bjope wrote: > @bjope It looks like the InstCombine changes enable IndVars to perform LFTR, > which is unprofitable in this case. Though the `umax(1)` call is actually > completely unnecessary here, but SCEV doesn't realize it. I've put up #75039 > to fix that. Does that improve things for you?
@nikic , thanks! I was thinking that maybe I would need to do something to please the SCEVExpander::isHighCostExpansionHelper to prevent the IndVar transform. But with #75039 I no longer see the problem umax instructions appearing (and increasing the IR instruction count). I still see some other regressions, but that could be due to limitations in the backend. I haven't found anything more that I can blame on this patch. https://github.com/llvm/llvm-project/pull/73662 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
