floraachy
/
OpenBLAS
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

fix defines for CORTEX-X

This commit is contained in:
Martin Kroeker 2022-03-28 17:37:06 +02:00 committed by GitHub
parent 48e421934f
commit d93cf7f23c
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
1 changed files with 3 additions and 3 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

6
param.h
View File

@ -3128,7 +3128,7 @@ USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
#define SYMV_P 16 #define SYMV_P 16
#if defined(CORTEXA57) || \ #if defined(CORTEXA57) || defined(CORTEXX1) \
defined(CORTEXA72) || defined(CORTEXA73) || \ defined(CORTEXA72) || defined(CORTEXA73) || \
defined(FALKOR) || defined(TSV110) || defined(EMAG8180) || defined(VORTEX) || defined(FT2000) defined(FALKOR) || defined(TSV110) || defined(EMAG8180) || defined(VORTEX) || defined(FT2000)
@ -3147,7 +3147,7 @@ USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
/*FIXME: this should be using the cache size, but there is currently no easy way to /*FIXME: this should be using the cache size, but there is currently no easy way to
query that on ARM. So if getarch counted more than 8 cores we simply assume the host query that on ARM. So if getarch counted more than 8 cores we simply assume the host
is a big desktop or server with abundant cache rather than a phone or embedded device */ is a big desktop or server with abundant cache rather than a phone or embedded device */
#if NUM_CORES > 8 || defined(TSV110) || defined(EMAG8180) || defined(VORTEX) #if NUM_CORES > 8 || defined(TSV110) || defined(EMAG8180) || defined(VORTEX)|| defined(CORTEXX1)
#define SGEMM_DEFAULT_P 512 #define SGEMM_DEFAULT_P 512
#define DGEMM_DEFAULT_P 256 #define DGEMM_DEFAULT_P 256
#define CGEMM_DEFAULT_P 256 #define CGEMM_DEFAULT_P 256
@ -3377,7 +3377,7 @@ is a big desktop or server with abundant cache rather than a phone or embedded d
#define CGEMM_DEFAULT_R 4096 #define CGEMM_DEFAULT_R 4096
#define ZGEMM_DEFAULT_R 4096 #define ZGEMM_DEFAULT_R 4096
#elif defined(ARMV8SVE) || defined(A64FX) || defined(ARMV9) || defined(CORTEXA510) #elif defined(ARMV8SVE) || defined(A64FX) || defined(ARMV9) || defined(CORTEXA510)|| defined(CORTEXA710) || defined(CORTEXX2)
/* When all BLAS3 routines are implemeted with SVE, SGEMM_DEFAULT_UNROLL_M should be "sve_vl". /* When all BLAS3 routines are implemeted with SVE, SGEMM_DEFAULT_UNROLL_M should be "sve_vl".
Until then, just keep it different than DGEMM_DEFAULT_UNROLL_N to keep copy routines in both directions seperated. */ Until then, just keep it different than DGEMM_DEFAULT_UNROLL_N to keep copy routines in both directions seperated. */