[PATCH] D126158: [MLIR][GPU] Replace fdiv on fp16 with promoted (fp32) multiplication with reciprocal plus one (conditional) Newton iteration.

[Artem Belevich via Phabricator via llvm-commits]

tra added a comment.

I would suggest separating it into separate LLVM and MLIR patches.

LLVM changes look OK to me. No idea about MLIR. we would probably want to lower fp16 fdiv the same way in LLVM, too, but that would also have to be a separate patch.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D126158/new/

https://reviews.llvm.org/D126158