Chris@10: /*
Chris@10:  * Copyright (c) 2003, 2007-11 Matteo Frigo
Chris@10:  * Copyright (c) 2003, 2007-11 Massachusetts Institute of Technology
Chris@10:  *
Chris@10:  * This program is free software; you can redistribute it and/or modify
Chris@10:  * it under the terms of the GNU General Public License as published by
Chris@10:  * the Free Software Foundation; either version 2 of the License, or
Chris@10:  * (at your option) any later version.
Chris@10:  *
Chris@10:  * This program is distributed in the hope that it will be useful,
Chris@10:  * but WITHOUT ANY WARRANTY; without even the implied warranty of
Chris@10:  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
Chris@10:  * GNU General Public License for more details.
Chris@10:  *
Chris@10:  * You should have received a copy of the GNU General Public License
Chris@10:  * along with this program; if not, write to the Free Software
Chris@10:  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02110-1301  USA
Chris@10:  *
Chris@10:  */
Chris@10: 
Chris@10: /* This file was automatically generated --- DO NOT EDIT */
Chris@10: /* Generated on Sun Nov 25 07:35:46 EST 2012 */
Chris@10: 
Chris@10: #include "codelet-dft.h"
Chris@10: 
Chris@10: #ifdef HAVE_FMA
Chris@10: 
Chris@10: /* Generated by: ../../../genfft/gen_notw.native -fma -reorder-insns -schedule-for-pipeline -compact -variables 4 -pipeline-latency 4 -n 20 -name n1_20 -include n.h */
Chris@10: 
Chris@10: /*
Chris@10:  * This function contains 208 FP additions, 72 FP multiplications,
Chris@10:  * (or, 136 additions, 0 multiplications, 72 fused multiply/add),
Chris@10:  * 86 stack variables, 4 constants, and 80 memory accesses
Chris@10:  */
Chris@10: #include "n.h"
Chris@10: 
Chris@10: static void n1_20(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs)
Chris@10: {
Chris@10:      DK(KP951056516, +0.951056516295153572116439333379382143405698634);
Chris@10:      DK(KP559016994, +0.559016994374947424102293417182819058860154590);
Chris@10:      DK(KP618033988, +0.618033988749894848204586834365638117720309180);
Chris@10:      DK(KP250000000, +0.250000000000000000000000000000000000000000000);
Chris@10:      {
Chris@10: 	  INT i;
Chris@10: 	  for (i = v; i > 0; i = i - 1, ri = ri + ivs, ii = ii + ivs, ro = ro + ovs, io = io + ovs, MAKE_VOLATILE_STRIDE(80, is), MAKE_VOLATILE_STRIDE(80, os)) {
Chris@10: 	       E T1Y, T1Z, T1W, T1V;
Chris@10: 	       {
Chris@10: 		    E T1d, TP, TD, T7, T3b, T2N, T2f, T1R, T2U, TB, T2P, T2A, T3d, T37, T3j;
Chris@10: 		    E TJ, T2n, T1b, T1T, T1y, T2b, T2h, T1j, T2V, Tm, T2O, T2H, T3c, T34, T1e;
Chris@10: 		    E T1f, T3i, TG, T2m, T10, T1S, T1J, T28, T2g;
Chris@10: 		    {
Chris@10: 			 E T4, T1N, T3, T2L, TN, T5, T1O, T1P, T1h, T1i;
Chris@10: 			 {
Chris@10: 			      E T1, T2, TL, TM;
Chris@10: 			      T1 = ri[0];
Chris@10: 			      T2 = ri[WS(is, 10)];
Chris@10: 			      TL = ii[0];
Chris@10: 			      TM = ii[WS(is, 10)];
Chris@10: 			      T4 = ri[WS(is, 5)];
Chris@10: 			      T1N = T1 - T2;
Chris@10: 			      T3 = T1 + T2;
Chris@10: 			      T2L = TL + TM;
Chris@10: 			      TN = TL - TM;
Chris@10: 			      T5 = ri[WS(is, 15)];
Chris@10: 			      T1O = ii[WS(is, 5)];
Chris@10: 			      T1P = ii[WS(is, 15)];
Chris@10: 			 }
Chris@10: 			 {
Chris@10: 			      E T1o, Tp, T2u, T13, T14, Ts, T2v, T1r, Tx, T1t, Tw, T2x, T18, Ty, T1u;
Chris@10: 			      E T1v;
Chris@10: 			      {
Chris@10: 				   E Tq, Tr, T1p, T1q;
Chris@10: 				   {
Chris@10: 					E Tn, To, T11, T12;
Chris@10: 					Tn = ri[WS(is, 8)];
Chris@10: 					{
Chris@10: 					     E TO, T6, T2M, T1Q;
Chris@10: 					     TO = T4 - T5;
Chris@10: 					     T6 = T4 + T5;
Chris@10: 					     T2M = T1O + T1P;
Chris@10: 					     T1Q = T1O - T1P;
Chris@10: 					     T1d = TO + TN;
Chris@10: 					     TP = TN - TO;
Chris@10: 					     TD = T3 + T6;
Chris@10: 					     T7 = T3 - T6;
Chris@10: 					     T3b = T2L + T2M;
Chris@10: 					     T2N = T2L - T2M;
Chris@10: 					     T2f = T1N + T1Q;
Chris@10: 					     T1R = T1N - T1Q;
Chris@10: 					     To = ri[WS(is, 18)];
Chris@10: 					}
Chris@10: 					T11 = ii[WS(is, 8)];
Chris@10: 					T12 = ii[WS(is, 18)];
Chris@10: 					Tq = ri[WS(is, 13)];
Chris@10: 					T1o = Tn - To;
Chris@10: 					Tp = Tn + To;
Chris@10: 					T2u = T11 + T12;
Chris@10: 					T13 = T11 - T12;
Chris@10: 					Tr = ri[WS(is, 3)];
Chris@10: 					T1p = ii[WS(is, 13)];
Chris@10: 					T1q = ii[WS(is, 3)];
Chris@10: 				   }
Chris@10: 				   {
Chris@10: 					E Tu, Tv, T16, T17;
Chris@10: 					Tu = ri[WS(is, 12)];
Chris@10: 					T14 = Tq - Tr;
Chris@10: 					Ts = Tq + Tr;
Chris@10: 					T2v = T1p + T1q;
Chris@10: 					T1r = T1p - T1q;
Chris@10: 					Tv = ri[WS(is, 2)];
Chris@10: 					T16 = ii[WS(is, 12)];
Chris@10: 					T17 = ii[WS(is, 2)];
Chris@10: 					Tx = ri[WS(is, 17)];
Chris@10: 					T1t = Tu - Tv;
Chris@10: 					Tw = Tu + Tv;
Chris@10: 					T2x = T16 + T17;
Chris@10: 					T18 = T16 - T17;
Chris@10: 					Ty = ri[WS(is, 7)];
Chris@10: 					T1u = ii[WS(is, 17)];
Chris@10: 					T1v = ii[WS(is, 7)];
Chris@10: 				   }
Chris@10: 			      }
Chris@10: 			      {
Chris@10: 				   E TH, T19, T1w, TI;
Chris@10: 				   {
Chris@10: 					E Tt, T2w, T35, TA, T2z, T36, Tz, T2y;
Chris@10: 					TH = Tp + Ts;
Chris@10: 					Tt = Tp - Ts;
Chris@10: 					T19 = Tx - Ty;
Chris@10: 					Tz = Tx + Ty;
Chris@10: 					T2y = T1u + T1v;
Chris@10: 					T1w = T1u - T1v;
Chris@10: 					T2w = T2u - T2v;
Chris@10: 					T35 = T2u + T2v;
Chris@10: 					TI = Tw + Tz;
Chris@10: 					TA = Tw - Tz;
Chris@10: 					T2z = T2x - T2y;
Chris@10: 					T36 = T2x + T2y;
Chris@10: 					T2U = Tt - TA;
Chris@10: 					TB = Tt + TA;
Chris@10: 					T2P = T2w + T2z;
Chris@10: 					T2A = T2w - T2z;
Chris@10: 					T3d = T35 + T36;
Chris@10: 					T37 = T35 - T36;
Chris@10: 				   }
Chris@10: 				   {
Chris@10: 					E T1s, T29, T1x, T2a, T15, T1a;
Chris@10: 					T15 = T13 - T14;
Chris@10: 					T1h = T14 + T13;
Chris@10: 					T1i = T19 + T18;
Chris@10: 					T1a = T18 - T19;
Chris@10: 					T1s = T1o - T1r;
Chris@10: 					T29 = T1o + T1r;
Chris@10: 					T3j = TH - TI;
Chris@10: 					TJ = TH + TI;
Chris@10: 					T1x = T1t - T1w;
Chris@10: 					T2a = T1t + T1w;
Chris@10: 					T2n = T15 - T1a;
Chris@10: 					T1b = T15 + T1a;
Chris@10: 					T1T = T1s + T1x;
Chris@10: 					T1y = T1s - T1x;
Chris@10: 					T2b = T29 - T2a;
Chris@10: 					T2h = T29 + T2a;
Chris@10: 				   }
Chris@10: 			      }
Chris@10: 			 }
Chris@10: 			 {
Chris@10: 			      E Ta, T1z, T2B, TS, TT, Td, T2C, T1C, Ti, T1E, Th, T2E, TX, Tj, T1F;
Chris@10: 			      E T1G;
Chris@10: 			      {
Chris@10: 				   E Tb, Tc, T1A, T1B;
Chris@10: 				   {
Chris@10: 					E TQ, TR, T8, T9;
Chris@10: 					T8 = ri[WS(is, 4)];
Chris@10: 					T9 = ri[WS(is, 14)];
Chris@10: 					T1j = T1h + T1i;
Chris@10: 					T1Y = T1h - T1i;
Chris@10: 					TQ = ii[WS(is, 4)];
Chris@10: 					TR = ii[WS(is, 14)];
Chris@10: 					Ta = T8 + T9;
Chris@10: 					T1z = T8 - T9;
Chris@10: 					Tb = ri[WS(is, 9)];
Chris@10: 					T2B = TQ + TR;
Chris@10: 					TS = TQ - TR;
Chris@10: 					Tc = ri[WS(is, 19)];
Chris@10: 					T1A = ii[WS(is, 9)];
Chris@10: 					T1B = ii[WS(is, 19)];
Chris@10: 				   }
Chris@10: 				   {
Chris@10: 					E Tf, Tg, TV, TW;
Chris@10: 					Tf = ri[WS(is, 16)];
Chris@10: 					TT = Tb - Tc;
Chris@10: 					Td = Tb + Tc;
Chris@10: 					T2C = T1A + T1B;
Chris@10: 					T1C = T1A - T1B;
Chris@10: 					Tg = ri[WS(is, 6)];
Chris@10: 					TV = ii[WS(is, 16)];
Chris@10: 					TW = ii[WS(is, 6)];
Chris@10: 					Ti = ri[WS(is, 1)];
Chris@10: 					T1E = Tf - Tg;
Chris@10: 					Th = Tf + Tg;
Chris@10: 					T2E = TV + TW;
Chris@10: 					TX = TV - TW;
Chris@10: 					Tj = ri[WS(is, 11)];
Chris@10: 					T1F = ii[WS(is, 1)];
Chris@10: 					T1G = ii[WS(is, 11)];
Chris@10: 				   }
Chris@10: 			      }
Chris@10: 			      {
Chris@10: 				   E TE, TY, T1H, TF;
Chris@10: 				   {
Chris@10: 					E Te, T2D, T32, Tl, T2G, T33, Tk, T2F;
Chris@10: 					TE = Ta + Td;
Chris@10: 					Te = Ta - Td;
Chris@10: 					TY = Ti - Tj;
Chris@10: 					Tk = Ti + Tj;
Chris@10: 					T2F = T1F + T1G;
Chris@10: 					T1H = T1F - T1G;
Chris@10: 					T2D = T2B - T2C;
Chris@10: 					T32 = T2B + T2C;
Chris@10: 					TF = Th + Tk;
Chris@10: 					Tl = Th - Tk;
Chris@10: 					T2G = T2E - T2F;
Chris@10: 					T33 = T2E + T2F;
Chris@10: 					T2V = Te - Tl;
Chris@10: 					Tm = Te + Tl;
Chris@10: 					T2O = T2D + T2G;
Chris@10: 					T2H = T2D - T2G;
Chris@10: 					T3c = T32 + T33;
Chris@10: 					T34 = T32 - T33;
Chris@10: 				   }
Chris@10: 				   {
Chris@10: 					E T1D, T26, T1I, T27, TU, TZ;
Chris@10: 					TU = TS - TT;
Chris@10: 					T1e = TT + TS;
Chris@10: 					T1f = TY + TX;
Chris@10: 					TZ = TX - TY;
Chris@10: 					T1D = T1z - T1C;
Chris@10: 					T26 = T1z + T1C;
Chris@10: 					T3i = TE - TF;
Chris@10: 					TG = TE + TF;
Chris@10: 					T1I = T1E - T1H;
Chris@10: 					T27 = T1E + T1H;
Chris@10: 					T2m = TU - TZ;
Chris@10: 					T10 = TU + TZ;
Chris@10: 					T1S = T1D + T1I;
Chris@10: 					T1J = T1D - T1I;
Chris@10: 					T28 = T26 - T27;
Chris@10: 					T2g = T26 + T27;
Chris@10: 				   }
Chris@10: 			      }
Chris@10: 			 }
Chris@10: 		    }
Chris@10: 		    {
Chris@10: 			 E T1g, T3g, T3f, T2S, T2R, T2k, T2j;
Chris@10: 			 {
Chris@10: 			      E T2s, T2r, TC, T2Q;
Chris@10: 			      T2s = Tm - TB;
Chris@10: 			      TC = Tm + TB;
Chris@10: 			      T1g = T1e + T1f;
Chris@10: 			      T1Z = T1e - T1f;
Chris@10: 			      T2r = FNMS(KP250000000, TC, T7);
Chris@10: 			      ro[WS(os, 10)] = T7 + TC;
Chris@10: 			      T2Q = T2O + T2P;
Chris@10: 			      T2S = T2O - T2P;
Chris@10: 			      {
Chris@10: 				   E T2K, T2I, T2t, T2J;
Chris@10: 				   T2K = FMA(KP618033988, T2A, T2H);
Chris@10: 				   T2I = FNMS(KP618033988, T2H, T2A);
Chris@10: 				   T2t = FNMS(KP559016994, T2s, T2r);
Chris@10: 				   T2J = FMA(KP559016994, T2s, T2r);
Chris@10: 				   ro[WS(os, 18)] = FMA(KP951056516, T2I, T2t);
Chris@10: 				   ro[WS(os, 2)] = FNMS(KP951056516, T2I, T2t);
Chris@10: 				   ro[WS(os, 6)] = FMA(KP951056516, T2K, T2J);
Chris@10: 				   ro[WS(os, 14)] = FNMS(KP951056516, T2K, T2J);
Chris@10: 				   T2R = FNMS(KP250000000, T2Q, T2N);
Chris@10: 			      }
Chris@10: 			      io[WS(os, 10)] = T2N + T2Q;
Chris@10: 			 }
Chris@10: 			 {
Chris@10: 			      E T30, T2Z, TK, T3e;
Chris@10: 			      TK = TG + TJ;
Chris@10: 			      T30 = TG - TJ;
Chris@10: 			      {
Chris@10: 				   E T2T, T2X, T2Y, T2W;
Chris@10: 				   T2T = FNMS(KP559016994, T2S, T2R);
Chris@10: 				   T2X = FMA(KP559016994, T2S, T2R);
Chris@10: 				   T2Y = FMA(KP618033988, T2U, T2V);
Chris@10: 				   T2W = FNMS(KP618033988, T2V, T2U);
Chris@10: 				   io[WS(os, 14)] = FMA(KP951056516, T2Y, T2X);
Chris@10: 				   io[WS(os, 6)] = FNMS(KP951056516, T2Y, T2X);
Chris@10: 				   io[WS(os, 18)] = FNMS(KP951056516, T2W, T2T);
Chris@10: 				   io[WS(os, 2)] = FMA(KP951056516, T2W, T2T);
Chris@10: 				   T2Z = FNMS(KP250000000, TK, TD);
Chris@10: 			      }
Chris@10: 			      ro[0] = TD + TK;
Chris@10: 			      T3e = T3c + T3d;
Chris@10: 			      T3g = T3c - T3d;
Chris@10: 			      {
Chris@10: 				   E T31, T39, T3a, T38;
Chris@10: 				   T31 = FMA(KP559016994, T30, T2Z);
Chris@10: 				   T39 = FNMS(KP559016994, T30, T2Z);
Chris@10: 				   T3a = FNMS(KP618033988, T34, T37);
Chris@10: 				   T38 = FMA(KP618033988, T37, T34);
Chris@10: 				   ro[WS(os, 8)] = FMA(KP951056516, T3a, T39);
Chris@10: 				   ro[WS(os, 12)] = FNMS(KP951056516, T3a, T39);
Chris@10: 				   ro[WS(os, 16)] = FMA(KP951056516, T38, T31);
Chris@10: 				   ro[WS(os, 4)] = FNMS(KP951056516, T38, T31);
Chris@10: 				   T3f = FNMS(KP250000000, T3e, T3b);
Chris@10: 			      }
Chris@10: 			      io[0] = T3b + T3e;
Chris@10: 			 }
Chris@10: 			 {
Chris@10: 			      E T24, T23, T1c, T2i;
Chris@10: 			      T1c = T10 + T1b;
Chris@10: 			      T24 = T10 - T1b;
Chris@10: 			      {
Chris@10: 				   E T3h, T3l, T3m, T3k;
Chris@10: 				   T3h = FMA(KP559016994, T3g, T3f);
Chris@10: 				   T3l = FNMS(KP559016994, T3g, T3f);
Chris@10: 				   T3m = FNMS(KP618033988, T3i, T3j);
Chris@10: 				   T3k = FMA(KP618033988, T3j, T3i);
Chris@10: 				   io[WS(os, 12)] = FMA(KP951056516, T3m, T3l);
Chris@10: 				   io[WS(os, 8)] = FNMS(KP951056516, T3m, T3l);
Chris@10: 				   io[WS(os, 16)] = FNMS(KP951056516, T3k, T3h);
Chris@10: 				   io[WS(os, 4)] = FMA(KP951056516, T3k, T3h);
Chris@10: 				   T23 = FNMS(KP250000000, T1c, TP);
Chris@10: 			      }
Chris@10: 			      io[WS(os, 5)] = TP + T1c;
Chris@10: 			      T2i = T2g + T2h;
Chris@10: 			      T2k = T2g - T2h;
Chris@10: 			      {
Chris@10: 				   E T25, T2d, T2e, T2c;
Chris@10: 				   T25 = FMA(KP559016994, T24, T23);
Chris@10: 				   T2d = FNMS(KP559016994, T24, T23);
Chris@10: 				   T2e = FNMS(KP618033988, T28, T2b);
Chris@10: 				   T2c = FMA(KP618033988, T2b, T28);
Chris@10: 				   io[WS(os, 17)] = FMA(KP951056516, T2e, T2d);
Chris@10: 				   io[WS(os, 13)] = FNMS(KP951056516, T2e, T2d);
Chris@10: 				   io[WS(os, 9)] = FMA(KP951056516, T2c, T25);
Chris@10: 				   io[WS(os, 1)] = FNMS(KP951056516, T2c, T25);
Chris@10: 				   T2j = FNMS(KP250000000, T2i, T2f);
Chris@10: 			      }
Chris@10: 			      ro[WS(os, 5)] = T2f + T2i;
Chris@10: 			 }
Chris@10: 			 {
Chris@10: 			      E T1m, T1l, T1k, T1U;
Chris@10: 			      T1k = T1g + T1j;
Chris@10: 			      T1m = T1g - T1j;
Chris@10: 			      {
Chris@10: 				   E T2l, T2p, T2q, T2o;
Chris@10: 				   T2l = FMA(KP559016994, T2k, T2j);
Chris@10: 				   T2p = FNMS(KP559016994, T2k, T2j);
Chris@10: 				   T2q = FNMS(KP618033988, T2m, T2n);
Chris@10: 				   T2o = FMA(KP618033988, T2n, T2m);
Chris@10: 				   ro[WS(os, 17)] = FNMS(KP951056516, T2q, T2p);
Chris@10: 				   ro[WS(os, 13)] = FMA(KP951056516, T2q, T2p);
Chris@10: 				   ro[WS(os, 9)] = FNMS(KP951056516, T2o, T2l);
Chris@10: 				   ro[WS(os, 1)] = FMA(KP951056516, T2o, T2l);
Chris@10: 				   T1l = FNMS(KP250000000, T1k, T1d);
Chris@10: 			      }
Chris@10: 			      io[WS(os, 15)] = T1d + T1k;
Chris@10: 			      T1U = T1S + T1T;
Chris@10: 			      T1W = T1S - T1T;
Chris@10: 			      {
Chris@10: 				   E T1n, T1L, T1M, T1K;
Chris@10: 				   T1n = FNMS(KP559016994, T1m, T1l);
Chris@10: 				   T1L = FMA(KP559016994, T1m, T1l);
Chris@10: 				   T1M = FMA(KP618033988, T1y, T1J);
Chris@10: 				   T1K = FNMS(KP618033988, T1J, T1y);
Chris@10: 				   io[WS(os, 19)] = FMA(KP951056516, T1M, T1L);
Chris@10: 				   io[WS(os, 11)] = FNMS(KP951056516, T1M, T1L);
Chris@10: 				   io[WS(os, 7)] = FMA(KP951056516, T1K, T1n);
Chris@10: 				   io[WS(os, 3)] = FNMS(KP951056516, T1K, T1n);
Chris@10: 				   T1V = FNMS(KP250000000, T1U, T1R);
Chris@10: 			      }
Chris@10: 			      ro[WS(os, 15)] = T1R + T1U;
Chris@10: 			 }
Chris@10: 		    }
Chris@10: 	       }
Chris@10: 	       {
Chris@10: 		    E T21, T1X, T20, T22;
Chris@10: 		    T21 = FMA(KP559016994, T1W, T1V);
Chris@10: 		    T1X = FNMS(KP559016994, T1W, T1V);
Chris@10: 		    T20 = FNMS(KP618033988, T1Z, T1Y);
Chris@10: 		    T22 = FMA(KP618033988, T1Y, T1Z);
Chris@10: 		    ro[WS(os, 19)] = FNMS(KP951056516, T22, T21);
Chris@10: 		    ro[WS(os, 11)] = FMA(KP951056516, T22, T21);
Chris@10: 		    ro[WS(os, 7)] = FNMS(KP951056516, T20, T1X);
Chris@10: 		    ro[WS(os, 3)] = FMA(KP951056516, T20, T1X);
Chris@10: 	       }
Chris@10: 	  }
Chris@10:      }
Chris@10: }
Chris@10: 
Chris@10: static const kdft_desc desc = { 20, "n1_20", {136, 0, 72, 0}, &GENUS, 0, 0, 0, 0 };
Chris@10: 
Chris@10: void X(codelet_n1_20) (planner *p) {
Chris@10:      X(kdft_register) (p, n1_20, &desc);
Chris@10: }
Chris@10: 
Chris@10: #else				/* HAVE_FMA */
Chris@10: 
Chris@10: /* Generated by: ../../../genfft/gen_notw.native -compact -variables 4 -pipeline-latency 4 -n 20 -name n1_20 -include n.h */
Chris@10: 
Chris@10: /*
Chris@10:  * This function contains 208 FP additions, 48 FP multiplications,
Chris@10:  * (or, 184 additions, 24 multiplications, 24 fused multiply/add),
Chris@10:  * 81 stack variables, 4 constants, and 80 memory accesses
Chris@10:  */
Chris@10: #include "n.h"
Chris@10: 
Chris@10: static void n1_20(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs)
Chris@10: {
Chris@10:      DK(KP587785252, +0.587785252292473129168705954639072768597652438);
Chris@10:      DK(KP951056516, +0.951056516295153572116439333379382143405698634);
Chris@10:      DK(KP250000000, +0.250000000000000000000000000000000000000000000);
Chris@10:      DK(KP559016994, +0.559016994374947424102293417182819058860154590);
Chris@10:      {
Chris@10: 	  INT i;
Chris@10: 	  for (i = v; i > 0; i = i - 1, ri = ri + ivs, ii = ii + ivs, ro = ro + ovs, io = io + ovs, MAKE_VOLATILE_STRIDE(80, is), MAKE_VOLATILE_STRIDE(80, os)) {
Chris@10: 	       E T7, T2Q, T3h, TD, TP, T1U, T2l, T1d, Tt, TA, TB, T2w, T2z, T2S, T35;
Chris@10: 	       E T36, T3f, TH, TI, TJ, T15, T1a, T1b, T1s, T1x, T1W, T29, T2a, T2j, T1h;
Chris@10: 	       E T1i, T1j, Te, Tl, Tm, T2D, T2G, T2R, T32, T33, T3e, TE, TF, TG, TU;
Chris@10: 	       E TZ, T10, T1D, T1I, T1V, T26, T27, T2i, T1e, T1f, T1g;
Chris@10: 	       {
Chris@10: 		    E T3, T1Q, TN, T2O, T6, TO, T1T, T2P;
Chris@10: 		    {
Chris@10: 			 E T1, T2, TL, TM;
Chris@10: 			 T1 = ri[0];
Chris@10: 			 T2 = ri[WS(is, 10)];
Chris@10: 			 T3 = T1 + T2;
Chris@10: 			 T1Q = T1 - T2;
Chris@10: 			 TL = ii[0];
Chris@10: 			 TM = ii[WS(is, 10)];
Chris@10: 			 TN = TL - TM;
Chris@10: 			 T2O = TL + TM;
Chris@10: 		    }
Chris@10: 		    {
Chris@10: 			 E T4, T5, T1R, T1S;
Chris@10: 			 T4 = ri[WS(is, 5)];
Chris@10: 			 T5 = ri[WS(is, 15)];
Chris@10: 			 T6 = T4 + T5;
Chris@10: 			 TO = T4 - T5;
Chris@10: 			 T1R = ii[WS(is, 5)];
Chris@10: 			 T1S = ii[WS(is, 15)];
Chris@10: 			 T1T = T1R - T1S;
Chris@10: 			 T2P = T1R + T1S;
Chris@10: 		    }
Chris@10: 		    T7 = T3 - T6;
Chris@10: 		    T2Q = T2O - T2P;
Chris@10: 		    T3h = T2O + T2P;
Chris@10: 		    TD = T3 + T6;
Chris@10: 		    TP = TN - TO;
Chris@10: 		    T1U = T1Q - T1T;
Chris@10: 		    T2l = T1Q + T1T;
Chris@10: 		    T1d = TO + TN;
Chris@10: 	       }
Chris@10: 	       {
Chris@10: 		    E Tp, T1o, T13, T2u, Ts, T14, T1r, T2v, Tw, T1t, T18, T2x, Tz, T19, T1w;
Chris@10: 		    E T2y;
Chris@10: 		    {
Chris@10: 			 E Tn, To, T11, T12;
Chris@10: 			 Tn = ri[WS(is, 8)];
Chris@10: 			 To = ri[WS(is, 18)];
Chris@10: 			 Tp = Tn + To;
Chris@10: 			 T1o = Tn - To;
Chris@10: 			 T11 = ii[WS(is, 8)];
Chris@10: 			 T12 = ii[WS(is, 18)];
Chris@10: 			 T13 = T11 - T12;
Chris@10: 			 T2u = T11 + T12;
Chris@10: 		    }
Chris@10: 		    {
Chris@10: 			 E Tq, Tr, T1p, T1q;
Chris@10: 			 Tq = ri[WS(is, 13)];
Chris@10: 			 Tr = ri[WS(is, 3)];
Chris@10: 			 Ts = Tq + Tr;
Chris@10: 			 T14 = Tq - Tr;
Chris@10: 			 T1p = ii[WS(is, 13)];
Chris@10: 			 T1q = ii[WS(is, 3)];
Chris@10: 			 T1r = T1p - T1q;
Chris@10: 			 T2v = T1p + T1q;
Chris@10: 		    }
Chris@10: 		    {
Chris@10: 			 E Tu, Tv, T16, T17;
Chris@10: 			 Tu = ri[WS(is, 12)];
Chris@10: 			 Tv = ri[WS(is, 2)];
Chris@10: 			 Tw = Tu + Tv;
Chris@10: 			 T1t = Tu - Tv;
Chris@10: 			 T16 = ii[WS(is, 12)];
Chris@10: 			 T17 = ii[WS(is, 2)];
Chris@10: 			 T18 = T16 - T17;
Chris@10: 			 T2x = T16 + T17;
Chris@10: 		    }
Chris@10: 		    {
Chris@10: 			 E Tx, Ty, T1u, T1v;
Chris@10: 			 Tx = ri[WS(is, 17)];
Chris@10: 			 Ty = ri[WS(is, 7)];
Chris@10: 			 Tz = Tx + Ty;
Chris@10: 			 T19 = Tx - Ty;
Chris@10: 			 T1u = ii[WS(is, 17)];
Chris@10: 			 T1v = ii[WS(is, 7)];
Chris@10: 			 T1w = T1u - T1v;
Chris@10: 			 T2y = T1u + T1v;
Chris@10: 		    }
Chris@10: 		    Tt = Tp - Ts;
Chris@10: 		    TA = Tw - Tz;
Chris@10: 		    TB = Tt + TA;
Chris@10: 		    T2w = T2u - T2v;
Chris@10: 		    T2z = T2x - T2y;
Chris@10: 		    T2S = T2w + T2z;
Chris@10: 		    T35 = T2u + T2v;
Chris@10: 		    T36 = T2x + T2y;
Chris@10: 		    T3f = T35 + T36;
Chris@10: 		    TH = Tp + Ts;
Chris@10: 		    TI = Tw + Tz;
Chris@10: 		    TJ = TH + TI;
Chris@10: 		    T15 = T13 - T14;
Chris@10: 		    T1a = T18 - T19;
Chris@10: 		    T1b = T15 + T1a;
Chris@10: 		    T1s = T1o - T1r;
Chris@10: 		    T1x = T1t - T1w;
Chris@10: 		    T1W = T1s + T1x;
Chris@10: 		    T29 = T1o + T1r;
Chris@10: 		    T2a = T1t + T1w;
Chris@10: 		    T2j = T29 + T2a;
Chris@10: 		    T1h = T14 + T13;
Chris@10: 		    T1i = T19 + T18;
Chris@10: 		    T1j = T1h + T1i;
Chris@10: 	       }
Chris@10: 	       {
Chris@10: 		    E Ta, T1z, TS, T2B, Td, TT, T1C, T2C, Th, T1E, TX, T2E, Tk, TY, T1H;
Chris@10: 		    E T2F;
Chris@10: 		    {
Chris@10: 			 E T8, T9, TQ, TR;
Chris@10: 			 T8 = ri[WS(is, 4)];
Chris@10: 			 T9 = ri[WS(is, 14)];
Chris@10: 			 Ta = T8 + T9;
Chris@10: 			 T1z = T8 - T9;
Chris@10: 			 TQ = ii[WS(is, 4)];
Chris@10: 			 TR = ii[WS(is, 14)];
Chris@10: 			 TS = TQ - TR;
Chris@10: 			 T2B = TQ + TR;
Chris@10: 		    }
Chris@10: 		    {
Chris@10: 			 E Tb, Tc, T1A, T1B;
Chris@10: 			 Tb = ri[WS(is, 9)];
Chris@10: 			 Tc = ri[WS(is, 19)];
Chris@10: 			 Td = Tb + Tc;
Chris@10: 			 TT = Tb - Tc;
Chris@10: 			 T1A = ii[WS(is, 9)];
Chris@10: 			 T1B = ii[WS(is, 19)];
Chris@10: 			 T1C = T1A - T1B;
Chris@10: 			 T2C = T1A + T1B;
Chris@10: 		    }
Chris@10: 		    {
Chris@10: 			 E Tf, Tg, TV, TW;
Chris@10: 			 Tf = ri[WS(is, 16)];
Chris@10: 			 Tg = ri[WS(is, 6)];
Chris@10: 			 Th = Tf + Tg;
Chris@10: 			 T1E = Tf - Tg;
Chris@10: 			 TV = ii[WS(is, 16)];
Chris@10: 			 TW = ii[WS(is, 6)];
Chris@10: 			 TX = TV - TW;
Chris@10: 			 T2E = TV + TW;
Chris@10: 		    }
Chris@10: 		    {
Chris@10: 			 E Ti, Tj, T1F, T1G;
Chris@10: 			 Ti = ri[WS(is, 1)];
Chris@10: 			 Tj = ri[WS(is, 11)];
Chris@10: 			 Tk = Ti + Tj;
Chris@10: 			 TY = Ti - Tj;
Chris@10: 			 T1F = ii[WS(is, 1)];
Chris@10: 			 T1G = ii[WS(is, 11)];
Chris@10: 			 T1H = T1F - T1G;
Chris@10: 			 T2F = T1F + T1G;
Chris@10: 		    }
Chris@10: 		    Te = Ta - Td;
Chris@10: 		    Tl = Th - Tk;
Chris@10: 		    Tm = Te + Tl;
Chris@10: 		    T2D = T2B - T2C;
Chris@10: 		    T2G = T2E - T2F;
Chris@10: 		    T2R = T2D + T2G;
Chris@10: 		    T32 = T2B + T2C;
Chris@10: 		    T33 = T2E + T2F;
Chris@10: 		    T3e = T32 + T33;
Chris@10: 		    TE = Ta + Td;
Chris@10: 		    TF = Th + Tk;
Chris@10: 		    TG = TE + TF;
Chris@10: 		    TU = TS - TT;
Chris@10: 		    TZ = TX - TY;
Chris@10: 		    T10 = TU + TZ;
Chris@10: 		    T1D = T1z - T1C;
Chris@10: 		    T1I = T1E - T1H;
Chris@10: 		    T1V = T1D + T1I;
Chris@10: 		    T26 = T1z + T1C;
Chris@10: 		    T27 = T1E + T1H;
Chris@10: 		    T2i = T26 + T27;
Chris@10: 		    T1e = TT + TS;
Chris@10: 		    T1f = TY + TX;
Chris@10: 		    T1g = T1e + T1f;
Chris@10: 	       }
Chris@10: 	       {
Chris@10: 		    E T2s, TC, T2r, T2I, T2K, T2A, T2H, T2J, T2t;
Chris@10: 		    T2s = KP559016994 * (Tm - TB);
Chris@10: 		    TC = Tm + TB;
Chris@10: 		    T2r = FNMS(KP250000000, TC, T7);
Chris@10: 		    T2A = T2w - T2z;
Chris@10: 		    T2H = T2D - T2G;
Chris@10: 		    T2I = FNMS(KP587785252, T2H, KP951056516 * T2A);
Chris@10: 		    T2K = FMA(KP951056516, T2H, KP587785252 * T2A);
Chris@10: 		    ro[WS(os, 10)] = T7 + TC;
Chris@10: 		    T2J = T2s + T2r;
Chris@10: 		    ro[WS(os, 14)] = T2J - T2K;
Chris@10: 		    ro[WS(os, 6)] = T2J + T2K;
Chris@10: 		    T2t = T2r - T2s;
Chris@10: 		    ro[WS(os, 2)] = T2t - T2I;
Chris@10: 		    ro[WS(os, 18)] = T2t + T2I;
Chris@10: 	       }
Chris@10: 	       {
Chris@10: 		    E T2V, T2T, T2U, T2N, T2Y, T2L, T2M, T2X, T2W;
Chris@10: 		    T2V = KP559016994 * (T2R - T2S);
Chris@10: 		    T2T = T2R + T2S;
Chris@10: 		    T2U = FNMS(KP250000000, T2T, T2Q);
Chris@10: 		    T2L = Tt - TA;
Chris@10: 		    T2M = Te - Tl;
Chris@10: 		    T2N = FNMS(KP587785252, T2M, KP951056516 * T2L);
Chris@10: 		    T2Y = FMA(KP951056516, T2M, KP587785252 * T2L);
Chris@10: 		    io[WS(os, 10)] = T2Q + T2T;
Chris@10: 		    T2X = T2V + T2U;
Chris@10: 		    io[WS(os, 6)] = T2X - T2Y;
Chris@10: 		    io[WS(os, 14)] = T2Y + T2X;
Chris@10: 		    T2W = T2U - T2V;
Chris@10: 		    io[WS(os, 2)] = T2N + T2W;
Chris@10: 		    io[WS(os, 18)] = T2W - T2N;
Chris@10: 	       }
Chris@10: 	       {
Chris@10: 		    E T2Z, TK, T30, T38, T3a, T34, T37, T39, T31;
Chris@10: 		    T2Z = KP559016994 * (TG - TJ);
Chris@10: 		    TK = TG + TJ;
Chris@10: 		    T30 = FNMS(KP250000000, TK, TD);
Chris@10: 		    T34 = T32 - T33;
Chris@10: 		    T37 = T35 - T36;
Chris@10: 		    T38 = FMA(KP951056516, T34, KP587785252 * T37);
Chris@10: 		    T3a = FNMS(KP587785252, T34, KP951056516 * T37);
Chris@10: 		    ro[0] = TD + TK;
Chris@10: 		    T39 = T30 - T2Z;
Chris@10: 		    ro[WS(os, 12)] = T39 - T3a;
Chris@10: 		    ro[WS(os, 8)] = T39 + T3a;
Chris@10: 		    T31 = T2Z + T30;
Chris@10: 		    ro[WS(os, 4)] = T31 - T38;
Chris@10: 		    ro[WS(os, 16)] = T31 + T38;
Chris@10: 	       }
Chris@10: 	       {
Chris@10: 		    E T3g, T3i, T3j, T3d, T3m, T3b, T3c, T3l, T3k;
Chris@10: 		    T3g = KP559016994 * (T3e - T3f);
Chris@10: 		    T3i = T3e + T3f;
Chris@10: 		    T3j = FNMS(KP250000000, T3i, T3h);
Chris@10: 		    T3b = TE - TF;
Chris@10: 		    T3c = TH - TI;
Chris@10: 		    T3d = FMA(KP951056516, T3b, KP587785252 * T3c);
Chris@10: 		    T3m = FNMS(KP587785252, T3b, KP951056516 * T3c);
Chris@10: 		    io[0] = T3h + T3i;
Chris@10: 		    T3l = T3j - T3g;
Chris@10: 		    io[WS(os, 8)] = T3l - T3m;
Chris@10: 		    io[WS(os, 12)] = T3m + T3l;
Chris@10: 		    T3k = T3g + T3j;
Chris@10: 		    io[WS(os, 4)] = T3d + T3k;
Chris@10: 		    io[WS(os, 16)] = T3k - T3d;
Chris@10: 	       }
Chris@10: 	       {
Chris@10: 		    E T23, T1c, T24, T2c, T2e, T28, T2b, T2d, T25;
Chris@10: 		    T23 = KP559016994 * (T10 - T1b);
Chris@10: 		    T1c = T10 + T1b;
Chris@10: 		    T24 = FNMS(KP250000000, T1c, TP);
Chris@10: 		    T28 = T26 - T27;
Chris@10: 		    T2b = T29 - T2a;
Chris@10: 		    T2c = FMA(KP951056516, T28, KP587785252 * T2b);
Chris@10: 		    T2e = FNMS(KP587785252, T28, KP951056516 * T2b);
Chris@10: 		    io[WS(os, 5)] = TP + T1c;
Chris@10: 		    T2d = T24 - T23;
Chris@10: 		    io[WS(os, 13)] = T2d - T2e;
Chris@10: 		    io[WS(os, 17)] = T2d + T2e;
Chris@10: 		    T25 = T23 + T24;
Chris@10: 		    io[WS(os, 1)] = T25 - T2c;
Chris@10: 		    io[WS(os, 9)] = T25 + T2c;
Chris@10: 	       }
Chris@10: 	       {
Chris@10: 		    E T2k, T2m, T2n, T2h, T2p, T2f, T2g, T2q, T2o;
Chris@10: 		    T2k = KP559016994 * (T2i - T2j);
Chris@10: 		    T2m = T2i + T2j;
Chris@10: 		    T2n = FNMS(KP250000000, T2m, T2l);
Chris@10: 		    T2f = TU - TZ;
Chris@10: 		    T2g = T15 - T1a;
Chris@10: 		    T2h = FMA(KP951056516, T2f, KP587785252 * T2g);
Chris@10: 		    T2p = FNMS(KP587785252, T2f, KP951056516 * T2g);
Chris@10: 		    ro[WS(os, 5)] = T2l + T2m;
Chris@10: 		    T2q = T2n - T2k;
Chris@10: 		    ro[WS(os, 13)] = T2p + T2q;
Chris@10: 		    ro[WS(os, 17)] = T2q - T2p;
Chris@10: 		    T2o = T2k + T2n;
Chris@10: 		    ro[WS(os, 1)] = T2h + T2o;
Chris@10: 		    ro[WS(os, 9)] = T2o - T2h;
Chris@10: 	       }
Chris@10: 	       {
Chris@10: 		    E T1m, T1k, T1l, T1K, T1M, T1y, T1J, T1L, T1n;
Chris@10: 		    T1m = KP559016994 * (T1g - T1j);
Chris@10: 		    T1k = T1g + T1j;
Chris@10: 		    T1l = FNMS(KP250000000, T1k, T1d);
Chris@10: 		    T1y = T1s - T1x;
Chris@10: 		    T1J = T1D - T1I;
Chris@10: 		    T1K = FNMS(KP587785252, T1J, KP951056516 * T1y);
Chris@10: 		    T1M = FMA(KP951056516, T1J, KP587785252 * T1y);
Chris@10: 		    io[WS(os, 15)] = T1d + T1k;
Chris@10: 		    T1L = T1m + T1l;
Chris@10: 		    io[WS(os, 11)] = T1L - T1M;
Chris@10: 		    io[WS(os, 19)] = T1L + T1M;
Chris@10: 		    T1n = T1l - T1m;
Chris@10: 		    io[WS(os, 3)] = T1n - T1K;
Chris@10: 		    io[WS(os, 7)] = T1n + T1K;
Chris@10: 	       }
Chris@10: 	       {
Chris@10: 		    E T1Z, T1X, T1Y, T1P, T21, T1N, T1O, T22, T20;
Chris@10: 		    T1Z = KP559016994 * (T1V - T1W);
Chris@10: 		    T1X = T1V + T1W;
Chris@10: 		    T1Y = FNMS(KP250000000, T1X, T1U);
Chris@10: 		    T1N = T1h - T1i;
Chris@10: 		    T1O = T1e - T1f;
Chris@10: 		    T1P = FNMS(KP587785252, T1O, KP951056516 * T1N);
Chris@10: 		    T21 = FMA(KP951056516, T1O, KP587785252 * T1N);
Chris@10: 		    ro[WS(os, 15)] = T1U + T1X;
Chris@10: 		    T22 = T1Z + T1Y;
Chris@10: 		    ro[WS(os, 11)] = T21 + T22;
Chris@10: 		    ro[WS(os, 19)] = T22 - T21;
Chris@10: 		    T20 = T1Y - T1Z;
Chris@10: 		    ro[WS(os, 3)] = T1P + T20;
Chris@10: 		    ro[WS(os, 7)] = T20 - T1P;
Chris@10: 	       }
Chris@10: 	  }
Chris@10:      }
Chris@10: }
Chris@10: 
Chris@10: static const kdft_desc desc = { 20, "n1_20", {184, 24, 24, 0}, &GENUS, 0, 0, 0, 0 };
Chris@10: 
Chris@10: void X(codelet_n1_20) (planner *p) {
Chris@10:      X(kdft_register) (p, n1_20, &desc);
Chris@10: }
Chris@10: 
Chris@10: #endif				/* HAVE_FMA */