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

bugfix for sbgemm_ncopy_8_neoversen2

This commit is contained in:
lilianhuang 2022-12-05 05:10:18 -05:00
parent fdac8a97c1
commit 729af6406f
1 changed files with 2 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
kernel/arm64/sbgemm_ncopy_8_neoversen2.c
View File

@ -83,13 +83,13 @@ int CNAME(BLASLONG m, BLASLONG n, IFLOAT *a, BLASLONG lda, IFLOAT *b) {
if (m & 3) { if (m & 3) {
BLASLONG rest = m & 3; BLASLONG rest = m & 3;
for (BLASLONG col = 0; col < 4; col++) { for (BLASLONG col = 0; col < 8; col++) {
b_offset[4 * col] = a_offsetx[col][0]; b_offset[4 * col] = a_offsetx[col][0];
b_offset[4 * col + 1] = rest == 1 ? 0 : a_offsetx[col][1]; b_offset[4 * col + 1] = rest == 1 ? 0 : a_offsetx[col][1];
b_offset[4 * col + 2] = rest <= 2 ? 0 : a_offsetx[col][2]; b_offset[4 * col + 2] = rest <= 2 ? 0 : a_offsetx[col][2];
b_offset[4 * col + 3] = rest <= 3 ? 0 : a_offsetx[col][3]; b_offset[4 * col + 3] = rest <= 3 ? 0 : a_offsetx[col][3];
} }
b_offset += 16; b_offset += 32;
} }
} }