power9 zgemm ztrmm optimized

This commit is contained in:
AbdelRauf 2019-06-05 10:30:57 +00:00
parent a469b32cf4
commit d0c3543c3f
5 changed files with 2671 additions and 1379 deletions

View File

@ -6,7 +6,7 @@
STRMMKERNEL = sgemm_kernel_power9.S STRMMKERNEL = sgemm_kernel_power9.S
DTRMMKERNEL = dgemm_kernel_power9.S DTRMMKERNEL = dgemm_kernel_power9.S
CTRMMKERNEL = ctrmm_kernel_8x4_power8.S CTRMMKERNEL = ctrmm_kernel_8x4_power8.S
ZTRMMKERNEL = ztrmm_kernel_8x2_power8.S ZTRMMKERNEL = zgemm_kernel_power9.S
SGEMMKERNEL = sgemm_kernel_power9.S SGEMMKERNEL = sgemm_kernel_power9.S
SGEMMINCOPY = ../generic/gemm_ncopy_16.c SGEMMINCOPY = ../generic/gemm_ncopy_16.c

View File

@ -63,7 +63,7 @@ USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
#define T8 r16 #define T8 r16
#define T5 r17 #define T5 r17
#define T2 r19 #define T2 r19
#define T9 r20 #define TEMP_REG r20
#define T6 r21 #define T6 r21
#define I r22 #define I r22
#define J r23 #define J r23

File diff suppressed because it is too large Load Diff

File diff suppressed because it is too large Load Diff

View File

@ -2256,7 +2256,7 @@ USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
#define SGEMM_DEFAULT_Q 1025 #define SGEMM_DEFAULT_Q 1025
#define DGEMM_DEFAULT_Q 384 #define DGEMM_DEFAULT_Q 384
#define CGEMM_DEFAULT_Q 640 #define CGEMM_DEFAULT_Q 640
#define ZGEMM_DEFAULT_Q 1025 #define ZGEMM_DEFAULT_Q 1026
#define SYMV_P 8 #define SYMV_P 8