-
Notifications
You must be signed in to change notification settings - Fork 4.9k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Regressions in System.Collections.Sort #90933
Comments
Only win-x64-Intel machine regressed, suspects are #90325 cc @tannergooding and #90318 cc @jakobbotsch |
Tagging subscribers to this area: @JulieLeeMSFT, @jakobbotsch Issue DetailsRun Information
Regressions in System.Collections.Sort<IntStruct>
ReproGeneral Docs link: https://github.com/dotnet/performance/blob/main/docs/benchmarking-workflow-dotnet-runtime.md git clone https://github.com/dotnet/performance.git
py .\performance\scripts\benchmarks_ci.py -f net8.0 --filter 'System.Collections.Sort<IntStruct>*' PayloadsSystem.Collections.Sort<IntStruct>.Array(Size: 512)ETL FilesHistogramDescription of detection logic
JIT DisasmsSystem.Collections.Sort<IntStruct>.List(Size: 512)ETL FilesHistogramDescription of detection logic
JIT DisasmsDocsProfiling workflow for dotnet/runtime repository
|
There's a bunch of RA diffs but I don't really see why they would cause perf diffs, maybe JCC erratum? There's a few new instances of those (but some old instances also disappear) when I check with FWIW, the diffs seem to be triggered by #90496 kicking in @EgorBo: Seems like we shouldn't be creating new locals for this case if the previous logic didn't need to (but it's not the cause of the diffs, I guess the cause is the early folding away of the BB?)
|
@jakobbotsch thanks for the investigation!! Indeed seems like my PR introduced a bit more conservative side-effect extraction than previously, I'll check whether I can relax that or handle downstream but looks like only one machine regressed according to ADX query: so I am going to move this to 9.0 as non-important. |
I didn't blame my PR initially because I thought it's Tier0 only but you correctly pointed out it's not |
Looks like the regression was either fixed or compensated by #98324 |
Run Information
Regressions in System.Collections.Sort<IntStruct>
Test Report
Repro
General Docs link: https://github.com/dotnet/performance/blob/main/docs/benchmarking-workflow-dotnet-runtime.md
Payloads
Baseline
Compare
System.Collections.Sort<IntStruct>.Array(Size: 512)
ETL Files
Histogram
Description of detection logic
JIT Disasms
System.Collections.Sort<IntStruct>.List(Size: 512)
ETL Files
Histogram
Description of detection logic
JIT Disasms
Docs
Profiling workflow for dotnet/runtime repository
Benchmarking workflow for dotnet/runtime repository
The text was updated successfully, but these errors were encountered: