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
|
/******************************************************************************
* Copyright (c) Intel Corporation - All rights reserved. *
* This file is part of the LIBXSMM library. *
* *
* For information on the license, see the LICENSE file. *
* Further information: https://github.com/hfp/libxsmm/ *
* SPDX-License-Identifier: BSD-3-Clause *
******************************************************************************/
/* Hans Pabst (Intel Corp.)
******************************************************************************/
#include <libxsmm.h>
#if defined(LIBXSMM_OFFLOAD_TARGET)
# pragma offload_attribute(push,target(LIBXSMM_OFFLOAD_TARGET))
#endif
#include <stdlib.h>
#include <string.h>
#include <stdio.h>
#if defined(_OPENMP)
# include <omp.h>
#endif
#if defined(__TBB)
# include <tbb/scalable_allocator.h>
#endif
#if defined(LIBXSMM_OFFLOAD_TARGET)
# pragma offload_attribute(pop)
#endif
#if defined(__TBB)
# define MALLOC scalable_malloc
# define FREE scalable_free
#elif defined(_OPENMP) && defined(LIBXSMM_INTEL_COMPILER) && (1901 > LIBXSMM_INTEL_COMPILER)
# define MALLOC kmp_malloc
# define FREE kmp_free
#elif 1
# define MALLOC malloc
# define FREE free
#endif
#if !defined(MAX_MALLOC_MB)
# define MAX_MALLOC_MB 100
#endif
#if !defined(MAX_MALLOC_N)
# define MAX_MALLOC_N 24
#endif
void* malloc_offsite(size_t size);
int main(int argc, char* argv[])
{
#if defined(_OPENMP)
const int max_nthreads = omp_get_max_threads();
#else
const int max_nthreads = 1;
#endif
const int ncycles = LIBXSMM_MAX(1 < argc ? atoi(argv[1]) : 100, 1);
const int max_nallocs = LIBXSMM_CLMP(2 < argc ? atoi(argv[2]) : 4, 1, MAX_MALLOC_N);
const int nthreads = LIBXSMM_CLMP(3 < argc ? atoi(argv[3]) : 1, 1, max_nthreads);
const char *const env_check = getenv("CHECK");
const double check = LIBXSMM_ABS(NULL == env_check ? 0 : atof(env_check));
unsigned int nallocs = 0, nerrors0 = 0, nerrors1 = 0;
int r[MAX_MALLOC_N], i;
int max_size = 0;
/* generate set of random numbers for parallel region */
for (i = 0; i < (MAX_MALLOC_N); ++i) r[i] = rand();
/* count number of calls according to randomized scheme */
for (i = 0; i < ncycles; ++i) {
const int count = r[i%(MAX_MALLOC_N)] % max_nallocs + 1;
int mbytes = 0, j;
for (j = 0; j < count; ++j) {
const int k = (i * count + j) % (MAX_MALLOC_N);
mbytes += (r[k] % (MAX_MALLOC_MB) + 1);
}
if (max_size < mbytes) max_size = mbytes;
nallocs += count;
}
assert(0 != nallocs);
fprintf(stdout, "Running %i cycles with max. %i malloc+free (%u calls) using %i thread%s...\n",
ncycles, max_nallocs, nallocs, 1 >= nthreads ? 1 : nthreads, 1 >= nthreads ? "" : "s");
#if defined(LIBXSMM_OFFLOAD_TARGET)
# pragma offload target(LIBXSMM_OFFLOAD_TARGET)
#endif
{
const char *const longlife_env = getenv("LONGLIFE");
const int enable_longlife = ((NULL == longlife_env || 0 == *longlife_env) ? 0 : atoi(longlife_env));
void* longlife = (0 == enable_longlife ? NULL : malloc_offsite((MAX_MALLOC_MB) << 20));
libxsmm_timer_tickint d0 = 0, d1 = 0;
libxsmm_scratch_info info;
int scratch = 0;
libxsmm_init();
#if defined(_OPENMP)
# pragma omp parallel for num_threads(nthreads) private(i) reduction(+:d1,nerrors1)
#endif
for (i = 0; i < ncycles; ++i) {
const int count = r[i%(MAX_MALLOC_N)] % max_nallocs + 1;
void* p[MAX_MALLOC_N];
int j;
assert(count <= MAX_MALLOC_N);
for (j = 0; j < count; ++j) {
const int k = (i * count + j) % (MAX_MALLOC_N);
const size_t nbytes = ((size_t)r[k] % (MAX_MALLOC_MB) + 1) << 20;
const libxsmm_timer_tickint t1 = libxsmm_timer_tick();
p[j] = libxsmm_aligned_scratch(nbytes, 0/*auto*/);
d1 += libxsmm_timer_ncycles(t1, libxsmm_timer_tick());
if (NULL == p[j]) {
++nerrors1;
}
else if (0 != check) {
memset(p[j], j, nbytes);
}
}
for (j = 0; j < count; ++j) {
libxsmm_free(p[j]);
}
}
libxsmm_free(longlife);
if (EXIT_SUCCESS == libxsmm_get_scratch_info(&info) && 0 < info.size) {
scratch = (int)(1.0 * LIBXSMM_MAX(info.size, info.local) / (1ULL << 20) + 0.5);
fprintf(stdout, "\nScratch: %i MB (mallocs=%lu, pools=%u)\n",
scratch, (unsigned long int)info.nmallocs, info.npools);
libxsmm_release_scratch(); /* suppress LIBXSMM's termination message about scratch */
}
#if (defined(MALLOC) && defined(FREE))
longlife = (0 == enable_longlife ? NULL : MALLOC((MAX_MALLOC_MB) << 20));
if (NULL == longlife) max_size += MAX_MALLOC_MB;
#if defined(_OPENMP)
# pragma omp parallel for num_threads(nthreads) private(i) reduction(+:d0,nerrors0)
#endif
for (i = 0; i < ncycles; ++i) {
const int count = r[i % (MAX_MALLOC_N)] % max_nallocs + 1;
void* p[MAX_MALLOC_N];
int j;
assert(count <= MAX_MALLOC_N);
for (j = 0; j < count; ++j) {
const int k = (i * count + j) % (MAX_MALLOC_N);
const size_t nbytes = ((size_t)r[k] % (MAX_MALLOC_MB) + 1) << 20;
const libxsmm_timer_tickint t1 = libxsmm_timer_tick();
p[j] = MALLOC(nbytes);
d0 += libxsmm_timer_ncycles(t1, libxsmm_timer_tick());
if (NULL == p[j]) {
++nerrors0;
}
else if (0 != check) {
memset(p[j], j, nbytes);
}
}
for (j = 0; j < count; ++j) FREE(p[j]);
}
FREE(longlife);
#endif /*(defined(MALLOC) && defined(FREE))*/
if (0 != d0 && 0 != d1 && 0 < nallocs) {
const double dcalls = libxsmm_timer_duration(0, d0);
const double dalloc = libxsmm_timer_duration(0, d1);
const double scratch_freq = 1E-3 * nallocs / dalloc;
const double malloc_freq = 1E-3 * nallocs / dcalls;
const double speedup = scratch_freq / malloc_freq;
fprintf(stdout, "\tlibxsmm scratch calls/s: %.1f kHz\n", scratch_freq);
fprintf(stdout, "Malloc: %i MB\n", max_size);
fprintf(stdout, "\tstd.malloc+free calls/s: %.1f kHz\n", malloc_freq);
fprintf(stdout, "Fair (size vs. speed): %.1fx\n", max_size * speedup / scratch);
fprintf(stdout, "Scratch Speedup: %.1fx\n", speedup);
}
}
if (0 != nerrors0 || 0 != nerrors1) {
fprintf(stdout, "FAILED (errors: malloc=%u libxsmm=%u)\n", nerrors0, nerrors1);
return EXIT_FAILURE;
}
return EXIT_SUCCESS;
}
void* malloc_offsite(size_t size) { return libxsmm_aligned_scratch(size, 0/*auto*/); }
|
