n2sv_32.c

快速fft变换

/*
 * Copyright (c) 2003, 2007-8 Matteo Frigo
 * Copyright (c) 2003, 2007-8 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 Sat Nov 15 20:46:32 EST 2008 */

#include "codelet-dft.h"

#ifdef HAVE_FMA

/* Generated by: ../../../genfft/gen_notw -fma -reorder-insns -schedule-for-pipeline -simd -compact -variables 4 -pipeline-latency 8 -n 32 -name n2sv_32 -with-ostride 1 -include n2s.h -store-multiple 4 */

/*
 * This function contains 372 FP additions, 136 FP multiplications,
 * (or, 236 additions, 0 multiplications, 136 fused multiply/add),
 * 194 stack variables, 7 constants, and 144 memory accesses
 */
#include "n2s.h"

static void n2sv_32(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs)
{
     DVK(KP980785280, +0.980785280403230449126182236134239036973933731);
     DVK(KP198912367, +0.198912367379658006911597622644676228597850501);
     DVK(KP831469612, +0.831469612302545237078788377617905756738560812);
     DVK(KP668178637, +0.668178637919298919997757686523080761552472251);
     DVK(KP923879532, +0.923879532511286756128183189396788286822416626);
     DVK(KP707106781, +0.707106781186547524400844362104849039284835938);
     DVK(KP414213562, +0.414213562373095048801688724209698078569671875);
     INT i;
     for (i = v; i > 0; i = i - (2 * VL), ri = ri + ((2 * VL) * ivs), ii = ii + ((2 * VL) * ivs), ro = ro + ((2 * VL) * ovs), io = io + ((2 * VL) * ovs), MAKE_VOLATILE_STRIDE(is), MAKE_VOLATILE_STRIDE(os)) {
	  V T61, T62, T63, T64, T65, T66, T67, T68, T69, T6a, T6b, T6c, T6d, T6e, T6f;
	  V T6g, T6h, T6i, T6j, T6k, T6l, T6m, T6n, T6o, T6p, T6q, T6r, T6s, T6t, T6u;
	  V T6v, T6w, T3g, T3f, T6x, T6y, T6z, T6A, T6B, T6C, T6D, T6E, T4p, T49, T4l;
	  V T4j, T6F, T6G, T6H, T6I, T6J, T6K, T6L, T6M, T3n, T3b, T3r, T3l, T3o, T3e;
	  V T4q, T4o, T4k, T4g, T3h, T3p;
	  {
	       V T2T, T3T, T4r, T7, T3t, T1z, T18, T4Z, Te, T50, T1f, T4s, T1G, T3U, T2W;
	       V T3u, Tm, T1n, T3X, T3y, T2Z, T1O, T53, T4w, Tt, T1u, T3W, T3B, T2Y, T1V;
	       V T52, T4z, T3O, T2t, T3L, T2K, TZ, T5F, T4R, T5k, T5j, T4W, T5I, T5X, T2E;
	       V T3M, T2N, T3P, T3H, T22, T3E, T2j, T4G, T5h, TK, T5A, T5D, T5W, T2d, T3F;
	       V T4L, T5g, T3I, T2m;
	       {
		    V T1L, T1j, T1k, T1l, T4v, T1K, T3w;
		    {
			 V T1, T2, T12, T13, T4, T5, T15, T16;
			 T1 = LD(&(ri[0]), ivs, &(ri[0]));
			 T2 = LD(&(ri[WS(is, 16)]), ivs, &(ri[0]));
			 T12 = LD(&(ii[0]), ivs, &(ii[0]));
			 T13 = LD(&(ii[WS(is, 16)]), ivs, &(ii[0]));
			 T4 = LD(&(ri[WS(is, 8)]), ivs, &(ri[0]));
			 T5 = LD(&(ri[WS(is, 24)]), ivs, &(ri[0]));
			 T15 = LD(&(ii[WS(is, 8)]), ivs, &(ii[0]));
			 T16 = LD(&(ii[WS(is, 24)]), ivs, &(ii[0]));
			 {
			      V Tb, T1A, Ta, T1B, T1b, Tc, T1c, T1d;
			      {
				   V T8, T1x, T3, T2R, T14, T2S, T6, T1y, T17, T9, T19, T1a;
				   T8 = LD(&(ri[WS(is, 4)]), ivs, &(ri[0]));
				   T1x = VSUB(T1, T2);
				   T3 = VADD(T1, T2);
				   T2R = VSUB(T12, T13);
				   T14 = VADD(T12, T13);
				   T2S = VSUB(T4, T5);
				   T6 = VADD(T4, T5);
				   T1y = VSUB(T15, T16);
				   T17 = VADD(T15, T16);
				   T9 = LD(&(ri[WS(is, 20)]), ivs, &(ri[0]));
				   T19 = LD(&(ii[WS(is, 4)]), ivs, &(ii[0]));
				   T1a = LD(&(ii[WS(is, 20)]), ivs, &(ii[0]));
				   Tb = LD(&(ri[WS(is, 28)]), ivs, &(ri[0]));
				   T2T = VSUB(T2R, T2S);
				   T3T = VADD(T2S, T2R);
				   T4r = VSUB(T3, T6);
				   T7 = VADD(T3, T6);
				   T3t = VSUB(T1x, T1y);
				   T1z = VADD(T1x, T1y);
				   T18 = VADD(T14, T17);
				   T4Z = VSUB(T14, T17);
				   T1A = VSUB(T8, T9);
				   Ta = VADD(T8, T9);
				   T1B = VSUB(T19, T1a);
				   T1b = VADD(T19, T1a);
				   Tc = LD(&(ri[WS(is, 12)]), ivs, &(ri[0]));
				   T1c = LD(&(ii[WS(is, 28)]), ivs, &(ii[0]));
				   T1d = LD(&(ii[WS(is, 12)]), ivs, &(ii[0]));
			      }
			      {
				   V Ti, T1I, T1J, Tl;
				   {
					V T1h, T1C, T2U, T1D, Td, T1E, T1e, T1i, Tg, Th;
					Tg = LD(&(ri[WS(is, 2)]), ivs, &(ri[0]));
					Th = LD(&(ri[WS(is, 18)]), ivs, &(ri[0]));
					T1h = LD(&(ii[WS(is, 2)]), ivs, &(ii[0]));
					T1C = VADD(T1A, T1B);
					T2U = VSUB(T1B, T1A);
					T1D = VSUB(Tb, Tc);
					Td = VADD(Tb, Tc);
					T1E = VSUB(T1c, T1d);
					T1e = VADD(T1c, T1d);
					T1L = VSUB(Tg, Th);
					Ti = VADD(Tg, Th);
					T1i = LD(&(ii[WS(is, 18)]), ivs, &(ii[0]));
					{
					     V T2V, T1F, Tj, Tk;
					     Tj = LD(&(ri[WS(is, 10)]), ivs, &(ri[0]));
					     Tk = LD(&(ri[WS(is, 26)]), ivs, &(ri[0]));
					     Te = VADD(Ta, Td);
					     T50 = VSUB(Td, Ta);
					     T2V = VADD(T1D, T1E);
					     T1F = VSUB(T1D, T1E);
					     T1f = VADD(T1b, T1e);
					     T4s = VSUB(T1b, T1e);
					     T1j = VADD(T1h, T1i);
					     T1I = VSUB(T1h, T1i);
					     T1J = VSUB(Tj, Tk);
					     Tl = VADD(Tj, Tk);
					     T1G = VADD(T1C, T1F);
					     T3U = VSUB(T1F, T1C);
					     T2W = VADD(T2U, T2V);
					     T3u = VSUB(T2U, T2V);
					     T1k = LD(&(ii[WS(is, 10)]), ivs, &(ii[0]));
					     T1l = LD(&(ii[WS(is, 26)]), ivs, &(ii[0]));
					}
				   }
				   T4v = VSUB(Ti, Tl);
				   Tm = VADD(Ti, Tl);
				   T1K = VSUB(T1I, T1J);
				   T3w = VADD(T1J, T1I);
			      }
			 }
		    }
		    {
			 V T1r, T1S, T1q, T1s, T4x, T1R, T3z;
			 {
			      V Tp, T1P, T1Q, Ts;
			      {
				   V Tn, To, T1o, T1M, T1m, T1p;
				   Tn = LD(&(ri[WS(is, 30)]), ivs, &(ri[0]));
				   To = LD(&(ri[WS(is, 14)]), ivs, &(ri[0]));
				   T1o = LD(&(ii[WS(is, 30)]), ivs, &(ii[0]));
				   T1M = VSUB(T1k, T1l);
				   T1m = VADD(T1k, T1l);
				   T1p = LD(&(ii[WS(is, 14)]), ivs, &(ii[0]));
				   {
					V Tq, Tr, T3x, T1N, T4u;
					Tq = LD(&(ri[WS(is, 6)]), ivs, &(ri[0]));
					Tr = LD(&(ri[WS(is, 22)]), ivs, &(ri[0]));
					T1r = LD(&(ii[WS(is, 6)]), ivs, &(ii[0]));
					T1S = VSUB(Tn, To);
					Tp = VADD(Tn, To);
					T3x = VSUB(T1L, T1M);
					T1N = VADD(T1L, T1M);
					T4u = VSUB(T1j, T1m);
					T1n = VADD(T1j, T1m);
					T1P = VSUB(T1o, T1p);
					T1q = VADD(T1o, T1p);
					T1Q = VSUB(Tq, Tr);
					Ts = VADD(Tq, Tr);
					T3X = VFNMS(LDK(KP414213562), T3w, T3x);
					T3y = VFMA(LDK(KP414213562), T3x, T3w);
					T2Z = VFMA(LDK(KP414213562), T1K, T1N);
					T1O = VFNMS(LDK(KP414213562), T1N, T1K);
					T53 = VADD(T4v, T4u);
					T4w = VSUB(T4u, T4v);
					T1s = LD(&(ii[WS(is, 22)]), ivs, &(ii[0]));
				   }
			      }
			      T4x = VSUB(Tp, Ts);
			      Tt = VADD(Tp, Ts);
			      T1R = VSUB(T1P, T1Q);
			      T3z = VADD(T1Q, T1P);
			 }
			 {
			      V T4S, T5G, T2y, T2L, T4V, T5H, T2D, T2M;
			      {
				   V T2G, TN, T4N, T2r, T2s, TQ, T2A, T4O, T2J, T2x, TU, T4T, T2w, T2z, TX;
				   V T2B, T2H, T2I, TR;
				   {
					V TL, TM, T2p, T1T, T1t, T2q;
					TL = LD(&(ri[WS(is, 31)]), ivs, &(ri[WS(is, 1)]));
					TM = LD(&(ri[WS(is, 15)]), ivs, &(ri[WS(is, 1)]));
					T2p = LD(&(ii[WS(is, 31)]), ivs, &(ii[WS(is, 1)]));
					T1T = VSUB(T1r, T1s);
					T1t = VADD(T1r, T1s);
					T2q = LD(&(ii[WS(is, 15)]), ivs, &(ii[WS(is, 1)]));
					{
					     V TO, TP, T3A, T1U, T4y;
					     TO = LD(&(ri[WS(is, 7)]), ivs, &(ri[WS(is, 1)]));
					     TP = LD(&(ri[WS(is, 23)]), ivs, &(ri[WS(is, 1)]));
					     T2H = LD(&(ii[WS(is, 7)]), ivs, &(ii[WS(is, 1)]));
					     T2G = VSUB(TL, TM);
					     TN = VADD(TL, TM);
					     T3A = VSUB(T1S, T1T);
					     T1U = VADD(T1S, T1T);
					     T4y = VSUB(T1q, T1t);
					     T1u = VADD(T1q, T1t);
					     T4N = VADD(T2p, T2q);
					     T2r = VSUB(T2p, T2q);
					     T2s = VSUB(TO, TP);
					     TQ = VADD(TO, TP);
					     T3W = VFMA(LDK(KP414213562), T3z, T3A);
					     T3B = VFNMS(LDK(KP414213562), T3A, T3z);
					     T2Y = VFNMS(LDK(KP414213562), T1R, T1U);
					     T1V = VFMA(LDK(KP414213562), T1U, T1R);
					     T52 = VSUB(T4x, T4y);
					     T4z = VADD(T4x, T4y);
					     T2I = LD(&(ii[WS(is, 23)]), ivs, &(ii[WS(is, 1)]));
					}
				   }
				   {
					V TS, TT, T2u, T2v, TV, TW;
					TS = LD(&(ri[WS(is, 3)]), ivs, &(ri[WS(is, 1)]));
					TT = LD(&(ri[WS(is, 19)]), ivs, &(ri[WS(is, 1)]));
					T2u = LD(&(ii[WS(is, 3)]), ivs, &(ii[WS(is, 1)]));
					T2v = LD(&(ii[WS(is, 19)]), ivs, &(ii[WS(is, 1)]));
					TV = LD(&(ri[WS(is, 27)]), ivs, &(ri[WS(is, 1)]));
					TW = LD(&(ri[WS(is, 11)]), ivs, &(ri[WS(is, 1)]));
					T2A = LD(&(ii[WS(is, 27)]), ivs, &(ii[WS(is, 1)]));
					T4O = VADD(T2H, T2I);
					T2J = VSUB(T2H, T2I);
					T2x = VSUB(TS, TT);
					TU = VADD(TS, TT);
					T4T = VADD(T2u, T2v);
					T2w = VSUB(T2u, T2v);
					T2z = VSUB(TV, TW);
					TX = VADD(TV, TW);
					T2B = LD(&(ii[WS(is, 11)]), ivs, &(ii[WS(is, 1)]));
				   }
				   T3O = VADD(T2s, T2r);
				   T2t = VSUB(T2r, T2s);
				   T3L = VSUB(T2G, T2J);
				   T2K = VADD(T2G, T2J);
				   T4S = VSUB(TN, TQ);
				   TR = VADD(TN, TQ);
				   {
					V T4P, T4Q, TY, T4U, T2C;
					T5G = VADD(T4N, T4O);
					T4P = VSUB(T4N, T4O);
					T4Q = VSUB(TX, TU);
					TY = VADD(TU, TX);
					T4U = VADD(T2A, T2B);
					T2C = VSUB(T2A, T2B);
					T2y = VSUB(T2w, T2x);
					T2L = VADD(T2x, T2w);
					TZ = VADD(TR, TY);
					T5F = VSUB(TR, TY);
					T4V = VSUB(T4T, T4U);
					T5H = VADD(T4T, T4U);
					T2D = VADD(T2z, T2C);
					T2M = VSUB(T2z, T2C);
					T4R = VSUB(T4P, T4Q);
					T5k = VADD(T4Q, T4P);
				   }
			      }
			      {
				   V T2f, Ty, T23, T4C, T20, T21, TB, T4D, T2i, T26, TF, T24, TG, TH, T29;
				   V T2a;
				   {
					V T1Y, T1Z, Tz, TA, T2g, T2h, Tw, Tx, TD, TE;
					Tw = LD(&(ri[WS(is, 1)]), ivs, &(ri[WS(is, 1)]));
					Tx = LD(&(ri[WS(is, 17)]), ivs, &(ri[WS(is, 1)]));
					T5j = VADD(T4S, T4V);
					T4W = VSUB(T4S, T4V);
					T5I = VSUB(T5G, T5H);
					T5X = VADD(T5G, T5H);
					T2E = VADD(T2y, T2D);
					T3M = VSUB(T2D, T2y);
					T2N = VADD(T2L, T2M);
					T3P = VSUB(T2L, T2M);
					T2f = VSUB(Tw, Tx);
					Ty = VADD(Tw, Tx);
					T1Y = LD(&(ii[WS(is, 1)]), ivs, &(ii[WS(is, 1)]));
					T1Z = LD(&(ii[WS(is, 17)]), ivs, &(ii[WS(is, 1)]));
					Tz = LD(&(ri[WS(is, 9)]), ivs, &(ri[WS(is, 1)]));
					TA = LD(&(ri[WS(is, 25)]), ivs, &(ri[WS(is, 1)]));
					T2g = LD(&(ii[WS(is, 9)]), ivs, &(ii[WS(is, 1)]));
					T2h = LD(&(ii[WS(is, 25)]), ivs, &(ii[WS(is, 1)]));
					TD = LD(&(ri[WS(is, 5)]), ivs, &(ri[WS(is, 1)]));
					TE = LD(&(ri[WS(is, 21)]), ivs, &(ri[WS(is, 1)]));
					T23 = LD(&(ii[WS(is, 5)]), ivs, &(ii[WS(is, 1)]));
					T4C = VADD(T1Y, T1Z);
					T20 = VSUB(T1Y, T1Z);
					T21 = VSUB(Tz, TA);
					TB = VADD(Tz, TA);
					T4D = VADD(T2g, T2h);
					T2i = VSUB(T2g, T2h);
					T26 = VSUB(TD, TE);
					TF = VADD(TD, TE);
					T24 = LD(&(ii[WS(is, 21)]), ivs, &(ii[WS(is, 1)]));
					TG = LD(&(ri[WS(is, 29)]), ivs, &(ri[WS(is, 1)]));
					TH = LD(&(ri[WS(is, 13)]), ivs, &(ri[WS(is, 1)]));
					T29 = LD(&(ii[WS(is, 29)]), ivs, &(ii[WS(is, 1)]));
					T2a = LD(&(ii[WS(is, 13)]), ivs, &(ii[WS(is, 1)]));
				   }
				   {
					V T4I, T25, T28, TI, T4J, T2b, T4H, TC, T5B, T4E;
					T3H = VADD(T21, T20);
					T22 = VSUB(T20, T21);
					T3E = VSUB(T2f, T2i);
					T2j = VADD(T2f, T2i);
					T4I = VADD(T23, T24);
					T25 = VSUB(T23, T24);
					T28 = VSUB(TG, TH);
					TI = VADD(TG, TH);
					T4J = VADD(T29, T2a);
					T2b = VSUB(T29, T2a);
					T4H = VSUB(Ty, TB);
					TC = VADD(Ty, TB);
					T5B = VADD(T4C, T4D);
					T4E = VSUB(T4C, T4D);
					{
					     V T27, T2k, TJ, T4F, T4K, T5C, T2c, T2l;
					     T27 = VSUB(T25, T26);
					     T2k = VADD(T26, T25);
					     TJ = VADD(TF, TI);
					     T4F = VSUB(TI, TF);
					     T4K = VSUB(T4I, T4J);
					     T5C = VADD(T4I, T4J);
					     T2c = VADD(T28, T2b);
					     T2l = VSUB(T28, T2b);
					     T4G = VSUB(T4E, T4F);
					     T5h = VADD(T4F, T4E);
					     TK = VADD(TC, TJ);
					     T5A = VSUB(TC, TJ);
					     T5D = VSUB(T5B, T5C);
					     T5W = VADD(T5B, T5C);
					     T2d = VADD(T27, T2c);
					     T3F = VSUB(T2c, T27);
					     T4L = VSUB(T4H, T4K);
					     T5g = VADD(T4H, T4K);
					     T3I = VSUB(T2k, T2l);
					     T2m = VADD(T2k, T2l);
					}
				   }
			      }
			 }
		    }
	       }
	       {
		    V T1v, T1g, T5V, Tv, T60, T5Y, T11, T10;
		    {
			 V T5o, T5n, T5i, T5r, T5f, T5l, T5w, T5u;
			 {
			      V T5d, T4t, T4A, T4X, T58, T51, T4M, T59, T54, T5e, T5b, T4B;
			      T5d = VADD(T4r, T4s);