mirror of
https://github.com/RPCS3/llvm-mirror.git
synced 2024-10-23 13:02:52 +02:00
dd9ab77318
Currently the default C calling convention functions are treated the same as compute kernels. Make this explicit so the default calling convention can be changed to a non-kernel. Converted with perl -pi -e 's/define void/define amdgpu_kernel void/' on the relevant test directories (and undoing in one place that actually wanted a non-kernel). llvm-svn: 298444
29 lines
1.3 KiB
LLVM
29 lines
1.3 KiB
LLVM
; RUN: llc -march=amdgcn -verify-machineinstrs < %s | FileCheck -check-prefix=GCN %s
|
|
; RUN: llc -march=amdgcn -mcpu=tonga -verify-machineinstrs < %s | FileCheck -check-prefix=GCN %s
|
|
|
|
; GCN-LABEL: {{^}}test_fmed3:
|
|
; GCN: v_med3_f32 v{{[0-9]+}}, s{{[0-9]+}}, v{{[0-9]+}}, v{{[0-9]+}}
|
|
define amdgpu_kernel void @test_fmed3(float addrspace(1)* %out, float %src0, float %src1, float %src2) #1 {
|
|
%mad = call float @llvm.amdgcn.fmed3.f32(float %src0, float %src1, float %src2)
|
|
store float %mad, float addrspace(1)* %out
|
|
ret void
|
|
}
|
|
|
|
; GCN-LABEL: {{^}}test_fmed3_srcmods:
|
|
; GCN: v_med3_f32 v{{[0-9]+}}, -s{{[0-9]+}}, |v{{[0-9]+}}|, -|v{{[0-9]+}}|
|
|
define amdgpu_kernel void @test_fmed3_srcmods(float addrspace(1)* %out, float %src0, float %src1, float %src2) #1 {
|
|
%src0.fneg = fsub float -0.0, %src0
|
|
%src1.fabs = call float @llvm.fabs.f32(float %src1)
|
|
%src2.fabs = call float @llvm.fabs.f32(float %src2)
|
|
%src2.fneg.fabs = fsub float -0.0, %src2.fabs
|
|
%mad = call float @llvm.amdgcn.fmed3.f32(float %src0.fneg, float %src1.fabs, float %src2.fneg.fabs)
|
|
store float %mad, float addrspace(1)* %out
|
|
ret void
|
|
}
|
|
|
|
declare float @llvm.amdgcn.fmed3.f32(float, float, float) #0
|
|
declare float @llvm.fabs.f32(float) #0
|
|
|
|
attributes #0 = { nounwind readnone }
|
|
attributes #1 = { nounwind }
|