kiper220/mimalloc
1
0
Fork 0
mirror of https://github.com/microsoft/mimalloc.git synced 2025-07-06 11:34:38 +03:00
Code Issues Projects Releases Packages Wiki Activity Actions

_mi_memcpy/_mi_memzero: tighten criteria for intrinsics for windows.

Browse source 
FSRM is better used for buffer <= 128 bytes and ERMS, if supported for larger
chunks.
This commit is contained in:
David Carlier 2024-11-16 09:38:26 +00:00
parent 54940a6a65
commit 9f0a7ac0be
WARNING! Although there is a key with this ID in the database it does not verify this commit! This commit is SUSPICIOUS.
GPG key ID: 8486F847B4B94EF1
2 changed files with 7 additions and 3 deletions
Download patch file Download diff file Expand all files Collapse all files

5
src/init.c
View file

@ -543,12 +543,15 @@ void _mi_process_load(void) {
#if defined(_WIN32) && (defined(_M_IX86) || defined(_M_X64))
#include <intrin.h>
mi_decl_cache_align bool _mi_cpu_has_fsrm = false;
mi_decl_cache_align bool _mi_cpu_has_erms = false;
static void mi_detect_cpu_features(void) {
// FSRM for fast rep movsb support (AMD Zen3+ (~2020) or Intel Ice Lake+ (~2017))
// FSRM for fast short rep movsb/stosb support (AMD Zen3+ (~2020) or Intel Ice Lake+ (~2017))
// EMRS for fast enhanced rep movsb/stosb support
int32_t cpu_info[4];
__cpuid(cpu_info, 7);
_mi_cpu_has_fsrm = ((cpu_info[3] & (1 << 4)) != 0); // bit 4 of EDX : see <https://en.wikipedia.org/wiki/CPUID#EAX=7,_ECX=0:_Extended_Features>
_mi_cpu_has_erms = ((cpu_info[2] & (1 << 9)) != 0); // bit 9 of ECX : see <https://en.wikipedia.org/wiki/CPUID#EAX=7,_ECX=0:_Extended_Features>
}
#else
static void mi_detect_cpu_features(void) {