[Perf] Regression in Span Indexing

[DrewScoggins]

Run Information

Architecture	x64
OS	ubuntu 18.04
Baseline	c3126d0b0e3038c1bcb9a6ec4200841be3de03ba
Compare	e0671e7ec6696449b61d61d19d3dd802b49a4fb3
Diff	Diff

Regressions in System.Memory.Span<Byte>

Benchmark	Baseline	Test	Test/Base	Test Quality	Baseline IR	Compare IR	IR Ratio	Baseline ETL	Compare ETL
IndexOfAnyFourValues - Duration of single invocation	42.03 ns	620.97 ns	14.78	0.51

Historical Data in Reporting System

Repro

git clone https://github.com/dotnet/performance.git
python3 .\performance\scripts\benchmarks_ci.py -f netcoreapp5.0 --filter 'System.Memory.Span<Byte>*'

Payloads

Baseline
Compare

Histogram

System.Memory.Span<Byte>.IndexOfAnyFourValues(Size: 512)

Docs

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

### Run Information

Architecture	x64
OS	ubuntu 18.04
Baseline	c3126d0b0e3038c1bcb9a6ec4200841be3de03ba
Compare	e0671e7ec6696449b61d61d19d3dd802b49a4fb3
Diff	Diff

Regressions in System.Memory.Span<Int32>

Benchmark	Baseline	Test	Test/Base	Test Quality	Baseline IR	Compare IR	IR Ratio	Baseline ETL	Compare ETL
IndexOfAnyFourValues - Duration of single invocation	286.54 ns	623.94 ns	2.18	0.01
BinarySearch - Duration of single invocation	15.75 ns	21.30 ns	1.35	0.06

Historical Data in Reporting System

Repro

git clone https://github.com/dotnet/performance.git
python3 .\performance\scripts\benchmarks_ci.py -f netcoreapp5.0 --filter 'System.Memory.Span<Int32>*'

Payloads

Baseline
Compare

Histogram

System.Memory.Span<Int32>.IndexOfAnyFourValues(Size: 512)

System.Memory.Span<Int32>.BinarySearch(Size: 512)

Docs

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

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

Issue Details

---

Run Information

Architecture	x64
OS	ubuntu 18.04
Baseline	c3126d0b0e3038c1bcb9a6ec4200841be3de03ba
Compare	e0671e7ec6696449b61d61d19d3dd802b49a4fb3
Diff	Diff

Regressions in System.Memory.Span<Byte>

Benchmark	Baseline	Test	Test/Base	Test Quality	Baseline IR	Compare IR	IR Ratio	Baseline ETL	Compare ETL
IndexOfAnyFourValues - Duration of single invocation	42.03 ns	620.97 ns	14.78	0.51

Historical Data in Reporting System

Repro

git clone https://github.com/dotnet/performance.git
python3 .\performance\scripts\benchmarks_ci.py -f netcoreapp5.0 --filter 'System.Memory.Span<Byte>*'

Payloads

Baseline
Compare

Histogram

System.Memory.Span<Byte>.IndexOfAnyFourValues(Size: 512)

Docs

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

### Run Information

Architecture	x64
OS	ubuntu 18.04
Baseline	c3126d0b0e3038c1bcb9a6ec4200841be3de03ba
Compare	e0671e7ec6696449b61d61d19d3dd802b49a4fb3
Diff	Diff

Regressions in System.Memory.Span<Int32>

Benchmark	Baseline	Test	Test/Base	Test Quality	Baseline IR	Compare IR	IR Ratio	Baseline ETL	Compare ETL
IndexOfAnyFourValues - Duration of single invocation	286.54 ns	623.94 ns	2.18	0.01
BinarySearch - Duration of single invocation	15.75 ns	21.30 ns	1.35	0.06

Historical Data in Reporting System

Repro

git clone https://github.com/dotnet/performance.git
python3 .\performance\scripts\benchmarks_ci.py -f netcoreapp5.0 --filter 'System.Memory.Span<Int32>*'

Payloads

Baseline
Compare

Histogram

System.Memory.Span<Int32>.IndexOfAnyFourValues(Size: 512)

System.Memory.Span<Int32>.BinarySearch(Size: 512)

Docs

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

Author:	DrewScoggins
Assignees:	-
Labels:	arch-x64, area-System.Memory, os-linux, tenet-performance, tenet-performance-benchmarks, untriaged
Milestone:	-

[GrabYourPitchforks]

Regression in IndexOfAnyFourValues is expected and is a consequence of #53652.

I don't see any recent changes to BinarySearch's source code. Perhaps codegen changed recently?

[DrewScoggins]

Looking at the diff, for BinarySearch it happens over a PGO update, so my guess is that is the most likely cause. @AndyAyersMS already has an issue open for that, so I will close this one as the other issue is by design.