We need to apply DXIL op attributes to the functions themselves, and all DXIL ops should have the `unwind` attribute. This matches the DXC behaviour and what consumers like warp's GPU-based validation expect. Fixes #193620
30 lines
1.2 KiB
LLVM
30 lines
1.2 KiB
LLVM
; RUN: opt -S -dxil-intrinsic-expansion -mtriple=dxil-pc-shadermodel6.3-library %s | FileCheck %s --check-prefixes=CHECK,EXPCHECK
|
|
; RUN: opt -S -dxil-intrinsic-expansion -dxil-op-lower -mtriple=dxil-pc-shadermodel6.3-library %s | FileCheck %s --check-prefixes=CHECK,DOPCHECK
|
|
|
|
; Make sure dxil operation function calls for log are generated.
|
|
|
|
define noundef float @log_float(float noundef %a) #0 {
|
|
entry:
|
|
; DOPCHECK: call float @dx.op.unary.f32(i32 23, float %{{.*}})
|
|
; EXPCHECK: call float @llvm.log2.f32(float %a)
|
|
; CHECK: fmul float 0x3FE62E4300000000, %{{.*}}
|
|
%elt.log = call float @llvm.log.f32(float %a)
|
|
ret float %elt.log
|
|
}
|
|
|
|
define noundef half @log_half(half noundef %a) #0 {
|
|
entry:
|
|
; DOPCHECK: call half @dx.op.unary.f16(i32 23, half %{{.*}})
|
|
; EXPCHECK: call half @llvm.log2.f16(half %a)
|
|
; CHECK: fmul half 0xH398C, %{{.*}}
|
|
%elt.log = call half @llvm.log.f16(half %a)
|
|
ret half %elt.log
|
|
}
|
|
|
|
; DOPCHECK-DAG: declare half @dx.op.unary.f16(i32, half) #[[#ATTR0:]]
|
|
; DOPCHECK-DAG: declare float @dx.op.unary.f32(i32, float) #[[#ATTR0]]
|
|
; DOPCHECK: attributes #[[#ATTR0]] = { nounwind memory(none) }
|
|
|
|
declare half @llvm.log.f16(half)
|
|
declare float @llvm.log.f32(float)
|