/* --------------------------------------------------------------------- * * -- PBLAS routine (version 2.0) -- * University of Tennessee, Knoxville, Oak Ridge National Laboratory, * and University of California, Berkeley. * April 1, 1998 * * --------------------------------------------------------------------- */ /* * Include files */ #include "pblas.h" #include "PBpblas.h" #include "PBtools.h" #include "PBblacs.h" #include "PBblas.h" #ifdef __STDC__ void pdsymm_( F_CHAR_T SIDE, F_CHAR_T UPLO, int * M, int * N, double * ALPHA, double * A, int * IA, int * JA, int * DESCA, double * B, int * IB, int * JB, int * DESCB, double * BETA, double * C, int * IC, int * JC, int * DESCC ) #else void pdsymm_( SIDE, UPLO, M, N, ALPHA, A, IA, JA, DESCA, B, IB, JB, DESCB, BETA, C, IC, JC, DESCC ) /* * .. Scalar Arguments .. */ F_CHAR_T SIDE, UPLO; int * IA, * IB, * IC, * JA, * JB, * JC, * M, * N; double * ALPHA, * BETA; /* * .. Array Arguments .. */ int * DESCA, * DESCB, * DESCC; double * A, * B, * C; #endif { /* * Purpose * ======= * * PDSYMM performs one of the matrix-matrix operations * * sub( C ) := alpha*sub( A )*sub( B ) + beta*sub( C ), * * or * * sub( C ) := alpha*sub( B )*sub( A ) + beta*sub( C ), * * where * * sub( C ) denotes C(IC:IC+M-1,JC:JC+N-1), * * sub( A ) denotes A(IA:IA+M-1,JA:JA+M-1) if SIDE = 'L', * A(IA:IA+N-1,JA:JA+N-1) if SIDE = 'R', and, * * sub( B ) denotes B(IB:IB+M-1,JB:JB+N-1). * * Alpha and beta are scalars, sub( A ) is a symmetric submatrix and * sub( B ) and sub( C ) are m by n submatrices. * * Notes * ===== * * A description vector is associated with each 2D block-cyclicly dis- * tributed matrix. This vector stores the information required to * establish the mapping between a matrix entry and its corresponding * process and memory location. * * In the following comments, the character _ should be read as * "of the distributed matrix". Let A be a generic term for any 2D * block cyclicly distributed matrix. Its description vector is DESC_A: * * NOTATION STORED IN EXPLANATION * ---------------- --------------- ------------------------------------ * DTYPE_A (global) DESCA[ DTYPE_ ] The descriptor type. * CTXT_A (global) DESCA[ CTXT_ ] The BLACS context handle, indicating * the NPROW x NPCOL BLACS process grid * A is distributed over. The context * itself is global, but the handle * (the integer value) may vary. * M_A (global) DESCA[ M_ ] The number of rows in the distribu- * ted matrix A, M_A >= 0. * N_A (global) DESCA[ N_ ] The number of columns in the distri- * buted matrix A, N_A >= 0. * IMB_A (global) DESCA[ IMB_ ] The number of rows of the upper left * block of the matrix A, IMB_A > 0. * INB_A (global) DESCA[ INB_ ] The number of columns of the upper * left block of the matrix A, * INB_A > 0. * MB_A (global) DESCA[ MB_ ] The blocking factor used to distri- * bute the last M_A-IMB_A rows of A, * MB_A > 0. * NB_A (global) DESCA[ NB_ ] The blocking factor used to distri- * bute the last N_A-INB_A columns of * A, NB_A > 0. * RSRC_A (global) DESCA[ RSRC_ ] The process row over which the first * row of the matrix A is distributed, * NPROW > RSRC_A >= 0. * CSRC_A (global) DESCA[ CSRC_ ] The process column over which the * first column of A is distributed. * NPCOL > CSRC_A >= 0. * LLD_A (local) DESCA[ LLD_ ] The leading dimension of the local * array storing the local blocks of * the distributed matrix A, * IF( Lc( 1, N_A ) > 0 ) * LLD_A >= MAX( 1, Lr( 1, M_A ) ) * ELSE * LLD_A >= 1. * * Let K be the number of rows of a matrix A starting at the global in- * dex IA,i.e, A( IA:IA+K-1, : ). Lr( IA, K ) denotes the number of rows * that the process of row coordinate MYROW ( 0 <= MYROW < NPROW ) would * receive if these K rows were distributed over NPROW processes. If K * is the number of columns of a matrix A starting at the global index * JA, i.e, A( :, JA:JA+K-1, : ), Lc( JA, K ) denotes the number of co- * lumns that the process MYCOL ( 0 <= MYCOL < NPCOL ) would receive if * these K columns were distributed over NPCOL processes. * * The values of Lr() and Lc() may be determined via a call to the func- * tion PB_Cnumroc: * Lr( IA, K ) = PB_Cnumroc( K, IA, IMB_A, MB_A, MYROW, RSRC_A, NPROW ) * Lc( JA, K ) = PB_Cnumroc( K, JA, INB_A, NB_A, MYCOL, CSRC_A, NPCOL ) * * Arguments * ========= * * SIDE (global input) CHARACTER*1 * On entry, SIDE specifies whether the symmetric submatrix * sub( A ) appears on the left or right in the operation as * follows: * * SIDE = 'L' or 'l' * sub( C ) := alpha*sub( A )*sub( B ) + beta*sub( C ), * * SIDE = 'R' or 'r' * sub( C ) := alpha*sub( B )*sub( A ) + beta*sub( C ). * * UPLO (global input) CHARACTER*1 * On entry, UPLO specifies whether the local pieces of * the array A containing the upper or lower triangular part * of the symmetric submatrix sub( A ) are to be referenced as * follows: * * UPLO = 'U' or 'u' Only the local pieces corresponding to * the upper triangular part of the * symmetric submatrix sub( A ) are to be * referenced, * * UPLO = 'L' or 'l' Only the local pieces corresponding to * the lower triangular part of the * symmetric submatrix sub( A ) are to be * referenced. * * M (global input) INTEGER * On entry, M specifies the number of rows of the submatrix * sub( C ). M must be at least zero. * * N (global input) INTEGER * On entry, N specifies the number of columns of the submatrix * sub( C ). N must be at least zero. * * ALPHA (global input) DOUBLE PRECISION * On entry, ALPHA specifies the scalar alpha. When ALPHA is * supplied as zero then the local entries of the arrays A and * B corresponding to the entries of the submatrices sub( A ) * and sub( B ) respectively need not be set on input. * * A (local input) DOUBLE PRECISION array * On entry, A is an array of dimension (LLD_A, Ka), where Ka is * at least Lc( 1, JA+M-1 ) when SIDE = 'L' or 'l' and is at * at least Lc( 1, JA+N-1 ) otherwise. Before entry, this array * contains the local entries of the matrix A. * Before entry with SIDE = 'L' or 'l', this array contains * the local entries corresponding to the entries of the m by m * symmetric submatrix sub( A ), such that when UPLO = 'U' or * 'u', this array contains the local entries of the upper tri- * angular part of the symmetric submatrix sub( A ), and the * local entries of the strictly lower triangular of sub( A ) * are not referenced, and when UPLO = 'L' or 'l', this array * contains the local entries of the lower triangular part of * the symmetric submatrix sub( A ), and the local entries of * the strictly upper triangular of sub( A ) are not referenced. * Before entry with SIDE = 'R' or 'r', this array contains * the local entries corresponding to the entries of the n by n * symmetric submatrix sub( A ), such that when UPLO = 'U' or * 'u', this array contains the local entries of the upper tri- * angular part of the symmetric submatrix sub( A ), and the * local entries of the strictly lower triangular of sub( A ) * are not referenced, and when UPLO = 'L' or 'l', this array * contains the local entries of the lower triangular part of * the symmetric submatrix sub( A ), and the local entries of * the strictly upper triangular of sub( A ) are not referenced. * * IA (global input) INTEGER * On entry, IA specifies A's global row index, which points to * the beginning of the submatrix sub( A ). * * JA (global input) INTEGER * On entry, JA specifies A's global column index, which points * to the beginning of the submatrix sub( A ). * * DESCA (global and local input) INTEGER array * On entry, DESCA is an integer array of dimension DLEN_. This * is the array descriptor for the matrix A. * * B (local input) DOUBLE PRECISION array * On entry, B is an array of dimension (LLD_B, Kb), where Kb is * at least Lc( 1, JB+N-1 ). Before entry, this array contains * the local entries of the matrix B. * * IB (global input) INTEGER * On entry, IB specifies B's global row index, which points to * the beginning of the submatrix sub( B ). * * JB (global input) INTEGER * On entry, JB specifies B's global column index, which points * to the beginning of the submatrix sub( B ). * * DESCB (global and local input) INTEGER array * On entry, DESCB is an integer array of dimension DLEN_. This * is the array descriptor for the matrix B. * * BETA (global input) DOUBLE PRECISION * On entry, BETA specifies the scalar beta. When BETA is * supplied as zero then the local entries of the array C * corresponding to the entries of the submatrix sub( C ) need * not be set on input. * * C (local input/local output) DOUBLE PRECISION array * On entry, C is an array of dimension (LLD_C, Kc), where Kc is * at least Lc( 1, JC+N-1 ). Before entry, this array contains * the local entries of the matrix C. * On exit, the entries of this array corresponding to the local * entries of the submatrix sub( C ) are overwritten by the * local entries of the m by n updated submatrix. * * IC (global input) INTEGER * On entry, IC specifies C's global row index, which points to * the beginning of the submatrix sub( C ). * * JC (global input) INTEGER * On entry, JC specifies C's global column index, which points * to the beginning of the submatrix sub( C ). * * DESCC (global and local input) INTEGER array * On entry, DESCC is an integer array of dimension DLEN_. This * is the array descriptor for the matrix C. * * -- Written on April 1, 1998 by * Antoine Petitet, University of Tennessee, Knoxville 37996, USA. * * --------------------------------------------------------------------- */ /* * .. Local Scalars .. */ char DirAB, SideOp, UploA, cbtop, cbtopsave, cctop, cctopsave, rbtop, rbtopsave, rctop, rctopsave; int Ai, Aj, Bi, Bj, ChooseABC, Ci, Cj, ForceTop, ctxt, info, lside, mycol, myrow, nb, npcol, nprow, upper; double ABCest, BCest, tmp1, tmp2, tmp3, tmp4; PBTYP_T * type; /* * .. Local Arrays .. */ int Ad[DLEN_], Bd[DLEN_], Cd[DLEN_]; /* .. * .. Executable Statements .. * */ lside = ( ( SideOp = Mupcase( F2C_CHAR( SIDE )[0] ) ) == CLEFT ); upper = ( ( UploA = Mupcase( F2C_CHAR( UPLO )[0] ) ) == CUPPER ); PB_CargFtoC( *IA, *JA, DESCA, &Ai, &Aj, Ad ); PB_CargFtoC( *IB, *JB, DESCB, &Bi, &Bj, Bd ); PB_CargFtoC( *IC, *JC, DESCC, &Ci, &Cj, Cd ); #ifndef NO_ARGCHK /* * Test the input parameters */ Cblacs_gridinfo( ( ctxt = Ad[CTXT_] ), &nprow, &npcol, &myrow, &mycol ); if( !( info = ( ( nprow == -1 ) ? -( 901 + CTXT_ ) : 0 ) ) ) { if( ( !lside ) && ( SideOp != CRIGHT ) ) { PB_Cwarn( ctxt, __LINE__, "PDSYMM", "Illegal SIDE = %c\n", SideOp ); info = -1; } else if( ( !upper ) && ( UploA != CLOWER ) ) { PB_Cwarn( ctxt, __LINE__, "PDSYMM", "Illegal UPLO = %c\n", UploA ); info = -2; } if( lside ) { PB_Cchkmat( ctxt, "PDSYMM", "A", *M, 3, *M, 3, Ai, Aj, Ad, 9, &info ); PB_Cchkmat( ctxt, "PDSYMM", "B", *M, 3, *N, 4, Bi, Bj, Bd, 13, &info ); } else { PB_Cchkmat( ctxt, "PDSYMM", "A", *N, 4, *N, 4, Ai, Aj, Ad, 9, &info ); PB_Cchkmat( ctxt, "PDSYMM", "B", *M, 3, *N, 4, Bi, Bj, Bd, 13, &info ); } PB_Cchkmat( ctxt, "PDSYMM", "C", *M, 3, *N, 4, Ci, Cj, Cd, 18, &info ); } if( info ) { PB_Cabort( ctxt, "PDSYMM", info ); return; } #endif /* * Quick return if possible */ if( ( *M == 0 ) || ( *N == 0 ) || ( ( ALPHA[REAL_PART] == ZERO ) && ( BETA[REAL_PART] == ONE ) ) ) return; /* * Get type structure */ type = PB_Cdtypeset(); /* * If alpha is zero, sub( C ) := beta * sub( C ). */ if( ALPHA[REAL_PART] == ZERO ) { if( BETA[REAL_PART] == ZERO ) { PB_Cplapad( type, ALL, NOCONJG, *M, *N, type->zero, type->zero, ((char * ) C), Ci, Cj, Cd ); } else if( !( BETA[REAL_PART] == ONE ) ) { PB_Cplascal( type, ALL, NOCONJG, *M, *N, ((char *) BETA), ((char *) C), Ci, Cj, Cd ); } return; } /* * Start the operations */ #ifdef NO_ARGCHK Cblacs_gridinfo( ( ctxt = Ad[CTXT_] ), &nprow, &npcol, &myrow, &mycol ); #endif /* * Algorithm selection is based on approximation of the communication volume * for distributed and aligned operands. * * ABCest: operands sub( A ), sub( B ) and sub( C ) are communicated (N >> M) * BCest : Both operands sub( B ) and sub( C ) are communicated (M >> N) */ if( lside ) { tmp1 = DNROC( *M, Ad[MB_], nprow ); tmp2 = DNROC( *N, Bd[NB_], npcol ); ABCest = (double)(*M) * ( ( ( ( Ad[CSRC_] == -1 ) || ( npcol == 1 ) ) ? ZERO : tmp1 / TWO ) + ( ( ( Bd[RSRC_] == -1 ) || ( nprow == 1 ) ) ? ZERO : tmp2 + tmp2 * CBRATIO ) ); tmp1 = DNROC( *M, Ad[MB_], nprow ); tmp2 = DNROC( *M, Ad[NB_], npcol ); tmp3 = DNROC( *M, Bd[MB_], nprow ); tmp4 = DNROC( *M, Cd[MB_], nprow ); BCest = (double)(*N) * ( CBRATIO * ( npcol == 1 ? ZERO : tmp1 ) + ( nprow == 1 ? ZERO : tmp2 ) + MAX( tmp2, tmp3 ) + ( ( ( Bd[CSRC_] == -1 ) || ( npcol == 1 ) ) ? ZERO : tmp1 ) + CBRATIO * ( nprow == 1 ? ZERO : tmp2 ) + MAX( tmp2, tmp4 ) ); } else { tmp1 = DNROC( *N, Ad[NB_], npcol ); tmp2 = DNROC( *M, Bd[MB_], nprow ); ABCest = (double)(*N) * ( ( ( ( Ad[RSRC_] == -1 ) || ( nprow == 1 ) ) ? ZERO : tmp1 / TWO ) + ( ( ( Bd[CSRC_] == -1 ) || ( npcol == 1 ) ) ? ZERO : tmp2 + tmp2 * CBRATIO ) ); tmp1 = DNROC( *N, Ad[MB_], nprow ); tmp2 = DNROC( *N, Ad[NB_], npcol ); tmp3 = DNROC( *N, Bd[NB_], npcol ); tmp4 = DNROC( *N, Cd[NB_], npcol ); BCest = (double)(*M) * ( ( npcol == 1 ? ZERO : tmp1 ) + MAX( tmp1, tmp3 ) + CBRATIO * ( nprow == 1 ? ZERO : tmp2 ) + ( ( ( Bd[RSRC_] == -1 ) || ( nprow == 1 ) ) ? ZERO : tmp2 ) + CBRATIO * ( npcol == 1 ? ZERO : tmp1 ) + MAX( tmp1, tmp4 ) ); } /* * Shift a little the cross-over point between both algorithms. */ ChooseABC = ( ( 1.5 * ABCest ) <= BCest ); /* * BLACS topologies are enforced iff M and N are strictly greater than the * logical block size returned by pilaenv_. Otherwise, it is assumed that the * routine calling this routine has already selected an adequate topology. */ nb = pilaenv_( &ctxt, C2F_CHAR( &type->type ) ); ForceTop = ( ( *M > nb ) && ( *N > nb ) ); rbtop = *PB_Ctop( &ctxt, BCAST, ROW, TOP_GET ); rctop = *PB_Ctop( &ctxt, COMBINE, ROW, TOP_GET ); cbtop = *PB_Ctop( &ctxt, BCAST, COLUMN, TOP_GET ); cctop = *PB_Ctop( &ctxt, COMBINE, COLUMN, TOP_GET ); if( ChooseABC ) { if( ForceTop ) { rbtopsave = rbtop; rctopsave = rctop; cbtopsave = cbtop; cctopsave = cctop; if( lside ) { /* * No clear winner for the ring topologies, so that if a ring topology is * already selected, keep it. */ if( ( rbtop != CTOP_DRING ) && ( rbtop != CTOP_IRING ) && ( rbtop != CTOP_SRING ) ) rbtop = *PB_Ctop( &ctxt, BCAST, ROW, TOP_IRING ); if( ( ( cbtop != CTOP_DRING ) && ( cbtop != CTOP_IRING ) && ( cbtop != CTOP_SRING ) ) || ( cbtop != cctop ) ) { cbtop = *PB_Ctop( &ctxt, BCAST, COLUMN, TOP_IRING ); cctop = *PB_Ctop( &ctxt, COMBINE, COLUMN, TOP_IRING ); /* * Remove the next 2 lines when the BLACS combine operations support ring * topologies */ rctop = *PB_Ctop( &ctxt, COMBINE, ROW, TOP_DEFAULT ); cctop = *PB_Ctop( &ctxt, COMBINE, COLUMN, TOP_DEFAULT ); } } else { /* * No clear winner for the ring topologies, so that if a ring topology is * already selected, keep it. */ if( ( cbtop != CTOP_DRING ) && ( cbtop != CTOP_IRING ) && ( cbtop != CTOP_SRING ) ) cbtop = *PB_Ctop( &ctxt, BCAST, COLUMN, TOP_IRING ); if( ( ( rbtop != CTOP_DRING ) && ( rbtop != CTOP_IRING ) && ( rbtop != CTOP_SRING ) ) || ( rbtop != rctop ) ) { rbtop = *PB_Ctop( &ctxt, BCAST, ROW, TOP_IRING ); rctop = *PB_Ctop( &ctxt, COMBINE, ROW, TOP_IRING ); /* * Remove the next 2 lines when the BLACS combine operations support ring * topologies */ rctop = *PB_Ctop( &ctxt, COMBINE, ROW, TOP_DEFAULT ); cctop = *PB_Ctop( &ctxt, COMBINE, COLUMN, TOP_DEFAULT ); } } } if( lside ) DirAB = ( rbtop == CTOP_DRING ? CBACKWARD : CFORWARD ); else DirAB = ( cbtop == CTOP_DRING ? CBACKWARD : CFORWARD ); PB_CpsymmAB( type, &DirAB, NOCONJG, &SideOp, &UploA, *M, *N, ((char *)ALPHA), ((char *)A), Ai, Aj, Ad, ((char *)B), Bi, Bj, Bd, ((char *)BETA), ((char *)C), Ci, Cj, Cd ); } else { if( ForceTop ) { rbtopsave = rbtop; rctopsave = rctop; cbtopsave = cbtop; cctopsave = cctop; if( lside ) { /* * No clear winner for the ring topologies, so that if a ring topology is * already selected, keep it. */ if( ( ( rbtop != CTOP_DRING ) && ( rbtop != CTOP_IRING ) && ( rbtop != CTOP_SRING ) ) || ( rbtop != rctop ) ) { rbtop = *PB_Ctop( &ctxt, BCAST, ROW, TOP_IRING ); rctop = *PB_Ctop( &ctxt, COMBINE, ROW, TOP_IRING ); /* * Remove the next 2 lines when the BLACS combine operations support ring * topologies */ rctop = *PB_Ctop( &ctxt, COMBINE, ROW, TOP_DEFAULT ); cctop = *PB_Ctop( &ctxt, COMBINE, COLUMN, TOP_DEFAULT ); } cbtop = *PB_Ctop( &ctxt, BCAST, COLUMN, TOP_DEFAULT ); cctop = *PB_Ctop( &ctxt, COMBINE, COLUMN, TOP_DEFAULT ); } else { /* * No clear winner for the ring topologies, so that if a ring topology is * already selected, keep it. */ if( ( ( cbtop != CTOP_DRING ) && ( cbtop != CTOP_IRING ) && ( cbtop != CTOP_SRING ) ) || ( cbtop != cctop ) ) { cbtop = *PB_Ctop( &ctxt, BCAST, COLUMN, TOP_IRING ); cctop = *PB_Ctop( &ctxt, COMBINE, COLUMN, TOP_IRING ); /* * Remove the next 2 lines when the BLACS combine operations support ring * topologies */ rctop = *PB_Ctop( &ctxt, COMBINE, ROW, TOP_DEFAULT ); cctop = *PB_Ctop( &ctxt, COMBINE, COLUMN, TOP_DEFAULT ); } rbtop = *PB_Ctop( &ctxt, BCAST, ROW, TOP_DEFAULT ); rctop = *PB_Ctop( &ctxt, COMBINE, ROW, TOP_DEFAULT ); } } if( lside ) DirAB = ( ( rbtop == CTOP_DRING || rctop == CTOP_DRING ) ? CBACKWARD : CFORWARD ); else DirAB = ( ( cbtop == CTOP_DRING || cctop == CTOP_DRING ) ? CBACKWARD : CFORWARD ); PB_CpsymmBC( type, &DirAB, NOCONJG, &SideOp, &UploA, *M, *N, ((char *)ALPHA), ((char *)A), Ai, Aj, Ad, ((char *)B), Bi, Bj, Bd, ((char *)BETA), ((char *)C), Ci, Cj, Cd ); } /* * Restore the BLACS topologies when necessary. */ if( ForceTop ) { rbtopsave = *PB_Ctop( &ctxt, BCAST, ROW, &rbtopsave ); rctopsave = *PB_Ctop( &ctxt, COMBINE, ROW, &rctopsave ); cbtopsave = *PB_Ctop( &ctxt, BCAST, COLUMN, &cbtopsave ); cctopsave = *PB_Ctop( &ctxt, COMBINE, COLUMN, &cctopsave ); } /* * End of PDSYMM */ }