AMDGPU: Fold away mbcnt.hi in wave32 mode

[arsenm]

Summary

This will allow libraries to drop some of the special casing based on
wave size.

[foad]

llvm/lib/Target/AMDGPU/AMDGPUInstCombineIntrinsic.cpp

906

Nit: I find the comment slightly misleading, since it suggests to me that something is being ANDed with exec which is not true. Maybe say something like "ThreadPosition < 32" instead of "exec_hi is all 0"?

[arsenm]

5ccfc4543d1784ed98cfae52364e438de5a993a6

llvm/lib/Target/AMDGPU/AMDGPUInstCombineIntrinsic.cpp

if (match(Src1, PatternMatch::m_Zero()) &&
                        ConstantInt::get(CC->getType(), SrcPred)};
       CallInst *NewCall = IC.Builder.CreateCall(NewF, Args);
       NewCall->takeName(&II);
       return IC.replaceInstUsesWith(II, NewCall);
     }
 
     break;
   }
+  case Intrinsic::amdgcn_mbcnt_hi: {
+    // exec_hi is all 0, so this is just a copy.
+    if (ST->isWave32())
+      return IC.replaceInstUsesWith(II, II.getArgOperand(1));
+    break;
+  }
   case Intrinsic::amdgcn_ballot: {
     if (auto *Src = dyn_cast<ConstantInt>(II.getArgOperand(0))) {
       if (Src->isZero()) {
         // amdgcn.ballot(i1 0) is zero.
         return IC.replaceInstUsesWith(II, Constant::getNullValue(II.getType()));
       }
     }
     break;

llvm/test/Transforms/InstCombine/AMDGPU/mbcnt.ll

 define i32 @mbcnt_hi(i32 %x, i32 %y) {
 ; DEFAULT-LABEL: define i32 @mbcnt_hi
 ; DEFAULT-SAME: (i32 [[X:%.*]], i32 [[Y:%.*]]) {
 ; DEFAULT-NEXT:    [[HI:%.*]] = call i32 @llvm.amdgcn.mbcnt.hi(i32 [[X]], i32 [[Y]])
 ; DEFAULT-NEXT:    ret i32 [[HI]]
 ;
 ; WAVE32-LABEL: define i32 @mbcnt_hi
 ; WAVE32-SAME: (i32 [[X:%.*]], i32 [[Y:%.*]]) #[[ATTR1]] {
-; WAVE32-NEXT:    [[HI:%.*]] = call i32 @llvm.amdgcn.mbcnt.hi(i32 [[X]], i32 [[Y]])
-; WAVE32-NEXT:    ret i32 [[HI]]
+; WAVE32-NEXT:    ret i32 [[Y]]
 ;
 ; WAVE64-LABEL: define i32 @mbcnt_hi
 ; WAVE64-SAME: (i32 [[X:%.*]], i32 [[Y:%.*]]) #[[ATTR1]] {
 ; WAVE64-NEXT:    [[HI:%.*]] = call i32 @llvm.amdgcn.mbcnt.hi(i32 [[X]], i32 [[Y]])
 ; WAVE64-NEXT:    ret i32 [[HI]]
 ;
   %hi = call i32 @llvm.amdgcn.mbcnt.hi(i32 %x, i32 %y)
   ret i32 %hi
 }
 
 define i32 @mbcnt_lo_hi(i32 %x, i32 %y, i32 %z) {
 ; DEFAULT-LABEL: define i32 @mbcnt_lo_hi
 ; DEFAULT-SAME: (i32 [[X:%.*]], i32 [[Y:%.*]], i32 [[Z:%.*]]) {
 ; DEFAULT-NEXT:    [[LO:%.*]] = call i32 @llvm.amdgcn.mbcnt.lo(i32 [[X]], i32 [[Y]])
 ; DEFAULT-NEXT:    [[HI:%.*]] = call i32 @llvm.amdgcn.mbcnt.hi(i32 [[Z]], i32 [[LO]])
 ; DEFAULT-NEXT:    ret i32 [[HI]]
 ;
 ; WAVE32-LABEL: define i32 @mbcnt_lo_hi
 ; WAVE32-SAME: (i32 [[X:%.*]], i32 [[Y:%.*]], i32 [[Z:%.*]]) #[[ATTR1]] {
 ; WAVE32-NEXT:    [[LO:%.*]] = call i32 @llvm.amdgcn.mbcnt.lo(i32 [[X]], i32 [[Y]])
-; WAVE32-NEXT:    [[HI:%.*]] = call i32 @llvm.amdgcn.mbcnt.hi(i32 [[Z]], i32 [[LO]])
-; WAVE32-NEXT:    ret i32 [[HI]]
+; WAVE32-NEXT:    ret i32 [[LO]]
 ;
 ; WAVE64-LABEL: define i32 @mbcnt_lo_hi
 ; WAVE64-SAME: (i32 [[X:%.*]], i32 [[Y:%.*]], i32 [[Z:%.*]]) #[[ATTR1]] {
 ; WAVE64-NEXT:    [[LO:%.*]] = call i32 @llvm.amdgcn.mbcnt.lo(i32 [[X]], i32 [[Y]])
 ; WAVE64-NEXT:    [[HI:%.*]] = call i32 @llvm.amdgcn.mbcnt.hi(i32 [[Z]], i32 [[LO]])
 ; WAVE64-NEXT:    ret i32 [[HI]]
 ;
   %lo = call i32 @llvm.amdgcn.mbcnt.lo(i32 %x, i32 %y)
   %hi = call i32 @llvm.amdgcn.mbcnt.hi(i32 %z, i32 %lo)
   ret i32 %hi
 }
 
 define i32 @ockl_lane_u32() {
 ; DEFAULT-LABEL: define i32 @ockl_lane_u32() {
 ; DEFAULT-NEXT:    [[LO:%.*]] = call i32 @llvm.amdgcn.mbcnt.lo(i32 -1, i32 0)
 ; DEFAULT-NEXT:    [[HI:%.*]] = call i32 @llvm.amdgcn.mbcnt.hi(i32 -1, i32 [[LO]])
 ; DEFAULT-NEXT:    ret i32 [[HI]]
 ;
 ; WAVE32-LABEL: define i32 @ockl_lane_u32
 ; WAVE32-SAME: () #[[ATTR1]] {
 ; WAVE32-NEXT:    [[LO:%.*]] = call i32 @llvm.amdgcn.mbcnt.lo(i32 -1, i32 0)
-; WAVE32-NEXT:    [[HI:%.*]] = call i32 @llvm.amdgcn.mbcnt.hi(i32 -1, i32 [[LO]])
-; WAVE32-NEXT:    ret i32 [[HI]]
+; WAVE32-NEXT:    ret i32 [[LO]]
 ;
 ; WAVE64-LABEL: define i32 @ockl_lane_u32
 ; WAVE64-SAME: () #[[ATTR1]] {
 ; WAVE64-NEXT:    [[LO:%.*]] = call i32 @llvm.amdgcn.mbcnt.lo(i32 -1, i32 0)
 ; WAVE64-NEXT:    [[HI:%.*]] = call i32 @llvm.amdgcn.mbcnt.hi(i32 -1, i32 [[LO]])
 ; WAVE64-NEXT:    ret i32 [[HI]]
 ;
   %lo = call i32 @llvm.amdgcn.mbcnt.lo(i32 -1, i32 0)
   %hi = call i32 @llvm.amdgcn.mbcnt.hi(i32 -1, i32 %lo)
   ret i32 %hi
 }
 
 ;; NOTE: These prefixes are unused and the list is autogenerated. Do not add tests below this line:
 ; CHECK: {{.*}}