[Perf] Linux/arm64: 1 Regression on 5/13/2024 4:59:01 PM

[performanceautofiler]

Run Information

Name	Value
Architecture	arm64
OS	ubuntu 22.04
Queue	AmpereUbuntu
Baseline	3de068c2be232960f4cb3b233dd2b4014e2ce7c3
Compare	8c1e16660727190ba5e013ebf01021b6586dd844
Diff	Diff
Configs	CompilationMode:tiered, RunKind:micro

Regressions in System.Text.Json.Serialization.Tests.WriteJson<LoginViewModel>

Benchmark	Baseline	Test	Test/Base	Test Quality	Edge Detector	Baseline IR	Compare IR	IR Ratio
SerializeToStream - Duration of single invocation 📝 - Benchmark Source ADX - Test Multi Config Graph	325.48 ns	344.01 ns	1.06	0.01	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.Text.Json.Serialization.Tests.WriteJson<LoginViewModel>*'

System.Text.Json.Serialization.Tests.WriteJson<LoginViewModel>.SerializeToStream(Mode: SourceGen)

ETL Files

Histogram

JIT Disasms

Docs

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

[dotnet-policy-service]

Tagging subscribers to this area: @dotnet/area-system-text-json, @gregsdennis
See info in area-owners.md if you want to be subscribed.

[EgorBo]

Looks like it regressed by #101461 cc @BrennanConroy

[stephentoub]

cc: @eiriktsarpalis

[BrennanConroy]

Darn, might be the struct copying. Could probably convert it to static methods instead. Although odd it's only 1 benchmark.

[eiriktsarpalis]

Seems to be impacting fast-path source gen only, which is probably expected -- infrastructural overhead is the dominant factor in that case.

Could probably convert it to static methods instead.

If you're talking about static abstracts, that wouldn't work with the older TFMs that STJ supports. If this is actually caused by struct copying, perhaps you could try converting the context to a ref parameter?