[Perf] Windows/x64: Regression in System.Buffers.Text.Tests.Base64EncodeDecodeInPlaceTests

[performanceautofiler]

Run Information

Name	Value
Architecture	x64
OS	Windows 10.0.22621
Queue	TigerWindows
Baseline	6af853d7757e55a09fd9bc6dc1d6aa12c0b454f2
Compare	b2b0db0c470eba512d358dcf3d359488083e82a8
Diff	Diff
Configs	CompilationMode:tiered, RunKind:micro

Regressions in System.Buffers.Text.Tests.Base64EncodeDecodeInPlaceTests

Benchmark	Baseline	Test	Test/Base	Test Quality	Edge Detector	Baseline IR	Compare IR	IR Ratio
Base64DecodeInPlace - Duration of single invocation 📝 - Benchmark Source ADX - Test Multi Config Graph	395.24 ms	442.64 ms	1.12	0.01	False

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
py .\performance\scripts\benchmarks_ci.py -f net8.0 --filter 'System.Buffers.Text.Tests.Base64EncodeDecodeInPlaceTests*'

System.Buffers.Text.Tests.Base64EncodeDecodeInPlaceTests.Base64DecodeInPlace(NumberOfBytes: 200000000)

ETL Files

Histogram

JIT Disasms

Docs

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

[LoopedBard3]

Comparison range: efebf20...eb765b7

[LoopedBard3]

Likely: #105262 @buyaa-n

[dotnet-policy-service]

Tagging subscribers to this area: @dotnet/area-system-memory
See info in area-owners.md if you want to be subscribed.

[buyaa-n]

Likely: #105262 @buyaa-n

#105262 should not really impact this much, I will try to repro when I got time

[buyaa-n]

#105262 should not really impact this much, I will try to repro when I got time

Surprisingly it was repro-ed locally, though there is nothing we can do for improving the changes made with #105262, the only change made for Base64.EncodeDecodeInPlace with the PR is

which should not impact the perf much.

Only thing I could try is review Base64.EncodeDecodeInPlace code entirely and try to improve the perf. So, I am gonna move this issue to v10