search for: lazyblockfrequencyinfo

...has profile summary or depend on the PSI.) With this, we > wouldn’t need to conditionally build pass pipelines and may work for the > new pass manager. But a similar wouldn’t work for the old pass manager > because we cannot conditionally depend on an analysis under it. > > There is LazyBlockFrequencyInfo. > Not sure how well it fits this idea. > Good point. LazyBlockFrequencyInfo seems usable with the old pass manager (save unnecessary BFI/BPI) and would work for function passes. I think the restriction still applies - a loop pass cannot still request (outer-scope) BFI, lazy or not, new or o...

...to conditionally build pass pipelines and >>> may work for the new pass manager. But a similar >>> wouldn’t work for the old pass manager because we cannot >>> conditionally depend on an analysis under it. >> There is LazyBlockFrequencyInfo. >> Not sure how well it fits this idea. >> >> >> Good point. LazyBlockFrequencyInfo seems usable with the old >> pass manager (save unnecessary BFI/BPI) and would work for >> function passes. I think the restriction still appl...

...and make them preserve BFI > through. This is to avoid pipeline interruptions and potential > invalidation/recomputation of BFI between the loop passes. We detect > PGO based on whether PSI has profile summary info. (For the old pass > manager, implement a similar approach by using LazyBlockFrequencyInfo.) There is already an optional analysis in LoopStandardAnalysisResults - MemorySSA. We can do the same for BFI/BPI. And, yes - preserving those through loop passes is a cornerstone to this approach. > > - Add a new proxy ModuleAnalysisManagerLoopProxy for a loop pass to be > able to get...

...or depend on the PSI.) With this, we >> wouldn’t need to conditionally build pass pipelines and may work for the >> new pass manager. But a similar wouldn’t work for the old pass manager >> because we cannot conditionally depend on an analysis under it. >> >> There is LazyBlockFrequencyInfo. >> Not sure how well it fits this idea. >> > > Good point. LazyBlockFrequencyInfo seems usable with the old pass manager > (save unnecessary BFI/BPI) and would work for function passes. I think t > he restriction still applies - a loop pass cannot still request > (outer-...

...pass >>>>> manager. But a similar wouldn’t work for the >>>>> old pass manager because we cannot >>>>> conditionally depend on an analysis under it. >>>> There is LazyBlockFrequencyInfo. >>>> Not sure how well it fits this idea. >>>> >>>> >>>> Good point. LazyBlockFrequencyInfo seems usable >>>> with the old pass manager (save unnecessary >>>>...

Hi all, To implement more profile-guided optimizations, we’d like to use ProfileSummaryInfo (PSI) and BlockFrequencyInfo (BFI) from more passes of various types, under the new pass manager. The following is what we came up with. Would appreciate feedback. Thanks. Issue It’s not obvious (to me) how to best do this, given that we cannot request an outer-scope analysis result from an inner-scope