Stream: t-compiler/wg-llvm

Topic: Inlining


Matthew Jasper (Oct 13 2019 at 17:18, on Zulip):

I've opened #65385 which tries to avoid some of the perf regression of #61430. Some of which was caused by llvm inlining a function (marked #[inline]) beyond the point where it made any sense. Does anyone have any ideas on how to avoid this sort of issue, or how I can debug llvm's inlining decisions?

gnzlbg (Oct 14 2019 at 10:08, on Zulip):

Does that perf regression also happens without LTO ?

Last update: Nov 15 2019 at 09:40UTC