diff options
author | cnugteren <web@cedricnugteren.nl> | 2016-03-30 21:37:56 -0700 |
---|---|---|
committer | cnugteren <web@cedricnugteren.nl> | 2016-03-30 21:37:56 -0700 |
commit | 8c3c6db7d07adaacb702fdaabfdf18f74fbfea13 (patch) | |
tree | f6dcd3f9d4f987ec74f87b1939c4b3600a7d42d0 /src/clblast.cc | |
parent | 6578102ae996ce0aa52b45704f38c1cd5a10d3c0 (diff) | |
parent | 5409f349a17f60ba68133fd0cc9789fb2918f790 (diff) |
Merge branch 'level1_routines' into development
Diffstat (limited to 'src/clblast.cc')
-rw-r--r-- | src/clblast.cc | 33 |
1 files changed, 33 insertions, 0 deletions
diff --git a/src/clblast.cc b/src/clblast.cc index 1b2c3a12..fc50ffae 100644 --- a/src/clblast.cc +++ b/src/clblast.cc @@ -26,6 +26,7 @@ #include "internal/routines/level1/xdot.h" #include "internal/routines/level1/xdotu.h" #include "internal/routines/level1/xdotc.h" +#include "internal/routines/level1/xnrm2.h" // BLAS level-2 includes #include "internal/routines/level2/xgemv.h" @@ -371,6 +372,38 @@ template StatusCode PUBLIC_API Dotc<double2>(const size_t, const cl_mem, const size_t, const size_t, cl_command_queue*, cl_event*); +// Euclidian norm of a vector: SNRM2/DNRM2/ScNRM2/DzNRM2 +template <typename T> +StatusCode Nrm2(const size_t n, + cl_mem nrm2_buffer, const size_t nrm2_offset, + const cl_mem x_buffer, const size_t x_offset, const size_t x_inc, + cl_command_queue* queue, cl_event* event) { + auto queue_cpp = Queue(*queue); + auto event_cpp = Event(event); + auto routine = Xnrm2<T>(queue_cpp, event_cpp); + auto status = routine.SetUp(); + if (status != StatusCode::kSuccess) { return status; } + return routine.DoNrm2(n, + Buffer<T>(nrm2_buffer), nrm2_offset, + Buffer<T>(x_buffer), x_offset, x_inc); +} +template StatusCode PUBLIC_API Nrm2<float>(const size_t, + cl_mem, const size_t, + const cl_mem, const size_t, const size_t, + cl_command_queue*, cl_event*); +template StatusCode PUBLIC_API Nrm2<double>(const size_t, + cl_mem, const size_t, + const cl_mem, const size_t, const size_t, + cl_command_queue*, cl_event*); +template StatusCode PUBLIC_API Nrm2<float2>(const size_t, + cl_mem, const size_t, + const cl_mem, const size_t, const size_t, + cl_command_queue*, cl_event*); +template StatusCode PUBLIC_API Nrm2<double2>(const size_t, + cl_mem, const size_t, + const cl_mem, const size_t, const size_t, + cl_command_queue*, cl_event*); + // ================================================================================================= // BLAS level-2 (matrix-vector) routines // ================================================================================================= |