[Perf] Linux/arm64: 3 Regressions on 4/21/2025 9:32:06 AM +00:00

[performanceautofiler]

Run Information

Name	Value
Architecture	arm64
OS	ubuntu 22.04
Queue	AmpereUbuntu
Baseline	a4848033324be5a3e0f7f0a901e222cdcadce07b
Compare	c0c5520b7b79747ca903a0ca59125bf3dede1307
Diff	Diff
Configs	CompilationMode:tiered, RunKind:micro

Regressions in System.Collections.CtorFromCollectionNonGeneric<Int32>

Benchmark	Baseline	Test	Test/Base	Test Quality	Edge Detector	Baseline IR	Compare IR	IR Ratio
Stack - Duration of single invocation 📝 - Benchmark Source ADX - Test Multi Config Graph	40.80 μs	49.89 μs	1.22	0.08	False
ArrayList - Duration of single invocation 📝 - Benchmark Source ADX - Test Multi Config Graph	37.76 μs	41.12 μs	1.09	0.12	False
Queue - Duration of single invocation 📝 - Benchmark Source ADX - Test Multi Config Graph	39.98 μs	50.31 μs	1.26	0.08	False

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.Collections.CtorFromCollectionNonGeneric<Int32>*'

System.Collections.CtorFromCollectionNonGeneric<Int32>.Stack(Size: 512)

ETL Files

Histogram

JIT Disasms

System.Collections.CtorFromCollectionNonGeneric<Int32>.ArrayList(Size: 512)

ETL Files

Histogram

JIT Disasms

System.Collections.CtorFromCollectionNonGeneric<Int32>.Queue(Size: 512)

ETL Files

Histogram

JIT Disasms

Docs

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

[AndyAyersMS]

Possibly dotnet/runtime#114754? That should have been no-diff (but maybe perturbed native PGO).

[LoopedBard3]

The regression point seemed to be noise and the test has improved greatly shortly after (~45ms to ~12ms).