[llvm-branch-commits] [llvm] [CodeGen][NPM] Remove "LowerConstantIntrinsicsPass" from the pipeline (PR #172794)

[Aiden Grossman via llvm-branch-commits]

https://github.com/boomanaiden154 approved this pull request.

LGTM given #97727.

I'm kind of thinking we should have a more principled story around `LowerConstantIntrinsics` though. It also exists in the optimization pipeline fairly late, but before an instance of the function simplification pipeline. This shows up in a couple of test cases (`memcpy-constant-size.ll`). But I don't have any brilliant ideas on how to preserve the small performance wins there while pulling the constant intrinsic lowering out of `preiselintrinsiclowering`. Right now it seems like we're also running `LowerConstantIntrinsics` twice if we're using any of the optimization pipelines.

CC @aengelke 

https://github.com/llvm/llvm-project/pull/172794