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
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
|
/*
-- MAGMA (version 2.9.0) --
Univ. of Tennessee, Knoxville
Univ. of California, Berkeley
Univ. of Colorado, Denver
@date January 2025
@author Tingxing Dong
@author Azzam Haidar
@generated from testing/testing_zgeqrf_batched.cpp, normal z -> s, Wed Jan 22 14:40:43 2025
*/
// includes, system
#include <stdlib.h>
#include <stdio.h>
#include <string.h>
#include <math.h>
// includes, project
#include "flops.h"
#include "magma_v2.h"
#include "magma_lapack.h"
#include "testings.h"
#if defined(_OPENMP)
#include <omp.h>
#include "../control/magma_threadsetting.h" // internal header
#endif
void get_QR_error(magma_int_t M, magma_int_t N, magma_int_t min_mn,
float *h_R, float *h_A, magma_int_t lda,
float *tau,
float *Q, magma_int_t ldq,
float *R, magma_int_t ldr,
float *h_work, magma_int_t lwork,
float *work, float *error, float *error2)
{
/* h_R:input the factorized matrix by lapack QR,
h_A:input the original matrix copy
tau: input
*/
const float d_neg_one = MAGMA_D_NEG_ONE;
const float d_one = MAGMA_D_ONE;
const float c_neg_one = MAGMA_S_NEG_ONE;
const float c_one = MAGMA_S_ONE;
const float c_zero = MAGMA_S_ZERO;
float Anorm;
magma_int_t info;
// generate M by K matrix Q, where K = min(M,N)
lapackf77_slacpy( "Lower", &M, &min_mn, h_R, &lda, Q, &ldq );
lapackf77_sorgqr( &M, &min_mn, &min_mn, Q, &ldq, tau, h_work, &lwork, &info );
assert( info == 0 );
// copy K by N matrix R
lapackf77_slaset( "Lower", &min_mn, &N, &c_zero, &c_zero, R, &ldr );
lapackf77_slacpy( "Upper", &min_mn, &N, h_R, &lda, R, &ldr );
// error = || R - Q^H*A || / (N * ||A||)
blasf77_sgemm( "Conj", "NoTrans", &min_mn, &N, &M,
&c_neg_one, Q, &ldq, h_A, &lda, &c_one, R, &ldr );
Anorm = lapackf77_slange( "1", &M, &N, h_A, &lda, work );
*error = lapackf77_slange( "1", &min_mn, &N, R, &ldr, work );
if ( N > 0 && Anorm > 0 )
*error /= (N*Anorm);
// set R = I (K by K identity), then R = I - Q^H*Q
// error = || I - Q^H*Q || / N
lapackf77_slaset( "Upper", &min_mn, &min_mn, &c_zero, &c_one, R, &ldr );
blasf77_ssyrk( "Upper", "Conj", &min_mn, &M, &d_neg_one, Q, &ldq, &d_one, R, &ldr );
*error2 = safe_lapackf77_slansy( "1", "Upper", &min_mn, R, &ldr, work );
if ( N > 0 )
*error2 /= N;
}
/* ////////////////////////////////////////////////////////////////////////////
-- Testing sgeqrf_batched
*/
int main( int argc, char** argv)
{
TESTING_CHECK( magma_init() );
magma_print_environment();
real_Double_t gflops, magma_perf, magma_time, device_perf=0, device_time=0, cpu_perf, cpu_time;
float magma_error, cublas_error, magma_error2, cublas_error2;
float *h_A, *h_R, *h_Amagma, *tau, *h_work, tmp[1], unused[1];
float *d_A, *dtau_magma, *dtau_cublas;
float **dA_array = NULL;
float **dtau_array = NULL;
magma_int_t *dinfo_magma, *dinfo_cublas;
magma_int_t M, N, lda, ldda, lwork, n2, info, min_mn;
magma_int_t ione = 1;
magma_int_t ISEED[4] = {0,0,0,1};
int status = 0;
int device_info; // not magma_int_t
magma_int_t batchCount;
magma_int_t column;
magma_opts opts( MagmaOptsBatched );
opts.parse_opts( argc, argv );
batchCount = opts.batchcount;
float tol = opts.tolerance * lapackf77_slamch("E");
printf("%% BatchCount M N MAGMA Gflop/s (ms) %s Gflop/s (ms) CPU Gflop/s (ms) |R - Q^H*A|_mag |I - Q^H*Q|_mag |R - Q^H*A|_cub |I - Q^H*Q|_cub\n", g_platform_str);
printf("%%============================================================================================================================================================\n");
for( int itest = 0; itest < opts.ntest; ++itest ) {
for( int iter = 0; iter < opts.niter; ++iter ) {
M = opts.msize[itest];
N = opts.nsize[itest];
min_mn = min(M, N);
lda = M;
n2 = lda*N * batchCount;
ldda = M;
ldda = magma_roundup( M, opts.align ); // multiple of 32 by default
gflops = (FLOPS_SGEQRF( M, N ) + FLOPS_SGEQRT( M, N )) / 1e9 * batchCount;
/* Allocate memory for the matrix */
TESTING_CHECK( magma_smalloc_cpu( &tau, min_mn * batchCount ));
TESTING_CHECK( magma_smalloc_cpu( &h_A, n2 ));
TESTING_CHECK( magma_smalloc_cpu( &h_Amagma, n2 ));
TESTING_CHECK( magma_smalloc_pinned( &h_R, n2 ));
TESTING_CHECK( magma_smalloc( &d_A, ldda*N * batchCount ));
TESTING_CHECK( magma_smalloc( &dtau_magma, min_mn * batchCount ));
TESTING_CHECK( magma_smalloc( &dtau_cublas, min_mn * batchCount ));
TESTING_CHECK( magma_imalloc( &dinfo_magma, batchCount ));
TESTING_CHECK( magma_imalloc( &dinfo_cublas, batchCount ));
TESTING_CHECK( magma_malloc( (void**) &dA_array, batchCount * sizeof(float*) ));
TESTING_CHECK( magma_malloc( (void**) &dtau_array, batchCount * sizeof(float*) ));
// to determine the size of lwork
lwork = -1;
lapackf77_sgeqrf( &M, &N, unused, &M, unused, tmp, &lwork, &info );
lwork = (magma_int_t)MAGMA_S_REAL( tmp[0] );
TESTING_CHECK( magma_smalloc_cpu( &h_work, lwork * batchCount ));
column = N * batchCount;
/* Initialize the matrix */
lapackf77_slarnv( &ione, ISEED, &n2, h_A );
lapackf77_slacpy( MagmaFullStr, &M, &column, h_A, &lda, h_R, &lda );
/* ====================================================================
Performs operation using MAGMA
=================================================================== */
magma_ssetmatrix( M, column, h_R, lda, d_A, ldda, opts.queue );
magma_sset_pointer( dA_array, d_A, 1, 0, 0, ldda*N, batchCount, opts.queue );
magma_sset_pointer( dtau_array, dtau_magma, 1, 0, 0, min_mn, batchCount, opts.queue );
magma_time = magma_sync_wtime( opts.queue );
info = magma_sgeqrf_batched(M, N, dA_array, ldda, dtau_array, dinfo_magma, batchCount, opts.queue);
magma_time = magma_sync_wtime( opts.queue ) - magma_time;
magma_perf = gflops / magma_time;
magma_sgetmatrix( M, column, d_A, ldda, h_Amagma, lda, opts.queue );
if (info != 0) {
printf("magma_sgeqrf_batched returned error %lld: %s.\n",
(long long) info, magma_strerror( info ));
}
/* ====================================================================
Performs operation using CUBLAS
=================================================================== */
magma_ssetmatrix( M, column, h_R, lda, d_A, ldda, opts.queue );
magma_sset_pointer( dA_array, d_A, 1, 0, 0, ldda*N, batchCount, opts.queue );
magma_sset_pointer( dtau_array, dtau_cublas, 1, 0, 0, min_mn, batchCount, opts.queue );
device_time = magma_sync_wtime( opts.queue );
#ifdef MAGMA_HAVE_CUDA
/* cublasSgeqrfBatched is only available from CUBLAS v6.5 */
#if CUDA_VERSION >= 6050
cublasSgeqrfBatched( opts.handle, int(M), int(N),
dA_array, int(ldda), dtau_array,
&device_info, int(batchCount) );
#endif
#else
hipblasSgeqrfBatched( opts.handle, int(M), int(N),
(float**)dA_array, int(ldda),
(float**)dtau_array,
&device_info, int(batchCount) );
#endif
device_time = magma_sync_wtime( opts.queue ) - device_time;
device_perf = gflops / device_time;
if (device_info != 0) {
printf("cublasSgeqrfBatched returned error %lld: %s.\n",
(long long) device_info, magma_strerror( device_info ));
}
/* =====================================================================
Performs operation using LAPACK
=================================================================== */
if ( opts.lapack ) {
cpu_time = magma_wtime();
// #define BATCHED_DISABLE_PARCPU
#if !defined (BATCHED_DISABLE_PARCPU) && defined(_OPENMP)
magma_int_t nthreads = magma_get_lapack_numthreads();
magma_set_lapack_numthreads(1);
magma_set_omp_numthreads(nthreads);
#pragma omp parallel for schedule(dynamic)
#endif
for (magma_int_t s=0; s < batchCount; s++)
{
magma_int_t locinfo;
lapackf77_sgeqrf(&M, &N, h_A + s * lda * N, &lda, tau + s * min_mn, h_work + s * lwork, &lwork, &locinfo);
if (locinfo != 0) {
printf("lapackf77_sgeqrf matrix %lld returned error %lld: %s.\n",
(long long) s, (long long) locinfo, magma_strerror( locinfo ));
}
}
#if !defined (BATCHED_DISABLE_PARCPU) && defined(_OPENMP)
magma_set_lapack_numthreads(nthreads);
#endif
cpu_time = magma_wtime() - cpu_time;
cpu_perf = gflops / cpu_time;
if (info != 0) {
printf("lapackf77_sgeqrf returned error %lld: %s.\n",
(long long) info, magma_strerror( info ));
}
printf("%10lld %5lld %5lld %7.2f (%7.2f) %7.2f (%7.2f) %7.2f (%7.2f)",
(long long) batchCount, (long long) M, (long long) N,
magma_perf, 1000.*magma_time,
device_perf, 1000.*device_time,
cpu_perf, 1000.*cpu_time );
}
else {
printf("%10lld %5lld %5lld %7.2f (%7.2f) %7.2f (%7.2f) --- ( --- )",
(long long) batchCount, (long long) M, (long long) N,
magma_perf, 1000.*magma_time,
device_perf, 1000.*device_time );
}
if (opts.check) {
/* =====================================================================
Check the MAGMA and CUBLAS results by computing residuals
=================================================================== */
magma_int_t ldq = M;
magma_int_t ldr = min_mn;
float *Q, *R;
float *work;
TESTING_CHECK( magma_smalloc_cpu( &Q, batchCount*ldq*min_mn )); // M by K
TESTING_CHECK( magma_smalloc_cpu( &R, batchCount*ldr*N )); // K by N
TESTING_CHECK( magma_smalloc_cpu( &work, batchCount*min_mn ));
/* check magma result */
magma_error = 0;
magma_error2 = 0;
magma_sgetvector(min_mn*batchCount, dtau_magma, 1, tau, 1, opts.queue );
#pragma omp parallel for reduction(max:magma_error,magma_error2)
for (int i=0; i < batchCount; i++) {
float err, err2;
get_QR_error(M, N, min_mn,
h_Amagma + i*lda*N, h_R + i*lda*N, lda, tau + i*min_mn,
Q + i*ldq*min_mn, ldq, R + i*ldr*N, ldr, h_work + i*lwork, lwork,
work + i*min_mn, &err, &err2);
magma_error = magma_max_nan( err, magma_error );
magma_error2 = magma_max_nan( err2, magma_error2 );
}
/* check cublas result */
cublas_error = 0;
cublas_error2 = 0;
#if ((defined(MAGMA_HAVE_CUDA) && CUDA_VERSION >= 6050) || defined(MAGMA_HAVE_HIP))
magma_sgetvector(min_mn*batchCount, dtau_cublas, 1, tau, 1, opts.queue );
magma_sgetmatrix( M, column, d_A, ldda, h_A, lda, opts.queue );
#pragma omp parallel for reduction(max:cublas_error,cublas_error2)
for (int i=0; i < batchCount; i++) {
float err, err2;
get_QR_error(M, N, min_mn,
h_A + i*lda*N, h_R + i*lda*N, lda, tau + i*min_mn,
Q + i*ldq*min_mn, ldq, R + i*ldr*N, ldr, h_work + i*lwork, lwork,
work + i*min_mn, &err, &err2);
cublas_error = magma_max_nan( err, cublas_error );
cublas_error2 = magma_max_nan( err2, cublas_error2 );
}
#endif
magma_free_cpu( Q ); Q = NULL;
magma_free_cpu( R ); R = NULL;
magma_free_cpu( work ); work = NULL;
bool okay = (magma_error < tol && magma_error2 < tol);
status += ! okay;
printf(" %15.2e %15.2e %15.2e %15.2e %s\n",
magma_error, magma_error2,
cublas_error, cublas_error2,
(okay ? "ok" : "failed") );
}
else {
printf("\n");
}
magma_free_cpu( tau );
magma_free_cpu( h_A );
magma_free_cpu( h_Amagma );
magma_free_cpu( h_work );
magma_free_pinned( h_R );
magma_free( d_A );
magma_free( dtau_magma );
magma_free( dtau_cublas );
magma_free( dinfo_magma );
magma_free( dinfo_cublas );
magma_free( dA_array );
magma_free( dtau_array );
fflush( stdout );
}
if ( opts.niter > 1 ) {
printf( "\n" );
}
}
opts.cleanup();
TESTING_CHECK( magma_finalize() );
return status;
}
|
