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

modified configuration, to use power6 sgemm kernel for power8

This commit is contained in:
Werner Saar 2016-03-04 13:38:57 +01:00
parent 73f04c2c72
commit 91e1c5080c
2 changed files with 7 additions and 6 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

8
kernel/power/KERNEL.POWER8
View File

@ -3,14 +3,14 @@
#CGEMM_BETA = ../generic/zgemm_beta.c #CGEMM_BETA = ../generic/zgemm_beta.c
#ZGEMM_BETA = ../generic/zgemm_beta.c #ZGEMM_BETA = ../generic/zgemm_beta.c
STRMMKERNEL = ../generic/trmmkernel_2x2.c STRMMKERNEL = gemm_kernel_power6.S
DTRMMKERNEL = dtrmm_kernel_16x4_power8.S DTRMMKERNEL = dtrmm_kernel_16x4_power8.S
CTRMMKERNEL = ../generic/ztrmmkernel_2x2.c CTRMMKERNEL = ../generic/ztrmmkernel_2x2.c
ZTRMMKERNEL = ztrmm_kernel_8x2_power8.S ZTRMMKERNEL = ztrmm_kernel_8x2_power8.S
SGEMMKERNEL = ../generic/gemmkernel_2x2.c SGEMMKERNEL = gemm_kernel_power6.S
SGEMMONCOPY = ../generic/gemm_ncopy_2.c SGEMMONCOPY = ../generic/gemm_ncopy_4.c
SGEMMOTCOPY = ../generic/gemm_tcopy_2.c SGEMMOTCOPY = ../generic/gemm_tcopy_4.c
SGEMMONCOPYOBJ = sgemm_oncopy.o SGEMMONCOPYOBJ = sgemm_oncopy.o
SGEMMOTCOPYOBJ = sgemm_otcopy.o SGEMMOTCOPYOBJ = sgemm_otcopy.o

5
param.h
View File

@ -1968,8 +1968,8 @@ USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
#define GEMM_DEFAULT_OFFSET_B 1024 #define GEMM_DEFAULT_OFFSET_B 1024
#define GEMM_DEFAULT_ALIGN 0x03fffUL #define GEMM_DEFAULT_ALIGN 0x03fffUL
#define SGEMM_DEFAULT_UNROLL_M 2 #define SGEMM_DEFAULT_UNROLL_M 4
#define SGEMM_DEFAULT_UNROLL_N 2 #define SGEMM_DEFAULT_UNROLL_N 4
#define DGEMM_DEFAULT_UNROLL_M 16 #define DGEMM_DEFAULT_UNROLL_M 16
#define DGEMM_DEFAULT_UNROLL_N 4 #define DGEMM_DEFAULT_UNROLL_N 4
#define CGEMM_DEFAULT_UNROLL_M 2 #define CGEMM_DEFAULT_UNROLL_M 2
@ -1987,6 +1987,7 @@ USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
#define CGEMM_DEFAULT_Q 400 #define CGEMM_DEFAULT_Q 400
#define ZGEMM_DEFAULT_Q 360 #define ZGEMM_DEFAULT_Q 360
#define SGEMM_DEFAULT_R 28800
#define DGEMM_DEFAULT_R 14400 #define DGEMM_DEFAULT_R 14400
#define ZGEMM_DEFAULT_R 7200 #define ZGEMM_DEFAULT_R 7200