KungFuDonkey wrote:
```
opt.exe -S -scalarizer -dxil-op-lower -mtriple=dxil-pc-shadermodel6.3-library <
C:\develop\personal_gh\llvm-project\llvm\test\CodeGen\DirectX\WaveActiveMin.ll
| c:\develop\personal_gh\llvm-project\build\bin\filecheck.exe
C:\develop\personal_gh\llvm-project\llvm\test\CodeGen\DirectX\WaveActiveMin.ll
Intrinsic has incorrect return type!
ptr @llvm.dx.wave.reduce.min.f32
```
from:
```
define noundef float @wave_active_min_half(float noundef %expr) {
entry:
; CHECK: call float @dx.op.waveActiveOp.f32(i32 119, float %expr, i8 2, i8
0){{$}}
%ret = call float @llvm.dx.wave.reduce.min.f32(float %expr)
ret float %ret
}
```
This happens for all float types. I cannot find where dx is handling floats for
WaveActiveMax
https://github.com/llvm/llvm-project/pull/164385
_______________________________________________
cfe-commits mailing list
[email protected]
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
