diff options
| author | Alex MacLean <amaclean@nvidia.com> | 2025-05-27 13:55:20 -0700 |
|---|---|---|
| committer | GitHub <noreply@github.com> | 2025-05-27 13:55:20 -0700 |
| commit | 9cf4603bf9e3bc4249fbe0d183fa31f2bba1eb7e (patch) | |
| tree | 55337b7b33b927633ec092c07dd214c542db5a4a /clang/lib/CodeGen/CodeGenModule.cpp | |
| parent | 8821c38788d215dd5691774d6777cf58882a2037 (diff) | |
| download | llvm-9cf4603bf9e3bc4249fbe0d183fa31f2bba1eb7e.zip llvm-9cf4603bf9e3bc4249fbe0d183fa31f2bba1eb7e.tar.gz llvm-9cf4603bf9e3bc4249fbe0d183fa31f2bba1eb7e.tar.bz2 | |
[NVPTX] Add -nvptx-prec-divf32=3 to disable ftz for f32 fdiv (#141276)
This change allows a new value (3) to be specified to the
-nvptx-prec-divf32 command line option to force non-ftz div generation
in all cases. This can be useful when debugging precision issues in
NVPTX.
Diffstat (limited to 'clang/lib/CodeGen/CodeGenModule.cpp')
0 files changed, 0 insertions, 0 deletions