[llvm] [InstCombine] handle trunc to i1 in foldLogOpOfMaskedICmps. (PR #128861)

Andreas Jonson via llvm-commits llvm-commits at lists.llvm.org
Wed Feb 26 03:17:19 PST 2025


andjo403 wrote:

what I can see many of the regressions in llvm-opt-benchmark is that `icmp ne (and x, 1), 0` is not folded to `trunc x to i1` but the fold from this PR is one of the reasons for regressions for that fold.

https://github.com/llvm/llvm-project/pull/128861


More information about the llvm-commits mailing list