The previous WaveActiveBallot implementation did not account for the fact that the DXC implementation of the intrinsic returns a struct type with 4 uints, rather than a vector of 4 uints. This must be respected, otherwise the validator will reject the uses of WaveActiveBallot that return a vector of 4 uints. This PR updates the return type and adds the DXC-specific return type `fouri32` to use for the intrinsic.
38 lines
1.4 KiB
LLVM
38 lines
1.4 KiB
LLVM
; RUN: opt -S -scalarizer -dxil-op-lower %s | FileCheck %s
|
|
|
|
target datalayout = "e-m:e-p:32:32-i1:32-i8:8-i16:16-i32:32-i64:64-f16:16-f32:32-f64:64-n8:16:32:64-v48:16:16-v96:32:32-v192:64:64"
|
|
target triple = "dxilv1.3-pc-shadermodel6.3-compute"
|
|
|
|
; The definition of the custom type should be added
|
|
; CHECK: %dx.types.fouri32 = type { i32, i32, i32, i32 }
|
|
|
|
; Function Attrs: alwaysinline convergent mustprogress norecurse nounwind
|
|
define hidden noundef <4 x i32> @_Z4testb(i1 noundef %p1) {
|
|
entry:
|
|
%p1.addr = alloca i32, align 4
|
|
%storedv = zext i1 %p1 to i32
|
|
store i32 %storedv, ptr %p1.addr, align 4
|
|
%0 = load i32, ptr %p1.addr, align 4
|
|
%loadedv = trunc i32 %0 to i1
|
|
%1 = load i32, ptr %p1.addr, align 4
|
|
%loadedv1 = trunc i32 %1 to i1
|
|
|
|
; CHECK: call %dx.types.fouri32 @dx.op.waveActiveBallot(i32 116, i1 %loadedv1)
|
|
|
|
%2 = call { i32, i32, i32, i32 } @llvm.dx.wave.ballot.i32(i1 %loadedv1)
|
|
%3 = extractvalue { i32, i32, i32, i32 } %2, 0
|
|
%4 = insertelement <4 x i32> poison, i32 %3, i32 0
|
|
%5 = extractvalue { i32, i32, i32, i32 } %2, 1
|
|
%6 = insertelement <4 x i32> %4, i32 %5, i32 1
|
|
%7 = extractvalue { i32, i32, i32, i32 } %2, 2
|
|
%8 = insertelement <4 x i32> %6, i32 %7, i32 2
|
|
%9 = extractvalue { i32, i32, i32, i32 } %2, 3
|
|
%10 = insertelement <4 x i32> %8, i32 %9, i32 3
|
|
|
|
; CHECK-NOT: ret %dx.types.fouri32
|
|
; CHECK: ret <4 x i32>
|
|
ret <4 x i32> %10
|
|
}
|
|
|
|
declare { i32, i32, i32, i32 } @llvm.dx.wave.ballot.i32(i1)
|