diff options
author | Cedric Nugteren <web@cedricnugteren.nl> | 2016-06-16 18:07:46 +0200 |
---|---|---|
committer | Cedric Nugteren <web@cedricnugteren.nl> | 2016-06-16 18:07:46 +0200 |
commit | 52ccaf5b25e14c9ce032315e5e96b1f27886d481 (patch) | |
tree | 087288b7aebf2a06ffc4e7dcbcd4353f7a3be6a7 /include/clblast.h | |
parent | 39b7dbc5e37829abfbcfb77852b9138b31540b42 (diff) |
Added XOMATCOPY routines to perform out-of-place matrix scaling, copying, and/or transposing
Diffstat (limited to 'include/clblast.h')
-rw-r--r-- | include/clblast.h | 13 |
1 files changed, 13 insertions, 0 deletions
diff --git a/include/clblast.h b/include/clblast.h index 64b2610a..31a07423 100644 --- a/include/clblast.h +++ b/include/clblast.h @@ -560,6 +560,19 @@ StatusCode Trsm(const Layout layout, const Side side, const Triangle triangle, c cl_command_queue* queue, cl_event* event = nullptr); // ================================================================================================= +// Extra non-BLAS routines (level-X) +// ================================================================================================= + +// Scaling and out-place transpose/copy (non-BLAS function): SOMATCOPY/DOMATCOPY/COMATCOPY/ZOMATCOPY/HOMATCOPY +template <typename T> +StatusCode Omatcopy(const Layout layout, const Transpose a_transpose, + const size_t m, const size_t n, + const T alpha, + const cl_mem a_buffer, const size_t a_offset, const size_t a_ld, + cl_mem b_buffer, const size_t b_offset, const size_t b_ld, + cl_command_queue* queue, cl_event* event = nullptr); + +// ================================================================================================= // CLBlast stores binaries of compiled kernels into a cache in case the same kernel is used later on // for the same device. This cache can be cleared to free up system memory or in case of debugging. |