Vector Optimized Library of Kernels  3.0.0
Architecture-tuned implementations of math kernels
volk_32f_8u_polarbutterflypuppet_32f.h
Go to the documentation of this file.
1 /* -*- c++ -*- */
2 /*
3  * Copyright 2015 Free Software Foundation, Inc.
4  *
5  * This file is part of VOLK
6  *
7  * SPDX-License-Identifier: LGPL-3.0-or-later
8  */
9 
10 /*
11  * This puppet is for VOLK tests only.
12  * For documentation see 'kernels/volk/volk_32f_8u_polarbutterfly_32f.h'
13  */
14 
15 #ifndef VOLK_KERNELS_VOLK_VOLK_32F_8U_POLARBUTTERFLYPUPPET_32F_H_
16 #define VOLK_KERNELS_VOLK_VOLK_32F_8U_POLARBUTTERFLYPUPPET_32F_H_
17 
21 
22 
23 static inline void sanitize_bytes(unsigned char* u, const int elements)
24 {
25  int i;
26  unsigned char* u_ptr = u;
27  for (i = 0; i < elements; i++) {
28  *u_ptr = (*u_ptr & 0x01);
29  u_ptr++;
30  }
31 }
32 
33 static inline void clean_up_intermediate_values(float* llrs,
34  unsigned char* u,
35  const int frame_size,
36  const int elements)
37 {
38  memset(u + frame_size, 0, sizeof(unsigned char) * (elements - frame_size));
39  memset(llrs + frame_size, 0, sizeof(float) * (elements - frame_size));
40 }
41 
42 static inline void
43 generate_error_free_input_vector(float* llrs, unsigned char* u, const int frame_size)
44 {
45  memset(u, 0, frame_size);
46  unsigned char* target = u + frame_size;
47  volk_8u_x2_encodeframepolar_8u_generic(target, u + 2 * frame_size, frame_size);
48  float* ft = llrs;
49  int i;
50  for (i = 0; i < frame_size; i++) {
51  *ft = (-2 * ((float)*target++)) + 1.0f;
52  ft++;
53  }
54 }
55 
56 static inline void
57 print_llr_tree(const float* llrs, const int frame_size, const int frame_exp)
58 {
59  int s, e;
60  for (s = 0; s < frame_size; s++) {
61  for (e = 0; e < frame_exp + 1; e++) {
62  printf("%+4.2f ", llrs[e * frame_size + s]);
63  }
64  printf("\n");
65  if ((s + 1) % 8 == 0) {
66  printf("\n");
67  }
68  }
69 }
70 
71 static inline int maximum_frame_size(const int elements)
72 {
73  unsigned int frame_size = next_lower_power_of_two(elements);
74  unsigned int frame_exp = log2_of_power_of_2(frame_size);
75  return next_lower_power_of_two(frame_size / frame_exp);
76 }
77 
78 #ifdef LV_HAVE_GENERIC
79 static inline void volk_32f_8u_polarbutterflypuppet_32f_generic(float* llrs,
80  const float* input,
81  unsigned char* u,
82  const int elements)
83 {
84  unsigned int frame_size = maximum_frame_size(elements);
85  unsigned int frame_exp = log2_of_power_of_2(frame_size);
86 
87  sanitize_bytes(u, elements);
88  clean_up_intermediate_values(llrs, u, frame_size, elements);
89  generate_error_free_input_vector(llrs + frame_exp * frame_size, u, frame_size);
90 
91  unsigned int u_num = 0;
92  for (; u_num < frame_size; u_num++) {
93  volk_32f_8u_polarbutterfly_32f_generic(llrs, u, frame_exp, 0, u_num, u_num);
94  u[u_num] = llrs[u_num] > 0 ? 0 : 1;
95  }
96 
97  clean_up_intermediate_values(llrs, u, frame_size, elements);
98 }
99 #endif /* LV_HAVE_GENERIC */
100 
101 #ifdef LV_HAVE_AVX
102 static inline void volk_32f_8u_polarbutterflypuppet_32f_u_avx(float* llrs,
103  const float* input,
104  unsigned char* u,
105  const int elements)
106 {
107  unsigned int frame_size = maximum_frame_size(elements);
108  unsigned int frame_exp = log2_of_power_of_2(frame_size);
109 
110  sanitize_bytes(u, elements);
111  clean_up_intermediate_values(llrs, u, frame_size, elements);
112  generate_error_free_input_vector(llrs + frame_exp * frame_size, u, frame_size);
113 
114  unsigned int u_num = 0;
115  for (; u_num < frame_size; u_num++) {
116  volk_32f_8u_polarbutterfly_32f_u_avx(llrs, u, frame_exp, 0, u_num, u_num);
117  u[u_num] = llrs[u_num] > 0 ? 0 : 1;
118  }
119 
120  clean_up_intermediate_values(llrs, u, frame_size, elements);
121 }
122 #endif /* LV_HAVE_AVX */
123 
124 #ifdef LV_HAVE_AVX2
125 static inline void volk_32f_8u_polarbutterflypuppet_32f_u_avx2(float* llrs,
126  const float* input,
127  unsigned char* u,
128  const int elements)
129 {
130  unsigned int frame_size = maximum_frame_size(elements);
131  unsigned int frame_exp = log2_of_power_of_2(frame_size);
132 
133  sanitize_bytes(u, elements);
134  clean_up_intermediate_values(llrs, u, frame_size, elements);
135  generate_error_free_input_vector(llrs + frame_exp * frame_size, u, frame_size);
136 
137  unsigned int u_num = 0;
138  for (; u_num < frame_size; u_num++) {
139  volk_32f_8u_polarbutterfly_32f_u_avx2(llrs, u, frame_exp, 0, u_num, u_num);
140  u[u_num] = llrs[u_num] > 0 ? 0 : 1;
141  }
142 
143  clean_up_intermediate_values(llrs, u, frame_size, elements);
144 }
145 #endif /* LV_HAVE_AVX2 */
146 
147 
148 #endif /* VOLK_KERNELS_VOLK_VOLK_32F_8U_POLARBUTTERFLYPUPPET_32F_H_ */
static void volk_32f_8u_polarbutterfly_32f_generic(float *llrs, unsigned char *u, const int frame_exp, const int stage, const int u_num, const int row)
Definition: volk_32f_8u_polarbutterfly_32f.h:144
static void volk_32f_8u_polarbutterfly_32f_u_avx(float *llrs, unsigned char *u, const int frame_exp, const int stage, const int u_num, const int row)
Definition: volk_32f_8u_polarbutterfly_32f.h:200
static void clean_up_intermediate_values(float *llrs, unsigned char *u, const int frame_size, const int elements)
Definition: volk_32f_8u_polarbutterflypuppet_32f.h:33
static void print_llr_tree(const float *llrs, const int frame_size, const int frame_exp)
Definition: volk_32f_8u_polarbutterflypuppet_32f.h:57
static void volk_32f_8u_polarbutterflypuppet_32f_generic(float *llrs, const float *input, unsigned char *u, const int elements)
Definition: volk_32f_8u_polarbutterflypuppet_32f.h:79
static int maximum_frame_size(const int elements)
Definition: volk_32f_8u_polarbutterflypuppet_32f.h:71
static void generate_error_free_input_vector(float *llrs, unsigned char *u, const int frame_size)
Definition: volk_32f_8u_polarbutterflypuppet_32f.h:43
static void sanitize_bytes(unsigned char *u, const int elements)
Definition: volk_32f_8u_polarbutterflypuppet_32f.h:23
static void volk_32f_8u_polarbutterflypuppet_32f_u_avx(float *llrs, const float *input, unsigned char *u, const int elements)
Definition: volk_32f_8u_polarbutterflypuppet_32f.h:102
static void volk_8u_x2_encodeframepolar_8u_generic(unsigned char *frame, unsigned char *temp, unsigned int frame_size)
Definition: volk_8u_x2_encodeframepolar_8u.h:52
static unsigned int log2_of_power_of_2(unsigned int val)
Definition: volk_8u_x2_encodeframepolar_8u.h:18
static unsigned int next_lower_power_of_two(const unsigned int val)
Definition: volk_8u_x3_encodepolarpuppet_8u.h:19
for i
Definition: volk_config_fixed.tmpl.h:13