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
|
#ifndef _OP_BLEND_ARM_H_
#define _OP_BLEND_ARM_H_
////////////////////////////////////////////////////////////////////////////////
// Blend bgr555 color in 'uSrc' (foreground) with bgr555 color
// in 'uDst' (background), returning resulting color.
//
// INPUT:
// 'uSrc','uDst' input: -bbbbbgggggrrrrr
// ^ bit 16
// OUTPUT:
// u16 output: 0bbbbbgggggrrrrr
// ^ bit 16
// RETURNS:
// Where '0' is zero-padding, and '-' is don't care
////////////////////////////////////////////////////////////////////////////////
template <int BLENDMODE, bool SKIP_USRC_MSB_MASK>
GPU_INLINE uint_fast16_t gpuBlendingARM(uint_fast16_t uSrc, uint_fast16_t uDst)
{
// These use Blargg's bitwise modulo-clamping:
// http://blargg.8bitalley.com/info/rgb_mixing.html
// http://blargg.8bitalley.com/info/rgb_clamped_add.html
// http://blargg.8bitalley.com/info/rgb_clamped_sub.html
uint_fast16_t mix;
asm ("bic %[uDst], %[uDst], #0x8000" : [uDst] "+r" (uDst));
if (BLENDMODE == 3) {
asm ("and %[uSrc], %[mask], %[uSrc], lsr #0x2" : [uSrc] "+r" (uSrc) : [mask] "r" (0x1ce7));
} else if (!SKIP_USRC_MSB_MASK) {
asm ("bic %[uSrc], %[uSrc], #0x8000" : [uSrc] "+r" (uSrc));
}
// 0.5 x Back + 0.5 x Forward
if (BLENDMODE==0) {
// mix = ((uSrc + uDst) - ((uSrc ^ uDst) & 0x0421)) >> 1;
asm ("eor %[mix], %[uSrc], %[uDst]\n\t"
"and %[mix], %[mix], %[mask]\n\t"
"sub %[mix], %[uDst], %[mix]\n\t"
"add %[mix], %[uSrc], %[mix]\n\t"
"mov %[mix], %[mix], lsr #0x1\n\t"
: [mix] "=&r" (mix)
: [uSrc] "r" (uSrc), [uDst] "r" (uDst), [mask] "r" (0x0421));
}
if (BLENDMODE == 1 || BLENDMODE == 3) {
// u32 sum = uSrc + uDst;
// u32 low_bits = (uSrc ^ uDst) & 0x0421;
// u32 carries = (sum - low_bits) & 0x8420;
// u32 modulo = sum - carries;
// u32 clamp = carries - (carries >> 5);
// mix = modulo | clamp;
u32 sum;
asm ("add %[sum], %[uSrc], %[uDst]\n\t"
"eor %[mix], %[uSrc], %[uDst]\n\t"
"and %[mix], %[mix], %[mask]\n\t"
"sub %[mix], %[sum], %[mix]\n\t"
"and %[mix], %[mix], %[mask], lsl #0x05\n\t"
"sub %[sum], %[sum], %[mix] \n\t"
"sub %[mix], %[mix], %[mix], lsr #0x05\n\t"
"orr %[mix], %[sum], %[mix]"
: [sum] "=&r" (sum), [mix] "=&r" (mix)
: [uSrc] "r" (uSrc), [uDst] "r" (uDst), [mask] "r" (0x0421));
}
// 1.0 x Back - 1.0 x Forward
if (BLENDMODE==2) {
u32 diff;
// u32 diff = uDst - uSrc + 0x8420;
// u32 low_bits = (uDst ^ uSrc) & 0x8420;
// u32 borrows = (diff - low_bits) & 0x8420;
// u32 modulo = diff - borrows;
// u32 clamp = borrows - (borrows >> 5);
// mix = modulo & clamp;
asm ("sub %[diff], %[uDst], %[uSrc]\n\t"
"add %[diff], %[diff], %[mask]\n\t"
"eor %[mix], %[uDst], %[uSrc]\n\t"
"and %[mix], %[mix], %[mask]\n\t"
"sub %[mix], %[diff], %[mix]\n\t"
"and %[mix], %[mix], %[mask]\n\t"
"sub %[diff], %[diff], %[mix]\n\t"
"sub %[mix], %[mix], %[mix], lsr #0x05\n\t"
"and %[mix], %[diff], %[mix]"
: [diff] "=&r" (diff), [mix] "=&r" (mix)
: [uSrc] "r" (uSrc), [uDst] "r" (uDst), [mask] "r" (0x8420));
}
// There's not a case where we can get into this function,
// SKIP_USRC_MSB_MASK is false, and the msb of uSrc is unset.
if (!SKIP_USRC_MSB_MASK) {
asm ("orr %[mix], %[mix], #0x8000" : [mix] "+r" (mix));
}
return mix;
}
#endif //_OP_BLEND_ARM_H_
|
