Blas functions for execution on GPUs. More...

Namespaces
module	sirius
	Namespace of the SIRIUS library.

namespace	sirius::acc
	Namespace for accelerator-related functions.

namespace	sirius::acc::blas
	User facing interface to GPU blas functions.

namespace	sirius::acc::blas::xt
	Interface to cuBlasXt functions.

Macros
#define	CALL_GPU_BLAS(func__, args__)

Functions
const char *	sirius::acc::blas::error_message (acc::blas_api::status_t status)

acc::blas_api::operation_t	sirius::acc::blas::get_gpublasOperation_t (char c)

acc::blas_api::side_mode_t	sirius::acc::blas::get_gpublasSideMode_t (char c)

acc::blas_api::fill_mode_t	sirius::acc::blas::get_gpublasFillMode_t (char c)

acc::blas_api::diagonal_t	sirius::acc::blas::get_gpublasDiagonal_t (char c)

acc::blas_api::handle_t &	sirius::acc::blas::null_stream_handle ()
	Store the default (null) stream handler. More...

std::vector< acc::blas_api::handle_t > &	sirius::acc::blas::stream_handles ()
	Store the gpublas handlers associated with acc streams. More...

void	sirius::acc::blas::create_stream_handles ()

void	sirius::acc::blas::destroy_stream_handles ()

acc::blas_api::handle_t	sirius::acc::blas::stream_handle (int id__)

void	sirius::acc::blas::zgemv (char transa, int32_t m, int32_t n, acc_complex_double_t alpha, acc_complex_double_t a, int32_t lda, acc_complex_double_t x, int32_t incx, acc_complex_double_t beta, acc_complex_double_t *y, int32_t incy, int stream_id)

void	sirius::acc::blas::cgemm (char transa, char transb, int32_t m, int32_t n, int32_t k, acc_complex_float_t const alpha, acc_complex_float_t const a, int32_t lda, acc_complex_float_t const b, int32_t ldb, acc_complex_float_t const beta, acc_complex_float_t *c, int32_t ldc, int stream_id)

void	sirius::acc::blas::zgemm (char transa, char transb, int32_t m, int32_t n, int32_t k, acc_complex_double_t const alpha, acc_complex_double_t const a, int32_t lda, acc_complex_double_t const b, int32_t ldb, acc_complex_double_t const beta, acc_complex_double_t *c, int32_t ldc, int stream_id)

void	sirius::acc::blas::sgemm (char transa, char transb, int32_t m, int32_t n, int32_t k, float const alpha, float const a, int32_t lda, float const b, int32_t ldb, float const beta, float *c, int32_t ldc, int stream_id)

void	sirius::acc::blas::dgemm (char transa, char transb, int32_t m, int32_t n, int32_t k, double const alpha, double const a, int32_t lda, double const b, int32_t ldb, double const beta, double *c, int32_t ldc, int stream_id)

void	sirius::acc::blas::strmm (char side__, char uplo__, char transa__, char diag__, int m__, int n__, float const alpha__, float const A__, int lda__, float *B__, int ldb__, int stream_id)

void	sirius::acc::blas::dtrmm (char side__, char uplo__, char transa__, char diag__, int m__, int n__, double const alpha__, double const A__, int lda__, double *B__, int ldb__, int stream_id)

void	sirius::acc::blas::ctrmm (char side__, char uplo__, char transa__, char diag__, int m__, int n__, acc_complex_float_t const alpha__, acc_complex_float_t const A__, int lda__, acc_complex_float_t *B__, int ldb__, int stream_id)

void	sirius::acc::blas::ztrmm (char side__, char uplo__, char transa__, char diag__, int m__, int n__, acc_complex_double_t const alpha__, acc_complex_double_t const A__, int lda__, acc_complex_double_t *B__, int ldb__, int stream_id)

void	sirius::acc::blas::sger (int m, int n, float const alpha, float const x, int incx, float const y, int incy, float A, int lda, int stream_id)

void	sirius::acc::blas::dger (int m, int n, double const alpha, double const x, int incx, double const y, int incy, double A, int lda, int stream_id)

void	sirius::acc::blas::cgeru (int m, int n, acc_complex_float_t const alpha, acc_complex_float_t const x, int incx, acc_complex_float_t const y, int incy, acc_complex_float_t A, int lda, int stream_id)

void	sirius::acc::blas::zgeru (int m, int n, acc_complex_double_t const alpha, acc_complex_double_t const x, int incx, acc_complex_double_t const y, int incy, acc_complex_double_t A, int lda, int stream_id)

void	sirius::acc::blas::zaxpy (int n__, acc_complex_double_t const alpha__, acc_complex_double_t const x__, int incx__, acc_complex_double_t *y__, int incy__)

void	sirius::acc::blas::dscal (int n__, double const alpha__, double x__, int incx__)

void	sirius::acc::blas::sscal (int n__, float const alpha__, float x__, int incx__)

cublasXtHandle_t &	sirius::acc::blas::xt::cublasxt_handle ()

void	sirius::acc::blas::xt::create_handle ()

void	sirius::acc::blas::xt::destroy_handle ()

void	sirius::acc::blas::xt::cgemm (char transa, char transb, int32_t m, int32_t n, int32_t k, acc_complex_float_t const alpha, acc_complex_float_t const a, int32_t lda, acc_complex_float_t const b, int32_t ldb, acc_complex_float_t const beta, acc_complex_float_t *c, int32_t ldc)

void	sirius::acc::blas::xt::zgemm (char transa, char transb, int32_t m, int32_t n, int32_t k, acc_complex_double_t const alpha, acc_complex_double_t const a, int32_t lda, acc_complex_double_t const b, int32_t ldb, acc_complex_double_t const beta, acc_complex_double_t *c, int32_t ldc)

void	sirius::acc::blas::xt::sgemm (char transa, char transb, int32_t m, int32_t n, int32_t k, float const alpha, float const a, int32_t lda, float const b, int32_t ldb, float const beta, float *c, int32_t ldc)

void	sirius::acc::blas::xt::dgemm (char transa, char transb, int32_t m, int32_t n, int32_t k, double const alpha, double const a, int32_t lda, double const b, int32_t ldb, double const beta, double *c, int32_t ldc)

void	sirius::acc::blas::xt::strmm (char side__, char uplo__, char transa__, char diag__, int m__, int n__, float const alpha__, float const A__, int lda__, float *B__, int ldb__)

void	sirius::acc::blas::xt::dtrmm (char side__, char uplo__, char transa__, char diag__, int m__, int n__, double const alpha__, double const A__, int lda__, double *B__, int ldb__)

void	sirius::acc::blas::xt::ctrmm (char side__, char uplo__, char transa__, char diag__, int m__, int n__, acc_complex_float_t const alpha__, acc_complex_float_t const A__, int lda__, acc_complex_float_t *B__, int ldb__)

void	sirius::acc::blas::xt::ztrmm (char side__, char uplo__, char transa__, char diag__, int m__, int n__, acc_complex_double_t const alpha__, acc_complex_double_t const A__, int lda__, acc_complex_double_t *B__, int ldb__)

Detailed Description

Blas functions for execution on GPUs.

Definition in file acc_blas.hpp.

Macro Definition Documentation

◆ CALL_GPU_BLAS

#define CALL_GPU_BLAS	(	func__,
		args__
	)

Value:

    {                                                                                                                  \
        acc::blas_api::status_t status;                                                                                  \
        if ((status = func__ args__) != acc::blas_api::status::Success) {                                                \
            error_message(status);                                                                                     \
            char nm[1024];                                                                                             \
            gethostname(nm, 1024);                                                                                     \
            std::printf("hostname: %s\n", nm);                                                                         \
            std::printf("Error in %s at line %i of file %s\n", #func__, __LINE__, __FILE__);                           \
            acc::stack_backtrace();                                                                                         \
        }                                                                                                              \
    }

Definition at line 154 of file acc_blas.hpp.

Namespaces

Macros

Functions

Detailed Description

Macro Definition Documentation

◆ CALL_GPU_BLAS