diff options
author | Angus, Alexander <aangus@qti.qualcomm.com> | 2023-01-03 10:56:04 -0800 |
---|---|---|
committer | Angus, Alexander <aangus@qti.qualcomm.com> | 2023-01-03 10:56:04 -0800 |
commit | 4f394608a28f419dfd6091c704148d9e638a26f0 (patch) | |
tree | 4c0e042109c4d249ff5b700fc49a862169edec5a /src/kernels/level2/xher2.opencl | |
parent | 03cffa83c5f7742f8ec0c5e762bb7048e38952f3 (diff) |
implemented changes to boost Adreno performance according to https://jira-dc.qualcomm.com/jira/browse/OSR-8731
Diffstat (limited to 'src/kernels/level2/xher2.opencl')
-rw-r--r-- | src/kernels/level2/xher2.opencl | 6 |
1 files changed, 5 insertions, 1 deletions
diff --git a/src/kernels/level2/xher2.opencl b/src/kernels/level2/xher2.opencl index 73305149..c3e85c15 100644 --- a/src/kernels/level2/xher2.opencl +++ b/src/kernels/level2/xher2.opencl @@ -18,7 +18,11 @@ R"( // ================================================================================================= // Symmetric version of the rank-2 matrix update kernel (HER2, HPR2, SYR2, SPR2) -__kernel __attribute__((reqd_work_group_size(WGS1, WGS2, 1))) +#if RELAX_WORKGROUP_SIZE == 1 + __kernel +#elif + __kernel __attribute__((reqd_work_group_size(WGS1, WGS2, 1))) +#endif void Xher2(const int n, const real_arg arg_alpha, const __global real* restrict xgm, const int x_offset, const int x_inc, |