[Mlir-commits] [mlir] [mlir][x86vector] AVX512-BF16 Convert packed F32 to BF16 (PR #125685)

[Renato Golin]

https://github.com/rengolin approved this pull request.

Same as before. Not the best design, but following the existing pattern. This is helping us understand the design and will be target of an RFC soon on how to simplify CPU dialect lowering without needing a full blown VM dialect.

https://github.com/llvm/llvm-project/pull/125685