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 16:49:30 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_r2cb.native -fma -reorder-insns -schedule-for-pipeline -compact -variables 4 -pipeline-latency 4 -sign 1 -n 16 -name r2cb_16 -include r2cb.h */
Chris@42: 
Chris@42: /*
Chris@42:  * This function contains 58 FP additions, 32 FP multiplications,
Chris@42:  * (or, 26 additions, 0 multiplications, 32 fused multiply/add),
Chris@42:  * 47 stack variables, 4 constants, and 32 memory accesses
Chris@42:  */
Chris@42: #include "r2cb.h"
Chris@42: 
Chris@42: static void r2cb_16(R *R0, R *R1, R *Cr, R *Ci, stride rs, stride csr, stride csi, INT v, INT ivs, INT ovs)
Chris@42: {
Chris@42:      DK(KP1_847759065, +1.847759065022573512256366378793576573644833252);
Chris@42:      DK(KP414213562, +0.414213562373095048801688724209698078569671875);
Chris@42:      DK(KP1_414213562, +1.414213562373095048801688724209698078569671875);
Chris@42:      DK(KP2_000000000, +2.000000000000000000000000000000000000000000000);
Chris@42:      {
Chris@42: 	  INT i;
Chris@42: 	  for (i = v; i > 0; i = i - 1, R0 = R0 + ovs, R1 = R1 + ovs, Cr = Cr + ivs, Ci = Ci + ivs, MAKE_VOLATILE_STRIDE(64, rs), MAKE_VOLATILE_STRIDE(64, csr), MAKE_VOLATILE_STRIDE(64, csi)) {
Chris@42: 	       E TN, TS, TF, TI;
Chris@42: 	       {
Chris@42: 		    E T8, TD, Tj, TL, T5, TM, TE, To, Td, Tq, Tc, TP, Ty, Te, Tr;
Chris@42: 		    E Ts;
Chris@42: 		    {
Chris@42: 			 E T4, Ti, T1, T2;
Chris@42: 			 T4 = Cr[WS(csr, 4)];
Chris@42: 			 Ti = Ci[WS(csi, 4)];
Chris@42: 			 T1 = Cr[0];
Chris@42: 			 T2 = Cr[WS(csr, 8)];
Chris@42: 			 {
Chris@42: 			      E Tk, Tn, T6, T7;
Chris@42: 			      T6 = Cr[WS(csr, 2)];
Chris@42: 			      T7 = Cr[WS(csr, 6)];
Chris@42: 			      {
Chris@42: 				   E Tl, Th, T3, Tm;
Chris@42: 				   Tl = Ci[WS(csi, 2)];
Chris@42: 				   Th = T1 - T2;
Chris@42: 				   T3 = T1 + T2;
Chris@42: 				   Tk = T6 - T7;
Chris@42: 				   T8 = T6 + T7;
Chris@42: 				   Tm = Ci[WS(csi, 6)];
Chris@42: 				   TD = FMA(KP2_000000000, Ti, Th);
Chris@42: 				   Tj = FNMS(KP2_000000000, Ti, Th);
Chris@42: 				   TL = FNMS(KP2_000000000, T4, T3);
Chris@42: 				   T5 = FMA(KP2_000000000, T4, T3);
Chris@42: 				   Tn = Tl + Tm;
Chris@42: 				   TM = Tl - Tm;
Chris@42: 			      }
Chris@42: 			      {
Chris@42: 				   E Ta, Tb, Tw, Tx;
Chris@42: 				   Ta = Cr[WS(csr, 1)];
Chris@42: 				   TE = Tk + Tn;
Chris@42: 				   To = Tk - Tn;
Chris@42: 				   Tb = Cr[WS(csr, 7)];
Chris@42: 				   Tw = Ci[WS(csi, 1)];
Chris@42: 				   Tx = Ci[WS(csi, 7)];
Chris@42: 				   Td = Cr[WS(csr, 5)];
Chris@42: 				   Tq = Ta - Tb;
Chris@42: 				   Tc = Ta + Tb;
Chris@42: 				   TP = Tw - Tx;
Chris@42: 				   Ty = Tw + Tx;
Chris@42: 				   Te = Cr[WS(csr, 3)];
Chris@42: 				   Tr = Ci[WS(csi, 5)];
Chris@42: 				   Ts = Ci[WS(csi, 3)];
Chris@42: 			      }
Chris@42: 			 }
Chris@42: 		    }
Chris@42: 		    {
Chris@42: 			 E TV, TG, TW, TH, TB, Tp, TA, TC, TJ, TK;
Chris@42: 			 {
Chris@42: 			      E T9, Tz, Tg, Tu, TT, TU, TO, TR;
Chris@42: 			      TV = FNMS(KP2_000000000, T8, T5);
Chris@42: 			      T9 = FMA(KP2_000000000, T8, T5);
Chris@42: 			      {
Chris@42: 				   E Tv, Tf, TQ, Tt;
Chris@42: 				   Tv = Td - Te;
Chris@42: 				   Tf = Td + Te;
Chris@42: 				   TQ = Tr - Ts;
Chris@42: 				   Tt = Tr + Ts;
Chris@42: 				   TG = Ty - Tv;
Chris@42: 				   Tz = Tv + Ty;
Chris@42: 				   TO = Tc - Tf;
Chris@42: 				   Tg = Tc + Tf;
Chris@42: 				   TW = TQ + TP;
Chris@42: 				   TR = TP - TQ;
Chris@42: 				   TH = Tq + Tt;
Chris@42: 				   Tu = Tq - Tt;
Chris@42: 			      }
Chris@42: 			      TN = FNMS(KP2_000000000, TM, TL);
Chris@42: 			      TT = FMA(KP2_000000000, TM, TL);
Chris@42: 			      TU = TO + TR;
Chris@42: 			      TS = TO - TR;
Chris@42: 			      R0[0] = FMA(KP2_000000000, Tg, T9);
Chris@42: 			      R0[WS(rs, 4)] = FNMS(KP2_000000000, Tg, T9);
Chris@42: 			      R0[WS(rs, 7)] = FMA(KP1_414213562, TU, TT);
Chris@42: 			      R0[WS(rs, 3)] = FNMS(KP1_414213562, TU, TT);
Chris@42: 			      TB = FNMS(KP1_414213562, To, Tj);
Chris@42: 			      Tp = FMA(KP1_414213562, To, Tj);
Chris@42: 			      TA = FNMS(KP414213562, Tz, Tu);
Chris@42: 			      TC = FMA(KP414213562, Tu, Tz);
Chris@42: 			 }
Chris@42: 			 R0[WS(rs, 6)] = FMA(KP2_000000000, TW, TV);
Chris@42: 			 R0[WS(rs, 2)] = FNMS(KP2_000000000, TW, TV);
Chris@42: 			 R1[0] = FMA(KP1_847759065, TA, Tp);
Chris@42: 			 R1[WS(rs, 4)] = FNMS(KP1_847759065, TA, Tp);
Chris@42: 			 TF = FNMS(KP1_414213562, TE, TD);
Chris@42: 			 TJ = FMA(KP1_414213562, TE, TD);
Chris@42: 			 TK = FMA(KP414213562, TG, TH);
Chris@42: 			 TI = FNMS(KP414213562, TH, TG);
Chris@42: 			 R1[WS(rs, 6)] = FMA(KP1_847759065, TC, TB);
Chris@42: 			 R1[WS(rs, 2)] = FNMS(KP1_847759065, TC, TB);
Chris@42: 			 R1[WS(rs, 7)] = FMA(KP1_847759065, TK, TJ);
Chris@42: 			 R1[WS(rs, 3)] = FNMS(KP1_847759065, TK, TJ);
Chris@42: 		    }
Chris@42: 	       }
Chris@42: 	       R0[WS(rs, 1)] = FMA(KP1_414213562, TS, TN);
Chris@42: 	       R0[WS(rs, 5)] = FNMS(KP1_414213562, TS, TN);
Chris@42: 	       R1[WS(rs, 5)] = FMA(KP1_847759065, TI, TF);
Chris@42: 	       R1[WS(rs, 1)] = FNMS(KP1_847759065, TI, TF);
Chris@42: 	  }
Chris@42:      }
Chris@42: }
Chris@42: 
Chris@42: static const kr2c_desc desc = { 16, "r2cb_16", {26, 0, 32, 0}, &GENUS };
Chris@42: 
Chris@42: void X(codelet_r2cb_16) (planner *p) {
Chris@42:      X(kr2c_register) (p, r2cb_16, &desc);
Chris@42: }
Chris@42: 
Chris@42: #else				/* HAVE_FMA */
Chris@42: 
Chris@42: /* Generated by: ../../../genfft/gen_r2cb.native -compact -variables 4 -pipeline-latency 4 -sign 1 -n 16 -name r2cb_16 -include r2cb.h */
Chris@42: 
Chris@42: /*
Chris@42:  * This function contains 58 FP additions, 18 FP multiplications,
Chris@42:  * (or, 54 additions, 14 multiplications, 4 fused multiply/add),
Chris@42:  * 31 stack variables, 4 constants, and 32 memory accesses
Chris@42:  */
Chris@42: #include "r2cb.h"
Chris@42: 
Chris@42: static void r2cb_16(R *R0, R *R1, R *Cr, R *Ci, stride rs, stride csr, stride csi, INT v, INT ivs, INT ovs)
Chris@42: {
Chris@42:      DK(KP1_847759065, +1.847759065022573512256366378793576573644833252);
Chris@42:      DK(KP765366864, +0.765366864730179543456919968060797733522689125);
Chris@42:      DK(KP1_414213562, +1.414213562373095048801688724209698078569671875);
Chris@42:      DK(KP2_000000000, +2.000000000000000000000000000000000000000000000);
Chris@42:      {
Chris@42: 	  INT i;
Chris@42: 	  for (i = v; i > 0; i = i - 1, R0 = R0 + ovs, R1 = R1 + ovs, Cr = Cr + ivs, Ci = Ci + ivs, MAKE_VOLATILE_STRIDE(64, rs), MAKE_VOLATILE_STRIDE(64, csr), MAKE_VOLATILE_STRIDE(64, csi)) {
Chris@42: 	       E T9, TS, Tl, TG, T6, TR, Ti, TD, Td, Tq, Tg, Tt, Tn, Tu, TV;
Chris@42: 	       E TU, TN, TK;
Chris@42: 	       {
Chris@42: 		    E T7, T8, TE, Tj, Tk, TF;
Chris@42: 		    T7 = Cr[WS(csr, 2)];
Chris@42: 		    T8 = Cr[WS(csr, 6)];
Chris@42: 		    TE = T7 - T8;
Chris@42: 		    Tj = Ci[WS(csi, 2)];
Chris@42: 		    Tk = Ci[WS(csi, 6)];
Chris@42: 		    TF = Tj + Tk;
Chris@42: 		    T9 = KP2_000000000 * (T7 + T8);
Chris@42: 		    TS = KP1_414213562 * (TE + TF);
Chris@42: 		    Tl = KP2_000000000 * (Tj - Tk);
Chris@42: 		    TG = KP1_414213562 * (TE - TF);
Chris@42: 	       }
Chris@42: 	       {
Chris@42: 		    E T5, TC, T3, TA;
Chris@42: 		    {
Chris@42: 			 E T4, TB, T1, T2;
Chris@42: 			 T4 = Cr[WS(csr, 4)];
Chris@42: 			 T5 = KP2_000000000 * T4;
Chris@42: 			 TB = Ci[WS(csi, 4)];
Chris@42: 			 TC = KP2_000000000 * TB;
Chris@42: 			 T1 = Cr[0];
Chris@42: 			 T2 = Cr[WS(csr, 8)];
Chris@42: 			 T3 = T1 + T2;
Chris@42: 			 TA = T1 - T2;
Chris@42: 		    }
Chris@42: 		    T6 = T3 + T5;
Chris@42: 		    TR = TA + TC;
Chris@42: 		    Ti = T3 - T5;
Chris@42: 		    TD = TA - TC;
Chris@42: 	       }
Chris@42: 	       {
Chris@42: 		    E TI, TM, TL, TJ;
Chris@42: 		    {
Chris@42: 			 E Tb, Tc, To, Tp;
Chris@42: 			 Tb = Cr[WS(csr, 1)];
Chris@42: 			 Tc = Cr[WS(csr, 7)];
Chris@42: 			 Td = Tb + Tc;
Chris@42: 			 TI = Tb - Tc;
Chris@42: 			 To = Ci[WS(csi, 1)];
Chris@42: 			 Tp = Ci[WS(csi, 7)];
Chris@42: 			 Tq = To - Tp;
Chris@42: 			 TM = To + Tp;
Chris@42: 		    }
Chris@42: 		    {
Chris@42: 			 E Te, Tf, Tr, Ts;
Chris@42: 			 Te = Cr[WS(csr, 5)];
Chris@42: 			 Tf = Cr[WS(csr, 3)];
Chris@42: 			 Tg = Te + Tf;
Chris@42: 			 TL = Te - Tf;
Chris@42: 			 Tr = Ci[WS(csi, 5)];
Chris@42: 			 Ts = Ci[WS(csi, 3)];
Chris@42: 			 Tt = Tr - Ts;
Chris@42: 			 TJ = Tr + Ts;
Chris@42: 		    }
Chris@42: 		    Tn = Td - Tg;
Chris@42: 		    Tu = Tq - Tt;
Chris@42: 		    TV = TM - TL;
Chris@42: 		    TU = TI + TJ;
Chris@42: 		    TN = TL + TM;
Chris@42: 		    TK = TI - TJ;
Chris@42: 	       }
Chris@42: 	       {
Chris@42: 		    E Ta, Th, TT, TW;
Chris@42: 		    Ta = T6 + T9;
Chris@42: 		    Th = KP2_000000000 * (Td + Tg);
Chris@42: 		    R0[WS(rs, 4)] = Ta - Th;
Chris@42: 		    R0[0] = Ta + Th;
Chris@42: 		    TT = TR - TS;
Chris@42: 		    TW = FNMS(KP1_847759065, TV, KP765366864 * TU);
Chris@42: 		    R1[WS(rs, 5)] = TT - TW;
Chris@42: 		    R1[WS(rs, 1)] = TT + TW;
Chris@42: 	       }
Chris@42: 	       {
Chris@42: 		    E TX, TY, Tm, Tv;
Chris@42: 		    TX = TR + TS;
Chris@42: 		    TY = FMA(KP1_847759065, TU, KP765366864 * TV);
Chris@42: 		    R1[WS(rs, 3)] = TX - TY;
Chris@42: 		    R1[WS(rs, 7)] = TX + TY;
Chris@42: 		    Tm = Ti - Tl;
Chris@42: 		    Tv = KP1_414213562 * (Tn - Tu);
Chris@42: 		    R0[WS(rs, 5)] = Tm - Tv;
Chris@42: 		    R0[WS(rs, 1)] = Tm + Tv;
Chris@42: 	       }
Chris@42: 	       {
Chris@42: 		    E Tw, Tx, TH, TO;
Chris@42: 		    Tw = Ti + Tl;
Chris@42: 		    Tx = KP1_414213562 * (Tn + Tu);
Chris@42: 		    R0[WS(rs, 3)] = Tw - Tx;
Chris@42: 		    R0[WS(rs, 7)] = Tw + Tx;
Chris@42: 		    TH = TD + TG;
Chris@42: 		    TO = FNMS(KP765366864, TN, KP1_847759065 * TK);
Chris@42: 		    R1[WS(rs, 4)] = TH - TO;
Chris@42: 		    R1[0] = TH + TO;
Chris@42: 	       }
Chris@42: 	       {
Chris@42: 		    E TP, TQ, Ty, Tz;
Chris@42: 		    TP = TD - TG;
Chris@42: 		    TQ = FMA(KP765366864, TK, KP1_847759065 * TN);
Chris@42: 		    R1[WS(rs, 2)] = TP - TQ;
Chris@42: 		    R1[WS(rs, 6)] = TP + TQ;
Chris@42: 		    Ty = T6 - T9;
Chris@42: 		    Tz = KP2_000000000 * (Tt + Tq);
Chris@42: 		    R0[WS(rs, 2)] = Ty - Tz;
Chris@42: 		    R0[WS(rs, 6)] = Ty + Tz;
Chris@42: 	       }
Chris@42: 	  }
Chris@42:      }
Chris@42: }
Chris@42: 
Chris@42: static const kr2c_desc desc = { 16, "r2cb_16", {54, 14, 4, 0}, &GENUS };
Chris@42: 
Chris@42: void X(codelet_r2cb_16) (planner *p) {
Chris@42:      X(kr2c_register) (p, r2cb_16, &desc);
Chris@42: }
Chris@42: 
Chris@42: #endif				/* HAVE_FMA */