/
fft_v_fftw.cc
217 lines (192 loc) · 6.82 KB
/
fft_v_fftw.cc
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
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
/* -*- c++ -*- */
/*
* Copyright 2004,2007,2008,2010,2012,2020 Free Software Foundation, Inc.
*
* This file is part of GNU Radio
*
* SPDX-License-Identifier: GPL-3.0-or-later
*
*/
#ifdef HAVE_CONFIG_H
#include "config.h"
#endif
#include "fft_v_fftw.h"
#include <volk/volk.h>
#include <cmath>
#include <cstring>
namespace gr {
namespace fft {
template <class T, bool forward>
typename fft_v<T, forward>::sptr fft_v<T, forward>::make(int fft_size,
const std::vector<float>& window,
bool shift,
int nthreads)
{
return gnuradio::get_initial_sptr(
new fft_v_fftw<T, forward>(fft_size, window, shift, nthreads));
}
template <class T, bool forward>
fft_v_fftw<T, forward>::fft_v_fftw(int fft_size,
const std::vector<float>& window,
bool shift,
int nthreads)
: sync_block("fft_v_fftw",
io_signature::make(1, 1, fft_size * sizeof(T)),
io_signature::make(1, 1, fft_size * sizeof(gr_complex))),
d_fft_size(fft_size),
d_fft(fft_size, nthreads),
d_shift(shift)
{
if (!set_window(window))
throw std::runtime_error("fft_v: window not the same length as fft_size");
}
template <class T, bool forward>
void fft_v_fftw<T, forward>::set_nthreads(int n)
{
d_fft.set_nthreads(n);
}
template <class T, bool forward>
int fft_v_fftw<T, forward>::nthreads() const
{
return d_fft.nthreads();
}
template <class T, bool forward>
bool fft_v_fftw<T, forward>::set_window(const std::vector<float>& window)
{
if (window.empty() || window.size() == d_fft_size) {
d_window = window;
return true;
} else
return false;
}
template <>
void fft_v_fftw<gr_complex, true>::fft_and_shift(const gr_complex* in, gr_complex* out)
{
if (!d_window.empty()) {
gr_complex* dst = d_fft.get_inbuf();
volk_32fc_32f_multiply_32fc(&dst[0], in, &d_window[0], d_fft_size);
} else {
memcpy(d_fft.get_inbuf(), in, sizeof(gr_complex) * d_fft_size);
}
d_fft.execute();
if (d_shift) {
unsigned int len = (unsigned int)(ceil(d_fft_size / 2.0));
memcpy(
&out[0], &d_fft.get_outbuf()[len], sizeof(gr_complex) * (d_fft_size - len));
memcpy(&out[d_fft_size - len], &d_fft.get_outbuf()[0], sizeof(gr_complex) * len);
} else {
memcpy(out, d_fft.get_outbuf(), sizeof(gr_complex) * d_fft_size);
}
}
template <>
void fft_v_fftw<gr_complex, false>::fft_and_shift(const gr_complex* in, gr_complex* out)
{
if (!d_window.empty()) {
gr_complex* dst = d_fft.get_inbuf();
if (d_shift) {
unsigned int offset = d_fft_size / 2;
int fft_m_offset = d_fft_size - offset;
volk_32fc_32f_multiply_32fc(&dst[fft_m_offset], &in[0], &d_window[0], offset);
volk_32fc_32f_multiply_32fc(
&dst[0], &in[offset], &d_window[offset], d_fft_size - offset);
} else {
volk_32fc_32f_multiply_32fc(&dst[0], in, &d_window[0], d_fft_size);
}
} else {
if (d_shift) { // apply an ifft shift on the data
gr_complex* dst = d_fft.get_inbuf();
unsigned int len =
(unsigned int)(floor(d_fft_size / 2.0)); // half length of complex array
memcpy(&dst[0], &in[len], sizeof(gr_complex) * (d_fft_size - len));
memcpy(&dst[d_fft_size - len], &in[0], sizeof(gr_complex) * len);
} else {
memcpy(d_fft.get_inbuf(), in, sizeof(gr_complex) * d_fft_size);
}
}
d_fft.execute();
memcpy(out, d_fft.get_outbuf(), sizeof(gr_complex) * d_fft_size);
}
template <>
void fft_v_fftw<float, true>::fft_and_shift(const float* in, gr_complex* out)
{
// copy input into optimally aligned buffer
if (!d_window.empty()) {
gr_complex* dst = d_fft.get_inbuf();
for (unsigned int i = 0; i < d_fft_size; i++) // apply window
dst[i] = in[i] * d_window[i];
} else {
gr_complex* dst = d_fft.get_inbuf();
for (unsigned int i = 0; i < d_fft_size; i++) // float to complex conversion
dst[i] = in[i];
}
d_fft.execute();
if (d_shift) {
unsigned int len = (unsigned int)(ceil(d_fft_size / 2.0));
memcpy(
&out[0], &d_fft.get_outbuf()[len], sizeof(gr_complex) * (d_fft_size - len));
memcpy(&out[d_fft_size - len], &d_fft.get_outbuf()[0], sizeof(gr_complex) * len);
} else {
memcpy(out, d_fft.get_outbuf(), sizeof(gr_complex) * d_fft_size);
}
}
template <>
void fft_v_fftw<float, false>::fft_and_shift(const float* in, gr_complex* out)
{
// copy input into optimally aligned buffer
if (!d_window.empty()) {
gr_complex* dst = d_fft.get_inbuf();
if (d_shift) {
unsigned int len =
(unsigned int)(floor(d_fft_size / 2.0)); // half length of complex array
for (unsigned int i = 0; i < len; i++) {
dst[i] = in[len + i] * d_window[len + i];
}
for (unsigned int i = len; i < d_fft_size; i++) {
dst[i] = in[i - len] * d_window[i - len];
}
} else {
for (unsigned int i = 0; i < d_fft_size; i++) // apply window
dst[i] = in[i] * d_window[i];
}
} else {
gr_complex* dst = d_fft.get_inbuf();
if (d_shift) {
unsigned int len =
(unsigned int)(floor(d_fft_size / 2.0)); // half length of complex array
for (unsigned int i = 0; i < len; i++) {
dst[i] = in[len + i];
}
for (unsigned int i = len; i < d_fft_size; i++) {
dst[i] = in[i - len];
}
} else {
for (unsigned int i = 0; i < d_fft_size; i++) // float to complex conversion
dst[i] = in[i];
}
}
// compute the fft
d_fft.execute();
// copy result to output stream
memcpy(out, d_fft.get_outbuf(), sizeof(gr_complex) * d_fft_size);
}
template <class T, bool forward>
int fft_v_fftw<T, forward>::work(int noutput_items,
gr_vector_const_void_star& input_items,
gr_vector_void_star& output_items)
{
auto in = reinterpret_cast<const T*>(input_items[0]);
auto out = reinterpret_cast<gr_complex*>(output_items[0]);
int count = 0;
while (count++ < noutput_items) {
fft_and_shift(in, out);
in += d_fft_size;
out += d_fft_size;
}
return noutput_items;
}
template class fft_v<gr_complex, true>;
template class fft_v<gr_complex, false>;
template class fft_v<float, true>;
template class fft_v<float, false>;
} /* namespace fft */
} /* namespace gr */