Benchmarks for #113575 (EgorBo)

[EgorBot]

Processing dotnet/runtime#113575 (comment) command:

Command

-amd -arm -profiler

using BenchmarkDotNet.Attributes;
using BenchmarkDotNet.Running;

public class Bench
{
    byte[] _arr1 = new byte[2000];
    byte[] _arr2 = new byte[2000];

    [Benchmark]
    [Arguments(1000)]
    public void CopyN(int elems)
    {
        Span<byte> span1 = _arr1;
        Span<byte> span2 = _arr2;

        for (int i = 0; i < elems; i++)
            span1[i] = span2[i];
    }


    [Benchmark]
    [Arguments(1000)]
    public void ReversedIter(int elems)
    {
        Span<byte> span = _arr1;

        // Reversed iteration
        for (int i = span.Length - 1; i >= 0; i--)
            span[i] = 42;
    }
}

(EgorBot will reply in this issue)

[EgorBot]

Benchmark results on linux_azure_genoa

BenchmarkDotNet v0.14.0, Ubuntu 24.04.2 LTS (Noble Numbat)
AMD EPYC 9V74, 1 CPU, 16 logical and 8 physical cores
  Job-DBDRMK : .NET 10.0.0 (42.42.42.42424), X64 RyuJIT AVX-512F+CD+BW+DQ+VL+VBMI
  Job-NDYHMJ : .NET 10.0.0 (42.42.42.42424), X64 RyuJIT AVX-512F+CD+BW+DQ+VL+VBMI

Method	Toolchain	elems	Mean	Error	Ratio
CopyN	Main	1000	414.3 ns	0.03 ns	1.00
CopyN	PR	1000	277.0 ns	0.03 ns	0.67
ReversedIter	Main	1000	552.5 ns	0.24 ns	1.00
ReversedIter	PR	1000	552.0 ns	0.08 ns	1.00

BDN_Artifacts.zip

Profiling for Bench_CopyN:

SVG flamegraph: Main vs PR 🔥
Speedscope flamegraph: Main vs PR
Hot asm code: Main vs PR
HW counters: Main vs PR

Profiling for Bench_ReversedIter:

SVG flamegraph: Main vs PR 🔥
Speedscope flamegraph: Main vs PR
Hot asm code: Main vs PR
HW counters: Main vs PR

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

[EgorBot]

Benchmark results on linux_azure_cobalt100

BenchmarkDotNet v0.14.0, Ubuntu 24.04.2 LTS (Noble Numbat)
azure_cobalt100
  Job-QLIBBY : .NET 10.0.0 (42.42.42.42424), Arm64 RyuJIT AdvSIMD
  Job-NQRDYW : .NET 10.0.0 (42.42.42.42424), Arm64 RyuJIT AdvSIMD

Method	Toolchain	elems	Mean	Error	Ratio
CopyN	Main	1000	476.4 ns	0.37 ns	1.00
CopyN	PR	1000	358.4 ns	0.23 ns	0.75
ReversedIter	Main	1000	599.2 ns	3.77 ns	1.00
ReversedIter	PR	1000	596.3 ns	0.02 ns	1.00

BDN_Artifacts.zip

Profiling for Bench_CopyN:

SVG flamegraph: Main vs PR 🔥
Speedscope flamegraph: Main vs PR
Hot asm code: Main vs PR
HW counters: Main vs PR

Profiling for Bench_ReversedIter:

SVG flamegraph: Main vs PR 🔥
Speedscope flamegraph: Main vs PR
Hot asm code: Main vs PR
HW counters: Main vs PR

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