|
Packit |
67cb25 |
/* blas/source_spmv.h
|
|
Packit |
67cb25 |
*
|
|
Packit |
67cb25 |
* Copyright (C) 1996, 1997, 1998, 1999, 2000 Gerard Jungman
|
|
Packit |
67cb25 |
*
|
|
Packit |
67cb25 |
* This program is free software; you can redistribute it and/or modify
|
|
Packit |
67cb25 |
* it under the terms of the GNU General Public License as published by
|
|
Packit |
67cb25 |
* the Free Software Foundation; either version 3 of the License, or (at
|
|
Packit |
67cb25 |
* your option) any later version.
|
|
Packit |
67cb25 |
*
|
|
Packit |
67cb25 |
* This program is distributed in the hope that it will be useful, but
|
|
Packit |
67cb25 |
* WITHOUT ANY WARRANTY; without even the implied warranty of
|
|
Packit |
67cb25 |
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
|
|
Packit |
67cb25 |
* General Public License for more details.
|
|
Packit |
67cb25 |
*
|
|
Packit |
67cb25 |
* You should have received a copy of the GNU General Public License
|
|
Packit |
67cb25 |
* along with this program; if not, write to the Free Software
|
|
Packit |
67cb25 |
* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
|
|
Packit |
67cb25 |
*/
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
{
|
|
Packit |
67cb25 |
INDEX i, j;
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
CHECK_ARGS10(SD_SPMV,order,Uplo,N,alpha,Ap,X,incX,beta,Y,incY);
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
if (alpha == 0.0 && beta == 1.0)
|
|
Packit |
67cb25 |
return;
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
/* form y := beta*y */
|
|
Packit |
67cb25 |
if (beta == 0.0) {
|
|
Packit |
67cb25 |
INDEX iy = OFFSET(N, incY);
|
|
Packit |
67cb25 |
for (i = 0; i < N; i++) {
|
|
Packit |
67cb25 |
Y[iy] = 0.0;
|
|
Packit |
67cb25 |
iy += incY;
|
|
Packit |
67cb25 |
}
|
|
Packit |
67cb25 |
} else if (beta != 1.0) {
|
|
Packit |
67cb25 |
INDEX iy = OFFSET(N, incY);
|
|
Packit |
67cb25 |
for (i = 0; i < N; i++) {
|
|
Packit |
67cb25 |
Y[iy] *= beta;
|
|
Packit |
67cb25 |
iy += incY;
|
|
Packit |
67cb25 |
}
|
|
Packit |
67cb25 |
}
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
if (alpha == 0.0)
|
|
Packit |
67cb25 |
return;
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
/* form y := alpha*A*x + y */
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
if ((order == CblasRowMajor && Uplo == CblasUpper)
|
|
Packit |
67cb25 |
|| (order == CblasColMajor && Uplo == CblasLower)) {
|
|
Packit |
67cb25 |
INDEX ix = OFFSET(N, incX);
|
|
Packit |
67cb25 |
INDEX iy = OFFSET(N, incY);
|
|
Packit |
67cb25 |
for (i = 0; i < N; i++) {
|
|
Packit |
67cb25 |
BASE tmp1 = alpha * X[ix];
|
|
Packit |
67cb25 |
BASE tmp2 = 0.0;
|
|
Packit |
67cb25 |
const INDEX j_min = i + 1;
|
|
Packit |
67cb25 |
const INDEX j_max = N;
|
|
Packit |
67cb25 |
INDEX jx = OFFSET(N, incX) + j_min * incX;
|
|
Packit |
67cb25 |
INDEX jy = OFFSET(N, incY) + j_min * incY;
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
Y[iy] += tmp1 * Ap[TPUP(N, i, i)];
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
for (j = j_min; j < j_max; j++) {
|
|
Packit |
67cb25 |
const BASE apk = Ap[TPUP(N, i, j)];
|
|
Packit |
67cb25 |
Y[jy] += tmp1 * apk;
|
|
Packit |
67cb25 |
tmp2 += apk * X[jx];
|
|
Packit |
67cb25 |
jy += incY;
|
|
Packit |
67cb25 |
jx += incX;
|
|
Packit |
67cb25 |
}
|
|
Packit |
67cb25 |
Y[iy] += alpha * tmp2;
|
|
Packit |
67cb25 |
ix += incX;
|
|
Packit |
67cb25 |
iy += incY;
|
|
Packit |
67cb25 |
}
|
|
Packit |
67cb25 |
} else if ((order == CblasRowMajor && Uplo == CblasLower)
|
|
Packit |
67cb25 |
|| (order == CblasColMajor && Uplo == CblasUpper)) {
|
|
Packit |
67cb25 |
INDEX ix = OFFSET(N, incX);
|
|
Packit |
67cb25 |
INDEX iy = OFFSET(N, incY);
|
|
Packit |
67cb25 |
for (i = 0; i < N; i++) {
|
|
Packit |
67cb25 |
BASE tmp1 = alpha * X[ix];
|
|
Packit |
67cb25 |
BASE tmp2 = 0.0;
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
const INDEX j_min = 0;
|
|
Packit |
67cb25 |
const INDEX j_max = i;
|
|
Packit |
67cb25 |
INDEX jx = OFFSET(N, incX) + j_min * incX;
|
|
Packit |
67cb25 |
INDEX jy = OFFSET(N, incY) + j_min * incY;
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
Y[iy] += tmp1 * Ap[TPLO(N, i, i)];
|
|
Packit |
67cb25 |
|
|
Packit |
67cb25 |
for (j = j_min; j < j_max; j++) {
|
|
Packit |
67cb25 |
const BASE apk = Ap[TPLO(N, i, j)];
|
|
Packit |
67cb25 |
Y[jy] += tmp1 * apk;
|
|
Packit |
67cb25 |
tmp2 += apk * X[jx];
|
|
Packit |
67cb25 |
jy += incY;
|
|
Packit |
67cb25 |
jx += incX;
|
|
Packit |
67cb25 |
}
|
|
Packit |
67cb25 |
Y[iy] += alpha * tmp2;
|
|
Packit |
67cb25 |
ix += incX;
|
|
Packit |
67cb25 |
iy += incY;
|
|
Packit |
67cb25 |
}
|
|
Packit |
67cb25 |
} else {
|
|
Packit |
67cb25 |
BLAS_ERROR("unrecognized operation");
|
|
Packit |
67cb25 |
}
|
|
Packit |
67cb25 |
}
|