Add 64-bit AArch64 asimd SIMD support
[gromacs.git] / src / gromacs / fft / fft.c
blob26ac29acbe2da25a871ebc14d53c0a6d13c14972
1 /*
2 * This file is part of the GROMACS molecular simulation package.
4 * Copyright (c) 1991-2003 Erik Lindahl, David van der Spoel, University of Groningen.
5 * Copyright (c) 2013,2014, by the GROMACS development team, led by
6 * Mark Abraham, David van der Spoel, Berk Hess, and Erik Lindahl,
7 * and including many others, as listed in the AUTHORS file in the
8 * top-level source directory and at http://www.gromacs.org.
10 * GROMACS is free software; you can redistribute it and/or
11 * modify it under the terms of the GNU Lesser General Public License
12 * as published by the Free Software Foundation; either version 2.1
13 * of the License, or (at your option) any later version.
15 * GROMACS is distributed in the hope that it will be useful,
16 * but WITHOUT ANY WARRANTY; without even the implied warranty of
17 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
18 * Lesser General Public License for more details.
20 * You should have received a copy of the GNU Lesser General Public
21 * License along with GROMACS; if not, see
22 * http://www.gnu.org/licenses, or write to the Free Software Foundation,
23 * Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
25 * If you want to redistribute modifications to GROMACS, please
26 * consider that scientific software is very special. Version
27 * control is crucial - bugs must be traceable. We will be happy to
28 * consider code for inclusion in the official distribution, but
29 * derived work must not be called official GROMACS. Details are found
30 * in the README & COPYING files - if they are missing, get the
31 * official version at http://www.gromacs.org.
33 * To help us fund GROMACS development, we humbly ask that you cite
34 * the research papers on the package. Check out http://www.gromacs.org.
36 #include "gmxpre.h"
38 #include "gromacs/fft/fft.h"
40 #include "config.h"
42 #include <errno.h>
43 #include <stdio.h>
44 #include <stdlib.h>
45 #include <string.h>
47 #include "gromacs/math/gmxcomplex.h"
48 #include "gromacs/utility/fatalerror.h"
49 #include "gromacs/utility/real.h"
51 /* This file contains common fft utility functions, but not
52 * the actual transform implementations. Check the
53 * files like fft_fftw3.c or fft_mkl.c for that.
56 #ifndef GMX_FFT_FFTW3
58 struct gmx_many_fft {
59 int howmany;
60 int dist;
61 gmx_fft_t fft;
64 typedef struct gmx_many_fft* gmx_many_fft_t;
66 int
67 gmx_fft_init_many_1d(gmx_fft_t * pfft,
68 int nx,
69 int howmany,
70 gmx_fft_flag flags)
72 gmx_many_fft_t fft;
73 if (pfft == NULL)
75 gmx_fatal(FARGS, "Invalid opaque FFT datatype pointer.");
76 return EINVAL;
78 *pfft = NULL;
80 if ( (fft = (gmx_many_fft_t)malloc(sizeof(struct gmx_many_fft))) == NULL)
82 return ENOMEM;
85 gmx_fft_init_1d(&fft->fft, nx, flags);
86 fft->howmany = howmany;
87 fft->dist = 2*nx;
89 *pfft = (gmx_fft_t)fft;
90 return 0;
93 int
94 gmx_fft_init_many_1d_real(gmx_fft_t * pfft,
95 int nx,
96 int howmany,
97 gmx_fft_flag flags)
99 gmx_many_fft_t fft;
100 if (pfft == NULL)
102 gmx_fatal(FARGS, "Invalid opaque FFT datatype pointer.");
103 return EINVAL;
105 *pfft = NULL;
107 if ( (fft = (gmx_many_fft_t)malloc(sizeof(struct gmx_many_fft))) == NULL)
109 return ENOMEM;
112 gmx_fft_init_1d_real(&fft->fft, nx, flags);
113 fft->howmany = howmany;
114 fft->dist = 2*(nx/2+1);
116 *pfft = (gmx_fft_t)fft;
117 return 0;
121 gmx_fft_many_1d (gmx_fft_t fft,
122 enum gmx_fft_direction dir,
123 void * in_data,
124 void * out_data)
126 gmx_many_fft_t mfft = (gmx_many_fft_t)fft;
127 int i, ret;
128 for (i = 0; i < mfft->howmany; i++)
130 ret = gmx_fft_1d(mfft->fft, dir, in_data, out_data);
131 if (ret != 0)
133 return ret;
135 in_data = (real*)in_data+mfft->dist;
136 out_data = (real*)out_data+mfft->dist;
138 return 0;
142 gmx_fft_many_1d_real (gmx_fft_t fft,
143 enum gmx_fft_direction dir,
144 void * in_data,
145 void * out_data)
147 gmx_many_fft_t mfft = (gmx_many_fft_t)fft;
148 int i, ret;
149 for (i = 0; i < mfft->howmany; i++)
151 ret = gmx_fft_1d_real(mfft->fft, dir, in_data, out_data);
152 if (ret != 0)
154 return ret;
156 in_data = (real*)in_data+mfft->dist;
157 out_data = (real*)out_data+mfft->dist;
159 return 0;
163 void
164 gmx_many_fft_destroy(gmx_fft_t fft)
166 gmx_many_fft_t mfft = (gmx_many_fft_t)fft;
167 if (mfft != NULL)
169 if (mfft->fft != NULL)
171 gmx_fft_destroy(mfft->fft);
173 free(mfft);
177 #endif //not GMX_FFT_FFTW3
179 int gmx_fft_transpose_2d(t_complex * in_data,
180 t_complex * out_data,
181 int nx,
182 int ny)
184 int i, j, k, im, n, ncount, done1, done2;
185 int i1, i1c, i2, i2c, kmi, max;
187 t_complex tmp1, tmp2, tmp3;
188 t_complex *data;
190 /* Use 500 bytes on stack to indicate moves.
191 * This is just for optimization, it does not limit any dimensions.
193 char move[500];
194 int nmove = 500;
196 if (nx < 2 || ny < 2)
198 if (in_data != out_data)
200 memcpy(out_data, in_data, sizeof(t_complex)*nx*ny);
202 return 0;
205 /* Out-of-place transposes are easy */
206 if (in_data != out_data)
208 for (i = 0; i < nx; i++)
210 for (j = 0; j < ny; j++)
212 out_data[j*nx+i].re = in_data[i*ny+j].re;
213 out_data[j*nx+i].im = in_data[i*ny+j].im;
216 return 0;
219 /* In-place transform. in_data=out_data=data */
220 data = in_data;
222 if (nx == ny)
224 /* trivial case, just swap elements */
225 for (i = 0; i < nx; i++)
227 for (j = i+1; j < nx; j++)
229 tmp1.re = data[i*nx+j].re;
230 tmp1.im = data[i*nx+j].im;
231 data[i*nx+j].re = data[j*nx+i].re;
232 data[i*nx+j].im = data[j*nx+i].im;
233 data[j*nx+i].re = tmp1.re;
234 data[j*nx+i].im = tmp1.im;
237 return 0;
240 for (i = 0; i < nmove; i++)
242 move[i] = 0;
245 ncount = 2;
247 if (nx > 2 && ny > 2)
249 i = nx-1;
250 j = ny-1;
253 k = i % j;
254 i = j;
255 j = k;
257 while (k);
258 ncount += i-1;
261 n = nx*ny;
262 k = n - 1;
263 i = 1;
264 im = ny;
266 done1 = 0;
269 i1 = i;
270 kmi = k-i;
271 tmp1.re = data[i1].re;
272 tmp1.im = data[i1].im;
273 i1c = kmi;
274 tmp2.re = data[i1c].re;
275 tmp2.im = data[i1c].im;
277 done2 = 0;
280 i2 = ny*i1-k*(i1/nx);
281 i2c = k-i2;
282 if (i1 < nmove)
284 move[i1] = 1;
286 if (i1c < nmove)
288 move[i1c] = 1;
290 ncount += 2;
291 if (i2 == i)
293 done2 = 1;
295 else if (i2 == kmi)
297 tmp3.re = tmp1.re;
298 tmp3.im = tmp1.im;
299 tmp1.re = tmp2.re;
300 tmp1.im = tmp2.im;
301 tmp2.re = tmp3.re;
302 tmp2.im = tmp3.im;
303 done2 = 1;
305 else
307 data[i1].re = data[i2].re;
308 data[i1].im = data[i2].im;
309 data[i1c].re = data[i2c].re;
310 data[i1c].im = data[i2c].im;
311 i1 = i2;
312 i1c = i2c;
315 while (!done2);
317 data[i1].re = tmp1.re;
318 data[i1].im = tmp1.im;
319 data[i1c].re = tmp2.re;
320 data[i1c].im = tmp2.im;
322 if (ncount >= n)
324 done1 = 1;
326 else
328 done2 = 0;
331 max = k-i;
332 i++;
333 im += ny;
334 if (im > k)
336 im -= k;
338 i2 = im;
339 if (i != i2)
341 if (i >= nmove)
343 while (i2 > i && i2 < max)
345 i1 = i2;
346 i2 = ny*i1-k*(i1/nx);
348 if (i2 == i)
350 done2 = 1;
353 else if (!move[i])
355 done2 = 1;
359 while (!done2);
362 while (!done1);
364 return 0;