301 lines
		
	
	
		
			8.6 KiB
		
	
	
	
		
			C
		
	
	
	
	
	
			
		
		
	
	
			301 lines
		
	
	
		
			8.6 KiB
		
	
	
	
		
			C
		
	
	
	
	
	
| /*
 | |
|  * Copyright (c) 2003, 2007-14 Matteo Frigo
 | |
|  * Copyright (c) 2003, 2007-14 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., 51 Franklin Street, Fifth Floor, Boston, MA  02110-1301  USA
 | |
|  *
 | |
|  */
 | |
| 
 | |
| 
 | |
| /* Do an R{E,O}DFT11 problem via an R2HC problem of the same *odd* size,
 | |
|    with some permutations and post-processing, as described in:
 | |
| 
 | |
|      S. C. Chan and K. L. Ho, "Fast algorithms for computing the
 | |
|      discrete cosine transform," IEEE Trans. Circuits Systems II:
 | |
|      Analog & Digital Sig. Proc. 39 (3), 185--190 (1992).
 | |
| 
 | |
|    (For even sizes, see reodft11e-radix2.c.)  
 | |
| 
 | |
|    This algorithm is related to the 8 x n prime-factor-algorithm (PFA)
 | |
|    decomposition of the size 8n "logical" DFT corresponding to the
 | |
|    R{EO}DFT11.
 | |
| 
 | |
|    Aside from very confusing notation (several symbols are redefined
 | |
|    from one line to the next), be aware that this paper has some
 | |
|    errors.  In particular, the signs are wrong in Eqs. (34-35).  Also,
 | |
|    Eqs. (36-37) should be simply C(k) = C(2k + 1 mod N), and similarly
 | |
|    for S (or, equivalently, the second cases should have 2*N - 2*k - 1
 | |
|    instead of N - k - 1).  Note also that in their definition of the
 | |
|    DFT, similarly to FFTW's, the exponent's sign is -1, but they
 | |
|    forgot to correspondingly multiply S (the sine terms) by -1.
 | |
| */
 | |
| 
 | |
| #include "reodft/reodft.h"
 | |
| 
 | |
| typedef struct {
 | |
|      solver super;
 | |
| } S;
 | |
| 
 | |
| typedef struct {
 | |
|      plan_rdft super;
 | |
|      plan *cld;
 | |
|      INT is, os;
 | |
|      INT n;
 | |
|      INT vl;
 | |
|      INT ivs, ovs;
 | |
|      rdft_kind kind;
 | |
| } P;
 | |
| 
 | |
| static DK(SQRT2, +1.4142135623730950488016887242096980785696718753769);
 | |
| 
 | |
| #define SGN_SET(x, i) ((i) % 2 ? -(x) : (x))
 | |
| 
 | |
| static void apply_re11(const plan *ego_, R *I, R *O)
 | |
| {
 | |
|      const P *ego = (const P *) ego_;
 | |
|      INT is = ego->is, os = ego->os;
 | |
|      INT i, n = ego->n, n2 = n/2;
 | |
|      INT iv, vl = ego->vl;
 | |
|      INT ivs = ego->ivs, ovs = ego->ovs;
 | |
|      R *buf;
 | |
| 
 | |
|      buf = (R *) MALLOC(sizeof(R) * n, BUFFERS);
 | |
| 
 | |
|      for (iv = 0; iv < vl; ++iv, I += ivs, O += ovs) {
 | |
| 	  {
 | |
| 	       INT m;
 | |
| 	       for (i = 0, m = n2; m < n; ++i, m += 4)
 | |
| 		    buf[i] = I[is * m];
 | |
| 	       for (; m < 2 * n; ++i, m += 4)
 | |
| 		    buf[i] = -I[is * (2*n - m - 1)];
 | |
| 	       for (; m < 3 * n; ++i, m += 4)
 | |
| 		    buf[i] = -I[is * (m - 2*n)];
 | |
| 	       for (; m < 4 * n; ++i, m += 4)
 | |
| 		    buf[i] = I[is * (4*n - m - 1)];
 | |
| 	       m -= 4 * n;
 | |
| 	       for (; i < n; ++i, m += 4)
 | |
| 		    buf[i] = I[is * m];
 | |
| 	  }
 | |
| 
 | |
| 	  { /* child plan: R2HC of size n */
 | |
| 	       plan_rdft *cld = (plan_rdft *) ego->cld;
 | |
| 	       cld->apply((plan *) cld, buf, buf);
 | |
| 	  }
 | |
| 	  
 | |
| 	  /* FIXME: strength-reduce loop by 4 to eliminate ugly sgn_set? */
 | |
| 	  for (i = 0; i + i + 1 < n2; ++i) {
 | |
| 	       INT k = i + i + 1;
 | |
| 	       E c1, s1;
 | |
| 	       E c2, s2;
 | |
| 	       c1 = buf[k];
 | |
| 	       c2 = buf[k + 1];
 | |
| 	       s2 = buf[n - (k + 1)];
 | |
| 	       s1 = buf[n - k];
 | |
| 	       
 | |
| 	       O[os * i] = SQRT2 * (SGN_SET(c1, (i+1)/2) +
 | |
| 				    SGN_SET(s1, i/2));
 | |
| 	       O[os * (n - (i+1))] = SQRT2 * (SGN_SET(c1, (n-i)/2) -
 | |
| 					      SGN_SET(s1, (n-(i+1))/2));
 | |
| 	       
 | |
| 	       O[os * (n2 - (i+1))] = SQRT2 * (SGN_SET(c2, (n2-i)/2) -
 | |
| 					       SGN_SET(s2, (n2-(i+1))/2));
 | |
| 	       O[os * (n2 + (i+1))] = SQRT2 * (SGN_SET(c2, (n2+i+2)/2) +
 | |
| 					       SGN_SET(s2, (n2+(i+1))/2));
 | |
| 	  }
 | |
| 	  if (i + i + 1 == n2) {
 | |
| 	       E c, s;
 | |
| 	       c = buf[n2];
 | |
| 	       s = buf[n - n2];
 | |
| 	       O[os * i] = SQRT2 * (SGN_SET(c, (i+1)/2) +
 | |
| 				    SGN_SET(s, i/2));
 | |
| 	       O[os * (n - (i+1))] = SQRT2 * (SGN_SET(c, (i+2)/2) +
 | |
| 					      SGN_SET(s, (i+1)/2));
 | |
| 	  }
 | |
| 	  O[os * n2] = SQRT2 * SGN_SET(buf[0], (n2+1)/2);
 | |
|      }
 | |
| 
 | |
|      X(ifree)(buf);
 | |
| }
 | |
| 
 | |
| /* like for rodft01, rodft11 is obtained from redft11 by
 | |
|    reversing the input and flipping the sign of every other output. */
 | |
| static void apply_ro11(const plan *ego_, R *I, R *O)
 | |
| {
 | |
|      const P *ego = (const P *) ego_;
 | |
|      INT is = ego->is, os = ego->os;
 | |
|      INT i, n = ego->n, n2 = n/2;
 | |
|      INT iv, vl = ego->vl;
 | |
|      INT ivs = ego->ivs, ovs = ego->ovs;
 | |
|      R *buf;
 | |
| 
 | |
|      buf = (R *) MALLOC(sizeof(R) * n, BUFFERS);
 | |
| 
 | |
|      for (iv = 0; iv < vl; ++iv, I += ivs, O += ovs) {
 | |
| 	  {
 | |
| 	       INT m;
 | |
| 	       for (i = 0, m = n2; m < n; ++i, m += 4)
 | |
| 		    buf[i] = I[is * (n - 1 - m)];
 | |
| 	       for (; m < 2 * n; ++i, m += 4)
 | |
| 		    buf[i] = -I[is * (m - n)];
 | |
| 	       for (; m < 3 * n; ++i, m += 4)
 | |
| 		    buf[i] = -I[is * (3*n - 1 - m)];
 | |
| 	       for (; m < 4 * n; ++i, m += 4)
 | |
| 		    buf[i] = I[is * (m - 3*n)];
 | |
| 	       m -= 4 * n;
 | |
| 	       for (; i < n; ++i, m += 4)
 | |
| 		    buf[i] = I[is * (n - 1 - m)];
 | |
| 	  }
 | |
| 
 | |
| 	  { /* child plan: R2HC of size n */
 | |
| 	       plan_rdft *cld = (plan_rdft *) ego->cld;
 | |
| 	       cld->apply((plan *) cld, buf, buf);
 | |
| 	  }
 | |
| 	  
 | |
| 	  /* FIXME: strength-reduce loop by 4 to eliminate ugly sgn_set? */
 | |
| 	  for (i = 0; i + i + 1 < n2; ++i) {
 | |
| 	       INT k = i + i + 1;
 | |
| 	       INT j;
 | |
| 	       E c1, s1;
 | |
| 	       E c2, s2;
 | |
| 	       c1 = buf[k];
 | |
| 	       c2 = buf[k + 1];
 | |
| 	       s2 = buf[n - (k + 1)];
 | |
| 	       s1 = buf[n - k];
 | |
| 	       
 | |
| 	       O[os * i] = SQRT2 * (SGN_SET(c1, (i+1)/2 + i) +
 | |
| 				    SGN_SET(s1, i/2 + i));
 | |
| 	       O[os * (n - (i+1))] = SQRT2 * (SGN_SET(c1, (n-i)/2 + i) -
 | |
| 					      SGN_SET(s1, (n-(i+1))/2 + i));
 | |
| 	       
 | |
| 	       j = n2 - (i+1);
 | |
| 	       O[os * j] = SQRT2 * (SGN_SET(c2, (n2-i)/2 + j) -
 | |
| 				    SGN_SET(s2, (n2-(i+1))/2 + j));
 | |
| 	       O[os * (n2 + (i+1))] = SQRT2 * (SGN_SET(c2, (n2+i+2)/2 + j) +
 | |
| 					       SGN_SET(s2, (n2+(i+1))/2 + j));
 | |
| 	  }
 | |
| 	  if (i + i + 1 == n2) {
 | |
| 	       E c, s;
 | |
| 	       c = buf[n2];
 | |
| 	       s = buf[n - n2];
 | |
| 	       O[os * i] = SQRT2 * (SGN_SET(c, (i+1)/2 + i) +
 | |
| 				    SGN_SET(s, i/2 + i));
 | |
| 	       O[os * (n - (i+1))] = SQRT2 * (SGN_SET(c, (i+2)/2 + i) +
 | |
| 					      SGN_SET(s, (i+1)/2 + i));
 | |
| 	  }
 | |
| 	  O[os * n2] = SQRT2 * SGN_SET(buf[0], (n2+1)/2 + n2);
 | |
|      }
 | |
| 
 | |
|      X(ifree)(buf);
 | |
| }
 | |
| 
 | |
| static void awake(plan *ego_, enum wakefulness wakefulness)
 | |
| {
 | |
|      P *ego = (P *) ego_;
 | |
|      X(plan_awake)(ego->cld, wakefulness);
 | |
| }
 | |
| 
 | |
| static void destroy(plan *ego_)
 | |
| {
 | |
|      P *ego = (P *) ego_;
 | |
|      X(plan_destroy_internal)(ego->cld);
 | |
| }
 | |
| 
 | |
| static void print(const plan *ego_, printer *p)
 | |
| {
 | |
|      const P *ego = (const P *) ego_;
 | |
|      p->print(p, "(%se-r2hc-odd-%D%v%(%p%))",
 | |
| 	      X(rdft_kind_str)(ego->kind), ego->n, ego->vl, ego->cld);
 | |
| }
 | |
| 
 | |
| static int applicable0(const solver *ego_, const problem *p_)
 | |
| {
 | |
|      const problem_rdft *p = (const problem_rdft *) p_;
 | |
|      UNUSED(ego_);
 | |
| 
 | |
|      return (1
 | |
| 	     && p->sz->rnk == 1
 | |
| 	     && p->vecsz->rnk <= 1
 | |
| 	     && p->sz->dims[0].n % 2 == 1
 | |
| 	     && (p->kind[0] == REDFT11 || p->kind[0] == RODFT11)
 | |
| 	  );
 | |
| }
 | |
| 
 | |
| static int applicable(const solver *ego, const problem *p, const planner *plnr)
 | |
| {
 | |
|      return (!NO_SLOWP(plnr) && applicable0(ego, p));
 | |
| }
 | |
| 
 | |
| static plan *mkplan(const solver *ego_, const problem *p_, planner *plnr)
 | |
| {
 | |
|      P *pln;
 | |
|      const problem_rdft *p;
 | |
|      plan *cld;
 | |
|      R *buf;
 | |
|      INT n;
 | |
|      opcnt ops;
 | |
| 
 | |
|      static const plan_adt padt = {
 | |
| 	  X(rdft_solve), awake, print, destroy
 | |
|      };
 | |
| 
 | |
|      if (!applicable(ego_, p_, plnr))
 | |
|           return (plan *)0;
 | |
| 
 | |
|      p = (const problem_rdft *) p_;
 | |
| 
 | |
|      n = p->sz->dims[0].n;
 | |
|      buf = (R *) MALLOC(sizeof(R) * n, BUFFERS);
 | |
| 
 | |
|      cld = X(mkplan_d)(plnr, X(mkproblem_rdft_1_d)(X(mktensor_1d)(n, 1, 1),
 | |
|                                                    X(mktensor_0d)(),
 | |
|                                                    buf, buf, R2HC));
 | |
|      X(ifree)(buf);
 | |
|      if (!cld)
 | |
|           return (plan *)0;
 | |
| 
 | |
|      pln = MKPLAN_RDFT(P, &padt, p->kind[0]==REDFT11 ? apply_re11:apply_ro11);
 | |
|      pln->n = n;
 | |
|      pln->is = p->sz->dims[0].is;
 | |
|      pln->os = p->sz->dims[0].os;
 | |
|      pln->cld = cld;
 | |
|      pln->kind = p->kind[0];
 | |
|      
 | |
|      X(tensor_tornk1)(p->vecsz, &pln->vl, &pln->ivs, &pln->ovs);
 | |
|      
 | |
|      X(ops_zero)(&ops);
 | |
|      ops.add = n - 1;
 | |
|      ops.mul = n;
 | |
|      ops.other = 4*n;
 | |
| 
 | |
|      X(ops_zero)(&pln->super.super.ops);
 | |
|      X(ops_madd2)(pln->vl, &ops, &pln->super.super.ops);
 | |
|      X(ops_madd2)(pln->vl, &cld->ops, &pln->super.super.ops);
 | |
| 
 | |
|      return &(pln->super.super);
 | |
| }
 | |
| 
 | |
| /* constructor */
 | |
| static solver *mksolver(void)
 | |
| {
 | |
|      static const solver_adt sadt = { PROBLEM_RDFT, mkplan, 0 };
 | |
|      S *slv = MKSOLVER(S, &sadt);
 | |
|      return &(slv->super);
 | |
| }
 | |
| 
 | |
| void X(reodft11e_r2hc_odd_register)(planner *p)
 | |
| {
 | |
|      REGISTER_SOLVER(p, mksolver());
 | |
| }
 | 
