Skip to content

Pick matching SVE fallback kernel in ARM64 DYNAMIC_ARCH by vector length #6130

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
martin-frbg merged 8 commits into from
Oct 17, 2026
Merged

Pick matching SVE fallback kernel in ARM64 DYNAMIC_ARCH by vector length #6130

Changes from all commits
Commits
Show all changes
8 commits
Select commit Hold shift + click to select a range
3ce8746
Pick SVE fallback kernel by vector length instead of defaulting to AR…
martin-frbg Oct 14, 2026
429a5bf
Use inline asm for sve vector length
martin-frbg Oct 15, 2026
53a1a57
fix quotes
martin-frbg Oct 15, 2026
31b7bf8
use correct mnemonic
martin-frbg Oct 15, 2026
def9aee
Fall back to prctl on Linux if sve not available to the compiler
martin-frbg Oct 15, 2026
df61124
Update dynamic_arm64.c
martin-frbg Oct 15, 2026
8d91f94
Scale prctl result to match cntw unit
martin-frbg Oct 15, 2026
912f266
Fall back to ARMV8 if veclen of unknown SVE-capable cpu is not among …
martin-frbg Oct 16, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
27 changes: 25 additions & 2 deletions driver/others/dynamic_arm64.c
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -48,6 +48,7 @@
#if (defined OS_LINUX || defined OS_ANDROID)
#include <asm/hwcap.h>
#include <sys/auxv.h>
#include <sys/prctl.h>
#endif

#ifdef __APPLE__
Expand Down Expand Up @@ -231,7 +232,7 @@ extern gotoblas_t gotoblas_VORTEXM4;
extern void openblas_warning(int verbose, const char * msg);
#define FALLBACK_VERBOSE 1
#define NEOVERSEN1_FALLBACK "OpenBLAS : Your OS does not support SVE instructions. OpenBLAS is using Neoverse N1 kernels as a fallback, which may give poorer performance.\n"

#define VECLEN_FALLBACK "OpenBLAS : None of the supported targets matches the SVE vector length of your system. Falling back to plain ARMV8 which may give poorer performance.\n"
#define NUM_CORETYPES 21

/*
Expand Down Expand Up @@ -550,7 +551,29 @@ static gotoblas_t *get_coretype(void) {

#ifndef NO_SVE
if ((getauxval(AT_HWCAP) & HWCAP_SVE)) {
return &gotoblas_ARMV8SVE;
int veclen;
#ifdef __ARM_FEATURE_SVE
asm("cntw %[SIZE_]" : [SIZE_] "=r" (veclen) : : );
#elif defined(OS_LINUX)
veclen = prctl(PR_SVE_GET_VL);
if (veclen > -1) veclen = (veclen & PR_SVE_VL_LEN_MASK) / 4;
else veclen = 4;
#else
veclen = 4;
#endif
switch(veclen) {
case 4:
return &gotoblas_ARMV8SVE;
break;
case 8:
return &gotoblas_NEOVERSEV1;

@almayne almayne Oct 15, 2026

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Note that KERNEL.NEOVERSEV1 includes KERNEL.ARMV8SVE

@martin-frbg martin-frbg Oct 15, 2026

Copy link
Copy Markdown
Collaborator Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

It does, but V1 has the GEMM_UNROLL_MN set appropriately, which should take care of the L3 packing issues

break;
case 16:
return &gotoblas_A64FX;
default:
openblas_warning(FALLBACK_VERBOSE, VECLEN_FALLBACK);
return &gotoblas_ARMV8;
}
}
#endif

Expand Down
Loading