diff options
| author | Pierre van Houtryve <pierre.vanhoutryve@amd.com> | 2025-05-16 10:31:03 +0200 |
|---|---|---|
| committer | GitHub <noreply@github.com> | 2025-05-16 10:31:03 +0200 |
| commit | aacebaeab5448b4ef614aa8aca52ca210e451f79 (patch) | |
| tree | 140211e8c660b789482901537142a2d773baed87 /llvm/lib/Target/AMDGPU/AMDGPUCodeGenPrepare.cpp | |
| parent | 5e7bc5e080fb393c3ddb954fa9fd3714284f72c2 (diff) | |
[AMDGPU] Do not promote uniform i16 operations to i32 in CGP (#140208)
For the majority of cases, this is a neutral or positive change.
There are even testcases that greatly benefit from it, but some regressions are possible.
There is #140040 for GlobalISel that'd need to be fixed but it's only a one instruction regression and I think it can be fixed later.
Solves #64591
Diffstat (limited to 'llvm/lib/Target/AMDGPU/AMDGPUCodeGenPrepare.cpp')
| -rw-r--r-- | llvm/lib/Target/AMDGPU/AMDGPUCodeGenPrepare.cpp | 8 |
1 files changed, 4 insertions, 4 deletions
diff --git a/llvm/lib/Target/AMDGPU/AMDGPUCodeGenPrepare.cpp b/llvm/lib/Target/AMDGPU/AMDGPUCodeGenPrepare.cpp index efb2894aaf64..70e80f9fe14e 100644 --- a/llvm/lib/Target/AMDGPU/AMDGPUCodeGenPrepare.cpp +++ b/llvm/lib/Target/AMDGPU/AMDGPUCodeGenPrepare.cpp @@ -47,10 +47,10 @@ static cl::opt<bool> WidenLoads( cl::init(false)); static cl::opt<bool> Widen16BitOps( - "amdgpu-codegenprepare-widen-16-bit-ops", - cl::desc("Widen uniform 16-bit instructions to 32-bit in AMDGPUCodeGenPrepare"), - cl::ReallyHidden, - cl::init(true)); + "amdgpu-codegenprepare-widen-16-bit-ops", + cl::desc( + "Widen uniform 16-bit instructions to 32-bit in AMDGPUCodeGenPrepare"), + cl::ReallyHidden, cl::init(false)); static cl::opt<bool> BreakLargePHIs("amdgpu-codegenprepare-break-large-phis", |