Skip to content

Commit 0c2ac76

Browse files
authored
Merge pull request #4734 from XiWeiGu/loongarch64_small_matrix
LoongArch: DGEMM small matrix opt
2 parents 913be34 + 8ab2e9e commit 0c2ac76

7 files changed

+2279
-0
lines changed

Makefile.system

Lines changed: 3 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -269,6 +269,9 @@ else ifeq ($(ARCH), power)
269269
SMALL_MATRIX_OPT = 1
270270
BUILD_BFLOAT16 = 1
271271
endif
272+
ifeq ($(ARCH), loongarch64)
273+
SMALL_MATRIX_OPT = 1
274+
endif
272275
ifeq ($(SMALL_MATRIX_OPT), 1)
273276
CCOMMON_OPT += -DSMALL_MATRIX_OPT
274277
endif

kernel/loongarch64/KERNEL.LOONGSON3R5

Lines changed: 10 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -162,4 +162,14 @@ STRSMKERNEL_LN = ../generic/trsm_kernel_LN.c
162162
STRSMKERNEL_LT = ../generic/trsm_kernel_LT.c
163163
STRSMKERNEL_RN = ../generic/trsm_kernel_RN.c
164164
STRSMKERNEL_RT = ../generic/trsm_kernel_RT.c
165+
166+
DGEMM_SMALL_M_PERMIT = dgemm_small_matrix_permit.c
167+
DGEMM_SMALL_K_NN = dgemm_small_kernel_nn_lasx.S
168+
DGEMM_SMALL_K_B0_NN = dgemm_small_kernel_nn_lasx.S
169+
DGEMM_SMALL_K_NT = dgemm_small_kernel_nt_lasx.S
170+
DGEMM_SMALL_K_B0_NT = dgemm_small_kernel_nt_lasx.S
171+
DGEMM_SMALL_K_TN = dgemm_small_kernel_tn_lasx.S
172+
DGEMM_SMALL_K_B0_TN = dgemm_small_kernel_tn_lasx.S
173+
DGEMM_SMALL_K_TT = dgemm_small_kernel_tt_lasx.S
174+
DGEMM_SMALL_K_B0_TT = dgemm_small_kernel_tt_lasx.S
165175
endif

0 commit comments

Comments
 (0)