forked from OSchip/llvm-project
f7b9ac8245
Currently, we use different names for block and grid dimensions in the GPU vs. NVVM dialect, which is confusing. With this change, the NVVM names match the ones from the GPU dialect. The NVVM intrinsics use different names anyway, so it does not add new confusion. PiperOrigin-RevId: 251389281 |
||
---|---|---|
mlir |