[Perf] Windows/x64: 3 Regressions on 1/18/2024 8:54:27 PM

[performanceautofiler]

Run Information

Name	Value
Architecture	x64
OS	Windows 10.0.18362
Queue	TigerWindows
Baseline	77d643d8678d0cae1144557f106bb693a1cbca25
Compare	59a38f18b777155f01c66afde35f2c0a48850cb0
Diff	Diff
Configs	CompilationMode:tiered, RunKind:micro

Regressions in SeekUnroll

Benchmark	Baseline	Test	Test/Base	Test Quality	Edge Detector	Baseline IR	Compare IR	IR Ratio
Test - Duration of single invocation 📝 - Benchmark Source 📈 - ADX Test Multi Config Graph	1.09 secs	1.25 secs	1.14	0.01	False
Test - Duration of single invocation 📝 - Benchmark Source 📈 - ADX Test Multi Config Graph	1.09 secs	1.25 secs	1.14	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 'SeekUnroll*'

Payloads

Baseline
Compare

SeekUnroll.Test(boxedIndex: 3)

ETL Files

Histogram

JIT Disasms

SeekUnroll.Test(boxedIndex: 1)

ETL Files

Histogram

JIT Disasms

Docs

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

---

Run Information

Name	Value
Architecture	x64
OS	Windows 10.0.18362
Queue	TigerWindows
Baseline	77d643d8678d0cae1144557f106bb693a1cbca25
Compare	59a38f18b777155f01c66afde35f2c0a48850cb0
Diff	Diff
Configs	CompilationMode:tiered, RunKind:micro

Regressions in System.IO.Tests.Perf_FileStream

Benchmark	Baseline	Test	Test/Base	Test Quality	Edge Detector	Baseline IR	Compare IR	IR Ratio
GetLength - Duration of single invocation 📝 - Benchmark Source 📈 - ADX Test Multi Config Graph	25.77 ns	30.12 ns	1.17	0.55	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.IO.Tests.Perf_FileStream*'

Payloads

Baseline
Compare

System.IO.Tests.Perf_FileStream.GetLength(options: Read)

ETL Files

Histogram

JIT Disasms

Docs

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

[EgorBo]

The first two are long running benchmarks, we suspect it's OSR related and hence, #97122 cc @jakobbotsch

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

Issue Details

---

Run Information

Name	Value
Architecture	x64
OS	Windows 10.0.18362
Queue	TigerWindows
Baseline	77d643d8678d0cae1144557f106bb693a1cbca25
Compare	59a38f18b777155f01c66afde35f2c0a48850cb0
Diff	Diff
Configs	CompilationMode:tiered, RunKind:micro

Regressions in SeekUnroll

Benchmark	Baseline	Test	Test/Base	Test Quality	Edge Detector	Baseline IR	Compare IR	IR Ratio
Test - Duration of single invocation 📝 - Benchmark Source 📈 - ADX Test Multi Config Graph	1.09 secs	1.25 secs	1.14	0.01	False
Test - Duration of single invocation 📝 - Benchmark Source 📈 - ADX Test Multi Config Graph	1.09 secs	1.25 secs	1.14	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 'SeekUnroll*'

Payloads

Baseline
Compare

SeekUnroll.Test(boxedIndex: 3)

ETL Files

Histogram

JIT Disasms

SeekUnroll.Test(boxedIndex: 1)

ETL Files

Histogram

JIT Disasms

Docs

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

---

Run Information

Name	Value
Architecture	x64
OS	Windows 10.0.18362
Queue	TigerWindows
Baseline	77d643d8678d0cae1144557f106bb693a1cbca25
Compare	59a38f18b777155f01c66afde35f2c0a48850cb0
Diff	Diff
Configs	CompilationMode:tiered, RunKind:micro

Regressions in System.IO.Tests.Perf_FileStream

Benchmark	Baseline	Test	Test/Base	Test Quality	Edge Detector	Baseline IR	Compare IR	IR Ratio
GetLength - Duration of single invocation 📝 - Benchmark Source 📈 - ADX Test Multi Config Graph	25.77 ns	30.12 ns	1.17	0.55	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.IO.Tests.Perf_FileStream*'

Payloads

Baseline
Compare

System.IO.Tests.Perf_FileStream.GetLength(options: Read)

ETL Files

Histogram

JIT Disasms

Docs

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

Author:	performanceautofiler[bot]
Assignees:	jakobbotsch
Labels:	os-windows, arch-x64, area-CodeGen-coreclr, untriaged, runtime-coreclr
Milestone:	-

[jakobbotsch]

@EgorBot -intel -amd -commit 2f63046 vs previous --disasm --envvars "DOTNET_JitDisasm:Test"

// Licensed to the .NET Foundation under one or more agreements.
// The .NET Foundation licenses this file to you under the MIT license.
// See the LICENSE file in the project root for more information.

using System.Numerics;
using System.Runtime.CompilerServices;
using BenchmarkDotNet.Attributes;

public class SeekUnroll
{
    // The purpose of this micro-benchmark is to measure the effect of unrolling
    // on this loop (taken from https://github.com/aspnet/KestrelHttpServer/pull/1138)
    [MethodImpl(MethodImplOptions.AggressiveInlining)]
    static int FindByte(ref Vector<byte> byteEquals)
    {
        var vector64 = Vector.AsVectorInt64(byteEquals);
        long longValue = 0;
        var i = 0;
        for (; i < Vector<long>.Count; i++)
        {
            longValue = vector64[i];
            if (longValue == 0) continue;
            break;
        }

        // Flag least significant power of two bit
        var powerOfTwoFlag = (ulong)(longValue ^ (longValue - 1));
        // Shift all powers of two into the high byte and extract
        var foundByteIndex = (int)((powerOfTwoFlag * _xorPowerOfTwoToHighByte) >> 57);
        // Single LEA instruction with jitted const (using function result)
        return i * 8 + foundByteIndex;
    }

    // Magic constant used in FindByte
    const ulong _xorPowerOfTwoToHighByte = (0x07ul |
                                            0x06ul << 8 |
                                            0x05ul << 16 |
                                            0x04ul << 24 |
                                            0x03ul << 32 |
                                            0x02ul << 40 |
                                            0x01ul << 48) + 1;

    // Inner loop to repeatedly call FindByte
    [MethodImpl(MethodImplOptions.AggressiveInlining)]
    static void InnerLoop(ref int foundIndex, ref Vector<Byte> vector)
    {
        for (int i = 0; i < InnerIterations; i++)
        {
            foundIndex = FindByte(ref vector);
        }
    }

    static int InnerIterations = 1000000000;

    [Benchmark(Description = nameof(SeekUnroll))]
    [ArgumentsSource(nameof(ArrayedBoxedIndicesToTest))]
    [MemoryRandomization]
    public bool Test(int boxedIndex) 
    {
        int index = boxedIndex;
        if (index >= Vector<Byte>.Count)
        {
            // FindByte assumes index is in range
            index = 0;
        }
        var bytes = new Byte[Vector<Byte>.Count];
        bytes[index] = 255;
        Vector<Byte> vector = new Vector<Byte>(bytes);

        int foundIndex = -1;

        InnerLoop(ref foundIndex, ref vector);

        return (index == foundIndex);
    }

    public static IEnumerable<object> ArrayedBoxedIndicesToTest => new object[] { 1, 3, 11, 19, 27 };
}