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.

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

[Perf] Windows/arm64: 5 Improvements on 9/29/2023 1:50:32 PM #22705

Closed
performanceautofiler bot opened this issue Oct 5, 2023 · 1 comment
Closed

[Perf] Windows/arm64: 5 Improvements on 9/29/2023 1:50:32 PM #22705

performanceautofiler bot opened this issue Oct 5, 2023 · 1 comment
Labels
arch-arm64 branch-refs/heads/main kind-micro os-windows perf-improvement PGO Applied if there were any profile guided optimization updates in the observed interval. runtime-coreclr

Comments

@performanceautofiler
Copy link

performanceautofiler bot commented Oct 5, 2023

Run Information

Name Value
Architecture arm64
OS Windows 10.0.19041
Queue SurfaceWindows
Baseline e500dad300c739a35eb88edb3605f71eca18a4ac
Compare cc072997b68b2412c0f7667737b0619f4f04241d
Diff Diff
Configs CompilationMode:tiered, RunKind:micro

Improvements in Benchstone.MDBenchI.MDMulMatrix

Benchmark Baseline Test Test/Base Test Quality Edge Detector Baseline IR Compare IR IR Ratio
864.12 ms 767.01 ms 0.89 0.01 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
py .\performance\scripts\benchmarks_ci.py -f net8.0 --filter 'Benchstone.MDBenchI.MDMulMatrix*'

Payloads

Baseline
Compare

Benchstone.MDBenchI.MDMulMatrix.Test

ETL Files

Histogram

Description of detection logic

IsRegressionBase: Marked as not a regression because the compare was not 5% greater than the baseline, or the value was too small.
IsImprovementBase: Marked as improvement because the compare was 5% less than the baseline, and the value was not too small.
IsImprovementCheck: Marked as improvement because the three check build points were 0.05 less than the baseline.
IsRegressionBase: Marked as not a regression because the compare was not 5% greater than the baseline, or the value was too small.
IsImprovementBase: Marked as improvement because the compare was 5% less than the baseline, and the value was not too small.
IsImprovementCheck: Marked as improvement because the three check build points were 0.05 less than the baseline.
IsImprovementWindowed: Marked as improvement because we could not find enough baseline builds for window checking.
IsChangePoint: Marked as a change because one of 5/19/2023 4:22:12 AM, 5/26/2023 9:55:14 AM, 9/28/2023 11:34:37 PM, 10/4/2023 1:30:06 AM falls between 9/25/2023 11:21:57 AM and 10/4/2023 1:30:06 AM.
IsImprovementStdDev: Marked as improvement because 59.83473080767967 (T) = (0 -765699225.4761906) / Math.Sqrt((1034816784797317.1 / (299)) + (609128529647.6758 / (6))) is greater than 1.967824097752068 = MathNet.Numerics.Distributions.StudentT.InvCDF(0, 1, (299) + (6) - 2, .975) and 0.12853445703434066 = (878633965.1140555 - 765699225.4761906) / 878633965.1140555 is greater than 0.05.
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 Windows 10.0.19041
Queue SurfaceWindows
Baseline e500dad300c739a35eb88edb3605f71eca18a4ac
Compare cc072997b68b2412c0f7667737b0619f4f04241d
Diff Diff
Configs CompilationMode:tiered, RunKind:micro

Improvements in Benchstone.MDBenchI.MDGeneralArray

Benchmark Baseline Test Test/Base Test Quality Edge Detector Baseline IR Compare IR IR Ratio
19.83 ms 17.08 ms 0.86 0.00 False
19.49 ms 17.12 ms 0.88 0.00 False

graph
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
py .\performance\scripts\benchmarks_ci.py -f net8.0 --filter 'Benchstone.MDBenchI.MDGeneralArray*'

Payloads

Baseline
Compare

Benchstone.MDBenchI.MDGeneralArray.Test2

ETL Files

Histogram

Description of detection logic

IsRegressionBase: Marked as not a regression because the compare was not 5% greater than the baseline, or the value was too small.
IsImprovementBase: Marked as improvement because the compare was 5% less than the baseline, and the value was not too small.
IsImprovementCheck: Marked as improvement because the three check build points were 0.05 less than the baseline.
IsRegressionBase: Marked as not a regression because the compare was not 5% greater than the baseline, or the value was too small.
IsImprovementBase: Marked as improvement because the compare was 5% less than the baseline, and the value was not too small.
IsImprovementCheck: Marked as improvement because the three check build points were 0.05 less than the baseline.
IsImprovementWindowed: Marked as improvement because we could not find enough baseline builds for window checking.
IsChangePoint: Marked as a change because one of 5/19/2023 4:22:12 AM, 9/28/2023 11:34:37 PM, 10/4/2023 1:30:06 AM falls between 9/25/2023 11:21:57 AM and 10/4/2023 1:30:06 AM.
IsImprovementStdDev: Marked as improvement because 136.19311959860167 (T) = (0 -17118156.037414964) / Math.Sqrt((49808219493.3453 / (299)) + (840326518.6440237 / (6))) is greater than 1.967824097752068 = MathNet.Numerics.Distributions.StudentT.InvCDF(0, 1, (299) + (6) - 2, .975) and 0.12228274228950332 = (19503041.425968133 - 17118156.037414964) / 19503041.425968133 is greater than 0.05.
IsChangeEdgeDetector: Marked not as a regression because Edge Detector said so.

JIT Disasms

Benchstone.MDBenchI.MDGeneralArray.Test

ETL Files

Histogram

Description of detection logic

IsRegressionBase: Marked as not a regression because the compare was not 5% greater than the baseline, or the value was too small.
IsImprovementBase: Marked as improvement because the compare was 5% less than the baseline, and the value was not too small.
IsImprovementCheck: Marked as improvement because the three check build points were 0.05 less than the baseline.
IsRegressionBase: Marked as not a regression because the compare was not 5% greater than the baseline, or the value was too small.
IsImprovementBase: Marked as improvement because the compare was 5% less than the baseline, and the value was not too small.
IsImprovementCheck: Marked as improvement because the three check build points were 0.05 less than the baseline.
IsImprovementWindowed: Marked as improvement because we could not find enough baseline builds for window checking.
IsChangePoint: Marked as a change because one of 5/19/2023 4:22:12 AM, 9/28/2023 11:34:37 PM, 10/4/2023 1:30:06 AM falls between 9/25/2023 11:21:57 AM and 10/4/2023 1:30:06 AM.
IsImprovementStdDev: Marked as improvement because 161.3844481419925 (T) = (0 -17136012.81746032) / Math.Sqrt((44554998569.195694 / (299)) + (405716131.93686694 / (6))) is greater than 1.967824097752068 = MathNet.Numerics.Distributions.StudentT.InvCDF(0, 1, (299) + (6) - 2, .975) and 0.12174092174057617 = (19511341.518292412 - 17136012.81746032) / 19511341.518292412 is greater than 0.05.
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 Windows 10.0.19041
Queue SurfaceWindows
Baseline e500dad300c739a35eb88edb3605f71eca18a4ac
Compare cc072997b68b2412c0f7667737b0619f4f04241d
Diff Diff
Configs CompilationMode:tiered, RunKind:micro

Improvements in Benchstone.MDBenchI.MDXposMatrix

Benchmark Baseline Test Test/Base Test Quality Edge Detector Baseline IR Compare IR IR Ratio
44.93 μs 41.42 μs 0.92 0.00 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
py .\performance\scripts\benchmarks_ci.py -f net8.0 --filter 'Benchstone.MDBenchI.MDXposMatrix*'

Payloads

Baseline
Compare

Benchstone.MDBenchI.MDXposMatrix.Test

ETL Files

Histogram

Description of detection logic

IsRegressionBase: Marked as not a regression because the compare was not 5% greater than the baseline, or the value was too small.
IsImprovementBase: Marked as improvement because the compare was 5% less than the baseline, and the value was not too small.
IsImprovementCheck: Marked as improvement because the three check build points were 0.05 less than the baseline.
IsRegressionBase: Marked as not a regression because the compare was not 5% greater than the baseline, or the value was too small.
IsImprovementBase: Marked as improvement because the compare was 5% less than the baseline, and the value was not too small.
IsImprovementCheck: Marked as improvement because the three check build points were 0.05 less than the baseline.
IsImprovementWindowed: Marked as improvement because we could not find enough baseline builds for window checking.
IsChangePoint: Marked as a change because one of 9/28/2023 11:34:37 PM, 10/4/2023 1:30:06 AM falls between 9/25/2023 11:21:57 AM and 10/4/2023 1:30:06 AM.
IsImprovementStdDev: Marked as improvement because 202.31111554292247 (T) = (0 -41446.34695795105) / Math.Sqrt((12569.892553917614 / (299)) + (1672.142223478621 / (6))) is greater than 1.967824097752068 = MathNet.Numerics.Distributions.StudentT.InvCDF(0, 1, (299) + (6) - 2, .975) and 0.08039085680868933 = (45069.524661444964 - 41446.34695795105) / 45069.524661444964 is greater than 0.05.
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 Windows 10.0.19041
Queue SurfaceWindows
Baseline e500dad300c739a35eb88edb3605f71eca18a4ac
Compare cc072997b68b2412c0f7667737b0619f4f04241d
Diff Diff
Configs CompilationMode:tiered, RunKind:micro

Improvements in Benchstone.MDBenchI.MDArray2

Benchmark Baseline Test Test/Base Test Quality Edge Detector Baseline IR Compare IR IR Ratio
1.94 secs 1.71 secs 0.88 0.00 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
py .\performance\scripts\benchmarks_ci.py -f net8.0 --filter 'Benchstone.MDBenchI.MDArray2*'

Payloads

Baseline
Compare

Benchstone.MDBenchI.MDArray2.Test

ETL Files

Histogram

Description of detection logic

IsRegressionBase: Marked as not a regression because the compare was not 5% greater than the baseline, or the value was too small.
IsImprovementBase: Marked as improvement because the compare was 5% less than the baseline, and the value was not too small.
IsImprovementCheck: Marked as improvement because the three check build points were 0.05 less than the baseline.
IsRegressionBase: Marked as not a regression because the compare was not 5% greater than the baseline, or the value was too small.
IsImprovementBase: Marked as improvement because the compare was 5% less than the baseline, and the value was not too small.
IsImprovementCheck: Marked as improvement because the three check build points were 0.05 less than the baseline.
IsImprovementWindowed: Marked as improvement because we could not find enough baseline builds for window checking.
IsChangePoint: Marked as a change because one of 5/3/2023 1:18:01 AM, 6/21/2023 4:19:00 AM, 9/28/2023 11:34:37 PM, 10/4/2023 1:30:06 AM falls between 9/25/2023 11:21:57 AM and 10/4/2023 1:30:06 AM.
IsImprovementStdDev: Marked as improvement because 149.11949297251678 (T) = (0 -1712979988.4157507) / Math.Sqrt((575164561828289.6 / (299)) + (670673094432.7651 / (6))) is greater than 1.967824097752068 = MathNet.Numerics.Distributions.StudentT.InvCDF(0, 1, (299) + (6) - 2, .975) and 0.11047541134070307 = (1925725280.9588733 - 1712979988.4157507) / 1925725280.9588733 is greater than 0.05.
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

@cincuranet
Copy link

cincuranet commented Oct 5, 2023

The diff points to dotnet/runtime@62d2c91...0cc9f21. Likely coming from dotnet/runtime#92109.

Seems to be on Windows arm64 Surface and Linux arm64 Ampere.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
arch-arm64 branch-refs/heads/main kind-micro os-windows perf-improvement PGO Applied if there were any profile guided optimization updates in the observed interval. runtime-coreclr
Projects
None yet
Development

No branches or pull requests

1 participant