From 6397e6174696b1a2ac6f4f0698e0f67e48891721 Mon Sep 17 00:00:00 2001 From: Cedric Nugteren Date: Fri, 9 Mar 2018 21:40:22 +0100 Subject: Added several more tuner API functions --- include/clblast.h | 32 ++++++++++++++++++++++++++++++-- 1 file changed, 30 insertions(+), 2 deletions(-) (limited to 'include') diff --git a/include/clblast.h b/include/clblast.h index d6118e19..c5d2ddc2 100644 --- a/include/clblast.h +++ b/include/clblast.h @@ -706,8 +706,36 @@ StatusCode PUBLIC_API OverrideParameters(const cl_device_id device, const std::s // ================================================================================================= template -StatusCode PUBLIC_API TuneCopyMatrixFast(cl_command_queue* queue, const size_t m, const size_t n, - const double fraction, std::unordered_map ¶meters); +StatusCode PUBLIC_API TuneXaxpy(cl_command_queue* queue, const size_t n, + const double fraction, std::unordered_map ¶meters); + +template +StatusCode PUBLIC_API TuneXdot(cl_command_queue* queue, const size_t n, + const double fraction, std::unordered_map ¶meters); + +template +StatusCode PUBLIC_API TuneXgemv(cl_command_queue* queue, const size_t m, const size_t n, + const double fraction, std::unordered_map ¶meters); + +template +StatusCode PUBLIC_API TuneXger(cl_command_queue* queue, const size_t m, const size_t n, + const double fraction, std::unordered_map ¶meters); + +template +StatusCode PUBLIC_API TuneCopy(cl_command_queue* queue, const size_t m, const size_t n, + const double fraction, std::unordered_map ¶meters); + +template +StatusCode PUBLIC_API TunePad(cl_command_queue* queue, const size_t m, const size_t n, + const double fraction, std::unordered_map ¶meters); + +template +StatusCode PUBLIC_API TuneTranspose(cl_command_queue* queue, const size_t m, const size_t n, + const double fraction, std::unordered_map ¶meters); + +template +StatusCode PUBLIC_API TunePadtranspose(cl_command_queue* queue, const size_t m, const size_t n, + const double fraction, std::unordered_map ¶meters); // ================================================================================================= -- cgit v1.2.3