hc2cfdft_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 21:03:19 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 -n 8 -dit -name hc2cfdft_8 -include hc2cf.h */

/*
 * This function contains 82 FP additions, 52 FP multiplications,
 * (or, 60 additions, 30 multiplications, 22 fused multiply/add),
 * 55 stack variables, 2 constants, and 32 memory accesses
 */
#include "hc2cf.h"

static void hc2cfdft_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) * 14); m < me; m = m + 1, Rp = Rp + ms, Ip = Ip + ms, Rm = Rm - ms, Im = Im - ms, W = W + 14, MAKE_VOLATILE_STRIDE(rs)) {
	  E T1A, T1w, T1z, T1x, T1H, T1v, T1L, T1F;
	  {
	       E Ty, T14, TO, T1o, Tv, TG, T16, T1m, Ta, T19, T1h, TV, T10, TX, TZ;
	       E Tk, T1i, TY, T1b, TF, TB, T1l;
	       {
		    E TH, TN, TK, TM;
		    {
			 E Tw, Tx, TI, TJ;
			 Tw = Ip[0];
			 Tx = Im[0];
			 TI = Rm[0];
			 TJ = Rp[0];
			 TH = W[0];
			 Ty = Tw - Tx;
			 TN = Tw + Tx;
			 T14 = TJ + TI;
			 TK = TI - TJ;
			 TM = W[1];
		    }
		    {
			 E Ts, Tp, Tt, Tm, Tr;
			 {
			      E Tn, To, TL, T1n;
			      Tn = Ip[WS(rs, 2)];
			      To = Im[WS(rs, 2)];
			      TL = TH * TK;
			      T1n = TM * TK;
			      Ts = Rp[WS(rs, 2)];
			      TF = Tn + To;
			      Tp = Tn - To;
			      TO = FNMS(TM, TN, TL);
			      T1o = FMA(TH, TN, T1n);
			      Tt = Rm[WS(rs, 2)];
			 }
			 Tm = W[6];
			 Tr = W[7];
			 {
			      E TE, TD, T15, TC, Tu, Tq;
			      TB = W[8];
			      TC = Tt - Ts;
			      Tu = Ts + Tt;
			      Tq = Tm * Tp;
			      TE = W[9];
			      TD = TB * TC;
			      T15 = Tm * Tu;
			      Tv = FNMS(Tr, Tu, Tq);
			      T1l = TE * TC;
			      TG = FNMS(TE, TF, TD);
			      T16 = FMA(Tr, Tp, T15);
			 }
		    }
	       }
	       {
		    E TU, TR, TT, T1g, TS;
		    {
			 E T2, T3, T7, T8;
			 T2 = Ip[WS(rs, 1)];
			 T1m = FMA(TB, TF, T1l);
			 T3 = Im[WS(rs, 1)];
			 T7 = Rp[WS(rs, 1)];
			 T8 = Rm[WS(rs, 1)];
			 {
			      E T1, T4, T9, T6, T5, TQ, T18;
			      T1 = W[2];
			      TU = T2 + T3;
			      T4 = T2 - T3;
			      TR = T7 - T8;
			      T9 = T7 + T8;
			      T6 = W[3];
			      T5 = T1 * T4;
			      TQ = W[4];
			      T18 = T1 * T9;
			      TT = W[5];
			      Ta = FNMS(T6, T9, T5);
			      T1g = TQ * TU;
			      TS = TQ * TR;
			      T19 = FMA(T6, T4, T18);
			 }
		    }
		    {
			 E Tc, Td, Th, Ti;
			 Tc = Ip[WS(rs, 3)];
			 T1h = FNMS(TT, TR, T1g);
			 TV = FMA(TT, TU, TS);
			 Td = Im[WS(rs, 3)];
			 Th = Rp[WS(rs, 3)];
			 Ti = Rm[WS(rs, 3)];
			 {
			      E Tb, Te, Tj, Tg, Tf, TW, T1a;
			      Tb = W[10];
			      T10 = Tc + Td;
			      Te = Tc - Td;
			      TX = Th - Ti;
			      Tj = Th + Ti;
			      Tg = W[11];
			      Tf = Tb * Te;
			      TW = W[12];
			      T1a = Tb * Tj;
			      TZ = W[13];
			      Tk = FNMS(Tg, Tj, Tf);
			      T1i = TW * T10;
			      TY = TW * TX;
			      T1b = FMA(Tg, Te, T1a);
			 }
		    }
	       }
	       {
		    E T1E, T1t, TA, T1s, T1D, T1u, T1e, T13, T1r, T1d;
		    {
			 E TP, T1f, T1q, T12, T17, T1c;
			 {
			      E Tl, T11, Tz, T1p, T1k, T1j;
			      T1E = Ta - Tk;
			      Tl = Ta + Tk;
			      T1j = FNMS(TZ, TX, T1i);
			      T11 = FMA(TZ, T10, TY);
			      Tz = Tv + Ty;
			      T1t = Ty - Tv;
			      T1A = T1o - T1m;
			      T1p = T1m + T1o;
			      T1k = T1h + T1j;
			      T1w = T1j - T1h;
			      T1z = TO - TG;
			      TP = TG + TO;
			      T1f = Tz - Tl;
			      TA = Tl + Tz;
			      T1s = T1k + T1p;
			      T1q = T1k - T1p;
			      T12 = TV + T11;
			      T1x = TV - T11;
			      T1D = T14 - T16;
			      T17 = T14 + T16;
			      T1c = T19 + T1b;
			      T1u = T19 - T1b;
			 }
			 Im[WS(rs, 1)] = KP500000000 * (T1q - T1f);
			 T1e = T12 + TP;
			 T13 = TP - T12;
			 T1r = T17 + T1c;
			 T1d = T17 - T1c;
			 Ip[WS(rs, 2)] = KP500000000 * (T1f + T1q);
		    }
		    Im[WS(rs, 3)] = KP500000000 * (T13 - TA);
		    Ip[0] = KP500000000 * (TA + T13);
		    Rm[WS(rs, 3)] = KP500000000 * (T1r - T1s);
		    Rp[0] = KP500000000 * (T1r + T1s);
		    Rp[WS(rs, 2)] = KP500000000 * (T1d + T1e);
		    Rm[WS(rs, 1)] = KP500000000 * (T1d - T1e);
		    T1H = T1u + T1t;
		    T1v = T1t - T1u;
		    T1L = T1D + T1E;
		    T1F = T1D - T1E;
	       }
	  }
	  {
	       E T1y, T1I, T1B, T1J;
	       T1y = T1w + T1x;
	       T1I = T1w - T1x;
	       T1B = T1z - T1A;
	       T1J = T1z + T1A;
	       {
		    E T1M, T1K, T1C, T1G;
		    T1M = T1I + T1J;
		    T1K = T1I - T1J;
		    T1C = T1y + T1B;
		    T1G = T1B - T1y;
		    Im[0] = -(KP500000000 * (FNMS(KP707106781, T1K, T1H)));
		    Ip[WS(rs, 3)] = KP500000000 * (FMA(KP707106781, T1K, T1H));
		    Rp[WS(rs, 1)] = KP500000000 * (FMA(KP707106781, T1M, T1L));
		    Rm[WS(rs, 2)] = KP500000000 * (FNMS(KP707106781, T1M, T1L));
		    Rp[WS(rs, 3)] = KP500000000 * (FMA(KP707106781, T1G, T1F));
		    Rm[0] = KP500000000 * (FNMS(KP707106781, T1G, T1F));
		    Im[WS(rs, 2)] = -(KP500000000 * (FNMS(KP707106781, T1C, T1v)));
		    Ip[WS(rs, 1)] = KP500000000 * (FMA(KP707106781, T1C, T1v));
	       }
	  }
     }
}

static const tw_instr twinstr[] = {
     {TW_FULL, 1, 8},
     {TW_NEXT, 1, 0}
};

static const hc2c_desc desc = { 8, "hc2cfdft_8", twinstr, &GENUS, {60, 30, 22, 0} };

void X(codelet_hc2cfdft_8) (planner *p) {
     X(khc2c_register) (p, hc2cfdft_8, &desc, HC2C_VIA_DFT);
}
#else				/* HAVE_FMA */

/* Generated by: ../../../genfft/gen_hc2cdft -compact -variables 4 -pipeline-latency 4 -n 8 -dit -name hc2cfdft_8 -include hc2cf.h */

/*
 * This function contains 82 FP additions, 44 FP multiplications,
 * (or, 68 additions, 30 multiplications, 14 fused multiply/add),
 * 39 stack variables, 2 constants, and 32 memory accesses
 */
#include "hc2cf.h"

static void hc2cfdft_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) * 14); m < me; m = m + 1, Rp = Rp + ms, Ip = Ip + ms, Rm = Rm - ms, Im = Im - ms, W = W + 14, MAKE_VOLATILE_STRIDE(rs)) {
	  E Tv, TX, Ts, TY, TE, T1a, TJ, T19, T1l, T1m, T9, T10, Ti, T11, TP;
	  E T16, TU, T17, T1i, T1j;
	  {
	       E Tt, Tu, TD, Tz, TA, TB, Tn, TI, Tr, TG, Tk, To;
	       Tt = Ip[0];
	       Tu = Im[0];
	       TD = Tt + Tu;
	       Tz = Rm[0];
	       TA = Rp[0];
	       TB = Tz - TA;
	       {
		    E Tl, Tm, Tp, Tq;
		    Tl = Ip[WS(rs, 2)];
		    Tm = Im[WS(rs, 2)];
		    Tn = Tl - Tm;
		    TI = Tl + Tm;
		    Tp = Rp[WS(rs, 2)];
		    Tq = Rm[WS(rs, 2)];
		    Tr = Tp + Tq;
		    TG = Tp - Tq;
	       }
	       Tv = Tt - Tu;
	       TX = TA + Tz;
	       Tk = W[6];
	       To = W[7];
	       Ts = FNMS(To, Tr, Tk * Tn);
	       TY = FMA(Tk, Tr, To * Tn);
	       {
		    E Ty, TC, TF, TH;
		    Ty = W[0];
		    TC = W[1];
		    TE = FNMS(TC, TD, Ty * TB);
		    T1a = FMA(TC, TB, Ty * TD);
		    TF = W[8];
		    TH = W[9];
		    TJ = FMA(TF, TG, TH * TI);
		    T19 = FNMS(TH, TG, TF * TI);
	       }
	       T1l = TJ + TE;
	       T1m = T1a - T19;
	  }
	  {
	       E T4, TO, T8, TM, Td, TT, Th, TR;
	       {
		    E T2, T3, T6, T7;
		    T2 = Ip[WS(rs, 1)];
		    T3 = Im[WS(rs, 1)];
		    T4 = T2 - T3;
		    TO = T2 + T3;
		    T6 = Rp[WS(rs, 1)];
		    T7 = Rm[WS(rs, 1)];
		    T8 = T6 + T7;
		    TM = T6 - T7;
	       }
	       {
		    E Tb, Tc, Tf, Tg;
		    Tb = Ip[WS(rs, 3)];
		    Tc = Im[WS(rs, 3)];
		    Td = Tb - Tc;
		    TT = Tb + Tc;
		    Tf = Rp[WS(rs, 3)];
		    Tg = Rm[WS(rs, 3)];
		    Th = Tf + Tg;
		    TR = Tf - Tg;
	       }
	       {
		    E T1, T5, Ta, Te;
		    T1 = W[2];
		    T5 = W[3];
		    T9 = FNMS(T5, T8, T1 * T4);
		    T10 = FMA(T1, T8, T5 * T4);
		    Ta = W[10];
		    Te = W[11];
		    Ti = FNMS(Te, Th, Ta * Td);
		    T11 = FMA(Ta, Th, Te * Td);
		    {
			 E TL, TN, TQ, TS;
			 TL = W[4];
			 TN = W[5];
			 TP = FMA(TL, TM, TN * TO);
			 T16 = FNMS(TN, TM, TL * TO);
			 TQ = W[12];
			 TS = W[13];
			 TU = FMA(TQ, TR, TS * TT);
			 T17 = FNMS(TS, TR, TQ * TT);
		    }
		    T1i = T17 - T16;
		    T1j = TP - TU;
	       }
	  }
	  {
	       E T1h, T1t, T1w, T1y, T1o, T1s, T1r, T1x;
	       {
		    E T1f, T1g, T1u, T1v;
		    T1f = Tv - Ts;
		    T1g = T10 - T11;
		    T1h = KP500000000 * (T1f - T1g);
		    T1t = KP500000000 * (T1g + T1f);
		    T1u = T1i - T1j;
		    T1v = T1l + T1m;
		    T1w = KP353553390 * (T1u - T1v);
		    T1y = KP353553390 * (T1u + T1v);
	       }
	       {
		    E T1k, T1n, T1p, T1q;
		    T1k = T1i + T1j;
		    T1n = T1l - T1m;
		    T1o = KP353553390 * (T1k + T1n);
		    T1s = KP353553390 * (T1n - T1k);
		    T1p = TX - TY;
		    T1q = T9 - Ti;
		    T1r = KP500000000 * (T1p - T1q);
		    T1x = KP500000000 * (T1p + T1q);
	       }
	       Ip[WS(rs, 1)] = T1h + T1o;
	       Rp[WS(rs, 1)] = T1x + T1y;
	       Im[WS(rs, 2)] = T1o - T1h;
	       Rm[WS(rs, 2)] = T1x - T1y;
	       Rm[0] = T1r - T1s;
	       Im[0] = T1w - T1t;
	       Rp[WS(rs, 3)] = T1r + T1s;
	       Ip[WS(rs, 3)] = T1t + T1w;
	  }
	  {
	       E Tx, T15, T1c, T1e, TW, T14, T13, T1d;
	       {
		    E Tj, Tw, T18, T1b;
		    Tj = T9 + Ti;
		    Tw = Ts + Tv;
		    Tx = Tj + Tw;
		    T15 = Tw - Tj;
		    T18 = T16 + T17;
		    T1b = T19 + T1a;
		    T1c = T18 - T1b;
		    T1e = T18 + T1b;
	       }
	       {
		    E TK, TV, TZ, T12;
		    TK = TE - TJ;
		    TV = TP + TU;
		    TW = TK - TV;
		    T14 = TV + TK;
		    TZ = TX + TY;
		    T12 = T10 + T11;
		    T13 = TZ - T12;
		    T1d = TZ + T12;
	       }
	       Ip[0] = KP500000000 * (Tx + TW);
	       Rp[0] = KP500000000 * (T1d + T1e);
	       Im[WS(rs, 3)] = KP500000000 * (TW - Tx);
	       Rm[WS(rs, 3)] = KP500000000 * (T1d - T1e);
	       Rm[WS(rs, 1)] = KP500000000 * (T13 - T14);
	       Im[WS(rs, 1)] = KP500000000 * (T1c - T15);
	       Rp[WS(rs, 2)] = KP500000000 * (T13 + T14);
	       Ip[WS(rs, 2)] = KP500000000 * (T15 + T1c);
	  }
     }
}

static const tw_instr twinstr[] = {
     {TW_FULL, 1, 8},
     {TW_NEXT, 1, 0}
};

static const hc2c_desc desc = { 8, "hc2cfdft_8", twinstr, &GENUS, {68, 30, 14, 0} };

void X(codelet_hc2cfdft_8) (planner *p) {
     X(khc2c_register) (p, hc2cfdft_8, &desc, HC2C_VIA_DFT);
}
#endif				/* HAVE_FMA */