[PATCH] D142093: [InstCombine] trunc (fptoui|fptosi)

[Sanjay Patel via Phabricator via llvm-commits]

spatel added reviewers: kpn, efriedma.
spatel added a comment.

I realize it's unlikely in practice, but is there a reason not to support any FP type? For fptoui, the integer type just needs one more bit than the max exponent for a given FP semantic?

For the tests, it should be sufficient to have the intermediate integer width be one more than the minimum required type width, so "%i = fptoui half %x to i17".

Please pre-commit baseline tests (either locally or push to main) and label tests that should not change as negative tests (either in the function name or with a code comment).


CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D142093/new/

https://reviews.llvm.org/D142093