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
|
/*
* Copyright 2023 Advanced Micro Devices, Inc.
*
* SPDX-License-Identifier: MIT
*/
#include "thread_sched.h"
#include "u_cpu_detect.h"
#include "u_debug.h"
DEBUG_GET_ONCE_BOOL_OPTION(pin_threads, "mesa_pin_threads", false)
bool
util_thread_scheduler_enabled(void)
{
#if DETECT_ARCH_X86 || DETECT_ARCH_X86_64
return util_get_cpu_caps()->num_L3_caches > 1 ||
debug_get_option_pin_threads();
#else
return false;
#endif
}
void
util_thread_scheduler_init_state(unsigned *state)
{
*state = UINT32_MAX;
util_thread_sched_apply_policy(thrd_current(), UTIL_THREAD_APP_CALLER, 0,
NULL); /* keep as NULL */
}
/**
* Apply the optimal thread scheduling policy for the given thread.
*
* "name" determines which thread the policy is being applied to.
*
* "app_thread_cpu" is the CPU where the application thread currently
* resides.
*
* "sched_state" is a per-gl_context state that this function uses to track
* what happened in previous invocations.
*/
bool
util_thread_sched_apply_policy(thrd_t thread, enum util_thread_name name,
unsigned app_thread_cpu, unsigned *sched_state)
{
#if DETECT_ARCH_X86 || DETECT_ARCH_X86_64
if (debug_get_option_pin_threads()) {
/* Pin threads to a specific CPU. This is done only once. *sched_state
* is true if this is the first time we are doing it.
*/
if (sched_state && !*sched_state)
return false;
/* Each thread is assigned to a different CPU. */
unsigned mask = BITFIELD_BIT(name);
if (sched_state)
*sched_state = 0;
return util_set_thread_affinity(thread, &mask, NULL, 32);
}
/* Don't do anything for the app thread with the L3 chasing policy. */
if (name == UTIL_THREAD_APP_CALLER)
return false;
/* Move Mesa threads to the L3 core complex where the app thread
* resides. We call this "L3 chasing".
*
* This improves multithreading performance by up to 33% on Ryzen 3900X.
*/
const struct util_cpu_caps_t *caps = util_get_cpu_caps();
int L3_cache = caps->cpu_to_L3[app_thread_cpu];
/* Don't do anything if the app thread hasn't moved to a different
* core complex. (*sched_state contains the last set L3 index)
*/
if (L3_cache == U_CPU_INVALID_L3 ||
(sched_state && L3_cache == *sched_state))
return false;
/* Apply the policy. */
if (sched_state)
*sched_state = L3_cache;
return util_set_thread_affinity(thread, caps->L3_affinity_mask[L3_cache],
NULL, caps->num_cpu_mask_bits);
#else
return false;
#endif
}
|
