diff options
| author | Matt Arsenault <Matthew.Arsenault@amd.com> | 2019-06-20 21:58:24 +0000 |
|---|---|---|
| committer | Matt Arsenault <Matthew.Arsenault@amd.com> | 2019-06-20 21:58:24 +0000 |
| commit | d88db6d7fc942947ad4a068b38c5b5af7d5d1751 (patch) | |
| tree | 670dc13f6ccde2424eded38229f3b60061f5e6a2 /llvm/test/CodeGen/AMDGPU/wave32.ll | |
| parent | 25f08a17c318e8ffbd30ecbab4d3ea5c5105ddbc (diff) | |
| download | bcm5719-llvm-d88db6d7fc942947ad4a068b38c5b5af7d5d1751.tar.gz bcm5719-llvm-d88db6d7fc942947ad4a068b38c5b5af7d5d1751.zip | |
AMDGPU: Always use s33 for global scratch wave offset
Every called function could possibly need this to calculate the
absolute address of stack objectst, and this avoids inserting a copy
around every call site in the kernel. It's also somewhat cleaner to
keep this in a callee saved SGPR.
llvm-svn: 363990
Diffstat (limited to 'llvm/test/CodeGen/AMDGPU/wave32.ll')
| -rw-r--r-- | llvm/test/CodeGen/AMDGPU/wave32.ll | 12 |
1 files changed, 6 insertions, 6 deletions
diff --git a/llvm/test/CodeGen/AMDGPU/wave32.ll b/llvm/test/CodeGen/AMDGPU/wave32.ll index 7039ef5f85f..2fe4676e93d 100644 --- a/llvm/test/CodeGen/AMDGPU/wave32.ll +++ b/llvm/test/CodeGen/AMDGPU/wave32.ll @@ -1080,13 +1080,13 @@ declare void @external_void_func_void() #1 ; GFX1064-NEXT: s_mov_b64 exec, [[COPY_EXEC0]] ; GFX1032-NEXT: s_mov_b32 exec_lo, [[COPY_EXEC0]] -; GCN-DAG: v_writelane_b32 v32, s33, 0 -; GCN-DAG: v_writelane_b32 v32, s34, 1 -; GCN-DAG: s_mov_b32 s33, s5 +; GCN-DAG: v_writelane_b32 v32, s34, 0 +; GCN-DAG: v_writelane_b32 v32, s35, 1 +; GCN-DAG: s_mov_b32 [[COPY_FP:s[0-9]+]], s5 ; GCN: s_swappc_b64 -; GCN-DAG: s_mov_b32 s5, s33 -; GCN-DAG: v_readlane_b32 s34, v32, 1 -; GCN-DAG: v_readlane_b32 s33, v32, 0 +; GCN-DAG: s_mov_b32 s5, [[COPY_FP]] +; GCN-DAG: v_readlane_b32 s35, v32, 1 +; GCN-DAG: v_readlane_b32 s34, v32, 0 ; GFX1064: s_or_saveexec_b64 [[COPY_EXEC1:s\[[0-9]+:[0-9]+\]]], -1{{$}} ; GFX1032: s_or_saveexec_b32 [[COPY_EXEC1:s[0-9]]], -1{{$}} |
