[PATCH] D125588: [MachineCombiner] Improve MachineCombiner's cost model

Guozhi Wei via Phabricator via llvm-commits llvm-commits at lists.llvm.org
Tue Jun 21 18:58:44 PDT 2022


Carrot added a comment.

It sounds to me like the MADD is decoded into two uops, mul and add. Then the mul can be immediately executable once the two multipliers are available. Is this correct?

I agree in this case MADD is always preferred. But the problem is current MachineCombiner and latency computation doesn't model this behavior. It's an independent issue. I think it should be solved in another patch.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D125588/new/

https://reviews.llvm.org/D125588



More information about the llvm-commits mailing list