[Mlir-commits] [mlir] [mlir][NVVM] Add InferTypeOpInterface to NVVM ops with deterministic result types (PR #188173)

[Guray Ozen]

grypp wrote:

> Could you please run the GPU/CUDA integration tests locally and confirm if they are still good after this change? I ask because we do not have coverage in CI for these yet and fixing the issues after a while becomes tedious at times.

Also take a look at the NVDSL examples. 

https://github.com/llvm/llvm-project/pull/188173