diff options
Diffstat (limited to 'test/CodeGen/AMDGPU/fmax_legacy.ll')
| -rw-r--r-- | test/CodeGen/AMDGPU/fmax_legacy.ll | 40 |
1 files changed, 40 insertions, 0 deletions
diff --git a/test/CodeGen/AMDGPU/fmax_legacy.ll b/test/CodeGen/AMDGPU/fmax_legacy.ll index 413957d2982a..d374fb67350c 100644 --- a/test/CodeGen/AMDGPU/fmax_legacy.ll +++ b/test/CodeGen/AMDGPU/fmax_legacy.ll @@ -87,6 +87,46 @@ define void @test_fmax_legacy_ogt_f32(float addrspace(1)* %out, float addrspace( ret void } +; FUNC-LABEL: {{^}}test_fmax_legacy_ogt_v1f32: +; SI: buffer_load_dword [[A:v[0-9]+]], {{v\[[0-9]+:[0-9]+\]}}, {{s\[[0-9]+:[0-9]+\]}}, 0 addr64{{$}} +; SI: buffer_load_dword [[B:v[0-9]+]], {{v\[[0-9]+:[0-9]+\]}}, {{s\[[0-9]+:[0-9]+\]}}, 0 addr64 offset:4 +; SI-SAFE: v_max_legacy_f32_e32 {{v[0-9]+}}, [[A]], [[B]] +; SI-NONAN: v_max_f32_e32 {{v[0-9]+}}, [[B]], [[A]] +; EG: MAX +define void @test_fmax_legacy_ogt_v1f32(<1 x float> addrspace(1)* %out, <1 x float> addrspace(1)* %in) #0 { + %tid = call i32 @llvm.r600.read.tidig.x() #1 + %gep.0 = getelementptr <1 x float>, <1 x float> addrspace(1)* %in, i32 %tid + %gep.1 = getelementptr <1 x float>, <1 x float> addrspace(1)* %gep.0, i32 1 + + %a = load <1 x float>, <1 x float> addrspace(1)* %gep.0 + %b = load <1 x float>, <1 x float> addrspace(1)* %gep.1 + + %cmp = fcmp ogt <1 x float> %a, %b + %val = select <1 x i1> %cmp, <1 x float> %a, <1 x float> %b + store <1 x float> %val, <1 x float> addrspace(1)* %out + ret void +} + +; FUNC-LABEL: {{^}}test_fmax_legacy_ogt_v3f32: +; SI-SAFE: v_max_legacy_f32_e32 +; SI-SAFE: v_max_legacy_f32_e32 +; SI-SAFE: v_max_legacy_f32_e32 +; SI-NONAN: v_max_f32_e32 +; SI-NONAN: v_max_f32_e32 +; SI-NONAN: v_max_f32_e32 +define void @test_fmax_legacy_ogt_v3f32(<3 x float> addrspace(1)* %out, <3 x float> addrspace(1)* %in) #0 { + %tid = call i32 @llvm.r600.read.tidig.x() #1 + %gep.0 = getelementptr <3 x float>, <3 x float> addrspace(1)* %in, i32 %tid + %gep.1 = getelementptr <3 x float>, <3 x float> addrspace(1)* %gep.0, i32 1 + + %a = load <3 x float>, <3 x float> addrspace(1)* %gep.0 + %b = load <3 x float>, <3 x float> addrspace(1)* %gep.1 + + %cmp = fcmp ogt <3 x float> %a, %b + %val = select <3 x i1> %cmp, <3 x float> %a, <3 x float> %b + store <3 x float> %val, <3 x float> addrspace(1)* %out + ret void +} ; FUNC-LABEL: @test_fmax_legacy_ogt_f32_multi_use ; SI: buffer_load_dword [[A:v[0-9]+]], {{v\[[0-9]+:[0-9]+\]}}, {{s\[[0-9]+:[0-9]+\]}}, 0 addr64{{$}} |