Skip to content
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.

Sign up for GitHub

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

Jump to bottom

[Perf] Linux/arm64: 6 Regressions on 8/8/2023 7:20:25 PM #20608

Closed
performanceautofiler bot opened this issue Aug 17, 2023 · 3 comments
Closed

[Perf] Linux/arm64: 6 Regressions on 8/8/2023 7:20:25 PM #20608

performanceautofiler bot opened this issue Aug 17, 2023 · 3 comments
Labels
ampere arch-arm64 branch-refs/heads/main kind-micro os-linux perf-regression runtime-coreclr

Comments

@performanceautofiler
Copy link

performanceautofiler bot commented Aug 17, 2023 •
edited
Loading

Run Information

Name Value
Architecture arm64
OS ubuntu 20.04
Queue AmpereUbuntu
Baseline e857642fd03f8a6acdf42af7d9a814e320ec8432
Compare 88633ae045e7741fffa17710dc48e9032e519258
Diff Diff
Configs CompilationMode:tiered, RunKind:micro

Regressions in System.Memory.Span<Char>

Benchmark Baseline Test Test/Base Test Quality Edge Detector Baseline IR Compare IR IR Ratio
27.67 ns 34.10 ns 1.23 0.02 True

graph
Test Report

Repro

General Docs link: https://github.com/dotnet/performance/blob/main/docs/benchmarking-workflow-dotnet-runtime.md

git clone https://github.com/dotnet/performance.git
python3 .\performance\scripts\benchmarks_ci.py -f net8.0 --filter 'System.Memory.Span&lt;Char&gt;*'

Payloads

Baseline
Compare

System.Memory.Span<Char>.IndexOfValue(Size: 512)

ETL Files

Histogram

Description of detection logic

IsRegressionBase: Marked as regression because the compare was 5% greater than the baseline, and the value was not too small.
IsRegressionChecked: Marked as regression because the three check build points were 0.05 greater than the baseline.
IsImprovementBase: Marked as not an improvement because the compare was not 5% less than the baseline, or the value was too small.
IsRegressionBase: Marked as regression because the compare was 5% greater than the baseline, and the value was not too small.
IsRegressionChecked: Marked as regression because the three check build points were 0.05 greater than the baseline.
IsRegressionWindowed: Marked as regression because 34.09713285178929 > 29.216196220310195.
IsChangePoint: Marked as a change because one of 4/6/2023 11:23:56 PM, 8/8/2023 8:10:39 AM, 8/16/2023 6:53:02 PM falls between 8/7/2023 10:39:59 PM and 8/16/2023 6:53:02 PM.
IsRegressionStdDev: Marked as regression because -18.19892466119765 (T) = (0 -33.671609148303006) / Math.Sqrt((0.49036175073419497 / (6)) + (0.25602489480162904 / (14))) is less than -2.100922040224263 = MathNet.Numerics.Distributions.StudentT.InvCDF(0, 1, (6) + (14) - 2, .025) and -0.20616787259466696 = (27.916188047580636 - 33.671609148303006) / 27.916188047580636 is less than -0.05.
IsImprovementBase: Marked as not an improvement because the compare was not 5% less than the baseline, or the value was too small.
IsChangeEdgeDetector: Marked as regression because Edge Detector said so.

JIT Disasms

Docs

Profiling workflow for dotnet/runtime repository
Benchmarking workflow for dotnet/runtime repository

Run Information

Name Value
Architecture arm64
OS ubuntu 20.04
Queue AmpereUbuntu
Baseline e857642fd03f8a6acdf42af7d9a814e320ec8432
Compare 88633ae045e7741fffa17710dc48e9032e519258
Diff Diff
Configs CompilationMode:tiered, RunKind:micro

Regressions in System.Tests.Perf_Int32

Benchmark Baseline Test Test/Base Test Quality Edge Detector Baseline IR Compare IR IR Ratio
21.18 ns 23.84 ns 1.13 0.25 False

graph
Test Report

Repro

General Docs link: https://github.com/dotnet/performance/blob/main/docs/benchmarking-workflow-dotnet-runtime.md

git clone https://github.com/dotnet/performance.git
python3 .\performance\scripts\benchmarks_ci.py -f net8.0 --filter 'System.Tests.Perf_Int32*'

Payloads

Baseline
Compare

System.Tests.Perf_Int32.ToStringHex(value: -2147483648)

ETL Files

Histogram

Description of detection logic

IsRegressionBase: Marked as regression because the compare was 5% greater than the baseline, and the value was not too small.
IsRegressionChecked: Marked as regression because the three check build points were 0.05 greater than the baseline.
IsImprovementBase: Marked as not an improvement because the compare was not 5% less than the baseline, or the value was too small.
IsRegressionBase: Marked as regression because the compare was 5% greater than the baseline, and the value was not too small.
IsRegressionChecked: Marked as regression because the three check build points were 0.05 greater than the baseline.
IsRegressionWindowed: Marked as regression because 23.843453200002333 > 22.486947409526284.
IsChangePoint: Marked as a change because one of 4/12/2023 8:39:35 PM, 4/25/2023 6:46:56 PM, 5/19/2023 9:46:56 PM, 8/8/2023 8:10:39 AM, 8/16/2023 6:53:02 PM falls between 8/7/2023 10:39:59 PM and 8/16/2023 6:53:02 PM.
IsRegressionStdDev: Marked as regression because -8.802195343931237 (T) = (0 -24.360574077034883) / Math.Sqrt((0.48783129688978255 / (6)) + (0.21101071458223963 / (12))) is less than -2.1199052992212764 = MathNet.Numerics.Distributions.StudentT.InvCDF(0, 1, (6) + (12) - 2, .025) and -0.12819217995540996 = (21.592574837735267 - 24.360574077034883) / 21.592574837735267 is less than -0.05.
IsImprovementBase: Marked as not an improvement because the compare was not 5% less than the baseline, or the value was too small.
IsChangeEdgeDetector: Marked not as a regression because Edge Detector said so.

JIT Disasms

Docs

Profiling workflow for dotnet/runtime repository
Benchmarking workflow for dotnet/runtime repository

Run Information

Name Value
Architecture arm64
OS ubuntu 20.04
Queue AmpereUbuntu
Baseline e857642fd03f8a6acdf42af7d9a814e320ec8432
Compare 88633ae045e7741fffa17710dc48e9032e519258
Diff Diff
Configs CompilationMode:tiered, RunKind:micro

Regressions in System.Tests.Perf_DateTimeOffset

Benchmark Baseline Test Test/Base Test Quality Edge Detector Baseline IR Compare IR IR Ratio
44.38 ns 47.44 ns 1.07 0.05 False

graph
Test Report

Repro

General Docs link: https://github.com/dotnet/performance/blob/main/docs/benchmarking-workflow-dotnet-runtime.md

git clone https://github.com/dotnet/performance.git
python3 .\performance\scripts\benchmarks_ci.py -f net8.0 --filter 'System.Tests.Perf_DateTimeOffset*'

Payloads

Baseline
Compare

System.Tests.Perf_DateTimeOffset.ToString(format: "r")

ETL Files

Histogram

Description of detection logic

IsRegressionBase: Marked as regression because the compare was 5% greater than the baseline, and the value was not too small.
IsRegressionChecked: Marked as regression because the three check build points were 0.05 greater than the baseline.
IsImprovementBase: Marked as not an improvement because the compare was not 5% less than the baseline, or the value was too small.
IsRegressionBase: Marked as regression because the compare was 5% greater than the baseline, and the value was not too small.
IsRegressionChecked: Marked as regression because the three check build points were 0.05 greater than the baseline.
IsRegressionWindowed: Marked as regression because 47.44281585068164 > 45.75146968913372.
IsChangePoint: Marked as a change because one of 5/19/2023 4:22:12 AM, 8/8/2023 8:10:39 AM, 8/16/2023 6:53:02 PM falls between 8/7/2023 10:39:59 PM and 8/16/2023 6:53:02 PM.
IsRegressionStdDev: Marked as regression because -8.105692202865033 (T) = (0 -47.081586582009436) / Math.Sqrt((0.42581390567289124 / (6)) + (1.4052599605062253 / (14))) is less than -2.100922040224263 = MathNet.Numerics.Distributions.StudentT.InvCDF(0, 1, (6) + (14) - 2, .025) and -0.07673303742918847 = (43.726332289776856 - 47.081586582009436) / 43.726332289776856 is less than -0.05.
IsImprovementBase: Marked as not an improvement because the compare was not 5% less than the baseline, or the value was too small.
IsChangeEdgeDetector: Marked not as a regression because Edge Detector said so.

JIT Disasms

Docs

Profiling workflow for dotnet/runtime repository
Benchmarking workflow for dotnet/runtime repository

Run Information

Name Value
Architecture arm64
OS ubuntu 20.04
Queue AmpereUbuntu
Baseline e857642fd03f8a6acdf42af7d9a814e320ec8432
Compare 88633ae045e7741fffa17710dc48e9032e519258
Diff Diff
Configs CompilationMode:tiered, RunKind:micro

Regressions in System.Memory.ReadOnlySpan

Benchmark Baseline Test Test/Base Test Quality Edge Detector Baseline IR Compare IR IR Ratio
57.42 ns 69.50 ns 1.21 0.00 True
58.25 ns 69.51 ns 1.19 0.01 True

graph
Test Report

Repro

General Docs link: https://github.com/dotnet/performance/blob/main/docs/benchmarking-workflow-dotnet-runtime.md

git clone https://github.com/dotnet/performance.git
python3 .\performance\scripts\benchmarks_ci.py -f net8.0 --filter 'System.Memory.ReadOnlySpan*'

Payloads

Baseline
Compare

System.Memory.ReadOnlySpan.IndexOfString(input: "だだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだxだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだだ", value: "x", comparisonType: Ordinal)

ETL Files

Histogram

Description of detection logic

IsRegressionBase: Marked as regression because the compare was 5% greater than the baseline, and the value was not too small.
IsRegressionChecked: Marked as regression because the three check build points were 0.05 greater than the baseline.
IsImprovementBase: Marked as not an improvement because the compare was not 5% less than the baseline, or the value was too small.
IsRegressionBase: Marked as regression because the compare was 5% greater than the baseline, and the value was not too small.
IsRegressionChecked: Marked as regression because the three check build points were 0.05 greater than the baseline.
IsRegressionWindowed: Marked as regression because 69.50080772107812 > 60.331577144002786.
IsChangePoint: Marked as a change because one of 8/8/2023 8:10:39 AM, 8/16/2023 6:53:02 PM falls between 8/7/2023 10:39:59 PM and 8/16/2023 6:53:02 PM.
IsRegressionStdDev: Marked as regression because -145.34814893982872 (T) = (0 -69.51865422249324) / Math.Sqrt((0.010184047218936938 / (6)) + (0.06763045882450479 / (13))) is less than -2.109815577813699 = MathNet.Numerics.Distributions.StudentT.InvCDF(0, 1, (6) + (13) - 2, .025) and -0.21016907105246393 = (57.44540650178245 - 69.51865422249324) / 57.44540650178245 is less than -0.05.
IsImprovementBase: Marked as not an improvement because the compare was not 5% less than the baseline, or the value was too small.
IsChangeEdgeDetector: Marked as regression because Edge Detector said so.

JIT Disasms

System.Memory.ReadOnlySpan.IndexOfString(input: "AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAXAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA", value: "X", comparisonType: Ordinal)

ETL Files

Histogram

Description of detection logic

IsRegressionBase: Marked as regression because the compare was 5% greater than the baseline, and the value was not too small.
IsRegressionChecked: Marked as regression because the three check build points were 0.05 greater than the baseline.
IsImprovementBase: Marked as not an improvement because the compare was not 5% less than the baseline, or the value was too small.
IsRegressionBase: Marked as regression because the compare was 5% greater than the baseline, and the value was not too small.
IsRegressionChecked: Marked as regression because the three check build points were 0.05 greater than the baseline.
IsRegressionWindowed: Marked as regression because 69.50551808948008 > 60.42229249814034.
IsChangePoint: Marked as a change because one of 8/8/2023 8:10:39 AM, 8/16/2023 6:53:02 PM falls between 8/7/2023 10:39:59 PM and 8/16/2023 6:53:02 PM.
IsRegressionStdDev: Marked as regression because -75.57129858597224 (T) = (0 -69.47615295846397) / Math.Sqrt((0.14900991017497361 / (6)) + (0.0027130841379520753 / (14))) is less than -2.100922040224263 = MathNet.Numerics.Distributions.StudentT.InvCDF(0, 1, (6) + (14) - 2, .025) and -0.20785225995584988 = (57.520406478358126 - 69.47615295846397) / 57.520406478358126 is less than -0.05.
IsImprovementBase: Marked as not an improvement because the compare was not 5% less than the baseline, or the value was too small.
IsChangeEdgeDetector: Marked as regression because Edge Detector said so.

JIT Disasms

Docs

Profiling workflow for dotnet/runtime repository
Benchmarking workflow for dotnet/runtime repository

Run Information

Name Value
Architecture arm64
OS ubuntu 20.04
Queue AmpereUbuntu
Baseline e857642fd03f8a6acdf42af7d9a814e320ec8432
Compare 88633ae045e7741fffa17710dc48e9032e519258
Diff Diff
Configs CompilationMode:tiered, RunKind:micro

Regressions in System.Text.RegularExpressions.Tests.Perf_Regex_Industry_RustLang_Sherlock

Benchmark Baseline Test Test/Base Test Quality Edge Detector Baseline IR Compare IR IR Ratio
8.22 ms 10.92 ms 1.33 0.07 False

graph
Test Report

Repro

General Docs link: https://github.com/dotnet/performance/blob/main/docs/benchmarking-workflow-dotnet-runtime.md

git clone https://github.com/dotnet/performance.git
python3 .\performance\scripts\benchmarks_ci.py -f net8.0 --filter 'System.Text.RegularExpressions.Tests.Perf_Regex_Industry_RustLang_Sherlock*'

Payloads

Baseline
Compare

System.Text.RegularExpressions.Tests.Perf_Regex_Industry_RustLang_Sherlock.Count(Pattern: "\b\w+n\b", Options: NonBacktracking)

ETL Files

Histogram

Description of detection logic

IsRegressionBase: Marked as regression because the compare was 5% greater than the baseline, and the value was not too small.
IsRegressionChecked: Marked as regression because the three check build points were 0.05 greater than the baseline.
IsImprovementBase: Marked as not an improvement because the compare was not 5% less than the baseline, or the value was too small.
IsRegressionBase: Marked as regression because the compare was 5% greater than the baseline, and the value was not too small.
IsRegressionChecked: Marked as regression because the three check build points were 0.05 greater than the baseline.
IsRegressionWindowed: Marked as regression because 10.9249862 > 8.481155414752688.
IsChangePoint: Marked as a change because one of 4/12/2023 8:39:35 PM, 4/25/2023 6:46:56 PM, 8/8/2023 8:10:39 AM, 8/16/2023 6:53:02 PM falls between 8/7/2023 10:39:59 PM and 8/16/2023 6:53:02 PM.
IsRegressionStdDev: Marked as regression because -53.62409757774009 (T) = (0 -10973422.664423078) / Math.Sqrt((16395116994.142904 / (6)) + (3350547550.2124906 / (13))) is less than -2.109815577813699 = MathNet.Numerics.Distributions.StudentT.InvCDF(0, 1, (6) + (13) - 2, .025) and -0.3646694629921251 = (8041084.645041552 - 10973422.664423078) / 8041084.645041552 is less than -0.05.
IsImprovementBase: Marked as not an improvement because the compare was not 5% less than the baseline, or the value was too small.
IsChangeEdgeDetector: Marked not as a regression because Edge Detector said so.

JIT Disasms

Docs

Profiling workflow for dotnet/runtime repository
Benchmarking workflow for dotnet/runtime repository
@AndyAyersMS
Copy link
Member

Suspect this may be dotnet/runtime#87438 enabling new static PGO data to be pulled in.

@AndyAyersMS AndyAyersMS mentioned this issue Aug 17, 2023
Closed
@AndyAyersMS
Copy link
Member

Looks like the first test regressed across multiple configs

image

Also the IndexOfString("AAAA..."

image

@EgorBo
Copy link
Member

EgorBo commented Sep 14, 2023

Seems to be R2R sensitive, dotnet/runtime#87438

@EgorBo EgorBo closed this as completed Sep 14, 2023
@EgorBo EgorBo removed the Look Again label Sep 14, 2023
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Assignees
No one assigned
Labels
ampere arch-arm64 branch-refs/heads/main kind-micro os-linux perf-regression runtime-coreclr
Projects
None yet
Milestone
No milestone
Development

No branches or pull requests
2 participants
@EgorBo @AndyAyersMS