*/
/**
- * @file libavcodec/fft.c
+ * @file
* FFT/IFFT transforms.
*/
-#include "dsputil.h"
-
-#if CONFIG_HARDCODED_TABLES
-#define COSTABLE(size) \
- extern const DECLARE_ALIGNED_16(FFTSample, ff_cos_##size[size/2]);
-#else
-#define COSTABLE(size) \
- DECLARE_ALIGNED_16(FFTSample, ff_cos_##size[size/2]);
-#endif
+#include <stdlib.h>
+#include <string.h>
+#include "libavutil/mathematics.h"
+#include "fft.h"
/* cos(2*pi*x/n) for 0<=x<=n/4, followed by its reverse */
-COSTABLE(16)
-COSTABLE(32)
-COSTABLE(64)
-COSTABLE(128)
-COSTABLE(256)
-COSTABLE(512)
-COSTABLE(1024)
-COSTABLE(2048)
-COSTABLE(4096)
-COSTABLE(8192)
-COSTABLE(16384)
-COSTABLE(32768)
-COSTABLE(65536)
-#if CONFIG_HARDCODED_TABLES
-const
+#if !CONFIG_HARDCODED_TABLES
+COSTABLE(16);
+COSTABLE(32);
+COSTABLE(64);
+COSTABLE(128);
+COSTABLE(256);
+COSTABLE(512);
+COSTABLE(1024);
+COSTABLE(2048);
+COSTABLE(4096);
+COSTABLE(8192);
+COSTABLE(16384);
+COSTABLE(32768);
+COSTABLE(65536);
#endif
-FFTSample * const ff_cos_tabs[] = {
+COSTABLE_CONST FFTSample * const ff_cos_tabs[] = {
+ NULL, NULL, NULL, NULL,
ff_cos_16, ff_cos_32, ff_cos_64, ff_cos_128, ff_cos_256, ff_cos_512, ff_cos_1024,
ff_cos_2048, ff_cos_4096, ff_cos_8192, ff_cos_16384, ff_cos_32768, ff_cos_65536,
};
else return split_radix_permutation(i, m, inverse)*4 - 1;
}
+av_cold void ff_init_ff_cos_tabs(int index)
+{
+#if !CONFIG_HARDCODED_TABLES
+ int i;
+ int m = 1<<index;
+ double freq = 2*M_PI/m;
+ FFTSample *tab = ff_cos_tabs[index];
+ for(i=0; i<=m/4; i++)
+ tab[i] = cos(i*freq);
+ for(i=1; i<m/4; i++)
+ tab[m/2-i] = tab[i];
+#endif
+}
+
av_cold int ff_fft_init(FFTContext *s, int nbits, int inverse)
{
- int i, j, m, n;
- float alpha, c1, s1, s2;
- int av_unused has_vectors;
+ int i, j, n;
if (nbits < 2 || nbits > 16)
goto fail;
s->nbits = nbits;
n = 1 << nbits;
- s->tmp_buf = NULL;
- s->exptab = av_malloc((n / 2) * sizeof(FFTComplex));
- if (!s->exptab)
- goto fail;
s->revtab = av_malloc(n * sizeof(uint16_t));
if (!s->revtab)
goto fail;
+ s->tmp_buf = av_malloc(n * sizeof(FFTComplex));
+ if (!s->tmp_buf)
+ goto fail;
s->inverse = inverse;
- s2 = inverse ? 1.0 : -1.0;
-
s->fft_permute = ff_fft_permute_c;
s->fft_calc = ff_fft_calc_c;
+#if CONFIG_MDCT
s->imdct_calc = ff_imdct_calc_c;
s->imdct_half = ff_imdct_half_c;
s->mdct_calc = ff_mdct_calc_c;
- s->exptab1 = NULL;
- s->split_radix = 1;
+#endif
if (ARCH_ARM) ff_fft_init_arm(s);
if (HAVE_ALTIVEC) ff_fft_init_altivec(s);
if (HAVE_MMX) ff_fft_init_mmx(s);
- if (s->split_radix) {
-#if !CONFIG_HARDCODED_TABLES
- for(j=4; j<=nbits; j++) {
- int m = 1<<j;
- double freq = 2*M_PI/m;
- FFTSample *tab = ff_cos_tabs[j-4];
- for(i=0; i<=m/4; i++)
- tab[i] = cos(i*freq);
- for(i=1; i<m/4; i++)
- tab[m/2-i] = tab[i];
- }
-#endif
- for(i=0; i<n; i++)
- s->revtab[-split_radix_permutation(i, n, s->inverse) & (n-1)] = i;
- s->tmp_buf = av_malloc(n * sizeof(FFTComplex));
- } else {
- int np, nblocks, np2, l;
- FFTComplex *q;
-
- for(i=0; i<(n/2); i++) {
- alpha = 2 * M_PI * (float)i / (float)n;
- c1 = cos(alpha);
- s1 = sin(alpha) * s2;
- s->exptab[i].re = c1;
- s->exptab[i].im = s1;
- }
-
- np = 1 << nbits;
- nblocks = np >> 3;
- np2 = np >> 1;
- s->exptab1 = av_malloc(np * 2 * sizeof(FFTComplex));
- if (!s->exptab1)
- goto fail;
- q = s->exptab1;
- do {
- for(l = 0; l < np2; l += 2 * nblocks) {
- *q++ = s->exptab[l];
- *q++ = s->exptab[l + nblocks];
-
- q->re = -s->exptab[l].im;
- q->im = s->exptab[l].re;
- q++;
- q->re = -s->exptab[l + nblocks].im;
- q->im = s->exptab[l + nblocks].re;
- q++;
- }
- nblocks = nblocks >> 1;
- } while (nblocks != 0);
- av_freep(&s->exptab);
-
- /* compute bit reverse table */
- for(i=0;i<n;i++) {
- m=0;
- for(j=0;j<nbits;j++) {
- m |= ((i >> j) & 1) << (nbits-j-1);
- }
- s->revtab[i]=m;
- }
+ for(j=4; j<=nbits; j++) {
+ ff_init_ff_cos_tabs(j);
}
+ for(i=0; i<n; i++)
+ s->revtab[-split_radix_permutation(i, n, s->inverse) & (n-1)] = i;
return 0;
fail:
av_freep(&s->revtab);
- av_freep(&s->exptab);
- av_freep(&s->exptab1);
av_freep(&s->tmp_buf);
return -1;
}
void ff_fft_permute_c(FFTContext *s, FFTComplex *z)
{
- int j, k, np;
- FFTComplex tmp;
+ int j, np;
const uint16_t *revtab = s->revtab;
np = 1 << s->nbits;
-
- if (s->tmp_buf) {
- /* TODO: handle split-radix permute in a more optimal way, probably in-place */
- for(j=0;j<np;j++) s->tmp_buf[revtab[j]] = z[j];
- memcpy(z, s->tmp_buf, np * sizeof(FFTComplex));
- return;
- }
-
- /* reverse */
- for(j=0;j<np;j++) {
- k = revtab[j];
- if (k < j) {
- tmp = z[k];
- z[k] = z[j];
- z[j] = tmp;
- }
- }
+ /* TODO: handle split-radix permute in a more optimal way, probably in-place */
+ for(j=0;j<np;j++) s->tmp_buf[revtab[j]] = z[j];
+ memcpy(z, s->tmp_buf, np * sizeof(FFTComplex));
}
av_cold void ff_fft_end(FFTContext *s)
{
av_freep(&s->revtab);
- av_freep(&s->exptab);
- av_freep(&s->exptab1);
av_freep(&s->tmp_buf);
}
projects / ffmpeg / blobdiff