hc2r_11.c

最新的FFT程序

/*
 * Copyright (c) 2003, 2006 Matteo Frigo
 * Copyright (c) 2003, 2006 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
 *
 */

/* This file was automatically generated --- DO NOT EDIT */
/* Generated on Fri Jan 27 20:39:11 EST 2006 */

#include "codelet-rdft.h"

#ifdef HAVE_FMA

/* Generated by: ../../../genfft/gen_hc2r -fma -reorder-insns -schedule-for-pipeline -compact -variables 4 -pipeline-latency 4 -sign 1 -n 11 -name hc2r_11 -include hc2r.h */

/*
 * This function contains 60 FP additions, 56 FP multiplications,
 * (or, 4 additions, 0 multiplications, 56 fused multiply/add),
 * 53 stack variables, and 22 memory accesses
 */
/*
 * Generator Id's : 
 * $Id: algsimp.ml,v 1.8 2006-01-05 03:04:27 stevenj Exp $
 * $Id: fft.ml,v 1.4 2006-01-05 03:04:27 stevenj Exp $
 * $Id: gen_hc2r.ml,v 1.18 2006-01-05 03:04:27 stevenj Exp $
 */

#include "hc2r.h"

static void hc2r_11(const R *ri, const R *ii, R *O, stride ris, stride iis, stride os, INT v, INT ivs, INT ovs)
{
     DK(KP1_979642883, +1.979642883761865464752184075553437574753038744);
     DK(KP1_918985947, +1.918985947228994779780736114132655398124909697);
     DK(KP876768831, +0.876768831002589333891339807079336796764054852);
     DK(KP918985947, +0.918985947228994779780736114132655398124909697);
     DK(KP778434453, +0.778434453334651800608337670740821884709317477);
     DK(KP2_000000000, +2.000000000000000000000000000000000000000000000);
     DK(KP634356270, +0.634356270682424498893150776899916060542806975);
     DK(KP342584725, +0.342584725681637509502641509861112333758894680);
     DK(KP830830026, +0.830830026003772851058548298459246407048009821);
     DK(KP715370323, +0.715370323453429719112414662767260662417897278);
     DK(KP521108558, +0.521108558113202722944698153526659300680427422);
     INT i;
     for (i = v; i > 0; i = i - 1, ri = ri + ivs, ii = ii + ivs, O = O + ovs, MAKE_VOLATILE_STRIDE(ris), MAKE_VOLATILE_STRIDE(iis), MAKE_VOLATILE_STRIDE(os)) {
	  E Tf, Tq, Tt, Tu;
	  {
	       E T1, Td, Th, Te, Tg, T2, Ts, TK, TB, TT, Tj, T6, T3, T4, T5;
	       E Tr;
	       T1 = ri[0];
	       Td = ii[WS(iis, 3)];
	       Th = ii[WS(iis, 5)];
	       Te = ii[WS(iis, 2)];
	       Tf = ii[WS(iis, 4)];
	       Tg = ii[WS(iis, 1)];
	       Tr = FMA(KP521108558, Td, Th);
	       T2 = ri[WS(ris, 1)];
	       {
		    E TJ, TA, TS, Ti;
		    TJ = FMA(KP521108558, Tf, Td);
		    TA = FNMS(KP521108558, Te, Tf);
		    TS = FMS(KP521108558, Tg, Te);
		    Ti = FMA(KP521108558, Th, Tg);
		    Ts = FNMS(KP715370323, Tr, Te);
		    TK = FMA(KP715370323, TJ, Tg);
		    TB = FMA(KP715370323, TA, Th);
		    TT = FMA(KP715370323, TS, Td);
		    Tj = FMA(KP715370323, Ti, Tf);
		    T6 = ri[WS(ris, 5)];
	       }
	       T3 = ri[WS(ris, 2)];
	       T4 = ri[WS(ris, 3)];
	       T5 = ri[WS(ris, 4)];
	       {
		    E TG, Tx, To, Tl, Tb, TU, TQ, TP, Ta;
		    {
			 E Tk, TE, Tv, T8;
			 Tk = FMA(KP830830026, Tj, Te);
			 TE = FNMS(KP342584725, T3, T6);
			 Tv = FNMS(KP342584725, T2, T4);
			 T8 = FNMS(KP342584725, T4, T3);
			 {
			      E T7, Tm, TN, TF;
			      T7 = T2 + T3 + T4 + T5 + T6;
			      Tm = FNMS(KP342584725, T5, T2);
			      TN = FNMS(KP342584725, T6, T5);
			      TF = FNMS(KP634356270, TE, T2);
			      {
				   E Tw, T9, Tn, TO;
				   Tw = FNMS(KP634356270, Tv, T6);
				   T9 = FNMS(KP634356270, T8, T5);
				   O[0] = FMA(KP2_000000000, T7, T1);
				   Tn = FNMS(KP634356270, Tm, T3);
				   TO = FNMS(KP634356270, TN, T4);
				   TG = FNMS(KP778434453, TF, T4);
				   Tx = FNMS(KP778434453, Tw, T5);
				   Ta = FNMS(KP778434453, T9, T2);
				   To = FNMS(KP778434453, Tn, T6);
				   TP = FNMS(KP778434453, TO, T3);
				   Tl = FMA(KP918985947, Tk, Td);
			      }
			 }
		    }
		    Tb = FNMS(KP876768831, Ta, T6);
		    TU = FNMS(KP830830026, TT, Tf);
		    TQ = FNMS(KP876768831, TP, T2);
		    {
			 E TI, TL, Ty, TC;
			 {
			      E Tc, TV, TR, TH;
			      TH = FNMS(KP876768831, TG, T5);
			      Tc = FNMS(KP1_918985947, Tb, T1);
			      TV = FNMS(KP918985947, TU, Th);
			      TR = FNMS(KP1_918985947, TQ, T1);
			      TI = FNMS(KP1_918985947, TH, T1);
			      O[WS(os, 10)] = FMA(KP1_979642883, Tl, Tc);
			      O[WS(os, 1)] = FNMS(KP1_979642883, Tl, Tc);
			      O[WS(os, 6)] = FMA(KP1_979642883, TV, TR);
			      O[WS(os, 5)] = FNMS(KP1_979642883, TV, TR);
			      TL = FNMS(KP830830026, TK, Th);
			 }
			 Ty = FNMS(KP876768831, Tx, T3);
			 TC = FNMS(KP830830026, TB, Td);
			 {
			      E TM, Tz, TD, Tp;
			      Tp = FNMS(KP876768831, To, T4);
			      TM = FMA(KP918985947, TL, Te);
			      Tz = FNMS(KP1_918985947, Ty, T1);
			      TD = FNMS(KP918985947, TC, Tg);
			      Tq = FNMS(KP1_918985947, Tp, T1);
			      O[WS(os, 4)] = FMA(KP1_979642883, TM, TI);
			      O[WS(os, 7)] = FNMS(KP1_979642883, TM, TI);
			      O[WS(os, 8)] = FMA(KP1_979642883, TD, Tz);
			      O[WS(os, 3)] = FNMS(KP1_979642883, TD, Tz);
			      Tt = FMA(KP830830026, Ts, Tg);
			 }
		    }
	       }
	  }
	  Tu = FNMS(KP918985947, Tt, Tf);
	  O[WS(os, 2)] = FMA(KP1_979642883, Tu, Tq);
	  O[WS(os, 9)] = FNMS(KP1_979642883, Tu, Tq);
     }
}

static const khc2r_desc desc = { 11, "hc2r_11", {4, 0, 56, 0}, &GENUS, 0, 0, 0, 0, 0 };

void X(codelet_hc2r_11) (planner *p) {
     X(khc2r_register) (p, hc2r_11, &desc);
}

#else				/* HAVE_FMA */

/* Generated by: ../../../genfft/gen_hc2r -compact -variables 4 -pipeline-latency 4 -sign 1 -n 11 -name hc2r_11 -include hc2r.h */

/*
 * This function contains 60 FP additions, 51 FP multiplications,
 * (or, 19 additions, 10 multiplications, 41 fused multiply/add),
 * 33 stack variables, and 22 memory accesses
 */
/*
 * Generator Id's : 
 * $Id: algsimp.ml,v 1.8 2006-01-05 03:04:27 stevenj Exp $
 * $Id: fft.ml,v 1.4 2006-01-05 03:04:27 stevenj Exp $
 * $Id: gen_hc2r.ml,v 1.18 2006-01-05 03:04:27 stevenj Exp $
 */

#include "hc2r.h"

static void hc2r_11(const R *ri, const R *ii, R *O, stride ris, stride iis, stride os, INT v, INT ivs, INT ovs)
{
     DK(KP2_000000000, +2.000000000000000000000000000000000000000000000);
     DK(KP1_918985947, +1.918985947228994779780736114132655398124909697);
     DK(KP1_309721467, +1.309721467890570128113850144932587106367582399);
     DK(KP284629676, +0.284629676546570280887585337232739337582102722);
     DK(KP830830026, +0.830830026003772851058548298459246407048009821);
     DK(KP1_682507065, +1.682507065662362337723623297838735435026584997);
     DK(KP563465113, +0.563465113682859395422835830693233798071555798);
     DK(KP1_511499148, +1.511499148708516567548071687944688840359434890);
     DK(KP1_979642883, +1.979642883761865464752184075553437574753038744);
     DK(KP1_819263990, +1.819263990709036742823430766158056920120482102);
     DK(KP1_081281634, +1.081281634911195164215271908637383390863541216);
     INT i;
     for (i = v; i > 0; i = i - 1, ri = ri + ivs, ii = ii + ivs, O = O + ovs, MAKE_VOLATILE_STRIDE(ris), MAKE_VOLATILE_STRIDE(iis), MAKE_VOLATILE_STRIDE(os)) {
	  E Td, Tl, Tf, Th, Tj, T1, T2, T6, T5, T4, T3, T7, Tk, Te, Tg;
	  E Ti;
	  {
	       E T8, Tc, T9, Ta, Tb;
	       T8 = ii[WS(iis, 2)];
	       Tc = ii[WS(iis, 1)];
	       T9 = ii[WS(iis, 4)];
	       Ta = ii[WS(iis, 5)];
	       Tb = ii[WS(iis, 3)];
	       Td = FMA(KP1_081281634, T8, KP1_819263990 * T9) + FNMA(KP1_979642883, Ta, KP1_511499148 * Tb) - (KP563465113 * Tc);
	       Tl = FMA(KP1_979642883, T8, KP1_819263990 * Ta) + FNMA(KP563465113, T9, KP1_081281634 * Tb) - (KP1_511499148 * Tc);
	       Tf = FMA(KP563465113, T8, KP1_819263990 * Tb) + FNMA(KP1_511499148, Ta, KP1_081281634 * T9) - (KP1_979642883 * Tc);
	       Th = FMA(KP1_081281634, Tc, KP1_819263990 * T8) + FMA(KP1_979642883, Tb, KP1_511499148 * T9) + (KP563465113 * Ta);
	       Tj = FMA(KP563465113, Tb, KP1_979642883 * T9) + FNMS(KP1_511499148, T8, KP1_081281634 * Ta) - (KP1_819263990 * Tc);
	  }
	  T1 = ri[0];
	  T2 = ri[WS(ris, 1)];
	  T6 = ri[WS(ris, 5)];
	  T5 = ri[WS(ris, 4)];
	  T4 = ri[WS(ris, 3)];
	  T3 = ri[WS(ris, 2)];
	  T7 = FMA(KP1_682507065, T3, T1) + FNMS(KP284629676, T6, KP830830026 * T5) + FNMA(KP1_309721467, T4, KP1_918985947 * T2);
	  Tk = FMA(KP1_682507065, T4, T1) + FNMS(KP1_918985947, T5, KP830830026 * T6) + FNMA(KP284629676, T3, KP1_309721467 * T2);
	  Te = FMA(KP830830026, T4, T1) + FNMS(KP1_309721467, T6, KP1_682507065 * T5) + FNMA(KP1_918985947, T3, KP284629676 * T2);
	  Tg = FMA(KP1_682507065, T2, T1) + FNMS(KP1_918985947, T6, KP830830026 * T3) + FNMA(KP1_309721467, T5, KP284629676 * T4);
	  Ti = FMA(KP830830026, T2, T1) + FNMS(KP284629676, T5, KP1_682507065 * T6) + FNMA(KP1_918985947, T4, KP1_309721467 * T3);
	  O[WS(os, 6)] = T7 - Td;
	  O[WS(os, 8)] = Te - Tf;
	  O[WS(os, 4)] = Tk + Tl;
	  O[WS(os, 5)] = T7 + Td;
	  O[WS(os, 7)] = Tk - Tl;
	  O[WS(os, 2)] = Ti + Tj;
	  O[WS(os, 3)] = Te + Tf;
	  O[WS(os, 10)] = Tg + Th;
	  O[WS(os, 1)] = Tg - Th;
	  O[WS(os, 9)] = Ti - Tj;
	  O[0] = FMA(KP2_000000000, T2 + T3 + T4 + T5 + T6, T1);
     }
}

static const khc2r_desc desc = { 11, "hc2r_11", {19, 10, 41, 0}, &GENUS, 0, 0, 0, 0, 0 };

void X(codelet_hc2r_11) (planner *p) {
     X(khc2r_register) (p, hc2r_11, &desc);
}

#endif				/* HAVE_FMA */