[Perf] Windows/x64: 3 Regressions on 10/14/2025 2:34:35 AM +00:00

[performanceautofiler]

Run Information

Name	Value
Architecture	x64
OS	Windows 10.0.22631
Queue	ViperWindows
Baseline	457c6b709b54b354efd8d36a9b5c03db53494612
Compare	0327b0dc1b9581bfd3d8996ec222a9218211dbd3
Diff	Diff
Configs	CompilationMode:tiered, RunKind:micro

Regressions in Interop.ComWrappersTests

Benchmark	Baseline	Test	Test/Base	Test Quality	Edge Detector	Baseline IR	Compare IR	IR Ratio
ParallelRCWLookUp - Duration of single invocation 📝 - Benchmark Source ADX - Test Multi Config Graph	547.04 ms	1.10 secs	2.01	0.07	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
py .\performance\scripts\benchmarks_ci.py -f net8.0 --filter 'Interop.ComWrappersTests*'

Interop.ComWrappersTests.ParallelRCWLookUp

ETL Files

Histogram

JIT Disasms

Docs

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

---

Run Information

Name	Value
Architecture	x64
OS	Windows 10.0.22631
Queue	ViperWindows
Baseline	457c6b709b54b354efd8d36a9b5c03db53494612
Compare	0327b0dc1b9581bfd3d8996ec222a9218211dbd3
Diff	Diff
Configs	CompilationMode:tiered, RunKind:micro

Regressions in System.Text.Json.Tests.Perf_Deep

Benchmark	Baseline	Test	Test/Base	Test Quality	Edge Detector	Baseline IR	Compare IR	IR Ratio
WriteDeepUtf8 - Duration of single invocation 📝 - Benchmark Source ADX - Test Multi Config Graph	2.45 ms	2.96 ms	1.21	0.30	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
py .\performance\scripts\benchmarks_ci.py -f net8.0 --filter 'System.Text.Json.Tests.Perf_Deep*'

System.Text.Json.Tests.Perf_Deep.WriteDeepUtf8(Formatted: False, SkipValidation: False)

ETL Files

Histogram

JIT Disasms

Docs

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

---

Run Information

Name	Value
Architecture	x64
OS	Windows 10.0.22631
Queue	ViperWindows
Baseline	457c6b709b54b354efd8d36a9b5c03db53494612
Compare	0327b0dc1b9581bfd3d8996ec222a9218211dbd3
Diff	Diff
Configs	CompilationMode:tiered, RunKind:micro

Regressions in System.Text.Json.Tests.Perf_Strings

Benchmark	Baseline	Test	Test/Base	Test Quality	Edge Detector	Baseline IR	Compare IR	IR Ratio
WriteStringsUtf16 - Duration of single invocation 📝 - Benchmark Source ADX - Test Multi Config Graph	37.54 ms	46.92 ms	1.25	0.26	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
py .\performance\scripts\benchmarks_ci.py -f net8.0 --filter 'System.Text.Json.Tests.Perf_Strings*'

System.Text.Json.Tests.Perf_Strings.WriteStringsUtf16(Formatted: True, SkipValidation: True, Escaped: AllEscaped)

ETL Files

Histogram

JIT Disasms

Docs

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

[LoopedBard3]

ParallelRCWLookup Regression caused by: dotnet/runtime#120675, is this acceptable @jkoritzinsky? If not, feel free to transfer to runtime.

Nothing obvious in range for the others (dotnet/runtime@caeb3b4...0327b0d) and it is a very large range, so I am going to try to regather some of the missing data so we can take another look.

Related Regressions:

• [Perf] Linux/x64: 1 Regression on 10/14/2025 2:34:35 AM +00:00 #64016 (ParallelRCW Regression)
• [Perf] Linux/x64: 1 Regression on 10/14/2025 2:34:35 AM +00:00 #64014 (Related to the JSON tests)

[jkoritzinsky]

For the RCW one, can you add it to the existing open issue for the Windows perf regression?

There's definitely something weird going on since we're getting very different results for windows vs Linux