/* ************************************************************************ * Copyright 2016-2019 Advanced Micro Devices, Inc. * ************************************************************************ */ #pragma once #ifndef _ROCBLAS_FUNCTIONS_H_ #define _ROCBLAS_FUNCTIONS_H_ #include "rocblas-export.h" #include "rocblas-types.h" /*!\file * \brief rocblas_functions.h provides Basic Linear Algebra Subprograms of Level 1, 2 and 3, * using HIP optimized for AMD HCC-based GPU hardware. This library can also run on CUDA-based * NVIDIA GPUs. * This file exposes C89 BLAS interface */ /* * =========================================================================== * README: Please follow the naming convention * Big case for matrix, e.g. matrix A, B, C GEMM (C = A*B) * Lower case for vector, e.g. vector x, y GEMV (y = A*x) * =========================================================================== */ /* ROCBLAS_VA_OPT_PRAGMA(pragma, ...) creates a _Pragma with stringized pragma if the trailing argument list is non-empty. __VA_OPT__ support is automatically detected if it's available; otherwise, the GCC/Clang ##__VA_ARGS__ extension is used to emulate it. */ #define ROCBLAS_VA_OPT_3RD_ARG(_1, _2, _3, ...) _3 #define ROCBLAS_VA_OPT_SUPPORTED(...) ROCBLAS_VA_OPT_3RD_ARG(__VA_OPT__(, ), 1, 0, ) #if ROCBLAS_VA_OPT_SUPPORTED(?) #define ROCBLAS_VA_OPT_PRAGMA(pragma, ...) __VA_OPT__(_Pragma(#pragma)) #else // ROCBLAS_VA_OPT_SUPPORTED #define ROCBLAS_VA_OPT_COUNT_IMPL(X, _1, _2, _3, _4, _5, _6, _7, _8, _9, _10, N, ...) N #define ROCBLAS_VA_OPT_COUNT(...) \ ROCBLAS_VA_OPT_COUNT_IMPL(, ##__VA_ARGS__, N, N, N, N, N, N, N, N, N, N, 0) #define ROCBLAS_VA_OPT_PRAGMA_SELECT0(...) #define ROCBLAS_VA_OPT_PRAGMA_SELECTN(pragma, ...) _Pragma(#pragma) #define ROCBLAS_VA_OPT_PRAGMA_IMPL2(pragma, count) ROCBLAS_VA_OPT_PRAGMA_SELECT##count(pragma) #define ROCBLAS_VA_OPT_PRAGMA_IMPL(pragma, count) ROCBLAS_VA_OPT_PRAGMA_IMPL2(pragma, count) #define ROCBLAS_VA_OPT_PRAGMA(pragma, ...) \ ROCBLAS_VA_OPT_PRAGMA_IMPL(pragma, ROCBLAS_VA_OPT_COUNT(__VA_ARGS__)) #endif // ROCBLAS_VA_OPT_SUPPORTED #ifdef __cplusplus extern "C" { #endif /* * =========================================================================== * level 1 BLAS * =========================================================================== */ /*! \brief BLAS Level 1 API \details scal scal the vector x[i] with scalar alpha, for i = 1 , … , n x := alpha * x , @param[in] handle rocblas_handle. handle to the rocblas library context queue. @param[in] n rocblas_int. @param[in] alpha specifies the scalar alpha. @param[inout] x pointer storing vector x on the GPU. @param[in] incx specifies the increment for the elements of x. ********************************************************************/ ROCBLAS_EXPORT rocblas_status rocblas_sscal( rocblas_handle handle, rocblas_int n, const float* alpha, float* x, rocblas_int incx); ROCBLAS_EXPORT rocblas_status rocblas_dscal( rocblas_handle handle, rocblas_int n, const double* alpha, double* x, rocblas_int incx); ROCBLAS_EXPORT rocblas_status rocblas_cscal(rocblas_handle handle, rocblas_int n, const rocblas_float_complex* alpha, rocblas_float_complex* x, rocblas_int incx); ROCBLAS_EXPORT rocblas_status rocblas_zscal(rocblas_handle handle, rocblas_int n, const rocblas_double_complex* alpha, rocblas_double_complex* x, rocblas_int incx); ROCBLAS_EXPORT rocblas_status rocblas_csscal(rocblas_handle handle, rocblas_int n, const float* alpha, rocblas_float_complex* x, rocblas_int incx); ROCBLAS_EXPORT rocblas_status rocblas_zdscal(rocblas_handle handle, rocblas_int n, const double* alpha, rocblas_double_complex* x, rocblas_int incx); /*! \brief BLAS Level 1 API \details copy copies the vector x into the vector y, for i = 1 , … , n y := x, @param[in] handle rocblas_handle. handle to the rocblas library context queue. @param[in] n rocblas_int. @param[in] x pointer storing vector x on the GPU. @param[in] incx specifies the increment for the elements of x. @param[out] y pointer storing vector y on the GPU. @param[in] incy rocblas_int specifies the increment for the elements of y. ********************************************************************/ ROCBLAS_EXPORT rocblas_status rocblas_scopy(rocblas_handle handle, rocblas_int n, const float* x, rocblas_int incx, float* y, rocblas_int incy); ROCBLAS_EXPORT rocblas_status rocblas_dcopy(rocblas_handle handle, rocblas_int n, const double* x, rocblas_int incx, double* y, rocblas_int incy); ROCBLAS_EXPORT rocblas_status rocblas_ccopy(rocblas_handle handle, rocblas_int n, const rocblas_float_complex* x, rocblas_int incx, rocblas_float_complex* y, rocblas_int incy); ROCBLAS_EXPORT rocblas_status rocblas_zcopy(rocblas_handle handle, rocblas_int n, const rocblas_double_complex* x, rocblas_int incx, rocblas_double_complex* y, rocblas_int incy); /*! \brief BLAS Level 1 API \details dot(u) perform dot product of vector x and y result = x * y; dotc perform dot product of complex vector x and complex y result = conjugate (x) * y; @param[in] handle rocblas_handle. handle to the rocblas library context queue. @param[in] n rocblas_int. @param[in] x pointer storing vector x on the GPU. @param[in] incx rocblas_int specifies the increment for the elements of y. @param[inout] result store the dot product. either on the host CPU or device GPU. return is 0.0 if n <= 0. ********************************************************************/ ROCBLAS_EXPORT rocblas_status rocblas_sdot(rocblas_handle handle, rocblas_int n, const float* x, rocblas_int incx, const float* y, rocblas_int incy, float* result); ROCBLAS_EXPORT rocblas_status rocblas_ddot(rocblas_handle handle, rocblas_int n, const double* x, rocblas_int incx, const double* y, rocblas_int incy, double* result); ROCBLAS_EXPORT rocblas_status rocblas_cdotu(rocblas_handle handle, rocblas_int n, const rocblas_float_complex* x, rocblas_int incx, const rocblas_float_complex* y, rocblas_int incy, rocblas_float_complex* result); ROCBLAS_EXPORT rocblas_status rocblas_zdotu(rocblas_handle handle, rocblas_int n, const rocblas_double_complex* x, rocblas_int incx, const rocblas_double_complex* y, rocblas_int incy, rocblas_double_complex* result); ROCBLAS_EXPORT rocblas_status rocblas_cdotc(rocblas_handle handle, rocblas_int n, const rocblas_float_complex* x, rocblas_int incx, const rocblas_float_complex* y, rocblas_int incy, rocblas_float_complex* result); ROCBLAS_EXPORT rocblas_status rocblas_zdotc(rocblas_handle handle, rocblas_int n, const rocblas_double_complex* x, rocblas_int incx, const rocblas_double_complex* y, rocblas_int incy, rocblas_double_complex* result); /*! \brief BLAS Level 1 API \details swap interchange vector x[i] and y[i], for i = 1 , … , n y := x; x := y @param[in] handle rocblas_handle. handle to the rocblas library context queue. @param[in] n rocblas_int. @param[inout] x pointer storing vector x on the GPU. @param[in] incx specifies the increment for the elements of x. @param[inout] y pointer storing vector y on the GPU. @param[in] incy rocblas_int specifies the increment for the elements of y. ********************************************************************/ ROCBLAS_EXPORT rocblas_status rocblas_sswap( rocblas_handle handle, rocblas_int n, float* x, rocblas_int incx, float* y, rocblas_int incy); ROCBLAS_EXPORT rocblas_status rocblas_dswap( rocblas_handle handle, rocblas_int n, double* x, rocblas_int incx, double* y, rocblas_int incy); ROCBLAS_EXPORT rocblas_status rocblas_cswap(rocblas_handle handle, rocblas_int n, rocblas_float_complex* x, rocblas_int incx, rocblas_float_complex* y, rocblas_int incy); ROCBLAS_EXPORT rocblas_status rocblas_zswap(rocblas_handle handle, rocblas_int n, rocblas_double_complex* x, rocblas_int incx, rocblas_double_complex* y, rocblas_int incy); /*! \brief BLAS Level 1 API \details axpy compute y := alpha * x + y @param[in] handle rocblas_handle. handle to the rocblas library context queue. @param[in] n rocblas_int. @param[in] alpha specifies the scalar alpha. @param[in] x pointer storing vector x on the GPU. @param[in] incx rocblas_int specifies the increment for the elements of x. @param[out] y pointer storing vector y on the GPU. @param[inout] incy rocblas_int specifies the increment for the elements of y. ********************************************************************/ ROCBLAS_EXPORT rocblas_status rocblas_haxpy(rocblas_handle handle, rocblas_int n, const rocblas_half* alpha, const rocblas_half* x, rocblas_int incx, rocblas_half* y, rocblas_int incy); ROCBLAS_EXPORT rocblas_status rocblas_saxpy(rocblas_handle handle, rocblas_int n, const float* alpha, const float* x, rocblas_int incx, float* y, rocblas_int incy); ROCBLAS_EXPORT rocblas_status rocblas_daxpy(rocblas_handle handle, rocblas_int n, const double* alpha, const double* x, rocblas_int incx, double* y, rocblas_int incy); ROCBLAS_EXPORT rocblas_status rocblas_caxpy(rocblas_handle handle, rocblas_int n, const rocblas_float_complex* alpha, const rocblas_float_complex* x, rocblas_int incx, rocblas_float_complex* y, rocblas_int incy); ROCBLAS_EXPORT rocblas_status rocblas_zaxpy(rocblas_handle handle, rocblas_int n, const rocblas_double_complex* alpha, const rocblas_double_complex* x, rocblas_int incx, rocblas_double_complex* y, rocblas_int incy); /*! \brief BLAS Level 1 API \details asum computes the sum of the magnitudes of elements of a real vector x, or the sum of magnitudes of the real and imaginary parts of elements if x is a complex vector @param[in] handle rocblas_handle. handle to the rocblas library context queue. @param[in] n rocblas_int. @param[in] x pointer storing vector x on the GPU. @param[in] incx rocblas_int specifies the increment for the elements of y. @param[inout] result store the asum product. either on the host CPU or device GPU. return is 0.0 if n, incx<=0. ********************************************************************/ ROCBLAS_EXPORT rocblas_status rocblas_sasum( rocblas_handle handle, rocblas_int n, const float* x, rocblas_int incx, float* result); ROCBLAS_EXPORT rocblas_status rocblas_dasum( rocblas_handle handle, rocblas_int n, const double* x, rocblas_int incx, double* result); ROCBLAS_EXPORT rocblas_status rocblas_scasum(rocblas_handle handle, rocblas_int n, const rocblas_float_complex* x, rocblas_int incx, float* result); ROCBLAS_EXPORT rocblas_status rocblas_dzasum(rocblas_handle handle, rocblas_int n, const rocblas_double_complex* x, rocblas_int incx, double* result); /*! \brief BLAS Level 1 API \details nrm2 computes the euclidean norm of a real or complex vector := sqrt( x'*x ) for real vector := sqrt( x**H*x ) for complex vector @param[in] handle rocblas_handle. handle to the rocblas library context queue. @param[in] n rocblas_int. @param[in] x pointer storing vector x on the GPU. @param[in] incx rocblas_int specifies the increment for the elements of y. @param[inout] result store the nrm2 product. either on the host CPU or device GPU. return is 0.0 if n, incx<=0. ********************************************************************/ ROCBLAS_EXPORT rocblas_status rocblas_snrm2( rocblas_handle handle, rocblas_int n, const float* x, rocblas_int incx, float* result); ROCBLAS_EXPORT rocblas_status rocblas_dnrm2( rocblas_handle handle, rocblas_int n, const double* x, rocblas_int incx, double* result); ROCBLAS_EXPORT rocblas_status rocblas_scnrm2(rocblas_handle handle, rocblas_int n, const rocblas_float_complex* x, rocblas_int incx, float* result); ROCBLAS_EXPORT rocblas_status rocblas_dznrm2(rocblas_handle handle, rocblas_int n, const rocblas_double_complex* x, rocblas_int incx, double* result); /*! \brief BLAS Level 1 API \details amax finds the first index of the element of maximum magnitude of real vector x or the sum of magnitude of the real and imaginary parts of elements if x is a complex vector @param[in] handle rocblas_handle. handle to the rocblas library context queue. @param[in] n rocblas_int. @param[in] x pointer storing vector x on the GPU. @param[in] incx rocblas_int specifies the increment for the elements of y. @param[inout] result store the amax index. either on the host CPU or device GPU. return is 0.0 if n, incx<=0. ********************************************************************/ ROCBLAS_EXPORT rocblas_status rocblas_isamax( rocblas_handle handle, rocblas_int n, const float* x, rocblas_int incx, rocblas_int* result); ROCBLAS_EXPORT rocblas_status rocblas_idamax( rocblas_handle handle, rocblas_int n, const double* x, rocblas_int incx, rocblas_int* result); ROCBLAS_EXPORT rocblas_status rocblas_icamax(rocblas_handle handle, rocblas_int n, const rocblas_float_complex* x, rocblas_int incx, rocblas_int* result); ROCBLAS_EXPORT rocblas_status rocblas_izamax(rocblas_handle handle, rocblas_int n, const rocblas_double_complex* x, rocblas_int incx, rocblas_int* result); /*! \brief BLAS Level 1 API \details amin finds the first index of the element of minimum magnitude of real vector x or the sum of magnitude of the real and imaginary parts of elements if x is a complex vector @param[in] handle rocblas_handle. handle to the rocblas library context queue. @param[in] n rocblas_int. @param[in] x pointer storing vector x on the GPU. @param[in] incx rocblas_int specifies the increment for the elements of y. @param[inout] result store the amin index. either on the host CPU or device GPU. return is 0.0 if n, incx<=0. ********************************************************************/ ROCBLAS_EXPORT rocblas_status rocblas_isamin( rocblas_handle handle, rocblas_int n, const float* x, rocblas_int incx, rocblas_int* result); ROCBLAS_EXPORT rocblas_status rocblas_idamin( rocblas_handle handle, rocblas_int n, const double* x, rocblas_int incx, rocblas_int* result); ROCBLAS_EXPORT rocblas_status rocblas_icamin(rocblas_handle handle, rocblas_int n, const rocblas_float_complex* x, rocblas_int incx, rocblas_int* result); ROCBLAS_EXPORT rocblas_status rocblas_izamin(rocblas_handle handle, rocblas_int n, const rocblas_double_complex* x, rocblas_int incx, rocblas_int* result); /* * =========================================================================== * level 2 BLAS * =========================================================================== */ /*! \brief BLAS Level 2 API \details xGEMV performs one of the matrix-vector operations y := alpha*A*x + beta*y, or y := alpha*A**T*x + beta*y, or y := alpha*A**H*x + beta*y, where alpha and beta are scalars, x and y are vectors and A is an m by n matrix. @param[in] handle rocblas_handle. handle to the rocblas library context queue. @param[in] trans rocblas_operation indicates whether matrix A is tranposed (conjugated) or not @param[in] m rocblas_int number of rows of matrix A @param[in] n rocblas_int number of columns of matrix A @param[in] alpha specifies the scalar alpha. @param[in] A pointer storing matrix A on the GPU. @param[in] lda rocblas_int specifies the leading dimension of A. @param[in] x pointer storing vector x on the GPU. @param[in] incx rocblas_int specifies the increment for the elements of x. @param[in] beta specifies the scalar beta. @param[inout] y pointer storing vector y on the GPU. @param[in] incy rocblas_int specifies the increment for the elements of y. ********************************************************************/ ROCBLAS_EXPORT rocblas_status rocblas_sgemv(rocblas_handle handle, rocblas_operation trans, rocblas_int m, rocblas_int n, const float* alpha, const float* A, rocblas_int lda, const float* x, rocblas_int incx, const float* beta, float* y, rocblas_int incy); ROCBLAS_EXPORT rocblas_status rocblas_dgemv(rocblas_handle handle, rocblas_operation trans, rocblas_int m, rocblas_int n, const double* alpha, const double* A, rocblas_int lda, const double* x, rocblas_int incx, const double* beta, double* y, rocblas_int incy); ROCBLAS_EXPORT rocblas_status rocblas_cgemv(rocblas_handle handle, rocblas_operation trans, rocblas_int m, rocblas_int n, const rocblas_float_complex* alpha, const rocblas_float_complex* A, rocblas_int lda, const rocblas_float_complex* x, rocblas_int incx, const rocblas_float_complex* beta, rocblas_float_complex* y, rocblas_int incy); ROCBLAS_EXPORT rocblas_status rocblas_zgemv(rocblas_handle handle, rocblas_operation trans, rocblas_int m, rocblas_int n, const rocblas_double_complex* alpha, const rocblas_double_complex* A, rocblas_int lda, const rocblas_double_complex* x, rocblas_int incx, const rocblas_double_complex* beta, rocblas_double_complex* y, rocblas_int incy); /*! \brief BLAS Level 2 API \details xGEMV_BATCHED performs a batch of matrix-vector operations y_i := alpha*A_i*x_i + beta*y_i, or y_i := alpha*A_i**T*x_i + beta*y_i, or y_i := alpha*A_i**H*x_i + beta*y_i, where (A_i, x_i, y_i) is the i-th instance of the batch. alpha and beta are scalars, x_i and y_i are vectors and A_i is an m by n matrix. @param[in] handle rocblas_handle. handle to the rocblas library context queue. @param[in] trans rocblas_operation indicates whether matrices A_i are tranposed (conjugated) or not @param[in] m rocblas_int number of rows of matrices A_i @param[in] n rocblas_int number of columns of matrices A_i @param[in] alpha specifies the scalar alpha. @param[in] A array of pointers storing the different matrices A_i on the GPU. @param[in] lda rocblas_int specifies the leading dimension of matrices A_i. @param[in] x array of pointers storing the different vectors x_i on the GPU. @param[in] incx rocblas_int specifies the increment for the elements of vectors x_i. @param[in] beta specifies the scalar beta. @param[inout] y array of pointers storing the different vectors y_i on the GPU. @param[in] incy rocblas_int specifies the increment for the elements of vectors y_i. @param[in] batch_count rocblas_int number of instances in the batch ********************************************************************/ ROCBLAS_EXPORT rocblas_status rocblas_sgemv_batched(rocblas_handle handle, rocblas_operation trans, rocblas_int m, rocblas_int n, const float* alpha, const float* const A[], rocblas_int lda, const float* const x[], rocblas_int incx, const float* beta, float* const y[], rocblas_int incy, rocblas_int batch_count); ROCBLAS_EXPORT rocblas_status rocblas_dgemv_batched(rocblas_handle handle, rocblas_operation trans, rocblas_int m, rocblas_int n, const double* alpha, const double* const A[], rocblas_int lda, const double* const x[], rocblas_int incx, const double* beta, double* const y[], rocblas_int incy, rocblas_int batch_count); ROCBLAS_EXPORT rocblas_status rocblas_cgemv_batched(rocblas_handle handle, rocblas_operation trans, rocblas_int m, rocblas_int n, const rocblas_float_complex* alpha, const rocblas_float_complex* const A[], rocblas_int lda, const rocblas_float_complex* const x[], rocblas_int incx, const rocblas_float_complex* beta, rocblas_float_complex* const y[], rocblas_int incy, rocblas_int batch_count); ROCBLAS_EXPORT rocblas_status rocblas_zgemv_batched(rocblas_handle handle, rocblas_operation trans, rocblas_int m, rocblas_int n, const rocblas_double_complex* alpha, const rocblas_double_complex* const A[], rocblas_int lda, const rocblas_double_complex* const x[], rocblas_int incx, const rocblas_double_complex* beta, rocblas_double_complex* const y[], rocblas_int incy, rocblas_int batch_count); /*! \brief BLAS Level 2 API \details xGEMV_STRIDED_BATCHED performs a batch of matrix-vector operations y_i := alpha*A_i*x_i + beta*y_i, or y_i := alpha*A_i**T*x_i + beta*y_i, or y_i := alpha*A_i**H*x_i + beta*y_i, where (A_i, x_i, y_i) is the i-th instance of the batch. alpha and beta are scalars, x_i and y_i are vectors and A_i is an m by n matrix. @param[in] handle rocblas_handle. handle to the rocblas library context queue. @param[in] trans rocblas_operation indicates whether matrices A_i are tranposed (conjugated) or not @param[in] m rocblas_int number of rows of matrices A_i @param[in] n rocblas_int number of columns of matrices A_i @param[in] alpha specifies the scalar alpha. @param[in] A pointer to the first matrix (A_0) in the batch stored on the GPU. @param[in] lda rocblas_int specifies the leading dimension of matrices A_i. @param[in] strideA rocblas_int stride from the start of one matrix (A_i) and the next one (A_i+1) @param[in] x pointer to the first vector (x_0) in the batch stored on the GPU. @param[in] incx rocblas_int specifies the increment for the elements of vectors x_i. @param[in] stridex rocblas_int stride form the start of one vector (x_i) and the next one (x_i+1) @param[in] beta specifies the scalar beta. @param[inout] y pointer to the first vector (y_0) in the batch stored on the GPU. @param[in] incy rocblas_int specifies the increment for the elements of vectors y_i. @param[in] stridey rocblas_int stride from the start of one vector (y_i) and the next one (y_i+1) @param[in] batch_count rocblas_int number of instances in the batch ********************************************************************/ ROCBLAS_EXPORT rocblas_status rocblas_sgemv_strided_batched(rocblas_handle handle, rocblas_operation transA, rocblas_int m, rocblas_int n, const float* alpha, const float* A, rocblas_int lda, rocblas_int strideA, const float* x, rocblas_int incx, rocblas_int stridex, const float* beta, float* y, rocblas_int incy, rocblas_int stridey, rocblas_int batch_count); ROCBLAS_EXPORT rocblas_status rocblas_dgemv_strided_batched(rocblas_handle handle, rocblas_operation transA, rocblas_int m, rocblas_int n, const double* alpha, const double* A, rocblas_int lda, rocblas_int strideA, const double* x, rocblas_int incx, rocblas_int stridex, const double* beta, double* y, rocblas_int incy, rocblas_int stridey, rocblas_int batch_count); ROCBLAS_EXPORT rocblas_status rocblas_cgemv_strided_batched(rocblas_handle handle, rocblas_operation transA, rocblas_int m, rocblas_int n, const rocblas_float_complex* alpha, const rocblas_float_complex* A, rocblas_int lda, rocblas_int strideA, const rocblas_float_complex* x, rocblas_int incx, rocblas_int stridex, const rocblas_float_complex* beta, rocblas_float_complex* y, rocblas_int incy, rocblas_int stridey, rocblas_int batch_count); ROCBLAS_EXPORT rocblas_status rocblas_zgemv_strided_batched(rocblas_handle handle, rocblas_operation transA, rocblas_int m, rocblas_int n, const rocblas_double_complex* alpha, const rocblas_double_complex* A, rocblas_int lda, rocblas_int strideA, const rocblas_double_complex* x, rocblas_int incx, rocblas_int stridex, const rocblas_double_complex* beta, rocblas_double_complex* y, rocblas_int incy, rocblas_int stridey, rocblas_int batch_count); /*! \brief BLAS Level 2 API \details trsv solves A*x = alpha*b or A**T*x = alpha*b, where x and b are vectors and A is a triangular matrix. The vector x is overwritten on b. @param[in] handle rocblas_handle. handle to the rocblas library context queue. @param[in] uplo rocblas_fill. rocblas_fill_upper: A is an upper triangular matrix. rocblas_fill_lower: A is a lower triangular matrix. @param[in] transA rocblas_operation @param[in] diag rocblas_diagonal. rocblas_diagonal_unit: A is assumed to be unit triangular. rocblas_diagonal_non_unit: A is not assumed to be unit triangular. @param[in] m rocblas_int m specifies the number of rows of b. m >= 0. @param[in] alpha specifies the scalar alpha. @param[in] A pointer storing matrix A on the GPU, of dimension ( lda, m ) @param[in] lda rocblas_int specifies the leading dimension of A. lda = max( 1, m ). @param[in] x pointer storing vector x on the GPU. @param[in] incx specifies the increment for the elements of x. ********************************************************************/ ROCBLAS_EXPORT rocblas_status rocblas_strsv(rocblas_handle handle, rocblas_fill uplo, rocblas_operation transA, rocblas_diagonal diag, rocblas_int m, const float* A, rocblas_int lda, float* x, rocblas_int incx); ROCBLAS_EXPORT rocblas_status rocblas_dtrsv(rocblas_handle handle, rocblas_fill uplo, rocblas_operation transA, rocblas_diagonal diag, rocblas_int m, const double* A, rocblas_int lda, double* x, rocblas_int incx); /*! \brief BLAS Level 2 API \details xHE(SY)MV performs the matrix-vector operation: y := alpha*A*x + beta*y, where alpha and beta are scalars, x and y are n element vectors and A is an n by n Hermitian(Symmetric) matrix. @param[in] handle rocblas_handle. handle to the rocblas library context queue. @param[in] uplo rocblas_fill. specifies whether the upper or lower @param[in] n rocblas_int. @param[in] alpha specifies the scalar alpha. @param[in] A pointer storing matrix A on the GPU. @param[in] lda rocblas_int specifies the leading dimension of A. @param[in] x pointer storing vector x on the GPU. @param[in] incx specifies the increment for the elements of x. @param[in] beta specifies the scalar beta. @param[out] y pointer storing vector y on the GPU. @param[in] incy rocblas_int specifies the increment for the elements of y. ********************************************************************/ /* not implemented ROCBLAS_EXPORT rocblas_status rocblas_ssymv(rocblas_handle handle, rocblas_fill uplo, rocblas_int n, const float *alpha, const float *A, rocblas_int lda, const float *x, rocblas_int incx, const float *beta, float *y, rocblas_int incy); ROCBLAS_EXPORT rocblas_status rocblas_dsymv(rocblas_handle handle, rocblas_fill uplo, rocblas_int n, const double *alpha, const double *A, rocblas_int lda, const double *x, rocblas_int incx, const double *beta, double *y, rocblas_int incy); */ /* not implemented ROCBLAS_EXPORT rocblas_status rocblas_chemv(rocblas_handle handle, rocblas_fill uplo, rocblas_int n, const rocblas_float_complex *alpha, const rocblas_float_complex *A, rocblas_int lda, const rocblas_float_complex *x, rocblas_int incx, const rocblas_float_complex *beta, rocblas_float_complex *y, rocblas_int incy); ROCBLAS_EXPORT rocblas_status rocblas_zhemv(rocblas_handle handle, rocblas_fill uplo, rocblas_int n, const rocblas_double_complex *alpha, const rocblas_double_complex *A, rocblas_int lda, const rocblas_double_complex *x, rocblas_int incx, const rocblas_double_complex *beta, rocblas_double_complex *y, rocblas_int incy); */ /*! \brief BLAS Level 2 API \details xGER performs the matrix-vector operations A := A + alpha*x*y**T where alpha is a scalars, x and y are vectors, and A is an m by n matrix. @param[in] handle rocblas_handle. handle to the rocblas library context queue. @param[in] m rocblas_int @param[in] n rocblas_int @param[in] alpha specifies the scalar alpha. @param[in] x pointer storing vector x on the GPU. @param[in] incx rocblas_int specifies the increment for the elements of x. @param[in] y pointer storing vector y on the GPU. @param[in] incy rocblas_int specifies the increment for the elements of y. @param[inout] A pointer storing matrix A on the GPU. @param[in] lda rocblas_int specifies the leading dimension of A. ********************************************************************/ ROCBLAS_EXPORT rocblas_status rocblas_sger(rocblas_handle handle, rocblas_int m, rocblas_int n, const float* alpha, const float* x, rocblas_int incx, const float* y, rocblas_int incy, float* A, rocblas_int lda); ROCBLAS_EXPORT rocblas_status rocblas_dger(rocblas_handle handle, rocblas_int m, rocblas_int n, const double* alpha, const double* x, rocblas_int incx, const double* y, rocblas_int incy, double* A, rocblas_int lda); /* not implemented ROCBLAS_EXPORT rocblas_status rocblas_cger(rocblas_handle handle, rocblas_int m, rocblas_int n, const rocblas_float_complex *alpha, const rocblas_float_complex *x, rocblas_int incx, const rocblas_float_complex *y, rocblas_int incy, rocblas_float_complex *A, rocblas_int lda); ROCBLAS_EXPORT rocblas_status rocblas_zger(rocblas_handle handle, rocblas_int m, rocblas_int n, const rocblas_double_complex *alpha, const rocblas_double_complex *x, rocblas_int incx, const rocblas_double_complex *y, rocblas_int incy, rocblas_double_complex *A, rocblas_int lda); */ /*! \brief BLAS Level 2 API \details xSYR performs the matrix-vector operations A := A + alpha*x*x**T where alpha is a scalars, x is a vector, and A is an n by n symmetric matrix. @param[in] handle rocblas_handle. handle to the rocblas library context queue. @param[in] n rocblas_int @param[in] alpha specifies the scalar alpha. @param[in] x pointer storing vector x on the GPU. @param[in] incx rocblas_int specifies the increment for the elements of x. @param[inout] A pointer storing matrix A on the GPU. @param[in] lda rocblas_int specifies the leading dimension of A. ********************************************************************/ ROCBLAS_EXPORT rocblas_status rocblas_ssyr(rocblas_handle handle, rocblas_fill uplo, rocblas_int n, const float* alpha, const float* x, rocblas_int incx, float* A, rocblas_int lda); ROCBLAS_EXPORT rocblas_status rocblas_dsyr(rocblas_handle handle, rocblas_fill uplo, rocblas_int n, const double* alpha, const double* x, rocblas_int incx, double* A, rocblas_int lda); /* not implemented ROCBLAS_EXPORT rocblas_status rocblas_csyr(rocblas_handle handle, rocblas_int n, const rocblas_float_complex *alpha, const rocblas_float_complex *x, rocblas_int incx, rocblas_float_complex *A, rocblas_int lda); ROCBLAS_EXPORT rocblas_status rocblas_zsyr(rocblas_handle handle, rocblas_int n, const rocblas_double_complex *alpha, const rocblas_double_complex *x, rocblas_int incx, rocblas_double_complex *A, rocblas_int lda); */ /* * =========================================================================== * level 3 BLAS * =========================================================================== */ /*! \brief BLAS Level 3 API \details trtri compute the inverse of a matrix A, namely, invA and write the result into invA; @param[in] handle rocblas_handle. handle to the rocblas library context queue. @param[in] uplo rocblas_fill. specifies whether the upper 'rocblas_fill_upper' or lower 'rocblas_fill_lower' if rocblas_fill_upper, the lower part of A is not referenced if rocblas_fill_lower, the upper part of A is not referenced @param[in] diag rocblas_diagonal. = 'rocblas_diagonal_non_unit', A is non-unit triangular; = 'rocblas_diagonal_unit', A is unit triangular; @param[in] n rocblas_int. size of matrix A and invA @param[in] A pointer storing matrix A on the GPU. @param[in] lda rocblas_int specifies the leading dimension of A. @param[output] invA pointer storing matrix invA on the GPU. @param[in] ldinvA rocblas_int specifies the leading dimension of invA. ********************************************************************/ ROCBLAS_EXPORT rocblas_status rocblas_strtri(rocblas_handle handle, rocblas_fill uplo, rocblas_diagonal diag, rocblas_int n, const float* A, rocblas_int lda, float* invA, rocblas_int ldinvA); ROCBLAS_EXPORT rocblas_status rocblas_dtrtri(rocblas_handle handle, rocblas_fill uplo, rocblas_diagonal diag, rocblas_int n, const double* A, rocblas_int lda, double* invA, rocblas_int ldinvA); /*! \brief BLAS Level 3 API \details trtri compute the inverse of a matrix A inv(A); @param[in] handle rocblas_handle. handle to the rocblas library context queue. @param[in] uplo rocblas_fill. specifies whether the upper 'rocblas_fill_upper' or lower 'rocblas_fill_lower' @param[in] diag rocblas_diagonal. = 'rocblas_diagonal_non_unit', A is non-unit triangular; = 'rocblas_diagonal_unit', A is unit triangular; @param[in] n rocblas_int. @param[in] A pointer storing matrix A on the GPU. @param[in] lda rocblas_int specifies the leading dimension of A. @param[in] stride_a rocblas_int "batch stride a": stride from the start of one "A" matrix to the next @param[out] invA pointer storing the inverse matrix A on the GPU. Partial inplace operation is supported, see below. If UPLO = 'U', the leading N-by-N upper triangular part of the invA will store the inverse of the upper triangular matrix, and the strictly lower triangular part of invA is cleared. If UPLO = 'L', the leading N-by-N lower triangular part of the invA will store the inverse of the lower triangular matrix, and the strictly upper triangular part of invA is cleared. @param[in] ldinvA rocblas_int specifies the leading dimension of invA. @param[in] stride_invA rocblas_int "batch stride invA": stride from the start of one "invA" matrix to the next @param[in] batch_count rocblas_int numbers of matrices in the batch ********************************************************************/ ROCBLAS_EXPORT rocblas_status rocblas_strtri_batched(rocblas_handle handle, rocblas_fill uplo, rocblas_diagonal diag, rocblas_int n, const float* A, rocblas_int lda, rocblas_int stride_a, float* invA, rocblas_int ldinvA, rocblas_int stride_invA, rocblas_int batch_count); ROCBLAS_EXPORT rocblas_status rocblas_dtrtri_batched(rocblas_handle handle, rocblas_fill uplo, rocblas_diagonal diag, rocblas_int n, const double* A, rocblas_int lda, rocblas_int stride_a, double* invA, rocblas_int ldinvA, rocblas_int stride_invA, rocblas_int batch_count); /*! \brief BLAS Level 3 API \details trsm solves op(A)*X = alpha*B or X*op(A) = alpha*B, where alpha is a scalar, X and B are m by n matrices, A is triangular matrix and op(A) is one of op( A ) = A or op( A ) = A^T or op( A ) = A^H. The matrix X is overwritten on B. Note about memory allocation: When trsm is launched with a k evenly divisible by the internal block size of 128, and is no larger than 10 of these blocks, the API takes advantage of utilizing pre-allocated memory found in the handle to increase overall performance. This memory can be managed by using the environment variable WORKBUF_TRSM_B_CHNK. When this variable is not set the device memory used for temporary storage will default to 1 MB and may result in chunking, which in turn may reduce performance. Under these circumstances it is recommended that WORKBUF_TRSM_B_CHNK be set to the desired chunk of right hand sides to be used at a time. (where k is m when rocblas_side_left and is n when rocblas_side_right) @param[in] handle rocblas_handle. handle to the rocblas library context queue. @param[in] side rocblas_side. rocblas_side_left: op(A)*X = alpha*B. rocblas_side_right: X*op(A) = alpha*B. @param[in] uplo rocblas_fill. rocblas_fill_upper: A is an upper triangular matrix. rocblas_fill_lower: A is a lower triangular matrix. @param[in] transA rocblas_operation. transB: op(A) = A. rocblas_operation_transpose: op(A) = A^T. rocblas_operation_conjugate_transpose: op(A) = A^H. @param[in] diag rocblas_diagonal. rocblas_diagonal_unit: A is assumed to be unit triangular. rocblas_diagonal_non_unit: A is not assumed to be unit triangular. @param[in] m rocblas_int. m specifies the number of rows of B. m >= 0. @param[in] n rocblas_int. n specifies the number of columns of B. n >= 0. @param[in] alpha alpha specifies the scalar alpha. When alpha is &zero then A is not referenced and B need not be set before entry. @param[in] A pointer storing matrix A on the GPU. of dimension ( lda, k ), where k is m when rocblas_side_left and is n when rocblas_side_right only the upper/lower triangular part is accessed. @param[in] lda rocblas_int. lda specifies the first dimension of A. if side = rocblas_side_left, lda >= max( 1, m ), if side = rocblas_side_right, lda >= max( 1, n ). @param[in,output] B pointer storing matrix B on the GPU. @param[in] ldb rocblas_int. ldb specifies the first dimension of B. ldb >= max( 1, m ). ********************************************************************/ ROCBLAS_EXPORT rocblas_status rocblas_strsm(rocblas_handle handle, rocblas_side side, rocblas_fill uplo, rocblas_operation transA, rocblas_diagonal diag, rocblas_int m, rocblas_int n, const float* alpha, const float* A, rocblas_int lda, float* B, rocblas_int ldb); ROCBLAS_EXPORT rocblas_status rocblas_dtrsm(rocblas_handle handle, rocblas_side side, rocblas_fill uplo, rocblas_operation transA, rocblas_diagonal diag, rocblas_int m, rocblas_int n, const double* alpha, const double* A, rocblas_int lda, double* B, rocblas_int ldb); /*! \brief BLAS Level 3 API \details xGEMM performs one of the matrix-matrix operations C = alpha*op( A )*op( B ) + beta*C, where op( X ) is one of op( X ) = X or op( X ) = X**T or op( X ) = X**H, alpha and beta are scalars, and A, B and C are matrices, with op( A ) an m by k matrix, op( B ) a k by n matrix and C an m by n matrix. @param[in] handle rocblas_handle, handle to the rocblas library context queue. @param[in] transA rocblas_operation, specifies the form of op( A ) @param[in] transB rocblas_operation, specifies the form of op( B ) @param[in] m rocblas_int, number or rows of matrices op( A ) and C @param[in] n rocblas_int, number of columns of matrices op( B ) and C @param[in] k rocblas_int, number of columns of matrix op( A ) and number of rows of matrix op( B ) @param[in] alpha specifies the scalar alpha. @param[in] A pointer storing matrix A on the GPU. @param[in] lda rocblas_int, specifies the leading dimension of A. @param[in] B pointer storing matrix B on the GPU. @param[in] ldb rocblas_int, specifies the leading dimension of B. @param[in] beta specifies the scalar beta. @param[in, out] C pointer storing matrix C on the GPU. @param[in] ldc rocblas_int, specifies the leading dimension of C. ********************************************************************/ ROCBLAS_EXPORT rocblas_status rocblas_hgemm(rocblas_handle handle, rocblas_operation transa, rocblas_operation transb, rocblas_int m, rocblas_int n, rocblas_int k, const rocblas_half* alpha, const rocblas_half* A, rocblas_int lda, const rocblas_half* B, rocblas_int ldb, const rocblas_half* beta, rocblas_half* C, rocblas_int ldc); ROCBLAS_EXPORT rocblas_status rocblas_sgemm(rocblas_handle handle, rocblas_operation transa, rocblas_operation transb, rocblas_int m, rocblas_int n, rocblas_int k, const float* alpha, const float* A, rocblas_int lda, const float* B, rocblas_int ldb, const float* beta, float* C, rocblas_int ldc); ROCBLAS_EXPORT rocblas_status rocblas_dgemm(rocblas_handle handle, rocblas_operation transa, rocblas_operation transb, rocblas_int m, rocblas_int n, rocblas_int k, const double* alpha, const double* A, rocblas_int lda, const double* B, rocblas_int ldb, const double* beta, double* C, rocblas_int ldc); /* not implemented ROCBLAS_EXPORT rocblas_status rocblas_qgemm( rocblas_handle handle, rocblas_operation transa, rocblas_operation transb, rocblas_int m, rocblas_int n, rocblas_int k, const rocblas_half_complex *alpha, const rocblas_half_complex *A, rocblas_int lda, const rocblas_half_complex *B, rocblas_int ldb, const rocblas_half_complex *beta, rocblas_half_complex *C, rocblas_int ldc); */ /* not implemented ROCBLAS_EXPORT rocblas_status rocblas_cgemm( rocblas_handle handle, rocblas_operation transa, rocblas_operation transb, rocblas_int m, rocblas_int n, rocblas_int k, const rocblas_float_complex *alpha, const rocblas_float_complex *A, rocblas_int lda, const rocblas_float_complex *B, rocblas_int ldb, const rocblas_float_complex *beta, rocblas_float_complex *C, rocblas_int ldc); ROCBLAS_EXPORT rocblas_status rocblas_zgemm( rocblas_handle handle, rocblas_operation transa, rocblas_operation transb, rocblas_int m, rocblas_int n, rocblas_int k, const rocblas_double_complex *alpha, const rocblas_double_complex *A, rocblas_int lda, const rocblas_double_complex *B, rocblas_int ldb, const rocblas_double_complex *beta, rocblas_double_complex *C, rocblas_int ldc); */ /*************************************************************************** * batched * stride_a - "batch stride a": stride from the start of one "A" matrix to the next * stride_b * stride_c * batch_count - numbers of gemm's in the batch **************************************************************************/ /*! \brief BLAS Level 3 API \details xGEMM_STRIDED_BATCHED performs one of the strided batched matrix-matrix operations C[i*stride_c] = alpha*op( A[i*stride_a] )*op( B[i*stride_b] ) + beta*C[i*stride_c], for i in [0,batch_count-1] where op( X ) is one of op( X ) = X or op( X ) = X**T or op( X ) = X**H, alpha and beta are scalars, and A, B and C are strided batched matrices, with op( A ) an m by k by batch_count strided_batched matrix, op( B ) an k by n by batch_count strided_batched matrix and C an m by n by batch_count strided_batched matrix. @param[in] handle rocblas_handle. handle to the rocblas library context queue. @param[in] transA rocblas_operation specifies the form of op( A ) @param[in] transB rocblas_operation specifies the form of op( B ) @param[in] m rocblas_int. matrix dimention m. @param[in] n rocblas_int. matrix dimention n. @param[in] k rocblas_int. matrix dimention k. @param[in] alpha specifies the scalar alpha. @param[in] A pointer storing strided batched matrix A on the GPU. @param[in] lda rocblas_int specifies the leading dimension of "A". @param[in] stride_a rocblas_int stride from the start of one "A" matrix to the next @param[in] B pointer storing strided batched matrix B on the GPU. @param[in] ldb rocblas_int specifies the leading dimension of "B". @param[in] stride_b rocblas_int stride from the start of one "B" matrix to the next @param[in] beta specifies the scalar beta. @param[in, out] C pointer storing strided batched matrix C on the GPU. @param[in] ldc rocblas_int specifies the leading dimension of "C". @param[in] stride_c rocblas_int stride from the start of one "C" matrix to the next @param[in] batch_count rocblas_int number of gemm operatons in the batch ********************************************************************/ ROCBLAS_EXPORT rocblas_status rocblas_hgemm_strided_batched(rocblas_handle handle, rocblas_operation transa, rocblas_operation transb, rocblas_int m, rocblas_int n, rocblas_int k, const rocblas_half* alpha, const rocblas_half* A, rocblas_int lda, rocblas_int stride_a, const rocblas_half* B, rocblas_int ldb, rocblas_int stride_b, const rocblas_half* beta, rocblas_half* C, rocblas_int ldc, rocblas_int stride_c, rocblas_int batch_count); ROCBLAS_EXPORT rocblas_status rocblas_sgemm_strided_batched(rocblas_handle handle, rocblas_operation transa, rocblas_operation transb, rocblas_int m, rocblas_int n, rocblas_int k, const float* alpha, const float* A, rocblas_int lda, rocblas_int stride_a, const float* B, rocblas_int ldb, rocblas_int stride_b, const float* beta, float* C, rocblas_int ldc, rocblas_int stride_c, rocblas_int batch_count); ROCBLAS_EXPORT rocblas_status rocblas_dgemm_strided_batched(rocblas_handle handle, rocblas_operation transa, rocblas_operation transb, rocblas_int m, rocblas_int n, rocblas_int k, const double* alpha, const double* A, rocblas_int lda, rocblas_int stride_a, const double* B, rocblas_int ldb, rocblas_int stride_b, const double* beta, double* C, rocblas_int ldc, rocblas_int stride_c, rocblas_int batch_count); ROCBLAS_EXPORT rocblas_status rocblas_hgemm_kernel_name(rocblas_handle handle, rocblas_operation transa, rocblas_operation transb, rocblas_int m, rocblas_int n, rocblas_int k, const rocblas_half* alpha, const rocblas_half* A, rocblas_int lda, rocblas_int stride_a, const rocblas_half* B, rocblas_int ldb, rocblas_int stride_b, const rocblas_half* beta, rocblas_half* C, rocblas_int ldc, rocblas_int stride_c, rocblas_int batch_count); ROCBLAS_EXPORT rocblas_status rocblas_sgemm_kernel_name(rocblas_handle handle, rocblas_operation transa, rocblas_operation transb, rocblas_int m, rocblas_int n, rocblas_int k, const float* alpha, const float* A, rocblas_int lda, rocblas_int stride_a, const float* B, rocblas_int ldb, rocblas_int stride_b, const float* beta, float* C, rocblas_int ldc, rocblas_int stride_c, rocblas_int batch_count); ROCBLAS_EXPORT rocblas_status rocblas_dgemm_kernel_name(rocblas_handle handle, rocblas_operation transa, rocblas_operation transb, rocblas_int m, rocblas_int n, rocblas_int k, const double* alpha, const double* A, rocblas_int lda, rocblas_int stride_a, const double* B, rocblas_int ldb, rocblas_int stride_b, const double* beta, double* C, rocblas_int ldc, rocblas_int stride_c, rocblas_int batch_count); /* not implemented ROCBLAS_EXPORT rocblas_status rocblas_qgemm_strided_batched( rocblas_handle handle, rocblas_operation transa, rocblas_operation transb, rocblas_int m, rocblas_int n, rocblas_int k, const rocblas_half_complex *alpha, const rocblas_half_complex *A, rocblas_int lda, rocblas_int stride_a, const rocblas_half_complex *B, rocblas_int ldb, rocblas_int stride_b, const rocblas_half_complex *beta, rocblas_half_complex *C, rocblas_int ldc, rocblas_int stride_c, rocblas_int batch_count ); */ /* not implemented ROCBLAS_EXPORT rocblas_status rocblas_cgemm_strided_batched( rocblas_handle handle, rocblas_operation transa, rocblas_operation transb, rocblas_int m, rocblas_int n, rocblas_int k, const rocblas_float_complex *alpha, const rocblas_float_complex *A, rocblas_int lda, rocblas_int stride_a, const rocblas_float_complex *B, rocblas_int ldb, rocblas_int stride_b, const rocblas_float_complex *beta, rocblas_float_complex *C, rocblas_int ldc, rocblas_int stride_c, rocblas_int batch_count ); ROCBLAS_EXPORT rocblas_status rocblas_zgemm_strided_batched( rocblas_handle handle, rocblas_operation transa, rocblas_operation transb, rocblas_int m, rocblas_int n, rocblas_int k, const rocblas_double_complex *alpha, const rocblas_double_complex *A, rocblas_int lda, rocblas_int stride_a, const rocblas_double_complex *B, rocblas_int ldb, rocblas_int stride_b, const rocblas_double_complex *beta, rocblas_double_complex *C, rocblas_int ldc, rocblas_int stride_c, rocblas_int batch_count ); */ /*! \brief BLAS Level 3 API \details xGEAM performs one of the matrix-matrix operations C = alpha*op( A ) + beta*op( B ), where op( X ) is one of op( X ) = X or op( X ) = X**T or op( X ) = X**H, alpha and beta are scalars, and A, B and C are matrices, with op( A ) an m by n matrix, op( B ) an m by n matrix, and C an m by n matrix. @param[in] handle rocblas_handle. handle to the rocblas library context queue. @param[in] transA rocblas_operation specifies the form of op( A ) @param[in] transB rocblas_operation specifies the form of op( B ) @param[in] m rocblas_int. @param[in] n rocblas_int. @param[in] alpha specifies the scalar alpha. @param[in] A pointer storing matrix A on the GPU. @param[in] lda rocblas_int specifies the leading dimension of A. @param[in] beta specifies the scalar beta. @param[in] B pointer storing matrix B on the GPU. @param[in] ldb rocblas_int specifies the leading dimension of B. @param[in, out] C pointer storing matrix C on the GPU. @param[in] ldc rocblas_int specifies the leading dimension of C. ********************************************************************/ ROCBLAS_EXPORT rocblas_status rocblas_sgeam(rocblas_handle handle, rocblas_operation transa, rocblas_operation transb, rocblas_int m, rocblas_int n, const float* alpha, const float* A, rocblas_int lda, const float* beta, const float* B, rocblas_int ldb, float* C, rocblas_int ldc); ROCBLAS_EXPORT rocblas_status rocblas_dgeam(rocblas_handle handle, rocblas_operation transa, rocblas_operation transb, rocblas_int m, rocblas_int n, const double* alpha, const double* A, rocblas_int lda, const double* beta, const double* B, rocblas_int ldb, double* C, rocblas_int ldc); /* * =========================================================================== * BLAS extensions * =========================================================================== */ /*! \brief BLAS EX API \details GEMM_EX performs one of the matrix-matrix operations D = alpha*op( A )*op( B ) + beta*C, where op( X ) is one of op( X ) = X or op( X ) = X**T or op( X ) = X**H, alpha and beta are scalars, and A, B, C, and D are matrices, with op( A ) an m by k matrix, op( B ) a k by n matrix and C and D are m by n matrices. Supported types are as follows: - rocblas_datatype_f64_r = a_type = b_type = c_type = d_type = compute_type - rocblas_datatype_f32_r = a_type = b_type = c_type = d_type = compute_type - rocblas_datatype_f16_r = a_type = b_type = c_type = d_type = compute_type - rocblas_datatype_f16_r = a_type = b_type = c_type = d_type; rocblas_datatype_f32_r = compute_type - rocblas_datatype_bf16_r = a_type = b_type = c_type = d_type; rocblas_datatype_f32_r = compute_type - rocblas_datatype_i8_r = a_type = b_type; rocblas_datatype_i32_r = c_type = d_type = compute_type Below are restrictions for rocblas_datatype_i8_r = a_type = b_type; rocblas_datatype_i32_r = c_type = d_type = compute_type: - k must be a multiple of 4 - lda must be a multiple of 4 if transA == rocblas_operation_transpose - ldb must be a multiple of 4 if transB == rocblas_operation_none - for transA == rocblas_operation_transpose or transB == rocblas_operation_none the matrices A and B must have each 4 consecutive values in the k dimension packed. This packing can be achieved with the following pseudo-code. The code assumes the original matrices are in A and B, and the packed matrices are A_packed and B_packed. The size of the A_packed matrix is the same as the size of the A matrix, and the size of the B_packed matrix is the same as the size of the B matrix. @code if(transA == rocblas_operation_none) { int nb = 4; for(int i_m = 0; i_m < m; i_m++) { for(int i_k = 0; i_k < k; i_k++) { A_packed[i_k % nb + (i_m + (i_k / nb) * lda) * nb] = A[i_m + i_k * lda]; } } } else { A_packed = A; } if(trans_b == rocblas_operation_transpose) { int nb = 4; for(int i_n = 0; i_n < m; i_n++) { for(int i_k = 0; i_k < k; i_k++) { B_packed[i_k % nb + (i_n + (i_k / nb) * lda) * nb] = B[i_n + i_k * lda]; } } } else { B_packed = B; } @endcode @param[in] handle rocblas_handle. handle to the rocblas library context queue. @param[in] transA rocblas_operation. specifies the form of op( A ). @param[in] transB rocblas_operation specifies the form of op( B ). @param[in] m rocblas_int. matrix dimension m. @param[in] n rocblas_int. matrix dimension n. @param[in] k rocblas_int. matrix dimension k. @param[in] alpha const void *. specifies the scalar alpha. Same datatype as compute_type. @param[in] a void *. pointer storing matrix A on the GPU. @param[in] a_type rocblas_datatype. specifies the datatype of matrix A. @param[in] lda rocblas_int. specifies the leading dimension of A. @param[in] b void *. pointer storing matrix B on the GPU. @param[in] b_type rocblas_datatype. specifies the datatype of matrix B. @param[in] ldb rocblas_int. specifies the leading dimension of B. @param[in] beta const void *. specifies the scalar beta. Same datatype as compute_type. @param[in] c void *. pointer storing matrix C on the GPU. @param[in] c_type rocblas_datatype. specifies the datatype of matrix C. @param[in] ldc rocblas_int. specifies the leading dimension of C. @param[out] d void *. pointer storing matrix D on the GPU. @param[in] d_type rocblas_datatype. specifies the datatype of matrix D. @param[in] ldd rocblas_int. specifies the leading dimension of D. @param[in] compute_type rocblas_datatype. specifies the datatype of computation. @param[in] algo rocblas_gemm_algo. enumerant specifying the algorithm type. @param[in] solution_index int32_t. reserved for future use. @param[in] flags uint32_t. reserved for future use. ********************************************************************/ ROCBLAS_EXPORT rocblas_status rocblas_gemm_ex(rocblas_handle handle, rocblas_operation transA, rocblas_operation trans_b, rocblas_int m, rocblas_int n, rocblas_int k, const void* alpha, const void* a, rocblas_datatype a_type, rocblas_int lda, const void* b, rocblas_datatype b_type, rocblas_int ldb, const void* beta, const void* c, rocblas_datatype c_type, rocblas_int ldc, void* d, rocblas_datatype d_type, rocblas_int ldd, rocblas_datatype compute_type, rocblas_gemm_algo algo, int32_t solution_index, uint32_t flags); /* For backward compatiblity, unused workspace_size and workspace arguments are ignored */ // clang-format off #define rocblas_gemm_ex(handle, \ transA, \ trans_b, \ m, \ n, \ k, \ alpha, \ a, \ a_type, \ lda, \ b, \ b_type, \ ldb, \ beta, \ c, \ c_type, \ ldc, \ d, \ d_type, \ ldd, \ compute_type, \ algo, \ solution_index, \ flags, \ ...) \ ROCBLAS_VA_OPT_PRAGMA(GCC warning "rocblas_gemm_ex: The workspace_size and workspace arguments are obsolete, and will be ignored", __VA_ARGS__) \ rocblas_gemm_ex(handle, \ transA, \ trans_b, \ m, \ n, \ k, \ alpha, \ a, \ a_type, \ lda, \ b, \ b_type, \ ldb, \ beta, \ c, \ c_type, \ ldc, \ d, \ d_type, \ ldd, \ compute_type, \ algo, \ solution_index, \ flags) // clang-format on /*! \brief BLAS EX API \details GEMM_STRIDED_BATCHED_EX performs one of the strided_batched matrix-matrix operations D[i*stride_d] = alpha*op(A[i*stride_a])*op(B[i*stride_b]) + beta*C[i*stride_c], for i in [0,batch_count-1] where op( X ) is one of op( X ) = X or op( X ) = X**T or op( X ) = X**H, alpha and beta are scalars, and A, B, C, and D are strided_batched matrices, with op( A ) an m by k by batch_count strided_batched matrix, op( B ) a k by n by batch_count strided_batched matrix and C and D are m by n by batch_count strided_batched matrices. The strided_batched matrices are multiple matrices separated by a constant stride. The number of matrices is batch_count. Supported types are as follows: - rocblas_datatype_f64_r = a_type = b_type = c_type = d_type = compute_type - rocblas_datatype_f32_r = a_type = b_type = c_type = d_type = compute_type - rocblas_datatype_f16_r = a_type = b_type = c_type = d_type = compute_type - rocblas_datatype_f16_r = a_type = b_type = c_type = d_type; rocblas_datatype_f32_r = compute_type - rocblas_datatype_bf16_r = a_type = b_type = c_type = d_type; rocblas_datatype_f32_r = compute_type - rocblas_datatype_i8_r = a_type = b_type; rocblas_datatype_i32_r = c_type = d_type = compute_type Below are restrictions for rocblas_datatype_i8_r = a_type = b_type; rocblas_datatype_i32_r = c_type = d_type = compute_type: - k must be a multiple of 4 - lda must be a multiple of 4 if transA == rocblas_operation_transpose - ldb must be a multiple of 4 if transB == rocblas_operation_none - for transA == rocblas_operation_transpose or transB == rocblas_operation_none the matrices A and B must have each 4 consecutive values in the k dimension packed. This packing can be achieved with the following pseudo-code. The code assumes the original matrices are in A and B, and the packed matrices are A_packed and B_packed. The size of the A_packed matrix is the same as the size of the A matrix, and the size of the B_packed matrix is the same as the size of the B matrix. @code if(transA == rocblas_operation_none) { int nb = 4; for(int i_m = 0; i_m < m; i_m++) { for(int i_k = 0; i_k < k; i_k++) { A_packed[i_k % nb + (i_m + (i_k / nb) * lda) * nb] = A[i_m + i_k * lda]; } } } else { A_packed = A; } if(trans_b == rocblas_operation_transpose) { int nb = 4; for(int i_n = 0; i_n < m; i_n++) { for(int i_k = 0; i_k < k; i_k++) { B_packed[i_k % nb + (i_n + (i_k / nb) * lda) * nb] = B[i_n + i_k * lda]; } } } else { B_packed = B; } @endcode @param[in] handle rocblas_handle. handle to the rocblas library context queue. @param[in] transA rocblas_operation. specifies the form of op( A ). @param[in] transB rocblas_operation. specifies the form of op( B ). @param[in] m rocblas_int. matrix dimension m. @param[in] n rocblas_int. matrix dimension n. @param[in] k rocblas_int. matrix dimension k. @param[in] alpha const void *. specifies the scalar alpha. Same datatype as compute_type. @param[in] a void *. pointer storing matrix A on the GPU. @param[in] a_type rocblas_datatype. specifies the datatype of matrix A. @param[in] lda rocblas_int. specifies the leading dimension of A. @param[in] stride_a rocblas_long. specifies stride from start of one "A" matrix to the next. @param[in] b void *. pointer storing matrix B on the GPU. @param[in] b_type rocblas_datatype. specifies the datatype of matrix B. @param[in] ldb rocblas_int. specifies the leading dimension of B. @param[in] stride_b rocblas_long. specifies stride from start of one "B" matrix to the next. @param[in] beta const void *. specifies the scalar beta. Same datatype as compute_type. @param[in] c void *. pointer storing matrix C on the GPU. @param[in] c_type rocblas_datatype. specifies the datatype of matrix C. @param[in] ldc rocblas_int. specifies the leading dimension of C. @param[in] stride_c rocblas_long. specifies stride from start of one "C" matrix to the next. @param[out] d void *. pointer storing matrix D on the GPU. @param[in] d_type rocblas_datatype. specifies the datatype of matrix D. @param[in] ldd rocblas_int. specifies the leading dimension of D. @param[in] stride_d rocblas_long. specifies stride from start of one "D" matrix to the next. @param[in] batch_count rocblas_int. number of gemm operations in the batch. @param[in] compute_type rocblas_datatype. specifies the datatype of computation. @param[in] algo rocblas_gemm_algo. enumerant specifying the algorithm type. @param[in] solution_index int32_t. reserved for future use. @param[in] flags uint32_t. reserved for future use. ********************************************************************/ ROCBLAS_EXPORT rocblas_status rocblas_gemm_strided_batched_ex(rocblas_handle handle, rocblas_operation transA, rocblas_operation trans_b, rocblas_int m, rocblas_int n, rocblas_int k, const void* alpha, const void* a, rocblas_datatype a_type, rocblas_int lda, rocblas_long stride_a, const void* b, rocblas_datatype b_type, rocblas_int ldb, rocblas_long stride_b, const void* beta, const void* c, rocblas_datatype c_type, rocblas_int ldc, rocblas_long stride_c, void* d, rocblas_datatype d_type, rocblas_int ldd, rocblas_long stride_d, rocblas_int batch_count, rocblas_datatype compute_type, rocblas_gemm_algo algo, int32_t solution_index, uint32_t flags); /* For backward compatiblity, unused workspace_size and workspace arguments are ignored */ // clang-format off #define rocblas_gemm_strided_batched_ex(handle, \ transA, \ trans_b, \ m, \ n, \ k, \ alpha, \ a, \ a_type, \ lda, \ stride_a, \ b, \ b_type, \ ldb, \ stride_b, \ beta, \ c, \ c_type, \ ldc, \ stride_c, \ d, \ d_type, \ ldd, \ stride_d, \ batch_count, \ compute_type, \ algo, \ solution_index, \ flags, \ ...) \ ROCBLAS_VA_OPT_PRAGMA(GCC warning "rocblas_gemm_strided_batched_ex: The workspace_size and workspace arguments are obsolete, and will be ignored", __VA_ARGS__) \ rocblas_gemm_strided_batched_ex(handle, \ transA, \ trans_b, \ m, \ n, \ k, \ alpha, \ a, \ a_type, \ lda, \ stride_a, \ b, \ b_type, \ ldb, \ stride_b, \ beta, \ c, \ c_type, \ ldc, \ stride_c, \ d, \ d_type, \ ldd, \ stride_d, \ batch_count, \ compute_type, \ algo, \ solution_index, \ flags) // clang-format on /*! BLAS EX API \details TRSM_EX solves op(A)*X = alpha*B or X*op(A) = alpha*B, where alpha is a scalar, X and B are m by n matrices, A is triangular matrix and op(A) is one of op( A ) = A or op( A ) = A^T or op( A ) = A^H. The matrix X is overwritten on B. TRSM_EX gives the user the ability to reuse the invA matrix between runs. If invA == NULL, rocblas_trsm_ex will automatically calculate invA on every run. Setting up invA: The accepted invA matrix consists of the packed 128x128 inverses of the diagonal blocks of matrix A, followed by any smaller diagonal block that remains. To set up invA it is recommended that rocblas_trtri_batched be used with matrix A as the input. Device memory of size 128 x k should be allocated for invA ahead of time, where k is m when rocblas_side_left and is n when rocblas_side_right. The actual number of elements in invA should be passed as invA_size. To begin, rocblas_trtri_batched must be called on the full 128x128 sized diagonal blocks of matrix A. Below are the restricted parameters: - n = 128 - ldinvA = 128 - stride_invA = 128x128 - batch_count = k / 128, Then any remaining block may be added: - n = k % 128 - invA = invA + stride_invA * previous_batch_count - ldinvA = 128 - batch_count = 1 @param[in] handle rocblas_handle handle to the rocblas library context queue. @param[in] side rocblas_side rocblas_side_left: op(A)*X = alpha*B. rocblas_side_right: X*op(A) = alpha*B. @param[in] uplo rocblas_fill rocblas_fill_upper: A is an upper triangular matrix. rocblas_fill_lower: A is a lower triangular matrix. @param[in] transA rocblas_operation transB: op(A) = A. rocblas_operation_transpose: op(A) = A^T. rocblas_operation_conjugate_transpose: op(A) = A^H. @param[in] diag rocblas_diagonal rocblas_diagonal_unit: A is assumed to be unit triangular. rocblas_diagonal_non_unit: A is not assumed to be unit triangular. @param[in] m rocblas_int m specifies the number of rows of B. m >= 0. @param[in] n rocblas_int n specifies the number of columns of B. n >= 0. @param[in] alpha alpha specifies the scalar alpha. When alpha is &zero then A is not referenced, and B need not be set before entry. @param[in] A void * pointer storing matrix A on the GPU. of dimension ( lda, k ), where k is m when rocblas_side_left and is n when rocblas_side_right only the upper/lower triangular part is accessed. @param[in] lda rocblas_int lda specifies the first dimension of A. if side = rocblas_side_left, lda >= max( 1, m ), if side = rocblas_side_right, lda >= max( 1, n ). @param[in, out] B void * pointer storing matrix B on the GPU. B is of dimension ( ldb, n ). Before entry, the leading m by n part of the array B must contain the right-hand side matrix B, and on exit is overwritten by the solution matrix X. @param[in] ldb rocblas_int ldb specifies the first dimension of B. ldb >= max( 1, m ). @param[in] invA void * pointer storing the inverse diagonal blocks of A on the GPU. invA is of dimension ( ld_invA, k ), where k is m when rocblas_side_left and is n when rocblas_side_right. ld_invA must be equal to 128. @param[in] invA_size rocblas_int invA_size specifies the number of elements of device memory in invA. @param[in] compute_type rocblas_datatype specifies the datatype of computation ********************************************************************/ ROCBLAS_EXPORT rocblas_status rocblas_trsm_ex(rocblas_handle handle, rocblas_side side, rocblas_fill uplo, rocblas_operation transA, rocblas_diagonal diag, rocblas_int m, rocblas_int n, const void* alpha, const void* A, rocblas_int lda, void* B, rocblas_int ldb, const void* invA, rocblas_int invA_size, rocblas_datatype compute_type); /* For backward compatiblity, option, x_temp_size and x_temp_workspace arguments are ignored */ // clang-format off #define rocblas_trsm_ex(handle, \ side, \ uplo, \ transA, \ diag, \ m, \ n, \ alpha, \ A, \ lda, \ B, \ ldb, \ invA, \ invA_size, \ compute_type, \ ...) \ ROCBLAS_VA_OPT_PRAGMA(GCC warning "rocblas_trsm_ex: The option, x_temp_size and x_temp_workspace arguments are obsolete, and will be ignored", __VA_ARGS__) \ rocblas_trsm_ex(handle, \ side, \ uplo, \ transA, \ diag, \ m, \ n, \ alpha, \ A, \ lda, \ B, \ ldb, \ invA, \ invA_size, \ compute_type) // clang-format on /* * =========================================================================== * build information * =========================================================================== */ /*! \brief loads char* buf with the rocblas library version. size_t len is the maximum length of char* buf. \details @param[in, out] buf pointer to buffer for version string @param[in] len length of buf ******************************************************************************/ ROCBLAS_EXPORT rocblas_status rocblas_get_version_string(char* buf, size_t len); ROCBLAS_EXPORT rocblas_status rocblas_start_device_memory_size_query(rocblas_handle handle); ROCBLAS_EXPORT rocblas_status rocblas_stop_device_memory_size_query(rocblas_handle handle, size_t* size); ROCBLAS_EXPORT rocblas_status rocblas_get_device_memory_size(rocblas_handle handle, size_t* size); ROCBLAS_EXPORT rocblas_status rocblas_set_device_memory_size(rocblas_handle handle, size_t size); ROCBLAS_EXPORT bool rocblas_is_managing_device_memory(rocblas_handle handle); #ifdef __cplusplus } #endif #endif /* _ROCBLAS_FUNCTIONS_H_ */