EgorBot for EgorBo in #109570

[EgorBot]

Processing https://github.com/dotnet/runtime/pull/109570#issuecomment-2458281137 command:

Command

-intel -amd --envvars DOTNET_JitDisasm:Zero256_Align64 ```cs using System.Runtime.InteropServices; using BenchmarkDotNet.Attributes; using System.Runtime.CompilerServices; using BenchmarkDotNet.Running; BenchmarkSwitcher.FromAssembly(typeof(Bench).Assembly).Run(args); public unsafe class Bench { static byte* _srcAlign64; static byte* _srcAlign8; [GlobalSetup] public void Setup() { _srcAlign64 = (byte*)NativeMemory.AlignedAlloc(1024 * 1024, 64); _srcAlign8 = _srcAlign64 + 8; } [GlobalCleanup] public void Cleanup() => NativeMemory.AlignedFree(_srcAlign64); [Benchmark] public void Zero256_Align64() => Unsafe.InitBlockUnaligned(_srcAlign64, 0, 256); [Benchmark] public void Zero128_Align64() => Unsafe.InitBlockUnaligned(_srcAlign64, 0, 128); [Benchmark] public void Zero64_Align64() => Unsafe.InitBlockUnaligned(_srcAlign64, 0, 64); [Benchmark] public void Zero20_Align64() => Unsafe.InitBlockUnaligned(_srcAlign64, 0, 20); [Benchmark] public void Zero256_Align8() => Unsafe.InitBlockUnaligned(_srcAlign8, 0, 256); [Benchmark] public void Zero128_Align8() => Unsafe.InitBlockUnaligned(_srcAlign8, 0, 128); [Benchmark] public void Zero64_Align8() => Unsafe.InitBlockUnaligned(_srcAlign8, 0, 64); [Benchmark] public void Zero20_Align8() => Unsafe.InitBlockUnaligned(_srcAlign8, 0, 20); } ```

(EgorBot will reply in this issue)

[EgorBot]

Benchmark results on linux-genoa

BenchmarkDotNet v0.14.0, Ubuntu 24.04 LTS (Noble Numbat)
AMD EPYC 9R14, 1 CPU, 16 logical and 16 physical cores
  Job-JETMWX : .NET 10.0.0 (42.42.42.42424), X64 RyuJIT AVX-512F+CD+BW+DQ+VL+VBMI
  Job-ZSTWZB : .NET 10.0.0 (42.42.42.42424), X64 RyuJIT AVX-512F+CD+BW+DQ+VL+VBMI
EnvironmentVariables=DOTNET_JitDisasm=Zero256_Align64

Method	Toolchain	Mean	Error	Ratio
Zero256_Align64	Main	1.0433 ns	0.0007 ns	1.00
Zero256_Align64	PR	8.4146 ns	0.0026 ns	8.07
Zero128_Align64	Main	0.0363 ns	0.0130 ns	1.12
Zero128_Align64	PR	7.8806 ns	0.0014 ns	242.57
Zero64_Align64	Main	0.0002 ns	0.0002 ns	?
Zero64_Align64	PR	1.4106 ns	0.0059 ns	?
Zero20_Align64	Main	0.2728 ns	0.0006 ns	1.00
Zero20_Align64	PR	1.3129 ns	0.0020 ns	4.81
Zero256_Align8	Main	1.3508 ns	0.0031 ns	1.00
Zero256_Align8	PR	10.8843 ns	0.0133 ns	8.06
Zero128_Align8	Main	0.4998 ns	0.0023 ns	1.00
Zero128_Align8	PR	10.3437 ns	0.0051 ns	20.69
Zero64_Align8	Main	0.0002 ns	0.0002 ns	?
Zero64_Align8	PR	1.4034 ns	0.0028 ns	?
Zero20_Align8	Main	0.2734 ns	0.0012 ns	1.00
Zero20_Align8	PR	1.3146 ns	0.0007 ns	4.81

BDN_Artifacts.zip

[EgorBot]

cc @EgorBo (logs)

[EgorBot]

Benchmark results on linux-sapphirelake

BenchmarkDotNet v0.14.0, Ubuntu 24.04 LTS (Noble Numbat)
Intel Xeon Platinum 8488C, 1 CPU, 16 logical and 8 physical cores
  Job-SUFLFE : .NET 10.0.0 (42.42.42.42424), X64 RyuJIT AVX-512F+CD+BW+DQ+VL+VBMI
  Job-OIKMMJ : .NET 10.0.0 (42.42.42.42424), X64 RyuJIT AVX-512F+CD+BW+DQ+VL+VBMI
EnvironmentVariables=DOTNET_JitDisasm=Zero256_Align64

Method	Toolchain	Mean	Error	Ratio
Zero256_Align64	Main	0.0000 ns	0.0000 ns	?
Zero256_Align64	PR	7.6155 ns	0.0270 ns	?
Zero128_Align64	Main	0.0055 ns	0.0073 ns	?
Zero128_Align64	PR	0.9472 ns	0.0199 ns	?
Zero64_Align64	Main	0.0031 ns	0.0040 ns	?
Zero64_Align64	PR	0.9536 ns	0.0229 ns	?
Zero20_Align64	Main	0.2898 ns	0.0088 ns	1.00
Zero20_Align64	PR	0.8497 ns	0.0190 ns	2.93
Zero256_Align8	Main	0.4823 ns	0.0070 ns	1.00
Zero256_Align8	PR	7.7413 ns	0.1844 ns	16.05
Zero128_Align8	Main	0.0045 ns	0.0034 ns	?
Zero128_Align8	PR	1.0201 ns	0.0573 ns	?
Zero64_Align8	Main	0.0046 ns	0.0068 ns	?
Zero64_Align8	PR	0.9594 ns	0.0176 ns	?
Zero20_Align8	Main	0.2864 ns	0.0087 ns	1.00
Zero20_Align8	PR	0.8476 ns	0.0202 ns	2.96

BDN_Artifacts.zip

[EgorBot]

cc @EgorBo (logs)