diff options
| author | Alexander Timofeev <Alexander.Timofeev@amd.com> | 2017-07-04 17:32:00 +0000 |
|---|---|---|
| committer | Alexander Timofeev <Alexander.Timofeev@amd.com> | 2017-07-04 17:32:00 +0000 |
| commit | 982aee6a38ed7b416bfd2ba67f2144148251f733 (patch) | |
| tree | 46cc47fedf5dc80cd04d83a71845dfaaf09a3a23 /llvm/test/CodeGen/AMDGPU/fadd64.ll | |
| parent | ada4ddc0bc49d1837ad296e2d269568c8f632d35 (diff) | |
| download | bcm5719-llvm-982aee6a38ed7b416bfd2ba67f2144148251f733.tar.gz bcm5719-llvm-982aee6a38ed7b416bfd2ba67f2144148251f733.zip | |
[AMDGPU] Switch scalarize global loads ON by default
Differential revision: https://reviews.llvm.org/D34407
llvm-svn: 307097
Diffstat (limited to 'llvm/test/CodeGen/AMDGPU/fadd64.ll')
| -rw-r--r-- | llvm/test/CodeGen/AMDGPU/fadd64.ll | 12 |
1 files changed, 10 insertions, 2 deletions
diff --git a/llvm/test/CodeGen/AMDGPU/fadd64.ll b/llvm/test/CodeGen/AMDGPU/fadd64.ll index c936d98673b..8fd1f52006f 100644 --- a/llvm/test/CodeGen/AMDGPU/fadd64.ll +++ b/llvm/test/CodeGen/AMDGPU/fadd64.ll @@ -5,8 +5,11 @@ ; CHECK: v_add_f64 {{v[[0-9]+:[0-9]+]}}, {{v[[0-9]+:[0-9]+]}}, {{v[[0-9]+:[0-9]+]}} define amdgpu_kernel void @v_fadd_f64(double addrspace(1)* %out, double addrspace(1)* %in1, double addrspace(1)* %in2) { - %r0 = load double, double addrspace(1)* %in1 - %r1 = load double, double addrspace(1)* %in2 + %tid = call i32 @llvm.amdgcn.workitem.id.x() + %gep1 = getelementptr inbounds double, double addrspace(1)* %in1, i32 %tid + %gep2 = getelementptr inbounds double, double addrspace(1)* %in2, i32 %tid + %r0 = load double, double addrspace(1)* %gep1 + %r1 = load double, double addrspace(1)* %gep2 %r2 = fadd double %r0, %r1 store double %r2, double addrspace(1)* %out ret void @@ -42,3 +45,8 @@ define amdgpu_kernel void @s_fadd_v2f64(<2 x double> addrspace(1)* %out, <2 x do store <2 x double> %r2, <2 x double> addrspace(1)* %out ret void } + +declare i32 @llvm.amdgcn.workitem.id.x() #1 + +attributes #0 = { nounwind } +attributes #1 = { nounwind readnone } |
