File:  [local] / rpl / lapack / blas / dgemm.f
Revision 1.13: download - view: text, annotated - select for diffs - revision graph
Sat Aug 27 15:37:52 2016 UTC (7 years, 8 months ago) by bertrand
Branches: MAIN
CVS tags: rpl-4_1_25, HEAD
Cohérence Blas.

    1: *> \brief \b DGEMM
    2: *
    3: *  =========== DOCUMENTATION ===========
    4: *
    5: * Online html documentation available at 
    6: *            http://www.netlib.org/lapack/explore-html/ 
    7: *
    8: *  Definition:
    9: *  ===========
   10: *
   11: *       SUBROUTINE DGEMM(TRANSA,TRANSB,M,N,K,ALPHA,A,LDA,B,LDB,BETA,C,LDC)
   12:    13: *       .. Scalar Arguments ..
   14: *       DOUBLE PRECISION ALPHA,BETA
   15: *       INTEGER K,LDA,LDB,LDC,M,N
   16: *       CHARACTER TRANSA,TRANSB
   17: *       ..
   18: *       .. Array Arguments ..
   19: *       DOUBLE PRECISION A(LDA,*),B(LDB,*),C(LDC,*)
   20: *       ..
   21: *  
   22: *
   23: *> \par Purpose:
   24: *  =============
   25: *>
   26: *> \verbatim
   27: *>
   28: *> DGEMM  performs one of the matrix-matrix operations
   29: *>
   30: *>    C := alpha*op( A )*op( B ) + beta*C,
   31: *>
   32: *> where  op( X ) is one of
   33: *>
   34: *>    op( X ) = X   or   op( X ) = X**T,
   35: *>
   36: *> alpha and beta are scalars, and A, B and C are matrices, with op( A )
   37: *> an m by k matrix,  op( B )  a  k by n matrix and  C an m by n matrix.
   38: *> \endverbatim
   39: *
   40: *  Arguments:
   41: *  ==========
   42: *
   43: *> \param[in] TRANSA
   44: *> \verbatim
   45: *>          TRANSA is CHARACTER*1
   46: *>           On entry, TRANSA specifies the form of op( A ) to be used in
   47: *>           the matrix multiplication as follows:
   48: *>
   49: *>              TRANSA = 'N' or 'n',  op( A ) = A.
   50: *>
   51: *>              TRANSA = 'T' or 't',  op( A ) = A**T.
   52: *>
   53: *>              TRANSA = 'C' or 'c',  op( A ) = A**T.
   54: *> \endverbatim
   55: *>
   56: *> \param[in] TRANSB
   57: *> \verbatim
   58: *>          TRANSB is CHARACTER*1
   59: *>           On entry, TRANSB specifies the form of op( B ) to be used in
   60: *>           the matrix multiplication as follows:
   61: *>
   62: *>              TRANSB = 'N' or 'n',  op( B ) = B.
   63: *>
   64: *>              TRANSB = 'T' or 't',  op( B ) = B**T.
   65: *>
   66: *>              TRANSB = 'C' or 'c',  op( B ) = B**T.
   67: *> \endverbatim
   68: *>
   69: *> \param[in] M
   70: *> \verbatim
   71: *>          M is INTEGER
   72: *>           On entry,  M  specifies  the number  of rows  of the  matrix
   73: *>           op( A )  and of the  matrix  C.  M  must  be at least  zero.
   74: *> \endverbatim
   75: *>
   76: *> \param[in] N
   77: *> \verbatim
   78: *>          N is INTEGER
   79: *>           On entry,  N  specifies the number  of columns of the matrix
   80: *>           op( B ) and the number of columns of the matrix C. N must be
   81: *>           at least zero.
   82: *> \endverbatim
   83: *>
   84: *> \param[in] K
   85: *> \verbatim
   86: *>          K is INTEGER
   87: *>           On entry,  K  specifies  the number of columns of the matrix
   88: *>           op( A ) and the number of rows of the matrix op( B ). K must
   89: *>           be at least  zero.
   90: *> \endverbatim
   91: *>
   92: *> \param[in] ALPHA
   93: *> \verbatim
   94: *>          ALPHA is DOUBLE PRECISION.
   95: *>           On entry, ALPHA specifies the scalar alpha.
   96: *> \endverbatim
   97: *>
   98: *> \param[in] A
   99: *> \verbatim
  100: *>          A is DOUBLE PRECISION array of DIMENSION ( LDA, ka ), where ka is
  101: *>           k  when  TRANSA = 'N' or 'n',  and is  m  otherwise.
  102: *>           Before entry with  TRANSA = 'N' or 'n',  the leading  m by k
  103: *>           part of the array  A  must contain the matrix  A,  otherwise
  104: *>           the leading  k by m  part of the array  A  must contain  the
  105: *>           matrix A.
  106: *> \endverbatim
  107: *>
  108: *> \param[in] LDA
  109: *> \verbatim
  110: *>          LDA is INTEGER
  111: *>           On entry, LDA specifies the first dimension of A as declared
  112: *>           in the calling (sub) program. When  TRANSA = 'N' or 'n' then
  113: *>           LDA must be at least  max( 1, m ), otherwise  LDA must be at
  114: *>           least  max( 1, k ).
  115: *> \endverbatim
  116: *>
  117: *> \param[in] B
  118: *> \verbatim
  119: *>          B is DOUBLE PRECISION array of DIMENSION ( LDB, kb ), where kb is
  120: *>           n  when  TRANSB = 'N' or 'n',  and is  k  otherwise.
  121: *>           Before entry with  TRANSB = 'N' or 'n',  the leading  k by n
  122: *>           part of the array  B  must contain the matrix  B,  otherwise
  123: *>           the leading  n by k  part of the array  B  must contain  the
  124: *>           matrix B.
  125: *> \endverbatim
  126: *>
  127: *> \param[in] LDB
  128: *> \verbatim
  129: *>          LDB is INTEGER
  130: *>           On entry, LDB specifies the first dimension of B as declared
  131: *>           in the calling (sub) program. When  TRANSB = 'N' or 'n' then
  132: *>           LDB must be at least  max( 1, k ), otherwise  LDB must be at
  133: *>           least  max( 1, n ).
  134: *> \endverbatim
  135: *>
  136: *> \param[in] BETA
  137: *> \verbatim
  138: *>          BETA is DOUBLE PRECISION.
  139: *>           On entry,  BETA  specifies the scalar  beta.  When  BETA  is
  140: *>           supplied as zero then C need not be set on input.
  141: *> \endverbatim
  142: *>
  143: *> \param[in,out] C
  144: *> \verbatim
  145: *>          C is DOUBLE PRECISION array of DIMENSION ( LDC, n ).
  146: *>           Before entry, the leading  m by n  part of the array  C must
  147: *>           contain the matrix  C,  except when  beta  is zero, in which
  148: *>           case C need not be set on entry.
  149: *>           On exit, the array  C  is overwritten by the  m by n  matrix
  150: *>           ( alpha*op( A )*op( B ) + beta*C ).
  151: *> \endverbatim
  152: *>
  153: *> \param[in] LDC
  154: *> \verbatim
  155: *>          LDC is INTEGER
  156: *>           On entry, LDC specifies the first dimension of C as declared
  157: *>           in  the  calling  (sub)  program.   LDC  must  be  at  least
  158: *>           max( 1, m ).
  159: *> \endverbatim
  160: *
  161: *  Authors:
  162: *  ========
  163: *
  164: *> \author Univ. of Tennessee 
  165: *> \author Univ. of California Berkeley 
  166: *> \author Univ. of Colorado Denver 
  167: *> \author NAG Ltd. 
  168: *
  169: *> \date November 2015
  170: *
  171: *> \ingroup double_blas_level3
  172: *
  173: *> \par Further Details:
  174: *  =====================
  175: *>
  176: *> \verbatim
  177: *>
  178: *>  Level 3 Blas routine.
  179: *>
  180: *>  -- Written on 8-February-1989.
  181: *>     Jack Dongarra, Argonne National Laboratory.
  182: *>     Iain Duff, AERE Harwell.
  183: *>     Jeremy Du Croz, Numerical Algorithms Group Ltd.
  184: *>     Sven Hammarling, Numerical Algorithms Group Ltd.
  185: *> \endverbatim
  186: *>
  187: *  =====================================================================
  188:       SUBROUTINE DGEMM(TRANSA,TRANSB,M,N,K,ALPHA,A,LDA,B,LDB,BETA,C,LDC)
  189: *
  190: *  -- Reference BLAS level3 routine (version 3.6.0) --
  191: *  -- Reference BLAS is a software package provided by Univ. of Tennessee,    --
  192: *  -- Univ. of California Berkeley, Univ. of Colorado Denver and NAG Ltd..--
  193: *     November 2015
  194: *
  195: *     .. Scalar Arguments ..
  196:       DOUBLE PRECISION ALPHA,BETA
  197:       INTEGER K,LDA,LDB,LDC,M,N
  198:       CHARACTER TRANSA,TRANSB
  199: *     ..
  200: *     .. Array Arguments ..
  201:       DOUBLE PRECISION A(LDA,*),B(LDB,*),C(LDC,*)
  202: *     ..
  203: *
  204: *  =====================================================================
  205: *
  206: *     .. External Functions ..
  207:       LOGICAL LSAME
  208:       EXTERNAL LSAME
  209: *     ..
  210: *     .. External Subroutines ..
  211:       EXTERNAL XERBLA
  212: *     ..
  213: *     .. Intrinsic Functions ..
  214:       INTRINSIC MAX
  215: *     ..
  216: *     .. Local Scalars ..
  217:       DOUBLE PRECISION TEMP
  218:       INTEGER I,INFO,J,L,NCOLA,NROWA,NROWB
  219:       LOGICAL NOTA,NOTB
  220: *     ..
  221: *     .. Parameters ..
  222:       DOUBLE PRECISION ONE,ZERO
  223:       PARAMETER (ONE=1.0D+0,ZERO=0.0D+0)
  224: *     ..
  225: *
  226: *     Set  NOTA  and  NOTB  as  true if  A  and  B  respectively are not
  227: *     transposed and set  NROWA, NCOLA and  NROWB  as the number of rows
  228: *     and  columns of  A  and the  number of  rows  of  B  respectively.
  229: *
  230:       NOTA = LSAME(TRANSA,'N')
  231:       NOTB = LSAME(TRANSB,'N')
  232:       IF (NOTA) THEN
  233:           NROWA = M
  234:           NCOLA = K
  235:       ELSE
  236:           NROWA = K
  237:           NCOLA = M
  238:       END IF
  239:       IF (NOTB) THEN
  240:           NROWB = K
  241:       ELSE
  242:           NROWB = N
  243:       END IF
  244: *
  245: *     Test the input parameters.
  246: *
  247:       INFO = 0
  248:       IF ((.NOT.NOTA) .AND. (.NOT.LSAME(TRANSA,'C')) .AND.
  249:      +    (.NOT.LSAME(TRANSA,'T'))) THEN
  250:           INFO = 1
  251:       ELSE IF ((.NOT.NOTB) .AND. (.NOT.LSAME(TRANSB,'C')) .AND.
  252:      +         (.NOT.LSAME(TRANSB,'T'))) THEN
  253:           INFO = 2
  254:       ELSE IF (M.LT.0) THEN
  255:           INFO = 3
  256:       ELSE IF (N.LT.0) THEN
  257:           INFO = 4
  258:       ELSE IF (K.LT.0) THEN
  259:           INFO = 5
  260:       ELSE IF (LDA.LT.MAX(1,NROWA)) THEN
  261:           INFO = 8
  262:       ELSE IF (LDB.LT.MAX(1,NROWB)) THEN
  263:           INFO = 10
  264:       ELSE IF (LDC.LT.MAX(1,M)) THEN
  265:           INFO = 13
  266:       END IF
  267:       IF (INFO.NE.0) THEN
  268:           CALL XERBLA('DGEMM ',INFO)
  269:           RETURN
  270:       END IF
  271: *
  272: *     Quick return if possible.
  273: *
  274:       IF ((M.EQ.0) .OR. (N.EQ.0) .OR.
  275:      +    (((ALPHA.EQ.ZERO).OR. (K.EQ.0)).AND. (BETA.EQ.ONE))) RETURN
  276: *
  277: *     And if  alpha.eq.zero.
  278: *
  279:       IF (ALPHA.EQ.ZERO) THEN
  280:           IF (BETA.EQ.ZERO) THEN
  281:               DO 20 J = 1,N
  282:                   DO 10 I = 1,M
  283:                       C(I,J) = ZERO
  284:    10             CONTINUE
  285:    20         CONTINUE
  286:           ELSE
  287:               DO 40 J = 1,N
  288:                   DO 30 I = 1,M
  289:                       C(I,J) = BETA*C(I,J)
  290:    30             CONTINUE
  291:    40         CONTINUE
  292:           END IF
  293:           RETURN
  294:       END IF
  295: *
  296: *     Start the operations.
  297: *
  298:       IF (NOTB) THEN
  299:           IF (NOTA) THEN
  300: *
  301: *           Form  C := alpha*A*B + beta*C.
  302: *
  303:               DO 90 J = 1,N
  304:                   IF (BETA.EQ.ZERO) THEN
  305:                       DO 50 I = 1,M
  306:                           C(I,J) = ZERO
  307:    50                 CONTINUE
  308:                   ELSE IF (BETA.NE.ONE) THEN
  309:                       DO 60 I = 1,M
  310:                           C(I,J) = BETA*C(I,J)
  311:    60                 CONTINUE
  312:                   END IF
  313:                   DO 80 L = 1,K
  314:                       TEMP = ALPHA*B(L,J)
  315:                       DO 70 I = 1,M
  316:                           C(I,J) = C(I,J) + TEMP*A(I,L)
  317:    70                 CONTINUE
  318:    80             CONTINUE
  319:    90         CONTINUE
  320:           ELSE
  321: *
  322: *           Form  C := alpha*A**T*B + beta*C
  323: *
  324:               DO 120 J = 1,N
  325:                   DO 110 I = 1,M
  326:                       TEMP = ZERO
  327:                       DO 100 L = 1,K
  328:                           TEMP = TEMP + A(L,I)*B(L,J)
  329:   100                 CONTINUE
  330:                       IF (BETA.EQ.ZERO) THEN
  331:                           C(I,J) = ALPHA*TEMP
  332:                       ELSE
  333:                           C(I,J) = ALPHA*TEMP + BETA*C(I,J)
  334:                       END IF
  335:   110             CONTINUE
  336:   120         CONTINUE
  337:           END IF
  338:       ELSE
  339:           IF (NOTA) THEN
  340: *
  341: *           Form  C := alpha*A*B**T + beta*C
  342: *
  343:               DO 170 J = 1,N
  344:                   IF (BETA.EQ.ZERO) THEN
  345:                       DO 130 I = 1,M
  346:                           C(I,J) = ZERO
  347:   130                 CONTINUE
  348:                   ELSE IF (BETA.NE.ONE) THEN
  349:                       DO 140 I = 1,M
  350:                           C(I,J) = BETA*C(I,J)
  351:   140                 CONTINUE
  352:                   END IF
  353:                   DO 160 L = 1,K
  354:                       TEMP = ALPHA*B(J,L)
  355:                       DO 150 I = 1,M
  356:                           C(I,J) = C(I,J) + TEMP*A(I,L)
  357:   150                 CONTINUE
  358:   160             CONTINUE
  359:   170         CONTINUE
  360:           ELSE
  361: *
  362: *           Form  C := alpha*A**T*B**T + beta*C
  363: *
  364:               DO 200 J = 1,N
  365:                   DO 190 I = 1,M
  366:                       TEMP = ZERO
  367:                       DO 180 L = 1,K
  368:                           TEMP = TEMP + A(L,I)*B(J,L)
  369:   180                 CONTINUE
  370:                       IF (BETA.EQ.ZERO) THEN
  371:                           C(I,J) = ALPHA*TEMP
  372:                       ELSE
  373:                           C(I,J) = ALPHA*TEMP + BETA*C(I,J)
  374:                       END IF
  375:   190             CONTINUE
  376:   200         CONTINUE
  377:           END IF
  378:       END IF
  379: *
  380:       RETURN
  381: *
  382: *     End of DGEMM .
  383: *
  384:       END

CVSweb interface <joel.bertrand@systella.fr>