[Perf -11%] System.Collections.Tests.Add_Remove_SteadyState<Int32>.ConcurrentQueue

[DrewScoggins]

Run Information

Architecture	x64
OS	Windows 10.0.18362
Baseline	c9d3d3e60dd9bf049c47d39895f1c2d4fddeb5df
Compare	095539fd800775fc9542a07f411cbc1b7e72084e
Diff	Diff

Regressions in System.Collections.Tests.Add_Remove_SteadyState

Benchmark	Baseline	Test	Test/Base	Baseline IR	Compare IR	IR Ratio	Baseline ETL	Compare ETL
ConcurrentQueue	20.72 ns	23.01 ns	1.11	124.02802102440427	142.22494413403405	1.1467162255701013	Trace	Trace

Historical Data in Reporting System

Repro

git clone https://github.com/dotnet/performance.git
py .\performance\scripts\benchmarks_ci.py -f netcoreapp5.0 --filter 'System.Collections.Tests.Add_Remove_SteadyState<Int32>*'

Histogram

System.Collections.Tests.Add_Remove_SteadyState.ConcurrentQueue(Count: 512)

[20.662 ; 21.152) | @@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
[21.152 ; 21.643) | 
[21.643 ; 22.297) | 
[22.297 ; 22.843) | @@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
[22.843 ; 23.407) | @@

Docs

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

Tagging subscribers to this area: @eiriktsarpalis, @jeffhandley
See info in area-owners.md if you want to be subscribed.

[DrewScoggins]

We are also seeing regressions in other similar tests across the same commit diff. I will be posting them below.

[DrewScoggins]

Run Information

Architecture	x86
OS	Windows 10.0.18362
Baseline	c9d3d3e60dd9bf049c47d39895f1c2d4fddeb5df
Compare	095539fd800775fc9542a07f411cbc1b7e72084e
Diff	Diff

Regressions in System.Collections.Concurrent.Count

Benchmark	Baseline	Test	Test/Base	Baseline IR	Compare IR	IR Ratio	Baseline ETL	Compare ETL
Queue_EnqueueCountDequeue	28.06 ns	30.47 ns	1.09	191.38764338788695	209.92582783749634	1.0968619714494205	Trace	Trace

Historical Data in Reporting System

Repro

git clone https://github.com/dotnet/performance.git
py .\performance\scripts\benchmarks_ci.py -f netcoreapp5.0 --filter 'System.Collections.Concurrent.Count<Int32>*'

Histogram

System.Collections.Concurrent.Count.Queue_EnqueueCountDequeue(Size: 512)

[27.435 ; 28.034) | @
[28.034 ; 29.015) | @@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
[29.015 ; 29.815) | 
[29.815 ; 30.746) | @@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@

Docs

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

[DrewScoggins]

Run Information

Architecture	x64
OS	Windows 10.0.18362
Baseline	c9d3d3e60dd9bf049c47d39895f1c2d4fddeb5df
Compare	095539fd800775fc9542a07f411cbc1b7e72084e
Diff	Diff

Regressions in System.Threading.Channels.Tests.UnboundedChannelPerfTests

Benchmark	Baseline	Test	Test/Base	Baseline IR	Compare IR	IR Ratio	Baseline ETL	Compare ETL
TryWriteThenTryRead	44.32 ns	47.24 ns	1.07	266.96108897183603	284.0653911793766	1.0640703942039473	Trace	Trace

Repro

git clone https://github.com/dotnet/performance.git
py .\performance\scripts\benchmarks_ci.py -f netcoreapp5.0 --filter 'System.Threading.Channels.Tests.UnboundedChannelPerfTests*'

Histogram

System.Threading.Channels.Tests.UnboundedChannelPerfTests.TryWriteThenTryRead

[44.027 ; 45.843) | @@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
[45.843 ; 47.685) | @@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@

Docs

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

[DrewScoggins]

Run Information

Architecture	x64
OS	Windows 10.0.18362
Baseline	c9d3d3e60dd9bf049c47d39895f1c2d4fddeb5df
Compare	095539fd800775fc9542a07f411cbc1b7e72084e
Diff	Diff

Regressions in System.Collections.Concurrent.Count

Benchmark	Baseline	Test	Test/Base	Baseline IR	Compare IR	IR Ratio	Baseline ETL	Compare ETL
Queue_EnqueueCountDequeue	28.61 ns	35.81 ns	1.25	226.0080356935066	242.67086925105576	1.0737267305846856	Trace	Trace

Historical Data in Reporting System

Repro

git clone https://github.com/dotnet/performance.git
py .\performance\scripts\benchmarks_ci.py -f netcoreapp5.0 --filter 'System.Collections.Concurrent.Count<String>*'

Histogram

System.Collections.Concurrent.Count.Queue_EnqueueCountDequeue(Size: 512)

[28.174 ; 29.284) | @@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
[29.284 ; 30.044) | 
[30.044 ; 31.515) | @@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
[31.515 ; 32.626) | @@@@
[32.626 ; 33.736) | 
[33.736 ; 34.847) | 
[34.847 ; 36.362) | @
[36.362 ; 37.535) | 
[37.535 ; 38.646) | @
[38.646 ; 39.847) | @

Docs

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

[DrewScoggins]

My only guess is that this is related to #38225

[naricc]

So for the mono interpreter ones, it looks like there are actually two regressions in there, one caused by #38225 and another (possibly) caused by dd23c81.

I am going to create a separate issue for that.