diff options
| author | Matt Arsenault <Matthew.Arsenault@amd.com> | 2019-10-31 23:32:31 -0700 |
|---|---|---|
| committer | Matt Arsenault <arsenm2@gmail.com> | 2020-04-02 17:17:12 -0400 |
| commit | 75cf30918f328523fec6a4ed52ea610cfa13e21b (patch) | |
| tree | 8ee8e392f88a39185876253cd413a5c0154ba5b7 /llvm/tools/llvm-objdump/llvm-objdump.cpp | |
| parent | 0c85c488e2b5be54ce26c9f772e66d48a9290c51 (diff) | |
| download | llvm-75cf30918f328523fec6a4ed52ea610cfa13e21b.zip llvm-75cf30918f328523fec6a4ed52ea610cfa13e21b.tar.gz llvm-75cf30918f328523fec6a4ed52ea610cfa13e21b.tar.bz2 | |
AMDGPU: Assume f32 denormals are enabled by default
This will likely introduce catastrophic performance regressions on
older subtargets, but should be correct. A follow up change will
remove the old fp32-denormals subtarget features, and switch to using
the new denormal-fp-math/denormal-fp-math-f32 attributes. Frontends
should be making sure to add the denormal-fp-math-f32 attribute when
appropriate to avoid performance regressions.
Diffstat (limited to 'llvm/tools/llvm-objdump/llvm-objdump.cpp')
0 files changed, 0 insertions, 0 deletions