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
|
#include "ref.h"
void ref_cmplx_dot_prod_f32(
float32_t * pSrcA,
float32_t * pSrcB,
uint32_t numSamples,
float32_t * realResult,
float32_t * imagResult)
{
float32_t sumr, sumi;
uint32_t i;
sumr = 0;
sumi = 0;
for(i=0;i<numSamples*2;i+=2)
{
sumr += pSrcA[i] * pSrcB[i] - pSrcA[i+1] * pSrcB[i+1];
sumi += pSrcA[i] * pSrcB[i+1] + pSrcA[i+1] * pSrcB[i];
}
*realResult = sumr;
*imagResult = sumi;
}
void ref_cmplx_dot_prod_q31(
q31_t * pSrcA,
q31_t * pSrcB,
uint32_t numSamples,
q63_t * realResult,
q63_t * imagResult)
{
q63_t sumr, sumi;
uint32_t i;
sumr = 0;
sumi = 0;
for(i=0;i<numSamples*2;i+=2)
{
//shifting down 14 here to provide guard bits
sumr += (((q63_t)pSrcA[i] * pSrcB[i] ) >> 14) - (((q63_t)pSrcA[i+1] * pSrcB[i+1]) >> 14);
sumi += (((q63_t)pSrcA[i] * pSrcB[i+1]) >> 14) + (((q63_t)pSrcA[i+1] * pSrcB[i] ) >> 14);
}
*realResult = sumr;
*imagResult = sumi;
}
void ref_cmplx_dot_prod_q15(
q15_t * pSrcA,
q15_t * pSrcB,
uint32_t numSamples,
q31_t * realResult,
q31_t * imagResult)
{
q63_t sumr, sumi;
uint32_t i;
sumr = 0;
sumi = 0;
for(i=0;i<numSamples*2;i+=2)
{
sumr += (q31_t)pSrcA[i] * pSrcB[i] - (q31_t)pSrcA[i+1] * pSrcB[i+1];
sumi += (q31_t)pSrcA[i] * pSrcB[i+1] + (q31_t)pSrcA[i+1] * pSrcB[i];
}
//shifting down 6 at the end here because there are already 32 guard bits available, this method is more accurate
*realResult = (q31_t)(sumr >> 6);
*imagResult = (q31_t)(sumi >> 6);
}
|
