File: test_batch_trsm_device.cc

package info (click to toggle)
blaspp 2024.10.26-1
  • links: PTS, VCS
  • area: main
  • in suites: forky, sid, trixie
  • size: 2,636 kB
  • sloc: cpp: 29,332; ansic: 8,448; python: 2,192; xml: 182; perl: 101; makefile: 53; sh: 7
file content (248 lines) | stat: -rw-r--r-- 8,230 bytes parent folder | download
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
// 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 "cblas_wrappers.hh"
#include "lapack_wrappers.hh"
#include "blas/flops.hh"
#include "print_matrix.hh"
#include "check_gemm.hh"

// -----------------------------------------------------------------------------
template <typename TA, typename TB>
void test_device_batch_trsm_work( Params& params, bool run )
{
    using namespace testsweeper;
    using namespace blas::batch;
    using blas::Uplo;
    using blas::Side;
    using blas::Op;
    using blas::Layout;
    using blas::Diag;
    using scalar_t = blas::scalar_type< TA, TB >;
    using real_t   = blas::real_type< scalar_t >;
    using std::swap;

    // get & mark input values
    blas::Layout layout = params.layout();
    blas::Side side_    = params.side();
    blas::Uplo uplo_    = params.uplo();
    blas::Op trans_     = params.trans();
    blas::Diag diag_    = params.diag();
    scalar_t alpha_     = params.alpha.get<scalar_t>();
    int64_t m_          = params.dim.m();
    int64_t n_          = params.dim.n();
    size_t batch        = params.batch();
    int64_t device      = params.device();
    int64_t align       = params.align();
    int64_t verbose     = params.verbose();

    // mark non-standard output values
    params.gflops();
    params.ref_time();
    params.ref_gflops();

    if (! run)
        return;

    if (blas::get_device_count() == 0) {
        params.msg() = "skipping: no GPU devices or no GPU support";
        return;
    }

    // ----------
    // setup
    int64_t Am    = (side_ == Side::Left ? m_ : n_);
    int64_t Bm    = m_;
    int64_t Bn    = n_;
    if (layout == Layout::RowMajor)
        swap( Bm, Bn );
    int64_t lda_  = roundup( Am, align );
    int64_t ldb_  = roundup( Bm, align );
    size_t size_A = size_t(lda_)*Am;
    size_t size_B = size_t(ldb_)*Bn;
    TA* A         = new TA[ batch * size_A ];
    TB* B         = new TB[ batch * size_B ];
    TB* Bref      = new TB[ batch * size_B ];

    // device specifics
    blas::Queue queue( device );
    TA* dA;
    TB* dB;

    dA = blas::device_malloc<TA>( batch * size_A, queue );
    dB = blas::device_malloc<TB>( batch * size_B, queue );

    // pointer arrays
    std::vector<TA*>    Aarray( batch );
    std::vector<TB*>    Barray( batch );
    std::vector<TA*>   dAarray( batch );
    std::vector<TB*>   dBarray( batch );
    std::vector<TB*> Brefarray( batch );

    for (size_t i = 0; i < batch; ++i) {
         Aarray[i]   =  A   + i * size_A;
         Barray[i]   =  B   + i * size_B;
        dAarray[i]   = dA   + i * size_A;
        dBarray[i]   = dB   + i * size_B;
        Brefarray[i] = Bref + i * size_B;
    }

    // info
    std::vector<int64_t> info( batch );

    // wrap scalar arguments in std::vector
    std::vector<Side>     side(1, side_);
    std::vector<Uplo>     uplo(1, uplo_);
    std::vector<Op>       trans(1, trans_);
    std::vector<Diag>     diag(1, diag_);
    std::vector<int64_t>  m(1, m_);
    std::vector<int64_t>  n(1, n_);
    std::vector<int64_t>  ldda(1, lda_);
    std::vector<int64_t>  lddb(1, ldb_);
    std::vector<scalar_t> alpha(1, alpha_);

    int64_t idist = 1;
    int iseed[4] = { 0, 0, 0, 1 };
    lapack_larnv( idist, iseed, batch * size_A, A );  // TODO: generate
    lapack_larnv( idist, iseed, batch * size_B, B );  // TODO
    lapack_lacpy( "g", Bm, batch * Bn, B, ldb_, Bref, ldb_ );

    // set unused data to nan
    /*if (uplo_ == Uplo::Lower) {
        for (int64_t j = 0; j < Am; ++j)
            for (int64_t i = 0; i < j; ++i)  // upper
                A[ i + j*lda_ ] = nan("");
    }
    else {
        for (int64_t j = 0; j < Am; ++j)
            for (int64_t i = j+1; i < Am; ++i)  // lower
                A[ i + j*lda_ ] = nan("");
    }*/

    // Factor A into L L^H or U U^H to get a well-conditioned triangular matrix.
    // If diag_ == Unit, the diag_onal is replaced; this is still well-conditioned.
    for (size_t s = 0; s < batch; ++s) {
        TA* pA = Aarray[s];
        // First, brute force positive definiteness.
        for (int64_t i = 0; i < Am; ++i) {
            pA[ i + i*lda_ ] += Am;
        }
        int64_t potrf_info = 0;
        lapack_potrf( to_c_string( uplo_ ), Am, pA, lda_, &potrf_info );
        require( potrf_info == 0 );
    }

    // if row-major, transpose A
    if (layout == Layout::RowMajor) {
        for (size_t s = 0; s < batch; ++s) {
            for (int64_t j = 0; j < Am; ++j) {
                for (int64_t i = 0; i < j; ++i) {
                    swap( Aarray[s][ i + j*lda_ ], Aarray[s][ j + i*lda_ ] );
                }
            }
        }
    }

    blas::device_copy_matrix(Am, batch * Am, A, lda_, dA, lda_, queue);
    blas::device_copy_matrix(Bm, batch * Bn, B, ldb_, dB, ldb_, queue);
    queue.sync();

    // norms for error check
    real_t work[1];
    real_t* Anorm = new real_t[ batch ];
    real_t* Bnorm = new real_t[ batch ];
    for (size_t s = 0; s < batch; ++s) {
        Anorm[ s ] = lapack_lantr( "f", to_c_string( uplo_ ), to_c_string( diag_ ), Am, Am, Aarray[s], lda_, work );
        Bnorm[ s ] = lapack_lange( "f", Bm, Bn, Barray[s], ldb_, work );
    }

    // decide error checking mode
    info.resize( 0 );

    // run test
    testsweeper::flush_cache( params.cache() );
    double time = get_wtime();
    blas::batch::trsm( layout, side, uplo, trans, diag, m, n, alpha, dAarray, ldda, dBarray, lddb,
                       batch, info, queue );
    queue.sync();
    time = get_wtime() - time;

    double gflop = batch * blas::Gflop< scalar_t >::trsm( side_, m_, n_ );
    params.time()   = time;
    params.gflops() = gflop / time;

    blas::device_copy_matrix(Bm, batch * Bn, dB, ldb_, B, ldb_, queue);
    queue.sync();

    if (params.check() == 'y') {
        // run reference
        testsweeper::flush_cache( params.cache() );
        time = get_wtime();
        for (size_t i = 0; i < batch; ++i) {
            cblas_trsm( cblas_layout_const(layout),
                        cblas_side_const(side_),
                        cblas_uplo_const(uplo_),
                        cblas_trans_const(trans_),
                        cblas_diag_const(diag_),
                        m_, n_, alpha_, Aarray[i], lda_, Brefarray[i], ldb_ );
        }
        time = get_wtime() - time;

        params.ref_time()   = time;
        params.ref_gflops() = gflop / time;

        // check error compared to reference
        // Am is reduction dimension
        // beta = 0, Cnorm = 0 (initial).
        real_t err, error = 0;
        bool ok, okay = true;
        for (size_t i = 0; i < batch; ++i) {
            check_gemm( Bm, Bn, Am, alpha_, scalar_t(0), Anorm[i], Bnorm[i], real_t(0),
                        Brefarray[i], ldb_, Barray[i], ldb_, verbose, &err, &ok );
            error = std::max( error, err );
            okay &= ok;
        }
        params.error() = error;
        params.okay() = okay;
    }

    delete[] A;
    delete[] B;
    delete[] Bref;
    delete[] Anorm;
    delete[] Bnorm;

    blas::device_free( dA, queue );
    blas::device_free( dB, queue );
}

// -----------------------------------------------------------------------------
void test_batch_trsm_device( Params& params, bool run )
{
    switch (params.datatype()) {
        case testsweeper::DataType::Single:
            test_device_batch_trsm_work< float, float >( params, run );
            break;

        case testsweeper::DataType::Double:
            test_device_batch_trsm_work< double, double >( params, run );
            break;

        case testsweeper::DataType::SingleComplex:
            test_device_batch_trsm_work< std::complex<float>, std::complex<float> >
                ( params, run );
            break;

        case testsweeper::DataType::DoubleComplex:
            test_device_batch_trsm_work< std::complex<double>, std::complex<double> >
                ( params, run );
            break;

        default:
            throw std::exception();
            break;
    }
}