hc2cfdft2_8.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 21:04:01 EST 2008 */

#include "codelet-rdft.h"

#ifdef HAVE_FMA

/* Generated by: ../../../genfft/gen_hc2cdft -fma -reorder-insns -schedule-for-pipeline -compact -variables 4 -pipeline-latency 4 -twiddle-log3 -precompute-twiddles -n 8 -dit -name hc2cfdft2_8 -include hc2cf.h */

/*
 * This function contains 90 FP additions, 66 FP multiplications,
 * (or, 60 additions, 36 multiplications, 30 fused multiply/add),
 * 68 stack variables, 2 constants, and 32 memory accesses
 */
#include "hc2cf.h"

static void hc2cfdft2_8(R *Rp, R *Ip, R *Rm, R *Im, const R *W, stride rs, INT mb, INT me, INT ms)
{
     DK(KP707106781, +0.707106781186547524400844362104849039284835938);
     DK(KP500000000, +0.500000000000000000000000000000000000000000000);
     INT m;
     for (m = mb, W = W + ((mb - 1) * 6); m < me; m = m + 1, Rp = Rp + ms, Ip = Ip + ms, Rm = Rm - ms, Im = Im - ms, W = W + 6, MAKE_VOLATILE_STRIDE(rs)) {
	  E T1G, T1F, T1C, T1D, T1N, T1B, T1R, T1L;
	  {
	       E T1, T2, Th, Tj, T4, T3, Ti, Tp, T5;
	       T1 = W[0];
	       T2 = W[2];
	       Th = W[4];
	       Tj = W[5];
	       T4 = W[1];
	       T3 = T1 * T2;
	       Ti = T1 * Th;
	       Tp = T1 * Tj;
	       T5 = W[3];
	       {
		    E Tk, Tq, TI, T1a, T1u, TY, TF, TS, T1s, T1c, Tr, T1n, Tg, T16, Tn;
		    E T13, T1f, Ts, To, T1o;
		    {
			 E T6, Tw, Tc, TB, TQ, TM, TC, TR, Tz, TD, TA;
			 {
			      E TX, TV, TT, TU;
			      {
				   E TG, Tb, TH, TP, TL;
				   TG = Ip[0];
				   Tk = FMA(T4, Tj, Ti);
				   Tq = FNMS(T4, Th, Tp);
				   T6 = FMA(T4, T5, T3);
				   Tw = FNMS(T4, T5, T3);
				   Tb = T1 * T5;
				   TH = Im[0];
				   TT = Rm[0];
				   TP = T6 * Tj;
				   TL = T6 * Th;
				   Tc = FNMS(T4, T2, Tb);
				   TB = FMA(T4, T2, Tb);
				   TX = TG + TH;
				   TI = TG - TH;
				   TU = Rp[0];
				   TQ = FNMS(Tc, Th, TP);
				   TM = FMA(Tc, Tj, TL);
			      }
			      T1a = TU + TT;
			      TV = TT - TU;
			      {
				   E Tx, Ty, T1t, TW;
				   Tx = Ip[WS(rs, 2)];
				   Ty = Im[WS(rs, 2)];
				   T1t = T4 * TV;
				   TW = T1 * TV;
				   TC = Rp[WS(rs, 2)];
				   TR = Tx + Ty;
				   Tz = Tx - Ty;
				   T1u = FMA(T1, TX, T1t);
				   TY = FNMS(T4, TX, TW);
				   TD = Rm[WS(rs, 2)];
			      }
			      TA = Tw * Tz;
			 }
			 {
			      E Td, T9, T12, Te, Ta, T1m;
			      {
				   E T7, T8, TN, TE, TO, T1r, T1b;
				   T7 = Ip[WS(rs, 1)];
				   T8 = Im[WS(rs, 1)];
				   TN = TD - TC;
				   TE = TC + TD;
				   Td = Rp[WS(rs, 1)];
				   T9 = T7 - T8;
				   T12 = T7 + T8;
				   TO = TM * TN;
				   T1r = TQ * TN;
				   T1b = Tw * TE;
				   TF = FNMS(TB, TE, TA);
				   TS = FNMS(TQ, TR, TO);
				   T1s = FMA(TM, TR, T1r);
				   T1c = FMA(TB, Tz, T1b);
				   Te = Rm[WS(rs, 1)];
			      }
			      Ta = T6 * T9;
			      T1m = T2 * T12;
			      {
				   E Tl, T10, Tf, Tm, T11, T1e;
				   Tl = Ip[WS(rs, 3)];
				   T10 = Td - Te;
				   Tf = Td + Te;
				   Tm = Im[WS(rs, 3)];
				   Tr = Rp[WS(rs, 3)];
				   T11 = T2 * T10;
				   T1n = FNMS(T5, T10, T1m);
				   T1e = T6 * Tf;
				   Tg = FNMS(Tc, Tf, Ta);
				   T16 = Tl + Tm;
				   Tn = Tl - Tm;
				   T13 = FMA(T5, T12, T11);
				   T1f = FMA(Tc, T9, T1e);
				   Ts = Rm[WS(rs, 3)];
			      }
			      To = Tk * Tn;
			      T1o = Th * T16;
			 }
		    }
		    {
			 E T1z, T1K, T1y, T1k, T1J, T1A, T1x, T1j;
			 {
			      E T1w, TK, T1l, T19, T1d, T1i;
			      {
				   E TJ, T14, Tt, T1v, T1h;
				   T1z = TI - TF;
				   TJ = TF + TI;
				   T14 = Tr - Ts;
				   Tt = Tr + Ts;
				   T1v = T1s + T1u;
				   T1G = T1u - T1s;
				   {
					E TZ, T1q, Tv, T18, T15;
					T1F = TY - TS;
					TZ = TS + TY;
					T15 = Th * T14;
					{
					     E T1p, T1g, Tu, T17;
					     T1p = FNMS(Tj, T14, T1o);
					     T1g = Tk * Tt;
					     Tu = FNMS(Tq, Tt, To);
					     T17 = FMA(Tj, T16, T15);
					     T1C = T1p - T1n;
					     T1q = T1n + T1p;
					     T1h = FMA(Tq, Tn, T1g);
					     T1K = Tg - Tu;
					     Tv = Tg + Tu;
					     T18 = T13 + T17;
					     T1D = T13 - T17;
					}
					T1w = T1q - T1v;
					T1y = T1q + T1v;
					TK = Tv + TJ;
					T1l = TJ - Tv;
					T1k = T18 + TZ;
					T19 = TZ - T18;
				   }
				   T1J = T1a - T1c;
				   T1d = T1a + T1c;
				   T1i = T1f + T1h;
				   T1A = T1f - T1h;
			      }
			      Ip[0] = KP500000000 * (TK + T19);
			      Im[WS(rs, 3)] = KP500000000 * (T19 - TK);
			      Im[WS(rs, 1)] = KP500000000 * (T1w - T1l);
			      T1x = T1d + T1i;
			      T1j = T1d - T1i;
			      Ip[WS(rs, 2)] = KP500000000 * (T1l + T1w);
			 }
			 Rm[WS(rs, 3)] = KP500000000 * (T1x - T1y);
			 Rp[0] = KP500000000 * (T1x + T1y);
			 Rp[WS(rs, 2)] = KP500000000 * (T1j + T1k);
			 Rm[WS(rs, 1)] = KP500000000 * (T1j - T1k);
			 T1N = T1A + T1z;
			 T1B = T1z - T1A;
			 T1R = T1J + T1K;
			 T1L = T1J - T1K;
		    }
	       }
	  }
	  {
	       E T1E, T1O, T1H, T1P;
	       T1E = T1C + T1D;
	       T1O = T1C - T1D;
	       T1H = T1F - T1G;
	       T1P = T1F + T1G;
	       {
		    E T1S, T1Q, T1I, T1M;
		    T1S = T1O + T1P;
		    T1Q = T1O - T1P;
		    T1I = T1E + T1H;
		    T1M = T1H - T1E;
		    Im[0] = -(KP500000000 * (FNMS(KP707106781, T1Q, T1N)));
		    Ip[WS(rs, 3)] = KP500000000 * (FMA(KP707106781, T1Q, T1N));
		    Rp[WS(rs, 1)] = KP500000000 * (FMA(KP707106781, T1S, T1R));
		    Rm[WS(rs, 2)] = KP500000000 * (FNMS(KP707106781, T1S, T1R));
		    Rp[WS(rs, 3)] = KP500000000 * (FMA(KP707106781, T1M, T1L));
		    Rm[0] = KP500000000 * (FNMS(KP707106781, T1M, T1L));
		    Im[WS(rs, 2)] = -(KP500000000 * (FNMS(KP707106781, T1I, T1B)));
		    Ip[WS(rs, 1)] = KP500000000 * (FMA(KP707106781, T1I, T1B));
	       }
	  }
     }
}

static const tw_instr twinstr[] = {
     {TW_CEXP, 1, 1},
     {TW_CEXP, 1, 3},
     {TW_CEXP, 1, 7},
     {TW_NEXT, 1, 0}
};

static const hc2c_desc desc = { 8, "hc2cfdft2_8", twinstr, &GENUS, {60, 36, 30, 0} };

void X(codelet_hc2cfdft2_8) (planner *p) {
     X(khc2c_register) (p, hc2cfdft2_8, &desc, HC2C_VIA_DFT);
}
#else				/* HAVE_FMA */

/* Generated by: ../../../genfft/gen_hc2cdft -compact -variables 4 -pipeline-latency 4 -twiddle-log3 -precompute-twiddles -n 8 -dit -name hc2cfdft2_8 -include hc2cf.h */

/*
 * This function contains 90 FP additions, 56 FP multiplications,
 * (or, 72 additions, 38 multiplications, 18 fused multiply/add),
 * 51 stack variables, 2 constants, and 32 memory accesses
 */
#include "hc2cf.h"

static void hc2cfdft2_8(R *Rp, R *Ip, R *Rm, R *Im, const R *W, stride rs, INT mb, INT me, INT ms)
{
     DK(KP353553390, +0.353553390593273762200422181052424519642417969);
     DK(KP500000000, +0.500000000000000000000000000000000000000000000);
     INT m;
     for (m = mb, W = W + ((mb - 1) * 6); m < me; m = m + 1, Rp = Rp + ms, Ip = Ip + ms, Rm = Rm - ms, Im = Im - ms, W = W + 6, MAKE_VOLATILE_STRIDE(rs)) {
	  E T1, T4, T2, T5, Tu, Ty, T7, Td, Ti, Tj, Tk, TP, To, TN;
	  {
	       E T3, Tc, T6, Tb;
	       T1 = W[0];
	       T4 = W[1];
	       T2 = W[2];
	       T5 = W[3];
	       T3 = T1 * T2;
	       Tc = T4 * T2;
	       T6 = T4 * T5;
	       Tb = T1 * T5;
	       Tu = T3 - T6;
	       Ty = Tb + Tc;
	       T7 = T3 + T6;
	       Td = Tb - Tc;
	       Ti = W[4];
	       Tj = W[5];
	       Tk = FMA(T1, Ti, T4 * Tj);
	       TP = FNMS(Td, Ti, T7 * Tj);
	       To = FNMS(T4, Ti, T1 * Tj);
	       TN = FMA(T7, Ti, Td * Tj);
	  }
	  {
	       E TF, T11, TC, T12, T1d, T1e, T1q, TM, TR, T1p, Th, Ts, T15, T14, T1a;
	       E T1b, T1m, TV, TY, T1n;
	       {
		    E TD, TE, TL, TI, TJ, TK, Tx, TQ, TB, TO;
		    TD = Ip[0];
		    TE = Im[0];
		    TL = TD + TE;
		    TI = Rm[0];
		    TJ = Rp[0];
		    TK = TI - TJ;
		    {
			 E Tv, Tw, Tz, TA;
			 Tv = Ip[WS(rs, 2)];
			 Tw = Im[WS(rs, 2)];
			 Tx = Tv - Tw;
			 TQ = Tv + Tw;
			 Tz = Rp[WS(rs, 2)];
			 TA = Rm[WS(rs, 2)];
			 TB = Tz + TA;
			 TO = Tz - TA;
		    }
		    TF = TD - TE;
		    T11 = TJ + TI;
		    TC = FNMS(Ty, TB, Tu * Tx);
		    T12 = FMA(Tu, TB, Ty * Tx);
		    T1d = FNMS(TP, TO, TN * TQ);
		    T1e = FMA(T4, TK, T1 * TL);
		    T1q = T1e - T1d;
		    TM = FNMS(T4, TL, T1 * TK);
		    TR = FMA(TN, TO, TP * TQ);
		    T1p = TR + TM;
	       }
	       {
		    E Ta, TU, Tg, TT, Tn, TX, Tr, TW;
		    {
			 E T8, T9, Te, Tf;
			 T8 = Ip[WS(rs, 1)];
			 T9 = Im[WS(rs, 1)];
			 Ta = T8 - T9;
			 TU = T8 + T9;
			 Te = Rp[WS(rs, 1)];
			 Tf = Rm[WS(rs, 1)];
			 Tg = Te + Tf;
			 TT = Te - Tf;
		    }
		    {
			 E Tl, Tm, Tp, Tq;
			 Tl = Ip[WS(rs, 3)];
			 Tm = Im[WS(rs, 3)];
			 Tn = Tl - Tm;
			 TX = Tl + Tm;
			 Tp = Rp[WS(rs, 3)];
			 Tq = Rm[WS(rs, 3)];
			 Tr = Tp + Tq;
			 TW = Tp - Tq;
		    }
		    Th = FNMS(Td, Tg, T7 * Ta);
		    Ts = FNMS(To, Tr, Tk * Tn);
		    T15 = FMA(Tk, Tr, To * Tn);
		    T14 = FMA(T7, Tg, Td * Ta);
		    T1a = FNMS(T5, TT, T2 * TU);
		    T1b = FNMS(Tj, TW, Ti * TX);
		    T1m = T1b - T1a;
		    TV = FMA(T2, TT, T5 * TU);
		    TY = FMA(Ti, TW, Tj * TX);
		    T1n = TV - TY;
	       }
	       {
		    E T1l, T1x, T1A, T1C, T1s, T1w, T1v, T1B;
		    {
			 E T1j, T1k, T1y, T1z;
			 T1j = TF - TC;
			 T1k = T14 - T15;
			 T1l = KP500000000 * (T1j - T1k);
			 T1x = KP500000000 * (T1k + T1j);
			 T1y = T1m - T1n;
			 T1z = T1p + T1q;
			 T1A = KP353553390 * (T1y - T1z);
			 T1C = KP353553390 * (T1y + T1z);
		    }
		    {
			 E T1o, T1r, T1t, T1u;
			 T1o = T1m + T1n;
			 T1r = T1p - T1q;
			 T1s = KP353553390 * (T1o + T1r);
			 T1w = KP353553390 * (T1r - T1o);
			 T1t = T11 - T12;
			 T1u = Th - Ts;
			 T1v = KP500000000 * (T1t - T1u);
			 T1B = KP500000000 * (T1t + T1u);
		    }
		    Ip[WS(rs, 1)] = T1l + T1s;
		    Rp[WS(rs, 1)] = T1B + T1C;
		    Im[WS(rs, 2)] = T1s - T1l;
		    Rm[WS(rs, 2)] = T1B - T1C;
		    Rm[0] = T1v - T1w;
		    Im[0] = T1A - T1x;
		    Rp[WS(rs, 3)] = T1v + T1w;
		    Ip[WS(rs, 3)] = T1x + T1A;
	       }
	       {
		    E TH, T19, T1g, T1i, T10, T18, T17, T1h;
		    {
			 E Tt, TG, T1c, T1f;
			 Tt = Th + Ts;
			 TG = TC + TF;
			 TH = Tt + TG;
			 T19 = TG - Tt;
			 T1c = T1a + T1b;
			 T1f = T1d + T1e;
			 T1g = T1c - T1f;
			 T1i = T1c + T1f;
		    }
		    {
			 E TS, TZ, T13, T16;
			 TS = TM - TR;
			 TZ = TV + TY;
			 T10 = TS - TZ;
			 T18 = TZ + TS;
			 T13 = T11 + T12;
			 T16 = T14 + T15;
			 T17 = T13 - T16;
			 T1h = T13 + T16;
		    }
		    Ip[0] = KP500000000 * (TH + T10);
		    Rp[0] = KP500000000 * (T1h + T1i);
		    Im[WS(rs, 3)] = KP500000000 * (T10 - TH);
		    Rm[WS(rs, 3)] = KP500000000 * (T1h - T1i);
		    Rm[WS(rs, 1)] = KP500000000 * (T17 - T18);
		    Im[WS(rs, 1)] = KP500000000 * (T1g - T19);
		    Rp[WS(rs, 2)] = KP500000000 * (T17 + T18);
		    Ip[WS(rs, 2)] = KP500000000 * (T19 + T1g);
	       }
	  }
     }
}

static const tw_instr twinstr[] = {
     {TW_CEXP, 1, 1},
     {TW_CEXP, 1, 3},
     {TW_CEXP, 1, 7},
     {TW_NEXT, 1, 0}
};

static const hc2c_desc desc = { 8, "hc2cfdft2_8", twinstr, &GENUS, {72, 38, 18, 0} };

void X(codelet_hc2cfdft2_8) (planner *p) {
     X(khc2c_register) (p, hc2cfdft2_8, &desc, HC2C_VIA_DFT);
}
#endif				/* HAVE_FMA */