diff options
author | Cedric Nugteren <web@cedricnugteren.nl> | 2016-08-20 12:50:31 +0200 |
---|---|---|
committer | Cedric Nugteren <web@cedricnugteren.nl> | 2016-08-20 12:50:31 +0200 |
commit | 6eca53ee235e8105a2822f725beb23d4ab7a7bd6 (patch) | |
tree | 03c49539feb9b34d149714a49a2f35292cf7f29a /src/kernels/level1/xnrm2.opencl | |
parent | 1ec21421d7dedd715badbc7cf8f555d3b76a022b (diff) | |
parent | 57f1aa76857cf0566e05b43b9b2a98a3a6139c8b (diff) |
Merge branch 'master' of https://github.com/dvasschemacq/CLBlast into dvasschemacq-master
Conflicts:
src/kernels/level1/xaxpy.opencl
src/kernels/level2/xgemv.opencl
src/kernels/level2/xgemv_fast.opencl
src/kernels/level2/xger.opencl
src/kernels/level2/xher.opencl
src/kernels/level2/xher2.opencl
src/kernels/level3/xgemm_part2.opencl
Diffstat (limited to 'src/kernels/level1/xnrm2.opencl')
-rw-r--r-- | src/kernels/level1/xnrm2.opencl | 14 |
1 files changed, 7 insertions, 7 deletions
diff --git a/src/kernels/level1/xnrm2.opencl b/src/kernels/level1/xnrm2.opencl index 9803687a..f6d869cb 100644 --- a/src/kernels/level1/xnrm2.opencl +++ b/src/kernels/level1/xnrm2.opencl @@ -30,10 +30,10 @@ R"( // ================================================================================================= // The main reduction kernel, performing the multiplication and the majority of the operation -__attribute__((reqd_work_group_size(WGS1, 1, 1))) -__kernel void Xnrm2(const int n, - const __global real* restrict xgm, const int x_offset, const int x_inc, - __global real* output) { +__kernel __attribute__((reqd_work_group_size(WGS1, 1, 1))) +void Xnrm2(const int n, + const __global real* restrict xgm, const int x_offset, const int x_inc, + __global real* output) { __local real lm[WGS1]; const int lid = get_local_id(0); const int wgid = get_group_id(0); @@ -72,9 +72,9 @@ __kernel void Xnrm2(const int n, // The epilogue reduction kernel, performing the final bit of the operation. This kernel has to // be launched with a single workgroup only. -__attribute__((reqd_work_group_size(WGS2, 1, 1))) -__kernel void Xnrm2Epilogue(const __global real* restrict input, - __global real* nrm2, const int nrm2_offset) { +__kernel __attribute__((reqd_work_group_size(WGS2, 1, 1))) +void Xnrm2Epilogue(const __global real* restrict input, + __global real* nrm2, const int nrm2_offset) { __local real lm[WGS2]; const int lid = get_local_id(0); |