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
|
// 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, typename TC>
void test_batch_hemm_device_work( Params& params, bool run )
{
using namespace testsweeper;
using blas::Side;
using blas::Layout;
using scalar_t = blas::scalar_type< TA, TB, TC >;
using real_t = blas::real_type< scalar_t >;
// get & mark input values
blas::Layout layout = params.layout();
blas::Side side_ = params.side();
blas::Uplo uplo_ = params.uplo();
scalar_t alpha_ = params.alpha.get<scalar_t>();
scalar_t beta_ = params.beta.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 An = (side_ == Side::Left ? m_ : n_);
int64_t Cm = m_;
int64_t Cn = n_;
if (layout == Layout::RowMajor)
std::swap( Cm, Cn );
int64_t lda_ = roundup( An, align );
int64_t ldb_ = roundup( Cm, align );
int64_t ldc_ = roundup( Cm, align );
size_t size_A = size_t(lda_)*An;
size_t size_B = size_t(ldb_)*Cn;
size_t size_C = size_t(ldc_)*Cn;
TA* A = new TA[ batch * size_A ];
TB* B = new TB[ batch * size_B ];
TC* C = new TC[ batch * size_C ];
TC* Cref = new TC[ batch * size_C ];
// device specifics
blas::Queue queue( device );
TA* dA = blas::device_malloc<TA>( batch * size_A, queue );
TB* dB = blas::device_malloc<TB>( batch * size_B, queue );
TC* dC = blas::device_malloc<TC>( batch * size_C, queue );
// pointer arrays
std::vector<TA*> Aarray( batch );
std::vector<TB*> Barray( batch );
std::vector<TC*> Carray( batch );
std::vector<TC*> Crefarray( batch );
std::vector<TA*> dAarray( batch );
std::vector<TB*> dBarray( batch );
std::vector<TC*> dCarray( batch );
for (size_t s = 0; s < batch; ++s) {
Aarray[s] = A + s * size_A;
Barray[s] = B + s * size_B;
Carray[s] = C + s * size_C;
Crefarray[s] = Cref + s * size_C;
dAarray[s] = dA + s * size_A;
dBarray[s] = dB + s * size_B;
dCarray[s] = dC + s * size_C;
}
// info
std::vector<int64_t> info( batch );
// wrap scalar arguments in std::vector
std::vector<blas::Side> side(1, side_);
std::vector<blas::Uplo> uplo(1, uplo_);
std::vector<int64_t> m(1, m_);
std::vector<int64_t> n(1, n_);
std::vector<int64_t> lda(1, lda_);
std::vector<int64_t> ldb(1, ldb_);
std::vector<int64_t> ldc(1, ldc_);
std::vector<scalar_t> alpha(1, alpha_);
std::vector<scalar_t> beta(1, beta_);
int64_t idist = 1;
int iseed[4] = { 0, 0, 0, 1 };
lapack_larnv( idist, iseed, batch * size_A, A );
lapack_larnv( idist, iseed, batch * size_B, B );
lapack_larnv( idist, iseed, batch * size_C, C );
lapack_lacpy( "g", Cm, batch * Cn, C, ldc_, Cref, ldc_ );
blas::device_copy_matrix(An, batch * An, A, lda_, dA, lda_, queue);
blas::device_copy_matrix(Cm, batch * Cn, B, ldb_, dB, ldb_, queue);
blas::device_copy_matrix(Cm, batch * Cn, C, ldc_, dC, ldc_, 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 ];
real_t* Cnorm = new real_t[ batch ];
for (size_t s = 0; s < batch; ++s) {
Anorm[s] = lapack_lansy( "f", to_c_string( uplo_ ), An, Aarray[s], lda_, work );
Bnorm[s] = lapack_lange( "f", Cm, Cn, Barray[s], ldb_, work );
Cnorm[s] = lapack_lange( "f", Cm, Cn, Carray[s], ldc_, work );
}
// decide error checking mode
info.resize( 0 );
// run test
testsweeper::flush_cache( params.cache() );
double time = get_wtime();
blas::batch::hemm( layout, side, uplo, m, n, alpha, dAarray, lda, dBarray, ldb, beta, dCarray, ldc,
batch, info, queue);
queue.sync();
time = get_wtime() - time;
double gflop = batch * blas::Gflop< scalar_t >::hemm( side_, m_, n_ );
params.time() = time;
params.gflops() = gflop / time;
blas::device_copy_matrix(Cm, batch * Cn, dC, ldc_, C, ldc_, queue);
queue.sync();
if (params.ref() == 'y' || params.check() == 'y') {
// run reference
testsweeper::flush_cache( params.cache() );
time = get_wtime();
for (size_t s = 0; s < batch; ++s) {
cblas_hemm( cblas_layout_const(layout),
cblas_side_const(side_),
cblas_uplo_const(uplo_),
m_, n_, alpha_, Aarray[s], lda_, Barray[s], ldb_, beta_, Crefarray[s], ldc_ );
}
time = get_wtime() - time;
params.ref_time() = time;
params.ref_gflops() = gflop / time;
// check error compared to reference
real_t err, error = 0;
bool ok, okay = true;
for (size_t s = 0; s < batch; ++s) {
check_gemm( Cm, Cn, An, alpha_, beta_, Anorm[s], Bnorm[s], Cnorm[s],
Crefarray[s], ldc_, Carray[s], ldc_, verbose, &err, &ok );
error = std::max( error, err );
okay &= ok;
}
params.error() = error;
params.okay() = okay;
}
delete[] A;
delete[] B;
delete[] C;
delete[] Cref;
delete[] Anorm;
delete[] Bnorm;
delete[] Cnorm;
blas::device_free( dA, queue );
blas::device_free( dB, queue );
blas::device_free( dC, queue );
}
// -----------------------------------------------------------------------------
void test_batch_hemm_device( Params& params, bool run )
{
switch (params.datatype()) {
case testsweeper::DataType::Single:
test_batch_hemm_device_work< float, float, float >( params, run );
break;
case testsweeper::DataType::Double:
test_batch_hemm_device_work< double, double, double >( params, run );
break;
case testsweeper::DataType::SingleComplex:
test_batch_hemm_device_work< std::complex<float>, std::complex<float>,
std::complex<float> >( params, run );
break;
case testsweeper::DataType::DoubleComplex:
test_batch_hemm_device_work< std::complex<double>, std::complex<double>,
std::complex<double> >( params, run );
break;
default:
throw std::exception();
break;
}
}
|
