[Perf] Linux/arm64: 4 Regressions on 7/6/2026 7:38:21 AM +00:00

### Run Information

Name | Value
-- | --
Architecture | arm64
OS | azurelinux 3.0
Queue | CobaltAzureLinux
Baseline | [8f365c934088b8ae21a4dfce8073da0d3eb22a54](https://github.com/dotnet/runtime/commit/8f365c934088b8ae21a4dfce8073da0d3eb22a54)
Compare | [c473d80e67a0e090fa4a3e67fbc5bc9d122b172a](https://github.com/dotnet/runtime/commit/c473d80e67a0e090fa4a3e67fbc5bc9d122b172a)
Diff | [Diff](https://github.com/dotnet/runtime/compare/8f365c934088b8ae21a4dfce8073da0d3eb22a54...c473d80e67a0e090fa4a3e67fbc5bc9d122b172a)
Configs | CompilationMode:tiered, R2RType:r2r_interpreter, RunKind:micro

### Regressions in System.Numerics.Tests.Perf_VectorOf<Double>

Benchmark | Baseline | Test | Test/Base | Test Quality | Edge Detector | Baseline IR | Compare IR | IR Ratio
-- | -- | -- | -- | -- | -- | -- | -- | --
|

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

| 1.78 ns | 3.43 ns | 1.93 | 0.17 | False | | |
|

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

| 1.53 ns | 2.65 ns | 1.73 | 0.09 | 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.Numerics.Tests.Perf_VectorOf*'
```

### System.Numerics.Tests.Perf_VectorOf<Double>.ConditionalSelectBenchmark

#### ETL Files

#### Histogram

#### JIT Disasms

### System.Numerics.Tests.Perf_VectorOf<Double>.OnesComplementOperatorBenchmark

#### 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)

---

### Run Information

Name | Value
-- | --
Architecture | arm64
OS | azurelinux 3.0
Queue | CobaltAzureLinux
Baseline | [8f365c934088b8ae21a4dfce8073da0d3eb22a54](https://github.com/dotnet/runtime/commit/8f365c934088b8ae21a4dfce8073da0d3eb22a54)
Compare | [c473d80e67a0e090fa4a3e67fbc5bc9d122b172a](https://github.com/dotnet/runtime/commit/c473d80e67a0e090fa4a3e67fbc5bc9d122b172a)
Diff | [Diff](https://github.com/dotnet/runtime/compare/8f365c934088b8ae21a4dfce8073da0d3eb22a54...c473d80e67a0e090fa4a3e67fbc5bc9d122b172a)
Configs | CompilationMode:tiered, R2RType:r2r_interpreter, RunKind:micro

### Regressions in System.Runtime.Intrinsics.Tests.Perf_Vector128Of<Int64>

Benchmark | Baseline | Test | Test/Base | Test Quality | Edge Detector | Baseline IR | Compare IR | IR Ratio
-- | -- | -- | -- | -- | -- | -- | -- | --
|

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

| 1.78 ns | 3.48 ns | 1.96 | 0.17 | False | | |

![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.Runtime.Intrinsics.Tests.Perf_Vector128Of*'
```

### System.Runtime.Intrinsics.Tests.Perf_Vector128Of<Int64>.ConditionalSelectBenchmark

#### 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)

---

### Run Information

Name | Value
-- | --
Architecture | arm64
OS | azurelinux 3.0
Queue | CobaltAzureLinux
Baseline | [8f365c934088b8ae21a4dfce8073da0d3eb22a54](https://github.com/dotnet/runtime/commit/8f365c934088b8ae21a4dfce8073da0d3eb22a54)
Compare | [c473d80e67a0e090fa4a3e67fbc5bc9d122b172a](https://github.com/dotnet/runtime/commit/c473d80e67a0e090fa4a3e67fbc5bc9d122b172a)
Diff | [Diff](https://github.com/dotnet/runtime/compare/8f365c934088b8ae21a4dfce8073da0d3eb22a54...c473d80e67a0e090fa4a3e67fbc5bc9d122b172a)
Configs | CompilationMode:tiered, R2RType:r2r_interpreter, RunKind:micro

### Regressions in System.Numerics.Tests.Perf_VectorOf<Single>

Benchmark | Baseline | Test | Test/Base | Test Quality | Edge Detector | Baseline IR | Compare IR | IR Ratio
-- | -- | -- | -- | -- | -- | -- | -- | --
|

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

| 1.54 ns | 2.84 ns | 1.84 | 0.10 | False | | |

![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.Numerics.Tests.Perf_VectorOf*'
```

### System.Numerics.Tests.Perf_VectorOf<Single>.OnesComplementBenchmark

#### 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)

Start with the listed benchmark sources, including src/benchmarks/micro/libraries/System.Numerics.Vectors/Perf_VectorOfT.cs and src/benchmarks/micro/libraries/System.Runtime.Intrinsics/Perf_Vector128OfT.cs, then follow the benchmarking workflow linked in the issue. Reproduce the affected filters on ARM64 AzureLinux using the supplied benchmarks_ci.py command and compare the baseline and test results. Done means identifying and resolving the cause of the reported regressions, with the benchmark results no longer showing the slowdown.

Written by the indexing model from the issue text.