GNU Radio 3.6.4.1 C++ API
volk_32fc_s32fc_multiply_32fc_u.h
Go to the documentation of this file.
1 #ifndef INCLUDED_volk_32fc_s32fc_multiply_32fc_u_H
2 #define INCLUDED_volk_32fc_s32fc_multiply_32fc_u_H
3 
4 #include <inttypes.h>
5 #include <stdio.h>
6 #include <volk/volk_complex.h>
7 #include <float.h>
8 
9 #ifdef LV_HAVE_SSE3
10 #include <pmmintrin.h>
11 /*!
12  \brief Multiplies the input vector by a scalar and stores the results in the third vector
13  \param cVector The vector where the results will be stored
14  \param aVector The vector to be multiplied
15  \param scalar The complex scalar to multiply aVector
16  \param num_points The number of complex values in aVector and bVector to be multiplied together and stored into cVector
17 */
18 static inline void volk_32fc_s32fc_multiply_32fc_u_sse3(lv_32fc_t* cVector, const lv_32fc_t* aVector, const lv_32fc_t scalar, unsigned int num_points){
19  unsigned int number = 0;
20  const unsigned int halfPoints = num_points / 2;
21 
22  __m128 x, yl, yh, z, tmp1, tmp2;
23  lv_32fc_t* c = cVector;
24  const lv_32fc_t* a = aVector;
25 
26  // Set up constant scalar vector
27  yl = _mm_set_ps1(lv_creal(scalar));
28  yh = _mm_set_ps1(lv_cimag(scalar));
29 
30  for(;number < halfPoints; number++){
31 
32  x = _mm_loadu_ps((float*)a); // Load the ar + ai, br + bi as ar,ai,br,bi
33 
34  tmp1 = _mm_mul_ps(x,yl); // tmp1 = ar*cr,ai*cr,br*dr,bi*dr
35 
36  x = _mm_shuffle_ps(x,x,0xB1); // Re-arrange x to be ai,ar,bi,br
37 
38  tmp2 = _mm_mul_ps(x,yh); // tmp2 = ai*ci,ar*ci,bi*di,br*di
39 
40  z = _mm_addsub_ps(tmp1,tmp2); // ar*cr-ai*ci, ai*cr+ar*ci, br*dr-bi*di, bi*dr+br*di
41 
42  _mm_storeu_ps((float*)c,z); // Store the results back into the C container
43 
44  a += 2;
45  c += 2;
46  }
47 
48  if((num_points % 2) != 0) {
49  *c = (*a) * scalar;
50  }
51 }
52 #endif /* LV_HAVE_SSE */
53 
54 #ifdef LV_HAVE_GENERIC
55 /*!
56  \brief Multiplies the input vector by a scalar and stores the results in the third vector
57  \param cVector The vector where the results will be stored
58  \param aVector The vector to be multiplied
59  \param scalar The complex scalar to multiply aVector
60  \param num_points The number of complex values in aVector and bVector to be multiplied together and stored into cVector
61 */
62 static inline void volk_32fc_s32fc_multiply_32fc_u_generic(lv_32fc_t* cVector, const lv_32fc_t* aVector, const lv_32fc_t scalar, unsigned int num_points){
63  lv_32fc_t* cPtr = cVector;
64  const lv_32fc_t* aPtr = aVector;
65  unsigned int number = num_points;
66 
67  // unwrap loop
68  while (number >= 8){
69  *cPtr++ = (*aPtr++) * scalar;
70  *cPtr++ = (*aPtr++) * scalar;
71  *cPtr++ = (*aPtr++) * scalar;
72  *cPtr++ = (*aPtr++) * scalar;
73  *cPtr++ = (*aPtr++) * scalar;
74  *cPtr++ = (*aPtr++) * scalar;
75  *cPtr++ = (*aPtr++) * scalar;
76  *cPtr++ = (*aPtr++) * scalar;
77  number -= 8;
78  }
79 
80  // clean up any remaining
81  while (number-- > 0)
82  *cPtr++ = *aPtr++ * scalar;
83 }
84 #endif /* LV_HAVE_GENERIC */
85 
86 
87 #endif /* INCLUDED_volk_32fc_x2_multiply_32fc_u_H */