[Perf] Windows/x64: 3 Regressions on 2/19/2024 10:53:50 AM

[performanceautofiler]

Run Information

Name	Value
Architecture	x64
OS	Windows 10.0.18362
Queue	TigerWindows
Baseline	d972a19c077e899d0b3fff97d955968e50906396
Compare	5e1caf14fd94b4ed85357002303c8935fedacb8b
Diff	Diff
Configs	CompilationMode:tiered, RunKind:micro

Regressions in System.Tests.Perf_String

Benchmark	Baseline	Test	Test/Base	Test Quality	Edge Detector	Baseline IR	Compare IR	IR Ratio
IndexOfAny - Duration of single invocation 📝 - Benchmark Source 📈 - ADX Test Multi Config Graph	7.88 ns	11.71 ns	1.49	0.31	False
Trim - Duration of single invocation 📝 - Benchmark Source 📈 - ADX Test Multi Config Graph	14.48 ns	15.74 ns	1.09	0.09	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.Tests.Perf_String*'

Payloads

Baseline
Compare

System.Tests.Perf_String.IndexOfAny

ETL Files

Histogram

JIT Disasms

System.Tests.Perf_String.Trim(s: " Te st ")

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 10.0.18362
Queue	TigerWindows
Baseline	d972a19c077e899d0b3fff97d955968e50906396
Compare	5e1caf14fd94b4ed85357002303c8935fedacb8b
Diff	Diff
Configs	CompilationMode:tiered, RunKind:micro

Regressions in Microsoft.Extensions.Primitives.StringSegmentBenchmark

Benchmark	Baseline	Test	Test/Base	Test Quality	Edge Detector	Baseline IR	Compare IR	IR Ratio
IndexOfAny - Duration of single invocation 📝 - Benchmark Source 📈 - ADX Test Multi Config Graph	6.26 ns	7.91 ns	1.26	0.17	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 'Microsoft.Extensions.Primitives.StringSegmentBenchmark*'

Payloads

Baseline
Compare

Microsoft.Extensions.Primitives.StringSegmentBenchmark.IndexOfAny

ETL Files

Histogram

JIT Disasms

Docs

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

[DrewScoggins]

Related to #96588.

PTAL @MihaZupan

Improvements:

dotnet/perf-autofiling-issues#29844
dotnet/perf-autofiling-issues#29818
dotnet/perf-autofiling-issues#29802

[MihaZupan]

The string IndexOfAny regressions are very likely #96588 (comment) - extra per-call overhead when the JIT can't see the values as constants.
@stephentoub does this look like something we should care about given the values passed to here are usually going to be constants? The relative size of the regression is a bit sad

[ghost]

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

Issue Details

---

Run Information

Name	Value
Architecture	x64
OS	Windows 10.0.18362
Queue	TigerWindows
Baseline	d972a19c077e899d0b3fff97d955968e50906396
Compare	5e1caf14fd94b4ed85357002303c8935fedacb8b
Diff	Diff
Configs	CompilationMode:tiered, RunKind:micro

Regressions in System.Tests.Perf_String

Benchmark	Baseline	Test	Test/Base	Test Quality	Edge Detector	Baseline IR	Compare IR	IR Ratio
IndexOfAny - Duration of single invocation 📝 - Benchmark Source 📈 - ADX Test Multi Config Graph	7.88 ns	11.71 ns	1.49	0.31	False
Trim - Duration of single invocation 📝 - Benchmark Source 📈 - ADX Test Multi Config Graph	14.48 ns	15.74 ns	1.09	0.09	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.Tests.Perf_String*'

Payloads

Baseline
Compare

System.Tests.Perf_String.IndexOfAny

ETL Files

Histogram

JIT Disasms

System.Tests.Perf_String.Trim(s: " Te st ")

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 10.0.18362
Queue	TigerWindows
Baseline	d972a19c077e899d0b3fff97d955968e50906396
Compare	5e1caf14fd94b4ed85357002303c8935fedacb8b
Diff	Diff
Configs	CompilationMode:tiered, RunKind:micro

Regressions in Microsoft.Extensions.Primitives.StringSegmentBenchmark

Benchmark	Baseline	Test	Test/Base	Test Quality	Edge Detector	Baseline IR	Compare IR	IR Ratio
IndexOfAny - Duration of single invocation 📝 - Benchmark Source 📈 - ADX Test Multi Config Graph	6.26 ns	7.91 ns	1.26	0.17	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 'Microsoft.Extensions.Primitives.StringSegmentBenchmark*'

Payloads

Baseline
Compare

Microsoft.Extensions.Primitives.StringSegmentBenchmark.IndexOfAny

ETL Files

Histogram

JIT Disasms

Docs

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

Author:	performanceautofiler[bot]
Assignees:	-
Labels:	area-System.Buffers, os-windows, arch-x64, untriaged, runtime-coreclr, needs-area-label
Milestone:	-

[stephentoub]

given the values passed to here are usually going to be constants?

It's probably still the right tradeoff, but it'd be nice if we didn't have to make that tradeoff... there are some call sites that don't involve constants (either direct or that the JIT would be able to see via inlining). In the linked comment, you'd mentioned IsKnownConstant... could we follow-up on that to see if there's a possible solution there to let us have our cake and eat it, too?