Browse Source

Update dgemm_kernel_8x8_skylakex.c

tags/v0.3.8^2
wjc404 GitHub 6 years ago
parent
commit
0d669e04bb
No known key found for this signature in database GPG Key ID: 4AEE18F83AFDEB23
1 changed files with 4 additions and 0 deletions
  1. +4
    -0
      kernel/x86_64/dgemm_kernel_8x8_skylakex.c

+ 4
- 0
kernel/x86_64/dgemm_kernel_8x8_skylakex.c View File

@@ -1,6 +1,8 @@
#include "common.h"
#include <stdint.h>
#include <immintrin.h>

#define ICOPY_4
//register usage: zmm3 for alpha, zmm4-zmm7 for temporary use, zmm8-zmm31 for accumulators.
/* row-major c_block */
#define INNER_KERNEL_k1m1n8 \
@@ -743,6 +745,7 @@ static void KERNEL_EDGE(double *packed_a, double *packed_b, BLASLONG m, BLASLONG
}
}
}
#ifdef ICOPY_4
static void copy_4_to_8(double *src,double *dst,BLASLONG m,BLASLONG k){
BLASLONG m_count,k_count;double *src1,*dst1,*src2;__m256d tmp;
src1 = src; dst1 = dst; src2 = src1 + 4 * k;
@@ -760,6 +763,7 @@ static void copy_4_to_8(double *src,double *dst,BLASLONG m,BLASLONG k){
}
}
}
#endif
int __attribute__ ((noinline)) CNAME(BLASLONG m, BLASLONG n, BLASLONG k, double alpha, double * __restrict__ A, double * __restrict__ B, double * __restrict__ C, BLASLONG ldc){
if(m==0 || n==0 || k==0 || alpha == 0.0) return 0;
BLASLONG ndiv8 = n/8;double ALPHA = alpha;


Loading…
Cancel
Save