hc2cb_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 21:10:37 EST 2008 */

#include "codelet-rdft.h"

#ifdef HAVE_FMA

/* Generated by: ../../../genfft/gen_hc2c -fma -reorder-insns -schedule-for-pipeline -compact -variables 4 -pipeline-latency 4 -sign 1 -n 8 -dif -name hc2cb_8 -include hc2cb.h */

/*
 * This function contains 66 FP additions, 36 FP multiplications,
 * (or, 44 additions, 14 multiplications, 22 fused multiply/add),
 * 52 stack variables, 1 constants, and 32 memory accesses
 */
#include "hc2cb.h"

static void hc2cb_8(R *Rp, R *Ip, R *Rm, R *Im, const R *W, stride rs, INT mb, INT me, INT ms)
{
     DK(KP707106781, +0.707106781186547524400844362104849039284835938);
     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 Tw, TH, Tf, Ty, Tx, TI;
	  {
	       E TV, TD, T1i, T7, T1b, T1n, TQ, Tk, Tp, TE, Te, T1o, T1e, T1j, Tu;
	       E TF;
	       {
		    E T4, Tg, T3, T19, TC, T5, Th, Ti;
		    {
			 E T1, T2, TA, TB;
			 T1 = Rp[0];
			 T2 = Rm[WS(rs, 3)];
			 TA = Ip[0];
			 TB = Im[WS(rs, 3)];
			 T4 = Rp[WS(rs, 2)];
			 Tg = T1 - T2;
			 T3 = T1 + T2;
			 T19 = TA - TB;
			 TC = TA + TB;
			 T5 = Rm[WS(rs, 1)];
			 Th = Ip[WS(rs, 2)];
			 Ti = Im[WS(rs, 1)];
		    }
		    {
			 E Tb, Tl, Ta, T1c, To, Tc, Tr, Ts;
			 {
			      E T8, T9, Tm, Tn;
			      T8 = Rp[WS(rs, 1)];
			      {
				   E Tz, T6, T1a, Tj;
				   Tz = T4 - T5;
				   T6 = T4 + T5;
				   T1a = Th - Ti;
				   Tj = Th + Ti;
				   TV = TC - Tz;
				   TD = Tz + TC;
				   T1i = T3 - T6;
				   T7 = T3 + T6;
				   T1b = T19 + T1a;
				   T1n = T19 - T1a;
				   TQ = Tg + Tj;
				   Tk = Tg - Tj;
				   T9 = Rm[WS(rs, 2)];
			      }
			      Tm = Ip[WS(rs, 1)];
			      Tn = Im[WS(rs, 2)];
			      Tb = Rm[0];
			      Tl = T8 - T9;
			      Ta = T8 + T9;
			      T1c = Tm - Tn;
			      To = Tm + Tn;
			      Tc = Rp[WS(rs, 3)];
			      Tr = Ip[WS(rs, 3)];
			      Ts = Im[0];
			 }
			 {
			      E Tq, Td, T1d, Tt;
			      Tp = Tl - To;
			      TE = Tl + To;
			      Tq = Tb - Tc;
			      Td = Tb + Tc;
			      T1d = Tr - Ts;
			      Tt = Tr + Ts;
			      Te = Ta + Td;
			      T1o = Ta - Td;
			      T1e = T1c + T1d;
			      T1j = T1d - T1c;
			      Tu = Tq - Tt;
			      TF = Tq + Tt;
			 }
		    }
	       }
	       {
		    E TG, Tv, T10, T13, T1s, T1k, T1p, T1v, T1u, T1w, T1t, TR, TW;
		    Rp[0] = T7 + Te;
		    Rm[0] = T1b + T1e;
		    TG = TE - TF;
		    TR = TE + TF;
		    TW = Tp - Tu;
		    Tv = Tp + Tu;
		    {
			 E TP, TS, TX, TU, T1r, TT, TY;
			 TP = W[4];
			 T10 = FMA(KP707106781, TR, TQ);
			 TS = FNMS(KP707106781, TR, TQ);
			 TX = FMA(KP707106781, TW, TV);
			 T13 = FNMS(KP707106781, TW, TV);
			 TU = W[5];
			 T1s = T1i + T1j;
			 T1k = T1i - T1j;
			 TT = TP * TS;
			 TY = TP * TX;
			 T1p = T1n - T1o;
			 T1v = T1o + T1n;
			 T1r = W[2];
			 Ip[WS(rs, 1)] = FNMS(TU, TX, TT);
			 Im[WS(rs, 1)] = FMA(TU, TS, TY);
			 T1u = W[3];
			 T1w = T1r * T1v;
			 T1t = T1r * T1s;
		    }
		    {
			 E T1f, T15, T18, T17, T1g, T1h, T1m;
			 {
			      E TZ, T12, T16, T14, T11;
			      Rm[WS(rs, 1)] = FMA(T1u, T1s, T1w);
			      Rp[WS(rs, 1)] = FNMS(T1u, T1v, T1t);
			      TZ = W[12];
			      T12 = W[13];
			      T1f = T1b - T1e;
			      T16 = T7 - Te;
			      T14 = TZ * T13;
			      T11 = TZ * T10;
			      T15 = W[6];
			      T18 = W[7];
			      Im[WS(rs, 3)] = FMA(T12, T10, T14);
			      Ip[WS(rs, 3)] = FNMS(T12, T13, T11);
			      T17 = T15 * T16;
			      T1g = T18 * T16;
			 }
			 Rp[WS(rs, 2)] = FNMS(T18, T1f, T17);
			 Rm[WS(rs, 2)] = FMA(T15, T1f, T1g);
			 T1h = W[10];
			 T1m = W[11];
			 {
			      E TN, TJ, TM, TL, TO, TK, T1q, T1l;
			      Tw = FNMS(KP707106781, Tv, Tk);
			      TK = FMA(KP707106781, Tv, Tk);
			      T1q = T1h * T1p;
			      T1l = T1h * T1k;
			      TN = FMA(KP707106781, TG, TD);
			      TH = FNMS(KP707106781, TG, TD);
			      Rm[WS(rs, 3)] = FMA(T1m, T1k, T1q);
			      Rp[WS(rs, 3)] = FNMS(T1m, T1p, T1l);
			      TJ = W[0];
			      TM = W[1];
			      Tf = W[8];
			      TL = TJ * TK;
			      TO = TM * TK;
			      Ty = W[9];
			      Tx = Tf * Tw;
			      Ip[0] = FNMS(TM, TN, TL);
			      Im[0] = FMA(TJ, TN, TO);
			 }
		    }
	       }
	  }
	  Ip[WS(rs, 2)] = FNMS(Ty, TH, Tx);
	  TI = Ty * Tw;
	  Im[WS(rs, 2)] = FMA(Tf, TH, TI);
     }
}

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

static const hc2c_desc desc = { 8, "hc2cb_8", twinstr, &GENUS, {44, 14, 22, 0} };

void X(codelet_hc2cb_8) (planner *p) {
     X(khc2c_register) (p, hc2cb_8, &desc, HC2C_VIA_RDFT);
}
#else				/* HAVE_FMA */

/* Generated by: ../../../genfft/gen_hc2c -compact -variables 4 -pipeline-latency 4 -sign 1 -n 8 -dif -name hc2cb_8 -include hc2cb.h */

/*
 * This function contains 66 FP additions, 32 FP multiplications,
 * (or, 52 additions, 18 multiplications, 14 fused multiply/add),
 * 30 stack variables, 1 constants, and 32 memory accesses
 */
#include "hc2cb.h"

static void hc2cb_8(R *Rp, R *Ip, R *Rm, R *Im, const R *W, stride rs, INT mb, INT me, INT ms)
{
     DK(KP707106781, +0.707106781186547524400844362104849039284835938);
     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 T7, T18, T1c, To, Ty, TM, TY, TC, Te, TZ, T10, Tv, Tz, TP, TS;
	  E TD;
	  {
	       E T3, TK, Tk, TX, T6, TW, Tn, TL;
	       {
		    E T1, T2, Ti, Tj;
		    T1 = Rp[0];
		    T2 = Rm[WS(rs, 3)];
		    T3 = T1 + T2;
		    TK = T1 - T2;
		    Ti = Ip[0];
		    Tj = Im[WS(rs, 3)];
		    Tk = Ti - Tj;
		    TX = Ti + Tj;
	       }
	       {
		    E T4, T5, Tl, Tm;
		    T4 = Rp[WS(rs, 2)];
		    T5 = Rm[WS(rs, 1)];
		    T6 = T4 + T5;
		    TW = T4 - T5;
		    Tl = Ip[WS(rs, 2)];
		    Tm = Im[WS(rs, 1)];
		    Tn = Tl - Tm;
		    TL = Tl + Tm;
	       }
	       T7 = T3 + T6;
	       T18 = TK + TL;
	       T1c = TX - TW;
	       To = Tk + Tn;
	       Ty = T3 - T6;
	       TM = TK - TL;
	       TY = TW + TX;
	       TC = Tk - Tn;
	  }
	  {
	       E Ta, TN, Tr, TO, Td, TQ, Tu, TR;
	       {
		    E T8, T9, Tp, Tq;
		    T8 = Rp[WS(rs, 1)];
		    T9 = Rm[WS(rs, 2)];
		    Ta = T8 + T9;
		    TN = T8 - T9;
		    Tp = Ip[WS(rs, 1)];
		    Tq = Im[WS(rs, 2)];
		    Tr = Tp - Tq;
		    TO = Tp + Tq;
	       }
	       {
		    E Tb, Tc, Ts, Tt;
		    Tb = Rm[0];
		    Tc = Rp[WS(rs, 3)];
		    Td = Tb + Tc;
		    TQ = Tb - Tc;
		    Ts = Ip[WS(rs, 3)];
		    Tt = Im[0];
		    Tu = Ts - Tt;
		    TR = Ts + Tt;
	       }
	       Te = Ta + Td;
	       TZ = TN + TO;
	       T10 = TQ + TR;
	       Tv = Tr + Tu;
	       Tz = Tu - Tr;
	       TP = TN - TO;
	       TS = TQ - TR;
	       TD = Ta - Td;
	  }
	  Rp[0] = T7 + Te;
	  Rm[0] = To + Tv;
	  {
	       E Tg, Tw, Tf, Th;
	       Tg = T7 - Te;
	       Tw = To - Tv;
	       Tf = W[6];
	       Th = W[7];
	       Rp[WS(rs, 2)] = FNMS(Th, Tw, Tf * Tg);
	       Rm[WS(rs, 2)] = FMA(Th, Tg, Tf * Tw);
	  }
	  {
	       E TG, TI, TF, TH;
	       TG = Ty + Tz;
	       TI = TD + TC;
	       TF = W[2];
	       TH = W[3];
	       Rp[WS(rs, 1)] = FNMS(TH, TI, TF * TG);
	       Rm[WS(rs, 1)] = FMA(TF, TI, TH * TG);
	  }
	  {
	       E TA, TE, Tx, TB;
	       TA = Ty - Tz;
	       TE = TC - TD;
	       Tx = W[10];
	       TB = W[11];
	       Rp[WS(rs, 3)] = FNMS(TB, TE, Tx * TA);
	       Rm[WS(rs, 3)] = FMA(Tx, TE, TB * TA);
	  }
	  {
	       E T1a, T1g, T1e, T1i, T19, T1d;
	       T19 = KP707106781 * (TZ + T10);
	       T1a = T18 - T19;
	       T1g = T18 + T19;
	       T1d = KP707106781 * (TP - TS);
	       T1e = T1c + T1d;
	       T1i = T1c - T1d;
	       {
		    E T17, T1b, T1f, T1h;
		    T17 = W[4];
		    T1b = W[5];
		    Ip[WS(rs, 1)] = FNMS(T1b, T1e, T17 * T1a);
		    Im[WS(rs, 1)] = FMA(T17, T1e, T1b * T1a);
		    T1f = W[12];
		    T1h = W[13];
		    Ip[WS(rs, 3)] = FNMS(T1h, T1i, T1f * T1g);
		    Im[WS(rs, 3)] = FMA(T1f, T1i, T1h * T1g);
	       }
	  }
	  {
	       E TU, T14, T12, T16, TT, T11;
	       TT = KP707106781 * (TP + TS);
	       TU = TM - TT;
	       T14 = TM + TT;
	       T11 = KP707106781 * (TZ - T10);
	       T12 = TY - T11;
	       T16 = TY + T11;
	       {
		    E TJ, TV, T13, T15;
		    TJ = W[8];
		    TV = W[9];
		    Ip[WS(rs, 2)] = FNMS(TV, T12, TJ * TU);
		    Im[WS(rs, 2)] = FMA(TV, TU, TJ * T12);
		    T13 = W[0];
		    T15 = W[1];
		    Ip[0] = FNMS(T15, T16, T13 * T14);
		    Im[0] = FMA(T15, T14, T13 * T16);
	       }
	  }
     }
}

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

static const hc2c_desc desc = { 8, "hc2cb_8", twinstr, &GENUS, {52, 18, 14, 0} };

void X(codelet_hc2cb_8) (planner *p) {
     X(khc2c_register) (p, hc2cb_8, &desc, HC2C_VIA_RDFT);
}
#endif				/* HAVE_FMA */