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

sbgemm: cooperlake: fix bug in m64n12

This commit is contained in:
Wangyang Guo 2021-08-17 21:13:29 +08:00
parent 8356a604f0
commit cece3541ab
1 changed files with 1 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

3
kernel/x86_64/sbgemm_kernel_16x4_cooperlake.c
View File

@ -306,9 +306,8 @@ int CNAME (BLASLONG m, BLASLONG n, BLASLONG k, FLOAT alpha, IFLOAT * A, IFLOAT *
STORE_4X(0, 0, 1); STORE_4X(1, 0, 1); STORE_4X(0, 0, 1); STORE_4X(1, 0, 1);
STORE_4X(0, 0, 2); STORE_4X(1, 0, 2); STORE_4X(0, 0, 2); STORE_4X(1, 0, 2);
ptr_c += 16 * 2; ptr_c += 16 * 2;
}
if (m > 31) {
ptr_a0 = ptr_a1; ptr_a0 = ptr_a1;
ptr_a1 = ptr_a0 + 16 * k;
} }
for (; m_count > 15; m_count -= 16) { for (; m_count > 15; m_count -= 16) {
ptr_b0 = ptr_b00; ptr_b0 = ptr_b00;