search

dotnet / perf-autofiling-issues

A landing place for auto-filed performance issues before they receive triage
MIT License
9 stars 4 forks source link

[Perf] Linux/x64: 14 Improvements on 7/4/2024 8:57:20 PM #38099

Open performanceautofiler[bot] opened 2 months ago

performanceautofiler[bot] commented 2 months ago

Run Information

Name Value
Architecture x64
OS ubuntu 22.04
Queue TigerUbuntu
Baseline 898ffc5f74fad1593ed4036a69da039937224f99
Compare 0dac63563156579c7b3027ee15a51f1f19e590c2
Diff Diff
Configs AOT:true, CompilationMode:wasm, RunKind:micro

Improvements in System.Text.Json.Serialization.Tests.ReadJson<Nullable<DateTimeOffset>>

Benchmark Baseline Test Test/Base Test Quality Edge Detector Baseline IR Compare IR IR Ratio
1.18 μs 774.90 ns 0.66 0.18 False
1.18 μs 753.01 ns 0.64 0.12 False
1.08 μs 648.03 ns 0.60 0.15 False
1.41 μs 1.04 μs 0.74 0.08 False
1.06 μs 653.43 ns 0.62 0.15 False
1.46 μs 1.05 μs 0.72 0.09 False

graph graph graph graph graph graph 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
python3 .\performance\scripts\benchmarks_ci.py -f net8.0 --filter 'System.Text.Json.Serialization.Tests.ReadJson&lt;Nullable&lt;DateTimeOffset&gt;&gt;*'
### System.Text.Json.Serialization.Tests.ReadJson<Nullable<DateTimeOffset>>.DeserializeFromString(Mode: Reflection) #### ETL Files #### Histogram #### JIT Disasms ### System.Text.Json.Serialization.Tests.ReadJson<Nullable<DateTimeOffset>>.DeserializeFromString(Mode: SourceGen) #### ETL Files #### Histogram #### JIT Disasms ### System.Text.Json.Serialization.Tests.ReadJson<Nullable<DateTimeOffset>>.DeserializeFromUtf8Bytes(Mode: Reflection) #### ETL Files #### Histogram #### JIT Disasms ### System.Text.Json.Serialization.Tests.ReadJson<Nullable<DateTimeOffset>>.DeserializeFromReader(Mode: Reflection) #### ETL Files #### Histogram #### JIT Disasms ### System.Text.Json.Serialization.Tests.ReadJson<Nullable<DateTimeOffset>>.DeserializeFromUtf8Bytes(Mode: SourceGen) #### ETL Files #### Histogram #### JIT Disasms ### System.Text.Json.Serialization.Tests.ReadJson<Nullable<DateTimeOffset>>.DeserializeFromReader(Mode: SourceGen) #### 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 x64
OS ubuntu 22.04
Queue TigerUbuntu
Baseline dd6840d5e579b4a8a77814f0b96e5ace56da4932
Compare c52177b21cb358d45a18a1f14d5855aedb61d0ec
Diff Diff
Configs AOT:true, CompilationMode:wasm, RunKind:micro

Improvements in System.Text.Json.Serialization.Tests.ReadJson<SimpleStructWithProperties>

Benchmark Baseline Test Test/Base Test Quality Edge Detector Baseline IR Compare IR IR Ratio
3.31 μs 2.49 μs 0.75 0.19 False
3.87 μs 3.30 μs 0.85 0.15 False
3.58 μs 2.71 μs 0.76 0.23 False
3.41 μs 2.62 μs 0.77 0.17 False
3.07 μs 2.62 μs 0.86 0.24 False
3.93 μs 3.07 μs 0.78 0.17 False

graph graph graph graph graph graph 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
python3 .\performance\scripts\benchmarks_ci.py -f net8.0 --filter 'System.Text.Json.Serialization.Tests.ReadJson&lt;SimpleStructWithProperties&gt;*'
### System.Text.Json.Serialization.Tests.ReadJson<SimpleStructWithProperties>.DeserializeFromUtf8Bytes(Mode: SourceGen) #### ETL Files #### Histogram #### JIT Disasms ### System.Text.Json.Serialization.Tests.ReadJson<SimpleStructWithProperties>.DeserializeFromReader(Mode: Reflection) #### ETL Files #### Histogram #### JIT Disasms ### System.Text.Json.Serialization.Tests.ReadJson<SimpleStructWithProperties>.DeserializeFromString(Mode: Reflection) #### ETL Files #### Histogram #### JIT Disasms ### System.Text.Json.Serialization.Tests.ReadJson<SimpleStructWithProperties>.DeserializeFromString(Mode: SourceGen) #### ETL Files #### Histogram #### JIT Disasms ### System.Text.Json.Serialization.Tests.ReadJson<SimpleStructWithProperties>.DeserializeFromUtf8Bytes(Mode: Reflection) #### ETL Files #### Histogram #### JIT Disasms ### System.Text.Json.Serialization.Tests.ReadJson<SimpleStructWithProperties>.DeserializeFromReader(Mode: SourceGen) #### 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 x64
OS ubuntu 22.04
Queue TigerUbuntu
Baseline dd6840d5e579b4a8a77814f0b96e5ace56da4932
Compare 0dac63563156579c7b3027ee15a51f1f19e590c2
Diff Diff
Configs AOT:true, CompilationMode:wasm, RunKind:micro

Improvements in MicroBenchmarks.Serializers.Json_FromString<LoginViewModel>

Benchmark Baseline Test Test/Base Test Quality Edge Detector Baseline IR Compare IR IR Ratio
2.39 μs 1.97 μs 0.83 0.08 False
1.93 μs 1.53 μs 0.79 0.10 False

graph graph 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
python3 .\performance\scripts\benchmarks_ci.py -f net8.0 --filter 'MicroBenchmarks.Serializers.Json_FromString&lt;LoginViewModel&gt;*'
### MicroBenchmarks.Serializers.Json_FromString<LoginViewModel>.SystemTextJson_Reflection_ #### ETL Files #### Histogram #### JIT Disasms ### MicroBenchmarks.Serializers.Json_FromString<LoginViewModel>.SystemTextJson_SourceGen_ #### 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)
lewing commented 2 months ago

looks like improvements from https://github.com/dotnet/runtime/pull/104328 cc @eiriktsarpalis @stephentoub

stephentoub commented 2 months ago

looks like improvements from dotnet/runtime#104328 cc @eiriktsarpalis @stephentoub

Lovely... was this expected, @eiriktsarpalis?

Also, just looking at the above graphs, do we know what happened back in Nov/Dec where perf fell off a cliff? In some cases this seems to bring us back to the same levels as before that.

eiriktsarpalis commented 2 months ago

Lovely... was this expected, @eiriktsarpalis?

Nope.. given that this change is appending functionality I would have actually expected slight regressions. Perhaps it was caused by some of the refactorings I made in the reader class to use pattern matching, but more likely it's unrelated to this change?

Also, just looking at the above graphs, do we know what happened back in Nov/Dec where perf fell off a cliff?

I don't know. Don't recall a regression report being filed either.