Chris@42: /*
Chris@42:  * Copyright (c) 2003, 2007-14 Matteo Frigo
Chris@42:  * Copyright (c) 2003, 2007-14 Massachusetts Institute of Technology
Chris@42:  *
Chris@42:  * This program is free software; you can redistribute it and/or modify
Chris@42:  * it under the terms of the GNU General Public License as published by
Chris@42:  * the Free Software Foundation; either version 2 of the License, or
Chris@42:  * (at your option) any later version.
Chris@42:  *
Chris@42:  * This program is distributed in the hope that it will be useful,
Chris@42:  * but WITHOUT ANY WARRANTY; without even the implied warranty of
Chris@42:  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
Chris@42:  * GNU General Public License for more details.
Chris@42:  *
Chris@42:  * You should have received a copy of the GNU General Public License
Chris@42:  * along with this program; if not, write to the Free Software
Chris@42:  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02110-1301  USA
Chris@42:  *
Chris@42:  */
Chris@42: 
Chris@42: /* This file was automatically generated --- DO NOT EDIT */
Chris@42: /* Generated on Sat Jul 30 16:48:41 EDT 2016 */
Chris@42: 
Chris@42: #include "codelet-rdft.h"
Chris@42: 
Chris@42: #ifdef HAVE_FMA
Chris@42: 
Chris@42: /* Generated by: ../../../genfft/gen_hc2cdft.native -fma -reorder-insns -schedule-for-pipeline -compact -variables 4 -pipeline-latency 4 -n 6 -dit -name hc2cfdft_6 -include hc2cf.h */
Chris@42: 
Chris@42: /*
Chris@42:  * This function contains 58 FP additions, 44 FP multiplications,
Chris@42:  * (or, 36 additions, 22 multiplications, 22 fused multiply/add),
Chris@42:  * 42 stack variables, 2 constants, and 24 memory accesses
Chris@42:  */
Chris@42: #include "hc2cf.h"
Chris@42: 
Chris@42: static void hc2cfdft_6(R *Rp, R *Ip, R *Rm, R *Im, const R *W, stride rs, INT mb, INT me, INT ms)
Chris@42: {
Chris@42:      DK(KP866025403, +0.866025403784438646763723170752936183471402627);
Chris@42:      DK(KP500000000, +0.500000000000000000000000000000000000000000000);
Chris@42:      {
Chris@42: 	  INT m;
Chris@42: 	  for (m = mb, W = W + ((mb - 1) * 10); m < me; m = m + 1, Rp = Rp + ms, Ip = Ip + ms, Rm = Rm - ms, Im = Im - ms, W = W + 10, MAKE_VOLATILE_STRIDE(24, rs)) {
Chris@42: 	       E TP, TT, TN, TM, TY, T13;
Chris@42: 	       {
Chris@42: 		    E T3, TQ, TJ, T12, Tu, TB, TX, T10, Tj, Tf, Ti, Td, Th, TU, TS;
Chris@42: 		    {
Chris@42: 			 E TC, TI, TF, TH, TA, Tw, TZ;
Chris@42: 			 {
Chris@42: 			      E T1, T2, TD, TE;
Chris@42: 			      T1 = Ip[0];
Chris@42: 			      T2 = Im[0];
Chris@42: 			      TD = Rm[0];
Chris@42: 			      TE = Rp[0];
Chris@42: 			      TC = W[0];
Chris@42: 			      T3 = T1 - T2;
Chris@42: 			      TI = T1 + T2;
Chris@42: 			      TQ = TE + TD;
Chris@42: 			      TF = TD - TE;
Chris@42: 			      TH = W[1];
Chris@42: 			 }
Chris@42: 			 {
Chris@42: 			      E Tr, To, Ts, Tl, Tq;
Chris@42: 			      {
Chris@42: 				   E Tm, Tn, TG, T11;
Chris@42: 				   Tm = Rm[WS(rs, 2)];
Chris@42: 				   Tn = Rp[WS(rs, 2)];
Chris@42: 				   TG = TC * TF;
Chris@42: 				   T11 = TH * TF;
Chris@42: 				   Tr = Ip[WS(rs, 2)];
Chris@42: 				   TA = Tn + Tm;
Chris@42: 				   To = Tm - Tn;
Chris@42: 				   TJ = FNMS(TH, TI, TG);
Chris@42: 				   T12 = FMA(TC, TI, T11);
Chris@42: 				   Ts = Im[WS(rs, 2)];
Chris@42: 			      }
Chris@42: 			      Tl = W[8];
Chris@42: 			      Tq = W[9];
Chris@42: 			      {
Chris@42: 				   E Tz, Ty, TW, Tx, Tt, Tp;
Chris@42: 				   Tw = W[6];
Chris@42: 				   Tx = Tr - Ts;
Chris@42: 				   Tt = Tr + Ts;
Chris@42: 				   Tp = Tl * To;
Chris@42: 				   Tz = W[7];
Chris@42: 				   Ty = Tw * Tx;
Chris@42: 				   TW = Tl * Tt;
Chris@42: 				   Tu = FNMS(Tq, Tt, Tp);
Chris@42: 				   TZ = Tz * Tx;
Chris@42: 				   TB = FNMS(Tz, TA, Ty);
Chris@42: 				   TX = FMA(Tq, To, TW);
Chris@42: 			      }
Chris@42: 			 }
Chris@42: 			 {
Chris@42: 			      E T5, T6, Ta, Tb;
Chris@42: 			      T5 = Ip[WS(rs, 1)];
Chris@42: 			      T10 = FMA(Tw, TA, TZ);
Chris@42: 			      T6 = Im[WS(rs, 1)];
Chris@42: 			      Ta = Rp[WS(rs, 1)];
Chris@42: 			      Tb = Rm[WS(rs, 1)];
Chris@42: 			      {
Chris@42: 				   E T4, Tg, T7, Tc, T9, T8, TR;
Chris@42: 				   T4 = W[5];
Chris@42: 				   Tg = T5 - T6;
Chris@42: 				   T7 = T5 + T6;
Chris@42: 				   Tj = Ta + Tb;
Chris@42: 				   Tc = Ta - Tb;
Chris@42: 				   T9 = W[4];
Chris@42: 				   T8 = T4 * T7;
Chris@42: 				   Tf = W[2];
Chris@42: 				   Ti = W[3];
Chris@42: 				   TR = T9 * T7;
Chris@42: 				   Td = FMA(T9, Tc, T8);
Chris@42: 				   Th = Tf * Tg;
Chris@42: 				   TU = Ti * Tg;
Chris@42: 				   TS = FNMS(T4, Tc, TR);
Chris@42: 			      }
Chris@42: 			 }
Chris@42: 		    }
Chris@42: 		    {
Chris@42: 			 E Te, T1d, TK, Tv, T1a, T1b, Tk, TV;
Chris@42: 			 TP = Td + T3;
Chris@42: 			 Te = T3 - Td;
Chris@42: 			 Tk = FNMS(Ti, Tj, Th);
Chris@42: 			 TV = FMA(Tf, Tj, TU);
Chris@42: 			 T1d = TQ + TS;
Chris@42: 			 TT = TQ - TS;
Chris@42: 			 TN = TJ - TB;
Chris@42: 			 TK = TB + TJ;
Chris@42: 			 Tv = Tk + Tu;
Chris@42: 			 TM = Tu - Tk;
Chris@42: 			 TY = TV - TX;
Chris@42: 			 T1a = TV + TX;
Chris@42: 			 T1b = T10 + T12;
Chris@42: 			 T13 = T10 - T12;
Chris@42: 			 {
Chris@42: 			      E T1g, TL, T1e, T1c, T19, T1f;
Chris@42: 			      T1g = Tv - TK;
Chris@42: 			      TL = Tv + TK;
Chris@42: 			      T1e = T1a + T1b;
Chris@42: 			      T1c = T1a - T1b;
Chris@42: 			      T19 = FNMS(KP500000000, TL, Te);
Chris@42: 			      Ip[0] = KP500000000 * (Te + TL);
Chris@42: 			      T1f = FNMS(KP500000000, T1e, T1d);
Chris@42: 			      Rp[0] = KP500000000 * (T1d + T1e);
Chris@42: 			      Im[WS(rs, 1)] = -(KP500000000 * (FNMS(KP866025403, T1c, T19)));
Chris@42: 			      Ip[WS(rs, 2)] = KP500000000 * (FMA(KP866025403, T1c, T19));
Chris@42: 			      Rm[WS(rs, 1)] = KP500000000 * (FMA(KP866025403, T1g, T1f));
Chris@42: 			      Rp[WS(rs, 2)] = KP500000000 * (FNMS(KP866025403, T1g, T1f));
Chris@42: 			 }
Chris@42: 		    }
Chris@42: 	       }
Chris@42: 	       {
Chris@42: 		    E TO, T16, T14, T18, T17, T15;
Chris@42: 		    TO = TM + TN;
Chris@42: 		    T16 = TN - TM;
Chris@42: 		    T14 = TY + T13;
Chris@42: 		    T18 = T13 - TY;
Chris@42: 		    T17 = FMA(KP500000000, TO, TP);
Chris@42: 		    Im[WS(rs, 2)] = KP500000000 * (TO - TP);
Chris@42: 		    T15 = FNMS(KP500000000, T14, TT);
Chris@42: 		    Rm[WS(rs, 2)] = KP500000000 * (TT + T14);
Chris@42: 		    Im[0] = -(KP500000000 * (FNMS(KP866025403, T18, T17)));
Chris@42: 		    Ip[WS(rs, 1)] = KP500000000 * (FMA(KP866025403, T18, T17));
Chris@42: 		    Rm[0] = KP500000000 * (FNMS(KP866025403, T16, T15));
Chris@42: 		    Rp[WS(rs, 1)] = KP500000000 * (FMA(KP866025403, T16, T15));
Chris@42: 	       }
Chris@42: 	  }
Chris@42:      }
Chris@42: }
Chris@42: 
Chris@42: static const tw_instr twinstr[] = {
Chris@42:      {TW_FULL, 1, 6},
Chris@42:      {TW_NEXT, 1, 0}
Chris@42: };
Chris@42: 
Chris@42: static const hc2c_desc desc = { 6, "hc2cfdft_6", twinstr, &GENUS, {36, 22, 22, 0} };
Chris@42: 
Chris@42: void X(codelet_hc2cfdft_6) (planner *p) {
Chris@42:      X(khc2c_register) (p, hc2cfdft_6, &desc, HC2C_VIA_DFT);
Chris@42: }
Chris@42: #else				/* HAVE_FMA */
Chris@42: 
Chris@42: /* Generated by: ../../../genfft/gen_hc2cdft.native -compact -variables 4 -pipeline-latency 4 -n 6 -dit -name hc2cfdft_6 -include hc2cf.h */
Chris@42: 
Chris@42: /*
Chris@42:  * This function contains 58 FP additions, 36 FP multiplications,
Chris@42:  * (or, 44 additions, 22 multiplications, 14 fused multiply/add),
Chris@42:  * 40 stack variables, 3 constants, and 24 memory accesses
Chris@42:  */
Chris@42: #include "hc2cf.h"
Chris@42: 
Chris@42: static void hc2cfdft_6(R *Rp, R *Ip, R *Rm, R *Im, const R *W, stride rs, INT mb, INT me, INT ms)
Chris@42: {
Chris@42:      DK(KP250000000, +0.250000000000000000000000000000000000000000000);
Chris@42:      DK(KP500000000, +0.500000000000000000000000000000000000000000000);
Chris@42:      DK(KP433012701, +0.433012701892219323381861585376468091735701313);
Chris@42:      {
Chris@42: 	  INT m;
Chris@42: 	  for (m = mb, W = W + ((mb - 1) * 10); m < me; m = m + 1, Rp = Rp + ms, Ip = Ip + ms, Rm = Rm - ms, Im = Im - ms, W = W + 10, MAKE_VOLATILE_STRIDE(24, rs)) {
Chris@42: 	       E T3, TM, Tc, TN, Ts, T10, TI, TR, TF, T11, TH, TU;
Chris@42: 	       {
Chris@42: 		    E T1, T2, TD, Tz, TA, TB, T7, Tf, Tb, Th, Tq, Tw, Tm, Tu, T4;
Chris@42: 		    E T8;
Chris@42: 		    {
Chris@42: 			 E T5, T6, T9, Ta;
Chris@42: 			 T1 = Ip[0];
Chris@42: 			 T2 = Im[0];
Chris@42: 			 TD = T1 + T2;
Chris@42: 			 Tz = Rm[0];
Chris@42: 			 TA = Rp[0];
Chris@42: 			 TB = Tz - TA;
Chris@42: 			 T5 = Ip[WS(rs, 1)];
Chris@42: 			 T6 = Im[WS(rs, 1)];
Chris@42: 			 T7 = T5 + T6;
Chris@42: 			 Tf = T5 - T6;
Chris@42: 			 T9 = Rp[WS(rs, 1)];
Chris@42: 			 Ta = Rm[WS(rs, 1)];
Chris@42: 			 Tb = T9 - Ta;
Chris@42: 			 Th = T9 + Ta;
Chris@42: 			 {
Chris@42: 			      E To, Tp, Tk, Tl;
Chris@42: 			      To = Rp[WS(rs, 2)];
Chris@42: 			      Tp = Rm[WS(rs, 2)];
Chris@42: 			      Tq = To - Tp;
Chris@42: 			      Tw = To + Tp;
Chris@42: 			      Tk = Ip[WS(rs, 2)];
Chris@42: 			      Tl = Im[WS(rs, 2)];
Chris@42: 			      Tm = Tk + Tl;
Chris@42: 			      Tu = Tk - Tl;
Chris@42: 			 }
Chris@42: 		    }
Chris@42: 		    T3 = T1 - T2;
Chris@42: 		    TM = TA + Tz;
Chris@42: 		    T4 = W[5];
Chris@42: 		    T8 = W[4];
Chris@42: 		    Tc = FMA(T4, T7, T8 * Tb);
Chris@42: 		    TN = FNMS(T4, Tb, T8 * T7);
Chris@42: 		    {
Chris@42: 			 E Ti, TP, Tr, TQ;
Chris@42: 			 {
Chris@42: 			      E Te, Tg, Tj, Tn;
Chris@42: 			      Te = W[2];
Chris@42: 			      Tg = W[3];
Chris@42: 			      Ti = FNMS(Tg, Th, Te * Tf);
Chris@42: 			      TP = FMA(Tg, Tf, Te * Th);
Chris@42: 			      Tj = W[9];
Chris@42: 			      Tn = W[8];
Chris@42: 			      Tr = FMA(Tj, Tm, Tn * Tq);
Chris@42: 			      TQ = FNMS(Tj, Tq, Tn * Tm);
Chris@42: 			 }
Chris@42: 			 Ts = Ti - Tr;
Chris@42: 			 T10 = TP + TQ;
Chris@42: 			 TI = Ti + Tr;
Chris@42: 			 TR = TP - TQ;
Chris@42: 		    }
Chris@42: 		    {
Chris@42: 			 E Tx, TS, TE, TT;
Chris@42: 			 {
Chris@42: 			      E Tt, Tv, Ty, TC;
Chris@42: 			      Tt = W[6];
Chris@42: 			      Tv = W[7];
Chris@42: 			      Tx = FNMS(Tv, Tw, Tt * Tu);
Chris@42: 			      TS = FMA(Tv, Tu, Tt * Tw);
Chris@42: 			      Ty = W[0];
Chris@42: 			      TC = W[1];
Chris@42: 			      TE = FNMS(TC, TD, Ty * TB);
Chris@42: 			      TT = FMA(TC, TB, Ty * TD);
Chris@42: 			 }
Chris@42: 			 TF = Tx + TE;
Chris@42: 			 T11 = TS + TT;
Chris@42: 			 TH = TE - Tx;
Chris@42: 			 TU = TS - TT;
Chris@42: 		    }
Chris@42: 	       }
Chris@42: 	       {
Chris@42: 		    E T12, Td, TG, TZ;
Chris@42: 		    T12 = KP433012701 * (T10 - T11);
Chris@42: 		    Td = T3 - Tc;
Chris@42: 		    TG = Ts + TF;
Chris@42: 		    TZ = FNMS(KP250000000, TG, KP500000000 * Td);
Chris@42: 		    Ip[0] = KP500000000 * (Td + TG);
Chris@42: 		    Im[WS(rs, 1)] = T12 - TZ;
Chris@42: 		    Ip[WS(rs, 2)] = TZ + T12;
Chris@42: 	       }
Chris@42: 	       {
Chris@42: 		    E T16, T13, T14, T15;
Chris@42: 		    T16 = KP433012701 * (Ts - TF);
Chris@42: 		    T13 = TM + TN;
Chris@42: 		    T14 = T10 + T11;
Chris@42: 		    T15 = FNMS(KP250000000, T14, KP500000000 * T13);
Chris@42: 		    Rp[WS(rs, 2)] = T15 - T16;
Chris@42: 		    Rp[0] = KP500000000 * (T13 + T14);
Chris@42: 		    Rm[WS(rs, 1)] = T16 + T15;
Chris@42: 	       }
Chris@42: 	       {
Chris@42: 		    E TY, TJ, TK, TX;
Chris@42: 		    TY = KP433012701 * (TU - TR);
Chris@42: 		    TJ = TH - TI;
Chris@42: 		    TK = Tc + T3;
Chris@42: 		    TX = FMA(KP500000000, TK, KP250000000 * TJ);
Chris@42: 		    Im[WS(rs, 2)] = KP500000000 * (TJ - TK);
Chris@42: 		    Im[0] = TY - TX;
Chris@42: 		    Ip[WS(rs, 1)] = TX + TY;
Chris@42: 	       }
Chris@42: 	       {
Chris@42: 		    E TL, TO, TV, TW;
Chris@42: 		    TL = KP433012701 * (TI + TH);
Chris@42: 		    TO = TM - TN;
Chris@42: 		    TV = TR + TU;
Chris@42: 		    TW = FNMS(KP250000000, TV, KP500000000 * TO);
Chris@42: 		    Rp[WS(rs, 1)] = TL + TW;
Chris@42: 		    Rm[WS(rs, 2)] = KP500000000 * (TO + TV);
Chris@42: 		    Rm[0] = TW - TL;
Chris@42: 	       }
Chris@42: 	  }
Chris@42:      }
Chris@42: }
Chris@42: 
Chris@42: static const tw_instr twinstr[] = {
Chris@42:      {TW_FULL, 1, 6},
Chris@42:      {TW_NEXT, 1, 0}
Chris@42: };
Chris@42: 
Chris@42: static const hc2c_desc desc = { 6, "hc2cfdft_6", twinstr, &GENUS, {44, 22, 14, 0} };
Chris@42: 
Chris@42: void X(codelet_hc2cfdft_6) (planner *p) {
Chris@42:      X(khc2c_register) (p, hc2cfdft_6, &desc, HC2C_VIA_DFT);
Chris@42: }
Chris@42: #endif				/* HAVE_FMA */