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
|
#include <errno.h>
#include <inttypes.h>
#include <pthread.h>
#include <math.h>
#include <stdint.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <strings.h>
#include <unistd.h>
#include <sys/time.h>
#include <ck_pr.h>
#include "../../common.h"
/* 8! = 40320, evenly divide 1 .. 8 processor workload. */
#define WORKLOAD (40320 * 2056)
struct block {
unsigned int tid;
};
static struct affinity a;
static unsigned int ready;
static uint64_t *count;
static uint64_t nthr;
static uint64_t object[2] CK_CC_CACHELINE;
static void *
fairness(void *null)
{
struct block *context = null;
unsigned int i = context->tid;
if (aff_iterate(&a)) {
perror("ERROR: Could not affine thread");
exit(EXIT_FAILURE);
}
while (ck_pr_load_uint(&ready) == 0);
while (ck_pr_load_uint(&ready)) {
ATOMIC;
ATOMIC;
ATOMIC;
ATOMIC;
ck_pr_store_64(count + i, count[i] + 1);
}
return (NULL);
}
int
main(int argc, char *argv[])
{
uint64_t v, d;
unsigned int i;
pthread_t *threads;
struct block *context;
if (argc != 3) {
ck_error("Usage: " ATOMIC_STRING " <number of threads> <affinity delta>\n");
exit(EXIT_FAILURE);
}
nthr = atoi(argv[1]);
if (nthr <= 0) {
ck_error("ERROR: Number of threads must be greater than 0\n");
exit(EXIT_FAILURE);
}
threads = malloc(sizeof(pthread_t) * nthr);
if (threads == NULL) {
ck_error("ERROR: Could not allocate thread structures\n");
exit(EXIT_FAILURE);
}
context = malloc(sizeof(struct block) * nthr);
if (context == NULL) {
ck_error("ERROR: Could not allocate thread contexts\n");
exit(EXIT_FAILURE);
}
a.delta = atoi(argv[2]);
a.request = 0;
count = malloc(sizeof(uint64_t) * nthr);
if (count == NULL) {
ck_error("ERROR: Could not create acquisition buffer\n");
exit(EXIT_FAILURE);
}
memset(count, 0, sizeof(uint64_t) * nthr);
fprintf(stderr, "Creating threads (fairness)...");
for (i = 0; i < nthr; i++) {
context[i].tid = i;
if (pthread_create(&threads[i], NULL, fairness, context + i)) {
ck_error("ERROR: Could not create thread %d\n", i);
exit(EXIT_FAILURE);
}
}
fprintf(stderr, "done\n");
ck_pr_store_uint(&ready, 1);
common_sleep(10);
ck_pr_store_uint(&ready, 0);
fprintf(stderr, "Waiting for threads to finish acquisition regression...");
for (i = 0; i < nthr; i++)
pthread_join(threads[i], NULL);
fprintf(stderr, "done\n\n");
for (i = 0, v = 0; i < nthr; i++) {
printf("%d %15" PRIu64 "\n", i, count[i]);
v += count[i];
}
printf("\n# total : %15" PRIu64 "\n", v);
printf("# throughput : %15" PRIu64 " a/s\n", (v /= nthr) / 10);
for (i = 0, d = 0; i < nthr; i++)
d += (count[i] - v) * (count[i] - v);
printf("# average : %15" PRIu64 "\n", v);
printf("# deviation : %.2f (%.2f%%)\n\n", sqrt(d / nthr), (sqrt(d / nthr) / v) * 100.00);
return (0);
}
|
