    buffer_store_dwordx2 v[0:1], off, s[0:3], 0
 ; GFX7LESS-NEXT:    s_endpgm
 ;
 ; GFX8-LABEL: max_i64_constant:
 ; GFX8:       ; %bb.0: ; %entry
 ; GFX8-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
-; GFX8-NEXT:    v_cmp_ne_u32_e64 s[2:3], 1, 0
-; GFX8-NEXT:    v_mbcnt_lo_u32_b32 v0, s2, 0
-; GFX8-NEXT:    v_mbcnt_hi_u32_b32 v0, s3, v0
+; GFX8-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
+; GFX8-NEXT:    v_mbcnt_hi_u32_b32 v0, exec_hi, v0
 ; GFX8-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX8-NEXT:    ; implicit-def: $vgpr0_vgpr1
 ; GFX8-NEXT:    s_and_saveexec_b64 s[2:3], vcc
 ; GFX8-NEXT:    s_cbranch_execz BB18_2
 ; GFX8-NEXT:  ; %bb.1:
 ; GFX8-NEXT:    v_mov_b32_e32 v0, 5
 ; GFX8-NEXT:    v_mov_b32_e32 v2, local_var64@abs32@lo
 ; GFX8-NEXT:    v_mov_b32_e32 v1, 0

 ; GFX8-NEXT:    s_mov_b32 s2, -1
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX8-NEXT:    buffer_store_dwordx2 v[0:1], off, s[0:3], 0
 ; GFX8-NEXT:    s_endpgm
 ;
 ; GFX9-LABEL: max_i64_constant:
 ; GFX9:       ; %bb.0: ; %entry
 ; GFX9-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
-; GFX9-NEXT:    v_cmp_ne_u32_e64 s[2:3], 1, 0
-; GFX9-NEXT:    v_mbcnt_lo_u32_b32 v0, s2, 0
-; GFX9-NEXT:    v_mbcnt_hi_u32_b32 v0, s3, v0
+; GFX9-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
+; GFX9-NEXT:    v_mbcnt_hi_u32_b32 v0, exec_hi, v0
 ; GFX9-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX9-NEXT:    ; implicit-def: $vgpr0_vgpr1
 ; GFX9-NEXT:    s_and_saveexec_b64 s[2:3], vcc
 ; GFX9-NEXT:    s_cbranch_execz BB18_2
 ; GFX9-NEXT:  ; %bb.1:
 ; GFX9-NEXT:    v_mov_b32_e32 v0, 5
 ; GFX9-NEXT:    v_mov_b32_e32 v2, local_var64@abs32@lo
 ; GFX9-NEXT:    v_mov_b32_e32 v1, 0

 ; GFX9-NEXT:    s_mov_b32 s3, 0xf000
 ; GFX9-NEXT:    s_mov_b32 s2, -1
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:    buffer_store_dwordx2 v[0:1], off, s[0:3], 0
 ; GFX9-NEXT:    s_endpgm
 ;
 ; GFX1064-LABEL: max_i64_constant:
 ; GFX1064:       ; %bb.0: ; %entry
-; GFX1064-NEXT:    v_cmp_ne_u32_e64 s[2:3], 1, 0
 ; GFX1064-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
-; GFX1064-NEXT:    v_mbcnt_lo_u32_b32_e64 v0, s2, 0
-; GFX1064-NEXT:    v_mbcnt_hi_u32_b32_e64 v0, s3, v0
+; GFX1064-NEXT:    v_mbcnt_lo_u32_b32_e64 v0, exec_lo, 0
+; GFX1064-NEXT:    v_mbcnt_hi_u32_b32_e64 v0, exec_hi, v0
 ; GFX1064-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX1064-NEXT:    ; implicit-def: $vgpr0_vgpr1
 ; GFX1064-NEXT:    s_and_saveexec_b64 s[2:3], vcc
 ; GFX1064-NEXT:    s_cbranch_execz BB18_2
 ; GFX1064-NEXT:  ; %bb.1:
 ; GFX1064-NEXT:    v_mov_b32_e32 v0, 5
 ; GFX1064-NEXT:    v_mov_b32_e32 v2, local_var64@abs32@lo
 ; GFX1064-NEXT:    v_mov_b32_e32 v1, 0

 ; GFX1064-NEXT:    v_cndmask_b32_e64 v0, v0, s4, vcc
 ; GFX1064-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX1064-NEXT:    buffer_store_dwordx2 v[0:1], off, s[0:3], 0
 ; GFX1064-NEXT:    s_endpgm
 ;
 ; GFX1032-LABEL: max_i64_constant:
 ; GFX1032:       ; %bb.0: ; %entry
 ; GFX1032-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
-; GFX1032-NEXT:    v_cmp_ne_u32_e64 s2, 1, 0
+; GFX1032-NEXT:    v_mbcnt_lo_u32_b32_e64 v0, exec_lo, 0
 ; GFX1032-NEXT:    ; implicit-def: $vcc_hi
-; GFX1032-NEXT:    v_mbcnt_lo_u32_b32_e64 v0, s2, 0
 ; GFX1032-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 0, v0
 ; GFX1032-NEXT:    ; implicit-def: $vgpr0_vgpr1
 ; GFX1032-NEXT:    s_and_saveexec_b32 s2, vcc_lo
 ; GFX1032-NEXT:    s_cbranch_execz BB18_2
 ; GFX1032-NEXT:  ; %bb.1:
 ; GFX1032-NEXT:    v_mov_b32_e32 v0, 5
 ; GFX1032-NEXT:    v_mov_b32_e32 v2, local_var64@abs32@lo
 ; GFX1032-NEXT:    v_mov_b32_e32 v1, 0

 ; GFX1032-NEXT:    buffer_store_dwordx2 v[0:1], off, s[0:3], 0
 ; GFX1032-NEXT:    s_endpgm
 entry:
   %old = atomicrmw max i64 addrspace(3)* @local_var64, i64 5 acq_rel
   store i64 %old, i64 addrspace(1)* %out
   ret void
 }
 
-; GFX8MORE32: v_readlane_b32 s[[scalar_value:[0-9]+]], v{{[0-9]+}}, 31
-; GFX8MORE: v_mov_b32{{(_e[0-9]+)?}} v[[value:[0-9]+]], s[[scalar_value]]
-; GFX8MORE: ds_min_rtn_i32 v{{[0-9]+}}, v{{[0-9]+}}, v[[value]]
 define amdgpu_kernel void @min_i32_varying(i32 addrspace(1)* %out) {
 ;
 ;
 ; GFX7LESS-LABEL: min_i32_varying:
 ; GFX7LESS:       ; %bb.0: ; %entry
 ; GFX7LESS-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x9
 ; GFX7LESS-NEXT:    v_mov_b32_e32 v1, local_var32@abs32@lo
 ; GFX7LESS-NEXT:    s_mov_b32 m0, -1
 ; GFX7LESS-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX7LESS-NEXT:    ds_min_rtn_i32 v0, v1, v0
 ; GFX7LESS-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX7LESS-NEXT:    buffer_wbinvl1
 ; GFX7LESS-NEXT:    s_mov_b32 s3, 0xf000
 ; GFX7LESS-NEXT:    s_mov_b32 s2, -1
 ; GFX7LESS-NEXT:    buffer_store_dword v0, off, s[0:3], 0
 ; GFX7LESS-NEXT:    s_endpgm
 ;
 ; GFX8-LABEL: min_i32_varying:
 ; GFX8:       ; %bb.0: ; %entry
 ; GFX8-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
-; GFX8-NEXT:    v_cmp_ne_u32_e64 s[2:3], 1, 0
-; GFX8-NEXT:    v_mbcnt_lo_u32_b32 v3, s2, 0
-; GFX8-NEXT:    v_mbcnt_hi_u32_b32 v3, s3, v3
+; GFX8-NEXT:    v_mbcnt_lo_u32_b32 v3, exec_lo, 0
+; GFX8-NEXT:    v_mbcnt_hi_u32_b32 v3, exec_hi, v3
 ; GFX8-NEXT:    v_mov_b32_e32 v2, v0
 ; GFX8-NEXT:    s_or_saveexec_b64 s[2:3], -1
 ; GFX8-NEXT:    v_bfrev_b32_e32 v1, -2
 ; GFX8-NEXT:    s_mov_b64 exec, s[2:3]
 ; GFX8-NEXT:    s_not_b64 exec, exec
 ; GFX8-NEXT:    v_mov_b32_e32 v2, v1
 ; GFX8-NEXT:    s_not_b64 exec, exec
 ; GFX8-NEXT:    s_or_saveexec_b64 s[4:5], -1

 ; GFX8-NEXT:    s_nop 0
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX8-NEXT:    buffer_store_dword v0, off, s[0:3], 0
 ; GFX8-NEXT:    s_endpgm
 ;
 ; GFX9-LABEL: min_i32_varying:
 ; GFX9:       ; %bb.0: ; %entry
 ; GFX9-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
-; GFX9-NEXT:    v_cmp_ne_u32_e64 s[2:3], 1, 0
-; GFX9-NEXT:    v_mbcnt_lo_u32_b32 v3, s2, 0
-; GFX9-NEXT:    v_mbcnt_hi_u32_b32 v3, s3, v3
+; GFX9-NEXT:    v_mbcnt_lo_u32_b32 v3, exec_lo, 0
+; GFX9-NEXT:    v_mbcnt_hi_u32_b32 v3, exec_hi, v3
 ; GFX9-NEXT:    v_mov_b32_e32 v2, v0
 ; GFX9-NEXT:    s_or_saveexec_b64 s[2:3], -1
 ; GFX9-NEXT:    v_bfrev_b32_e32 v1, -2
 ; GFX9-NEXT:    s_mov_b64 exec, s[2:3]
 ; GFX9-NEXT:    s_not_b64 exec, exec
 ; GFX9-NEXT:    v_mov_b32_e32 v2, v1
 ; GFX9-NEXT:    s_not_b64 exec, exec
 ; GFX9-NEXT:    s_or_saveexec_b64 s[4:5], -1

 ; GFX9-NEXT:    s_nop 0
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:    buffer_store_dword v0, off, s[0:3], 0
 ; GFX9-NEXT:    s_endpgm
 ;
 ; GFX1064-LABEL: min_i32_varying:
 ; GFX1064:       ; %bb.0: ; %entry
 ; GFX1064-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
-; GFX1064-NEXT:    v_cmp_ne_u32_e64 s[2:3], 1, 0
+; GFX1064-NEXT:    v_mbcnt_lo_u32_b32_e64 v4, exec_lo, 0
 ; GFX1064-NEXT:    v_mov_b32_e32 v2, v0
-; GFX1064-NEXT:    v_mbcnt_lo_u32_b32_e64 v4, s2, 0
-; GFX1064-NEXT:    v_mbcnt_hi_u32_b32_e64 v4, s3, v4
+; GFX1064-NEXT:    v_mbcnt_hi_u32_b32_e64 v4, exec_hi, v4
 ; GFX1064-NEXT:    s_or_saveexec_b64 s[2:3], -1
 ; GFX1064-NEXT:    v_bfrev_b32_e32 v1, -2
 ; GFX1064-NEXT:    s_mov_b64 exec, s[2:3]
 ; GFX1064-NEXT:    s_not_b64 exec, exec
 ; GFX1064-NEXT:    v_mov_b32_e32 v2, v1
 ; GFX1064-NEXT:    s_not_b64 exec, exec
 ; GFX1064-NEXT:    s_or_saveexec_b64 s[4:5], -1
 ; GFX1064-NEXT:    v_min_i32_dpp v2, v2, v2 row_shr:1 row_mask:0xf bank_mask:0xf

 ; GFX1064-NEXT:    s_nop 1
 ; GFX1064-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX1064-NEXT:    buffer_store_dword v0, off, s[0:3], 0
 ; GFX1064-NEXT:    s_endpgm
 ;
 ; GFX1032-LABEL: min_i32_varying:
 ; GFX1032:       ; %bb.0: ; %entry
 ; GFX1032-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
-; GFX1032-NEXT:    v_cmp_ne_u32_e64 s2, 1, 0
+; GFX1032-NEXT:    v_mbcnt_lo_u32_b32_e64 v4, exec_lo, 0
 ; GFX1032-NEXT:    ; implicit-def: $vcc_hi
 ; GFX1032-NEXT:    v_mov_b32_e32 v2, v0
-; GFX1032-NEXT:    v_mbcnt_lo_u32_b32_e64 v4, s2, 0
 ; GFX1032-NEXT:    s_or_saveexec_b32 s2, -1
 ; GFX1032-NEXT:    v_bfrev_b32_e32 v1, -2
 ; GFX1032-NEXT:    s_mov_b32 exec_lo, s2
 ; GFX1032-NEXT:    s_not_b32 exec_lo, exec_lo
 ; GFX1032-NEXT:    v_mov_b32_e32 v2, v1
 ; GFX1032-NEXT:    s_not_b32 exec_lo, exec_lo
 ; GFX1032-NEXT:    s_or_saveexec_b32 s4, -1
 ; GFX1032-NEXT:    v_min_i32_dpp v2, v2, v2 row_shr:1 row_mask:0xf bank_mask:0xf

 }
 
 define amdgpu_kernel void @min_i64_constant(i64 addrspace(1)* %out) {
 ;
 ;
 ; GFX7LESS-LABEL: min_i64_constant:
 ; GFX7LESS:       ; %bb.0: ; %entry
 ; GFX7LESS-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x9
-; GFX7LESS-NEXT:    v_cmp_ne_u32_e64 s[2:3], 1, 0
-; GFX7LESS-NEXT:    v_mbcnt_lo_u32_b32_e64 v0, s2, 0
-; GFX7LESS-NEXT:    v_mbcnt_hi_u32_b32_e32 v0, s3, v0
+; GFX7LESS-NEXT:    v_mbcnt_lo_u32_b32_e64 v0, exec_lo, 0
+; GFX7LESS-NEXT:    v_mbcnt_hi_u32_b32_e32 v0, exec_hi, v0
 ; GFX7LESS-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX7LESS-NEXT:    ; implicit-def: $vgpr0_vgpr1
 ; GFX7LESS-NEXT:    s_and_saveexec_b64 s[2:3], vcc
 ; GFX7LESS-NEXT:    s_cbranch_execz BB20_2
 ; GFX7LESS-NEXT:  ; %bb.1:
 ; GFX7LESS-NEXT:    v_mov_b32_e32 v2, local_var64@abs32@lo
 ; GFX7LESS-NEXT:    v_mov_b32_e32 v0, 5
 ; GFX7LESS-NEXT:    v_mov_b32_e32 v1, 0

 ; GFX7LESS-NEXT:    s_mov_b32 s3, 0xf000
 ; GFX7LESS-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX7LESS-NEXT:    buffer_store_dwordx2 v[0:1], off, s[0:3], 0
 ; GFX7LESS-NEXT:    s_endpgm
 ;
 ; GFX8-LABEL: min_i64_constant:
 ; GFX8:       ; %bb.0: ; %entry
 ; GFX8-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
-; GFX8-NEXT:    v_cmp_ne_u32_e64 s[2:3], 1, 0
-; GFX8-NEXT:    v_mbcnt_lo_u32_b32 v0, s2, 0
-; GFX8-NEXT:    v_mbcnt_hi_u32_b32 v0, s3, v0
+; GFX8-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
+; GFX8-NEXT:    v_mbcnt_hi_u32_b32 v0, exec_hi, v0
 ; GFX8-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX8-NEXT:    ; implicit-def: $vgpr0_vgpr1
 ; GFX8-NEXT:    s_and_saveexec_b64 s[2:3], vcc
 ; GFX8-NEXT:    s_cbranch_execz BB20_2
 ; GFX8-NEXT:  ; %bb.1:
 ; GFX8-NEXT:    v_mov_b32_e32 v0, 5
 ; GFX8-NEXT:    v_mov_b32_e32 v2, local_var64@abs32@lo
 ; GFX8-NEXT:    v_mov_b32_e32 v1, 0

 ; GFX8-NEXT:    s_mov_b32 s3, 0xf000
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX8-NEXT:    buffer_store_dwordx2 v[0:1], off, s[0:3], 0
 ; GFX8-NEXT:    s_endpgm
 ;
 ; GFX9-LABEL: min_i64_constant:
 ; GFX9:       ; %bb.0: ; %entry
 ; GFX9-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
-; GFX9-NEXT:    v_cmp_ne_u32_e64 s[2:3], 1, 0
-; GFX9-NEXT:    v_mbcnt_lo_u32_b32 v0, s2, 0
-; GFX9-NEXT:    v_mbcnt_hi_u32_b32 v0, s3, v0
+; GFX9-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
+; GFX9-NEXT:    v_mbcnt_hi_u32_b32 v0, exec_hi, v0
 ; GFX9-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX9-NEXT:    ; implicit-def: $vgpr0_vgpr1
 ; GFX9-NEXT:    s_and_saveexec_b64 s[2:3], vcc
 ; GFX9-NEXT:    s_cbranch_execz BB20_2
 ; GFX9-NEXT:  ; %bb.1:
 ; GFX9-NEXT:    v_mov_b32_e32 v0, 5
 ; GFX9-NEXT:    v_mov_b32_e32 v2, local_var64@abs32@lo
 ; GFX9-NEXT:    v_mov_b32_e32 v1, 0

 ; GFX9-NEXT:    v_cndmask_b32_e32 v0, v0, v2, vcc
 ; GFX9-NEXT:    s_mov_b32 s3, 0xf000
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:    buffer_store_dwordx2 v[0:1], off, s[0:3], 0
 ; GFX9-NEXT:    s_endpgm
 ;
 ; GFX1064-LABEL: min_i64_constant:
 ; GFX1064:       ; %bb.0: ; %entry
-; GFX1064-NEXT:    v_cmp_ne_u32_e64 s[2:3], 1, 0
 ; GFX1064-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
-; GFX1064-NEXT:    v_mbcnt_lo_u32_b32_e64 v0, s2, 0
-; GFX1064-NEXT:    v_mbcnt_hi_u32_b32_e64 v0, s3, v0
+; GFX1064-NEXT:    v_mbcnt_lo_u32_b32_e64 v0, exec_lo, 0
+; GFX1064-NEXT:    v_mbcnt_hi_u32_b32_e64 v0, exec_hi, v0
 ; GFX1064-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX1064-NEXT:    ; implicit-def: $vgpr0_vgpr1
 ; GFX1064-NEXT:    s_and_saveexec_b64 s[2:3], vcc
 ; GFX1064-NEXT:    s_cbranch_execz BB20_2
 ; GFX1064-NEXT:  ; %bb.1:
 ; GFX1064-NEXT:    v_mov_b32_e32 v0, 5
 ; GFX1064-NEXT:    v_mov_b32_e32 v2, local_var64@abs32@lo
 ; GFX1064-NEXT:    v_mov_b32_e32 v1, 0

 ; GFX1064-NEXT:    v_cndmask_b32_e64 v0, v0, s4, vcc
 ; GFX1064-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX1064-NEXT:    buffer_store_dwordx2 v[0:1], off, s[0:3], 0
 ; GFX1064-NEXT:    s_endpgm
 ;
 ; GFX1032-LABEL: min_i64_constant:
 ; GFX1032:       ; %bb.0: ; %entry
 ; GFX1032-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
-; GFX1032-NEXT:    v_cmp_ne_u32_e64 s2, 1, 0
+; GFX1032-NEXT:    v_mbcnt_lo_u32_b32_e64 v0, exec_lo, 0
 ; GFX1032-NEXT:    ; implicit-def: $vcc_hi
-; GFX1032-NEXT:    v_mbcnt_lo_u32_b32_e64 v0, s2, 0
 ; GFX1032-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 0, v0
 ; GFX1032-NEXT:    ; implicit-def: $vgpr0_vgpr1
 ; GFX1032-NEXT:    s_and_saveexec_b32 s2, vcc_lo
 ; GFX1032-NEXT:    s_cbranch_execz BB20_2
 ; GFX1032-NEXT:  ; %bb.1:
 ; GFX1032-NEXT:    v_mov_b32_e32 v0, 5
 ; GFX1032-NEXT:    v_mov_b32_e32 v2, local_var64@abs32@lo
 ; GFX1032-NEXT:    v_mov_b32_e32 v1, 0

 ; GFX1032-NEXT:    buffer_store_dwordx2 v[0:1], off, s[0:3], 0
 ; GFX1032-NEXT:    s_endpgm
 entry:
   %old = atomicrmw min i64 addrspace(3)* @local_var64, i64 5 acq_rel
   store i64 %old, i64 addrspace(1)* %out
   ret void
 }
 
-; GFX8MORE32: v_readlane_b32 s[[scalar_value:[0-9]+]], v{{[0-9]+}}, 31
-; GFX8MORE: v_mov_b32{{(_e[0-9]+)?}} v[[value:[0-9]+]], s[[scalar_value]]
-; GFX8MORE: ds_max_rtn_u32 v{{[0-9]+}}, v{{[0-9]+}}, v[[value]]
 define amdgpu_kernel void @umax_i32_varying(i32 addrspace(1)* %out) {
 ;
 ;
 ; GFX7LESS-LABEL: umax_i32_varying:
 ; GFX7LESS:       ; %bb.0: ; %entry
 ; GFX7LESS-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x9
 ; GFX7LESS-NEXT:    v_mov_b32_e32 v1, local_var32@abs32@lo
 ; GFX7LESS-NEXT:    s_mov_b32 m0, -1
 ; GFX7LESS-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX7LESS-NEXT:    ds_max_rtn_u32 v0, v1, v0
 ; GFX7LESS-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX7LESS-NEXT:    buffer_wbinvl1
 ; GFX7LESS-NEXT:    s_mov_b32 s3, 0xf000
 ; GFX7LESS-NEXT:    s_mov_b32 s2, -1
 ; GFX7LESS-NEXT:    buffer_store_dword v0, off, s[0:3], 0
 ; GFX7LESS-NEXT:    s_endpgm
 ;
 ; GFX8-LABEL: umax_i32_varying:
 ; GFX8:       ; %bb.0: ; %entry
 ; GFX8-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
+; GFX8-NEXT:    s_mov_b64 s[2:3], exec
 ; GFX8-NEXT:    v_mov_b32_e32 v2, v0
-; GFX8-NEXT:    s_or_saveexec_b64 s[2:3], -1
+; GFX8-NEXT:    s_or_saveexec_b64 s[4:5], -1
 ; GFX8-NEXT:    v_mov_b32_e32 v1, 0
-; GFX8-NEXT:    s_mov_b64 exec, s[2:3]
-; GFX8-NEXT:    v_cmp_ne_u32_e64 s[2:3], 1, 0
+; GFX8-NEXT:    s_mov_b64 exec, s[4:5]
 ; GFX8-NEXT:    v_mbcnt_lo_u32_b32 v0, s2, 0
 ; GFX8-NEXT:    v_mbcnt_hi_u32_b32 v0, s3, v0
 ; GFX8-NEXT:    s_not_b64 exec, exec
 ; GFX8-NEXT:    v_mov_b32_e32 v2, 0
 ; GFX8-NEXT:    s_not_b64 exec, exec
 ; GFX8-NEXT:    s_or_saveexec_b64 s[4:5], -1
 ; GFX8-NEXT:    v_max_u32_dpp v2, v2, v2 row_shr:1 row_mask:0xf bank_mask:0xf bound_ctrl:0
 ; GFX8-NEXT:    s_nop 1

 ; GFX8-NEXT:    s_nop 0
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX8-NEXT:    buffer_store_dword v0, off, s[0:3], 0
 ; GFX8-NEXT:    s_endpgm
 ;
 ; GFX9-LABEL: umax_i32_varying:
 ; GFX9:       ; %bb.0: ; %entry
 ; GFX9-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
+; GFX9-NEXT:    s_mov_b64 s[2:3], exec
 ; GFX9-NEXT:    v_mov_b32_e32 v2, v0
-; GFX9-NEXT:    s_or_saveexec_b64 s[2:3], -1
+; GFX9-NEXT:    s_or_saveexec_b64 s[4:5], -1
 ; GFX9-NEXT:    v_mov_b32_e32 v1, 0
-; GFX9-NEXT:    s_mov_b64 exec, s[2:3]
-; GFX9-NEXT:    v_cmp_ne_u32_e64 s[2:3], 1, 0
+; GFX9-NEXT:    s_mov_b64 exec, s[4:5]
 ; GFX9-NEXT:    v_mbcnt_lo_u32_b32 v0, s2, 0
 ; GFX9-NEXT:    v_mbcnt_hi_u32_b32 v0, s3, v0
 ; GFX9-NEXT:    s_not_b64 exec, exec
 ; GFX9-NEXT:    v_mov_b32_e32 v2, 0
 ; GFX9-NEXT:    s_not_b64 exec, exec
 ; GFX9-NEXT:    s_or_saveexec_b64 s[4:5], -1
 ; GFX9-NEXT:    v_max_u32_dpp v2, v2, v2 row_shr:1 row_mask:0xf bank_mask:0xf bound_ctrl:0
 ; GFX9-NEXT:    s_nop 1

 ; GFX9-NEXT:    s_nop 0
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:    buffer_store_dword v0, off, s[0:3], 0
 ; GFX9-NEXT:    s_endpgm
 ;
 ; GFX1064-LABEL: umax_i32_varying:
 ; GFX1064:       ; %bb.0: ; %entry
 ; GFX1064-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
+; GFX1064-NEXT:    s_mov_b64 s[2:3], exec
 ; GFX1064-NEXT:    v_mov_b32_e32 v2, v0
-; GFX1064-NEXT:    s_or_saveexec_b64 s[2:3], -1
+; GFX1064-NEXT:    s_or_saveexec_b64 s[4:5], -1
 ; GFX1064-NEXT:    v_mov_b32_e32 v1, 0
-; GFX1064-NEXT:    s_mov_b64 exec, s[2:3]
-; GFX1064-NEXT:    v_cmp_ne_u32_e64 s[2:3], 1, 0
+; GFX1064-NEXT:    s_mov_b64 exec, s[4:5]
 ; GFX1064-NEXT:    v_mbcnt_lo_u32_b32_e64 v0, s2, 0
 ; GFX1064-NEXT:    v_mbcnt_hi_u32_b32_e64 v0, s3, v0
 ; GFX1064-NEXT:    s_not_b64 exec, exec
 ; GFX1064-NEXT:    v_mov_b32_e32 v2, 0
 ; GFX1064-NEXT:    s_not_b64 exec, exec
 ; GFX1064-NEXT:    s_or_saveexec_b64 s[4:5], -1
 ; GFX1064-NEXT:    v_max_u32_dpp v2, v2, v2 row_shr:1 row_mask:0xf bank_mask:0xf bound_ctrl:0
 ; GFX1064-NEXT:    v_max_u32_dpp v2, v2, v2 row_shr:2 row_mask:0xf bank_mask:0xf bound_ctrl:0

 ; GFX1064-NEXT:    s_nop 1
 ; GFX1064-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX1064-NEXT:    buffer_store_dword v0, off, s[0:3], 0
 ; GFX1064-NEXT:    s_endpgm
 ;
 ; GFX1032-LABEL: umax_i32_varying:
 ; GFX1032:       ; %bb.0: ; %entry
 ; GFX1032-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
+; GFX1032-NEXT:    s_mov_b32 s2, exec_lo
 ; GFX1032-NEXT:    ; implicit-def: $vcc_hi
 ; GFX1032-NEXT:    v_mov_b32_e32 v2, v0
-; GFX1032-NEXT:    s_or_saveexec_b32 s2, -1
+; GFX1032-NEXT:    s_or_saveexec_b32 s3, -1
 ; GFX1032-NEXT:    v_mov_b32_e32 v1, 0
-; GFX1032-NEXT:    s_mov_b32 exec_lo, s2
-; GFX1032-NEXT:    v_cmp_ne_u32_e64 s2, 1, 0
+; GFX1032-NEXT:    s_mov_b32 exec_lo, s3
 ; GFX1032-NEXT:    v_mbcnt_lo_u32_b32_e64 v0, s2, 0
 ; GFX1032-NEXT:    s_not_b32 exec_lo, exec_lo
 ; GFX1032-NEXT:    v_mov_b32_e32 v2, 0
 ; GFX1032-NEXT:    s_not_b32 exec_lo, exec_lo
 ; GFX1032-NEXT:    s_or_saveexec_b32 s4, -1
 ; GFX1032-NEXT:    v_max_u32_dpp v2, v2, v2 row_shr:1 row_mask:0xf bank_mask:0xf bound_ctrl:0
 ; GFX1032-NEXT:    s_mov_b32 s2, -1
 ; GFX1032-NEXT:    v_max_u32_dpp v2, v2, v2 row_shr:2 row_mask:0xf bank_mask:0xf bound_ctrl:0

 }
 
 define amdgpu_kernel void @umax_i64_constant(i64 addrspace(1)* %out) {
 ;
 ;
 ; GFX7LESS-LABEL: umax_i64_constant:
 ; GFX7LESS:       ; %bb.0: ; %entry
 ; GFX7LESS-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x9
-; GFX7LESS-NEXT:    v_cmp_ne_u32_e64 s[2:3], 1, 0
-; GFX7LESS-NEXT:    v_mbcnt_lo_u32_b32_e64 v0, s2, 0
-; GFX7LESS-NEXT:    v_mbcnt_hi_u32_b32_e32 v0, s3, v0
+; GFX7LESS-NEXT:    v_mbcnt_lo_u32_b32_e64 v0, exec_lo, 0
+; GFX7LESS-NEXT:    v_mbcnt_hi_u32_b32_e32 v0, exec_hi, v0
 ; GFX7LESS-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX7LESS-NEXT:    ; implicit-def: $vgpr0_vgpr1
 ; GFX7LESS-NEXT:    s_and_saveexec_b64 s[2:3], vcc
 ; GFX7LESS-NEXT:    s_cbranch_execz BB22_2
 ; GFX7LESS-NEXT:  ; %bb.1:
 ; GFX7LESS-NEXT:    v_mov_b32_e32 v2, local_var64@abs32@lo
 ; GFX7LESS-NEXT:    v_mov_b32_e32 v0, 5
 ; GFX7LESS-NEXT:    v_mov_b32_e32 v1, 0

 ; GFX7LESS-NEXT:    s_mov_b32 s2, -1
 ; GFX7LESS-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX7LESS-NEXT:    buffer_store_dwordx2 v[0:1], off, s[0:3], 0
 ; GFX7LESS-NEXT:    s_endpgm
 ;
 ; GFX8-LABEL: umax_i64_constant:
 ; GFX8:       ; %bb.0: ; %entry
 ; GFX8-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
-; GFX8-NEXT:    v_cmp_ne_u32_e64 s[2:3], 1, 0
-; GFX8-NEXT:    v_mbcnt_lo_u32_b32 v0, s2, 0
-; GFX8-NEXT:    v_mbcnt_hi_u32_b32 v0, s3, v0
+; GFX8-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
+; GFX8-NEXT:    v_mbcnt_hi_u32_b32 v0, exec_hi, v0
 ; GFX8-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX8-NEXT:    ; implicit-def: $vgpr0_vgpr1
 ; GFX8-NEXT:    s_and_saveexec_b64 s[2:3], vcc
 ; GFX8-NEXT:    s_cbranch_execz BB22_2
 ; GFX8-NEXT:  ; %bb.1:
 ; GFX8-NEXT:    v_mov_b32_e32 v0, 5
 ; GFX8-NEXT:    v_mov_b32_e32 v2, local_var64@abs32@lo
 ; GFX8-NEXT:    v_mov_b32_e32 v1, 0

 ; GFX8-NEXT:    s_mov_b32 s2, -1
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX8-NEXT:    buffer_store_dwordx2 v[0:1], off, s[0:3], 0
 ; GFX8-NEXT:    s_endpgm
 ;
 ; GFX9-LABEL: umax_i64_constant:
 ; GFX9:       ; %bb.0: ; %entry
 ; GFX9-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
-; GFX9-NEXT:    v_cmp_ne_u32_e64 s[2:3], 1, 0
-; GFX9-NEXT:    v_mbcnt_lo_u32_b32 v0, s2, 0
-; GFX9-NEXT:    v_mbcnt_hi_u32_b32 v0, s3, v0
+; GFX9-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
+; GFX9-NEXT:    v_mbcnt_hi_u32_b32 v0, exec_hi, v0
 ; GFX9-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX9-NEXT:    ; implicit-def: $vgpr0_vgpr1
 ; GFX9-NEXT:    s_and_saveexec_b64 s[2:3], vcc
 ; GFX9-NEXT:    s_cbranch_execz BB22_2
 ; GFX9-NEXT:  ; %bb.1:
 ; GFX9-NEXT:    v_mov_b32_e32 v0, 5
 ; GFX9-NEXT:    v_mov_b32_e32 v2, local_var64@abs32@lo
 ; GFX9-NEXT:    v_mov_b32_e32 v1, 0

 ; GFX9-NEXT:    s_mov_b32 s3, 0xf000
 ; GFX9-NEXT:    s_mov_b32 s2, -1
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:    buffer_store_dwordx2 v[0:1], off, s[0:3], 0
 ; GFX9-NEXT:    s_endpgm
 ;
 ; GFX1064-LABEL: umax_i64_constant:
 ; GFX1064:       ; %bb.0: ; %entry
-; GFX1064-NEXT:    v_cmp_ne_u32_e64 s[2:3], 1, 0
 ; GFX1064-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
-; GFX1064-NEXT:    v_mbcnt_lo_u32_b32_e64 v0, s2, 0
-; GFX1064-NEXT:    v_mbcnt_hi_u32_b32_e64 v0, s3, v0
+; GFX1064-NEXT:    v_mbcnt_lo_u32_b32_e64 v0, exec_lo, 0
+; GFX1064-NEXT:    v_mbcnt_hi_u32_b32_e64 v0, exec_hi, v0
 ; GFX1064-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX1064-NEXT:    ; implicit-def: $vgpr0_vgpr1
 ; GFX1064-NEXT:    s_and_saveexec_b64 s[2:3], vcc
 ; GFX1064-NEXT:    s_cbranch_execz BB22_2
 ; GFX1064-NEXT:  ; %bb.1:
 ; GFX1064-NEXT:    v_mov_b32_e32 v0, 5
 ; GFX1064-NEXT:    v_mov_b32_e32 v2, local_var64@abs32@lo
 ; GFX1064-NEXT:    v_mov_b32_e32 v1, 0

 ; GFX1064-NEXT:    v_cndmask_b32_e64 v1, 0, s5, vcc
 ; GFX1064-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX1064-NEXT:    buffer_store_dwordx2 v[0:1], off, s[0:3], 0
 ; GFX1064-NEXT:    s_endpgm
 ;
 ; GFX1032-LABEL: umax_i64_constant:
 ; GFX1032:       ; %bb.0: ; %entry
 ; GFX1032-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
-; GFX1032-NEXT:    v_cmp_ne_u32_e64 s2, 1, 0
+; GFX1032-NEXT:    v_mbcnt_lo_u32_b32_e64 v0, exec_lo, 0
 ; GFX1032-NEXT:    ; implicit-def: $vcc_hi
-; GFX1032-NEXT:    v_mbcnt_lo_u32_b32_e64 v0, s2, 0
 ; GFX1032-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 0, v0
 ; GFX1032-NEXT:    ; implicit-def: $vgpr0_vgpr1
 ; GFX1032-NEXT:    s_and_saveexec_b32 s2, vcc_lo
 ; GFX1032-NEXT:    s_cbranch_execz BB22_2
 ; GFX1032-NEXT:  ; %bb.1:
 ; GFX1032-NEXT:    v_mov_b32_e32 v0, 5
 ; GFX1032-NEXT:    v_mov_b32_e32 v2, local_var64@abs32@lo
 ; GFX1032-NEXT:    v_mov_b32_e32 v1, 0

 ; GFX1032-NEXT:    buffer_store_dwordx2 v[0:1], off, s[0:3], 0
 ; GFX1032-NEXT:    s_endpgm
 entry:
   %old = atomicrmw umax i64 addrspace(3)* @local_var64, i64 5 acq_rel
   store i64 %old, i64 addrspace(1)* %out
   ret void
 }
 
-; GFX8MORE32: v_readlane_b32 s[[scalar_value:[0-9]+]], v{{[0-9]+}}, 31
-; GFX8MORE: v_mov_b32{{(_e[0-9]+)?}} v[[value:[0-9]+]], s[[scalar_value]]
-; GFX8MORE: ds_min_rtn_u32 v{{[0-9]+}}, v{{[0-9]+}}, v[[value]]
 define amdgpu_kernel void @umin_i32_varying(i32 addrspace(1)* %out) {
 ;
 ;
 ; GFX7LESS-LABEL: umin_i32_varying:
 ; GFX7LESS:       ; %bb.0: ; %entry
 ; GFX7LESS-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x9
 ; GFX7LESS-NEXT:    v_mov_b32_e32 v1, local_var32@abs32@lo
 ; GFX7LESS-NEXT:    s_mov_b32 m0, -1
 ; GFX7LESS-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX7LESS-NEXT:    ds_min_rtn_u32 v0, v1, v0
 ; GFX7LESS-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
 ; GFX7LESS-NEXT:    buffer_wbinvl1
 ; GFX7LESS-NEXT:    s_mov_b32 s3, 0xf000
 ; GFX7LESS-NEXT:    s_mov_b32 s2, -1
 ; GFX7LESS-NEXT:    buffer_store_dword v0, off, s[0:3], 0
 ; GFX7LESS-NEXT:    s_endpgm
 ;
 ; GFX8-LABEL: umin_i32_varying:
 ; GFX8:       ; %bb.0: ; %entry
 ; GFX8-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
-; GFX8-NEXT:    v_cmp_ne_u32_e64 s[2:3], 1, 0
-; GFX8-NEXT:    v_mbcnt_lo_u32_b32 v3, s2, 0
-; GFX8-NEXT:    v_mbcnt_hi_u32_b32 v3, s3, v3
+; GFX8-NEXT:    v_mbcnt_lo_u32_b32 v3, exec_lo, 0
+; GFX8-NEXT:    v_mbcnt_hi_u32_b32 v3, exec_hi, v3
 ; GFX8-NEXT:    v_mov_b32_e32 v2, v0
 ; GFX8-NEXT:    s_or_saveexec_b64 s[2:3], -1
 ; GFX8-NEXT:    v_mov_b32_e32 v1, -1
 ; GFX8-NEXT:    s_mov_b64 exec, s[2:3]
 ; GFX8-NEXT:    s_not_b64 exec, exec
 ; GFX8-NEXT:    v_mov_b32_e32 v2, -1
 ; GFX8-NEXT:    s_not_b64 exec, exec
 ; GFX8-NEXT:    s_or_saveexec_b64 s[4:5], -1

 ; GFX8-NEXT:    s_nop 0
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX8-NEXT:    buffer_store_dword v0, off, s[0:3], 0
 ; GFX8-NEXT:    s_endpgm
 ;
 ; GFX9-LABEL: umin_i32_varying:
 ; GFX9:       ; %bb.0: ; %entry
 ; GFX9-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
-; GFX9-NEXT:    v_cmp_ne_u32_e64 s[2:3], 1, 0
-; GFX9-NEXT:    v_mbcnt_lo_u32_b32 v3, s2, 0
-; GFX9-NEXT:    v_mbcnt_hi_u32_b32 v3, s3, v3
+; GFX9-NEXT:    v_mbcnt_lo_u32_b32 v3, exec_lo, 0
+; GFX9-NEXT:    v_mbcnt_hi_u32_b32 v3, exec_hi, v3
 ; GFX9-NEXT:    v_mov_b32_e32 v2, v0
 ; GFX9-NEXT:    s_or_saveexec_b64 s[2:3], -1
 ; GFX9-NEXT:    v_mov_b32_e32 v1, -1
 ; GFX9-NEXT:    s_mov_b64 exec, s[2:3]
 ; GFX9-NEXT:    s_not_b64 exec, exec
 ; GFX9-NEXT:    v_mov_b32_e32 v2, -1
 ; GFX9-NEXT:    s_not_b64 exec, exec
 ; GFX9-NEXT:    s_or_saveexec_b64 s[4:5], -1

 ; GFX9-NEXT:    s_nop 0
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:    buffer_store_dword v0, off, s[0:3], 0
 ; GFX9-NEXT:    s_endpgm
 ;
 ; GFX1064-LABEL: umin_i32_varying:
 ; GFX1064:       ; %bb.0: ; %entry
 ; GFX1064-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
-; GFX1064-NEXT:    v_cmp_ne_u32_e64 s[2:3], 1, 0
+; GFX1064-NEXT:    v_mbcnt_lo_u32_b32_e64 v4, exec_lo, 0
 ; GFX1064-NEXT:    v_mov_b32_e32 v2, v0
-; GFX1064-NEXT:    v_mbcnt_lo_u32_b32_e64 v4, s2, 0
-; GFX1064-NEXT:    v_mbcnt_hi_u32_b32_e64 v4, s3, v4
+; GFX1064-NEXT:    v_mbcnt_hi_u32_b32_e64 v4, exec_hi, v4
 ; GFX1064-NEXT:    s_or_saveexec_b64 s[2:3], -1
 ; GFX1064-NEXT:    v_mov_b32_e32 v1, -1
 ; GFX1064-NEXT:    s_mov_b64 exec, s[2:3]
 ; GFX1064-NEXT:    s_not_b64 exec, exec
 ; GFX1064-NEXT:    v_mov_b32_e32 v2, -1
 ; GFX1064-NEXT:    s_not_b64 exec, exec
 ; GFX1064-NEXT:    s_or_saveexec_b64 s[4:5], -1
 ; GFX1064-NEXT:    v_min_u32_dpp v2, v2, v2 row_shr:1 row_mask:0xf bank_mask:0xf

 ; GFX1064-NEXT:    s_nop 1
 ; GFX1064-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX1064-NEXT:    buffer_store_dword v0, off, s[0:3], 0
 ; GFX1064-NEXT:    s_endpgm
 ;
 ; GFX1032-LABEL: umin_i32_varying:
 ; GFX1032:       ; %bb.0: ; %entry
 ; GFX1032-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
-; GFX1032-NEXT:    v_cmp_ne_u32_e64 s2, 1, 0
+; GFX1032-NEXT:    v_mbcnt_lo_u32_b32_e64 v4, exec_lo, 0
 ; GFX1032-NEXT:    ; implicit-def: $vcc_hi
 ; GFX1032-NEXT:    v_mov_b32_e32 v2, v0
-; GFX1032-NEXT:    v_mbcnt_lo_u32_b32_e64 v4, s2, 0
 ; GFX1032-NEXT:    s_or_saveexec_b32 s2, -1
 ; GFX1032-NEXT:    v_mov_b32_e32 v1, -1
 ; GFX1032-NEXT:    s_mov_b32 exec_lo, s2
 ; GFX1032-NEXT:    s_not_b32 exec_lo, exec_lo
 ; GFX1032-NEXT:    v_mov_b32_e32 v2, -1
 ; GFX1032-NEXT:    s_not_b32 exec_lo, exec_lo
 ; GFX1032-NEXT:    s_or_saveexec_b32 s4, -1
 ; GFX1032-NEXT:    v_min_u32_dpp v2, v2, v2 row_shr:1 row_mask:0xf bank_mask:0xf

 }
 
 define amdgpu_kernel void @umin_i64_constant(i64 addrspace(1)* %out) {
 ;
 ;
 ; GFX7LESS-LABEL: umin_i64_constant:
 ; GFX7LESS:       ; %bb.0: ; %entry
 ; GFX7LESS-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x9
-; GFX7LESS-NEXT:    v_cmp_ne_u32_e64 s[2:3], 1, 0
-; GFX7LESS-NEXT:    v_mbcnt_lo_u32_b32_e64 v0, s2, 0
-; GFX7LESS-NEXT:    v_mbcnt_hi_u32_b32_e32 v0, s3, v0
+; GFX7LESS-NEXT:    v_mbcnt_lo_u32_b32_e64 v0, exec_lo, 0
+; GFX7LESS-NEXT:    v_mbcnt_hi_u32_b32_e32 v0, exec_hi, v0
 ; GFX7LESS-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX7LESS-NEXT:    ; implicit-def: $vgpr0_vgpr1
 ; GFX7LESS-NEXT:    s_and_saveexec_b64 s[2:3], vcc
 ; GFX7LESS-NEXT:    s_cbranch_execz BB24_2
 ; GFX7LESS-NEXT:  ; %bb.1:
 ; GFX7LESS-NEXT:    v_mov_b32_e32 v2, local_var64@abs32@lo
 ; GFX7LESS-NEXT:    v_mov_b32_e32 v0, 5
 ; GFX7LESS-NEXT:    v_mov_b32_e32 v1, 0

 ; GFX7LESS-NEXT:    s_mov_b32 s3, 0xf000
 ; GFX7LESS-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX7LESS-NEXT:    buffer_store_dwordx2 v[0:1], off, s[0:3], 0
 ; GFX7LESS-NEXT:    s_endpgm
 ;
 ; GFX8-LABEL: umin_i64_constant:
 ; GFX8:       ; %bb.0: ; %entry
 ; GFX8-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
-; GFX8-NEXT:    v_cmp_ne_u32_e64 s[2:3], 1, 0
-; GFX8-NEXT:    v_mbcnt_lo_u32_b32 v0, s2, 0
-; GFX8-NEXT:    v_mbcnt_hi_u32_b32 v0, s3, v0
+; GFX8-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
+; GFX8-NEXT:    v_mbcnt_hi_u32_b32 v0, exec_hi, v0
 ; GFX8-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX8-NEXT:    ; implicit-def: $vgpr0_vgpr1
 ; GFX8-NEXT:    s_and_saveexec_b64 s[2:3], vcc
 ; GFX8-NEXT:    s_cbranch_execz BB24_2
 ; GFX8-NEXT:  ; %bb.1:
 ; GFX8-NEXT:    v_mov_b32_e32 v0, 5
 ; GFX8-NEXT:    v_mov_b32_e32 v2, local_var64@abs32@lo
 ; GFX8-NEXT:    v_mov_b32_e32 v1, 0

 ; GFX8-NEXT:    s_mov_b32 s3, 0xf000
 ; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX8-NEXT:    buffer_store_dwordx2 v[0:1], off, s[0:3], 0
 ; GFX8-NEXT:    s_endpgm
 ;
 ; GFX9-LABEL: umin_i64_constant:
 ; GFX9:       ; %bb.0: ; %entry
 ; GFX9-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
-; GFX9-NEXT:    v_cmp_ne_u32_e64 s[2:3], 1, 0
-; GFX9-NEXT:    v_mbcnt_lo_u32_b32 v0, s2, 0
-; GFX9-NEXT:    v_mbcnt_hi_u32_b32 v0, s3, v0
+; GFX9-NEXT:    v_mbcnt_lo_u32_b32 v0, exec_lo, 0
+; GFX9-NEXT:    v_mbcnt_hi_u32_b32 v0, exec_hi, v0
 ; GFX9-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX9-NEXT:    ; implicit-def: $vgpr0_vgpr1
 ; GFX9-NEXT:    s_and_saveexec_b64 s[2:3], vcc
 ; GFX9-NEXT:    s_cbranch_execz BB24_2
 ; GFX9-NEXT:  ; %bb.1:
 ; GFX9-NEXT:    v_mov_b32_e32 v0, 5
 ; GFX9-NEXT:    v_mov_b32_e32 v2, local_var64@abs32@lo
 ; GFX9-NEXT:    v_mov_b32_e32 v1, 0

 ; GFX9-NEXT:    v_cndmask_b32_e32 v0, v0, v2, vcc
 ; GFX9-NEXT:    s_mov_b32 s3, 0xf000
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:    buffer_store_dwordx2 v[0:1], off, s[0:3], 0
 ; GFX9-NEXT:    s_endpgm
 ;
 ; GFX1064-LABEL: umin_i64_constant:
 ; GFX1064:       ; %bb.0: ; %entry
-; GFX1064-NEXT:    v_cmp_ne_u32_e64 s[2:3], 1, 0
 ; GFX1064-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
-; GFX1064-NEXT:    v_mbcnt_lo_u32_b32_e64 v0, s2, 0
-; GFX1064-NEXT:    v_mbcnt_hi_u32_b32_e64 v0, s3, v0
+; GFX1064-NEXT:    v_mbcnt_lo_u32_b32_e64 v0, exec_lo, 0
+; GFX1064-NEXT:    v_mbcnt_hi_u32_b32_e64 v0, exec_hi, v0
 ; GFX1064-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX1064-NEXT:    ; implicit-def: $vgpr0_vgpr1
 ; GFX1064-NEXT:    s_and_saveexec_b64 s[2:3], vcc
 ; GFX1064-NEXT:    s_cbranch_execz BB24_2
 ; GFX1064-NEXT:  ; %bb.1:
 ; GFX1064-NEXT:    v_mov_b32_e32 v0, 5
 ; GFX1064-NEXT:    v_mov_b32_e32 v2, local_var64@abs32@lo
 ; GFX1064-NEXT:    v_mov_b32_e32 v1, 0

 ; GFX1064-NEXT:    v_cndmask_b32_e64 v0, v0, s4, vcc
 ; GFX1064-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX1064-NEXT:    buffer_store_dwordx2 v[0:1], off, s[0:3], 0
 ; GFX1064-NEXT:    s_endpgm
 ;
 ; GFX1032-LABEL: umin_i64_constant:
 ; GFX1032:       ; %bb.0: ; %entry
 ; GFX1032-NEXT:    s_load_dwordx2 s[0:1], s[0:1], 0x24
-; GFX1032-NEXT:    v_cmp_ne_u32_e64 s2, 1, 0
+; GFX1032-NEXT:    v_mbcnt_lo_u32_b32_e64 v0, exec_lo, 0
 ; GFX1032-NEXT:    ; implicit-def: $vcc_hi
-; GFX1032-NEXT:    v_mbcnt_lo_u32_b32_e64 v0, s2, 0
 ; GFX1032-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 0, v0
 ; GFX1032-NEXT:    ; implicit-def: $vgpr0_vgpr1
 ; GFX1032-NEXT:    s_and_saveexec_b32 s2, vcc_lo
 ; GFX1032-NEXT:    s_cbranch_execz BB24_2
 ; GFX1032-NEXT:  ; %bb.1:
 ; GFX1032-NEXT:    v_mov_b32_e32 v0, 5
 ; GFX1032-NEXT:    v_mov_b32_e32 v2, local_var64@abs32@lo
 ; GFX1032-NEXT:    v_mov_b32_e32 v1, 0

llvm/test/CodeGen/AMDGPU/atomic_optimizations_pixelshader.ll

 define amdgpu_ps void @add_i32_constant(<4 x i32> inreg %out, <4 x i32> inreg %inout) {
 ; GFX7-LABEL: add_i32_constant:
 ; GFX7:       ; %bb.0: ; %entry
 ; GFX7-NEXT:    s_mov_b64 s[10:11], exec
 ; GFX7-NEXT:    ; implicit-def: $vgpr0
 ; GFX7-NEXT:    s_and_saveexec_b64 s[8:9], s[10:11]
 ; GFX7-NEXT:    s_cbranch_execz BB0_4
 ; GFX7-NEXT:  ; %bb.1:
-; GFX7-NEXT:    v_cmp_ne_u32_e64 s[12:13], 1, 0
+; GFX7-NEXT:    s_mov_b64 s[12:13], exec
 ; GFX7-NEXT:    v_mbcnt_lo_u32_b32_e64 v0, s12, 0
 ; GFX7-NEXT:    v_mbcnt_hi_u32_b32_e32 v0, s13, v0
 ; GFX7-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX7-NEXT:    ; implicit-def: $vgpr1
 ; GFX7-NEXT:    s_and_saveexec_b64 s[10:11], vcc
 ; GFX7-NEXT:    s_cbranch_execz BB0_3
 ; GFX7-NEXT:  ; %bb.2:
 ; GFX7-NEXT:    s_bcnt1_i32_b64 s12, s[12:13]

 ;
 ; GFX8-LABEL: add_i32_constant:
 ; GFX8:       ; %bb.0: ; %entry
 ; GFX8-NEXT:    s_mov_b64 s[10:11], exec
 ; GFX8-NEXT:    ; implicit-def: $vgpr0
 ; GFX8-NEXT:    s_and_saveexec_b64 s[8:9], s[10:11]
 ; GFX8-NEXT:    s_cbranch_execz BB0_4
 ; GFX8-NEXT:  ; %bb.1:
-; GFX8-NEXT:    v_cmp_ne_u32_e64 s[12:13], 1, 0
+; GFX8-NEXT:    s_mov_b64 s[12:13], exec
 ; GFX8-NEXT:    v_mbcnt_lo_u32_b32 v0, s12, 0
 ; GFX8-NEXT:    v_mbcnt_hi_u32_b32 v0, s13, v0
 ; GFX8-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX8-NEXT:    ; implicit-def: $vgpr1
 ; GFX8-NEXT:    s_and_saveexec_b64 s[10:11], vcc
 ; GFX8-NEXT:    s_cbranch_execz BB0_3
 ; GFX8-NEXT:  ; %bb.2:
 ; GFX8-NEXT:    s_bcnt1_i32_b64 s12, s[12:13]

 ;
 ; GFX9-LABEL: add_i32_constant:
 ; GFX9:       ; %bb.0: ; %entry
 ; GFX9-NEXT:    s_mov_b64 s[10:11], exec
 ; GFX9-NEXT:    ; implicit-def: $vgpr0
 ; GFX9-NEXT:    s_and_saveexec_b64 s[8:9], s[10:11]
 ; GFX9-NEXT:    s_cbranch_execz BB0_4
 ; GFX9-NEXT:  ; %bb.1:
-; GFX9-NEXT:    v_cmp_ne_u32_e64 s[12:13], 1, 0
+; GFX9-NEXT:    s_mov_b64 s[12:13], exec
 ; GFX9-NEXT:    v_mbcnt_lo_u32_b32 v0, s12, 0
 ; GFX9-NEXT:    v_mbcnt_hi_u32_b32 v0, s13, v0
 ; GFX9-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX9-NEXT:    ; implicit-def: $vgpr1
 ; GFX9-NEXT:    s_and_saveexec_b64 s[10:11], vcc
 ; GFX9-NEXT:    s_cbranch_execz BB0_3
 ; GFX9-NEXT:  ; %bb.2:
 ; GFX9-NEXT:    s_bcnt1_i32_b64 s12, s[12:13]

 ;
 ; GFX1064-LABEL: add_i32_constant:
 ; GFX1064:       ; %bb.0: ; %entry
 ; GFX1064-NEXT:    s_mov_b64 s[10:11], exec
 ; GFX1064-NEXT:    ; implicit-def: $vgpr0
 ; GFX1064-NEXT:    s_and_saveexec_b64 s[8:9], s[10:11]
 ; GFX1064-NEXT:    s_cbranch_execz BB0_4
 ; GFX1064-NEXT:  ; %bb.1:
-; GFX1064-NEXT:    v_cmp_ne_u32_e64 s[12:13], 1, 0
+; GFX1064-NEXT:    s_mov_b64 s[12:13], exec
 ; GFX1064-NEXT:    ; implicit-def: $vgpr1
 ; GFX1064-NEXT:    v_mbcnt_lo_u32_b32_e64 v0, s12, 0
 ; GFX1064-NEXT:    v_mbcnt_hi_u32_b32_e64 v0, s13, v0
 ; GFX1064-NEXT:    v_cmp_eq_u32_e32 vcc, 0, v0
 ; GFX1064-NEXT:    s_and_saveexec_b64 s[30:31], vcc
 ; GFX1064-NEXT:    s_cbranch_execz BB0_3
 ; GFX1064-NEXT:  ; %bb.2:
 ; GFX1064-NEXT:    s_bcnt1_i32_b64 s12, s[12:13]

 ; GFX1032-LABEL: add_i32_constant:
 ; GFX1032:       ; %bb.0: ; %entry
 ; GFX1032-NEXT:    s_mov_b32 s9, exec_lo
 ; GFX1032-NEXT:    ; implicit-def: $vgpr0
 ; GFX1032-NEXT:    ; implicit-def: $vcc_hi
 ; GFX1032-NEXT:    s_and_saveexec_b32 s8, s9
 ; GFX1032-NEXT:    s_cbranch_execz BB0_4
 ; GFX1032-NEXT:  ; %bb.1:
-; GFX1032-NEXT:    v_cmp_ne_u32_e64 s10, 1, 0
+; GFX1032-NEXT:    s_mov_b32 s10, exec_lo
 ; GFX1032-NEXT:    ; implicit-def: $vgpr1
 ; GFX1032-NEXT:    v_mbcnt_lo_u32_b32_e64 v0, s10, 0
 ; GFX1032-NEXT:    v_cmp_eq_u32_e32 vcc_lo, 0, v0
 ; GFX1032-NEXT:    s_and_saveexec_b32 s9, vcc_lo
 ; GFX1032-NEXT:    s_cbranch_execz BB0_3
 ; GFX1032-NEXT:  ; %bb.2:
 ; GFX1032-NEXT:    s_bcnt1_i32_b32 s10, s10
 ; GFX1032-NEXT:    v_mul_u32_u24_e64 v1, s10, 5

 ; GFX8-LABEL: add_i32_varying:
 ; GFX8:       ; %bb.0: ; %entry
 ; GFX8-NEXT:    s_mov_b64 s[10:11], exec
 ; GFX8-NEXT:    ; implicit-def: $vgpr3
 ; GFX8-NEXT:    v_mov_b32_e32 v2, v0
 ; GFX8-NEXT:    s_and_saveexec_b64 s[8:9], s[10:11]
 ; GFX8-NEXT:    s_cbranch_execz BB1_4
 ; GFX8-NEXT:  ; %bb.1:
-; GFX8-NEXT:    s_or_saveexec_b64 s[10:11], -1
+; GFX8-NEXT:    s_mov_b64 s[10:11], exec
+; GFX8-NEXT:    s_or_saveexec_b64 s[12:13], -1
 ; GFX8-NEXT:    v_mov_b32_e32 v1, 0
-; GFX8-NEXT:    s_mov_b64 exec, s[10:11]
-; GFX8-NEXT:    v_cmp_ne_u32_e64 s[10:11], 1, 0
+; GFX8-NEXT:    s_mov_b64 exec, s[12:13]
 ; GFX8-NEXT:    v_mbcnt_lo_u32_b32 v0, s10, 0
 ; GFX8-NEXT:    v_mbcnt_hi_u32_b32 v0, s11, v0
 ; GFX8-NEXT:    s_not_b64 exec, exec
 ; GFX8-NEXT:    v_mov_b32_e32 v2, 0
 ; GFX8-NEXT:    s_not_b64 exec, exec
 ; GFX8-NEXT:    s_or_saveexec_b64 s[10:11], -1
 ; GFX8-NEXT:    s_nop 0
 ; GFX8-NEXT:    v_add_u32_dpp v2, vcc, v2, v2 row_shr:1 row_mask:0xf bank_mask:0xf bound_ctrl:0

 ; GFX9-LABEL: add_i32_varying:
 ; GFX9:       ; %bb.0: ; %entry
 ; GFX9-NEXT:    s_mov_b64 s[10:11], exec
 ; GFX9-NEXT:    ; implicit-def: $vgpr3
 ; GFX9-NEXT:    v_mov_b32_e32 v2, v0
 ; GFX9-NEXT:    s_and_saveexec_b64 s[8:9], s[10:11]
 ; GFX9-NEXT:    s_cbranch_execz BB1_4
 ; GFX9-NEXT:  ; %bb.1:
-; GFX9-NEXT:    s_or_saveexec_b64 s[10:11], -1
+; GFX9-NEXT:    s_mov_b64 s[10:11], exec
+; GFX9-NEXT:    s_or_saveexec_b64 s[12:13], -1
 ; GFX9-NEXT:    v_mov_b32_e32 v1, 0
-; GFX9-NEXT:    s_mov_b64 exec, s[10:11]
-; GFX9-NEXT:    v_cmp_ne_u32_e64 s[10:11], 1, 0
+; GFX9-NEXT:    s_mov_b64 exec, s[12:13]
 ; GFX9-NEXT:    v_mbcnt_lo_u32_b32 v0, s10, 0
 ; GFX9-NEXT:    v_mbcnt_hi_u32_b32 v0, s11, v0
 ; GFX9-NEXT:    s_not_b64 exec, exec
 ; GFX9-NEXT:    v_mov_b32_e32 v2, 0
 ; GFX9-NEXT:    s_not_b64 exec, exec
 ; GFX9-NEXT:    s_or_saveexec_b64 s[10:11], -1
 ; GFX9-NEXT:    s_nop 0
 ; GFX9-NEXT:    v_add_u32_dpp v2, v2, v2 row_shr:1 row_mask:0xf bank_mask:0xf bound_ctrl:0

 ; GFX1064-LABEL: add_i32_varying:
 ; GFX1064:       ; %bb.0: ; %entry
 ; GFX1064-NEXT:    s_mov_b64 s[10:11], exec
 ; GFX1064-NEXT:    ; implicit-def: $vgpr4
 ; GFX1064-NEXT:    v_mov_b32_e32 v2, v0
 ; GFX1064-NEXT:    s_and_saveexec_b64 s[8:9], s[10:11]
 ; GFX1064-NEXT:    s_cbranch_execz BB1_4
 ; GFX1064-NEXT:  ; %bb.1:
-; GFX1064-NEXT:    s_or_saveexec_b64 s[10:11], -1
+; GFX1064-NEXT:    s_mov_b64 s[10:11], exec
+; GFX1064-NEXT:    s_or_saveexec_b64 s[12:13], -1
 ; GFX1064-NEXT:    v_mov_b32_e32 v1, 0
-; GFX1064-NEXT:    s_mov_b64 exec, s[10:11]
-; GFX1064-NEXT:    v_cmp_ne_u32_e64 s[10:11], 1, 0
+; GFX1064-NEXT:    s_mov_b64 exec, s[12:13]
 ; GFX1064-NEXT:    v_mbcnt_lo_u32_b32_e64 v0, s10, 0
 ; GFX1064-NEXT:    v_mbcnt_hi_u32_b32_e64 v0, s11, v0
 ; GFX1064-NEXT:    s_not_b64 exec, exec
 ; GFX1064-NEXT:    v_mov_b32_e32 v2, 0
 ; GFX1064-NEXT:    s_not_b64 exec, exec
 ; GFX1064-NEXT:    s_or_saveexec_b64 s[10:11], -1
 ; GFX1064-NEXT:    v_add_nc_u32_dpp v2, v2, v2 row_shr:1 row_mask:0xf bank_mask:0xf bound_ctrl:0
 ; GFX1064-NEXT:    v_add_nc_u32_dpp v2, v2, v2 row_shr:2 row_mask:0xf bank_mask:0xf bound_ctrl:0

 ; GFX1032:       ; %bb.0: ; %entry
 ; GFX1032-NEXT:    s_mov_b32 s9, exec_lo
 ; GFX1032-NEXT:    ; implicit-def: $vgpr4
 ; GFX1032-NEXT:    ; implicit-def: $vcc_hi
 ; GFX1032-NEXT:    v_mov_b32_e32 v2, v0
 ; GFX1032-NEXT:    s_and_saveexec_b32 s8, s9
 ; GFX1032-NEXT:    s_cbranch_execz BB1_4
 ; GFX1032-NEXT:  ; %bb.1:
-; GFX1032-NEXT:    s_or_saveexec_b32 s9, -1
+; GFX1032-NEXT:    s_mov_b32 s9, exec_lo
+; GFX1032-NEXT:    s_or_saveexec_b32 s10, -1
 ; GFX1032-NEXT:    v_mov_b32_e32 v1, 0
-; GFX1032-NEXT:    s_mov_b32 exec_lo, s9
-; GFX1032-NEXT:    v_cmp_ne_u32_e64 s9, 1, 0
+; GFX1032-NEXT:    s_mov_b32 exec_lo, s10
 ; GFX1032-NEXT:    v_mbcnt_lo_u32_b32_e64 v0, s9, 0
 ; GFX1032-NEXT:    s_not_b32 exec_lo, exec_lo
 ; GFX1032-NEXT:    v_mov_b32_e32 v2, 0
 ; GFX1032-NEXT:    s_not_b32 exec_lo, exec_lo
 ; GFX1032-NEXT:    s_or_saveexec_b32 s9, -1
 ; GFX1032-NEXT:    v_add_nc_u32_dpp v2, v2, v2 row_shr:1 row_mask:0xf bank_mask:0xf bound_ctrl:0
 ; GFX1032-NEXT:    v_add_nc_u32_dpp v2, v2, v2 row_shr:2 row_mask:0xf bank_mask:0xf bound_ctrl:0
 ; GFX1032-NEXT:    v_add_nc_u32_dpp v2, v2, v2 row_shr:4 row_mask:0xf bank_mask:0xf bound_ctrl:0

llvm/test/CodeGen/AMDGPU/atomic_optimizations_raw_buffer.ll

 ; RUN: llc -march=amdgcn -mtriple=amdgcn---amdgiz -amdgpu-atomic-optimizations=true -verify-machineinstrs < %s | FileCheck -enable-var-scope -check-prefixes=GCN,GCN64,GFX7LESS %s
 ; RUN: llc -march=amdgcn -mtriple=amdgcn---amdgiz -mcpu=tonga -mattr=-flat-for-global -amdgpu-atomic-optimizations=true -verify-machineinstrs < %s | FileCheck -enable-var-scope -check-prefixes=GCN,GCN64,GFX8MORE,GFX8MORE64,DPPCOMB %s
 ; RUN: llc -march=amdgcn -mtriple=amdgcn---amdgiz -mcpu=gfx900 -mattr=-flat-for-global -amdgpu-atomic-optimizations=true -verify-machineinstrs < %s | FileCheck -enable-var-scope -check-prefixes=GCN,GCN64,GFX8MORE,GFX8MORE64,DPPCOMB %s
 ; RUN: llc -march=amdgcn -mtriple=amdgcn---amdgiz -mcpu=gfx1010 -mattr=-wavefrontsize32,+wavefrontsize64 -mattr=-flat-for-global -amdgpu-atomic-optimizations=true -verify-machineinstrs < %s | FileCheck -enable-var-scope -check-prefixes=GCN,GCN64,GFX8MORE,GFX8MORE64 %s
 ; RUN: llc -march=amdgcn -mtriple=amdgcn---amdgiz -mcpu=gfx1010 -mattr=+wavefrontsize32,-wavefrontsize64 -mattr=-flat-for-global -amdgpu-atomic-optimizations=true -verify-machineinstrs < %s | FileCheck -enable-var-scope -check-prefixes=GCN,GCN32,GFX8MORE,GFX8MORE32 %s
 
 declare i32 @llvm.amdgcn.workitem.id.x()
 declare i32 @llvm.amdgcn.raw.buffer.atomic.add(i32, <4 x i32>, i32, i32, i32)
 declare i32 @llvm.amdgcn.raw.buffer.atomic.sub(i32, <4 x i32>, i32, i32, i32)
 
-; Show that what the atomic optimization pass will do for raw buffers.
+; Show what the atomic optimization pass will do for raw buffers.
 
 ; GCN-LABEL: add_i32_constant:
-; GCN32: v_cmp_ne_u32_e64 s[[exec_lo:[0-9]+]], 1, 0
-; GCN64: v_cmp_ne_u32_e64 s{{\[}}[[exec_lo:[0-9]+]]:[[exec_hi:[0-9]+]]{{\]}}, 1, 0
+; GCN32: s_mov_b32 s[[exec_lo:[0-9]+]], exec_lo
+; GCN64: s_mov_b64 s{{\[}}[[exec_lo:[0-9]+]]:[[exec_hi:[0-9]+]]{{\]}}, exec
 ; GCN: v_mbcnt_lo_u32_b32{{(_e[0-9]+)?}} v[[mbcnt:[0-9]+]], s[[exec_lo]], 0
 ; GCN64: v_mbcnt_hi_u32_b32{{(_e[0-9]+)?}} v[[mbcnt]], s[[exec_hi]], v[[mbcnt]]
 ; GCN: v_cmp_eq_u32{{(_e[0-9]+)?}} vcc{{(_lo)?}}, 0, v[[mbcnt]]
+; GCN: s_and_saveexec_b{{32|64}} s[[exec:\[?[0-9:]+\]?]], vcc
 ; GCN32: s_bcnt1_i32_b32 s[[popcount:[0-9]+]], s[[exec_lo]]
 ; GCN64: s_bcnt1_i32_b64 s[[popcount:[0-9]+]], s{{\[}}[[exec_lo]]:[[exec_hi]]{{\]}}
 ; GCN: v_mul_u32_u24{{(_e[0-9]+)?}} v[[value:[0-9]+]], s[[popcount]], 5
 ; GCN: buffer_atomic_add v[[value]]
 define amdgpu_kernel void @add_i32_constant(i32 addrspace(1)* %out, <4 x i32> %inout) {
 entry:
   %old = call i32 @llvm.amdgcn.raw.buffer.atomic.add(i32 5, <4 x i32> %inout, i32 0, i32 0, i32 0)
   store i32 %old, i32 addrspace(1)* %out
   ret void
 }
 
 ; GCN-LABEL: add_i32_uniform:
-; GCN32: v_cmp_ne_u32_e64 s[[exec_lo:[0-9]+]], 1, 0
-; GCN64: v_cmp_ne_u32_e64 s{{\[}}[[exec_lo:[0-9]+]]:[[exec_hi:[0-9]+]]{{\]}}, 1, 0
+; GCN32: s_mov_b32 s[[exec_lo:[0-9]+]], exec_lo
+; GCN64: s_mov_b64 s{{\[}}[[exec_lo:[0-9]+]]:[[exec_hi:[0-9]+]]{{\]}}, exec
 ; GCN: v_mbcnt_lo_u32_b32{{(_e[0-9]+)?}} v[[mbcnt:[0-9]+]], s[[exec_lo]], 0
 ; GCN64: v_mbcnt_hi_u32_b32{{(_e[0-9]+)?}} v[[mbcnt]], s[[exec_hi]], v[[mbcnt]]
 ; GCN: v_cmp_eq_u32{{(_e[0-9]+)?}} vcc{{(_lo)?}}, 0, v[[mbcnt]]
+; GCN: s_and_saveexec_b{{32|64}} s[[exec:\[?[0-9:]+\]?]], vcc
 ; GCN32: s_bcnt1_i32_b32 s[[popcount:[0-9]+]], s[[exec_lo]]
 ; GCN64: s_bcnt1_i32_b64 s[[popcount:[0-9]+]], s{{\[}}[[exec_lo]]:[[exec_hi]]{{\]}}
 ; GCN: s_mul_i32 s[[scalar_value:[0-9]+]], s{{[0-9]+}}, s[[popcount]]
 ; GCN: v_mov_b32{{(_e[0-9]+)?}} v[[value:[0-9]+]], s[[scalar_value]]
 ; GCN: buffer_atomic_add v[[value]]
 define amdgpu_kernel void @add_i32_uniform(i32 addrspace(1)* %out, <4 x i32> %inout, i32 %additive) {
 entry:
   %old = call i32 @llvm.amdgcn.raw.buffer.atomic.add(i32 %additive, <4 x i32> %inout, i32 0, i32 0, i32 0)

 entry:
   %lane = call i32 @llvm.amdgcn.workitem.id.x()
   %old = call i32 @llvm.amdgcn.raw.buffer.atomic.add(i32 1, <4 x i32> %inout, i32 %lane, i32 0, i32 0)
   store i32 %old, i32 addrspace(1)* %out
   ret void
 }
 
 ; GCN-LABEL: sub_i32_constant:
-; GCN32: v_cmp_ne_u32_e64 s[[exec_lo:[0-9]+]], 1, 0
-; GCN64: v_cmp_ne_u32_e64 s{{\[}}[[exec_lo:[0-9]+]]:[[exec_hi:[0-9]+]]{{\]}}, 1, 0
+; GCN32: s_mov_b32 s[[exec_lo:[0-9]+]], exec_lo
+; GCN64: s_mov_b64 s{{\[}}[[exec_lo:[0-9]+]]:[[exec_hi:[0-9]+]]{{\]}}, exec
 ; GCN: v_mbcnt_lo_u32_b32{{(_e[0-9]+)?}} v[[mbcnt:[0-9]+]], s[[exec_lo]], 0
 ; GCN64: v_mbcnt_hi_u32_b32{{(_e[0-9]+)?}} v[[mbcnt]], s[[exec_hi]], v[[mbcnt]]
 ; GCN: v_cmp_eq_u32{{(_e[0-9]+)?}} vcc{{(_lo)?}}, 0, v[[mbcnt]]
+; GCN: s_and_saveexec_b{{32|64}} s[[exec:\[?[0-9:]+\]?]], vcc
 ; GCN32: s_bcnt1_i32_b32 s[[popcount:[0-9]+]], s[[exec_lo]]
 ; GCN64: s_bcnt1_i32_b64 s[[popcount:[0-9]+]], s{{\[}}[[exec_lo]]:[[exec_hi]]{{\]}}
 ; GCN: v_mul_u32_u24{{(_e[0-9]+)?}} v[[value:[0-9]+]], s[[popcount]], 5
 ; GCN: buffer_atomic_sub v[[value]]
 define amdgpu_kernel void @sub_i32_constant(i32 addrspace(1)* %out, <4 x i32> %inout) {
 entry:
   %old = call i32 @llvm.amdgcn.raw.buffer.atomic.sub(i32 5, <4 x i32> %inout, i32 0, i32 0, i32 0)
   store i32 %old, i32 addrspace(1)* %out
   ret void
 }
 
 ; GCN-LABEL: sub_i32_uniform:
-; GCN32: v_cmp_ne_u32_e64 s[[exec_lo:[0-9]+]], 1, 0
-; GCN64: v_cmp_ne_u32_e64 s{{\[}}[[exec_lo:[0-9]+]]:[[exec_hi:[0-9]+]]{{\]}}, 1, 0
+; GCN32: s_mov_b32 s[[exec_lo:[0-9]+]], exec_lo
+; GCN64: s_mov_b64 s{{\[}}[[exec_lo:[0-9]+]]:[[exec_hi:[0-9]+]]{{\]}}, exec
 ; GCN: v_mbcnt_lo_u32_b32{{(_e[0-9]+)?}} v[[mbcnt:[0-9]+]], s[[exec_lo]], 0
 ; GCN64: v_mbcnt_hi_u32_b32{{(_e[0-9]+)?}} v[[mbcnt]], s[[exec_hi]], v[[mbcnt]]
 ; GCN: v_cmp_eq_u32{{(_e[0-9]+)?}} vcc{{(_lo)?}}, 0, v[[mbcnt]]
+; GCN: s_and_saveexec_b{{32|64}} s[[exec:\[?[0-9:]+\]?]], vcc
 ; GCN32: s_bcnt1_i32_b32 s[[popcount:[0-9]+]], s[[exec_lo]]
 ; GCN64: s_bcnt1_i32_b64 s[[popcount:[0-9]+]], s{{\[}}[[exec_lo]]:[[exec_hi]]{{\]}}
 ; GCN: s_mul_i32 s[[scalar_value:[0-9]+]], s{{[0-9]+}}, s[[popcount]]
 ; GCN: v_mov_b32{{(_e[0-9]+)?}} v[[value:[0-9]+]], s[[scalar_value]]
 ; GCN: buffer_atomic_sub v[[value]]
 define amdgpu_kernel void @sub_i32_uniform(i32 addrspace(1)* %out, <4 x i32> %inout, i32 %subitive) {
 entry:
   %old = call i32 @llvm.amdgcn.raw.buffer.atomic.sub(i32 %subitive, <4 x i32> %inout, i32 0, i32 0, i32 0)

llvm/test/CodeGen/AMDGPU/atomic_optimizations_struct_buffer.ll

 ; RUN: llc -march=amdgcn -mtriple=amdgcn---amdgiz -amdgpu-atomic-optimizations=true -verify-machineinstrs < %s | FileCheck -enable-var-scope -check-prefixes=GCN,GCN64,GFX7LESS %s
 ; RUN: llc -march=amdgcn -mtriple=amdgcn---amdgiz -mcpu=tonga -mattr=-flat-for-global -amdgpu-atomic-optimizations=true -verify-machineinstrs < %s | FileCheck -enable-var-scope -check-prefixes=GCN,GCN64,GFX8MORE,GFX8MORE64,DPPCOMB %s
 ; RUN: llc -march=amdgcn -mtriple=amdgcn---amdgiz -mcpu=gfx900 -mattr=-flat-for-global -amdgpu-atomic-optimizations=true -verify-machineinstrs < %s | FileCheck -enable-var-scope -check-prefixes=GCN,GCN64,GFX8MORE,GFX8MORE64,DPPCOMB %s
 ; RUN: llc -march=amdgcn -mtriple=amdgcn---amdgiz -mcpu=gfx1010 -mattr=-wavefrontsize32,+wavefrontsize64 -mattr=-flat-for-global -amdgpu-atomic-optimizations=true -verify-machineinstrs < %s | FileCheck -enable-var-scope -check-prefixes=GCN,GCN64,GFX8MORE,GFX8MORE64 %s
 ; RUN: llc -march=amdgcn -mtriple=amdgcn---amdgiz -mcpu=gfx1010 -mattr=+wavefrontsize32,-wavefrontsize64 -mattr=-flat-for-global -amdgpu-atomic-optimizations=true -verify-machineinstrs < %s | FileCheck -enable-var-scope -check-prefixes=GCN,GCN32,GFX8MORE,GFX8MORE32 %s
 
 declare i32 @llvm.amdgcn.workitem.id.x()
 declare i32 @llvm.amdgcn.struct.buffer.atomic.add(i32, <4 x i32>, i32, i32, i32, i32)
 declare i32 @llvm.amdgcn.struct.buffer.atomic.sub(i32, <4 x i32>, i32, i32, i32, i32)
 
-; Show that what the atomic optimization pass will do for struct buffers.
+; Show what the atomic optimization pass will do for struct buffers.
 
 ; GCN-LABEL: add_i32_constant:
-; GCN32: v_cmp_ne_u32_e64 s[[exec_lo:[0-9]+]], 1, 0
-; GCN64: v_cmp_ne_u32_e64 s{{\[}}[[exec_lo:[0-9]+]]:[[exec_hi:[0-9]+]]{{\]}}, 1, 0
+; GCN32: s_mov_b32 s[[exec_lo:[0-9]+]], exec_lo
+; GCN64: s_mov_b64 s{{\[}}[[exec_lo:[0-9]+]]:[[exec_hi:[0-9]+]]{{\]}}, exec
 ; GCN: v_mbcnt_lo_u32_b32{{(_e[0-9]+)?}} v[[mbcnt:[0-9]+]], s[[exec_lo]], 0
 ; GCN64: v_mbcnt_hi_u32_b32{{(_e[0-9]+)?}} v[[mbcnt]], s[[exec_hi]], v[[mbcnt]]
 ; GCN: v_cmp_eq_u32{{(_e[0-9]+)?}} vcc{{(_lo)?}}, 0, v[[mbcnt]]
+; GCN: s_and_saveexec_b{{32|64}} s[[exec:\[?[0-9:]+\]?]], vcc
 ; GCN32: s_bcnt1_i32_b32 s[[popcount:[0-9]+]], s[[exec_lo]]
 ; GCN64: s_bcnt1_i32_b64 s[[popcount:[0-9]+]], s{{\[}}[[exec_lo]]:[[exec_hi]]{{\]}}
 ; GCN: v_mul_u32_u24{{(_e[0-9]+)?}} v[[value:[0-9]+]], s[[popcount]], 5
 ; GCN: buffer_atomic_add v[[value]]
 define amdgpu_kernel void @add_i32_constant(i32 addrspace(1)* %out, <4 x i32> %inout) {
 entry:
   %old = call i32 @llvm.amdgcn.struct.buffer.atomic.add(i32 5, <4 x i32> %inout, i32 0, i32 0, i32 0, i32 0)
   store i32 %old, i32 addrspace(1)* %out
   ret void
 }
 
 ; GCN-LABEL: add_i32_uniform:
-; GCN32: v_cmp_ne_u32_e64 s[[exec_lo:[0-9]+]], 1, 0
-; GCN64: v_cmp_ne_u32_e64 s{{\[}}[[exec_lo:[0-9]+]]:[[exec_hi:[0-9]+]]{{\]}}, 1, 0
+; GCN32: s_mov_b32 s[[exec_lo:[0-9]+]], exec_lo
+; GCN64: s_mov_b64 s{{\[}}[[exec_lo:[0-9]+]]:[[exec_hi:[0-9]+]]{{\]}}, exec
 ; GCN: v_mbcnt_lo_u32_b32{{(_e[0-9]+)?}} v[[mbcnt:[0-9]+]], s[[exec_lo]], 0
 ; GCN64: v_mbcnt_hi_u32_b32{{(_e[0-9]+)?}} v[[mbcnt]], s[[exec_hi]], v[[mbcnt]]
 ; GCN: v_cmp_eq_u32{{(_e[0-9]+)?}} vcc{{(_lo)?}}, 0, v[[mbcnt]]
+; GCN: s_and_saveexec_b{{32|64}} s[[exec:\[?[0-9:]+\]?]], vcc
 ; GCN32: s_bcnt1_i32_b32 s[[popcount:[0-9]+]], s[[exec_lo]]
 ; GCN64: s_bcnt1_i32_b64 s[[popcount:[0-9]+]], s{{\[}}[[exec_lo]]:[[exec_hi]]{{\]}}
 ; GCN: s_mul_i32 s[[scalar_value:[0-9]+]], s{{[0-9]+}}, s[[popcount]]
 ; GCN: v_mov_b32{{(_e[0-9]+)?}} v[[value:[0-9]+]], s[[scalar_value]]
 ; GCN: buffer_atomic_add v[[value]]
 define amdgpu_kernel void @add_i32_uniform(i32 addrspace(1)* %out, <4 x i32> %inout, i32 %additive) {
 entry:
   %old = call i32 @llvm.amdgcn.struct.buffer.atomic.add(i32 %additive, <4 x i32> %inout, i32 0, i32 0, i32 0, i32 0)

 entry:
   %lane = call i32 @llvm.amdgcn.workitem.id.x()
   %old = call i32 @llvm.amdgcn.struct.buffer.atomic.add(i32 1, <4 x i32> %inout, i32 0, i32 %lane, i32 0, i32 0)
   store i32 %old, i32 addrspace(1)* %out
   ret void
 }
 
 ; GCN-LABEL: sub_i32_constant:
-; GCN32: v_cmp_ne_u32_e64 s[[exec_lo:[0-9]+]], 1, 0
-; GCN64: v_cmp_ne_u32_e64 s{{\[}}[[exec_lo:[0-9]+]]:[[exec_hi:[0-9]+]]{{\]}}, 1, 0
+; GCN32: s_mov_b32 s[[exec_lo:[0-9]+]], exec_lo
+; GCN64: s_mov_b64 s{{\[}}[[exec_lo:[0-9]+]]:[[exec_hi:[0-9]+]]{{\]}}, exec
 ; GCN: v_mbcnt_lo_u32_b32{{(_e[0-9]+)?}} v[[mbcnt:[0-9]+]], s[[exec_lo]], 0
 ; GCN64: v_mbcnt_hi_u32_b32{{(_e[0-9]+)?}} v[[mbcnt]], s[[exec_hi]], v[[mbcnt]]
 ; GCN: v_cmp_eq_u32{{(_e[0-9]+)?}} vcc{{(_lo)?}}, 0, v[[mbcnt]]
+; GCN: s_and_saveexec_b{{32|64}} s[[exec:\[?[0-9:]+\]?]], vcc
 ; GCN32: s_bcnt1_i32_b32 s[[popcount:[0-9]+]], s[[exec_lo]]
 ; GCN64: s_bcnt1_i32_b64 s[[popcount:[0-9]+]], s{{\[}}[[exec_lo]]:[[exec_hi]]{{\]}}
 ; GCN: v_mul_u32_u24{{(_e[0-9]+)?}} v[[value:[0-9]+]], s[[popcount]], 5
 ; GCN: buffer_atomic_sub v[[value]]
 define amdgpu_kernel void @sub_i32_constant(i32 addrspace(1)* %out, <4 x i32> %inout) {
 entry:
   %old = call i32 @llvm.amdgcn.struct.buffer.atomic.sub(i32 5, <4 x i32> %inout, i32 0, i32 0, i32 0, i32 0)
   store i32 %old, i32 addrspace(1)* %out
   ret void
 }
 
 ; GCN-LABEL: sub_i32_uniform:
-; GCN32: v_cmp_ne_u32_e64 s[[exec_lo:[0-9]+]], 1, 0
-; GCN64: v_cmp_ne_u32_e64 s{{\[}}[[exec_lo:[0-9]+]]:[[exec_hi:[0-9]+]]{{\]}}, 1, 0
+; GCN32: s_mov_b32 s[[exec_lo:[0-9]+]], exec_lo
+; GCN64: s_mov_b64 s{{\[}}[[exec_lo:[0-9]+]]:[[exec_hi:[0-9]+]]{{\]}}, exec
 ; GCN: v_mbcnt_lo_u32_b32{{(_e[0-9]+)?}} v[[mbcnt:[0-9]+]], s[[exec_lo]], 0
 ; GCN64: v_mbcnt_hi_u32_b32{{(_e[0-9]+)?}} v[[mbcnt]], s[[exec_hi]], v[[mbcnt]]
 ; GCN: v_cmp_eq_u32{{(_e[0-9]+)?}} vcc{{(_lo)?}}, 0, v[[mbcnt]]
+; GCN: s_and_saveexec_b{{32|64}} s[[exec:\[?[0-9:]+\]?]], vcc
 ; GCN32: s_bcnt1_i32_b32 s[[popcount:[0-9]+]], s[[exec_lo]]
 ; GCN64: s_bcnt1_i32_b64 s[[popcount:[0-9]+]], s{{\[}}[[exec_lo]]:[[exec_hi]]{{\]}}
 ; GCN: s_mul_i32 s[[scalar_value:[0-9]+]], s{{[0-9]+}}, s[[popcount]]
 ; GCN: v_mov_b32{{(_e[0-9]+)?}} v[[value:[0-9]+]], s[[scalar_value]]
 ; GCN: buffer_atomic_sub v[[value]]
 define amdgpu_kernel void @sub_i32_uniform(i32 addrspace(1)* %out, <4 x i32> %inout, i32 %subitive) {
 entry:
   %old = call i32 @llvm.amdgcn.struct.buffer.atomic.sub(i32 %subitive, <4 x i32> %inout, i32 0, i32 0, i32 0, i32 0)

llvm/test/CodeGen/AMDGPU/llvm.amdgcn.ballot.i32.ll

+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc -march=amdgcn -mcpu=gfx1010 -mattr=+wavefrontsize32,-wavefrontsize64 < %s | FileCheck %s
+
+declare i32 @llvm.amdgcn.ballot.i32(i1)
+
+; Test ballot(0)
+
+define i32 @test0() {
+; CHECK-LABEL: test0:
+; CHECK:       ; %bb.0:
+; CHECK-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
+; CHECK-NEXT:    s_waitcnt_vscnt null, 0x0
+; CHECK-NEXT:    v_mov_b32_e32 v0, 0
+; CHECK-NEXT:    ; implicit-def: $vcc_hi
+; CHECK-NEXT:    s_setpc_b64 s[30:31]
+  %ballot = call i32 @llvm.amdgcn.ballot.i32(i1 0)
+  ret i32 %ballot
+}
+
+; Test ballot(1)
+
+define i32 @test1() {
+; CHECK-LABEL: test1:
+; CHECK:       ; %bb.0:
+; CHECK-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
+; CHECK-NEXT:    s_waitcnt_vscnt null, 0x0
+; CHECK-NEXT:    v_mov_b32_e32 v0, exec_lo
+; CHECK-NEXT:    ; implicit-def: $vcc_hi
+; CHECK-NEXT:    s_setpc_b64 s[30:31]
+  %ballot = call i32 @llvm.amdgcn.ballot.i32(i1 1)
+  ret i32 %ballot
+}
+
+; Test ballot of a non-comparison operation
+
+define i32 @test2(i32 %x) {
+; CHECK-LABEL: test2:
+; CHECK:       ; %bb.0:
+; CHECK-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
+; CHECK-NEXT:    s_waitcnt_vscnt null, 0x0
+; CHECK-NEXT:    v_and_b32_e32 v0, 1, v0
+; CHECK-NEXT:    ; implicit-def: $vcc_hi
+; CHECK-NEXT:    v_cmp_ne_u32_e64 s4, 0, v0
+; CHECK-NEXT:    v_mov_b32_e32 v0, s4
+; CHECK-NEXT:    s_setpc_b64 s[30:31]
+  %trunc = trunc i32 %x to i1
+  %ballot = call i32 @llvm.amdgcn.ballot.i32(i1 %trunc)
+  ret i32 %ballot
+}
+
+; Test ballot of comparisons
+
+define i32 @test3(i32 %x, i32 %y) {
+; CHECK-LABEL: test3:
+; CHECK:       ; %bb.0:
+; CHECK-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
+; CHECK-NEXT:    s_waitcnt_vscnt null, 0x0
+; CHECK-NEXT:    v_cmp_eq_u32_e64 s4, v0, v1
+; CHECK-NEXT:    ; implicit-def: $vcc_hi
+; CHECK-NEXT:    v_mov_b32_e32 v0, s4
+; CHECK-NEXT:    s_setpc_b64 s[30:31]
+  %cmp = icmp eq i32 %x, %y
+  %ballot = call i32 @llvm.amdgcn.ballot.i32(i1 %cmp)
+  ret i32 %ballot
+}
+
+define i32 @test4(i32 %x) {
+; CHECK-LABEL: test4:
+; CHECK:       ; %bb.0:
+; CHECK-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
+; CHECK-NEXT:    s_waitcnt_vscnt null, 0x0
+; CHECK-NEXT:    v_cmp_lt_i32_e64 s4, 0x62, v0
+; CHECK-NEXT:    ; implicit-def: $vcc_hi
+; CHECK-NEXT:    v_mov_b32_e32 v0, s4
+; CHECK-NEXT:    s_setpc_b64 s[30:31]
+  %cmp = icmp sge i32 %x, 99
+  %ballot = call i32 @llvm.amdgcn.ballot.i32(i1 %cmp)
+  ret i32 %ballot
+}
+
+define i32 @test5(float %x, float %y) {
+; CHECK-LABEL: test5:
+; CHECK:       ; %bb.0:
+; CHECK-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
+; CHECK-NEXT:    s_waitcnt_vscnt null, 0x0
+; CHECK-NEXT:    v_cmp_gt_f32_e64 s4, v0, v1
+; CHECK-NEXT:    ; implicit-def: $vcc_hi
+; CHECK-NEXT:    v_mov_b32_e32 v0, s4
+; CHECK-NEXT:    s_setpc_b64 s[30:31]
+  %cmp = fcmp ogt float %x, %y
+  %ballot = call i32 @llvm.amdgcn.ballot.i32(i1 %cmp)
+  ret i32 %ballot
+}

llvm/test/CodeGen/AMDGPU/llvm.amdgcn.ballot.i64.ll

+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc -march=amdgcn -mcpu=gfx900 < %s | FileCheck %s
+
+declare i64 @llvm.amdgcn.ballot.i64(i1)
+
+; Test ballot(0)
+
+define i64 @test0() {
+; CHECK-LABEL: test0:
+; CHECK:       ; %bb.0:
+; CHECK-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
+; CHECK-NEXT:    v_mov_b32_e32 v0, 0
+; CHECK-NEXT:    v_mov_b32_e32 v1, 0
+; CHECK-NEXT:    s_setpc_b64 s[30:31]
+  %ballot = call i64 @llvm.amdgcn.ballot.i64(i1 0)
+  ret i64 %ballot
+}
+
+; Test ballot(1)
+
+define i64 @test1() {
+; CHECK-LABEL: test1:
+; CHECK:       ; %bb.0:
+; CHECK-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
+; CHECK-NEXT:    v_mov_b32_e32 v0, exec_lo
+; CHECK-NEXT:    v_mov_b32_e32 v1, exec_hi
+; CHECK-NEXT:    s_setpc_b64 s[30:31]
+  %ballot = call i64 @llvm.amdgcn.ballot.i64(i1 1)
+  ret i64 %ballot
+}
+
+; Test ballot of a non-comparison operation
+
+define i64 @test2(i32 %x) {
+; CHECK-LABEL: test2:
+; CHECK:       ; %bb.0:
+; CHECK-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
+; CHECK-NEXT:    v_and_b32_e32 v0, 1, v0
+; CHECK-NEXT:    v_cmp_ne_u32_e64 s[4:5], 0, v0
+; CHECK-NEXT:    v_mov_b32_e32 v0, s4
+; CHECK-NEXT:    v_mov_b32_e32 v1, s5
+; CHECK-NEXT:    s_setpc_b64 s[30:31]
+  %trunc = trunc i32 %x to i1
+  %ballot = call i64 @llvm.amdgcn.ballot.i64(i1 %trunc)
+  ret i64 %ballot
+}
+
+; Test ballot of comparisons
+
+define i64 @test3(i32 %x, i32 %y) {
+; CHECK-LABEL: test3:
+; CHECK:       ; %bb.0:
+; CHECK-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
+; CHECK-NEXT:    v_cmp_eq_u32_e64 s[4:5], v0, v1
+; CHECK-NEXT:    v_mov_b32_e32 v0, s4
+; CHECK-NEXT:    v_mov_b32_e32 v1, s5
+; CHECK-NEXT:    s_setpc_b64 s[30:31]
+  %cmp = icmp eq i32 %x, %y
+  %ballot = call i64 @llvm.amdgcn.ballot.i64(i1 %cmp)
+  ret i64 %ballot
+}
+
+define i64 @test4(i32 %x) {
+; CHECK-LABEL: test4:
+; CHECK:       ; %bb.0:
+; CHECK-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
+; CHECK-NEXT:    s_movk_i32 s4, 0x62
+; CHECK-NEXT:    v_cmp_lt_i32_e64 s[4:5], s4, v0
+; CHECK-NEXT:    v_mov_b32_e32 v0, s4
+; CHECK-NEXT:    v_mov_b32_e32 v1, s5
+; CHECK-NEXT:    s_setpc_b64 s[30:31]
+  %cmp = icmp sge i32 %x, 99
+  %ballot = call i64 @llvm.amdgcn.ballot.i64(i1 %cmp)
+  ret i64 %ballot
+}
+
+define i64 @test5(float %x, float %y) {
+; CHECK-LABEL: test5:
+; CHECK:       ; %bb.0:
+; CHECK-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
+; CHECK-NEXT:    v_cmp_gt_f32_e64 s[4:5], v0, v1
+; CHECK-NEXT:    v_mov_b32_e32 v0, s4
+; CHECK-NEXT:    v_mov_b32_e32 v1, s5
+; CHECK-NEXT:    s_setpc_b64 s[30:31]
+  %cmp = fcmp ogt float %x, %y
+  %ballot = call i64 @llvm.amdgcn.ballot.i64(i1 %cmp)
+  ret i64 %ballot
+}

llvm/test/Transforms/InstCombine/AMDGPU/amdgcn-intrinsics.ll

 ; CHECK-NEXT:    [[RESULT:%.*]] = call i64 @llvm.amdgcn.fcmp.i64.f32(float [[X:%.*]], float 4.000000e+00, i32 2)
 ; CHECK-NEXT:    ret i64 [[RESULT]]
 ;
   %result = call i64 @llvm.amdgcn.fcmp.i64.f32(float 4.0, float %x, i32 4)
   ret i64 %result
 }
 
 ; --------------------------------------------------------------------
+; llvm.amdgcn.ballot
+; --------------------------------------------------------------------
+
+declare i64 @llvm.amdgcn.ballot.i64(i1) nounwind readnone convergent
+declare i32 @llvm.amdgcn.ballot.i32(i1) nounwind readnone convergent
+
+define i64 @ballot_nocombine_64(i1 %i) {
+; CHECK-LABEL: @ballot_nocombine_64(
+; CHECK-NEXT:    %b = call i64 @llvm.amdgcn.ballot.i64(i1 %i)
+; CHECK-NEXT:    ret i64 %b
+;
+  %b = call i64 @llvm.amdgcn.ballot.i64(i1 %i)
+  ret i64 %b
+}
+
+define i64 @ballot_zero_64() {
+; CHECK-LABEL: @ballot_zero_64(
+; CHECK-NEXT:    ret i64 0
+;
+  %b = call i64 @llvm.amdgcn.ballot.i64(i1 0)
+  ret i64 %b
+}
+
+define i64 @ballot_one_64() {
+; CHECK-LABEL: @ballot_one_64(
+; CHECK-NEXT:    %b = call i64 @llvm.read_register.i64(metadata !0) [[CONVERGENT]]
+; CHECK-NEXT:    ret i64 %b
+;
+  %b = call i64 @llvm.amdgcn.ballot.i64(i1 1)
+  ret i64 %b
+}
+
+define i32 @ballot_nocombine_32(i1 %i) {
+; CHECK-LABEL: @ballot_nocombine_32(
+; CHECK-NEXT:    %b = call i32 @llvm.amdgcn.ballot.i32(i1 %i)
+; CHECK-NEXT:    ret i32 %b
+;
+  %b = call i32 @llvm.amdgcn.ballot.i32(i1 %i)
+  ret i32 %b
+}
+
+define i32 @ballot_zero_32() {
+; CHECK-LABEL: @ballot_zero_32(
+; CHECK-NEXT:    ret i32 0
+;
+  %b = call i32 @llvm.amdgcn.ballot.i32(i1 0)
+  ret i32 %b
+}
+
+define i32 @ballot_one_32() {
+; CHECK-LABEL: @ballot_one_32(
+; CHECK-NEXT:    %b = call i32 @llvm.read_register.i32(metadata !1) [[CONVERGENT]]
+; CHECK-NEXT:    ret i32 %b
+;
+  %b = call i32 @llvm.amdgcn.ballot.i32(i1 1)
+  ret i32 %b
+}
+
+; --------------------------------------------------------------------
 ; llvm.amdgcn.wqm.vote
 ; --------------------------------------------------------------------
 
 declare i1 @llvm.amdgcn.wqm.vote(i1)
 
 define float @wqm_vote_true() {
 ; CHECK-LABEL: @wqm_vote_true(
 ; CHECK-NEXT:  main_body: