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

Modifying a couple paramaters in the "POWER10"-specific section of param.h, for performance enhancements for SGEMM and DGEMM.

This commit is contained in:
austinpagan 2021-03-10 18:19:12 -05:00
parent ef0238ba2b
commit 9579bd47e5
1 changed files with 4 additions and 4 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

8
param.h
View File

@ -2455,13 +2455,13 @@ USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
#define ZGEMM_DEFAULT_UNROLL_M 8
#define ZGEMM_DEFAULT_UNROLL_N 2
#define SGEMM_DEFAULT_P 832
#define DGEMM_DEFAULT_P 320
#define SGEMM_DEFAULT_P 512
#define DGEMM_DEFAULT_P 384
#define CGEMM_DEFAULT_P 512
#define ZGEMM_DEFAULT_P 256
#define SGEMM_DEFAULT_Q 1026
#define DGEMM_DEFAULT_Q 960
#define SGEMM_DEFAULT_Q 512
#define DGEMM_DEFAULT_Q 512
#define CGEMM_DEFAULT_Q 1026
#define ZGEMM_DEFAULT_Q 1026