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
199
200
201
202
203
204
205
206
207
208
209
|
// Copyright (c) 2017-2023, University of Tennessee. All rights reserved.
// SPDX-License-Identifier: BSD-3-Clause
// This program is free software: you can redistribute it and/or modify it under
// the terms of the BSD 3-Clause license. See the accompanying LICENSE file.
#include "test.hh"
#include "lapack.hh"
#include "lapack/flops.hh"
#include "print_matrix.hh"
#include "error.hh"
#include "lapacke_wrappers.hh"
#include <vector>
// -----------------------------------------------------------------------------
template< typename scalar_t >
void test_gglse_work( Params& params, bool run )
{
using real_t = blas::real_type< scalar_t >;
// get & mark input values
int64_t m = params.dim.m();
int64_t n = params.dim.n();
// TODO int64_t p = params.p();
int64_t p = params.dim.k();
int64_t align = params.align();
params.matrix.mark();
params.matrixB.mark();
int64_t verbose = params.verbose();
real_t eps = std::numeric_limits< real_t >::epsilon();
real_t tol = params.tol() * eps;
// mark non-standard output values
params.error2();
params.ref_time();
// params.ref_gflops();
// params.gflops();
params.msg();
if (! run) {
// Use well-conditioned matrices per LAWN 41.
params.matrix .kind.set_default( "svd" );
params.matrixB.kind.set_default( "svd" );
params.matrix .cond() = 100;
params.matrixB.cond() = 10;
return;
}
// skip invalid sizes
if (! ((0 <= p) && (p <= n) && (n <= m+p))) {
params.msg() = "skipping: requires 0 <= p <= n <= m+p";
return;
}
// ---------- setup
bool consistent = true;
int64_t lda = roundup( blas::max( 1, m ), align );
int64_t ldb = roundup( blas::max( 1, p ), align );
size_t size_A = (size_t) ( lda * n );
size_t size_B = (size_t) ( ldb * n );
size_t size_C = (size_t) (m);
size_t size_D = (size_t) (p);
size_t size_X = (size_t) (n);
std::vector< scalar_t > A_tst( size_A );
std::vector< scalar_t > A_ref( size_A );
std::vector< scalar_t > B_tst( size_B );
std::vector< scalar_t > B_ref( size_B );
std::vector< scalar_t > C_tst( size_C );
std::vector< scalar_t > C_ref( size_C );
std::vector< scalar_t > D_tst( size_D );
std::vector< scalar_t > D_ref( size_D );
std::vector< scalar_t > X_tst( size_X );
std::vector< scalar_t > X_ref( size_X );
lapack::generate_matrix( params.matrix, m, n, &A_tst[0], lda );
lapack::generate_matrix( params.matrixB, p, n, &B_tst[0], ldb );
int64_t idist = 1;
int64_t iseed[4] = { 0, 1, 2, 3 };
if (consistent) {
// Generate random X, then set C = A*X and D = B*X.
lapack::larnv( idist, iseed, X_ref.size(), &X_ref[0] );
blas::gemv( blas::Layout::ColMajor, blas::Op::NoTrans, m, n,
1.0, &A_tst[0], lda, &X_ref[0], 1,
0.0, &C_tst[0], 1 );
blas::gemv( blas::Layout::ColMajor, blas::Op::NoTrans, p, n,
1.0, &B_tst[0], ldb, &X_ref[0], 1,
0.0, &D_tst[0], 1 );
}
else {
// Generate random C, D.
lapack::larnv( idist, iseed, C_tst.size(), &C_tst[0] );
lapack::larnv( idist, iseed, D_tst.size(), &D_tst[0] );
}
A_ref = A_tst;
B_ref = B_tst;
C_ref = C_tst;
D_ref = D_tst;
if (verbose >= 2) {
printf( "A = " ); print_matrix( m, n, &A_tst[0], lda );
printf( "B = " ); print_matrix( p, n, &B_tst[0], ldb );
printf( "c = " ); print_vector( m, &C_tst[0], 1 );
printf( "d = " ); print_vector( p, &D_tst[0], 1 );
}
// ---------- run test
// minimize || c - A*x ||_2 subject to B*x = d
// A is M-by-N matrix, B is P-by-N matrix, c is M-vector, and d is P-vector
testsweeper::flush_cache( params.cache() );
double time = testsweeper::get_wtime();
int64_t info_tst = lapack::gglse( m, n, p, &A_tst[0], lda, &B_tst[0], ldb, &C_tst[0], &D_tst[0], &X_tst[0] );
time = testsweeper::get_wtime() - time;
if (info_tst != 0) {
fprintf( stderr, "lapack::gglse returned error %lld\n", llong( info_tst ) );
}
params.time() = time;
// double gflop = lapack::Gflop< scalar_t >::gglse( m, n );
// params.gflops() = gflop / time;
if (verbose >= 2) {
printf( "x_tst = " ); print_vector( n, &X_tst[0], 1 );
}
if (params.check() == 'y') {
// ---------- check error
real_t x_norm = blas::asum( n, &X_tst[0], 1 );
// r1 = Ax - c is small only for a consistent system.
if (consistent) {
C_tst = C_ref;
blas::gemv( blas::Layout::ColMajor, blas::Op::NoTrans, m, n,
1.0, &A_ref[0], lda, &X_tst[0], 1,
-1.0, &C_tst[0], 1 );
real_t error1 = blas::asum( m, &C_tst[0], 1 );
real_t A_norm = lapack::lange( lapack::Norm::One, m, n, &A_ref[0], lda );
if (A_norm != 0)
error1 /= A_norm;
if (x_norm != 0)
error1 /= x_norm;
params.error() = error1;
params.okay() = (error1 < tol);
}
// r2 = Bx - d should always be small.
D_tst = D_ref;
blas::gemv( blas::Layout::ColMajor, blas::Op::NoTrans, p, n,
1.0, &B_ref[0], ldb, &X_tst[0], 1,
-1.0, &D_tst[0], 1 );
real_t error2 = blas::asum( p, &D_tst[0], 1 );
real_t B_norm = lapack::lange( lapack::Norm::One, p, n, &B_ref[0], ldb );
error2 /= n;
if (B_norm != 0)
error2 /= B_norm;
if (x_norm != 0)
error2 /= x_norm;
params.error2() = error2;
params.okay() = params.okay() && (error2 < tol);
}
if (params.ref() == 'y') {
// ---------- run reference
testsweeper::flush_cache( params.cache() );
time = testsweeper::get_wtime();
int64_t info_ref = LAPACKE_gglse( m, n, p, &A_ref[0], lda, &B_ref[0], ldb, &C_ref[0], &D_ref[0], &X_ref[0] );
time = testsweeper::get_wtime() - time;
if (info_ref != 0) {
fprintf( stderr, "LAPACKE_gglse returned error %lld\n", llong( info_ref ) );
}
if (verbose >= 2) {
printf( "x_ref = " ); print_vector( n, &X_ref[0], 1 );
}
params.ref_time() = time;
// params.ref_gflops() = gflop / time;
}
}
// -----------------------------------------------------------------------------
void test_gglse( Params& params, bool run )
{
switch (params.datatype()) {
case testsweeper::DataType::Single:
test_gglse_work< float >( params, run );
break;
case testsweeper::DataType::Double:
test_gglse_work< double >( params, run );
break;
case testsweeper::DataType::SingleComplex:
test_gglse_work< std::complex<float> >( params, run );
break;
case testsweeper::DataType::DoubleComplex:
test_gglse_work< std::complex<double> >( params, run );
break;
default:
throw std::runtime_error( "unknown datatype" );
break;
}
}
|
