[PATCH] D51276: [New PM][PassTiming] implement -time-passes for the new pass manager

Yuanfang Chen via Phabricator via llvm-commits llvm-commits at lists.llvm.org
Mon Nov 23 15:27:54 PST 2020

ychen added a comment.

In D51276#1676924 <https://reviews.llvm.org/D51276#1676924>, @tejohnson wrote:

> In D51276#1254958 <https://reviews.llvm.org/D51276#1254958>, @chandlerc wrote:
>> In D51276#1254913 <https://reviews.llvm.org/D51276#1254913>, @fedor.sergeev wrote:
>>> as for the per-pass-name aggregation - I was going to supply a follow-up patch that changes -time-passes from boolean option to -time-passes=<aggregation-type>.
>>> Perhaps aggregation-type should be 'aggregate' by default, and 'invocations' providing per-invocation info.
>>> Yet I dont want to just remove the possibility of getting a more fine-grained data that we have in this patch.
>> Yeah, i definitely like having the per-invocation information. I'm just suggesting that (eventually, happy for this to come in follow-ups) we'll probably want the default to be aggregated a bit.
> Ping on getting this aggregation. I just tried running -ftime-report on a large file, and got >2 million lines of output, with 41K lines for just the InlinerPass alone. This isn't particularly helpful when trying to understand overall impacts of changes on various passes.

Observed the same thing recently. I'll work on a fix.




More information about the llvm-commits mailing list