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 07:39:47 EST 2012 */
Chris@10: 
Chris@10: #include "codelet-rdft.h"
Chris@10: 
Chris@10: #ifdef HAVE_FMA
Chris@10: 
Chris@10: /* Generated by: ../../../genfft/gen_r2cf.native -fma -reorder-insns -schedule-for-pipeline -compact -variables 4 -pipeline-latency 4 -n 32 -name r2cf_32 -include r2cf.h */
Chris@10: 
Chris@10: /*
Chris@10:  * This function contains 156 FP additions, 68 FP multiplications,
Chris@10:  * (or, 88 additions, 0 multiplications, 68 fused multiply/add),
Chris@10:  * 89 stack variables, 7 constants, and 64 memory accesses
Chris@10:  */
Chris@10: #include "r2cf.h"
Chris@10: 
Chris@10: static void r2cf_32(R *R0, R *R1, R *Cr, R *Ci, stride rs, stride csr, stride csi, INT v, INT ivs, INT ovs)
Chris@10: {
Chris@10:      DK(KP831469612, +0.831469612302545237078788377617905756738560812);
Chris@10:      DK(KP668178637, +0.668178637919298919997757686523080761552472251);
Chris@10:      DK(KP980785280, +0.980785280403230449126182236134239036973933731);
Chris@10:      DK(KP198912367, +0.198912367379658006911597622644676228597850501);
Chris@10:      DK(KP923879532, +0.923879532511286756128183189396788286822416626);
Chris@10:      DK(KP707106781, +0.707106781186547524400844362104849039284835938);
Chris@10:      DK(KP414213562, +0.414213562373095048801688724209698078569671875);
Chris@10:      {
Chris@10: 	  INT i;
Chris@10: 	  for (i = v; i > 0; i = i - 1, R0 = R0 + ivs, R1 = R1 + ivs, Cr = Cr + ovs, Ci = Ci + ovs, MAKE_VOLATILE_STRIDE(128, rs), MAKE_VOLATILE_STRIDE(128, csr), MAKE_VOLATILE_STRIDE(128, csi)) {
Chris@10: 	       E T1x, T1M, T1I, T1E, T1J, T1H;
Chris@10: 	       {
Chris@10: 		    E Tv, T1h, T7, T2b, Te, T2n, Ty, T1i, T1l, TF, T2d, Tt, T1k, TC, T2c;
Chris@10: 		    E Tm, T2j, T1Z, T2k, T22, TK, T1B, T19, T1C, T1e, TO, TV, T1T, TN, TP;
Chris@10: 		    E T2g, T1S;
Chris@10: 		    {
Chris@10: 			 E TD, Tp, Tq, Tr;
Chris@10: 			 {
Chris@10: 			      E T1, T2, T4, T5;
Chris@10: 			      T1 = R0[0];
Chris@10: 			      T2 = R0[WS(rs, 8)];
Chris@10: 			      T4 = R0[WS(rs, 4)];
Chris@10: 			      T5 = R0[WS(rs, 12)];
Chris@10: 			      {
Chris@10: 				   E Ta, Tw, Tx, Td, Tn, To;
Chris@10: 				   {
Chris@10: 					E T8, T3, T6, T9, Tb, Tc;
Chris@10: 					T8 = R0[WS(rs, 2)];
Chris@10: 					Tv = T1 - T2;
Chris@10: 					T3 = T1 + T2;
Chris@10: 					T1h = T4 - T5;
Chris@10: 					T6 = T4 + T5;
Chris@10: 					T9 = R0[WS(rs, 10)];
Chris@10: 					Tb = R0[WS(rs, 14)];
Chris@10: 					Tc = R0[WS(rs, 6)];
Chris@10: 					T7 = T3 + T6;
Chris@10: 					T2b = T3 - T6;
Chris@10: 					Ta = T8 + T9;
Chris@10: 					Tw = T8 - T9;
Chris@10: 					Tx = Tb - Tc;
Chris@10: 					Td = Tb + Tc;
Chris@10: 				   }
Chris@10: 				   Tn = R0[WS(rs, 15)];
Chris@10: 				   To = R0[WS(rs, 7)];
Chris@10: 				   Te = Ta + Td;
Chris@10: 				   T2n = Td - Ta;
Chris@10: 				   Ty = Tw + Tx;
Chris@10: 				   T1i = Tx - Tw;
Chris@10: 				   TD = Tn - To;
Chris@10: 				   Tp = Tn + To;
Chris@10: 				   Tq = R0[WS(rs, 3)];
Chris@10: 				   Tr = R0[WS(rs, 11)];
Chris@10: 			      }
Chris@10: 			 }
Chris@10: 			 {
Chris@10: 			      E Tj, TA, Ti, Tk;
Chris@10: 			      {
Chris@10: 				   E Tg, Th, TE, Ts;
Chris@10: 				   Tg = R0[WS(rs, 1)];
Chris@10: 				   Th = R0[WS(rs, 9)];
Chris@10: 				   Tj = R0[WS(rs, 5)];
Chris@10: 				   TE = Tq - Tr;
Chris@10: 				   Ts = Tq + Tr;
Chris@10: 				   TA = Tg - Th;
Chris@10: 				   Ti = Tg + Th;
Chris@10: 				   T1l = FNMS(KP414213562, TD, TE);
Chris@10: 				   TF = FMA(KP414213562, TE, TD);
Chris@10: 				   T2d = Tp - Ts;
Chris@10: 				   Tt = Tp + Ts;
Chris@10: 				   Tk = R0[WS(rs, 13)];
Chris@10: 			      }
Chris@10: 			      {
Chris@10: 				   E T11, T15, T1c, T20, T14, T16, T1X, T1Y, T1Q, T1R;
Chris@10: 				   {
Chris@10: 					E T1a, T1b, T12, T13;
Chris@10: 					{
Chris@10: 					     E TZ, T10, TB, Tl;
Chris@10: 					     TZ = R1[WS(rs, 15)];
Chris@10: 					     T10 = R1[WS(rs, 7)];
Chris@10: 					     T1a = R1[WS(rs, 11)];
Chris@10: 					     TB = Tj - Tk;
Chris@10: 					     Tl = Tj + Tk;
Chris@10: 					     T1X = TZ + T10;
Chris@10: 					     T11 = TZ - T10;
Chris@10: 					     T1k = FMA(KP414213562, TA, TB);
Chris@10: 					     TC = FNMS(KP414213562, TB, TA);
Chris@10: 					     T2c = Ti - Tl;
Chris@10: 					     Tm = Ti + Tl;
Chris@10: 					     T1b = R1[WS(rs, 3)];
Chris@10: 					}
Chris@10: 					T12 = R1[WS(rs, 1)];
Chris@10: 					T13 = R1[WS(rs, 9)];
Chris@10: 					T15 = R1[WS(rs, 13)];
Chris@10: 					T1Y = T1b + T1a;
Chris@10: 					T1c = T1a - T1b;
Chris@10: 					T20 = T12 + T13;
Chris@10: 					T14 = T12 - T13;
Chris@10: 					T16 = R1[WS(rs, 5)];
Chris@10: 				   }
Chris@10: 				   T2j = T1X - T1Y;
Chris@10: 				   T1Z = T1X + T1Y;
Chris@10: 				   {
Chris@10: 					E TT, TU, TL, TM;
Chris@10: 					{
Chris@10: 					     E TI, T21, T17, TJ, T18, T1d;
Chris@10: 					     TI = R1[0];
Chris@10: 					     T21 = T15 + T16;
Chris@10: 					     T17 = T15 - T16;
Chris@10: 					     TJ = R1[WS(rs, 8)];
Chris@10: 					     TT = R1[WS(rs, 4)];
Chris@10: 					     T2k = T21 - T20;
Chris@10: 					     T22 = T20 + T21;
Chris@10: 					     T18 = T14 + T17;
Chris@10: 					     T1d = T17 - T14;
Chris@10: 					     T1Q = TI + TJ;
Chris@10: 					     TK = TI - TJ;
Chris@10: 					     T1B = FNMS(KP707106781, T18, T11);
Chris@10: 					     T19 = FMA(KP707106781, T18, T11);
Chris@10: 					     T1C = FNMS(KP707106781, T1d, T1c);
Chris@10: 					     T1e = FMA(KP707106781, T1d, T1c);
Chris@10: 					     TU = R1[WS(rs, 12)];
Chris@10: 					}
Chris@10: 					TL = R1[WS(rs, 2)];
Chris@10: 					TM = R1[WS(rs, 10)];
Chris@10: 					TO = R1[WS(rs, 14)];
Chris@10: 					T1R = TT + TU;
Chris@10: 					TV = TT - TU;
Chris@10: 					T1T = TL + TM;
Chris@10: 					TN = TL - TM;
Chris@10: 					TP = R1[WS(rs, 6)];
Chris@10: 				   }
Chris@10: 				   T2g = T1Q - T1R;
Chris@10: 				   T1S = T1Q + T1R;
Chris@10: 			      }
Chris@10: 			 }
Chris@10: 		    }
Chris@10: 		    {
Chris@10: 			 E T1P, T25, T23, T2h, T1W, T1y, TS, T1z, TX, T27, T2a;
Chris@10: 			 {
Chris@10: 			      E Tf, Tu, T29, T28;
Chris@10: 			      {
Chris@10: 				   E T1U, TQ, T1V, TR, TW;
Chris@10: 				   T1P = T7 - Te;
Chris@10: 				   Tf = T7 + Te;
Chris@10: 				   T1U = TO + TP;
Chris@10: 				   TQ = TO - TP;
Chris@10: 				   Tu = Tm + Tt;
Chris@10: 				   T25 = Tt - Tm;
Chris@10: 				   T23 = T1Z - T22;
Chris@10: 				   T29 = T1Z + T22;
Chris@10: 				   T2h = T1U - T1T;
Chris@10: 				   T1V = T1T + T1U;
Chris@10: 				   TR = TN + TQ;
Chris@10: 				   TW = TN - TQ;
Chris@10: 				   T27 = Tf + Tu;
Chris@10: 				   T1W = T1S - T1V;
Chris@10: 				   T28 = T1S + T1V;
Chris@10: 				   T1y = FNMS(KP707106781, TR, TK);
Chris@10: 				   TS = FMA(KP707106781, TR, TK);
Chris@10: 				   T1z = FNMS(KP707106781, TW, TV);
Chris@10: 				   TX = FMA(KP707106781, TW, TV);
Chris@10: 				   T2a = T28 + T29;
Chris@10: 			      }
Chris@10: 			      Cr[WS(csr, 8)] = Tf - Tu;
Chris@10: 			      Ci[WS(csi, 8)] = T29 - T28;
Chris@10: 			 }
Chris@10: 			 Cr[0] = T27 + T2a;
Chris@10: 			 Cr[WS(csr, 16)] = T27 - T2a;
Chris@10: 			 {
Chris@10: 			      E T2s, T2i, T2v, T2f, T2r, T2p, T2l, T2t;
Chris@10: 			      {
Chris@10: 				   E T2o, T2e, T26, T24;
Chris@10: 				   T2o = T2d - T2c;
Chris@10: 				   T2e = T2c + T2d;
Chris@10: 				   T2s = FNMS(KP414213562, T2g, T2h);
Chris@10: 				   T2i = FMA(KP414213562, T2h, T2g);
Chris@10: 				   T26 = T23 - T1W;
Chris@10: 				   T24 = T1W + T23;
Chris@10: 				   T2v = FNMS(KP707106781, T2e, T2b);
Chris@10: 				   T2f = FMA(KP707106781, T2e, T2b);
Chris@10: 				   T2r = FMA(KP707106781, T2o, T2n);
Chris@10: 				   T2p = FNMS(KP707106781, T2o, T2n);
Chris@10: 				   Ci[WS(csi, 4)] = FMA(KP707106781, T26, T25);
Chris@10: 				   Ci[WS(csi, 12)] = FMS(KP707106781, T26, T25);
Chris@10: 				   Cr[WS(csr, 4)] = FMA(KP707106781, T24, T1P);
Chris@10: 				   Cr[WS(csr, 12)] = FNMS(KP707106781, T24, T1P);
Chris@10: 				   T2l = FNMS(KP414213562, T2k, T2j);
Chris@10: 				   T2t = FMA(KP414213562, T2j, T2k);
Chris@10: 			      }
Chris@10: 			      {
Chris@10: 				   E T1v, T1G, TH, T1s, T1F, T1w, T1o, T1g, T1p, T1n;
Chris@10: 				   {
Chris@10: 					E T1f, TY, T1t, T1u, T1j, T1m;
Chris@10: 					{
Chris@10: 					     E Tz, TG, T1q, T1r;
Chris@10: 					     T1v = FNMS(KP707106781, Ty, Tv);
Chris@10: 					     Tz = FMA(KP707106781, Ty, Tv);
Chris@10: 					     {
Chris@10: 						  E T2q, T2m, T2w, T2u;
Chris@10: 						  T2q = T2l - T2i;
Chris@10: 						  T2m = T2i + T2l;
Chris@10: 						  T2w = T2t - T2s;
Chris@10: 						  T2u = T2s + T2t;
Chris@10: 						  Ci[WS(csi, 10)] = FMA(KP923879532, T2q, T2p);
Chris@10: 						  Ci[WS(csi, 6)] = FMS(KP923879532, T2q, T2p);
Chris@10: 						  Cr[WS(csr, 2)] = FMA(KP923879532, T2m, T2f);
Chris@10: 						  Cr[WS(csr, 14)] = FNMS(KP923879532, T2m, T2f);
Chris@10: 						  Cr[WS(csr, 10)] = FNMS(KP923879532, T2w, T2v);
Chris@10: 						  Cr[WS(csr, 6)] = FMA(KP923879532, T2w, T2v);
Chris@10: 						  Ci[WS(csi, 2)] = FMA(KP923879532, T2u, T2r);
Chris@10: 						  Ci[WS(csi, 14)] = FMS(KP923879532, T2u, T2r);
Chris@10: 						  TG = TC + TF;
Chris@10: 						  T1G = TF - TC;
Chris@10: 					     }
Chris@10: 					     T1f = FNMS(KP198912367, T1e, T19);
Chris@10: 					     T1q = FMA(KP198912367, T19, T1e);
Chris@10: 					     T1r = FMA(KP198912367, TS, TX);
Chris@10: 					     TY = FNMS(KP198912367, TX, TS);
Chris@10: 					     T1t = FNMS(KP923879532, TG, Tz);
Chris@10: 					     TH = FMA(KP923879532, TG, Tz);
Chris@10: 					     T1u = T1r + T1q;
Chris@10: 					     T1s = T1q - T1r;
Chris@10: 					     T1F = FMA(KP707106781, T1i, T1h);
Chris@10: 					     T1j = FNMS(KP707106781, T1i, T1h);
Chris@10: 					     T1m = T1k + T1l;
Chris@10: 					     T1w = T1k - T1l;
Chris@10: 					}
Chris@10: 					Cr[WS(csr, 7)] = FMA(KP980785280, T1u, T1t);
Chris@10: 					T1o = T1f - TY;
Chris@10: 					T1g = TY + T1f;
Chris@10: 					T1p = FMA(KP923879532, T1m, T1j);
Chris@10: 					T1n = FNMS(KP923879532, T1m, T1j);
Chris@10: 					Cr[WS(csr, 9)] = FNMS(KP980785280, T1u, T1t);
Chris@10: 				   }
Chris@10: 				   Cr[WS(csr, 1)] = FMA(KP980785280, T1g, TH);
Chris@10: 				   Cr[WS(csr, 15)] = FNMS(KP980785280, T1g, TH);
Chris@10: 				   Ci[WS(csi, 1)] = FMS(KP980785280, T1s, T1p);
Chris@10: 				   Ci[WS(csi, 15)] = FMA(KP980785280, T1s, T1p);
Chris@10: 				   Ci[WS(csi, 9)] = FMS(KP980785280, T1o, T1n);
Chris@10: 				   Ci[WS(csi, 7)] = FMA(KP980785280, T1o, T1n);
Chris@10: 				   {
Chris@10: 					E T1A, T1D, T1N, T1O, T1K, T1L;
Chris@10: 					T1A = FMA(KP668178637, T1z, T1y);
Chris@10: 					T1K = FNMS(KP668178637, T1y, T1z);
Chris@10: 					T1L = FNMS(KP668178637, T1B, T1C);
Chris@10: 					T1D = FMA(KP668178637, T1C, T1B);
Chris@10: 					T1N = FNMS(KP923879532, T1w, T1v);
Chris@10: 					T1x = FMA(KP923879532, T1w, T1v);
Chris@10: 					T1O = T1K + T1L;
Chris@10: 					T1M = T1K - T1L;
Chris@10: 					Cr[WS(csr, 5)] = FNMS(KP831469612, T1O, T1N);
Chris@10: 					T1I = T1D - T1A;
Chris@10: 					T1E = T1A + T1D;
Chris@10: 					T1J = FMA(KP923879532, T1G, T1F);
Chris@10: 					T1H = FNMS(KP923879532, T1G, T1F);
Chris@10: 					Cr[WS(csr, 11)] = FMA(KP831469612, T1O, T1N);
Chris@10: 				   }
Chris@10: 			      }
Chris@10: 			 }
Chris@10: 		    }
Chris@10: 	       }
Chris@10: 	       Ci[WS(csi, 3)] = FMA(KP831469612, T1M, T1J);
Chris@10: 	       Cr[WS(csr, 3)] = FMA(KP831469612, T1E, T1x);
Chris@10: 	       Ci[WS(csi, 13)] = FMS(KP831469612, T1M, T1J);
Chris@10: 	       Cr[WS(csr, 13)] = FNMS(KP831469612, T1E, T1x);
Chris@10: 	       Ci[WS(csi, 11)] = FMA(KP831469612, T1I, T1H);
Chris@10: 	       Ci[WS(csi, 5)] = FMS(KP831469612, T1I, T1H);
Chris@10: 	  }
Chris@10:      }
Chris@10: }
Chris@10: 
Chris@10: static const kr2c_desc desc = { 32, "r2cf_32", {88, 0, 68, 0}, &GENUS };
Chris@10: 
Chris@10: void X(codelet_r2cf_32) (planner *p) {
Chris@10:      X(kr2c_register) (p, r2cf_32, &desc);
Chris@10: }
Chris@10: 
Chris@10: #else				/* HAVE_FMA */
Chris@10: 
Chris@10: /* Generated by: ../../../genfft/gen_r2cf.native -compact -variables 4 -pipeline-latency 4 -n 32 -name r2cf_32 -include r2cf.h */
Chris@10: 
Chris@10: /*
Chris@10:  * This function contains 156 FP additions, 42 FP multiplications,
Chris@10:  * (or, 140 additions, 26 multiplications, 16 fused multiply/add),
Chris@10:  * 54 stack variables, 7 constants, and 64 memory accesses
Chris@10:  */
Chris@10: #include "r2cf.h"
Chris@10: 
Chris@10: static void r2cf_32(R *R0, R *R1, R *Cr, R *Ci, stride rs, stride csr, stride csi, INT v, INT ivs, INT ovs)
Chris@10: {
Chris@10:      DK(KP555570233, +0.555570233019602224742830813948532874374937191);
Chris@10:      DK(KP831469612, +0.831469612302545237078788377617905756738560812);
Chris@10:      DK(KP195090322, +0.195090322016128267848284868477022240927691618);
Chris@10:      DK(KP980785280, +0.980785280403230449126182236134239036973933731);
Chris@10:      DK(KP382683432, +0.382683432365089771728459984030398866761344562);
Chris@10:      DK(KP923879532, +0.923879532511286756128183189396788286822416626);
Chris@10:      DK(KP707106781, +0.707106781186547524400844362104849039284835938);
Chris@10:      {
Chris@10: 	  INT i;
Chris@10: 	  for (i = v; i > 0; i = i - 1, R0 = R0 + ivs, R1 = R1 + ivs, Cr = Cr + ovs, Ci = Ci + ovs, MAKE_VOLATILE_STRIDE(128, rs), MAKE_VOLATILE_STRIDE(128, csr), MAKE_VOLATILE_STRIDE(128, csi)) {
Chris@10: 	       E T7, T2b, Tv, T1l, Te, T2o, Ty, T1k, Tt, T2d, TF, T1h, Tm, T2c, TC;
Chris@10: 	       E T1i, T1Z, T22, T2k, T2j, T1e, T1C, T19, T1B, T1S, T1V, T2h, T2g, TX, T1z;
Chris@10: 	       E TS, T1y;
Chris@10: 	       {
Chris@10: 		    E T1, T2, T3, T4, T5, T6;
Chris@10: 		    T1 = R0[0];
Chris@10: 		    T2 = R0[WS(rs, 8)];
Chris@10: 		    T3 = T1 + T2;
Chris@10: 		    T4 = R0[WS(rs, 4)];
Chris@10: 		    T5 = R0[WS(rs, 12)];
Chris@10: 		    T6 = T4 + T5;
Chris@10: 		    T7 = T3 + T6;
Chris@10: 		    T2b = T3 - T6;
Chris@10: 		    Tv = T1 - T2;
Chris@10: 		    T1l = T4 - T5;
Chris@10: 	       }
Chris@10: 	       {
Chris@10: 		    E Ta, Tw, Td, Tx;
Chris@10: 		    {
Chris@10: 			 E T8, T9, Tb, Tc;
Chris@10: 			 T8 = R0[WS(rs, 2)];
Chris@10: 			 T9 = R0[WS(rs, 10)];
Chris@10: 			 Ta = T8 + T9;
Chris@10: 			 Tw = T8 - T9;
Chris@10: 			 Tb = R0[WS(rs, 14)];
Chris@10: 			 Tc = R0[WS(rs, 6)];
Chris@10: 			 Td = Tb + Tc;
Chris@10: 			 Tx = Tb - Tc;
Chris@10: 		    }
Chris@10: 		    Te = Ta + Td;
Chris@10: 		    T2o = Td - Ta;
Chris@10: 		    Ty = KP707106781 * (Tw + Tx);
Chris@10: 		    T1k = KP707106781 * (Tx - Tw);
Chris@10: 	       }
Chris@10: 	       {
Chris@10: 		    E Tp, TD, Ts, TE;
Chris@10: 		    {
Chris@10: 			 E Tn, To, Tq, Tr;
Chris@10: 			 Tn = R0[WS(rs, 15)];
Chris@10: 			 To = R0[WS(rs, 7)];
Chris@10: 			 Tp = Tn + To;
Chris@10: 			 TD = Tn - To;
Chris@10: 			 Tq = R0[WS(rs, 3)];
Chris@10: 			 Tr = R0[WS(rs, 11)];
Chris@10: 			 Ts = Tq + Tr;
Chris@10: 			 TE = Tq - Tr;
Chris@10: 		    }
Chris@10: 		    Tt = Tp + Ts;
Chris@10: 		    T2d = Tp - Ts;
Chris@10: 		    TF = FMA(KP923879532, TD, KP382683432 * TE);
Chris@10: 		    T1h = FNMS(KP923879532, TE, KP382683432 * TD);
Chris@10: 	       }
Chris@10: 	       {
Chris@10: 		    E Ti, TA, Tl, TB;
Chris@10: 		    {
Chris@10: 			 E Tg, Th, Tj, Tk;
Chris@10: 			 Tg = R0[WS(rs, 1)];
Chris@10: 			 Th = R0[WS(rs, 9)];
Chris@10: 			 Ti = Tg + Th;
Chris@10: 			 TA = Tg - Th;
Chris@10: 			 Tj = R0[WS(rs, 5)];
Chris@10: 			 Tk = R0[WS(rs, 13)];
Chris@10: 			 Tl = Tj + Tk;
Chris@10: 			 TB = Tj - Tk;
Chris@10: 		    }
Chris@10: 		    Tm = Ti + Tl;
Chris@10: 		    T2c = Ti - Tl;
Chris@10: 		    TC = FNMS(KP382683432, TB, KP923879532 * TA);
Chris@10: 		    T1i = FMA(KP382683432, TA, KP923879532 * TB);
Chris@10: 	       }
Chris@10: 	       {
Chris@10: 		    E T11, T1X, T1d, T1Y, T14, T20, T17, T21, T1a, T18;
Chris@10: 		    {
Chris@10: 			 E TZ, T10, T1b, T1c;
Chris@10: 			 TZ = R1[WS(rs, 15)];
Chris@10: 			 T10 = R1[WS(rs, 7)];
Chris@10: 			 T11 = TZ - T10;
Chris@10: 			 T1X = TZ + T10;
Chris@10: 			 T1b = R1[WS(rs, 3)];
Chris@10: 			 T1c = R1[WS(rs, 11)];
Chris@10: 			 T1d = T1b - T1c;
Chris@10: 			 T1Y = T1b + T1c;
Chris@10: 		    }
Chris@10: 		    {
Chris@10: 			 E T12, T13, T15, T16;
Chris@10: 			 T12 = R1[WS(rs, 1)];
Chris@10: 			 T13 = R1[WS(rs, 9)];
Chris@10: 			 T14 = T12 - T13;
Chris@10: 			 T20 = T12 + T13;
Chris@10: 			 T15 = R1[WS(rs, 13)];
Chris@10: 			 T16 = R1[WS(rs, 5)];
Chris@10: 			 T17 = T15 - T16;
Chris@10: 			 T21 = T15 + T16;
Chris@10: 		    }
Chris@10: 		    T1Z = T1X + T1Y;
Chris@10: 		    T22 = T20 + T21;
Chris@10: 		    T2k = T21 - T20;
Chris@10: 		    T2j = T1X - T1Y;
Chris@10: 		    T1a = KP707106781 * (T17 - T14);
Chris@10: 		    T1e = T1a - T1d;
Chris@10: 		    T1C = T1d + T1a;
Chris@10: 		    T18 = KP707106781 * (T14 + T17);
Chris@10: 		    T19 = T11 + T18;
Chris@10: 		    T1B = T11 - T18;
Chris@10: 	       }
Chris@10: 	       {
Chris@10: 		    E TK, T1Q, TW, T1R, TN, T1T, TQ, T1U, TT, TR;
Chris@10: 		    {
Chris@10: 			 E TI, TJ, TU, TV;
Chris@10: 			 TI = R1[0];
Chris@10: 			 TJ = R1[WS(rs, 8)];
Chris@10: 			 TK = TI - TJ;
Chris@10: 			 T1Q = TI + TJ;
Chris@10: 			 TU = R1[WS(rs, 4)];
Chris@10: 			 TV = R1[WS(rs, 12)];
Chris@10: 			 TW = TU - TV;
Chris@10: 			 T1R = TU + TV;
Chris@10: 		    }
Chris@10: 		    {
Chris@10: 			 E TL, TM, TO, TP;
Chris@10: 			 TL = R1[WS(rs, 2)];
Chris@10: 			 TM = R1[WS(rs, 10)];
Chris@10: 			 TN = TL - TM;
Chris@10: 			 T1T = TL + TM;
Chris@10: 			 TO = R1[WS(rs, 14)];
Chris@10: 			 TP = R1[WS(rs, 6)];
Chris@10: 			 TQ = TO - TP;
Chris@10: 			 T1U = TO + TP;
Chris@10: 		    }
Chris@10: 		    T1S = T1Q + T1R;
Chris@10: 		    T1V = T1T + T1U;
Chris@10: 		    T2h = T1U - T1T;
Chris@10: 		    T2g = T1Q - T1R;
Chris@10: 		    TT = KP707106781 * (TQ - TN);
Chris@10: 		    TX = TT - TW;
Chris@10: 		    T1z = TW + TT;
Chris@10: 		    TR = KP707106781 * (TN + TQ);
Chris@10: 		    TS = TK + TR;
Chris@10: 		    T1y = TK - TR;
Chris@10: 	       }
Chris@10: 	       {
Chris@10: 		    E Tf, Tu, T27, T28, T29, T2a;
Chris@10: 		    Tf = T7 + Te;
Chris@10: 		    Tu = Tm + Tt;
Chris@10: 		    T27 = Tf + Tu;
Chris@10: 		    T28 = T1S + T1V;
Chris@10: 		    T29 = T1Z + T22;
Chris@10: 		    T2a = T28 + T29;
Chris@10: 		    Cr[WS(csr, 8)] = Tf - Tu;
Chris@10: 		    Ci[WS(csi, 8)] = T29 - T28;
Chris@10: 		    Cr[WS(csr, 16)] = T27 - T2a;
Chris@10: 		    Cr[0] = T27 + T2a;
Chris@10: 	       }
Chris@10: 	       {
Chris@10: 		    E T1P, T25, T24, T26, T1W, T23;
Chris@10: 		    T1P = T7 - Te;
Chris@10: 		    T25 = Tt - Tm;
Chris@10: 		    T1W = T1S - T1V;
Chris@10: 		    T23 = T1Z - T22;
Chris@10: 		    T24 = KP707106781 * (T1W + T23);
Chris@10: 		    T26 = KP707106781 * (T23 - T1W);
Chris@10: 		    Cr[WS(csr, 12)] = T1P - T24;
Chris@10: 		    Ci[WS(csi, 12)] = T26 - T25;
Chris@10: 		    Cr[WS(csr, 4)] = T1P + T24;
Chris@10: 		    Ci[WS(csi, 4)] = T25 + T26;
Chris@10: 	       }
Chris@10: 	       {
Chris@10: 		    E T2f, T2v, T2p, T2r, T2m, T2q, T2u, T2w, T2e, T2n;
Chris@10: 		    T2e = KP707106781 * (T2c + T2d);
Chris@10: 		    T2f = T2b + T2e;
Chris@10: 		    T2v = T2b - T2e;
Chris@10: 		    T2n = KP707106781 * (T2d - T2c);
Chris@10: 		    T2p = T2n - T2o;
Chris@10: 		    T2r = T2o + T2n;
Chris@10: 		    {
Chris@10: 			 E T2i, T2l, T2s, T2t;
Chris@10: 			 T2i = FMA(KP923879532, T2g, KP382683432 * T2h);
Chris@10: 			 T2l = FNMS(KP382683432, T2k, KP923879532 * T2j);
Chris@10: 			 T2m = T2i + T2l;
Chris@10: 			 T2q = T2l - T2i;
Chris@10: 			 T2s = FNMS(KP382683432, T2g, KP923879532 * T2h);
Chris@10: 			 T2t = FMA(KP382683432, T2j, KP923879532 * T2k);
Chris@10: 			 T2u = T2s + T2t;
Chris@10: 			 T2w = T2t - T2s;
Chris@10: 		    }
Chris@10: 		    Cr[WS(csr, 14)] = T2f - T2m;
Chris@10: 		    Ci[WS(csi, 14)] = T2u - T2r;
Chris@10: 		    Cr[WS(csr, 2)] = T2f + T2m;
Chris@10: 		    Ci[WS(csi, 2)] = T2r + T2u;
Chris@10: 		    Ci[WS(csi, 6)] = T2p + T2q;
Chris@10: 		    Cr[WS(csr, 6)] = T2v + T2w;
Chris@10: 		    Ci[WS(csi, 10)] = T2q - T2p;
Chris@10: 		    Cr[WS(csr, 10)] = T2v - T2w;
Chris@10: 	       }
Chris@10: 	       {
Chris@10: 		    E TH, T1t, T1s, T1u, T1g, T1o, T1n, T1p;
Chris@10: 		    {
Chris@10: 			 E Tz, TG, T1q, T1r;
Chris@10: 			 Tz = Tv + Ty;
Chris@10: 			 TG = TC + TF;
Chris@10: 			 TH = Tz + TG;
Chris@10: 			 T1t = Tz - TG;
Chris@10: 			 T1q = FNMS(KP195090322, TS, KP980785280 * TX);
Chris@10: 			 T1r = FMA(KP195090322, T19, KP980785280 * T1e);
Chris@10: 			 T1s = T1q + T1r;
Chris@10: 			 T1u = T1r - T1q;
Chris@10: 		    }
Chris@10: 		    {
Chris@10: 			 E TY, T1f, T1j, T1m;
Chris@10: 			 TY = FMA(KP980785280, TS, KP195090322 * TX);
Chris@10: 			 T1f = FNMS(KP195090322, T1e, KP980785280 * T19);
Chris@10: 			 T1g = TY + T1f;
Chris@10: 			 T1o = T1f - TY;
Chris@10: 			 T1j = T1h - T1i;
Chris@10: 			 T1m = T1k - T1l;
Chris@10: 			 T1n = T1j - T1m;
Chris@10: 			 T1p = T1m + T1j;
Chris@10: 		    }
Chris@10: 		    Cr[WS(csr, 15)] = TH - T1g;
Chris@10: 		    Ci[WS(csi, 15)] = T1s - T1p;
Chris@10: 		    Cr[WS(csr, 1)] = TH + T1g;
Chris@10: 		    Ci[WS(csi, 1)] = T1p + T1s;
Chris@10: 		    Ci[WS(csi, 7)] = T1n + T1o;
Chris@10: 		    Cr[WS(csr, 7)] = T1t + T1u;
Chris@10: 		    Ci[WS(csi, 9)] = T1o - T1n;
Chris@10: 		    Cr[WS(csr, 9)] = T1t - T1u;
Chris@10: 	       }
Chris@10: 	       {
Chris@10: 		    E T1x, T1N, T1M, T1O, T1E, T1I, T1H, T1J;
Chris@10: 		    {
Chris@10: 			 E T1v, T1w, T1K, T1L;
Chris@10: 			 T1v = Tv - Ty;
Chris@10: 			 T1w = T1i + T1h;
Chris@10: 			 T1x = T1v + T1w;
Chris@10: 			 T1N = T1v - T1w;
Chris@10: 			 T1K = FNMS(KP555570233, T1y, KP831469612 * T1z);
Chris@10: 			 T1L = FMA(KP555570233, T1B, KP831469612 * T1C);
Chris@10: 			 T1M = T1K + T1L;
Chris@10: 			 T1O = T1L - T1K;
Chris@10: 		    }
Chris@10: 		    {
Chris@10: 			 E T1A, T1D, T1F, T1G;
Chris@10: 			 T1A = FMA(KP831469612, T1y, KP555570233 * T1z);
Chris@10: 			 T1D = FNMS(KP555570233, T1C, KP831469612 * T1B);
Chris@10: 			 T1E = T1A + T1D;
Chris@10: 			 T1I = T1D - T1A;
Chris@10: 			 T1F = TF - TC;
Chris@10: 			 T1G = T1l + T1k;
Chris@10: 			 T1H = T1F - T1G;
Chris@10: 			 T1J = T1G + T1F;
Chris@10: 		    }
Chris@10: 		    Cr[WS(csr, 13)] = T1x - T1E;
Chris@10: 		    Ci[WS(csi, 13)] = T1M - T1J;
Chris@10: 		    Cr[WS(csr, 3)] = T1x + T1E;
Chris@10: 		    Ci[WS(csi, 3)] = T1J + T1M;
Chris@10: 		    Ci[WS(csi, 5)] = T1H + T1I;
Chris@10: 		    Cr[WS(csr, 5)] = T1N + T1O;
Chris@10: 		    Ci[WS(csi, 11)] = T1I - T1H;
Chris@10: 		    Cr[WS(csr, 11)] = T1N - T1O;
Chris@10: 	       }
Chris@10: 	  }
Chris@10:      }
Chris@10: }
Chris@10: 
Chris@10: static const kr2c_desc desc = { 32, "r2cf_32", {140, 26, 16, 0}, &GENUS };
Chris@10: 
Chris@10: void X(codelet_r2cf_32) (planner *p) {
Chris@10:      X(kr2c_register) (p, r2cf_32, &desc);
Chris@10: }
Chris@10: 
Chris@10: #endif				/* HAVE_FMA */