Rev 107 |
Blame |
Compare with Previous |
Last modification |
View Log
| RSS feed
/*
* Copyright (c) 1997-1999 Massachusetts Institute of Technology
*
* This program is free software; you can redistribute it and/or modify
* it under the terms of the GNU General Public License as published by
* the Free Software Foundation; either version 2 of the License, or
* (at your option) any later version.
*
* This program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* GNU General Public License for more details.
*
* You should have received a copy of the GNU General Public License
* along with this program; if not, write to the Free Software
* Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
*
*/
/* $Id: fftwnd.c,v 1.2 2003-03-24 11:14:50 pj Exp $ */
#include <fftw-int.h>
/* the number of buffers to use for buffered transforms: */
#define FFTWND_NBUFFERS 8
/* the default number of buffers to use: */
#define FFTWND_DEFAULT_NBUFFERS 0
/* the number of "padding" elements between consecutive buffer lines */
#define FFTWND_BUFFER_PADDING 8
static void destroy_plan_array(int rank, fftw_plan *plans);
static void init_test_array(fftw_complex *arr, int stride, int n)
{
int j;
for (j = 0; j < n; ++j) {
c_re(arr[stride * j]) = 0.0;
c_im(arr[stride * j]) = 0.0;
}
}
/*
* Same as fftw_measure_runtime, except for fftwnd plan.
*/
double fftwnd_measure_runtime(fftwnd_plan plan,
fftw_complex *in, int istride,
fftw_complex *out, int ostride)
{
fftw_time begin, end, start;
double t, tmax, tmin;
int i, iter;
int n;
int repeat;
if (plan->rank == 0)
return 0.0;
n = 1;
for (i = 0; i < plan->rank; ++i)
n *= plan->n[i];
iter = 1;
for (;;) {
tmin = 1.0E10;
tmax = -1.0E10;
init_test_array(in, istride, n);
start = fftw_get_time();
/* repeat the measurement FFTW_TIME_REPEAT times */
for (repeat = 0; repeat < FFTW_TIME_REPEAT; ++repeat) {
begin = fftw_get_time();
for (i = 0; i < iter; ++i) {
fftwnd(plan, 1, in, istride, 0, out, ostride, 0);
}
end = fftw_get_time();
t = fftw_time_to_sec(fftw_time_diff(end, begin));
if (t < tmin)
tmin = t;
if (t > tmax)
tmax = t;
/* do not run for too long */
t = fftw_time_to_sec(fftw_time_diff(end, start));
if (t > FFTW_TIME_LIMIT)
break;
}
if (tmin >= FFTW_TIME_MIN)
break;
iter *= 2;
}
tmin /= (double) iter;
tmax /= (double) iter;
return tmin;
}
/********************** Initializing the FFTWND Plan ***********************/
/* Initialize everything except for the 1D plans and the work array: */
fftwnd_plan fftwnd_create_plan_aux(int rank, const int *n,
fftw_direction dir, int flags)
{
int i;
fftwnd_plan p;
if (rank < 0)
return 0;
for (i = 0; i < rank; ++i)
if (n[i] <= 0)
return 0;
p = (fftwnd_plan) fftw_malloc(sizeof(fftwnd_data));
p->n = 0;
p->n_before = 0;
p->n_after = 0;
p->plans = 0;
p->work = 0;
p->dir = dir;
p->rank = rank;
p->is_in_place = flags & FFTW_IN_PLACE;
p->nwork = 0;
p->nbuffers = 0;
if (rank == 0)
return 0;
p->n = (int *) fftw_malloc(sizeof(int) * rank);
p->n_before = (int *) fftw_malloc(sizeof(int) * rank);
p->n_after = (int *) fftw_malloc(sizeof(int) * rank);
p->n_before[0] = 1;
p->n_after[rank - 1] = 1;
for (i = 0; i < rank; ++i) {
p->n[i] = n[i];
if (i) {
p->n_before[i] = p->n_before[i - 1] * n[i - 1];
p->n_after[rank - 1 - i] = p->n_after[rank - i] * n[rank - i];
}
}
return p;
}
/* create an empty new array of rank 1d plans */
fftw_plan *fftwnd_new_plan_array(int rank)
{
fftw_plan *plans;
int i;
plans = (fftw_plan *) fftw_malloc(rank * sizeof(fftw_plan));
if (!plans)
return 0;
for (i = 0; i < rank; ++i)
plans[i] = 0;
return plans;
}
/*
* create an array of plans using the ordinary 1d fftw_create_plan,
* which allocates its own array and creates plans optimized for
* contiguous data.
*/
fftw_plan *fftwnd_create_plans_generic(fftw_plan *plans,
int rank, const int *n,
fftw_direction dir, int flags)
{
if (rank <= 0)
return 0;
if (plans) {
int i, j;
int cur_flags;
for (i = 0; i < rank; ++i) {
if (i < rank - 1 || (flags & FFTW_IN_PLACE)) {
/*
* fft's except the last dimension are always in-place
*/
cur_flags = flags | FFTW_IN_PLACE;
for (j = i - 1; j >= 0 && n[i] != n[j]; --j);
} else {
cur_flags = flags;
/*
* we must create a separate plan for the last
* dimension
*/
j = -1;
}
if (j >= 0) {
/*
* If a plan already exists for this size
* array, reuse it:
*/
plans[i] = plans[j];
} else {
/* generate a new plan: */
plans[i] = fftw_create_plan(n[i], dir, cur_flags);
if (!plans[i]) {
destroy_plan_array(rank, plans);
return 0;
}
}
}
}
return plans;
}
static int get_maxdim(int rank, const int *n, int flags)
{
int i;
int maxdim = 0;
for (i = 0; i < rank - 1; ++i)
if (n[i] > maxdim)
maxdim = n[i];
if (rank > 0 && flags & FFTW_IN_PLACE && n[rank - 1] > maxdim)
maxdim = n[rank - 1];
return maxdim;
}
/* compute number of elements required for work array (has to
be big enough to hold ncopies of the largest dimension in
n that will need an in-place transform. */
int fftwnd_work_size(int rank, const int *n, int flags, int ncopies)
{
return (ncopies * get_maxdim(rank, n, flags)
+ (ncopies - 1) * FFTWND_BUFFER_PADDING);
}
/*
* create plans using the fftw_create_plan_specific planner, which
* allows us to create plans for each dimension that are specialized
* for the strides that we are going to use.
*/
fftw_plan *fftwnd_create_plans_specific(fftw_plan *plans,
int rank, const int *n,
const int *n_after,
fftw_direction dir, int flags,
fftw_complex *in, int istride,
fftw_complex *out, int ostride)
{
if (rank <= 0)
return 0;
if (plans) {
int i, stride, cur_flags;
fftw_complex *work = 0;
int nwork;
nwork = fftwnd_work_size(rank, n, flags, 1);
if (nwork)
work = (fftw_complex*)fftw_malloc(nwork * sizeof(fftw_complex));
for (i = 0; i < rank; ++i) {
/* fft's except the last dimension are always in-place */
if (i < rank - 1)
cur_flags = flags | FFTW_IN_PLACE;
else
cur_flags = flags;
/* stride for transforming ith dimension */
stride = n_after[i];
if (cur_flags & FFTW_IN_PLACE)
plans[i] = fftw_create_plan_specific(n[i], dir, cur_flags,
in, istride * stride,
work, 1);
else
plans[i] = fftw_create_plan_specific(n[i], dir, cur_flags,
in, istride * stride,
out, ostride * stride);
if (!plans[i]) {
destroy_plan_array(rank, plans);
fftw_free(work);
return 0;
}
}
if (work)
fftw_free(work);
}
return plans;
}
/*
* Create an fftwnd_plan specialized for specific arrays. (These
* arrays are ignored, however, if they are NULL or if the flags do
* not include FFTW_MEASURE.) The main advantage of being provided
* arrays like this is that we can do runtime timing measurements of
* our options, without worrying about allocating excessive scratch
* space.
*/
fftwnd_plan fftwnd_create_plan_specific(int rank, const int *n,
fftw_direction dir, int flags,
fftw_complex *in, int istride,
fftw_complex *out, int ostride)
{
fftwnd_plan p;
if (!(p = fftwnd_create_plan_aux(rank, n, dir, flags)))
return 0;
if (!(flags & FFTW_MEASURE) || in == 0
|| (!p->is_in_place && out == 0)) {
/**** use default plan ****/
p->plans = fftwnd_create_plans_generic(fftwnd_new_plan_array(rank),
rank, n, dir, flags);
if (!p->plans) {
fftwnd_destroy_plan(p);
return 0;
}
if (flags & FFTWND_FORCE_BUFFERED)
p->nbuffers = FFTWND_NBUFFERS;
else
p->nbuffers = FFTWND_DEFAULT_NBUFFERS;
p->nwork = fftwnd_work_size(rank, n, flags, p->nbuffers + 1);
if (p->nwork && !(flags & FFTW_THREADSAFE)) {
p->work = (fftw_complex*) fftw_malloc(p->nwork
* sizeof(fftw_complex));
if (!p->work) {
fftwnd_destroy_plan(p);
return 0;
}
}
} else {
/**** use runtime measurements to pick plan ****/
fftw_plan *plans_buf, *plans_nobuf;
double t_buf, t_nobuf;
p->nwork = fftwnd_work_size(rank, n, flags, FFTWND_NBUFFERS + 1);
if (p->nwork && !(flags & FFTW_THREADSAFE)) {
p->work = (fftw_complex*) fftw_malloc(p->nwork
* sizeof(fftw_complex));
if (!p->work) {
fftwnd_destroy_plan(p);
return 0;
}
}
else
p->work = (fftw_complex*) NULL;
/* two possible sets of 1D plans: */
plans_buf = fftwnd_create_plans_generic(fftwnd_new_plan_array(rank),
rank, n, dir, flags);
plans_nobuf =
fftwnd_create_plans_specific(fftwnd_new_plan_array(rank),
rank, n, p->n_after, dir,
flags, in, istride,
out, ostride);
if (!plans_buf || !plans_nobuf) {
destroy_plan_array(rank, plans_nobuf);
destroy_plan_array(rank, plans_buf);
fftwnd_destroy_plan(p);
return 0;
}
/* time the two possible plans */
p->plans = plans_nobuf;
p->nbuffers = 0;
p->nwork = fftwnd_work_size(rank, n, flags, p->nbuffers + 1);
t_nobuf = fftwnd_measure_runtime(p, in, istride, out, ostride);
p->plans = plans_buf;
p->nbuffers = FFTWND_NBUFFERS;
p->nwork = fftwnd_work_size(rank, n, flags, p->nbuffers + 1);
t_buf = fftwnd_measure_runtime(p, in, istride, out, ostride);
/* pick the better one: */
if (t_nobuf < t_buf) { /* use unbuffered transform */
p->plans = plans_nobuf;
p->nbuffers = 0;
/* work array is unnecessarily large */
if (p->work)
fftw_free(p->work);
p->work = 0;
destroy_plan_array(rank, plans_buf);
/* allocate a work array of the correct size: */
p->nwork = fftwnd_work_size(rank, n, flags, p->nbuffers + 1);
if (p->nwork && !(flags & FFTW_THREADSAFE)) {
p->work = (fftw_complex*) fftw_malloc(p->nwork
* sizeof(fftw_complex));
if (!p->work) {
fftwnd_destroy_plan(p);
return 0;
}
}
} else { /* use buffered transform */
destroy_plan_array(rank, plans_nobuf);
}
}
return p;
}
fftwnd_plan fftw2d_create_plan_specific(int nx, int ny,
fftw_direction dir, int flags,
fftw_complex *in, int istride,
fftw_complex *out, int ostride)
{
int n[2];
n[0] = nx;
n[1] = ny;
return fftwnd_create_plan_specific(2, n, dir, flags,
in, istride, out, ostride);
}
fftwnd_plan fftw3d_create_plan_specific(int nx, int ny, int nz,
fftw_direction dir, int flags,
fftw_complex *in, int istride,
fftw_complex *out, int ostride)
{
int n[3];
n[0] = nx;
n[1] = ny;
n[2] = nz;
return fftwnd_create_plan_specific(3, n, dir, flags,
in, istride, out, ostride);
}
/* Create a generic fftwnd plan: */
fftwnd_plan fftwnd_create_plan(int rank, const int *n,
fftw_direction dir, int flags)
{
return fftwnd_create_plan_specific(rank, n, dir, flags, 0, 1, 0, 1);
}
fftwnd_plan fftw2d_create_plan(int nx, int ny,
fftw_direction dir, int flags)
{
return fftw2d_create_plan_specific(nx, ny, dir, flags, 0, 1, 0, 1);
}
fftwnd_plan fftw3d_create_plan(int nx, int ny, int nz,
fftw_direction dir, int flags)
{
return fftw3d_create_plan_specific(nx, ny, nz, dir, flags, 0, 1, 0, 1);
}
/************************ Freeing the FFTWND Plan ************************/
static void destroy_plan_array(int rank, fftw_plan *plans)
{
if (plans) {
int i, j;
for (i = 0; i < rank; ++i) {
for (j = i - 1;
j >= 0 && plans[i] != plans[j];
--j);
if (j < 0 && plans[i])
fftw_destroy_plan(plans[i]);
}
fftw_free(plans);
}
}
void fftwnd_destroy_plan(fftwnd_plan plan)
{
if (plan) {
destroy_plan_array(plan->rank, plan->plans);
if (plan->n)
fftw_free(plan->n);
if (plan->n_before)
fftw_free(plan->n_before);
if (plan->n_after)
fftw_free(plan->n_after);
if (plan->work)
fftw_free(plan->work);
fftw_free(plan);
}
}
/************************ Printing the FFTWND Plan ************************/
/*
void fftwnd_fprint_plan(FILE *f, fftwnd_plan plan)
{
if (plan) {
int i, j;
if (plan->rank == 0) {
fprintf(f, "plan for rank 0 (null) transform.\n");
return;
}
fprintf(f, "plan for ");
for (i = 0; i < plan->rank; ++i)
fprintf(f, "%s%d", i ? "x" : "", plan->n[i]);
fprintf(f, " transform:\n");
if (plan->nbuffers > 0)
fprintf(f, " -- using buffered transforms (%d buffers)\n",
plan->nbuffers);
else
fprintf(f, " -- using unbuffered transform\n");
for (i = 0; i < plan->rank; ++i) {
fprintf(f, "* dimension %d (size %d) ", i, plan->n[i]);
for (j = i - 1; j >= 0; --j)
if (plan->plans[j] == plan->plans[i])
break;
if (j < 0)
fftw_fprint_plan(f, plan->plans[i]);
else
fprintf(f, "plan is same as dimension %d plan.\n", j);
}
}
}
void fftwnd_print_plan(fftwnd_plan plan)
{
fftwnd_fprint_plan(stdout, plan);
}
*/
/********************* Buffered FFTW (in-place) *********************/
void fftw_buffered(fftw_plan p, int howmany,
fftw_complex *in, int istride, int idist,
fftw_complex *work,
int nbuffers, fftw_complex *buffers)
{
int i = 0, n, nb;
n = p->n;
nb = n + FFTWND_BUFFER_PADDING;
do {
for (; i <= howmany - nbuffers; i += nbuffers) {
fftw_complex *cur_in = in + i * idist;
int j, buf;
/*
* First, copy nbuffers strided arrays to the
* contiguous buffer arrays (reading consecutive
* locations, assuming that idist is 1):
*/
for (j = 0; j < n; ++j) {
fftw_complex *cur_in2 = cur_in + j * istride;
fftw_complex *cur_buffers = buffers + j;
for (buf = 0; buf <= nbuffers - 4; buf += 4) {
*cur_buffers = *cur_in2;
*(cur_buffers += nb) = *(cur_in2 += idist);
*(cur_buffers += nb) = *(cur_in2 += idist);
*(cur_buffers += nb) = *(cur_in2 += idist);
cur_buffers += nb;
cur_in2 += idist;
}
for (; buf < nbuffers; ++buf) {
*cur_buffers = *cur_in2;
cur_buffers += nb;
cur_in2 += idist;
}
}
/*
* Now, compute the FFTs in the buffers (in-place
* using work):
*/
fftw(p, nbuffers, buffers, 1, nb, work, 1, 0);
/*
* Finally, copy the results back from the contiguous
* buffers to the strided arrays (writing consecutive
* locations):
*/
for (j = 0; j < n; ++j) {
fftw_complex *cur_in2 = cur_in + j * istride;
fftw_complex *cur_buffers = buffers + j;
for (buf = 0; buf <= nbuffers - 4; buf += 4) {
*cur_in2 = *cur_buffers;
*(cur_in2 += idist) = *(cur_buffers += nb);
*(cur_in2 += idist) = *(cur_buffers += nb);
*(cur_in2 += idist) = *(cur_buffers += nb);
cur_buffers += nb;
cur_in2 += idist;
}
for (; buf < nbuffers; ++buf) {
*cur_in2 = *cur_buffers;
cur_buffers += nb;
cur_in2 += idist;
}
}
}
/*
* we skip howmany % nbuffers ffts at the end of the loop,
* so we have to go back and do them:
*/
nbuffers = howmany - i;
} while (i < howmany);
}
/********************* Computing the N-Dimensional FFT *********************/
void fftwnd_aux(fftwnd_plan p, int cur_dim,
fftw_complex *in, int istride,
fftw_complex *out, int ostride,
fftw_complex *work)
{
int n_after = p->n_after[cur_dim], n = p->n[cur_dim];
if (cur_dim == p->rank - 2) {
/* just do the last dimension directly: */
if (p->is_in_place)
fftw(p->plans[p->rank - 1], n,
in, istride, n_after * istride,
work, 1, 0);
else
fftw(p->plans[p->rank - 1], n,
in, istride, n_after * istride,
out, ostride, n_after * ostride);
} else { /* we have at least two dimensions to go */
int i;
/*
* process the subsequent dimensions recursively, in hyperslabs,
* to get maximum locality:
*/
for (i = 0; i < n; ++i)
fftwnd_aux(p, cur_dim + 1,
in + i * n_after * istride, istride,
out + i * n_after * ostride, ostride, work);
}
/* do the current dimension (in-place): */
if (p->nbuffers == 0) {
fftw(p->plans[cur_dim], n_after,
out, n_after * ostride, ostride,
work, 1, 0);
} else /* using contiguous copy buffers: */
fftw_buffered(p->plans[cur_dim], n_after,
out, n_after * ostride, ostride,
work, p->nbuffers, work + n);
}
/*
* alternate version of fftwnd_aux -- this version pushes the howmany
* loop down to the leaves of the computation, for greater locality in
* cases where dist < stride
*/
void fftwnd_aux_howmany(fftwnd_plan p, int cur_dim,
int howmany,
fftw_complex *in, int istride, int idist,
fftw_complex *out, int ostride, int odist,
fftw_complex *work)
{
int n_after = p->n_after[cur_dim], n = p->n[cur_dim];
int k;
if (cur_dim == p->rank - 2) {
/* just do the last dimension directly: */
if (p->is_in_place)
for (k = 0; k < n; ++k)
fftw(p->plans[p->rank - 1], howmany,
in + k * n_after * istride, istride, idist,
work, 1, 0);
else
for (k = 0; k < n; ++k)
fftw(p->plans[p->rank - 1], howmany,
in + k * n_after * istride, istride, idist,
out + k * n_after * ostride, ostride, odist);
} else { /* we have at least two dimensions to go */
int i;
/*
* process the subsequent dimensions recursively, in
* hyperslabs, to get maximum locality:
*/
for (i = 0; i < n; ++i)
fftwnd_aux_howmany(p, cur_dim + 1, howmany,
in + i * n_after * istride, istride, idist,
out + i * n_after * ostride, ostride, odist,
work);
}
/* do the current dimension (in-place): */
if (p->nbuffers == 0)
for (k = 0; k < n_after; ++k)
fftw(p->plans[cur_dim], howmany,
out + k * ostride, n_after * ostride, odist,
work, 1, 0);
else /* using contiguous copy buffers: */
for (k = 0; k < n_after; ++k)
fftw_buffered(p->plans[cur_dim], howmany,
out + k * ostride, n_after * ostride, odist,
work, p->nbuffers, work + n);
}
void fftwnd(fftwnd_plan p, int howmany,
fftw_complex *in, int istride, int idist,
fftw_complex *out, int ostride, int odist)
{
fftw_complex *work;
#ifdef FFTW_DEBUG
if (p->rank > 0 && (p->plans[0]->flags & FFTW_THREADSAFE)
&& p->nwork && p->work)
fftw_die("bug with FFTW_THREADSAFE flag");
#endif
if (p->nwork && !p->work)
work = (fftw_complex *) fftw_malloc(p->nwork * sizeof(fftw_complex));
else
work = p->work;
switch (p->rank) {
case 0:
break;
case 1:
if (p->is_in_place) /* fft is in-place */
fftw(p->plans[0], howmany, in, istride, idist,
work, 1, 0);
else
fftw(p->plans[0], howmany, in, istride, idist,
out, ostride, odist);
break;
default: /* rank >= 2 */
{
if (p->is_in_place) {
out = in;
ostride = istride;
odist = idist;
}
if (howmany > 1 && odist < ostride)
fftwnd_aux_howmany(p, 0, howmany,
in, istride, idist,
out, ostride, odist,
work);
else {
int i;
for (i = 0; i < howmany; ++i)
fftwnd_aux(p, 0,
in + i * idist, istride,
out + i * odist, ostride,
work);
}
}
}
if (p->nwork && !p->work)
fftw_free(work);
}
void fftwnd_one(fftwnd_plan p, fftw_complex *in, fftw_complex *out)
{
fftwnd(p, 1, in, 1, 1, out, 1, 1);
}