[Perf] Windows/x64: 1 Regression on 6/21/2026 8:49:02 AM +00:00

### Run Information

Name | Value
-- | --
Architecture | x64
OS | Windows 10.0.22631
Queue | ViperWindows
Baseline | [a065c06c0c5f5647d8e312fc70dec3f5ca77b830](https://github.com/dotnet/sdk/commit/a065c06c0c5f5647d8e312fc70dec3f5ca77b830)
Compare | [b756a8d8a0e64905e16c81911bb09a8110bc3395](https://github.com/dotnet/sdk/commit/b756a8d8a0e64905e16c81911bb09a8110bc3395)
Diff | [Diff](https://github.com/dotnet/sdk/compare/a065c06c0c5f5647d8e312fc70dec3f5ca77b830...b756a8d8a0e64905e16c81911bb09a8110bc3395)
Configs | CompilationMode:tiered, RunKind:mlnet

### Regressions in Microsoft.ML.Benchmarks.HashBench

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

• [HashVectorFloat - Duration of single invocation]()
• 📝 - [Benchmark Source]()
• [ADX - Test Multi Config 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 'Microsoft.ML.Benchmarks.HashBench*'
```

### Microsoft.ML.Benchmarks.HashBench.HashVectorFloat

#### 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 src/benchmarks/real-world/Microsoft.ML.Benchmarks/HashBench.cs around lines 173-174 and reproduce Microsoft.ML.Benchmarks.HashBench.HashVectorFloat using the provided benchmarks_ci.py command. Compare the baseline and compare commits and inspect the linked benchmark and report; done means identifying the cause of the Windows/x64 regression and validating that the benchmark no longer regresses.

Written by the indexing model from the issue text.