[PATCH] D102615: [LoopDeletion] Break backedge if we can prove that the loop is exited on 1st iteration
Max Kazantsev via Phabricator via llvm-commits
llvm-commits at lists.llvm.org
Wed May 26 02:13:44 PDT 2021
mkazantsev added a comment.
@nikic I already regret that I spent time improving implication engine of SCEV... :)
Actually, thinking a bit more about it, the "at" part wasn't needed for my initial motivating example. So I can rework this tests in a way it doesn't need "at".
However, this tendency is worrysome. We intentionally give up the optimization power trying to sustain compile time, instead of solving the problem with SCEV caching once and for all. Maybe this is what we should think about. If even LoopDeletion can cause compile time regression, I refuse to understand what doesn't. :)
Repository:
rG LLVM Github Monorepo
CHANGES SINCE LAST ACTION
https://reviews.llvm.org/D102615/new/
https://reviews.llvm.org/D102615
More information about the llvm-commits
mailing list