Book a Demo!
CoCalc Logo Icon
StoreFeaturesDocsShareSupportNewsAboutPoliciesSign UpSign In
Download
52868 views
1
/*****************************************************************************
2
* cpu-a.S: arm cpu detection
3
*****************************************************************************
4
* Copyright (C) 2009-2016 x264 project
5
*
6
* Authors: David Conrad <[email protected]>
7
*
8
* This program is free software; you can redistribute it and/or modify
9
* it under the terms of the GNU General Public License as published by
10
* the Free Software Foundation; either version 2 of the License, or
11
* (at your option) any later version.
12
*
13
* This program is distributed in the hope that it will be useful,
14
* but WITHOUT ANY WARRANTY; without even the implied warranty of
15
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
16
* GNU General Public License for more details.
17
*
18
* You should have received a copy of the GNU General Public License
19
* along with this program; if not, write to the Free Software
20
* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02111, USA.
21
*
22
* This program is also available under a commercial proprietary license.
23
* For more information, contact us at [email protected].
24
*****************************************************************************/
25
26
#include "asm.S"
27
28
.align 2
29
30
// done in gas because .fpu neon overrides the refusal to assemble
31
// instructions the selected -march/-mcpu doesn't support
32
function x264_cpu_neon_test
33
vadd.i16 q0, q0, q0
34
bx lr
35
endfunc
36
37
// return: 0 on success
38
// 1 if counters were already enabled
39
// 9 if lo-res counters were already enabled
40
function x264_cpu_enable_armv7_counter, export=0
41
mrc p15, 0, r2, c9, c12, 0 // read PMNC
42
ands r0, r2, #1
43
andne r0, r2, #9
44
45
orr r2, r2, #1 // enable counters
46
bic r2, r2, #8 // full resolution
47
mcreq p15, 0, r2, c9, c12, 0 // write PMNC
48
mov r2, #1 << 31 // enable cycle counter
49
mcr p15, 0, r2, c9, c12, 1 // write CNTENS
50
bx lr
51
endfunc
52
53
function x264_cpu_disable_armv7_counter, export=0
54
mrc p15, 0, r0, c9, c12, 0 // read PMNC
55
bic r0, r0, #1 // disable counters
56
mcr p15, 0, r0, c9, c12, 0 // write PMNC
57
bx lr
58
endfunc
59
60
61
.macro READ_TIME r
62
mrc p15, 0, \r, c9, c13, 0
63
.endm
64
65
// return: 0 if transfers neon -> arm transfers take more than 10 cycles
66
// nonzero otherwise
67
function x264_cpu_fast_neon_mrc_test
68
// check for user access to performance counters
69
mrc p15, 0, r0, c9, c14, 0
70
cmp r0, #0
71
bxeq lr
72
73
push {r4-r6,lr}
74
bl x264_cpu_enable_armv7_counter
75
ands r1, r0, #8
76
mov r3, #0
77
mov ip, #4
78
mov r6, #4
79
moveq r5, #1
80
movne r5, #64
81
82
average_loop:
83
mov r4, r5
84
READ_TIME r1
85
1: subs r4, r4, #1
86
.rept 8
87
vmov.u32 lr, d0[0]
88
add lr, lr, lr
89
.endr
90
bgt 1b
91
READ_TIME r2
92
93
subs r6, r6, #1
94
sub r2, r2, r1
95
cmpgt r2, #30 << 3 // assume context switch if it took over 30 cycles
96
addle r3, r3, r2
97
subsle ip, ip, #1
98
bgt average_loop
99
100
// disable counters if we enabled them
101
ands r0, r0, #1
102
bleq x264_cpu_disable_armv7_counter
103
104
lsr r0, r3, #5
105
cmp r0, #10
106
movgt r0, #0
107
pop {r4-r6,pc}
108
endfunc
109
110