Sponsor:

Your company here, and a link to your site. Click to find out more.

HPL_dgemm - Man Page

C := alpha * op(A) * op(B) + beta * C.

Synopsis

#include "hpl.h"
 void HPL_dgemm( const enum HPL_ORDER ORDER, const enum HPL_TRANS TRANSA, const enum HPL_TRANS TRANSB, const int M, const int N, const int K, const double ALPHA, const double * A, const int LDA, const double * B, const int LDB, const double BETA, double * C, const int LDC );

Description

HPL_dgemm 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.
 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.

Arguments

ORDER   (local input)           const enum HPL_ORDER

On entry, ORDER  specifies the storage format of the operands as follows:                                                  
  ORDER = HplRowMajor,                                      
  ORDER = HplColumnMajor.                                  

TRANSA  (local input)           const enum HPL_TRANS

On entry, TRANSA  specifies the form of  op(A)  to be used in the matrix-matrix operation follows:                        
  TRANSA==HplNoTrans    : op( A ) = A,                    
  TRANSA==HplTrans      : op( A ) = A^T,                  
  TRANSA==HplConjTrans  : op( A ) = A^T.                  

TRANSB  (local input)           const enum HPL_TRANS

On entry, TRANSB  specifies the form of  op(B)  to be used in the matrix-matrix operation follows:                        
  TRANSB==HplNoTrans    : op( B ) = B,                    
  TRANSB==HplTrans      : op( B ) = B^T,                  
  TRANSB==HplConjTrans  : op( B ) = B^T.                  

M       (local input)           const int

On entry,  M  specifies  the  number  of rows  of the  matrix op(A)  and  of  the  matrix  C.  M  must  be  at least  zero.

N       (local input)           const int

On entry,  N  specifies  the number  of columns of the matrix op(B)  and  the number of columns of the matrix  C. N must be at least zero.

K       (local input)           const int

On entry,  K  specifies  the  number of columns of the matrix op(A) and the number of rows of the matrix op(B).  K  must be be at least  zero.

ALPHA   (local input)           const double

On entry, ALPHA specifies the scalar alpha.   When  ALPHA  is supplied  as  zero  then the elements of the matrices A and B need not be set on input.

A       (local input)           const double *

On entry,  A  is an array of dimension (LDA,ka),  where ka is k  when   TRANSA==HplNoTrans,  and  is  m  otherwise.  Before entry  with  TRANSA==HplNoTrans, the  leading  m by k part of the array  A must contain the matrix A, otherwise the leading k  by  m  part of the array  A  must  contain the  matrix  A.

LDA     (local input)           const int

On entry, LDA  specifies the first dimension of A as declared in the  calling (sub) program. When  TRANSA==HplNoTrans  then LDA must be at least max(1,m), otherwise LDA must be at least max(1,k).

B       (local input)           const double *

On entry, B is an array of dimension (LDB,kb),  where  kb  is n   when  TRANSB==HplNoTrans, and  is  k  otherwise.   Before entry with TRANSB==HplNoTrans,  the  leading  k by n  part of the array  B must contain the matrix B, otherwise the leading n  by  k  part of the array  B  must  contain  the matrix  B.

LDB     (local input)           const int

On entry, LDB  specifies the first dimension of B as declared in the  calling (sub) program. When  TRANSB==HplNoTrans  then LDB must be at least max(1,k), otherwise LDB must be at least max(1,n).

BETA    (local input)           const double

On entry,  BETA  specifies the scalar  beta.   When  BETA  is supplied  as  zero  then  the  elements of the matrix C  need not be set on input.

C       (local input/output)    double *

On entry,  C  is an array of dimension (LDC,n). Before entry, the  leading m by n part  of  the  array  C  must contain the matrix C,  except when beta is zero, in which case C need not be set on entry. On exit, the array  C  is overwritten by the m by n  matrix ( alpha*op( A )*op( B ) + beta*C ).

LDC     (local input)           const int

On entry, LDC  specifies the first dimension of C as declared in  the   calling  (sub)  program.   LDC  must  be  at  least max(1,m).

Example

#include "hpl.h"
 int main(int argc, char *argv[])
{
  double a[2*2], b[2*2], c[2*2];
  a[0] = 1.0; a[1] = 2.0; a[2] = 3.0; a[3] = 3.0;
  b[0] = 2.0; b[1] = 1.0; b[2] = 1.0; b[3] = 2.0;
  c[0] = 4.0; c[1] = 3.0; c[2] = 2.0; c[3] = 1.0;
  HPL_dgemm( HplColumnMajor, HplNoTrans, HplNoTrans,
             2, 2, 2, 2.0, a, 2, b, 2, -1.0, c, 2 );
  printf("  [%f,%f]\n", c[0], c[2]);
  printf("c=[%f,%f]\n", c[1], c[3]);
  exit(0); return(0);
}

See Also

HPL_dtrsm (3).

Referenced By

HPL_dtrsm(3).

February 24, 2016 HPL 2.2 HPL Library Functions