diff options
| author | Srinivasa Ravi <srinivasar@nvidia.com> | 2025-05-06 12:46:41 +0530 |
|---|---|---|
| committer | GitHub <noreply@github.com> | 2025-05-06 12:46:41 +0530 |
| commit | bb2aa1adcc4ad1508b477a2f53aa4ab922b4f99f (patch) | |
| tree | e7b53b46dfffe7a2f27bb7499ea5081a35be671e /flang/lib/Frontend/CompilerInvocation.cpp | |
| parent | 3f1eafaa04f1c04ae5c7aae3e452eb75c507584d (diff) | |
| download | llvm-bb2aa1adcc4ad1508b477a2f53aa4ab922b4f99f.zip llvm-bb2aa1adcc4ad1508b477a2f53aa4ab922b4f99f.tar.gz llvm-bb2aa1adcc4ad1508b477a2f53aa4ab922b4f99f.tar.bz2 | |
[MLIR][NVVM] Update support for conversions to f8x2 and f6x2 types (#137781)
This change:
- Adds the `cvt.f32x2.to.f8x2`, `cvt.f16x2.to.f8x2`, and
`cvt.bf16x2.to.f8x2`
Ops to the NVVM dialect for the conversions to `.e4m3x2`, `e5m2x2`,
and `.ue8m0x2` types.
- Renames the recently added `cvt.to.f6x2` Op to `cvt.f32x2.to.f6x2`
for consistency with the other conversion Ops.
For more information, see PTX ISA:
https://docs.nvidia.com/cuda/parallel-thread-execution/#data-movement-and-conversion-instructions-cvt
Diffstat (limited to 'flang/lib/Frontend/CompilerInvocation.cpp')
0 files changed, 0 insertions, 0 deletions