Skip to content

Add dynamic support for Arm(R) Neoverse(TM) V2 processor #4444

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
merged 1 commit into from
Jan 20, 2024
Merged

Add dynamic support for Arm(R) Neoverse(TM) V2 processor #4444

Changes from all commits
Commits
Show all changes
1 commit
Select commit
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
31 changes: 21 additions & 10 deletions driver/others/dynamic_arm64.c
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,6 +1,6 @@
/*********************************************************************/
/* Copyright 2009, 2010 The University of Texas at Austin. */
/* Copyright 2023 The OpenBLAS Project */
/* Copyright 2023-2024 The OpenBLAS Project */
/* All rights reserved. */
/* */
/* Redistribution and use in source and binary forms, with or */
Expand Down Expand Up @@ -143,12 +143,13 @@ extern gotoblas_t gotoblas_ARMV8SVE;
#endif
extern gotoblas_t gotoblas_THUNDERX3T110;
#endif
#define gotoblas_NEOVERSEV2 gotoblas_NEOVERSEV1

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 NUM_CORETYPES 16
#define NUM_CORETYPES 17

/*
* In case asm/hwcap.h is outdated on the build system, make sure
Expand Down Expand Up @@ -178,6 +179,7 @@ static char *corename[] = {
"emag8180",
"neoversen1",
"neoversev1",
"neoversev2",
"neoversen2",
"thunderx3t110",
"cortexa55",
Expand All @@ -198,10 +200,11 @@ char *gotoblas_corename(void) {
if (gotoblas == &gotoblas_EMAG8180) return corename[ 9];
if (gotoblas == &gotoblas_NEOVERSEN1) return corename[10];
if (gotoblas == &gotoblas_NEOVERSEV1) return corename[11];
if (gotoblas == &gotoblas_NEOVERSEN2) return corename[12];
if (gotoblas == &gotoblas_THUNDERX3T110) return corename[13];
if (gotoblas == &gotoblas_CORTEXA55) return corename[14];
if (gotoblas == &gotoblas_ARMV8SVE) return corename[15];
if (gotoblas == &gotoblas_NEOVERSEV2) return corename[12];
if (gotoblas == &gotoblas_NEOVERSEN2) return corename[13];
if (gotoblas == &gotoblas_THUNDERX3T110) return corename[14];
if (gotoblas == &gotoblas_CORTEXA55) return corename[15];
if (gotoblas == &gotoblas_ARMV8SVE) return corename[16];
return corename[NUM_CORETYPES];
}

Expand Down Expand Up @@ -233,10 +236,11 @@ static gotoblas_t *force_coretype(char *coretype) {
case 9: return (&gotoblas_EMAG8180);
case 10: return (&gotoblas_NEOVERSEN1);
case 11: return (&gotoblas_NEOVERSEV1);
case 12: return (&gotoblas_NEOVERSEN2);
case 13: return (&gotoblas_THUNDERX3T110);
case 14: return (&gotoblas_CORTEXA55);
case 15: return (&gotoblas_ARMV8SVE);
case 12: return (&gotoblas_NEOVERSEV2);
case 13: return (&gotoblas_NEOVERSEN2);
case 14: return (&gotoblas_THUNDERX3T110);
case 15: return (&gotoblas_CORTEXA55);
case 16: return (&gotoblas_ARMV8SVE);
}
snprintf(message, 128, "Core not found: %s\n", coretype);
openblas_warning(1, message);
Expand Down Expand Up @@ -312,6 +316,13 @@ static gotoblas_t *get_coretype(void) {
return &gotoblas_NEOVERSEN1;
}else
return &gotoblas_NEOVERSEV1;
case 0xd4f:
if (!(getauxval(AT_HWCAP) & HWCAP_SVE)) {
openblas_warning(FALLBACK_VERBOSE, NEOVERSEN1_FALLBACK);
return &gotoblas_NEOVERSEN1;
} else {
return &gotoblas_NEOVERSEV2;
}
#endif
case 0xd05: // Cortex A55
return &gotoblas_CORTEXA55;
Expand Down