-
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.Net.Http.Tests.SocketsHttpHandlerPerfTest #66343
Comments
Tagging subscribers to this area: @dotnet/ncl Issue DetailsRun Information
Regressions in System.Net.Http.Tests.SocketsHttpHandlerPerfTest
Reprogit clone https://github.com/dotnet/performance.git
py .\performance\scripts\benchmarks_ci.py -f net6.0 --filter 'System.Net.Http.Tests.SocketsHttpHandlerPerfTest*' PayloadsHistogramSystem.Net.Http.Tests.SocketsHttpHandlerPerfTest.Get(ssl: False, chunkedResponse: True, responseLength: 1)
Description of detection logic
Description of detection logic
Description of detection logic
DocsProfiling workflow for dotnet/runtime repository
|
Could be, will take a look |
We are also seeing a regression on Arm64 Run Information
Regressions in System.Net.Http.Tests.SocketsHttpHandlerPerfTest
Reprogit clone https://github.com/dotnet/performance.git
py .\performance\scripts\benchmarks_ci.py -f net6.0 --filter 'System.Net.Http.Tests.SocketsHttpHandlerPerfTest*' PayloadsHistogramSystem.Net.Http.Tests.SocketsHttpHandlerPerfTest.Get(ssl: False, chunkedResponse: False, responseLength: 1)
Description of detection logic
Description of detection logic
Description of detection logic
DocsProfiling workflow for dotnet/runtime repository |
@kouvel did you confirm if it is introduced by that PR? Or should we investigate? |
Same on Windows-x64 dotnet/perf-autofiling-issues#3987 |
I'm seeing some regressions from #64834 on my machine. The % diff I see is not as much as above, but I suspect that PR is the main cause. |
Tagging subscribers to this area: @mangod9 Issue DetailsRun Information
Regressions in System.Net.Http.Tests.SocketsHttpHandlerPerfTest
Reprogit clone https://github.com/dotnet/performance.git
py .\performance\scripts\benchmarks_ci.py -f net6.0 --filter 'System.Net.Http.Tests.SocketsHttpHandlerPerfTest*' PayloadsHistogramSystem.Net.Http.Tests.SocketsHttpHandlerPerfTest.Get(ssl: False, chunkedResponse: True, responseLength: 1)
Description of detection logic
Description of detection logic
Description of detection logic
DocsProfiling workflow for dotnet/runtime repository
|
There are some known tradeoffs in #64834 where there are few dedicated IO-polling threads like on Unixes, as it introduces an extra thread hop - the polling thread wakes and queues up an IO event, then a worker thread wakes and processes it. That seems to be the difference I'm seeing on my machine, as when
I'll go ahead and close this for now, as based on what is being measured this is an expected difference. |
More regressions here: dotnet/perf-autofiling-issues#3962 |
This regression has shown up on the 7 vs 6 perf report and seems to have persisted on some configs: https://pvscmdupload.blob.core.windows.net/reports/allTestHistory/refs/heads/main_x64_Windows%2010.0.18362/System.Net.Http.Tests.SocketsHttpHandlerPerfTest.Get_EnumerateHeaders_Validated(ssl%3a%20True%2c%20chunkedResponse%3a%20False%2c%20responseLength%3a%201).html. It seems like these regressions are "by design" based on above explanation. |
Run Information
Regressions in System.Net.Http.Tests.SocketsHttpHandlerPerfTest
Test Report
Repro
Payloads
Baseline
Compare
Histogram
System.Net.Http.Tests.SocketsHttpHandlerPerfTest.Get(ssl: False, chunkedResponse: True, responseLength: 1)
Description of detection logic
Description of detection logic
Description of detection logic
Docs
Profiling workflow for dotnet/runtime repository
Benchmarking workflow for dotnet/runtime repository
The text was updated successfully, but these errors were encountered: