[Perf] Changes at 10/13/2021 4:29:43 PM

[performanceautofiler]

Run Information

Architecture	x64
OS	Windows 10.0.19042
Baseline	845e66cd08240dc77bce98f40cecaaa0f0dadc65
Compare	5b1ebf72747f530f59427783ce73501cd840a279
Diff	Diff

Improvemnts in System.IO.Tests.Perf_File

Benchmark	Baseline	Test	Test/Base	Test Quality	Edge Detector	Baseline IR	Compare IR	IR Ratio	Baseline ETL	Compare ETL
Exists - Duration of single invocation	17.63 μs	12.49 μs	0.71	0.08	False
AppendAllTextAsync - Duration of single invocation	278.12 μs	215.42 μs	0.77	0.01	True
WriteAllTextAsync - Duration of single invocation	238.72 μs	212.88 μs	0.89	0.02	False

Test Report

Repro

git clone https://github.com/dotnet/performance.git
py .\performance\scripts\benchmarks_ci.py -f net6.0 --filter 'System.IO.Tests.Perf_File*'

Payloads

Baseline
Compare

Histogram

System.IO.Tests.Perf_File.Exists

System.IO.Tests.Perf_File.AppendAllTextAsync(size: 10000)

System.IO.Tests.Perf_File.WriteAllTextAsync(size: 10000)

Docs

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

### Run Information

Architecture	x64
OS	Windows 10.0.19042
Baseline	845e66cd08240dc77bce98f40cecaaa0f0dadc65
Compare	5b1ebf72747f530f59427783ce73501cd840a279
Diff	Diff

Improvemnts in System.IO.MemoryMappedFiles.Tests.Perf_MemoryMappedFile

Benchmark	Baseline	Test	Test/Base	Test Quality	Edge Detector	Baseline IR	Compare IR	IR Ratio	Baseline ETL	Compare ETL
CreateFromFile - Duration of single invocation	62.45 μs	50.98 μs	0.82	0.09	False

Test Report

Repro

git clone https://github.com/dotnet/performance.git
py .\performance\scripts\benchmarks_ci.py -f net6.0 --filter 'System.IO.MemoryMappedFiles.Tests.Perf_MemoryMappedFile*'

Payloads

Baseline
Compare

Histogram

System.IO.MemoryMappedFiles.Tests.Perf_MemoryMappedFile.CreateFromFile(capacity: 10000)

Docs

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

[danmoseley]

@kunalspathak is the 'diff' link at the top correct? it excludes dotnet/runtime@cf49643 which we're assuming caused the improvements.

[kunalspathak]

@kunalspathak is the 'diff' link at the top correct? it excludes dotnet/runtime@cf49643 which we're assuming caused the improvements.

I am not sure. @DrewScoggins - can you double check?

[danmoseley]

perhaps it was dotnet/runtime#58434 which IS in the bracket.

[DrewScoggins]

Sorry for the delayed response here, I was OOF most of last week. If you look at the three System.IO perf tests you can see that the first one improves slightly before the other two. You can also see that the issue title has the date of the improvement for the first test and not the other two. What happened here is that for the first test we see a big drop at 10/13 16:29, and in the other two we see a small drop at that time. So all three registered as improvements for that time frame. That is why they all get lumped together here, and if you look at the range of the improvement on the first test you will see that it lines up with the baseline and compare builds, and the diff link.