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
|
#include <inttypes.h>
#include "../config.h"
#include "rgb2rgb.h"
#ifdef HAVE_MMX
#include "mmx.h"
#endif
#include "../mmx_defs.h"
void rgb24to32(uint8_t *src,uint8_t *dst,uint32_t src_size)
{
uint8_t *dest = dst;
uint8_t *s = src;
uint8_t *end;
#ifdef HAVE_MMX
const uint64_t mask32 = 0x00FFFFFF00FFFFFFULL;
uint8_t *mm_end;
#endif
end = s + src_size;
#ifdef HAVE_MMX
__asm __volatile(PREFETCH" %0\n\t"::"m"(*s):"memory");
mm_end = (uint8_t*)((((unsigned long)end)/(MMREG_SIZE*2))*(MMREG_SIZE*2));
__asm __volatile("movq %0, %%mm7"::"m"(mask32):"memory");
if(mm_end == end) mm_end -= MMREG_SIZE*2;
while(s < mm_end)
{
__asm __volatile(
PREFETCH" 32%1\n\t"
"movd %1, %%mm0\n\t"
"movd 3%1, %%mm1\n\t"
"movd 6%1, %%mm2\n\t"
"movd 9%1, %%mm3\n\t"
"punpckldq %%mm1, %%mm0\n\t"
"punpckldq %%mm3, %%mm2\n\t"
"pand %%mm7, %%mm0\n\t"
"pand %%mm7, %%mm2\n\t"
MOVNTQ" %%mm0, %0\n\t"
MOVNTQ" %%mm2, 8%0"
:"=m"(*dest)
:"m"(*s)
:"memory");
dest += 16;
s += 12;
}
__asm __volatile(SFENCE:::"memory");
__asm __volatile(EMMS:::"memory");
#endif
while(s < end)
{
*dest++ = *s++;
*dest++ = *s++;
*dest++ = *s++;
*dest++ = 0;
}
}
void rgb32to24(uint8_t *src,uint8_t *dst,uint32_t src_size)
{
uint8_t *dest = dst;
uint8_t *s = src;
uint8_t *end;
#ifdef HAVE_MMX
const uint64_t mask24l = 0x0000000000FFFFFFULL;
const uint64_t mask24h = 0x0000FFFFFF000000ULL;
uint8_t *mm_end;
#endif
end = s + src_size;
#ifdef HAVE_MMX
__asm __volatile(PREFETCH" %0\n\t"::"m"(*s):"memory");
mm_end = (uint8_t*)((((unsigned long)end)/(MMREG_SIZE*2))*(MMREG_SIZE*2));
__asm __volatile(
"movq %0, %%mm7\n\t"
"movq %1, %%mm6\n\t"
::"m"(mask24l),"m"(mask24h):"memory");
if(mm_end == end) mm_end -= MMREG_SIZE*2;
while(s < mm_end)
{
__asm __volatile(
PREFETCH" 32%1\n\t"
"movq %1, %%mm0\n\t"
"movq 8%1, %%mm1\n\t"
"movq %%mm0, %%mm2\n\t"
"movq %%mm1, %%mm3\n\t"
"psrlq $8, %%mm2\n\t"
"psrlq $8, %%mm3\n\t"
"pand %%mm7, %%mm0\n\t"
"pand %%mm7, %%mm1\n\t"
"pand %%mm6, %%mm2\n\t"
"pand %%mm6, %%mm3\n\t"
"por %%mm2, %%mm0\n\t"
"por %%mm3, %%mm1\n\t"
MOVNTQ" %%mm0, %0\n\t"
MOVNTQ" %%mm1, 6%0"
:"=m"(*dest)
:"m"(*s)
:"memory");
dest += 12;
s += 16;
}
__asm __volatile(SFENCE:::"memory");
__asm __volatile(EMMS:::"memory");
#endif
while(s < end)
{
*dest++ = *s++;
*dest++ = *s++;
*dest++ = *s++;
s++;
}
}
/* TODO: 3DNOW, MMX2 optimization */
/* Original by Strepto/Astral
ported to gcc & bugfixed : A'rpi */
void rgb15to16(uint8_t *src,uint8_t *dst,uint32_t src_size)
{
#ifdef HAVE_MMX
static uint64_t mask_b = 0x001F001F001F001FLL; // 00000000 00011111 xxB
static uint64_t mask_rg = 0x7FE07FE07FE07FE0LL; // 01111111 11100000 RGx
register char* s=src+src_size;
register char* d=dst+src_size;
register int offs=-src_size;
movq_m2r (mask_b, mm4);
movq_m2r (mask_rg, mm5);
while(offs<0){
movq_m2r (*(s+offs), mm0);
movq_r2r (mm0, mm1);
movq_m2r (*(s+8+offs), mm2);
movq_r2r (mm2, mm3);
pand_r2r (mm4, mm0);
pand_r2r (mm5, mm1);
psllq_i2r(1,mm1);
pand_r2r (mm4, mm2);
pand_r2r (mm5, mm3);
por_r2r (mm1, mm0);
psllq_i2r(1,mm3);
movq_r2m (mm0,*(d+offs));
por_r2r (mm3,mm2);
movq_r2m (mm2,*(d+8+offs));
offs+=16;
}
emms();
#else
uint16_t *s1=( uint16_t * )src;
uint16_t *d1=( uint16_t * )dst;
uint16_t *e=((uint8_t *)s1)+src_size;
while( s1<e ){
register int x=*( s1++ );
/* rrrrrggggggbbbbb
0rrrrrgggggbbbbb
0111 1111 1110 0000=0x7FE0
00000000000001 1111=0x001F */
*( d1++ )=( x&0x001F )|( ( x&0x7FE0 )<<1 );
}
#endif
}
|
