[llvm] [SCEV] Collect and merge loop guards through PHI nodes with multiple incoming values (PR #113915)

[Nikita Popov via llvm-commits]

nikic wrote:

> One surprising thing is that there's a notable change in compile-time in `stage1-O0-g`, but nothing in the O0 pipeline should use the loop guards AFAIK. @nikic do you have any idea what could be the cause for the changes there?

This is very weird. No idea what is happening there without profiling it. I did another run with this patch to see whether this might just be very unlucky noise, but that doesn't seem to be the case: https://llvm-compile-time-tracker.com/compare.php?from=dd116369f646d023a2e7e5c145a1bed5dcf9a45c&to=83ad7da76ee34ada80255cc4e39f99d2afd9e8d3&stat=instructions:u

https://github.com/llvm/llvm-project/pull/113915