1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
|
/*! \file
Copyright (c) 2003, The Regents of the University of California, through
Lawrence Berkeley National Laboratory (subject to receipt of any required
approvals from U.S. Dept. of Energy)
All rights reserved.
The source code is distributed under BSD license, see the file License.txt
at the top-level directory.
*/
/*! @file cmyblas2.c
* \brief Level 2 Blas operations
*
* <pre>
* -- SuperLU routine (version 2.0) --
* Univ. of California Berkeley, Xerox Palo Alto Research Center,
* and Lawrence Berkeley National Lab.
* November 15, 1997
* </pre>
* Purpose:
* Level 2 BLAS operations: solves and matvec, written in C.
* Note:
* This is only used when the system lacks an efficient BLAS library.
* </pre>
*/
/*
* File name: cmyblas2.c
*/
#include "slu_scomplex.h"
/*! \brief Solves a dense UNIT lower triangular system
*
* The unit lower
* triangular matrix is stored in a 2D array M(1:nrow,1:ncol).
* The solution will be returned in the rhs vector.
*/
void clsolve ( int ldm, int ncol, complex *M, complex *rhs )
{
int k;
complex x0, x1, x2, x3, temp;
complex *M0;
complex *Mki0, *Mki1, *Mki2, *Mki3;
register int firstcol = 0;
M0 = &M[0];
while ( firstcol < ncol - 3 ) { /* Do 4 columns */
Mki0 = M0 + 1;
Mki1 = Mki0 + ldm + 1;
Mki2 = Mki1 + ldm + 1;
Mki3 = Mki2 + ldm + 1;
x0 = rhs[firstcol];
cc_mult(&temp, &x0, Mki0); Mki0++;
c_sub(&x1, &rhs[firstcol+1], &temp);
cc_mult(&temp, &x0, Mki0); Mki0++;
c_sub(&x2, &rhs[firstcol+2], &temp);
cc_mult(&temp, &x1, Mki1); Mki1++;
c_sub(&x2, &x2, &temp);
cc_mult(&temp, &x0, Mki0); Mki0++;
c_sub(&x3, &rhs[firstcol+3], &temp);
cc_mult(&temp, &x1, Mki1); Mki1++;
c_sub(&x3, &x3, &temp);
cc_mult(&temp, &x2, Mki2); Mki2++;
c_sub(&x3, &x3, &temp);
rhs[++firstcol] = x1;
rhs[++firstcol] = x2;
rhs[++firstcol] = x3;
++firstcol;
for (k = firstcol; k < ncol; k++) {
cc_mult(&temp, &x0, Mki0); Mki0++;
c_sub(&rhs[k], &rhs[k], &temp);
cc_mult(&temp, &x1, Mki1); Mki1++;
c_sub(&rhs[k], &rhs[k], &temp);
cc_mult(&temp, &x2, Mki2); Mki2++;
c_sub(&rhs[k], &rhs[k], &temp);
cc_mult(&temp, &x3, Mki3); Mki3++;
c_sub(&rhs[k], &rhs[k], &temp);
}
M0 += 4 * ldm + 4;
}
if ( firstcol < ncol - 1 ) { /* Do 2 columns */
Mki0 = M0 + 1;
Mki1 = Mki0 + ldm + 1;
x0 = rhs[firstcol];
cc_mult(&temp, &x0, Mki0); Mki0++;
c_sub(&x1, &rhs[firstcol+1], &temp);
rhs[++firstcol] = x1;
++firstcol;
for (k = firstcol; k < ncol; k++) {
cc_mult(&temp, &x0, Mki0); Mki0++;
c_sub(&rhs[k], &rhs[k], &temp);
cc_mult(&temp, &x1, Mki1); Mki1++;
c_sub(&rhs[k], &rhs[k], &temp);
}
}
}
/*! \brief Solves a dense upper triangular system.
*
* The upper triangular matrix is
* stored in a 2-dim array M(1:ldm,1:ncol). The solution will be returned
* in the rhs vector.
*/
void
cusolve ( ldm, ncol, M, rhs )
int ldm; /* in */
int ncol; /* in */
complex *M; /* in */
complex *rhs; /* modified */
{
complex xj, temp;
int jcol, j, irow;
jcol = ncol - 1;
for (j = 0; j < ncol; j++) {
c_div(&xj, &rhs[jcol], &M[jcol + jcol*ldm]); /* M(jcol, jcol) */
rhs[jcol] = xj;
for (irow = 0; irow < jcol; irow++) {
cc_mult(&temp, &xj, &M[irow+jcol*ldm]); /* M(irow, jcol) */
c_sub(&rhs[irow], &rhs[irow], &temp);
}
jcol--;
}
}
/*! \brief Performs a dense matrix-vector multiply: Mxvec = Mxvec + M * vec.
*
* The input matrix is M(1:nrow,1:ncol); The product is returned in Mxvec[].
*/
void cmatvec ( ldm, nrow, ncol, M, vec, Mxvec )
int ldm; /* in -- leading dimension of M */
int nrow; /* in */
int ncol; /* in */
complex *M; /* in */
complex *vec; /* in */
complex *Mxvec; /* in/out */
{
complex vi0, vi1, vi2, vi3;
complex *M0, temp;
complex *Mki0, *Mki1, *Mki2, *Mki3;
register int firstcol = 0;
int k;
M0 = &M[0];
while ( firstcol < ncol - 3 ) { /* Do 4 columns */
Mki0 = M0;
Mki1 = Mki0 + ldm;
Mki2 = Mki1 + ldm;
Mki3 = Mki2 + ldm;
vi0 = vec[firstcol++];
vi1 = vec[firstcol++];
vi2 = vec[firstcol++];
vi3 = vec[firstcol++];
for (k = 0; k < nrow; k++) {
cc_mult(&temp, &vi0, Mki0); Mki0++;
c_add(&Mxvec[k], &Mxvec[k], &temp);
cc_mult(&temp, &vi1, Mki1); Mki1++;
c_add(&Mxvec[k], &Mxvec[k], &temp);
cc_mult(&temp, &vi2, Mki2); Mki2++;
c_add(&Mxvec[k], &Mxvec[k], &temp);
cc_mult(&temp, &vi3, Mki3); Mki3++;
c_add(&Mxvec[k], &Mxvec[k], &temp);
}
M0 += 4 * ldm;
}
while ( firstcol < ncol ) { /* Do 1 column */
Mki0 = M0;
vi0 = vec[firstcol++];
for (k = 0; k < nrow; k++) {
cc_mult(&temp, &vi0, Mki0); Mki0++;
c_add(&Mxvec[k], &Mxvec[k], &temp);
}
M0 += ldm;
}
}
|
