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

Fixed #27. Temporarily walk around axpy's low performance issue with small imput size & multithreads.

This commit is contained in:
Xianyi Zhang 2011-06-19 11:55:29 +08:00
parent b3d1887745
commit aeed8d6225
2 changed files with 6 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

1
Changelog.txt
View File

@ -25,6 +25,7 @@ x86/x86_64:
* Fixed #28 a wrong result of dsdot on x86_64. * Fixed #28 a wrong result of dsdot on x86_64.
* Fixed #32 a SEGFAULT bug of zdotc with gcc-4.6. * Fixed #32 a SEGFAULT bug of zdotc with gcc-4.6.
* Fixed #33 ztrmm bug on Nehalem. * Fixed #33 ztrmm bug on Nehalem.
* Walk round #27 the low performance axpy issue with small imput size & multithreads.
MIPS64: MIPS64:
* Fixed #28 a wrong result of dsdot on Loongson3A/MIPS64. * Fixed #28 a wrong result of dsdot on Loongson3A/MIPS64.

6
interface/axpy.c
View File

@ -85,7 +85,11 @@ void CNAME(blasint n, FLOAT alpha, FLOAT *x, blasint incx, FLOAT *y, blasint inc
//In that case, the threads would be dependent. //In that case, the threads would be dependent.
if (incx == 0 || incy == 0) if (incx == 0 || incy == 0)
nthreads = 1; nthreads = 1;
//Temporarily walk around the low performance issue with small imput size & multithreads.
if (n <= 10000)
nthreads = 1;
if (nthreads == 1) { if (nthreads == 1) {
#endif #endif