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 /src/clblast_c.cc | |
parent | 39b7dbc5e37829abfbcfb77852b9138b31540b42 (diff) |
Added XOMATCOPY routines to perform out-of-place matrix scaling, copying, and/or transposing
Diffstat (limited to 'src/clblast_c.cc')
-rw-r--r-- | src/clblast_c.cc | 81 |
1 files changed, 81 insertions, 0 deletions
diff --git a/src/clblast_c.cc b/src/clblast_c.cc index 2aac907a..22cb2192 100644 --- a/src/clblast_c.cc +++ b/src/clblast_c.cc @@ -2832,6 +2832,87 @@ StatusCode CLBlastHtrsm(const Layout layout, const Side side, const Triangle tri } // ================================================================================================= +// Extra non-BLAS routines (level-X) +// ================================================================================================= + +// OMATCOPY +StatusCode CLBlastSomatcopy(const Layout layout, const Transpose a_transpose, + const size_t m, const size_t n, + const float 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) { + auto status = clblast::Omatcopy(static_cast<clblast::Layout>(layout), + static_cast<clblast::Transpose>(a_transpose), + m, n, + alpha, + a_buffer, a_offset, a_ld, + b_buffer, b_offset, b_ld, + queue, event); + return static_cast<StatusCode>(status); +} +StatusCode CLBlastDomatcopy(const Layout layout, const Transpose a_transpose, + const size_t m, const size_t n, + const double 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) { + auto status = clblast::Omatcopy(static_cast<clblast::Layout>(layout), + static_cast<clblast::Transpose>(a_transpose), + m, n, + alpha, + a_buffer, a_offset, a_ld, + b_buffer, b_offset, b_ld, + queue, event); + return static_cast<StatusCode>(status); +} +StatusCode CLBlastComatcopy(const Layout layout, const Transpose a_transpose, + const size_t m, const size_t n, + const cl_float2 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) { + auto status = clblast::Omatcopy(static_cast<clblast::Layout>(layout), + static_cast<clblast::Transpose>(a_transpose), + m, n, + float2{alpha.s[0], alpha.s[1]}, + a_buffer, a_offset, a_ld, + b_buffer, b_offset, b_ld, + queue, event); + return static_cast<StatusCode>(status); +} +StatusCode CLBlastZomatcopy(const Layout layout, const Transpose a_transpose, + const size_t m, const size_t n, + const cl_double2 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) { + auto status = clblast::Omatcopy(static_cast<clblast::Layout>(layout), + static_cast<clblast::Transpose>(a_transpose), + m, n, + double2{alpha.s[0], alpha.s[1]}, + a_buffer, a_offset, a_ld, + b_buffer, b_offset, b_ld, + queue, event); + return static_cast<StatusCode>(status); +} +StatusCode CLBlastHomatcopy(const Layout layout, const Transpose a_transpose, + const size_t m, const size_t n, + const cl_half 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) { + auto status = clblast::Omatcopy(static_cast<clblast::Layout>(layout), + static_cast<clblast::Transpose>(a_transpose), + m, n, + alpha, + a_buffer, a_offset, a_ld, + b_buffer, b_offset, b_ld, + queue, event); + return static_cast<StatusCode>(status); +} + +// ================================================================================================= // Clears the cache of stored binaries StatusCode CLBlastClearCache() { |