[Perf] Windows/x64: 6 Regressions on 6/21/2025 12:11:31 PM +00:00

[performanceautofiler]

Run Information

Name	Value
Architecture	x64
OS	Windows 10.0.22621
Queue	TigerWindows
Baseline	953ccfbc21d653efaf2441ff3e2dbc288fde415d
Compare	f0168ee80ba9aca18a7e7140b2bb436defda623c
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
WriteDeepUtf16 - Duration of single invocation 📝 - Benchmark Source ADX - Test Multi Config Graph	3.86 ms	5.19 ms	1.34	0.19	False
WriteDeepUtf16 - Duration of single invocation 📝 - Benchmark Source ADX - Test Multi Config Graph	3.75 ms	4.99 ms	1.33	0.20	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*'

Details

System.Text.Json.Tests.Perf_Deep.WriteDeepUtf16(Formatted: False, SkipValidation: True)

ETL Files

Histogram

JIT Disasms

System.Text.Json.Tests.Perf_Deep.WriteDeepUtf16(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 10.0.22621
Queue	TigerWindows
Baseline	953ccfbc21d653efaf2441ff3e2dbc288fde415d
Compare	f0168ee80ba9aca18a7e7140b2bb436defda623c
Diff	Diff
Configs	CompilationMode:tiered, RunKind:micro

Regressions in System.MathBenchmarks.Single

Benchmark	Baseline	Test	Test/Base	Test Quality	Edge Detector	Baseline IR	Compare IR	IR Ratio
Max - Duration of single invocation 📝 - Benchmark Source ADX - Test Multi Config Graph	6.29 μs	9.31 μs	1.48	0.00	False
Min - Duration of single invocation 📝 - Benchmark Source ADX - Test Multi Config Graph	6.30 μs	8.87 μs	1.41	0.00	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.MathBenchmarks.Single*'

Details

System.MathBenchmarks.Single.Max

ETL Files

Histogram

JIT Disasms

System.MathBenchmarks.Single.Min

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 10.0.22621
Queue	TigerWindows
Baseline	953ccfbc21d653efaf2441ff3e2dbc288fde415d
Compare	f0168ee80ba9aca18a7e7140b2bb436defda623c
Diff	Diff
Configs	CompilationMode:tiered, RunKind:micro

Regressions in System.MathBenchmarks.Double

Benchmark	Baseline	Test	Test/Base	Test Quality	Edge Detector	Baseline IR	Compare IR	IR Ratio
Max - Duration of single invocation 📝 - Benchmark Source ADX - Test Multi Config Graph	6.30 μs	9.33 μs	1.48	0.00	False
Min - Duration of single invocation 📝 - Benchmark Source ADX - Test Multi Config Graph	6.30 μs	9.06 μs	1.44	0.00	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.MathBenchmarks.Double*'

Details

System.MathBenchmarks.Double.Max

ETL Files

Histogram

JIT Disasms

System.MathBenchmarks.Double.Min

ETL Files

Histogram

JIT Disasms

Docs

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

[DrewScoggins]

First two tests are likely not related, but the other four are from #116804

Regressions:

• [Perf] Linux/x64: 4 Regressions on 6/21/2025 12:11:31 PM +00:00 perf-autofiling-issues#58139

[dotnet-policy-service]

Tagging subscribers to this area: @JulieLeeMSFT, @jakobbotsch
See info in area-owners.md if you want to be subscribed.

[tannergooding]

There's been no actual change to the codegen here, just a slight change to the registers used and order in T0.

I cannot reproduce any change locally on any of my Intel or AMD machines.

We correspondingly had good improvements for TensorPrimitives, which is a more realistic set of benchmarks.