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
|
/*
* Copyright © 2015 Intel Corporation
*
* Permission is hereby granted, free of charge, to any person obtaining a
* copy of this software and associated documentation files (the "Software"),
* to deal in the Software without restriction, including without limitation
* the rights to use, copy, modify, merge, publish, distribute, sublicense,
* and/or sell copies of the Software, and to permit persons to whom the
* Software is furnished to do so, subject to the following conditions:
*
* The above copyright notice and this permission notice (including the next
* paragraph) shall be included in all copies or substantial portions of the
* Software.
*
* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
* IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
* FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
* THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
* LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
* FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
* IN THE SOFTWARE.
*/
#include <pthread.h>
#include "anv_private.h"
#define NUM_THREADS 16
#define STATES_PER_THREAD 1024
#define NUM_RUNS 64
struct job {
pthread_t thread;
unsigned id;
struct anv_state_pool *pool;
uint32_t offsets[STATES_PER_THREAD];
} jobs[NUM_THREADS];
pthread_barrier_t barrier;
static void *alloc_states(void *_job)
{
struct job *job = _job;
pthread_barrier_wait(&barrier);
for (unsigned i = 0; i < STATES_PER_THREAD; i++) {
struct anv_state state = anv_state_pool_alloc(job->pool, 16, 16);
job->offsets[i] = state.offset;
}
return NULL;
}
static void run_test()
{
struct anv_instance instance;
struct anv_device device = {
.instance = &instance,
};
struct anv_state_pool state_pool;
pthread_mutex_init(&device.mutex, NULL);
anv_state_pool_init(&state_pool, &device, 64, 0);
pthread_barrier_init(&barrier, NULL, NUM_THREADS);
for (unsigned i = 0; i < NUM_THREADS; i++) {
jobs[i].pool = &state_pool;
jobs[i].id = i;
pthread_create(&jobs[i].thread, NULL, alloc_states, &jobs[i]);
}
for (unsigned i = 0; i < NUM_THREADS; i++)
pthread_join(jobs[i].thread, NULL);
/* A list of indices, one per thread */
unsigned next[NUM_THREADS];
memset(next, 0, sizeof(next));
int highest = -1;
while (true) {
/* First, we find which thread has the highest next element */
int thread_max = -1;
int max_thread_idx = -1;
for (unsigned i = 0; i < NUM_THREADS; i++) {
if (next[i] >= STATES_PER_THREAD)
continue;
if (thread_max < jobs[i].offsets[next[i]]) {
thread_max = jobs[i].offsets[next[i]];
max_thread_idx = i;
}
}
/* The only way this can happen is if all of the next[] values are at
* BLOCKS_PER_THREAD, in which case, we're done.
*/
if (thread_max == -1)
break;
/* That next element had better be higher than the previous highest */
assert(jobs[max_thread_idx].offsets[next[max_thread_idx]] > highest);
highest = jobs[max_thread_idx].offsets[next[max_thread_idx]];
next[max_thread_idx]++;
}
anv_state_pool_finish(&state_pool);
pthread_mutex_destroy(&device.mutex);
}
int main(int argc, char **argv)
{
for (unsigned i = 0; i < NUM_RUNS; i++)
run_test();
}
|
