Details | Last modification | View Log | RSS feed
| Rev | Author | Line No. | Line |
|---|---|---|---|
| 2 | pj | 1 | |
| 2 | /* |
||
| 3 | * Copyright (c) 1997-1999 Massachusetts Institute of Technology |
||
| 4 | * |
||
| 5 | * This program is free software; you can redistribute it and/or modify |
||
| 6 | * it under the terms of the GNU General Public License as published by |
||
| 7 | * the Free Software Foundation; either version 2 of the License, or |
||
| 8 | * (at your option) any later version. |
||
| 9 | * |
||
| 10 | * This program is distributed in the hope that it will be useful, |
||
| 11 | * but WITHOUT ANY WARRANTY; without even the implied warranty of |
||
| 12 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the |
||
| 13 | * GNU General Public License for more details. |
||
| 14 | * |
||
| 15 | * You should have received a copy of the GNU General Public License |
||
| 16 | * along with this program; if not, write to the Free Software |
||
| 17 | * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA |
||
| 18 | * |
||
| 19 | */ |
||
| 20 | |||
| 21 | /* |
||
| 22 | * rgeneric.c -- "generic" rfftw codelets. They work for all n (and |
||
| 23 | * they are slow) |
||
| 24 | */ |
||
| 25 | #include <ports/fftw-int.h> |
||
| 26 | #include <ports/rfftw.h> |
||
| 27 | |||
| 28 | /* this code assumes that r and m are both odd */ |
||
| 29 | void fftw_hc2hc_forward_generic(fftw_real *A, const fftw_complex *W, |
||
| 30 | int m, int r, int n, int dist) |
||
| 31 | { |
||
| 32 | int i, j, k; |
||
| 33 | fftw_complex *tmp = (fftw_complex *) |
||
| 34 | fftw_malloc(r * sizeof(fftw_complex)); |
||
| 35 | fftw_real rsum, isum; |
||
| 36 | fftw_real *X, *YO, *YI; |
||
| 37 | int wp, wincr; |
||
| 38 | int iostride = m * dist; |
||
| 39 | X = A; |
||
| 40 | YO = A + r * iostride; |
||
| 41 | YI = A + iostride; |
||
| 42 | |||
| 43 | /* compute the transform of the r 0th elements (which are real) */ |
||
| 44 | for (i = 0; i + i < r; ++i) { |
||
| 45 | rsum = 0.0; |
||
| 46 | isum = 0.0; |
||
| 47 | wincr = m * i; |
||
| 48 | for (j = 0, wp = 0; j < r; ++j) { |
||
| 49 | fftw_real tw_r = c_re(W[wp]); |
||
| 50 | fftw_real tw_i = c_im(W[wp]); |
||
| 51 | fftw_real re = X[j * iostride]; |
||
| 52 | rsum += re * tw_r; |
||
| 53 | isum += re * tw_i; |
||
| 54 | wp += wincr; |
||
| 55 | if (wp >= n) |
||
| 56 | wp -= n; |
||
| 57 | } |
||
| 58 | c_re(tmp[i]) = rsum; |
||
| 59 | c_im(tmp[i]) = isum; |
||
| 60 | } |
||
| 61 | |||
| 62 | /* store the transform back onto the A array */ |
||
| 63 | X[0] = c_re(tmp[0]); |
||
| 64 | for (i = 1; i + i < r; ++i) { |
||
| 65 | X[i * iostride] = c_re(tmp[i]); |
||
| 66 | YO[-i * iostride] = c_im(tmp[i]); |
||
| 67 | } |
||
| 68 | |||
| 69 | X += dist; |
||
| 70 | YI -= dist; |
||
| 71 | YO -= dist; |
||
| 72 | |||
| 73 | /* compute the transform of the middle elements (which are complex) */ |
||
| 74 | for (k = 1; k + k < m; ++k, X += dist, YI -= dist, YO -= dist) { |
||
| 75 | for (i = 0; i < r; ++i) { |
||
| 76 | rsum = 0.0; |
||
| 77 | isum = 0.0; |
||
| 78 | wincr = k + m * i; |
||
| 79 | for (j = 0, wp = 0; j < r; ++j) { |
||
| 80 | fftw_real tw_r = c_re(W[wp]); |
||
| 81 | fftw_real tw_i = c_im(W[wp]); |
||
| 82 | fftw_real re = X[j * iostride]; |
||
| 83 | fftw_real im = YI[j * iostride]; |
||
| 84 | rsum += re * tw_r - im * tw_i; |
||
| 85 | isum += re * tw_i + im * tw_r; |
||
| 86 | wp += wincr; |
||
| 87 | if (wp >= n) |
||
| 88 | wp -= n; |
||
| 89 | } |
||
| 90 | c_re(tmp[i]) = rsum; |
||
| 91 | c_im(tmp[i]) = isum; |
||
| 92 | } |
||
| 93 | |||
| 94 | /* store the transform back onto the A array */ |
||
| 95 | for (i = 0; i + i < r; ++i) { |
||
| 96 | X[i * iostride] = c_re(tmp[i]); |
||
| 97 | YO[-i * iostride] = c_im(tmp[i]); |
||
| 98 | } |
||
| 99 | for (; i < r; ++i) { |
||
| 100 | X[i * iostride] = -c_im(tmp[i]); |
||
| 101 | YO[-i * iostride] = c_re(tmp[i]); |
||
| 102 | } |
||
| 103 | } |
||
| 104 | |||
| 105 | /* no final element, since m is odd */ |
||
| 106 | fftw_free(tmp); |
||
| 107 | } |
||
| 108 | |||
| 109 | void fftw_hc2hc_backward_generic(fftw_real *A, const fftw_complex *W, |
||
| 110 | int m, int r, int n, int dist) |
||
| 111 | { |
||
| 112 | int i, j, k; |
||
| 113 | int wp, wincr; |
||
| 114 | fftw_complex *tmp = (fftw_complex *) |
||
| 115 | fftw_malloc(r * sizeof(fftw_complex)); |
||
| 116 | fftw_real rsum, isum; |
||
| 117 | fftw_real *X, *YO, *YI; |
||
| 118 | int iostride = m * dist; |
||
| 119 | X = A; |
||
| 120 | YO = A + iostride; |
||
| 121 | YI = A + r * iostride; |
||
| 122 | |||
| 123 | /* |
||
| 124 | * compute the transform of the r 0th elements (which are halfcomplex) |
||
| 125 | * yielding real numbers |
||
| 126 | */ |
||
| 127 | /* copy the input into the temporary array */ |
||
| 128 | c_re(tmp[0]) = X[0]; |
||
| 129 | for (i = 1; i + i < r; ++i) { |
||
| 130 | c_re(tmp[i]) = X[i * iostride]; |
||
| 131 | c_im(tmp[i]) = YI[-i * iostride]; |
||
| 132 | } |
||
| 133 | |||
| 134 | for (i = 0; i < r; ++i) { |
||
| 135 | rsum = 0.0; |
||
| 136 | wincr = m * i; |
||
| 137 | for (j = 1, wp = wincr; j + j < r; ++j) { |
||
| 138 | fftw_real tw_r = c_re(W[wp]); |
||
| 139 | fftw_real tw_i = c_im(W[wp]); |
||
| 140 | fftw_real re = c_re(tmp[j]); |
||
| 141 | fftw_real im = c_im(tmp[j]); |
||
| 142 | rsum += re * tw_r + im * tw_i; |
||
| 143 | wp += wincr; |
||
| 144 | if (wp >= n) |
||
| 145 | wp -= n; |
||
| 146 | } |
||
| 147 | X[i * iostride] = 2.0 * rsum + c_re(tmp[0]); |
||
| 148 | } |
||
| 149 | |||
| 150 | X += dist; |
||
| 151 | YI -= dist; |
||
| 152 | YO -= dist; |
||
| 153 | |||
| 154 | /* compute the transform of the middle elements (which are complex) */ |
||
| 155 | for (k = 1; k + k < m; ++k, X += dist, YI -= dist, YO -= dist) { |
||
| 156 | /* copy the input into the temporary array */ |
||
| 157 | for (i = 0; i + i < r; ++i) { |
||
| 158 | c_re(tmp[i]) = X[i * iostride]; |
||
| 159 | c_im(tmp[i]) = YI[-i * iostride]; |
||
| 160 | } |
||
| 161 | for (; i < r; ++i) { |
||
| 162 | c_im(tmp[i]) = -X[i * iostride]; |
||
| 163 | c_re(tmp[i]) = YI[-i * iostride]; |
||
| 164 | } |
||
| 165 | |||
| 166 | for (i = 0; i < r; ++i) { |
||
| 167 | rsum = 0.0; |
||
| 168 | isum = 0.0; |
||
| 169 | wincr = m * i; |
||
| 170 | for (j = 0, wp = k * i; j < r; ++j) { |
||
| 171 | fftw_real tw_r = c_re(W[wp]); |
||
| 172 | fftw_real tw_i = c_im(W[wp]); |
||
| 173 | fftw_real re = c_re(tmp[j]); |
||
| 174 | fftw_real im = c_im(tmp[j]); |
||
| 175 | rsum += re * tw_r + im * tw_i; |
||
| 176 | isum += im * tw_r - re * tw_i; |
||
| 177 | wp += wincr; |
||
| 178 | if (wp >= n) |
||
| 179 | wp -= n; |
||
| 180 | } |
||
| 181 | X[i * iostride] = rsum; |
||
| 182 | YO[i * iostride] = isum; |
||
| 183 | } |
||
| 184 | } |
||
| 185 | |||
| 186 | /* no final element, since m is odd */ |
||
| 187 | fftw_free(tmp); |
||
| 188 | } |