Skip to content

Commit 52ec7fa

Browse files
authored
Merge pull request #5554 from hideaki-motoki/issue5553_gemm_default_pqr_for_a64fx
Setting optimized `[SD]GEMM_DEFAULT_[PQR]` parameters for `A64FX`
2 parents 1ffea2b + 5f07358 commit 52ec7fa

File tree

1 file changed

+6
-6
lines changed

1 file changed

+6
-6
lines changed

param.h

Lines changed: 6 additions & 6 deletions
Original file line numberDiff line numberDiff line change
@@ -3778,18 +3778,18 @@ Until then, just keep it different than DGEMM_DEFAULT_UNROLL_N to keep copy rout
37783778
#define ZGEMM_DEFAULT_UNROLL_N 4
37793779
#define ZGEMM_DEFAULT_UNROLL_MN 16
37803780

3781-
#define SGEMM_DEFAULT_P 128
3782-
#define DGEMM_DEFAULT_P 160
3781+
#define SGEMM_DEFAULT_P 128
3782+
#define DGEMM_DEFAULT_P 128
37833783
#define CGEMM_DEFAULT_P 128
37843784
#define ZGEMM_DEFAULT_P 128
37853785

3786-
#define SGEMM_DEFAULT_Q 352
3787-
#define DGEMM_DEFAULT_Q 128
3786+
#define SGEMM_DEFAULT_Q 896
3787+
#define DGEMM_DEFAULT_Q 448
37883788
#define CGEMM_DEFAULT_Q 224
37893789
#define ZGEMM_DEFAULT_Q 112
37903790

3791-
#define SGEMM_DEFAULT_R 4096
3792-
#define DGEMM_DEFAULT_R 4096
3791+
#define SGEMM_DEFAULT_R 3072
3792+
#define DGEMM_DEFAULT_R 3072
37933793
#define CGEMM_DEFAULT_R 4096
37943794
#define ZGEMM_DEFAULT_R 4096
37953795

0 commit comments

Comments
 (0)