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
|
#include "main.h"
#include <math.h>
#pragma intrinsic(fabs)
/*
#ifndef _WIN64
__inline static int lrint(double flt)
{
int intgr;
_asm
{
fld flt
fistp intgr
}
return intgr;
}
#else
__inline static int lrint(double flt)
{
return (int)flt;
}
#endif
*/
#define PA_CLIP_( val, min, max )\
{ val = ((val) < (min)) ? (min) : (((val) > (max)) ? (max) : (val)); }
void Float32_To_Int16_Clip(
void *destinationBuffer, signed int destinationStride,
void *sourceBuffer, signed int sourceStride,
unsigned int count)
{
float *src = (float*)sourceBuffer;
signed short *dest = (signed short*)destinationBuffer;
while( count-- )
{
long samp = lrint((*src * (32768.0)));
PA_CLIP_( samp, -0x8000, 0x7FFF );
*dest = (signed short) samp;
src += sourceStride;
dest += destinationStride;
}
}
inline static double clip(double x, double a, double b)
{
double x1 = fabs (x-a);
double x2 = fabs (x-b);
x = x1 + (a+b);
x -= x2;
x *= 0.5;
return x;
}
/*
benski> this might be faster than what the compiler spits out for the above function,
but we should benchmark
inline static double clip(double x, double a, double b)
{
const double zero_point_five = 0.5;
__asm
{
fld x
fld a
fld b
fld st(2)
fsub st(0),st(2) // x-b
fabs
fadd st(0),st(2)
fadd st(0),st(1)
fld st(3)
fsub st(0), st(2)
fabs
fsubp st(1), st(0)
fmul zero_point_five
ffree st(4)
ffree st(3)
ffree st(2)
ffree st(1)
}
}
*/
void Float32_To_Int24_Clip(
void *destinationBuffer, signed int destinationStride,
void *sourceBuffer, signed int sourceStride,
unsigned int count)
{
float *src = (float*)sourceBuffer;
unsigned char *dest = (unsigned char*)destinationBuffer;
while( count-- )
{
/* convert to 32 bit and drop the low 8 bits */
double scaled = *src * 0x7FFFFFFF;
scaled=clip( scaled, -2147483648., 2147483647. );
signed long temp = (signed long) scaled;
dest[0] = (unsigned char)(temp >> 8);
dest[1] = (unsigned char)(temp >> 16);
dest[2] = (unsigned char)(temp >> 24);
src += sourceStride;
dest += destinationStride * 3;
}
}
|
