diff options
| author | Jefferson Le Quellec <jefferson.lequellec@codeplay.com> | 2024-12-12 13:35:07 +0100 |
|---|---|---|
| committer | GitHub <noreply@github.com> | 2024-12-12 13:35:07 +0100 |
| commit | 81825687b4b45e0a6839fd05cad7bedf18205315 (patch) | |
| tree | 9d6b92cacefdf60b2ae12c1548da9eb1b5ec38bb /llvm/docs/tutorial | |
| parent | 0cbdad4bd2396b740742d9ae94cba7d7b8a32cb5 (diff) | |
| download | llvm-81825687b4b45e0a6839fd05cad7bedf18205315.zip llvm-81825687b4b45e0a6839fd05cad7bedf18205315.tar.gz llvm-81825687b4b45e0a6839fd05cad7bedf18205315.tar.bz2 | |
[MLIR][GPUToLLVMSPV] Update ConvertGpuOpsToLLVMSPVOps's option (#118818)
## Description
This PR updates the `ConvertGpuOpsToLLVMSPVOps`'s option by replacing
the `index-bitwidth` with a boolean option `use-64bit-index` (similar to
the `ConvertGPUToSPIRV` option).
The reason for this modification is because the
`ConvertGpuOpsToLLVMSPVOps`:
> Generate LLVM operations to be ingested by a SPIR-V backend for gpu
operations
In the context of SPIR-V specifications only two physical addressing
models are allowed: `Physical32` and `Physical64`.
This change guarantees output sanity by preventing invalid or
unsupported index bitwidths from being specified.
Diffstat (limited to 'llvm/docs/tutorial')
0 files changed, 0 insertions, 0 deletions