[PATCH] D76194: [BasicAA] Use known lower bounds for index values for size based check.

[Florian Hahn via Phabricator via llvm-commits]

fhahn added a comment.

In D76194#2065065 <https://reviews.llvm.org/D76194#2065065>, @nikic wrote:

> This change caused a 0.5% compile-time regression <http://llvm-compile-time-tracker.com/compare.php?from=e6aba43cda848f4a8cfa5ce9f174b77def10e9df&to=d99a1848c4f8ca164c0c0768e10eafc850b2a68a&stat=instructions>. The largest regression is 1.6% for lencod (with the q_matrix.c file at 3-4%). As the code-size changes are <= 0.01%, this optimization does not seem to trigger much in practice (and not at all for lencod). This seems like a somewhat unfavorable trade-off between compile-time and optimization power.


Hm yeah that does not look great. I'll check to see if there's something we can do to limit the amount of extra work.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D76194/new/

https://reviews.llvm.org/D76194