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

Disable FMA intrinsics in the srot kernel when the compiler is PGI/NVIDIA

This commit is contained in:
Martin Kroeker 2020-12-19 22:15:58 +01:00 committed by GitHub
parent b37e5fa2f8
commit 114eb159a4
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
1 changed files with 1 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
kernel/x86_64/srot.c
View File

@ -13,7 +13,7 @@ static void srot_kernel(BLASLONG n, FLOAT *x, FLOAT *y, FLOAT c, FLOAT s)
{
BLASLONG i = 0;
#if V_SIMD && (defined(HAVE_FMA3) || V_SIMD > 128)
#if V_SIMD && !defined(C_PGI) && (defined(HAVE_FMA3) || V_SIMD > 128)
const int vstep = v_nlanes_f32;
const int unrollx4 = n & (-vstep * 4);
const int unrollx = n & -vstep;