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
|
/* Ergo, version 3.8.2, a program for linear scaling electronic structure
* calculations.
* Copyright (C) 2023 Elias Rudberg, Emanuel H. Rubensson, Pawel Salek,
* and Anastasia Kruchinina.
*
* This program is free software: you can redistribute it and/or modify
* it under the terms of the GNU General Public License as published by
* the Free Software Foundation, either version 3 of the License, or
* (at your option) any later version.
*
* This program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* GNU General Public License for more details.
*
* You should have received a copy of the GNU General Public License
* along with this program. If not, see <http://www.gnu.org/licenses/>.
*
* Primary academic reference:
* Ergo: An open-source program for linear-scaling electronic structure
* calculations,
* Elias Rudberg, Emanuel H. Rubensson, Pawel Salek, and Anastasia
* Kruchinina,
* SoftwareX 7, 107 (2018),
* <http://dx.doi.org/10.1016/j.softx.2018.03.005>
*
* For further information about Ergo, see <http://www.ergoscf.org>.
*/
/** @file bench_gemm_only.cc Benchmark of the matrix library with
* input parameters specifying block sizes, parallel level etc.
*/
#include <fstream> /* For ifstream */
#include <iomanip> /* For setprecision in fstream */
#include <iostream>
#include <cmath>
#include <stdio.h> /* For FILE */
#include <sys/time.h>
#include "Matrix.h"
#include "Vector.h"
#include "MatrixSymmetric.h"
#include "MatrixTriangular.h"
#include "MatrixGeneral.h"
#include "VectorGeneral.h"
#include "mat_gblas.h"
#include "Lanczos.h"
static double get_wall_seconds() {
struct timeval tv;
if(gettimeofday(&tv, NULL) != 0)
throw std::runtime_error("Error in get_wall_seconds(), in gettimeofday().");
double seconds = tv.tv_sec + (double)tv.tv_usec / 1000000;
return seconds;
}
using namespace mat;
template<typename real>
int mainFun(int size, int parallel_level, int block_size,
int block_factor_1, int block_factor_2, int block_factor_3) {
typedef Matrix<real, real> Mat_1;
typedef Matrix<real, Mat_1> Mat_2;
typedef Matrix<real, Mat_2> Mat_3;
typedef Matrix<real, Mat_3> Mat_4;
typedef Mat_4 matri;
typedef MatrixGeneral<real, matri> normalMatrix;
try {
/********** Initialization of SizesAndBlocks */
int nlevels = 4;
std::vector<int> blockSizes(nlevels);
blockSizes[3] = 1;
blockSizes[2] = blockSizes[3] * block_size;
blockSizes[1] = blockSizes[2] * block_factor_1;
blockSizes[0] = blockSizes[1] * block_factor_2;
for(int i = 0; i < nlevels; i++)
std::cout << "blockSizes[" << i << "] = " << blockSizes[i] << std::endl;
SizesAndBlocks rows(blockSizes, size);
SizesAndBlocks cols(blockSizes, size);
real alpha = 0.77;
{
normalMatrix A, B, C;
A.resetSizesAndBlocks(rows,cols);
B.resetSizesAndBlocks(rows,cols);
A.random();
B.random();
double startTime = get_wall_seconds();
C = alpha * A * B;
double secondsTaken = get_wall_seconds() - startTime;
std::cout << "Operation C = alpha * A * B took " << secondsTaken << " wall seconds." << std::endl;
}
}
catch (Failure & e) {
std::cout << "Failure caught: "<<e.what() << std::endl;
std::exit(1);
}
catch (std::exception & e) {
std::cout << "Exception caught: "<<e.what() << std::endl;
std::exit(1);
}
return 0;
}
int main(int argc,char* argv[]){
double startTime;
int do_single = 1;
int do_double = 1;
int do_longdouble = 1;
int size = 100;
int parallel_level = 2;
int block_size = 10;
int block_factor_1 = 8;
int block_factor_2 = 8;
int block_factor_3 = 8;
if(argc != 10) {
std::cout << "argc != 10, using default parameters." << std::endl;
}
else {
do_single = atoi(argv[1]);
do_double = atoi(argv[2]);
do_longdouble = atoi(argv[3]);
size = atoi(argv[4]);
parallel_level = atoi(argv[5]);
block_size = atoi(argv[6]);
block_factor_1 = atoi(argv[7]);
block_factor_2 = atoi(argv[8]);
block_factor_3 = atoi(argv[9]);
}
std::cout << "Parameter values:" << std::endl;
std::cout << "do_single = " << do_single << std::endl;
std::cout << "do_double = " << do_double << std::endl;
std::cout << "do_longdouble = " << do_longdouble << std::endl;
std::cout << "size = " << size << std::endl;
std::cout << "parallel_level = " << parallel_level << std::endl;
std::cout << "block_size = " << block_size << std::endl;
std::cout << "block_factor_1 = " << block_factor_1 << std::endl;
std::cout << "block_factor_2 = " << block_factor_2 << std::endl;
std::cout << "block_factor_3 = " << block_factor_3 << std::endl;
std::cout << std::endl;
#ifdef _OPENMP
int defThreads;
const char *env = getenv("OMP_NUM_THREADS");
if ( !(env && (defThreads=atoi(env)) > 0) ) {
defThreads = 1;
}
mat::Params::setNProcs(defThreads);
mat::Params::setMatrixParallelLevel(parallel_level);
std::cout<<"OpenMP is used, number of threads set to "
<<mat::Params::getNProcs()<<". Matrix parallel level: "
<<mat::Params::getMatrixParallelLevel()<<"."<<std::endl;
#endif
if(do_single == 1) {
std::cout<<"Benchmark of matrix library with single precision:" <<std::endl;
startTime = get_wall_seconds();
if (!mainFun<float>(size, parallel_level, block_size,
block_factor_1, block_factor_2, block_factor_3))
std::cout<<"Matrix library benchmark with single precision completed "
"successfully.\n"
<<"Wall time: "
<< get_wall_seconds() - startTime << " seconds.\n\n";
}
if(do_double == 1) {
std::cout<<"Benchmark of matrix library with double precision:" <<std::endl;
startTime = get_wall_seconds();
if (!mainFun<double>(size, parallel_level, block_size,
block_factor_1, block_factor_2, block_factor_3))
std::cout<<"Matrix library benchmark with double precision completed "
"successfully.\n"
<<"Wall time: "
<< get_wall_seconds() - startTime << " seconds.\n\n";
}
if(do_longdouble == 1) {
std::cout<<"Benchmark of matrix library with long double precision:" <<std::endl;
startTime = get_wall_seconds();
if (!mainFun<long double>(size, parallel_level, block_size,
block_factor_1, block_factor_2, block_factor_3))
std::cout<<"Matrix library benchmark with long double precision completed "
"successfully.\n"
<<"Wall time: "
<< get_wall_seconds() - startTime << " seconds.\n\n";
}
std::exit(0);
}
|
