diff options
| author | Srinivasa Ravi <srinivasar@nvidia.com> | 2025-10-15 11:55:55 +0530 |
|---|---|---|
| committer | GitHub <noreply@github.com> | 2025-10-15 11:55:55 +0530 |
| commit | e98de2ed7c1db5144b52c5284118a95229d0a8b2 (patch) | |
| tree | 3a7b8b4959892535c63fde03769352d9db5cbbb8 /clang/lib/Serialization/ModuleCache.cpp | |
| parent | de9013fa18a12f982a223ea30819b523e86667d4 (diff) | |
| download | llvm-e98de2ed7c1db5144b52c5284118a95229d0a8b2.zip llvm-e98de2ed7c1db5144b52c5284118a95229d0a8b2.tar.gz llvm-e98de2ed7c1db5144b52c5284118a95229d0a8b2.tar.bz2 | |
[MLIR][NVVM] Add support for f32x2 to f4x2 conversion (#162273)
This change adds the `convert.f32x2.to.f4x2` op to the NVVM Dialect
for converting a pair of f32 values to an f4x2 (`e2m1x2`) value.
PTX reference:
https://docs.nvidia.com/cuda/parallel-thread-execution/index.html#data-movement-and-conversion-instructions-cvt
Diffstat (limited to 'clang/lib/Serialization/ModuleCache.cpp')
0 files changed, 0 insertions, 0 deletions