Regressions in System.Linq.Tests.Perf_Enumerable

[performanceautofiler[bot]]

Run Information

Name	Value
Architecture	arm64
OS	Windows 10.0.19041
Queue	SurfaceWindows
Baseline	1a985d9d113b3873f337cead6c3a136e1876cd2c
Compare	72c4c57a3c8c1b267447f253fdc14f9d14d91a2b
Diff	Diff
Configs	CompilationMode:tiered, RunKind:micro

Regressions in System.Linq.Tests.Perf_Enumerable

Benchmark	Baseline	Test	Test/Base	Test Quality	Edge Detector	Baseline IR	Compare IR	IR Ratio
[WhereAny_LastElementMatches - Duration of single invocation](<https://pvscmdupload.blob.core.windows.net/reports/allTestHistory/refs/heads/main_arm64_Windows 10.0.19041/System.Linq.Tests.Perf_Enumerable.WhereAny_LastElementMatches(input%3a%20List).html>) 📝 - Benchmark Source 📈 - ADX Test Multi Config Graph	301.14 ns	323.99 ns	1.08	0.01	False
[EmptyTakeSelectToArray - Duration of single invocation](<https://pvscmdupload.blob.core.windows.net/reports/allTestHistory/refs/heads/main_arm64_Windows 10.0.19041/System.Linq.Tests.Perf_Enumerable.EmptyTakeSelectToArray.html>) 📝 - Benchmark Source 📈 - ADX Test Multi Config Graph	10.37 ns	28.45 ns	2.74	0.27	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.Linq.Tests.Perf_Enumerable*'

### Payloads [Baseline]() [Compare]() ### System.Linq.Tests.Perf_Enumerable.WhereAny_LastElementMatches(input: List) #### ETL Files #### Histogram #### JIT Disasms ### System.Linq.Tests.Perf_Enumerable.EmptyTakeSelectToArray #### ETL Files #### Histogram #### JIT Disasms ### Docs [Profiling workflow for dotnet/runtime repository](https://github.com/dotnet/performance/blob/master/docs/profiling-workflow-dotnet-runtime.md) [Benchmarking workflow for dotnet/runtime repository](https://github.com/dotnet/performance/blob/master/docs/benchmarking-workflow-dotnet-runtime.md)

[cincuranet]

Only EmptyTakeSelectToArray seems to be valid. Diff.

Regression is on all platforms:

[ghost]

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

Issue Details

---

### Run Information Name | Value -- | -- Architecture | arm64 OS | Windows 10.0.19041 Queue | SurfaceWindows Baseline | [1a985d9d113b3873f337cead6c3a136e1876cd2c](https://github.com/dotnet/runtime/commit/1a985d9d113b3873f337cead6c3a136e1876cd2c) Compare | [72c4c57a3c8c1b267447f253fdc14f9d14d91a2b](https://github.com/dotnet/runtime/commit/72c4c57a3c8c1b267447f253fdc14f9d14d91a2b) Diff | [Diff](https://github.com/dotnet/runtime/compare/1a985d9d113b3873f337cead6c3a136e1876cd2c...72c4c57a3c8c1b267447f253fdc14f9d14d91a2b) Configs | CompilationMode:tiered, RunKind:micro ### Regressions in System.Linq.Tests.Perf_Enumerable Benchmark | Baseline | Test | Test/Base | Test Quality | Edge Detector | Baseline IR | Compare IR | IR Ratio -- | -- | -- | -- | -- | -- | -- | -- | -- |

• [WhereAny_LastElementMatches - Duration of single invocation]()
• 📝 - [Benchmark Source]()
• [📈 - ADX Test Multi Config Graph]()

| 301.14 ns | 323.99 ns | 1.08 | 0.01 | False | | | |

• [EmptyTakeSelectToArray - Duration of single invocation]()
• 📝 - [Benchmark Source]()
• [📈 - ADX Test Multi Config Graph]()

| 10.37 ns | 28.45 ns | 2.74 | 0.27 | False | | | ![graph]() ![graph]() [Test Report]() ### Repro General Docs link: https://github.com/dotnet/performance/blob/main/docs/benchmarking-workflow-dotnet-runtime.md ```cmd git clone https://github.com/dotnet/performance.git py .\performance\scripts\benchmarks_ci.py -f net8.0 --filter 'System.Linq.Tests.Perf_Enumerable*' ```

### Payloads [Baseline]() [Compare]() ### System.Linq.Tests.Perf_Enumerable.WhereAny_LastElementMatches(input: List) #### ETL Files #### Histogram #### JIT Disasms ### System.Linq.Tests.Perf_Enumerable.EmptyTakeSelectToArray #### ETL Files #### Histogram #### JIT Disasms ### Docs [Profiling workflow for dotnet/runtime repository](https://github.com/dotnet/performance/blob/master/docs/profiling-workflow-dotnet-runtime.md) [Benchmarking workflow for dotnet/runtime repository](https://github.com/dotnet/performance/blob/master/docs/benchmarking-workflow-dotnet-runtime.md)

Author:	performanceautofiler[bot]
Assignees:	-
Labels:	`arch-arm64`, `area-System.Linq`, `os-windows`, `untriaged`, `runtime-coreclr`
Milestone:	-

[stephentoub]

Only EmptyTakeSelectToArray seems to be valid. Diff.

I think we need to accept this one. The gap has come back down a bit, to less than 10ns. The issue is we're now checking for empty arrays rather than a dedicated empty sentinel, and the perf test is using T == int, which requires covariance checks (since eg a uint[] will also pass as an int[]). So the couple of times on this path that we're testing for empty, that check is a bit more expensive than it used to be. However, it's also more comprehensive than it used to be, so whereas previously we'd only have special-cased empty inputs that came from Enumerable.Empty<T>, now we'll special-case both that and inputs from Array.Empty<T>() (which is what collection literals produce) and new T[i] where i == 0, etc.