Browse Source

Ref #458: Backport, sandybrigde uses nehalem zgemm kernel

tags/v0.2.14^2
Werner Saar 10 years ago
parent
commit
587e16fba3
2 changed files with 7 additions and 7 deletions
  1. +6
    -6
      kernel/x86_64/KERNEL.SANDYBRIDGE
  2. +1
    -1
      param.h

+ 6
- 6
kernel/x86_64/KERNEL.SANDYBRIDGE View File

@@ -34,17 +34,17 @@ CGEMMITCOPYOBJ = cgemm_itcopy$(TSUFFIX).$(SUFFIX)
CGEMMONCOPYOBJ = cgemm_oncopy$(TSUFFIX).$(SUFFIX) CGEMMONCOPYOBJ = cgemm_oncopy$(TSUFFIX).$(SUFFIX)
CGEMMOTCOPYOBJ = cgemm_otcopy$(TSUFFIX).$(SUFFIX) CGEMMOTCOPYOBJ = cgemm_otcopy$(TSUFFIX).$(SUFFIX)



ZGEMMKERNEL = zgemm_kernel_4x4_sandy.S
ZGEMMINCOPY =
ZGEMMITCOPY =
ZGEMMKERNEL = zgemm_kernel_1x4_nehalem.S
ZGEMMINCOPY = zgemm_ncopy_1.S
ZGEMMITCOPY = zgemm_tcopy_1.S
ZGEMMONCOPY = ../generic/zgemm_ncopy_4.c ZGEMMONCOPY = ../generic/zgemm_ncopy_4.c
ZGEMMOTCOPY = ../generic/zgemm_tcopy_4.c ZGEMMOTCOPY = ../generic/zgemm_tcopy_4.c
ZGEMMINCOPYOBJ =
ZGEMMITCOPYOBJ =
ZGEMMINCOPYOBJ = zgemm_incopy$(TSUFFIX).$(SUFFIX)
ZGEMMITCOPYOBJ = zgemm_itcopy$(TSUFFIX).$(SUFFIX)
ZGEMMONCOPYOBJ = zgemm_oncopy$(TSUFFIX).$(SUFFIX) ZGEMMONCOPYOBJ = zgemm_oncopy$(TSUFFIX).$(SUFFIX)
ZGEMMOTCOPYOBJ = zgemm_otcopy$(TSUFFIX).$(SUFFIX) ZGEMMOTCOPYOBJ = zgemm_otcopy$(TSUFFIX).$(SUFFIX)



#STRSMKERNEL_LN = trsm_kernel_LN_4x8_nehalem.S #STRSMKERNEL_LN = trsm_kernel_LN_4x8_nehalem.S
#STRSMKERNEL_LT = trsm_kernel_LT_4x8_nehalem.S #STRSMKERNEL_LT = trsm_kernel_LT_4x8_nehalem.S
#STRSMKERNEL_RN = trsm_kernel_LT_4x8_nehalem.S #STRSMKERNEL_RN = trsm_kernel_LT_4x8_nehalem.S


+ 1
- 1
param.h View File

@@ -1129,7 +1129,7 @@ USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
#define DGEMM_DEFAULT_UNROLL_M 8 #define DGEMM_DEFAULT_UNROLL_M 8
#define QGEMM_DEFAULT_UNROLL_M 2 #define QGEMM_DEFAULT_UNROLL_M 2
#define CGEMM_DEFAULT_UNROLL_M 8 #define CGEMM_DEFAULT_UNROLL_M 8
#define ZGEMM_DEFAULT_UNROLL_M 4
#define ZGEMM_DEFAULT_UNROLL_M 1
#define XGEMM_DEFAULT_UNROLL_M 1 #define XGEMM_DEFAULT_UNROLL_M 1


#define SGEMM_DEFAULT_UNROLL_N 4 #define SGEMM_DEFAULT_UNROLL_N 4


Loading…
Cancel
Save