Chris@19: /*
Chris@19:  * Copyright (c) 2003, 2007-14 Matteo Frigo
Chris@19:  * Copyright (c) 2003, 2007-14 Massachusetts Institute of Technology
Chris@19:  *
Chris@19:  * This program is free software; you can redistribute it and/or modify
Chris@19:  * it under the terms of the GNU General Public License as published by
Chris@19:  * the Free Software Foundation; either version 2 of the License, or
Chris@19:  * (at your option) any later version.
Chris@19:  *
Chris@19:  * This program is distributed in the hope that it will be useful,
Chris@19:  * but WITHOUT ANY WARRANTY; without even the implied warranty of
Chris@19:  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
Chris@19:  * GNU General Public License for more details.
Chris@19:  *
Chris@19:  * You should have received a copy of the GNU General Public License
Chris@19:  * along with this program; if not, write to the Free Software
Chris@19:  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02110-1301  USA
Chris@19:  *
Chris@19:  */
Chris@19: 
Chris@19: /* This file was automatically generated --- DO NOT EDIT */
Chris@19: /* Generated on Tue Mar  4 13:50:28 EST 2014 */
Chris@19: 
Chris@19: #include "codelet-rdft.h"
Chris@19: 
Chris@19: #ifdef HAVE_FMA
Chris@19: 
Chris@19: /* Generated by: ../../../genfft/gen_hc2hc.native -fma -reorder-insns -schedule-for-pipeline -compact -variables 4 -pipeline-latency 4 -sign 1 -twiddle-log3 -precompute-twiddles -n 8 -dif -name hb2_8 -include hb.h */
Chris@19: 
Chris@19: /*
Chris@19:  * This function contains 74 FP additions, 50 FP multiplications,
Chris@19:  * (or, 44 additions, 20 multiplications, 30 fused multiply/add),
Chris@19:  * 77 stack variables, 1 constants, and 32 memory accesses
Chris@19:  */
Chris@19: #include "hb.h"
Chris@19: 
Chris@19: static void hb2_8(R *cr, R *ci, const R *W, stride rs, INT mb, INT me, INT ms)
Chris@19: {
Chris@19:      DK(KP707106781, +0.707106781186547524400844362104849039284835938);
Chris@19:      {
Chris@19: 	  INT m;
Chris@19: 	  for (m = mb, W = W + ((mb - 1) * 6); m < me; m = m + 1, cr = cr + ms, ci = ci - ms, W = W + 6, MAKE_VOLATILE_STRIDE(16, rs)) {
Chris@19: 	       E Tf, Tg, Tl, Tp, Ti, Tj, T1o, T1u, Tk, T1b, To, T1e, TK, Tq, T13;
Chris@19: 	       E TP, T1p, T7, T1h, T1v, TZ, Tv, Tw, Ta, Tx, T1j, TE, TB, Td, Ty;
Chris@19: 	       E Th, T1n, T1t;
Chris@19: 	       Tf = W[0];
Chris@19: 	       Tg = W[2];
Chris@19: 	       Tl = W[4];
Chris@19: 	       Tp = W[5];
Chris@19: 	       Ti = W[1];
Chris@19: 	       Th = Tf * Tg;
Chris@19: 	       T1n = Tf * Tl;
Chris@19: 	       T1t = Tf * Tp;
Chris@19: 	       Tj = W[3];
Chris@19: 	       {
Chris@19: 		    E Tr, T3, Ts, T1f, TO, TL, T6, Tt;
Chris@19: 		    {
Chris@19: 			 E TM, TN, T4, T5;
Chris@19: 			 {
Chris@19: 			      E T1, Tn, T2, TJ, Tm;
Chris@19: 			      T1 = cr[0];
Chris@19: 			      T1o = FMA(Ti, Tp, T1n);
Chris@19: 			      T1u = FNMS(Ti, Tl, T1t);
Chris@19: 			      Tk = FMA(Ti, Tj, Th);
Chris@19: 			      T1b = FNMS(Ti, Tj, Th);
Chris@19: 			      Tn = Tf * Tj;
Chris@19: 			      T2 = ci[WS(rs, 3)];
Chris@19: 			      TM = ci[WS(rs, 7)];
Chris@19: 			      TJ = Tk * Tp;
Chris@19: 			      Tm = Tk * Tl;
Chris@19: 			      To = FNMS(Ti, Tg, Tn);
Chris@19: 			      T1e = FMA(Ti, Tg, Tn);
Chris@19: 			      Tr = T1 - T2;
Chris@19: 			      T3 = T1 + T2;
Chris@19: 			      TK = FNMS(To, Tl, TJ);
Chris@19: 			      Tq = FMA(To, Tp, Tm);
Chris@19: 			      TN = cr[WS(rs, 4)];
Chris@19: 			 }
Chris@19: 			 T4 = cr[WS(rs, 2)];
Chris@19: 			 T5 = ci[WS(rs, 1)];
Chris@19: 			 Ts = ci[WS(rs, 5)];
Chris@19: 			 T1f = TM - TN;
Chris@19: 			 TO = TM + TN;
Chris@19: 			 TL = T4 - T5;
Chris@19: 			 T6 = T4 + T5;
Chris@19: 			 Tt = cr[WS(rs, 6)];
Chris@19: 		    }
Chris@19: 		    {
Chris@19: 			 E TC, TD, Tb, Tc;
Chris@19: 			 {
Chris@19: 			      E T8, T1g, Tu, T9;
Chris@19: 			      T8 = cr[WS(rs, 1)];
Chris@19: 			      T13 = TO - TL;
Chris@19: 			      TP = TL + TO;
Chris@19: 			      T1p = T3 - T6;
Chris@19: 			      T7 = T3 + T6;
Chris@19: 			      T1g = Ts - Tt;
Chris@19: 			      Tu = Ts + Tt;
Chris@19: 			      T9 = ci[WS(rs, 2)];
Chris@19: 			      TC = ci[WS(rs, 4)];
Chris@19: 			      T1h = T1f + T1g;
Chris@19: 			      T1v = T1f - T1g;
Chris@19: 			      TZ = Tr + Tu;
Chris@19: 			      Tv = Tr - Tu;
Chris@19: 			      Tw = T8 - T9;
Chris@19: 			      Ta = T8 + T9;
Chris@19: 			      TD = cr[WS(rs, 7)];
Chris@19: 			 }
Chris@19: 			 Tb = ci[0];
Chris@19: 			 Tc = cr[WS(rs, 3)];
Chris@19: 			 Tx = ci[WS(rs, 6)];
Chris@19: 			 T1j = TC - TD;
Chris@19: 			 TE = TC + TD;
Chris@19: 			 TB = Tb - Tc;
Chris@19: 			 Td = Tb + Tc;
Chris@19: 			 Ty = cr[WS(rs, 5)];
Chris@19: 		    }
Chris@19: 	       }
Chris@19: 	       {
Chris@19: 		    E TR, TF, Te, T1w;
Chris@19: 		    TR = TB + TE;
Chris@19: 		    TF = TB - TE;
Chris@19: 		    Te = Ta + Td;
Chris@19: 		    T1w = Ta - Td;
Chris@19: 		    {
Chris@19: 			 E Tz, T1i, T1B, T1x, T1c;
Chris@19: 			 Tz = Tx + Ty;
Chris@19: 			 T1i = Tx - Ty;
Chris@19: 			 T1B = T1w + T1v;
Chris@19: 			 T1x = T1v - T1w;
Chris@19: 			 T1c = T7 - Te;
Chris@19: 			 cr[0] = T7 + Te;
Chris@19: 			 {
Chris@19: 			      E T1k, T1q, TQ, TA;
Chris@19: 			      T1k = T1i + T1j;
Chris@19: 			      T1q = T1j - T1i;
Chris@19: 			      TQ = Tw + Tz;
Chris@19: 			      TA = Tw - Tz;
Chris@19: 			      {
Chris@19: 				   E T1y, T1C, T1m, T1d;
Chris@19: 				   T1y = T1o * T1x;
Chris@19: 				   T1C = Tk * T1B;
Chris@19: 				   T1m = T1e * T1c;
Chris@19: 				   T1d = T1b * T1c;
Chris@19: 				   {
Chris@19: 					E T1z, T1r, T1l, TG, T14;
Chris@19: 					T1z = T1p + T1q;
Chris@19: 					T1r = T1p - T1q;
Chris@19: 					T1l = T1h - T1k;
Chris@19: 					ci[0] = T1h + T1k;
Chris@19: 					TG = TA + TF;
Chris@19: 					T14 = TA - TF;
Chris@19: 					{
Chris@19: 					     E T10, TS, T1s, T1A;
Chris@19: 					     T10 = TQ + TR;
Chris@19: 					     TS = TQ - TR;
Chris@19: 					     ci[WS(rs, 6)] = FMA(T1u, T1r, T1y);
Chris@19: 					     T1s = T1o * T1r;
Chris@19: 					     ci[WS(rs, 2)] = FMA(To, T1z, T1C);
Chris@19: 					     T1A = Tk * T1z;
Chris@19: 					     ci[WS(rs, 4)] = FMA(T1b, T1l, T1m);
Chris@19: 					     cr[WS(rs, 4)] = FNMS(T1e, T1l, T1d);
Chris@19: 					     {
Chris@19: 						  E T15, T19, TV, TH;
Chris@19: 						  T15 = FMA(KP707106781, T14, T13);
Chris@19: 						  T19 = FNMS(KP707106781, T14, T13);
Chris@19: 						  TV = FMA(KP707106781, TG, Tv);
Chris@19: 						  TH = FNMS(KP707106781, TG, Tv);
Chris@19: 						  {
Chris@19: 						       E TT, TX, T11, T17;
Chris@19: 						       TT = FNMS(KP707106781, TS, TP);
Chris@19: 						       TX = FMA(KP707106781, TS, TP);
Chris@19: 						       T11 = FNMS(KP707106781, T10, TZ);
Chris@19: 						       T17 = FMA(KP707106781, T10, TZ);
Chris@19: 						       cr[WS(rs, 6)] = FNMS(T1u, T1x, T1s);
Chris@19: 						       cr[WS(rs, 2)] = FNMS(To, T1B, T1A);
Chris@19: 						       {
Chris@19: 							    E T1a, T16, TU, TI;
Chris@19: 							    T1a = Tl * T19;
Chris@19: 							    T16 = Tg * T15;
Chris@19: 							    TU = TK * TH;
Chris@19: 							    TI = Tq * TH;
Chris@19: 							    {
Chris@19: 								 E TY, TW, T18, T12;
Chris@19: 								 TY = Ti * TV;
Chris@19: 								 TW = Tf * TV;
Chris@19: 								 T18 = Tl * T17;
Chris@19: 								 T12 = Tg * T11;
Chris@19: 								 ci[WS(rs, 7)] = FMA(Tp, T17, T1a);
Chris@19: 								 ci[WS(rs, 3)] = FMA(Tj, T11, T16);
Chris@19: 								 ci[WS(rs, 5)] = FMA(Tq, TT, TU);
Chris@19: 								 cr[WS(rs, 5)] = FNMS(TK, TT, TI);
Chris@19: 								 ci[WS(rs, 1)] = FMA(Tf, TX, TY);
Chris@19: 								 cr[WS(rs, 1)] = FNMS(Ti, TX, TW);
Chris@19: 								 cr[WS(rs, 7)] = FNMS(Tp, T19, T18);
Chris@19: 								 cr[WS(rs, 3)] = FNMS(Tj, T15, T12);
Chris@19: 							    }
Chris@19: 						       }
Chris@19: 						  }
Chris@19: 					     }
Chris@19: 					}
Chris@19: 				   }
Chris@19: 			      }
Chris@19: 			 }
Chris@19: 		    }
Chris@19: 	       }
Chris@19: 	  }
Chris@19:      }
Chris@19: }
Chris@19: 
Chris@19: static const tw_instr twinstr[] = {
Chris@19:      {TW_CEXP, 1, 1},
Chris@19:      {TW_CEXP, 1, 3},
Chris@19:      {TW_CEXP, 1, 7},
Chris@19:      {TW_NEXT, 1, 0}
Chris@19: };
Chris@19: 
Chris@19: static const hc2hc_desc desc = { 8, "hb2_8", twinstr, &GENUS, {44, 20, 30, 0} };
Chris@19: 
Chris@19: void X(codelet_hb2_8) (planner *p) {
Chris@19:      X(khc2hc_register) (p, hb2_8, &desc);
Chris@19: }
Chris@19: #else				/* HAVE_FMA */
Chris@19: 
Chris@19: /* Generated by: ../../../genfft/gen_hc2hc.native -compact -variables 4 -pipeline-latency 4 -sign 1 -twiddle-log3 -precompute-twiddles -n 8 -dif -name hb2_8 -include hb.h */
Chris@19: 
Chris@19: /*
Chris@19:  * This function contains 74 FP additions, 44 FP multiplications,
Chris@19:  * (or, 56 additions, 26 multiplications, 18 fused multiply/add),
Chris@19:  * 46 stack variables, 1 constants, and 32 memory accesses
Chris@19:  */
Chris@19: #include "hb.h"
Chris@19: 
Chris@19: static void hb2_8(R *cr, R *ci, const R *W, stride rs, INT mb, INT me, INT ms)
Chris@19: {
Chris@19:      DK(KP707106781, +0.707106781186547524400844362104849039284835938);
Chris@19:      {
Chris@19: 	  INT m;
Chris@19: 	  for (m = mb, W = W + ((mb - 1) * 6); m < me; m = m + 1, cr = cr + ms, ci = ci - ms, W = W + 6, MAKE_VOLATILE_STRIDE(16, rs)) {
Chris@19: 	       E Tf, Ti, Tg, Tj, Tl, Tp, TP, TR, TF, TG, TH, T15, TL, TT;
Chris@19: 	       {
Chris@19: 		    E Th, To, Tk, Tn;
Chris@19: 		    Tf = W[0];
Chris@19: 		    Ti = W[1];
Chris@19: 		    Tg = W[2];
Chris@19: 		    Tj = W[3];
Chris@19: 		    Th = Tf * Tg;
Chris@19: 		    To = Ti * Tg;
Chris@19: 		    Tk = Ti * Tj;
Chris@19: 		    Tn = Tf * Tj;
Chris@19: 		    Tl = Th - Tk;
Chris@19: 		    Tp = Tn + To;
Chris@19: 		    TP = Th + Tk;
Chris@19: 		    TR = Tn - To;
Chris@19: 		    TF = W[4];
Chris@19: 		    TG = W[5];
Chris@19: 		    TH = FMA(Tf, TF, Ti * TG);
Chris@19: 		    T15 = FNMS(TR, TF, TP * TG);
Chris@19: 		    TL = FNMS(Ti, TF, Tf * TG);
Chris@19: 		    TT = FMA(TP, TF, TR * TG);
Chris@19: 	       }
Chris@19: 	       {
Chris@19: 		    E T7, T1f, T1i, Tw, TI, TW, T18, TM, Te, T19, T1a, TD, TJ, TZ, T12;
Chris@19: 		    E TN, Tm, TE;
Chris@19: 		    {
Chris@19: 			 E T3, TU, Tv, TV, T6, T16, Ts, T17;
Chris@19: 			 {
Chris@19: 			      E T1, T2, Tt, Tu;
Chris@19: 			      T1 = cr[0];
Chris@19: 			      T2 = ci[WS(rs, 3)];
Chris@19: 			      T3 = T1 + T2;
Chris@19: 			      TU = T1 - T2;
Chris@19: 			      Tt = ci[WS(rs, 5)];
Chris@19: 			      Tu = cr[WS(rs, 6)];
Chris@19: 			      Tv = Tt - Tu;
Chris@19: 			      TV = Tt + Tu;
Chris@19: 			 }
Chris@19: 			 {
Chris@19: 			      E T4, T5, Tq, Tr;
Chris@19: 			      T4 = cr[WS(rs, 2)];
Chris@19: 			      T5 = ci[WS(rs, 1)];
Chris@19: 			      T6 = T4 + T5;
Chris@19: 			      T16 = T4 - T5;
Chris@19: 			      Tq = ci[WS(rs, 7)];
Chris@19: 			      Tr = cr[WS(rs, 4)];
Chris@19: 			      Ts = Tq - Tr;
Chris@19: 			      T17 = Tq + Tr;
Chris@19: 			 }
Chris@19: 			 T7 = T3 + T6;
Chris@19: 			 T1f = TU + TV;
Chris@19: 			 T1i = T17 - T16;
Chris@19: 			 Tw = Ts + Tv;
Chris@19: 			 TI = T3 - T6;
Chris@19: 			 TW = TU - TV;
Chris@19: 			 T18 = T16 + T17;
Chris@19: 			 TM = Ts - Tv;
Chris@19: 		    }
Chris@19: 		    {
Chris@19: 			 E Ta, TX, TC, T11, Td, T10, Tz, TY;
Chris@19: 			 {
Chris@19: 			      E T8, T9, TA, TB;
Chris@19: 			      T8 = cr[WS(rs, 1)];
Chris@19: 			      T9 = ci[WS(rs, 2)];
Chris@19: 			      Ta = T8 + T9;
Chris@19: 			      TX = T8 - T9;
Chris@19: 			      TA = ci[WS(rs, 4)];
Chris@19: 			      TB = cr[WS(rs, 7)];
Chris@19: 			      TC = TA - TB;
Chris@19: 			      T11 = TA + TB;
Chris@19: 			 }
Chris@19: 			 {
Chris@19: 			      E Tb, Tc, Tx, Ty;
Chris@19: 			      Tb = ci[0];
Chris@19: 			      Tc = cr[WS(rs, 3)];
Chris@19: 			      Td = Tb + Tc;
Chris@19: 			      T10 = Tb - Tc;
Chris@19: 			      Tx = ci[WS(rs, 6)];
Chris@19: 			      Ty = cr[WS(rs, 5)];
Chris@19: 			      Tz = Tx - Ty;
Chris@19: 			      TY = Tx + Ty;
Chris@19: 			 }
Chris@19: 			 Te = Ta + Td;
Chris@19: 			 T19 = TX + TY;
Chris@19: 			 T1a = T10 + T11;
Chris@19: 			 TD = Tz + TC;
Chris@19: 			 TJ = TC - Tz;
Chris@19: 			 TZ = TX - TY;
Chris@19: 			 T12 = T10 - T11;
Chris@19: 			 TN = Ta - Td;
Chris@19: 		    }
Chris@19: 		    cr[0] = T7 + Te;
Chris@19: 		    ci[0] = Tw + TD;
Chris@19: 		    Tm = T7 - Te;
Chris@19: 		    TE = Tw - TD;
Chris@19: 		    cr[WS(rs, 4)] = FNMS(Tp, TE, Tl * Tm);
Chris@19: 		    ci[WS(rs, 4)] = FMA(Tp, Tm, Tl * TE);
Chris@19: 		    {
Chris@19: 			 E TQ, TS, TK, TO;
Chris@19: 			 TQ = TI + TJ;
Chris@19: 			 TS = TN + TM;
Chris@19: 			 cr[WS(rs, 2)] = FNMS(TR, TS, TP * TQ);
Chris@19: 			 ci[WS(rs, 2)] = FMA(TP, TS, TR * TQ);
Chris@19: 			 TK = TI - TJ;
Chris@19: 			 TO = TM - TN;
Chris@19: 			 cr[WS(rs, 6)] = FNMS(TL, TO, TH * TK);
Chris@19: 			 ci[WS(rs, 6)] = FMA(TH, TO, TL * TK);
Chris@19: 		    }
Chris@19: 		    {
Chris@19: 			 E T1h, T1l, T1k, T1m, T1g, T1j;
Chris@19: 			 T1g = KP707106781 * (T19 + T1a);
Chris@19: 			 T1h = T1f - T1g;
Chris@19: 			 T1l = T1f + T1g;
Chris@19: 			 T1j = KP707106781 * (TZ - T12);
Chris@19: 			 T1k = T1i + T1j;
Chris@19: 			 T1m = T1i - T1j;
Chris@19: 			 cr[WS(rs, 3)] = FNMS(Tj, T1k, Tg * T1h);
Chris@19: 			 ci[WS(rs, 3)] = FMA(Tg, T1k, Tj * T1h);
Chris@19: 			 cr[WS(rs, 7)] = FNMS(TG, T1m, TF * T1l);
Chris@19: 			 ci[WS(rs, 7)] = FMA(TF, T1m, TG * T1l);
Chris@19: 		    }
Chris@19: 		    {
Chris@19: 			 E T14, T1d, T1c, T1e, T13, T1b;
Chris@19: 			 T13 = KP707106781 * (TZ + T12);
Chris@19: 			 T14 = TW - T13;
Chris@19: 			 T1d = TW + T13;
Chris@19: 			 T1b = KP707106781 * (T19 - T1a);
Chris@19: 			 T1c = T18 - T1b;
Chris@19: 			 T1e = T18 + T1b;
Chris@19: 			 cr[WS(rs, 5)] = FNMS(T15, T1c, TT * T14);
Chris@19: 			 ci[WS(rs, 5)] = FMA(T15, T14, TT * T1c);
Chris@19: 			 cr[WS(rs, 1)] = FNMS(Ti, T1e, Tf * T1d);
Chris@19: 			 ci[WS(rs, 1)] = FMA(Ti, T1d, Tf * T1e);
Chris@19: 		    }
Chris@19: 	       }
Chris@19: 	  }
Chris@19:      }
Chris@19: }
Chris@19: 
Chris@19: static const tw_instr twinstr[] = {
Chris@19:      {TW_CEXP, 1, 1},
Chris@19:      {TW_CEXP, 1, 3},
Chris@19:      {TW_CEXP, 1, 7},
Chris@19:      {TW_NEXT, 1, 0}
Chris@19: };
Chris@19: 
Chris@19: static const hc2hc_desc desc = { 8, "hb2_8", twinstr, &GENUS, {56, 26, 18, 0} };
Chris@19: 
Chris@19: void X(codelet_hb2_8) (planner *p) {
Chris@19:      X(khc2hc_register) (p, hb2_8, &desc);
Chris@19: }
Chris@19: #endif				/* HAVE_FMA */