EgorBot for AndyAyersMS in #109209

[EgorBot]

Processing https://github.com/dotnet/runtime/pull/109209#issuecomment-2440134776 command:

Command

-intel -arm64 ```cs using BenchmarkDotNet.Attributes; using BenchmarkDotNet.Running; using System.Runtime.CompilerServices; BenchmarkSwitcher.FromAssembly(typeof(Bench).Assembly).Run(args); public class Bench { string[] s_ro_str_array = new string[512]; [Benchmark] public int Foreach() { IEnumerable e = s_ro_str_array; int sum = 0; foreach (string s in e) sum += s == null ? 0 : s.Length; return sum; } [Benchmark] public int Count() => CountInner(s_ro_str_array); [MethodImpl(MethodImplOptions.NoInlining)] int CountInner(ICollection c) => c.Count; } ```

(EgorBot will reply in this issue)

[EgorBot]

Benchmark results on Arm64

BenchmarkDotNet v0.14.0, Ubuntu 24.04 LTS (Noble Numbat)
Arm64
  Job-TDCDWJ : .NET 10.0.0 (42.42.42.42424), Arm64 RyuJIT AdvSIMD
  Job-OGRUHE : .NET 10.0.0 (42.42.42.42424), Arm64 RyuJIT AdvSIMD

Method	Toolchain	Mean	Error	Ratio
Foreach	Main	1,297.250 ns	0.1176 ns	1.00
Foreach	PR	1,110.222 ns	0.2542 ns	0.86
Count	Main	1.112 ns	0.0002 ns	1.00
Count	PR	1.066 ns	0.0002 ns	0.96

BDN_Artifacts.zip

[EgorBot]

cc @AndyAyersMS (logs)

[EgorBot]

Benchmark results on Intel

BenchmarkDotNet v0.14.0, Ubuntu 24.04 LTS (Noble Numbat)
Intel Xeon Platinum 8488C, 1 CPU, 8 logical and 4 physical cores
  Job-OEFINY : .NET 10.0.0 (42.42.42.42424), X64 RyuJIT AVX-512F+CD+BW+DQ+VL+VBMI
  Job-VDHZTA : .NET 10.0.0 (42.42.42.42424), X64 RyuJIT AVX-512F+CD+BW+DQ+VL+VBMI

Method	Toolchain	Mean	Error	Ratio
Foreach	Main	397.2479 ns	2.3665 ns	1.00
Foreach	PR	442.7244 ns	4.5621 ns	1.11
Count	Main	3.8438 ns	0.0332 ns	1.00
Count	PR	0.5762 ns	0.0139 ns	0.15

BDN_Artifacts.zip

[EgorBot]

cc @AndyAyersMS (logs)

[AndyAyersMS]

Hmm, Foreach is slower on intel. Let me dig in to that.