diff options
| author | Konstantin Zhuravlyov <kzhuravl_dev@outlook.com> | 2018-09-12 18:50:47 +0000 |
|---|---|---|
| committer | Konstantin Zhuravlyov <kzhuravl_dev@outlook.com> | 2018-09-12 18:50:47 +0000 |
| commit | 71e43ee47d05293fccdecc61bc118301088b2466 (patch) | |
| tree | 50643334b8857b7899a0cce15b7e9eaaae5dcf34 /llvm/test/CodeGen/AMDGPU | |
| parent | 0090753ef00944036c86811ade14a7eb049e9b52 (diff) | |
| download | bcm5719-llvm-71e43ee47d05293fccdecc61bc118301088b2466.tar.gz bcm5719-llvm-71e43ee47d05293fccdecc61bc118301088b2466.zip | |
AMDGPU: Re-apply r341982 after fixing the layering issue
Move isa version determination into TargetParser.
Also switch away from target features to CPU string when
determining isa version. This fixes an issue when we
output wrong isa version in the object code when features
of a particular CPU are altered (i.e. gfx902 w/o xnack
used to result in gfx900).
llvm-svn: 342069
Diffstat (limited to 'llvm/test/CodeGen/AMDGPU')
| -rw-r--r-- | llvm/test/CodeGen/AMDGPU/gfx902-without-xnack.ll | 8 |
1 files changed, 8 insertions, 0 deletions
diff --git a/llvm/test/CodeGen/AMDGPU/gfx902-without-xnack.ll b/llvm/test/CodeGen/AMDGPU/gfx902-without-xnack.ll new file mode 100644 index 00000000000..445e112a301 --- /dev/null +++ b/llvm/test/CodeGen/AMDGPU/gfx902-without-xnack.ll @@ -0,0 +1,8 @@ +; RUN: llc -march=amdgcn -mtriple=amdgcn-amd-amdhsa -mcpu=gfx902 -mattr=-xnack < %s | FileCheck %s + +; CHECK: .hsa_code_object_isa 9,0,2,"AMD","AMDGPU" +define amdgpu_kernel void @test_kernel(float addrspace(1)* %out0, double addrspace(1)* %out1) nounwind { + store float 0.0, float addrspace(1)* %out0 + ret void +} + |
