[Perf] Changes: about 10k size increase in system.test.json.dll

[Lxiamail]

Description

about 10k size increase in system.test.json.dll

Configuration

All

Regression?

yes

Data

Run Information

Architecture	x64
OS	ubuntu 18.04
Baseline	a8926ba
Compare	5842db2
Diff	Diff

Historical Data in Reporting System

Repro: git clone https://github.com/dotnet/performance.git python3 .\performance\scripts\benchmarks_ci.py -f netcoreapp5.0 --filter 'SOD - Minimum Blazor Template - Publish*'

[ghost]

Tagging subscribers to 'size-reduction': @eerhardt, @SamMonoRT, @marek-safar
See info in area-owners.md if you want to be subscribed.

Issue Details

---

Description

about 10k size increase in system.test.json.dll

Configuration

All

Regression?

yes

Data

Run Information

Architecture	x64
OS	ubuntu 18.04
Baseline	a8926ba
Compare	5842db2
Diff	Diff

Historical Data in Reporting System

Repro: git clone https://github.com/dotnet/performance.git python3 .\performance\scripts\benchmarks_ci.py -f netcoreapp5.0 --filter 'SOD - Minimum Blazor Template - Publish*'

Author:	Lxiamail
Assignees:	-
Labels:	size-reduction, tenet-performance, tenet-performance-benchmarks
Milestone:	-

[eerhardt]

This issue doesn't have as much information as previous issues.

1. Do we have before and after .zips of the publish folder?

2. It also doesn't have the before/after ratio table either, for example:

cc @DrewScoggins

[ghost]

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

Issue Details

---

Description

about 10k size increase in system.test.json.dll

Configuration

All

Regression?

yes

Data

Run Information

Architecture	x64
OS	ubuntu 18.04
Baseline	a8926ba
Compare	5842db2
Diff	Diff

Historical Data in Reporting System

Repro: git clone https://github.com/dotnet/performance.git python3 .\performance\scripts\benchmarks_ci.py -f netcoreapp5.0 --filter 'SOD - Minimum Blazor Template - Publish*'

Author:	Lxiamail
Assignees:	-
Labels:	area-System.Text.Json, size-reduction, tenet-performance, tenet-performance-benchmarks, untriaged
Milestone:	-

[eerhardt]

This could very well be #55108, as it is in the before and after commit range.

cc @eiriktsarpalis

[DrewScoggins]

Run Information

Architecture	x64
OS	ubuntu 18.04
Baseline	a8926ba6ecadb259694e09650225d9094081a7df
Compare	5842db2c301cb1752ae854232a4fbae51d4a9ed0
Diff	Diff

Regressions in SOD - New Blazor Template - Publish

Benchmark	Baseline	Test	Test/Base	Test Quality	Edge Detector	Baseline IR	Compare IR	IR Ratio	Baseline ETL	Compare ETL
SOD - New Blazor Template - Publish - pub/wwwroot/_framework/System.Text.Json.dll	250.50 KB	261.00 KB	1.04	0.00	True
SOD - New Blazor Template - Publish - pub/wwwroot/_framework/System.ComponentModel.dll.br	1.31 KB	1.33 KB	1.01	0.00	True
SOD - New Blazor Template - Publish - pub/wwwroot/_framework/System.Private.Runtime.InteropServices.JavaScript.dll.gz	9.70 KB	9.94 KB	1.02	0.00	True
SOD - New Blazor Template - Publish - pub/wwwroot/_framework/System.Private.Runtime.InteropServices.JavaScript.dll	20.50 KB	21.00 KB	1.02	0.00	True
SOD - New Blazor Template - Publish - pub/wwwroot/_framework/System.Text.Json.dll.br	83.34 KB	86.19 KB	1.03	0.00	True
SOD - New Blazor Template - Publish - pub/wwwroot/_framework/System.Private.Runtime.InteropServices.JavaScript.dll.br	8.63 KB	8.80 KB	1.02	0.00	True
SOD - New Blazor Template - Publish - pub/wwwroot/_framework/System.Text.Json.dll.gz	97.35 KB	100.66 KB	1.03	0.00	True

Historical Data in Reporting System

Repro

git clone https://github.com/dotnet/performance.git
python3 .\performance\scripts\benchmarks_ci.py -f netcoreapp5.0 --filter 'SOD - New Blazor Template - Publish*'

Payloads

Baseline
Compare

Histogram

SOD - New Blazor Template - Publish

Docs

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

[eerhardt]

Diffing the baseline and compare assemblies, this was caused mostly by #55108. There were also changes from #55846 in there, but those weren't as large.

@eiriktsarpalis @layomia

[eiriktsarpalis]

The reported size regression is an expected side-effect of new features being added to the serializer. I have submitted #55991 which somewhat mitigates the issue (contributing ~2K of uncompressed size reduction, ~100B compressed).

After conversation with @eerhardt, I'm going to close this issue as by-design.