diff options
| -rw-r--r-- | CHANGELOG | 1 | ||||
| -rw-r--r-- | CMakeLists.txt | 19 | ||||
| -rw-r--r-- | README.md | 2 | ||||
| -rw-r--r-- | include/clblast_c.h | 397 | ||||
| -rw-r--r-- | samples/sgemm.c | 108 | ||||
| -rw-r--r-- | src/clblast_c.cc | 758 |
6 files changed, 1278 insertions, 7 deletions
@@ -2,6 +2,7 @@ Development version (next release) - Now using the Claduc C++11 interface to OpenCL - Removed clBLAS sources, it should now be installed separately for testing +- Added plain C API for increased compatibility (clblast_c.h) - Added level-2 routines: * CHEMV/ZHEMV * SSYMV/DSYMV diff --git a/CMakeLists.txt b/CMakeLists.txt index 8ca225b2..8b598bf8 100644 --- a/CMakeLists.txt +++ b/CMakeLists.txt @@ -11,7 +11,7 @@ # CMake project details cmake_minimum_required(VERSION 2.8.10) -project("clblast" CXX) +project("clblast" C CXX) set(clblast_VERSION_MAJOR 0) set(clblast_VERSION_MINOR 3) set(clblast_VERSION_PATCH 0) @@ -103,7 +103,8 @@ include_directories(${clblast_SOURCE_DIR}/include ${OPENCL_INCLUDE_DIRS}) # Sets the supported routines and the used kernels. New routines and kernels should be added here. set(KERNELS copy pad transpose padtranspose xaxpy xgemv xgemm) -set(SAMPLE_PROGRAMS sgemm) +set(SAMPLE_PROGRAMS_CPP sgemm) +set(SAMPLE_PROGRAMS_C sgemm) set(LEVEL1_ROUTINES xaxpy) set(LEVEL2_ROUTINES xgemv xhemv xsymv) set(LEVEL3_ROUTINES xgemm xsymm xhemm xsyrk xherk xsyr2k xher2k xtrmm) @@ -112,7 +113,7 @@ set(ROUTINES ${LEVEL1_ROUTINES} ${LEVEL2_ROUTINES} ${LEVEL3_ROUTINES}) # ================================================================================================== # Gathers all source-files -set(SOURCES src/clblast.cc src/database.cc src/routine.cc src/utilities.cc) +set(SOURCES src/clblast.cc src/database.cc src/routine.cc src/utilities.cc src/clblast_c.cc) foreach(ROUTINE ${LEVEL1_ROUTINES}) set(SOURCES ${SOURCES} src/routines/level1/${ROUTINE}.cc) endforeach() @@ -130,19 +131,27 @@ target_link_libraries(clblast ${OPENCL_LIBRARIES}) # Installs the library install(TARGETS clblast DESTINATION lib) install(FILES include/clblast.h DESTINATION include) +install(FILES include/clblast_c.h DESTINATION include) # ================================================================================================== # This section contains all the code related to the examples if(SAMPLES) - # Adds sample programs - foreach(SAMPLE ${SAMPLE_PROGRAMS}) + # Adds sample programs (C++) + foreach(SAMPLE ${SAMPLE_PROGRAMS_CPP}) add_executable(sample_${SAMPLE} samples/${SAMPLE}.cc) target_link_libraries(sample_${SAMPLE} clblast ${OPENCL_LIBRARIES}) install(TARGETS sample_${SAMPLE} DESTINATION bin) endforeach() + # Adds sample programs (C) + foreach(SAMPLE ${SAMPLE_PROGRAMS_C}) + add_executable(sample_${SAMPLE}_c samples/${SAMPLE}.c) + target_link_libraries(sample_${SAMPLE}_c clblast ${OPENCL_LIBRARIES}) + install(TARGETS sample_${SAMPLE}_c DESTINATION bin) + endforeach() + endif() # ================================================================================================== @@ -208,8 +208,6 @@ To-do list before release of version 1.0 - Increase the functionality: * Support all routines supported by clBLAS * Allow the user control over events and synchronization - * Add an interface with OpenCL C++ data-types - * Add an old-style C compatible interface * Add half-precision routines (e.g. HGEMM) - Improve host performance: * Allow initialization to pre-compile kernels and store to disk diff --git a/include/clblast_c.h b/include/clblast_c.h new file mode 100644 index 00000000..c25e5880 --- /dev/null +++ b/include/clblast_c.h @@ -0,0 +1,397 @@ + +// ================================================================================================= +// This file is part of the CLBlast project. The project is licensed under Apache Version 2.0. This +// project loosely follows the Google C++ styleguide and uses a tab-size of two spaces and a max- +// width of 100 characters per line. +// +// Author(s): +// Cedric Nugteren <www.cedricnugteren.nl> +// +// This file contains the plain C interface to the CLBlast BLAS routines, the counter-part of the +// normal 'clblast.h' C++ header. +// +// ================================================================================================= + +#ifndef CLBLAST_CLBLAST_C_H_ +#define CLBLAST_CLBLAST_C_H_ + +// Includes the normal OpenCL C header +#if defined(__APPLE__) || defined(__MACOSX) + #include <OpenCL/opencl.h> +#else + #include <CL/opencl.h> +#endif + +// ================================================================================================= + +// Status codes. These codes can be returned by functions declared in this header file. The error +// codes match either the standard OpenCL error codes or the clBLAS error codes. +typedef enum StatusCode_ { + + // Status codes in common with the OpenCL standard + kSuccess = 0, // CL_SUCCESS + kTempBufferAllocFailure = -4, // CL_MEM_OBJECT_ALLOCATION_FAILURE + kBuildProgramFailure = -11, // CL_BUILD_PROGRAM_FAILURE: OpenCL compilation error + kInvalidBinary = -42, // CL_INVALID_BINARY + kInvalidKernel = -48, // CL_INVALID_KERNEL + kInvalidLocalNumDimensions = -53, // CL_INVALID_WORK_DIMENSION: Too many thread dimensions + kInvalidLocalThreadsTotal = -54, // CL_INVALID_WORK_GROUP_SIZE: Too many threads in total + kInvalidLocalThreadsDim = -55, // CL_INVALID_WORK_ITEM_SIZE: ... or for a specific dimension + kInvalidTempBufferSize = -61, // CL_INVALID_BUFFER_SIZE + + // Status codes in common with the clBLAS library + kNotImplemented = -1024, // Routine or functionality not implemented yet + kInvalidMatrixA = -1022, // Matrix A is not a valid OpenCL buffer + kInvalidMatrixB = -1021, // Matrix B is not a valid OpenCL buffer + kInvalidMatrixC = -1020, // Matrix C is not a valid OpenCL buffer + kInvalidVectorX = -1019, // Vector X is not a valid OpenCL buffer + kInvalidVectorY = -1018, // Vector Y is not a valid OpenCL buffer + kInvalidDimension = -1017, // Dimensions M, N, and K have to be larger than zero + kInvalidLeadDimA = -1016, // LD of A is smaller than the matrix's first dimension + kInvalidLeadDimB = -1015, // LD of B is smaller than the matrix's first dimension + kInvalidLeadDimC = -1014, // LD of C is smaller than the matrix's first dimension + kInvalidIncrementX = -1013, // Increment of vector X cannot be zero + kInvalidIncrementY = -1012, // Increment of vector Y cannot be zero + kInsufficientMemoryA = -1011, // Matrix A's OpenCL buffer is too small + kInsufficientMemoryB = -1010, // Matrix B's OpenCL buffer is too small + kInsufficientMemoryC = -1009, // Matrix C's OpenCL buffer is too small + kInsufficientMemoryX = -1008, // Vector X's OpenCL buffer is too small + kInsufficientMemoryY = -1007, // Vector Y's OpenCL buffer is too small + + // Custom additional status codes for CLBlast + kKernelLaunchError = -2048, // Problem occurred when enqueuing the kernel + kKernelRunError = -2047, // Problem occurred while running the kernel + kInvalidLocalMemUsage = -2046, // Not enough local memory available on this device + kNoHalfPrecision = -2045, // Half precision (16-bits) not supported by the device + kNoDoublePrecision = -2044, // Double precision (64-bits) not supported by the device +} StatusCode; + +// Matrix layout and transpose types +typedef enum Layout_ { kRowMajor, kColMajor } Layout; +typedef enum Transpose_ { kNo, kYes, kConjugate } Transpose; +typedef enum Side_ { kLeft, kRight } Side; +typedef enum Triangle_ { kUpper, kLower } Triangle; +typedef enum Diagonal_ { kUnit, kNonUnit } Diagonal; + +// Precision scoped enum (values in bits) +typedef enum Precision_ { kHalf = 16, kSingle = 32, kDouble = 64, + kComplexSingle = 3232, kComplexDouble = 6464 } Precision; + +// ================================================================================================= +// BLAS level-1 (vector-vector) routines +// ================================================================================================= + +// Vector-times-constant plus vector: SAXPY/DAXPY/CAXPY/ZAXPY +StatusCode CLBlastSaxpy(const size_t n, + const float alpha, + const cl_mem x_buffer, const size_t x_offset, const size_t x_inc, + cl_mem y_buffer, const size_t y_offset, const size_t y_inc, + cl_command_queue* queue, cl_event* event); +StatusCode CLBlastDaxpy(const size_t n, + const double alpha, + const cl_mem x_buffer, const size_t x_offset, const size_t x_inc, + cl_mem y_buffer, const size_t y_offset, const size_t y_inc, + cl_command_queue* queue, cl_event* event); +StatusCode CLBlastCaxpy(const size_t n, + const cl_float2 alpha, + const cl_mem x_buffer, const size_t x_offset, const size_t x_inc, + cl_mem y_buffer, const size_t y_offset, const size_t y_inc, + cl_command_queue* queue, cl_event* event); +StatusCode CLBlastZaxpy(const size_t n, + const cl_double2 alpha, + const cl_mem x_buffer, const size_t x_offset, const size_t x_inc, + cl_mem y_buffer, const size_t y_offset, const size_t y_inc, + cl_command_queue* queue, cl_event* event); + +// ================================================================================================= +// BLAS level-2 (matrix-vector) routines +// ================================================================================================= + +// Generalized matrix-vector multiplication: SGEMV/DGEMV/CGEMV/ZGEMV +StatusCode CLBlastSgemv(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, + const cl_mem x_buffer, const size_t x_offset, const size_t x_inc, + const float beta, + cl_mem y_buffer, const size_t y_offset, const size_t y_inc, + cl_command_queue* queue, cl_event* event); +StatusCode CLBlastDgemv(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, + const cl_mem x_buffer, const size_t x_offset, const size_t x_inc, + const double beta, + cl_mem y_buffer, const size_t y_offset, const size_t y_inc, + cl_command_queue* queue, cl_event* event); +StatusCode CLBlastCgemv(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, + const cl_mem x_buffer, const size_t x_offset, const size_t x_inc, + const cl_float2 beta, + cl_mem y_buffer, const size_t y_offset, const size_t y_inc, + cl_command_queue* queue, cl_event* event); +StatusCode CLBlastZgemv(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, + const cl_mem x_buffer, const size_t x_offset, const size_t x_inc, + const cl_double2 beta, + cl_mem y_buffer, const size_t y_offset, const size_t y_inc, + cl_command_queue* queue, cl_event* event); + +// Hermitian matrix-vector multiplication: CHEMV/ZHEMV +StatusCode CLBlastChemv(const Layout layout, const Triangle triangle, + const size_t n, + const cl_float2 alpha, + const cl_mem a_buffer, const size_t a_offset, const size_t a_ld, + const cl_mem x_buffer, const size_t x_offset, const size_t x_inc, + const cl_float2 beta, + cl_mem y_buffer, const size_t y_offset, const size_t y_inc, + cl_command_queue* queue, cl_event* event); +StatusCode CLBlastZhemv(const Layout layout, const Triangle triangle, + const size_t n, + const cl_double2 alpha, + const cl_mem a_buffer, const size_t a_offset, const size_t a_ld, + const cl_mem x_buffer, const size_t x_offset, const size_t x_inc, + const cl_double2 beta, + cl_mem y_buffer, const size_t y_offset, const size_t y_inc, + cl_command_queue* queue, cl_event* event); + +// Symmetric matrix-vector multiplication: SSYMV/DSYMV +StatusCode CLBlastSsymv(const Layout layout, const Triangle triangle, + const size_t n, + const float alpha, + const cl_mem a_buffer, const size_t a_offset, const size_t a_ld, + const cl_mem x_buffer, const size_t x_offset, const size_t x_inc, + const float beta, + cl_mem y_buffer, const size_t y_offset, const size_t y_inc, + cl_command_queue* queue, cl_event* event); +StatusCode CLBlastDsymv(const Layout layout, const Triangle triangle, + const size_t n, + const double alpha, + const cl_mem a_buffer, const size_t a_offset, const size_t a_ld, + const cl_mem x_buffer, const size_t x_offset, const size_t x_inc, + const double beta, + cl_mem y_buffer, const size_t y_offset, const size_t y_inc, + cl_command_queue* queue, cl_event* event); + +// ================================================================================================= +// BLAS level-3 (matrix-matrix) routines +// ================================================================================================= + +// Generalized matrix-matrix multiplication: SGEMM/DGEMM/CGEMM/ZGEMM +StatusCode CLBlastSgemm(const Layout layout, const Transpose a_transpose, const Transpose b_transpose, + const size_t m, const size_t n, const size_t k, + const float alpha, + const cl_mem a_buffer, const size_t a_offset, const size_t a_ld, + const cl_mem b_buffer, const size_t b_offset, const size_t b_ld, + const float beta, + cl_mem c_buffer, const size_t c_offset, const size_t c_ld, + cl_command_queue* queue, cl_event* event); +StatusCode CLBlastDgemm(const Layout layout, const Transpose a_transpose, const Transpose b_transpose, + const size_t m, const size_t n, const size_t k, + const double alpha, + const cl_mem a_buffer, const size_t a_offset, const size_t a_ld, + const cl_mem b_buffer, const size_t b_offset, const size_t b_ld, + const double beta, + cl_mem c_buffer, const size_t c_offset, const size_t c_ld, + cl_command_queue* queue, cl_event* event); +StatusCode CLBlastCgemm(const Layout layout, const Transpose a_transpose, const Transpose b_transpose, + const size_t m, const size_t n, const size_t k, + const cl_float2 alpha, + const cl_mem a_buffer, const size_t a_offset, const size_t a_ld, + const cl_mem b_buffer, const size_t b_offset, const size_t b_ld, + const cl_float2 beta, + cl_mem c_buffer, const size_t c_offset, const size_t c_ld, + cl_command_queue* queue, cl_event* event); +StatusCode CLBlastZgemm(const Layout layout, const Transpose a_transpose, const Transpose b_transpose, + const size_t m, const size_t n, const size_t k, + const cl_double2 alpha, + const cl_mem a_buffer, const size_t a_offset, const size_t a_ld, + const cl_mem b_buffer, const size_t b_offset, const size_t b_ld, + const cl_double2 beta, + cl_mem c_buffer, const size_t c_offset, const size_t c_ld, + cl_command_queue* queue, cl_event* event); + +// Symmetric matrix-matrix multiplication: SSYMM/DSYMM/CSYMM/ZSYMM +StatusCode CLBlastSsymm(const Layout layout, const Side side, const Triangle triangle, + 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, + const cl_mem b_buffer, const size_t b_offset, const size_t b_ld, + const float beta, + cl_mem c_buffer, const size_t c_offset, const size_t c_ld, + cl_command_queue* queue, cl_event* event); +StatusCode CLBlastDsymm(const Layout layout, const Side side, const Triangle triangle, + 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, + const cl_mem b_buffer, const size_t b_offset, const size_t b_ld, + const double beta, + cl_mem c_buffer, const size_t c_offset, const size_t c_ld, + cl_command_queue* queue, cl_event* event); +StatusCode CLBlastCsymm(const Layout layout, const Side side, const Triangle triangle, + 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, + const cl_mem b_buffer, const size_t b_offset, const size_t b_ld, + const cl_float2 beta, + cl_mem c_buffer, const size_t c_offset, const size_t c_ld, + cl_command_queue* queue, cl_event* event); +StatusCode CLBlastZsymm(const Layout layout, const Side side, const Triangle triangle, + 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, + const cl_mem b_buffer, const size_t b_offset, const size_t b_ld, + const cl_double2 beta, + cl_mem c_buffer, const size_t c_offset, const size_t c_ld, + cl_command_queue* queue, cl_event* event); + +// Hermitian matrix-matrix multiplication: CHEMM/ZHEMM +StatusCode CLBlastChemm(const Layout layout, const Side side, const Triangle triangle, + 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, + const cl_mem b_buffer, const size_t b_offset, const size_t b_ld, + const cl_float2 beta, + cl_mem c_buffer, const size_t c_offset, const size_t c_ld, + cl_command_queue* queue, cl_event* event); +StatusCode CLBlastZhemm(const Layout layout, const Side side, const Triangle triangle, + 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, + const cl_mem b_buffer, const size_t b_offset, const size_t b_ld, + const cl_double2 beta, + cl_mem c_buffer, const size_t c_offset, const size_t c_ld, + cl_command_queue* queue, cl_event* event); + +// Rank-K update of a symmetric matrix: SSYRK/DSYRK/CSYRK/ZSYRK +StatusCode CLBlastSsyrk(const Layout layout, const Triangle triangle, const Transpose a_transpose, + const size_t n, const size_t k, + const float alpha, + const cl_mem a_buffer, const size_t a_offset, const size_t a_ld, + const float beta, + cl_mem c_buffer, const size_t c_offset, const size_t c_ld, + cl_command_queue* queue, cl_event* event); +StatusCode CLBlastDsyrk(const Layout layout, const Triangle triangle, const Transpose a_transpose, + const size_t n, const size_t k, + const double alpha, + const cl_mem a_buffer, const size_t a_offset, const size_t a_ld, + const double beta, + cl_mem c_buffer, const size_t c_offset, const size_t c_ld, + cl_command_queue* queue, cl_event* event); +StatusCode CLBlastCsyrk(const Layout layout, const Triangle triangle, const Transpose a_transpose, + const size_t n, const size_t k, + const cl_float2 alpha, + const cl_mem a_buffer, const size_t a_offset, const size_t a_ld, + const cl_float2 beta, + cl_mem c_buffer, const size_t c_offset, const size_t c_ld, + cl_command_queue* queue, cl_event* event); +StatusCode CLBlastZsyrk(const Layout layout, const Triangle triangle, const Transpose a_transpose, + const size_t n, const size_t k, + const cl_double2 alpha, + const cl_mem a_buffer, const size_t a_offset, const size_t a_ld, + const cl_double2 beta, + cl_mem c_buffer, const size_t c_offset, const size_t c_ld, + cl_command_queue* queue, cl_event* event); + +// Rank-K update of a hermitian matrix: CHERK/ZHERK +StatusCode CLBlastCherk(const Layout layout, const Triangle triangle, const Transpose a_transpose, + const size_t n, const size_t k, + const float alpha, + const cl_mem a_buffer, const size_t a_offset, const size_t a_ld, + const float beta, + cl_mem c_buffer, const size_t c_offset, const size_t c_ld, + cl_command_queue* queue, cl_event* event); +StatusCode CLBlastZherk(const Layout layout, const Triangle triangle, const Transpose a_transpose, + const size_t n, const size_t k, + const double alpha, + const cl_mem a_buffer, const size_t a_offset, const size_t a_ld, + const double beta, + cl_mem c_buffer, const size_t c_offset, const size_t c_ld, + cl_command_queue* queue, cl_event* event); + +// Rank-2K update of a symmetric matrix: SSYR2K/DSYR2K/CSYR2K/ZSYR2K +StatusCode CLBlastSsyr2k(const Layout layout, const Triangle triangle, const Transpose ab_transpose, + const size_t n, const size_t k, + const float alpha, + const cl_mem a_buffer, const size_t a_offset, const size_t a_ld, + const cl_mem b_buffer, const size_t b_offset, const size_t b_ld, + const float beta, + cl_mem c_buffer, const size_t c_offset, const size_t c_ld, + cl_command_queue* queue, cl_event* event); +StatusCode CLBlastDsyr2k(const Layout layout, const Triangle triangle, const Transpose ab_transpose, + const size_t n, const size_t k, + const double alpha, + const cl_mem a_buffer, const size_t a_offset, const size_t a_ld, + const cl_mem b_buffer, const size_t b_offset, const size_t b_ld, + const double beta, + cl_mem c_buffer, const size_t c_offset, const size_t c_ld, + cl_command_queue* queue, cl_event* event); +StatusCode CLBlastCsyr2k(const Layout layout, const Triangle triangle, const Transpose ab_transpose, + const size_t n, const size_t k, + const cl_float2 alpha, + const cl_mem a_buffer, const size_t a_offset, const size_t a_ld, + const cl_mem b_buffer, const size_t b_offset, const size_t b_ld, + const cl_float2 beta, + cl_mem c_buffer, const size_t c_offset, const size_t c_ld, + cl_command_queue* queue, cl_event* event); +StatusCode CLBlastZsyr2k(const Layout layout, const Triangle triangle, const Transpose ab_transpose, + const size_t n, const size_t k, + const cl_double2 alpha, + const cl_mem a_buffer, const size_t a_offset, const size_t a_ld, + const cl_mem b_buffer, const size_t b_offset, const size_t b_ld, + const cl_double2 beta, + cl_mem c_buffer, const size_t c_offset, const size_t c_ld, + cl_command_queue* queue, cl_event* event); + +// Rank-2K update of a hermitian matrix: CHER2K/ZHER2K +StatusCode CLBlastCher2k(const Layout layout, const Triangle triangle, const Transpose ab_transpose, + const size_t n, const size_t k, + const cl_float2 alpha, + const cl_mem a_buffer, const size_t a_offset, const size_t a_ld, + const cl_mem b_buffer, const size_t b_offset, const size_t b_ld, + const float beta, + cl_mem c_buffer, const size_t c_offset, const size_t c_ld, + cl_command_queue* queue, cl_event* event); +StatusCode CLBlastZher2k(const Layout layout, const Triangle triangle, const Transpose ab_transpose, + const size_t n, const size_t k, + const cl_double2 alpha, + const cl_mem a_buffer, const size_t a_offset, const size_t a_ld, + const cl_mem b_buffer, const size_t b_offset, const size_t b_ld, + const double beta, + cl_mem c_buffer, const size_t c_offset, const size_t c_ld, + cl_command_queue* queue, cl_event* event); + +// Triangular matrix-matrix multiplication: STRMM/DTRMM/CTRMM/ZTRMM +StatusCode CLBlastStrmm(const Layout layout, const Side side, const Triangle triangle, const Transpose a_transpose, const Diagonal diagonal, + 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); +StatusCode CLBlastDtrmm(const Layout layout, const Side side, const Triangle triangle, const Transpose a_transpose, const Diagonal diagonal, + 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); +StatusCode CLBlastCtrmm(const Layout layout, const Side side, const Triangle triangle, const Transpose a_transpose, const Diagonal diagonal, + 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); +StatusCode CLBlastZtrmm(const Layout layout, const Side side, const Triangle triangle, const Transpose a_transpose, const Diagonal diagonal, + 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); + +// ================================================================================================= + +// CLBLAST_CLBLAST_C_H_ +#endif diff --git a/samples/sgemm.c b/samples/sgemm.c new file mode 100644 index 00000000..f43fb147 --- /dev/null +++ b/samples/sgemm.c @@ -0,0 +1,108 @@ + +// ================================================================================================= +// This file is part of the CLBlast project. The project is licensed under Apache Version 2.0. This +// project loosely follows the Google C++ styleguide and uses a tab-size of two spaces and a max- +// width of 100 characters per line. +// +// Author(s): +// Cedric Nugteren <www.cedricnugteren.nl> +// +// This file demonstrates the use of the SGEMM routine. It is pure C99 and demonstrates the use of +// the C API to the CLBlast library. +// +// Note that this example is meant for illustration purposes only. CLBlast provides other programs +// for performance benchmarking ('client_xxxxx') and for correctness testing ('test_xxxxx'). +// +// ================================================================================================= + +#include <stdio.h> +#include <string.h> + +// Includes the CLBlast library (C interface) +#include <clblast_c.h> + +// ================================================================================================= + +// Example use of the single-precision routine SGEMM +int main(void) { + + // OpenCL platform/device settings + const size_t platform_id = 0; + const size_t device_id = 0; + + // Example SGEMM arguments + const size_t m = 128; + const size_t n = 64; + const size_t k = 512; + const float alpha = 0.7f; + const float beta = 1.0f; + const size_t a_ld = k; + const size_t b_ld = n; + const size_t c_ld = n; + + // Initializes the OpenCL platform + cl_uint num_platforms; + clGetPlatformIDs(0, NULL, &num_platforms); + cl_platform_id* platforms = (cl_platform_id*)malloc(num_platforms*sizeof(cl_platform_id)); + clGetPlatformIDs(num_platforms, platforms, NULL); + cl_platform_id platform = platforms[platform_id]; + + // Initializes the OpenCL device (note: example for GPU devices only) + cl_uint num_devices; + clGetDeviceIDs(platform, CL_DEVICE_TYPE_GPU, 0, NULL, &num_devices); + cl_device_id* devices = (cl_device_id*)malloc(num_devices*sizeof(cl_device_id)); + clGetDeviceIDs(platform, CL_DEVICE_TYPE_GPU, num_devices, devices, NULL); + cl_device_id device = devices[device_id]; + + // Creates the OpenCL context, queue, and an event + cl_context context = clCreateContext(NULL, 1, &device, NULL, NULL, NULL); + cl_command_queue queue = clCreateCommandQueue(context, device, 0, NULL); + cl_event event = NULL; + + // Populate host matrices with some example data + float* host_a = (float*)malloc(sizeof(float)*m*k); + float* host_b = (float*)malloc(sizeof(float)*n*k); + float* host_c = (float*)malloc(sizeof(float)*m*n); + for (size_t i=0; i<m*k; ++i) { host_a[i] = 12.193f; } + for (size_t i=0; i<n*k; ++i) { host_b[i] = -8.199f; } + for (size_t i=0; i<m*n; ++i) { host_c[i] = 0.0f; } + + // Copy the matrices to the device + cl_mem device_a = clCreateBuffer(context, CL_MEM_READ_WRITE, m*k*sizeof(float), NULL, NULL); + cl_mem device_b = clCreateBuffer(context, CL_MEM_READ_WRITE, n*k*sizeof(float), NULL, NULL); + cl_mem device_c = clCreateBuffer(context, CL_MEM_READ_WRITE, m*n*sizeof(float), NULL, NULL); + clEnqueueWriteBuffer(queue, device_a, CL_TRUE, 0, m*k*sizeof(float), host_a, 0, NULL, NULL); + clEnqueueWriteBuffer(queue, device_b, CL_TRUE, 0, n*k*sizeof(float), host_b, 0, NULL, NULL); + clEnqueueWriteBuffer(queue, device_c, CL_TRUE, 0, m*n*sizeof(float), host_c, 0, NULL, NULL); + + // Call the SGEMM routine. + StatusCode status = CLBlastSgemm(kRowMajor, kNo, kNo, + m, n, k, + alpha, + device_a, 0, a_ld, + device_b, 0, b_ld, + beta, + device_c, 0, c_ld, + &queue, &event); + + // Wait for completion + clWaitForEvents(1, &event); + + // Example completed. See "clblast_c.h" for status codes (0 -> success). + printf("Completed with status %d\n", status); + + // Clean-up + free(platforms); + free(devices); + free(host_a); + free(host_b); + free(host_c); + clReleaseMemObject(device_a); + clReleaseMemObject(device_b); + clReleaseMemObject(device_c); + clReleaseCommandQueue(queue); + clReleaseContext(context); + return 0; +} + +// ================================================================================================= diff --git a/src/clblast_c.cc b/src/clblast_c.cc new file mode 100644 index 00000000..3b437aff --- /dev/null +++ b/src/clblast_c.cc @@ -0,0 +1,758 @@ + +// ================================================================================================= +// This file is part of the CLBlast project. The project is licensed under Apache Version 2.0. This +// project loosely follows the Google C++ styleguide and uses a tab-size of two spaces and a max- +// width of 100 characters per line. +// +// Author(s): +// Cedric Nugteren <www.cedricnugteren.nl> +// +// This file implements all the plain C BLAS API calls. This forwards the calls to the C++ API. +// +// ================================================================================================= + +#include <string> + +extern "C" { + #include "clblast_c.h" +} +#include "clblast.h" +#include "internal/utilities.h" + +// ================================================================================================= +// BLAS level-1 (vector-vector) routines +// ================================================================================================= + +// AXPY +StatusCode CLBlastSaxpy(const size_t n, + const float alpha, + const cl_mem x_buffer, const size_t x_offset, const size_t x_inc, + cl_mem y_buffer, const size_t y_offset, const size_t y_inc, + cl_command_queue* queue, cl_event* event) { + auto status = clblast::Axpy(n, + alpha, + x_buffer, x_offset, x_inc, + y_buffer, y_offset, y_inc, + queue, event); + return static_cast<StatusCode>(status); +} +StatusCode CLBlastDaxpy(const size_t n, + const double alpha, + const cl_mem x_buffer, const size_t x_offset, const size_t x_inc, + cl_mem y_buffer, const size_t y_offset, const size_t y_inc, + cl_command_queue* queue, cl_event* event) { + auto status = clblast::Axpy(n, + alpha, + x_buffer, x_offset, x_inc, + y_buffer, y_offset, y_inc, + queue, event); + return static_cast<StatusCode>(status); +} +StatusCode CLBlastCaxpy(const size_t n, + const cl_float2 alpha, + const cl_mem x_buffer, const size_t x_offset, const size_t x_inc, + cl_mem y_buffer, const size_t y_offset, const size_t y_inc, + cl_command_queue* queue, cl_event* event) { + auto status = clblast::Axpy(n, + clblast::float2{alpha.s[0], alpha.s[1]}, + x_buffer, x_offset, x_inc, + y_buffer, y_offset, y_inc, + queue, event); + return static_cast<StatusCode>(status); +} +StatusCode CLBlastZaxpy(const size_t n, + const cl_double2 alpha, + const cl_mem x_buffer, const size_t x_offset, const size_t x_inc, + cl_mem y_buffer, const size_t y_offset, const size_t y_inc, + cl_command_queue* queue, cl_event* event) { + auto status = clblast::Axpy(n, + clblast::double2{alpha.s[0], alpha.s[1]}, + x_buffer, x_offset, x_inc, + y_buffer, y_offset, y_inc, + queue, event); + return static_cast<StatusCode>(status); +} + +// ================================================================================================= +// BLAS level-2 (matrix-vector) routines +// ================================================================================================= + +// GEMV +StatusCode CLBlastSgemv(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, + const cl_mem x_buffer, const size_t x_offset, const size_t x_inc, + const float beta, + cl_mem y_buffer, const size_t y_offset, const size_t y_inc, + cl_command_queue* queue, cl_event* event) { + auto status = clblast::Gemv(static_cast<clblast::Layout>(layout), + static_cast<clblast::Transpose>(a_transpose), + m, n, + alpha, + a_buffer, a_offset, a_ld, + x_buffer, x_offset, x_inc, + beta, + y_buffer, y_offset, y_inc, + queue, event); + return static_cast<StatusCode>(status); +} +StatusCode CLBlastDgemv(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, + const cl_mem x_buffer, const size_t x_offset, const size_t x_inc, + const double beta, + cl_mem y_buffer, const size_t y_offset, const size_t y_inc, + cl_command_queue* queue, cl_event* event) { + auto status = clblast::Gemv(static_cast<clblast::Layout>(layout), + static_cast<clblast::Transpose>(a_transpose), + m, n, + alpha, + a_buffer, a_offset, a_ld, + x_buffer, x_offset, x_inc, + beta, + y_buffer, y_offset, y_inc, + queue, event); + return static_cast<StatusCode>(status); +} +StatusCode CLBlastCgemv(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, + const cl_mem x_buffer, const size_t x_offset, const size_t x_inc, + const cl_float2 beta, + cl_mem y_buffer, const size_t y_offset, const size_t y_inc, + cl_command_queue* queue, cl_event* event) { + auto status = clblast::Gemv(static_cast<clblast::Layout>(layout), + static_cast<clblast::Transpose>(a_transpose), + m, n, + clblast::float2{alpha.s[0], alpha.s[1]}, + a_buffer, a_offset, a_ld, + x_buffer, x_offset, x_inc, + clblast::float2{beta.s[0], beta.s[1]}, + y_buffer, y_offset, y_inc, + queue, event); + return static_cast<StatusCode>(status); +} +StatusCode CLBlastZgemv(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, + const cl_mem x_buffer, const size_t x_offset, const size_t x_inc, + const cl_double2 beta, + cl_mem y_buffer, const size_t y_offset, const size_t y_inc, + cl_command_queue* queue, cl_event* event) { + auto status = clblast::Gemv(static_cast<clblast::Layout>(layout), + static_cast<clblast::Transpose>(a_transpose), + m, n, + clblast::double2{alpha.s[0], alpha.s[1]}, + a_buffer, a_offset, a_ld, + x_buffer, x_offset, x_inc, + clblast::double2{beta.s[0], beta.s[1]}, + y_buffer, y_offset, y_inc, + queue, event); + return static_cast<StatusCode>(status); +} + +// HEMV +StatusCode CLBlastChemv(const Layout layout, const Triangle triangle, + const size_t n, + const cl_float2 alpha, + const cl_mem a_buffer, const size_t a_offset, const size_t a_ld, + const cl_mem x_buffer, const size_t x_offset, const size_t x_inc, + const cl_float2 beta, + cl_mem y_buffer, const size_t y_offset, const size_t y_inc, + cl_command_queue* queue, cl_event* event) { + auto status = clblast::Hemv(static_cast<clblast::Layout>(layout), + static_cast<clblast::Triangle>(triangle), + n, + clblast::float2{alpha.s[0], alpha.s[1]}, + a_buffer, a_offset, a_ld, + x_buffer, x_offset, x_inc, + clblast::float2{beta.s[0], beta.s[1]}, + y_buffer, y_offset, y_inc, + queue, event); + return static_cast<StatusCode>(status); +} +StatusCode CLBlastZhemv(const Layout layout, const Triangle triangle, + const size_t n, + const cl_double2 alpha, + const cl_mem a_buffer, const size_t a_offset, const size_t a_ld, + const cl_mem x_buffer, const size_t x_offset, const size_t x_inc, + const cl_double2 beta, + cl_mem y_buffer, const size_t y_offset, const size_t y_inc, + cl_command_queue* queue, cl_event* event) { + auto status = clblast::Hemv(static_cast<clblast::Layout>(layout), + static_cast<clblast::Triangle>(triangle), + n, + clblast::double2{alpha.s[0], alpha.s[1]}, + a_buffer, a_offset, a_ld, + x_buffer, x_offset, x_inc, + clblast::double2{beta.s[0], beta.s[1]}, + y_buffer, y_offset, y_inc, + queue, event); + return static_cast<StatusCode>(status); +} + +// SYMV +StatusCode CLBlastSsymv(const Layout layout, const Triangle triangle, + const size_t n, + const float alpha, + const cl_mem a_buffer, const size_t a_offset, const size_t a_ld, + const cl_mem x_buffer, const size_t x_offset, const size_t x_inc, + const float beta, + cl_mem y_buffer, const size_t y_offset, const size_t y_inc, + cl_command_queue* queue, cl_event* event) { + auto status = clblast::Symv(static_cast<clblast::Layout>(layout), + static_cast<clblast::Triangle>(triangle), + n, + alpha, + a_buffer, a_offset, a_ld, + x_buffer, x_offset, x_inc, + beta, + y_buffer, y_offset, y_inc, + queue, event); + return static_cast<StatusCode>(status); +} +StatusCode CLBlastDsymv(const Layout layout, const Triangle triangle, + const size_t n, + const double alpha, + const cl_mem a_buffer, const size_t a_offset, const size_t a_ld, + const cl_mem x_buffer, const size_t x_offset, const size_t x_inc, + const double beta, + cl_mem y_buffer, const size_t y_offset, const size_t y_inc, + cl_command_queue* queue, cl_event* event) { + auto status = clblast::Symv(static_cast<clblast::Layout>(layout), + static_cast<clblast::Triangle>(triangle), + n, + alpha, + a_buffer, a_offset, a_ld, + x_buffer, x_offset, x_inc, + beta, + y_buffer, y_offset, y_inc, + queue, event); + return static_cast<StatusCode>(status); +} + +// ================================================================================================= +// BLAS level-3 (matrix-matrix) routines +// ================================================================================================= + +// GEMM +StatusCode CLBlastSgemm(const Layout layout, const Transpose a_transpose, const Transpose b_transpose, + const size_t m, const size_t n, const size_t k, + const float alpha, + const cl_mem a_buffer, const size_t a_offset, const size_t a_ld, + const cl_mem b_buffer, const size_t b_offset, const size_t b_ld, + const float beta, + cl_mem c_buffer, const size_t c_offset, const size_t c_ld, + cl_command_queue* queue, cl_event* event) { + auto status = clblast::Gemm(static_cast<clblast::Layout>(layout), + static_cast<clblast::Transpose>(a_transpose), + static_cast<clblast::Transpose>(b_transpose), + m, n, k, + alpha, + a_buffer, a_offset, a_ld, + b_buffer, b_offset, b_ld, + beta, + c_buffer, c_offset, c_ld, + queue, event); + return static_cast<StatusCode>(status); +} +StatusCode CLBlastDgemm(const Layout layout, const Transpose a_transpose, const Transpose b_transpose, + const size_t m, const size_t n, const size_t k, + const double alpha, + const cl_mem a_buffer, const size_t a_offset, const size_t a_ld, + const cl_mem b_buffer, const size_t b_offset, const size_t b_ld, + const double beta, + cl_mem c_buffer, const size_t c_offset, const size_t c_ld, + cl_command_queue* queue, cl_event* event) { + auto status = clblast::Gemm(static_cast<clblast::Layout>(layout), + static_cast<clblast::Transpose>(a_transpose), + static_cast<clblast::Transpose>(b_transpose), + m, n, k, + alpha, + a_buffer, a_offset, a_ld, + b_buffer, b_offset, b_ld, + beta, + c_buffer, c_offset, c_ld, + queue, event); + return static_cast<StatusCode>(status); +} +StatusCode CLBlastCgemm(const Layout layout, const Transpose a_transpose, const Transpose b_transpose, + const size_t m, const size_t n, const size_t k, + const cl_float2 alpha, + const cl_mem a_buffer, const size_t a_offset, const size_t a_ld, + const cl_mem b_buffer, const size_t b_offset, const size_t b_ld, + const cl_float2 beta, + cl_mem c_buffer, const size_t c_offset, const size_t c_ld, + cl_command_queue* queue, cl_event* event) { + auto status = clblast::Gemm(static_cast<clblast::Layout>(layout), + static_cast<clblast::Transpose>(a_transpose), + static_cast<clblast::Transpose>(b_transpose), + m, n, k, + clblast::float2{alpha.s[0], alpha.s[1]}, + a_buffer, a_offset, a_ld, + b_buffer, b_offset, b_ld, + clblast::float2{beta.s[0], beta.s[1]}, + c_buffer, c_offset, c_ld, + queue, event); + return static_cast<StatusCode>(status); +} +StatusCode CLBlastZgemm(const Layout layout, const Transpose a_transpose, const Transpose b_transpose, + const size_t m, const size_t n, const size_t k, + const cl_double2 alpha, + const cl_mem a_buffer, const size_t a_offset, const size_t a_ld, + const cl_mem b_buffer, const size_t b_offset, const size_t b_ld, + const cl_double2 beta, + cl_mem c_buffer, const size_t c_offset, const size_t c_ld, + cl_command_queue* queue, cl_event* event) { + auto status = clblast::Gemm(static_cast<clblast::Layout>(layout), + static_cast<clblast::Transpose>(a_transpose), + static_cast<clblast::Transpose>(b_transpose), + m, n, k, + clblast::double2{alpha.s[0], alpha.s[1]}, + a_buffer, a_offset, a_ld, + b_buffer, b_offset, b_ld, + clblast::double2{beta.s[0], beta.s[1]}, + c_buffer, c_offset, c_ld, + queue, event); + return static_cast<StatusCode>(status); +} + +// SYMM +StatusCode CLBlastSsymm(const Layout layout, const Side side, const Triangle triangle, + 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, + const cl_mem b_buffer, const size_t b_offset, const size_t b_ld, + const float beta, + cl_mem c_buffer, const size_t c_offset, const size_t c_ld, + cl_command_queue* queue, cl_event* event) { + auto status = clblast::Symm(static_cast<clblast::Layout>(layout), + static_cast<clblast::Side>(side), + static_cast<clblast::Triangle>(triangle), + m, n, + alpha, + a_buffer, a_offset, a_ld, + b_buffer, b_offset, b_ld, + beta, + c_buffer, c_offset, c_ld, + queue, event); + return static_cast<StatusCode>(status); +} +StatusCode CLBlastDsymm(const Layout layout, const Side side, const Triangle triangle, + 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, + const cl_mem b_buffer, const size_t b_offset, const size_t b_ld, + const double beta, + cl_mem c_buffer, const size_t c_offset, const size_t c_ld, + cl_command_queue* queue, cl_event* event) { + auto status = clblast::Symm(static_cast<clblast::Layout>(layout), + static_cast<clblast::Side>(side), + static_cast<clblast::Triangle>(triangle), + m, n, + alpha, + a_buffer, a_offset, a_ld, + b_buffer, b_offset, b_ld, + beta, + c_buffer, c_offset, c_ld, + queue, event); + return static_cast<StatusCode>(status); +} +StatusCode CLBlastCsymm(const Layout layout, const Side side, const Triangle triangle, + 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, + const cl_mem b_buffer, const size_t b_offset, const size_t b_ld, + const cl_float2 beta, + cl_mem c_buffer, const size_t c_offset, const size_t c_ld, + cl_command_queue* queue, cl_event* event) { + auto status = clblast::Symm(static_cast<clblast::Layout>(layout), + static_cast<clblast::Side>(side), + static_cast<clblast::Triangle>(triangle), + m, n, + clblast::float2{alpha.s[0], alpha.s[1]}, + a_buffer, a_offset, a_ld, + b_buffer, b_offset, b_ld, + clblast::float2{beta.s[0], beta.s[1]}, + c_buffer, c_offset, c_ld, + queue, event); + return static_cast<StatusCode>(status); +} +StatusCode CLBlastZsymm(const Layout layout, const Side side, const Triangle triangle, + 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, + const cl_mem b_buffer, const size_t b_offset, const size_t b_ld, + const cl_double2 beta, + cl_mem c_buffer, const size_t c_offset, const size_t c_ld, + cl_command_queue* queue, cl_event* event) { + auto status = clblast::Symm(static_cast<clblast::Layout>(layout), + static_cast<clblast::Side>(side), + static_cast<clblast::Triangle>(triangle), + m, n, + clblast::double2{alpha.s[0], alpha.s[1]}, + a_buffer, a_offset, a_ld, + b_buffer, b_offset, b_ld, + clblast::double2{beta.s[0], beta.s[1]}, + c_buffer, c_offset, c_ld, + queue, event); + return static_cast<StatusCode>(status); +} + +// HEMM +StatusCode CLBlastChemm(const Layout layout, const Side side, const Triangle triangle, + 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, + const cl_mem b_buffer, const size_t b_offset, const size_t b_ld, + const cl_float2 beta, + cl_mem c_buffer, const size_t c_offset, const size_t c_ld, + cl_command_queue* queue, cl_event* event) { + auto status = clblast::Hemm(static_cast<clblast::Layout>(layout), + static_cast<clblast::Side>(side), + static_cast<clblast::Triangle>(triangle), + m, n, + clblast::float2{alpha.s[0], alpha.s[1]}, + a_buffer, a_offset, a_ld, + b_buffer, b_offset, b_ld, + clblast::float2{beta.s[0], beta.s[1]}, + c_buffer, c_offset, c_ld, + queue, event); + return static_cast<StatusCode>(status); +} +StatusCode CLBlastZhemm(const Layout layout, const Side side, const Triangle triangle, + 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, + const cl_mem b_buffer, const size_t b_offset, const size_t b_ld, + const cl_double2 beta, + cl_mem c_buffer, const size_t c_offset, const size_t c_ld, + cl_command_queue* queue, cl_event* event) { + auto status = clblast::Hemm(static_cast<clblast::Layout>(layout), + static_cast<clblast::Side>(side), + static_cast<clblast::Triangle>(triangle), + m, n, + clblast::double2{alpha.s[0], alpha.s[1]}, + a_buffer, a_offset, a_ld, + b_buffer, b_offset, b_ld, + clblast::double2{beta.s[0], beta.s[1]}, + c_buffer, c_offset, c_ld, + queue, event); + return static_cast<StatusCode>(status); +} + +// SYRK +StatusCode CLBlastSsyrk(const Layout layout, const Triangle triangle, const Transpose a_transpose, + const size_t n, const size_t k, + const float alpha, + const cl_mem a_buffer, const size_t a_offset, const size_t a_ld, + const float beta, + cl_mem c_buffer, const size_t c_offset, const size_t c_ld, + cl_command_queue* queue, cl_event* event) { + auto status = clblast::Syrk(static_cast<clblast::Layout>(layout), + static_cast<clblast::Triangle>(triangle), + static_cast<clblast::Transpose>(a_transpose), + n, k, + alpha, + a_buffer, a_offset, a_ld, + beta, + c_buffer, c_offset, c_ld, + queue, event); + return static_cast<StatusCode>(status); +} +StatusCode CLBlastDsyrk(const Layout layout, const Triangle triangle, const Transpose a_transpose, + const size_t n, const size_t k, + const double alpha, + const cl_mem a_buffer, const size_t a_offset, const size_t a_ld, + const double beta, + cl_mem c_buffer, const size_t c_offset, const size_t c_ld, + cl_command_queue* queue, cl_event* event) { + auto status = clblast::Syrk(static_cast<clblast::Layout>(layout), + static_cast<clblast::Triangle>(triangle), + static_cast<clblast::Transpose>(a_transpose), + n, k, + alpha, + a_buffer, a_offset, a_ld, + beta, + c_buffer, c_offset, c_ld, + queue, event); + return static_cast<StatusCode>(status); +} +StatusCode CLBlastCsyrk(const Layout layout, const Triangle triangle, const Transpose a_transpose, + const size_t n, const size_t k, + const cl_float2 alpha, + const cl_mem a_buffer, const size_t a_offset, const size_t a_ld, + const cl_float2 beta, + cl_mem c_buffer, const size_t c_offset, const size_t c_ld, + cl_command_queue* queue, cl_event* event) { + auto status = clblast::Syrk(static_cast<clblast::Layout>(layout), + static_cast<clblast::Triangle>(triangle), + static_cast<clblast::Transpose>(a_transpose), + n, k, + clblast::float2{alpha.s[0], alpha.s[1]}, + a_buffer, a_offset, a_ld, + clblast::float2{beta.s[0], beta.s[1]}, + c_buffer, c_offset, c_ld, + queue, event); + return static_cast<StatusCode>(status); +} +StatusCode CLBlastZsyrk(const Layout layout, const Triangle triangle, const Transpose a_transpose, + const size_t n, const size_t k, + const cl_double2 alpha, + const cl_mem a_buffer, const size_t a_offset, const size_t a_ld, + const cl_double2 beta, + cl_mem c_buffer, const size_t c_offset, const size_t c_ld, + cl_command_queue* queue, cl_event* event) { + auto status = clblast::Syrk(static_cast<clblast::Layout>(layout), + static_cast<clblast::Triangle>(triangle), + static_cast<clblast::Transpose>(a_transpose), + n, k, + clblast::double2{alpha.s[0], alpha.s[1]}, + a_buffer, a_offset, a_ld, + clblast::double2{beta.s[0], beta.s[1]}, + c_buffer, c_offset, c_ld, + queue, event); + return static_cast<StatusCode>(status); +} + +// HERK +StatusCode CLBlastCherk(const Layout layout, const Triangle triangle, const Transpose a_transpose, + const size_t n, const size_t k, + const float alpha, + const cl_mem a_buffer, const size_t a_offset, const size_t a_ld, + const float beta, + cl_mem c_buffer, const size_t c_offset, const size_t c_ld, + cl_command_queue* queue, cl_event* event) { + auto status = clblast::Herk(static_cast<clblast::Layout>(layout), + static_cast<clblast::Triangle>(triangle), + static_cast<clblast::Transpose>(a_transpose), + n, k, + alpha, + a_buffer, a_offset, a_ld, + beta, + c_buffer, c_offset, c_ld, + queue, event); + return static_cast<StatusCode>(status); +} +StatusCode CLBlastZherk(const Layout layout, const Triangle triangle, const Transpose a_transpose, + const size_t n, const size_t k, + const double alpha, + const cl_mem a_buffer, const size_t a_offset, const size_t a_ld, + const double beta, + cl_mem c_buffer, const size_t c_offset, const size_t c_ld, + cl_command_queue* queue, cl_event* event) { + auto status = clblast::Herk(static_cast<clblast::Layout>(layout), + static_cast<clblast::Triangle>(triangle), + static_cast<clblast::Transpose>(a_transpose), + n, k, + alpha, + a_buffer, a_offset, a_ld, + beta, + c_buffer, c_offset, c_ld, + queue, event); + return static_cast<StatusCode>(status); +} + +// SYR2K +StatusCode CLBlastSsyr2k(const Layout layout, const Triangle triangle, const Transpose ab_transpose, + const size_t n, const size_t k, + const float alpha, + const cl_mem a_buffer, const size_t a_offset, const size_t a_ld, + const cl_mem b_buffer, const size_t b_offset, const size_t b_ld, + const float beta, + cl_mem c_buffer, const size_t c_offset, const size_t c_ld, + cl_command_queue* queue, cl_event* event) { + auto status = clblast::Syr2k(static_cast<clblast::Layout>(layout), + static_cast<clblast::Triangle>(triangle), + static_cast<clblast::Transpose>(ab_transpose), + n, k, + alpha, + a_buffer, a_offset, a_ld, + b_buffer, b_offset, b_ld, + beta, + c_buffer, c_offset, c_ld, + queue, event); + return static_cast<StatusCode>(status); +} +StatusCode CLBlastDsyr2k(const Layout layout, const Triangle triangle, const Transpose ab_transpose, + const size_t n, const size_t k, + const double alpha, + const cl_mem a_buffer, const size_t a_offset, const size_t a_ld, + const cl_mem b_buffer, const size_t b_offset, const size_t b_ld, + const double beta, + cl_mem c_buffer, const size_t c_offset, const size_t c_ld, + cl_command_queue* queue, cl_event* event) { + auto status = clblast::Syr2k(static_cast<clblast::Layout>(layout), + static_cast<clblast::Triangle>(triangle), + static_cast<clblast::Transpose>(ab_transpose), + n, k, + alpha, + a_buffer, a_offset, a_ld, + b_buffer, b_offset, b_ld, + beta, + c_buffer, c_offset, c_ld, + queue, event); + return static_cast<StatusCode>(status); +} +StatusCode CLBlastCsyr2k(const Layout layout, const Triangle triangle, const Transpose ab_transpose, + const size_t n, const size_t k, + const cl_float2 alpha, + const cl_mem a_buffer, const size_t a_offset, const size_t a_ld, + const cl_mem b_buffer, const size_t b_offset, const size_t b_ld, + const cl_float2 beta, + cl_mem c_buffer, const size_t c_offset, const size_t c_ld, + cl_command_queue* queue, cl_event* event) { + auto status = clblast::Syr2k(static_cast<clblast::Layout>(layout), + static_cast<clblast::Triangle>(triangle), + static_cast<clblast::Transpose>(ab_transpose), + n, k, + clblast::float2{alpha.s[0], alpha.s[1]}, + a_buffer, a_offset, a_ld, + b_buffer, b_offset, b_ld, + clblast::float2{beta.s[0], beta.s[1]}, + c_buffer, c_offset, c_ld, + queue, event); + return static_cast<StatusCode>(status); +} +StatusCode CLBlastZsyr2k(const Layout layout, const Triangle triangle, const Transpose ab_transpose, + const size_t n, const size_t k, + const cl_double2 alpha, + const cl_mem a_buffer, const size_t a_offset, const size_t a_ld, + const cl_mem b_buffer, const size_t b_offset, const size_t b_ld, + const cl_double2 beta, + cl_mem c_buffer, const size_t c_offset, const size_t c_ld, + cl_command_queue* queue, cl_event* event) { + auto status = clblast::Syr2k(static_cast<clblast::Layout>(layout), + static_cast<clblast::Triangle>(triangle), + static_cast<clblast::Transpose>(ab_transpose), + n, k, + clblast::double2{alpha.s[0], alpha.s[1]}, + a_buffer, a_offset, a_ld, + b_buffer, b_offset, b_ld, + clblast::double2{beta.s[0], beta.s[1]}, + c_buffer, c_offset, c_ld, + queue, event); + return static_cast<StatusCode>(status); +} + +// HER2K +StatusCode CLBlastCher2k(const Layout layout, const Triangle triangle, const Transpose ab_transpose, + const size_t n, const size_t k, + const cl_float2 alpha, + const cl_mem a_buffer, const size_t a_offset, const size_t a_ld, + const cl_mem b_buffer, const size_t b_offset, const size_t b_ld, + const float beta, + cl_mem c_buffer, const size_t c_offset, const size_t c_ld, + cl_command_queue* queue, cl_event* event) { + auto status = clblast::Her2k(static_cast<clblast::Layout>(layout), + static_cast<clblast::Triangle>(triangle), + static_cast<clblast::Transpose>(ab_transpose), + n, k, + clblast::float2{alpha.s[0], alpha.s[1]}, + a_buffer, a_offset, a_ld, + b_buffer, b_offset, b_ld, + beta, + c_buffer, c_offset, c_ld, + queue, event); + return static_cast<StatusCode>(status); +} +StatusCode CLBlastZher2k(const Layout layout, const Triangle triangle, const Transpose ab_transpose, + const size_t n, const size_t k, + const cl_double2 alpha, + const cl_mem a_buffer, const size_t a_offset, const size_t a_ld, + const cl_mem b_buffer, const size_t b_offset, const size_t b_ld, + const double beta, + cl_mem c_buffer, const size_t c_offset, const size_t c_ld, + cl_command_queue* queue, cl_event* event) { + auto status = clblast::Her2k(static_cast<clblast::Layout>(layout), + static_cast<clblast::Triangle>(triangle), + static_cast<clblast::Transpose>(ab_transpose), + n, k, + clblast::double2{alpha.s[0], alpha.s[1]}, + a_buffer, a_offset, a_ld, + b_buffer, b_offset, b_ld, + beta, + c_buffer, c_offset, c_ld, + queue, event); + return static_cast<StatusCode>(status); +} + +// TRMM +StatusCode CLBlastStrmm(const Layout layout, const Side side, const Triangle triangle, const Transpose a_transpose, const Diagonal diagonal, + 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::Trmm(static_cast<clblast::Layout>(layout), + static_cast<clblast::Side>(side), + static_cast<clblast::Triangle>(triangle), + static_cast<clblast::Transpose>(a_transpose), + static_cast<clblast::Diagonal>(diagonal), + m, n, + alpha, + a_buffer, a_offset, a_ld, + b_buffer, b_offset, b_ld, + queue, event); + return static_cast<StatusCode>(status); +} +StatusCode CLBlastDtrmm(const Layout layout, const Side side, const Triangle triangle, const Transpose a_transpose, const Diagonal diagonal, + 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::Trmm(static_cast<clblast::Layout>(layout), + static_cast<clblast::Side>(side), + static_cast<clblast::Triangle>(triangle), + static_cast<clblast::Transpose>(a_transpose), + static_cast<clblast::Diagonal>(diagonal), + m, n, + alpha, + a_buffer, a_offset, a_ld, + b_buffer, b_offset, b_ld, + queue, event); + return static_cast<StatusCode>(status); +} +StatusCode CLBlastCtrmm(const Layout layout, const Side side, const Triangle triangle, const Transpose a_transpose, const Diagonal diagonal, + 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::Trmm(static_cast<clblast::Layout>(layout), + static_cast<clblast::Side>(side), + static_cast<clblast::Triangle>(triangle), + static_cast<clblast::Transpose>(a_transpose), + static_cast<clblast::Diagonal>(diagonal), + m, n, + clblast::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 CLBlastZtrmm(const Layout layout, const Side side, const Triangle triangle, const Transpose a_transpose, const Diagonal diagonal, + 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::Trmm(static_cast<clblast::Layout>(layout), + static_cast<clblast::Side>(side), + static_cast<clblast::Triangle>(triangle), + static_cast<clblast::Transpose>(a_transpose), + static_cast<clblast::Diagonal>(diagonal), + m, n, + clblast::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); +} + +// ================================================================================================= |