annotate src/fftw-3.3.3/dft/scalar/codelets/n1_6.c @ 95:89f5e221ed7b

Add FFTW3

/*
 * Copyright (c) 2003, 2007-11 Matteo Frigo
 * Copyright (c) 2003, 2007-11 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., 51 Franklin Street, Fifth Floor, Boston, MA  02110-1301  USA
 *
 */

/* This file was automatically generated --- DO NOT EDIT */
/* Generated on Sun Nov 25 07:35:42 EST 2012 */

#include "codelet-dft.h"

#ifdef HAVE_FMA

/* Generated by: ../../../genfft/gen_notw.native -fma -reorder-insns -schedule-for-pipeline -compact -variables 4 -pipeline-latency 4 -n 6 -name n1_6 -include n.h */

/*
 * This function contains 36 FP additions, 12 FP multiplications,
 * (or, 24 additions, 0 multiplications, 12 fused multiply/add),
 * 30 stack variables, 2 constants, and 24 memory accesses
 */
#include "n.h"

static void n1_6(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs)
{
     DK(KP866025403, +0.866025403784438646763723170752936183471402627);
     DK(KP500000000, +0.500000000000000000000000000000000000000000000);
     {
	  INT i;
	  for (i = v; i > 0; i = i - 1, ri = ri + ivs, ii = ii + ivs, ro = ro + ovs, io = io + ovs, MAKE_VOLATILE_STRIDE(24, is), MAKE_VOLATILE_STRIDE(24, os)) {
	       E TA, Tz;
	       {
		    E Tb, T3, Tx, Tp, Tj, Te, Ts, Ta, Tu, Ti, Tk;
		    {
			 E T1, T2, Tn, To;
			 T1 = ri[0];
			 T2 = ri[WS(is, 3)];
			 Tn = ii[0];
			 To = ii[WS(is, 3)];
			 {
			      E T4, T5, T7, T8;
			      T4 = ri[WS(is, 2)];
			      Tb = T1 + T2;
			      T3 = T1 - T2;
			      Tx = Tn + To;
			      Tp = Tn - To;
			      T5 = ri[WS(is, 5)];
			      T7 = ri[WS(is, 4)];
			      T8 = ri[WS(is, 1)];
			      {
				   E Tg, Tc, T6, Td, T9, Th;
				   Tg = ii[WS(is, 2)];
				   Tc = T4 + T5;
				   T6 = T4 - T5;
				   Td = T7 + T8;
				   T9 = T7 - T8;
				   Th = ii[WS(is, 5)];
				   Tj = ii[WS(is, 4)];
				   Te = Tc + Td;
				   TA = Td - Tc;
				   Ts = T9 - T6;
				   Ta = T6 + T9;
				   Tu = Tg + Th;
				   Ti = Tg - Th;
				   Tk = ii[WS(is, 1)];
			      }
			 }
		    }
		    ro[WS(os, 3)] = T3 + Ta;
		    ro[0] = Tb + Te;
		    {
			 E Tf, Tv, Tl, Ty, Tr;
			 Tf = FNMS(KP500000000, Ta, T3);
			 Tv = Tj + Tk;
			 Tl = Tj - Tk;
			 {
			      E Tt, Tw, Tq, Tm;
			      Tt = FNMS(KP500000000, Te, Tb);
			      Ty = Tu + Tv;
			      Tw = Tu - Tv;
			      Tq = Ti + Tl;
			      Tm = Ti - Tl;
			      io[0] = Tx + Ty;
			      ro[WS(os, 1)] = FMA(KP866025403, Tm, Tf);
			      ro[WS(os, 5)] = FNMS(KP866025403, Tm, Tf);
			      Tr = FNMS(KP500000000, Tq, Tp);
			      io[WS(os, 3)] = Tp + Tq;
			      ro[WS(os, 2)] = FNMS(KP866025403, Tw, Tt);
			      ro[WS(os, 4)] = FMA(KP866025403, Tw, Tt);
			 }
			 io[WS(os, 5)] = FNMS(KP866025403, Ts, Tr);
			 io[WS(os, 1)] = FMA(KP866025403, Ts, Tr);
			 Tz = FNMS(KP500000000, Ty, Tx);
		    }
	       }
	       io[WS(os, 4)] = FMA(KP866025403, TA, Tz);
	       io[WS(os, 2)] = FNMS(KP866025403, TA, Tz);
	  }
     }
}

static const kdft_desc desc = { 6, "n1_6", {24, 0, 12, 0}, &GENUS, 0, 0, 0, 0 };

void X(codelet_n1_6) (planner *p) {
     X(kdft_register) (p, n1_6, &desc);
}

#else				/* HAVE_FMA */

/* Generated by: ../../../genfft/gen_notw.native -compact -variables 4 -pipeline-latency 4 -n 6 -name n1_6 -include n.h */

/*
 * This function contains 36 FP additions, 8 FP multiplications,
 * (or, 32 additions, 4 multiplications, 4 fused multiply/add),
 * 23 stack variables, 2 constants, and 24 memory accesses
 */
#include "n.h"

static void n1_6(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs)
{
     DK(KP866025403, +0.866025403784438646763723170752936183471402627);
     DK(KP500000000, +0.500000000000000000000000000000000000000000000);
     {
	  INT i;
	  for (i = v; i > 0; i = i - 1, ri = ri + ivs, ii = ii + ivs, ro = ro + ovs, io = io + ovs, MAKE_VOLATILE_STRIDE(24, is), MAKE_VOLATILE_STRIDE(24, os)) {
	       E T3, Tb, Tq, Tx, T6, Tc, T9, Td, Ta, Te, Ti, Tu, Tl, Tv, Tr;
	       E Ty;
	       {
		    E T1, T2, To, Tp;
		    T1 = ri[0];
		    T2 = ri[WS(is, 3)];
		    T3 = T1 - T2;
		    Tb = T1 + T2;
		    To = ii[0];
		    Tp = ii[WS(is, 3)];
		    Tq = To - Tp;
		    Tx = To + Tp;
	       }
	       {
		    E T4, T5, T7, T8;
		    T4 = ri[WS(is, 2)];
		    T5 = ri[WS(is, 5)];
		    T6 = T4 - T5;
		    Tc = T4 + T5;
		    T7 = ri[WS(is, 4)];
		    T8 = ri[WS(is, 1)];
		    T9 = T7 - T8;
		    Td = T7 + T8;
	       }
	       Ta = T6 + T9;
	       Te = Tc + Td;
	       {
		    E Tg, Th, Tj, Tk;
		    Tg = ii[WS(is, 2)];
		    Th = ii[WS(is, 5)];
		    Ti = Tg - Th;
		    Tu = Tg + Th;
		    Tj = ii[WS(is, 4)];
		    Tk = ii[WS(is, 1)];
		    Tl = Tj - Tk;
		    Tv = Tj + Tk;
	       }
	       Tr = Ti + Tl;
	       Ty = Tu + Tv;
	       ro[WS(os, 3)] = T3 + Ta;
	       io[WS(os, 3)] = Tq + Tr;
	       ro[0] = Tb + Te;
	       io[0] = Tx + Ty;
	       {
		    E Tf, Tm, Tn, Ts;
		    Tf = FNMS(KP500000000, Ta, T3);
		    Tm = KP866025403 * (Ti - Tl);
		    ro[WS(os, 5)] = Tf - Tm;
		    ro[WS(os, 1)] = Tf + Tm;
		    Tn = KP866025403 * (T9 - T6);
		    Ts = FNMS(KP500000000, Tr, Tq);
		    io[WS(os, 1)] = Tn + Ts;
		    io[WS(os, 5)] = Ts - Tn;
	       }
	       {
		    E Tt, Tw, Tz, TA;
		    Tt = FNMS(KP500000000, Te, Tb);
		    Tw = KP866025403 * (Tu - Tv);
		    ro[WS(os, 2)] = Tt - Tw;
		    ro[WS(os, 4)] = Tt + Tw;
		    Tz = FNMS(KP500000000, Ty, Tx);
		    TA = KP866025403 * (Td - Tc);
		    io[WS(os, 2)] = Tz - TA;
		    io[WS(os, 4)] = TA + Tz;
	       }
	  }
     }
}

static const kdft_desc desc = { 6, "n1_6", {32, 4, 4, 0}, &GENUS, 0, 0, 0, 0 };

void X(codelet_n1_6) (planner *p) {
     X(kdft_register) (p, n1_6, &desc);
}

#endif				/* HAVE_FMA */