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

sgemm_direct_skylakex: fix 75eeb26 regression. 

The
`#if defined(SKYLAKEX) || defined (COOPERLAKE)`
from that commit was before #include "common.h" so caused the
compiled function to be empty, returning garbage results for
qualifying sgemm's on those architectures.

Closes #2914
This commit is contained in:
Bart Oldeman 2020-10-18 19:50:38 +00:00
parent efd47b0104
commit 03e781b766
1 changed files with 2 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

3
kernel/x86_64/sgemm_direct_skylakex.c
View File

@ -1,7 +1,8 @@
#if defined(SKYLAKEX) || defined (COOPERLAKE)
/* the direct sgemm code written by Arjan van der Ven */ /* the direct sgemm code written by Arjan van der Ven */
#include <immintrin.h> #include <immintrin.h>
#include "common.h" #include "common.h"
#if defined(SKYLAKEX) || defined (COOPERLAKE)
/* /*
* "Direct sgemm" code. This code operates directly on the inputs and outputs * "Direct sgemm" code. This code operates directly on the inputs and outputs
* of the sgemm call, avoiding the copies, memory realignments and threading, * of the sgemm call, avoiding the copies, memory realignments and threading,