2 Copyright (c) 2003-2004, Mark Borgerding
6 Redistribution and use in source and binary forms, with or without modification, are permitted provided that the following conditions are met:
8 * Redistributions of source code must retain the above copyright notice, this list of conditions and the following disclaimer.
9 * Redistributions in binary form must reproduce the above copyright notice, this list of conditions and the following disclaimer in the documentation and/or other materials provided with the distribution.
10 * Neither the author nor the names of any contributors may be used to endorse or promote products derived from this software without specific prior written permission.
12 THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
20 #include "_kiss_fft_guts.h"
23 /* The guts header contains all the multiplication and addition macros that are defined for
24 fixed or floating point complex numbers. It also delares the kf_ internal functions.
27 static kiss_fft_cpx *scratchbuf=NULL;
28 static size_t nscratchbuf=0;
29 static kiss_fft_cpx *tmpbuf=NULL;
30 static size_t ntmpbuf=0;
32 #define CHECKBUF(buf,nbuf,n) \
34 if ( nbuf < (size_t)(n) ) {\
36 buf = (kiss_fft_cpx*)KISS_FFT_MALLOC(sizeof(kiss_fft_cpx)*(n)); \
44 const kiss_fft_cfg st,
49 kiss_fft_cpx * tw1 = st->twiddles;
57 x[i].r = SHR(x[i].r,1);
58 x[i].i = SHR(x[i].i,1);
63 C_MUL (t, *Fout2 , *tw1);
65 C_SUB( *Fout2 , *Fout , t );
75 const kiss_fft_cfg st,
79 kiss_fft_cpx *tw1,*tw2,*tw3;
80 kiss_fft_cpx scratch[6];
85 tw3 = tw2 = tw1 = st->twiddles;
92 x[i].r = PSHR16(x[i].r,2);
93 x[i].i = PSHR16(x[i].i,2);
99 C_MUL(scratch[0],Fout[m] , *tw1 );
100 C_MUL(scratch[1],Fout[m2] , *tw2 );
101 C_MUL(scratch[2],Fout[m3] , *tw3 );
103 C_SUB( scratch[5] , *Fout, scratch[1] );
104 C_ADDTO(*Fout, scratch[1]);
105 C_ADD( scratch[3] , scratch[0] , scratch[2] );
106 C_SUB( scratch[4] , scratch[0] , scratch[2] );
107 C_SUB( Fout[m2], *Fout, scratch[3] );
111 C_ADDTO( *Fout , scratch[3] );
113 Fout[m].r = scratch[5].r - scratch[4].i;
114 Fout[m].i = scratch[5].i + scratch[4].r;
115 Fout[m3].r = scratch[5].r + scratch[4].i;
116 Fout[m3].i = scratch[5].i - scratch[4].r;
122 C_MUL(scratch[0],Fout[m] , *tw1 );
123 C_MUL(scratch[1],Fout[m2] , *tw2 );
124 C_MUL(scratch[2],Fout[m3] , *tw3 );
126 C_SUB( scratch[5] , *Fout, scratch[1] );
127 C_ADDTO(*Fout, scratch[1]);
128 C_ADD( scratch[3] , scratch[0] , scratch[2] );
129 C_SUB( scratch[4] , scratch[0] , scratch[2] );
130 C_SUB( Fout[m2], *Fout, scratch[3] );
134 C_ADDTO( *Fout , scratch[3] );
136 Fout[m].r = scratch[5].r + scratch[4].i;
137 Fout[m].i = scratch[5].i - scratch[4].r;
138 Fout[m3].r = scratch[5].r - scratch[4].i;
139 Fout[m3].i = scratch[5].i + scratch[4].r;
145 static void kf_bfly3(
147 const size_t fstride,
148 const kiss_fft_cfg st,
153 const size_t m2 = 2*m;
154 kiss_fft_cpx *tw1,*tw2;
155 kiss_fft_cpx scratch[5];
157 epi3 = st->twiddles[fstride*m];
159 tw1=tw2=st->twiddles;
163 C_FIXDIV(*Fout,3); C_FIXDIV(Fout[m],3); C_FIXDIV(Fout[m2],3);
166 C_MUL(scratch[1],Fout[m] , *tw1);
167 C_MUL(scratch[2],Fout[m2] , *tw2);
169 C_ADD(scratch[3],scratch[1],scratch[2]);
170 C_SUB(scratch[0],scratch[1],scratch[2]);
174 Fout[m].r = Fout->r - HALF_OF(scratch[3].r);
175 Fout[m].i = Fout->i - HALF_OF(scratch[3].i);
177 C_MULBYSCALAR( scratch[0] , epi3.i );
179 C_ADDTO(*Fout,scratch[3]);
181 Fout[m2].r = Fout[m].r + scratch[0].i;
182 Fout[m2].i = Fout[m].i - scratch[0].r;
184 Fout[m].r -= scratch[0].i;
185 Fout[m].i += scratch[0].r;
191 static void kf_bfly5(
193 const size_t fstride,
194 const kiss_fft_cfg st,
198 kiss_fft_cpx *Fout0,*Fout1,*Fout2,*Fout3,*Fout4;
200 kiss_fft_cpx scratch[13];
201 kiss_fft_cpx * twiddles = st->twiddles;
204 ya = twiddles[fstride*m];
205 yb = twiddles[fstride*2*m];
214 for ( u=0; u<m; ++u ) {
216 C_FIXDIV( *Fout0,5); C_FIXDIV( *Fout1,5); C_FIXDIV( *Fout2,5); C_FIXDIV( *Fout3,5); C_FIXDIV( *Fout4,5);
220 C_MUL(scratch[1] ,*Fout1, tw[u*fstride]);
221 C_MUL(scratch[2] ,*Fout2, tw[2*u*fstride]);
222 C_MUL(scratch[3] ,*Fout3, tw[3*u*fstride]);
223 C_MUL(scratch[4] ,*Fout4, tw[4*u*fstride]);
225 C_ADD( scratch[7],scratch[1],scratch[4]);
226 C_SUB( scratch[10],scratch[1],scratch[4]);
227 C_ADD( scratch[8],scratch[2],scratch[3]);
228 C_SUB( scratch[9],scratch[2],scratch[3]);
230 Fout0->r += scratch[7].r + scratch[8].r;
231 Fout0->i += scratch[7].i + scratch[8].i;
233 scratch[5].r = scratch[0].r + S_MUL(scratch[7].r,ya.r) + S_MUL(scratch[8].r,yb.r);
234 scratch[5].i = scratch[0].i + S_MUL(scratch[7].i,ya.r) + S_MUL(scratch[8].i,yb.r);
236 scratch[6].r = S_MUL(scratch[10].i,ya.i) + S_MUL(scratch[9].i,yb.i);
237 scratch[6].i = -S_MUL(scratch[10].r,ya.i) - S_MUL(scratch[9].r,yb.i);
239 C_SUB(*Fout1,scratch[5],scratch[6]);
240 C_ADD(*Fout4,scratch[5],scratch[6]);
242 scratch[11].r = scratch[0].r + S_MUL(scratch[7].r,yb.r) + S_MUL(scratch[8].r,ya.r);
243 scratch[11].i = scratch[0].i + S_MUL(scratch[7].i,yb.r) + S_MUL(scratch[8].i,ya.r);
244 scratch[12].r = - S_MUL(scratch[10].i,yb.i) + S_MUL(scratch[9].i,ya.i);
245 scratch[12].i = S_MUL(scratch[10].r,yb.i) - S_MUL(scratch[9].r,ya.i);
247 C_ADD(*Fout2,scratch[11],scratch[12]);
248 C_SUB(*Fout3,scratch[11],scratch[12]);
250 ++Fout0;++Fout1;++Fout2;++Fout3;++Fout4;
254 /* perform the butterfly for one stage of a mixed radix FFT */
255 static void kf_bfly_generic(
257 const size_t fstride,
258 const kiss_fft_cfg st,
264 kiss_fft_cpx * twiddles = st->twiddles;
266 int Norig = st->nfft;
268 CHECKBUF(scratchbuf,nscratchbuf,p);
270 for ( u=0; u<m; ++u ) {
272 for ( q1=0 ; q1<p ; ++q1 ) {
273 scratchbuf[q1] = Fout[ k ];
275 C_FIXDIV(scratchbuf[q1],p);
281 for ( q1=0 ; q1<p ; ++q1 ) {
283 Fout[ k ] = scratchbuf[0];
285 twidx += fstride * k;
286 if (twidx>=Norig) twidx-=Norig;
287 C_MUL(t,scratchbuf[q] , twiddles[twidx] );
288 C_ADDTO( Fout[ k ] ,t);
298 const kiss_fft_cpx * f,
299 const size_t fstride,
302 const kiss_fft_cfg st
305 kiss_fft_cpx * Fout_beg=Fout;
306 const int p=*factors++; /* the radix */
307 const int m=*factors++; /* stage's fft length/p */
308 const kiss_fft_cpx * Fout_end = Fout + p*m;
313 f += fstride*in_stride;
314 }while(++Fout != Fout_end );
317 kf_work( Fout , f, fstride*p, in_stride, factors,st);
318 f += fstride*in_stride;
319 }while( (Fout += m) != Fout_end );
325 case 2: kf_bfly2(Fout,fstride,st,m); break;
326 case 3: kf_bfly3(Fout,fstride,st,m); break;
327 case 4: kf_bfly4(Fout,fstride,st,m); break;
328 case 5: kf_bfly5(Fout,fstride,st,m); break;
329 default: kf_bfly_generic(Fout,fstride,st,m,p); break;
333 /* facbuf is populated by p1,m1,p2,m2, ...
338 void kf_factor(int n,int * facbuf)
342 floor_sqrt = floor( sqrt((double)n) );
344 /*factor out powers of 4, powers of 2, then any remaining primes */
348 case 4: p = 2; break;
349 case 2: p = 3; break;
350 default: p += 2; break;
353 p = n; /* no more factors, skip to end */
363 * User-callable function to allocate all necessary storage space for the fft.
365 * The return value is a contiguous block of memory, allocated with malloc. As such,
366 * It can be freed with free(), rather than a kiss_fft-specific function.
368 kiss_fft_cfg kiss_fft_alloc(int nfft,int inverse_fft,void * mem,size_t * lenmem )
370 kiss_fft_cfg st=NULL;
371 size_t memneeded = sizeof(struct kiss_fft_state)
372 + sizeof(kiss_fft_cpx)*(nfft-1); /* twiddle factors*/
374 if ( lenmem==NULL ) {
375 st = ( kiss_fft_cfg)KISS_FFT_MALLOC( memneeded );
377 if (mem != NULL && *lenmem >= memneeded)
378 st = (kiss_fft_cfg)mem;
384 st->inverse = inverse_fft;
386 for (i=0;i<nfft;++i) {
387 const double pi=3.14159265358979323846264338327;
388 double phase = ( -2*pi /nfft ) * i;
391 kf_cexp(st->twiddles+i, phase );
394 kf_factor(nfft,st->factors);
402 void kiss_fft_stride(kiss_fft_cfg st,const kiss_fft_cpx *fin,kiss_fft_cpx *fout,int in_stride)
405 CHECKBUF(tmpbuf,ntmpbuf,st->nfft);
406 kf_work(tmpbuf,fin,1,in_stride, st->factors,st);
407 speex_move(fout,tmpbuf,sizeof(kiss_fft_cpx)*st->nfft);
409 kf_work( fout, fin, 1,in_stride, st->factors,st );
413 void kiss_fft(kiss_fft_cfg cfg,const kiss_fft_cpx *fin,kiss_fft_cpx *fout)
415 kiss_fft_stride(cfg,fin,fout,1);
419 /* not really necessary to call, but if someone is doing in-place ffts, they may want to free the
420 buffers from CHECKBUF
422 void kiss_fft_cleanup(void)
424 speex_free(scratchbuf);