EgorBot for hamarb123 in #99596

[EgorBot]

Processing dotnet/runtime#99596 (comment) command:

Command

-intel -amd

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

public class Benchmarks
{
    private int[] source = null!;
    private int[] dest = null!;

    [Params(1, 4, 8, 16, 64, 256, 1024, 16384)]
    public int N;

    [GlobalSetup]
    public void Setup()
    {
        source = new int[N];
        dest = new int[N];
        new Random(42).NextBytes(MemoryMarshal.AsBytes(source.AsSpan()));
    }

    //test to see if the new codegen using shuf over perm is meaningfully faster
    [Benchmark]
    public void TestReverseEndianness() => System.Buffers.Binary.BinaryPrimitives.ReverseEndianness(source, dest);
}

(EgorBot will reply in this issue)

[EgorBot]

Benchmark results on linux_azure_genoa

BenchmarkDotNet v0.14.0, Ubuntu 24.04.1 LTS (Noble Numbat)
AMD EPYC 9V74, 1 CPU, 8 logical and 4 physical cores
  Job-NZEGIC : .NET 10.0.0 (42.42.42.42424), X64 RyuJIT AVX-512F+CD+BW+DQ+VL+VBMI
  Job-KRDJMP : .NET 10.0.0 (42.42.42.42424), X64 RyuJIT AVX-512F+CD+BW+DQ+VL+VBMI

Method	Toolchain	N	Mean	Error	Ratio
TestReverseEndianness	Main	1	2.130 ns	0.0005 ns	1.00
TestReverseEndianness	PR	1	2.181 ns	0.0005 ns	1.02
TestReverseEndianness	Main	4	2.112 ns	0.0011 ns	1.00
TestReverseEndianness	PR	4	2.112 ns	0.0005 ns	1.00
TestReverseEndianness	Main	8	1.846 ns	0.0045 ns	1.00
TestReverseEndianness	PR	8	2.115 ns	0.0011 ns	1.15
TestReverseEndianness	Main	16	2.129 ns	0.0104 ns	1.00
TestReverseEndianness	PR	16	2.223 ns	0.0107 ns	1.04
TestReverseEndianness	Main	64	4.905 ns	0.0009 ns	1.00
TestReverseEndianness	PR	64	4.906 ns	0.0015 ns	1.00
TestReverseEndianness	Main	256	20.959 ns	0.4269 ns	1.00
TestReverseEndianness	PR	256	21.124 ns	0.4491 ns	1.01
TestReverseEndianness	Main	1024	74.340 ns	0.0685 ns	1.00
TestReverseEndianness	PR	1024	73.255 ns	0.0610 ns	0.99
TestReverseEndianness	Main	16384	1,288.386 ns	0.1772 ns	1.00
TestReverseEndianness	PR	16384	1,272.751 ns	0.1428 ns	0.99

BDN_Artifacts.zip

cc @hamarb123 (agent_logs.txt). EgorBot manual: link.

[EgorBot]

Benchmark results on linux_azure_cascadelake

BenchmarkDotNet v0.14.0, Ubuntu 24.04.1 LTS (Noble Numbat)
Intel Xeon Platinum 8370C CPU 2.80GHz, 1 CPU, 8 logical and 4 physical cores
  Job-QECZCG : .NET 10.0.0 (42.42.42.42424), X64 RyuJIT AVX-512F+CD+BW+DQ+VL+VBMI
  Job-UTMNHM : .NET 10.0.0 (42.42.42.42424), X64 RyuJIT AVX-512F+CD+BW+DQ+VL+VBMI

Method	Toolchain	N	Mean	Error	Ratio
TestReverseEndianness	Main	1	2.672 ns	0.0039 ns	1.00
TestReverseEndianness	PR	1	2.876 ns	0.0007 ns	1.08
TestReverseEndianness	Main	4	2.412 ns	0.0029 ns	1.00
TestReverseEndianness	PR	4	2.414 ns	0.0075 ns	1.00
TestReverseEndianness	Main	8	2.044 ns	0.0139 ns	1.00
TestReverseEndianness	PR	8	2.045 ns	0.0030 ns	1.00
TestReverseEndianness	Main	16	2.433 ns	0.0069 ns	1.00
TestReverseEndianness	PR	16	2.482 ns	0.0062 ns	1.02
TestReverseEndianness	Main	64	6.432 ns	0.0311 ns	1.00
TestReverseEndianness	PR	64	6.425 ns	0.0652 ns	1.00
TestReverseEndianness	Main	256	19.812 ns	0.0287 ns	1.00
TestReverseEndianness	PR	256	19.923 ns	0.0283 ns	1.01
TestReverseEndianness	Main	1024	77.652 ns	0.0253 ns	1.00
TestReverseEndianness	PR	1024	77.094 ns	0.0158 ns	0.99
TestReverseEndianness	Main	16384	1,732.283 ns	0.2492 ns	1.00
TestReverseEndianness	PR	16384	1,735.780 ns	0.0807 ns	1.00

BDN_Artifacts.zip

cc @hamarb123 (agent_logs.txt). EgorBot manual: link.