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 07:41:08 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_r2cb.native -fma -reorder-insns -schedule-for-pipeline -compact -variables 4 -pipeline-latency 4 -sign 1 -n 14 -name r2cb_14 -include r2cb.h */
Chris@10: 
Chris@10: /*
Chris@10:  * This function contains 62 FP additions, 44 FP multiplications,
Chris@10:  * (or, 18 additions, 0 multiplications, 44 fused multiply/add),
Chris@10:  * 58 stack variables, 7 constants, and 28 memory accesses
Chris@10:  */
Chris@10: #include "r2cb.h"
Chris@10: 
Chris@10: static void r2cb_14(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(KP1_949855824, +1.949855824363647214036263365987862434465571601);
Chris@10:      DK(KP1_801937735, +1.801937735804838252472204639014890102331838324);
Chris@10:      DK(KP692021471, +0.692021471630095869627814897002069140197260599);
Chris@10:      DK(KP2_000000000, +2.000000000000000000000000000000000000000000000);
Chris@10:      DK(KP356895867, +0.356895867892209443894399510021300583399127187);
Chris@10:      DK(KP801937735, +0.801937735804838252472204639014890102331838324);
Chris@10:      DK(KP554958132, +0.554958132087371191422194871006410481067288862);
Chris@10:      {
Chris@10: 	  INT i;
Chris@10: 	  for (i = v; i > 0; i = i - 1, R0 = R0 + ovs, R1 = R1 + ovs, Cr = Cr + ivs, Ci = Ci + ivs, MAKE_VOLATILE_STRIDE(56, rs), MAKE_VOLATILE_STRIDE(56, csr), MAKE_VOLATILE_STRIDE(56, csi)) {
Chris@10: 	       E Te, TO, TT, TG, TJ, TD, TR, TE;
Chris@10: 	       {
Chris@10: 		    E T3, TK, To, TM, Tu, TL, Tr, TS, TA, TN, TX, TF, Tv, T7, Tf;
Chris@10: 		    E T6, Th, Tc, T8, T1, T2;
Chris@10: 		    T1 = Cr[0];
Chris@10: 		    T2 = Cr[WS(csr, 7)];
Chris@10: 		    {
Chris@10: 			 E Ts, Tt, Tp, Tq, Tm, Tn;
Chris@10: 			 Tm = Ci[WS(csi, 4)];
Chris@10: 			 Tn = Ci[WS(csi, 3)];
Chris@10: 			 Ts = Ci[WS(csi, 6)];
Chris@10: 			 Te = T1 + T2;
Chris@10: 			 T3 = T1 - T2;
Chris@10: 			 TK = Tm + Tn;
Chris@10: 			 To = Tm - Tn;
Chris@10: 			 Tt = Ci[WS(csi, 1)];
Chris@10: 			 Tp = Ci[WS(csi, 2)];
Chris@10: 			 Tq = Ci[WS(csi, 5)];
Chris@10: 			 {
Chris@10: 			      E T4, T5, Ta, Tb;
Chris@10: 			      T4 = Cr[WS(csr, 2)];
Chris@10: 			      TM = Ts + Tt;
Chris@10: 			      Tu = Ts - Tt;
Chris@10: 			      TL = Tp + Tq;
Chris@10: 			      Tr = Tp - Tq;
Chris@10: 			      TS = FMA(KP554958132, TK, TM);
Chris@10: 			      TA = FMA(KP554958132, To, Tu);
Chris@10: 			      TN = FMA(KP554958132, TM, TL);
Chris@10: 			      TX = FNMS(KP554958132, TL, TK);
Chris@10: 			      TF = FNMS(KP554958132, Tr, To);
Chris@10: 			      Tv = FMA(KP554958132, Tu, Tr);
Chris@10: 			      T5 = Cr[WS(csr, 5)];
Chris@10: 			      Ta = Cr[WS(csr, 6)];
Chris@10: 			      Tb = Cr[WS(csr, 1)];
Chris@10: 			      T7 = Cr[WS(csr, 4)];
Chris@10: 			      Tf = T4 + T5;
Chris@10: 			      T6 = T4 - T5;
Chris@10: 			      Th = Ta + Tb;
Chris@10: 			      Tc = Ta - Tb;
Chris@10: 			      T8 = Cr[WS(csr, 3)];
Chris@10: 			 }
Chris@10: 		    }
Chris@10: 		    {
Chris@10: 			 E Tw, Tx, TP, Tg, T9, TY, TC, TI, TQ;
Chris@10: 			 Tw = FMA(KP801937735, Tv, To);
Chris@10: 			 Tx = FNMS(KP356895867, Tf, Th);
Chris@10: 			 TP = FNMS(KP356895867, T6, Tc);
Chris@10: 			 Tg = T7 + T8;
Chris@10: 			 T9 = T7 - T8;
Chris@10: 			 TY = FNMS(KP801937735, TX, TM);
Chris@10: 			 {
Chris@10: 			      E TB, TH, TV, Ty, Tl, Ti, TW, Tz;
Chris@10: 			      TB = FNMS(KP801937735, TA, Tr);
Chris@10: 			      Ti = Tf + Tg + Th;
Chris@10: 			      TC = FNMS(KP356895867, Th, Tg);
Chris@10: 			      {
Chris@10: 				   E Tj, Td, TU, Tk;
Chris@10: 				   Tj = FNMS(KP356895867, Tg, Tf);
Chris@10: 				   Td = T6 + T9 + Tc;
Chris@10: 				   TH = FNMS(KP356895867, T9, T6);
Chris@10: 				   TU = FNMS(KP356895867, Tc, T9);
Chris@10: 				   R0[0] = FMA(KP2_000000000, Ti, Te);
Chris@10: 				   Tk = FNMS(KP692021471, Tj, Th);
Chris@10: 				   R1[WS(rs, 3)] = FMA(KP2_000000000, Td, T3);
Chris@10: 				   TV = FNMS(KP692021471, TU, T6);
Chris@10: 				   Ty = FNMS(KP692021471, Tx, Tg);
Chris@10: 				   Tl = FNMS(KP1_801937735, Tk, Te);
Chris@10: 			      }
Chris@10: 			      TO = FMA(KP801937735, TN, TK);
Chris@10: 			      TW = FNMS(KP1_801937735, TV, T3);
Chris@10: 			      Tz = FNMS(KP1_801937735, Ty, Te);
Chris@10: 			      R0[WS(rs, 3)] = FMA(KP1_949855824, Tw, Tl);
Chris@10: 			      R0[WS(rs, 4)] = FNMS(KP1_949855824, Tw, Tl);
Chris@10: 			      R1[WS(rs, 5)] = FMA(KP1_949855824, TY, TW);
Chris@10: 			      R1[WS(rs, 1)] = FNMS(KP1_949855824, TY, TW);
Chris@10: 			      R0[WS(rs, 6)] = FMA(KP1_949855824, TB, Tz);
Chris@10: 			      R0[WS(rs, 1)] = FNMS(KP1_949855824, TB, Tz);
Chris@10: 			      TI = FNMS(KP692021471, TH, Tc);
Chris@10: 			 }
Chris@10: 			 TT = FNMS(KP801937735, TS, TL);
Chris@10: 			 TQ = FNMS(KP692021471, TP, T9);
Chris@10: 			 TG = FNMS(KP801937735, TF, Tu);
Chris@10: 			 TJ = FNMS(KP1_801937735, TI, T3);
Chris@10: 			 TD = FNMS(KP692021471, TC, Tf);
Chris@10: 			 TR = FNMS(KP1_801937735, TQ, T3);
Chris@10: 		    }
Chris@10: 	       }
Chris@10: 	       R1[WS(rs, 6)] = FMA(KP1_949855824, TO, TJ);
Chris@10: 	       R1[0] = FNMS(KP1_949855824, TO, TJ);
Chris@10: 	       TE = FNMS(KP1_801937735, TD, Te);
Chris@10: 	       R1[WS(rs, 2)] = FMA(KP1_949855824, TT, TR);
Chris@10: 	       R1[WS(rs, 4)] = FNMS(KP1_949855824, TT, TR);
Chris@10: 	       R0[WS(rs, 2)] = FMA(KP1_949855824, TG, TE);
Chris@10: 	       R0[WS(rs, 5)] = FNMS(KP1_949855824, TG, TE);
Chris@10: 	  }
Chris@10:      }
Chris@10: }
Chris@10: 
Chris@10: static const kr2c_desc desc = { 14, "r2cb_14", {18, 0, 44, 0}, &GENUS };
Chris@10: 
Chris@10: void X(codelet_r2cb_14) (planner *p) {
Chris@10:      X(kr2c_register) (p, r2cb_14, &desc);
Chris@10: }
Chris@10: 
Chris@10: #else				/* HAVE_FMA */
Chris@10: 
Chris@10: /* Generated by: ../../../genfft/gen_r2cb.native -compact -variables 4 -pipeline-latency 4 -sign 1 -n 14 -name r2cb_14 -include r2cb.h */
Chris@10: 
Chris@10: /*
Chris@10:  * This function contains 62 FP additions, 38 FP multiplications,
Chris@10:  * (or, 36 additions, 12 multiplications, 26 fused multiply/add),
Chris@10:  * 28 stack variables, 7 constants, and 28 memory accesses
Chris@10:  */
Chris@10: #include "r2cb.h"
Chris@10: 
Chris@10: static void r2cb_14(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(KP1_801937735, +1.801937735804838252472204639014890102331838324);
Chris@10:      DK(KP445041867, +0.445041867912628808577805128993589518932711138);
Chris@10:      DK(KP1_246979603, +1.246979603717467061050009768008479621264549462);
Chris@10:      DK(KP867767478, +0.867767478235116240951536665696717509219981456);
Chris@10:      DK(KP1_949855824, +1.949855824363647214036263365987862434465571601);
Chris@10:      DK(KP1_563662964, +1.563662964936059617416889053348115500464669037);
Chris@10:      DK(KP2_000000000, +2.000000000000000000000000000000000000000000000);
Chris@10:      {
Chris@10: 	  INT i;
Chris@10: 	  for (i = v; i > 0; i = i - 1, R0 = R0 + ovs, R1 = R1 + ovs, Cr = Cr + ivs, Ci = Ci + ivs, MAKE_VOLATILE_STRIDE(56, rs), MAKE_VOLATILE_STRIDE(56, csr), MAKE_VOLATILE_STRIDE(56, csi)) {
Chris@10: 	       E T3, Td, T6, Te, Tq, Tz, Tn, Ty, Tc, Tg, Tk, Tx, T9, Tf, T1;
Chris@10: 	       E T2;
Chris@10: 	       T1 = Cr[0];
Chris@10: 	       T2 = Cr[WS(csr, 7)];
Chris@10: 	       T3 = T1 - T2;
Chris@10: 	       Td = T1 + T2;
Chris@10: 	       {
Chris@10: 		    E T4, T5, To, Tp;
Chris@10: 		    T4 = Cr[WS(csr, 2)];
Chris@10: 		    T5 = Cr[WS(csr, 5)];
Chris@10: 		    T6 = T4 - T5;
Chris@10: 		    Te = T4 + T5;
Chris@10: 		    To = Ci[WS(csi, 2)];
Chris@10: 		    Tp = Ci[WS(csi, 5)];
Chris@10: 		    Tq = To - Tp;
Chris@10: 		    Tz = To + Tp;
Chris@10: 	       }
Chris@10: 	       {
Chris@10: 		    E Tl, Tm, Ta, Tb;
Chris@10: 		    Tl = Ci[WS(csi, 6)];
Chris@10: 		    Tm = Ci[WS(csi, 1)];
Chris@10: 		    Tn = Tl - Tm;
Chris@10: 		    Ty = Tl + Tm;
Chris@10: 		    Ta = Cr[WS(csr, 6)];
Chris@10: 		    Tb = Cr[WS(csr, 1)];
Chris@10: 		    Tc = Ta - Tb;
Chris@10: 		    Tg = Ta + Tb;
Chris@10: 	       }
Chris@10: 	       {
Chris@10: 		    E Ti, Tj, T7, T8;
Chris@10: 		    Ti = Ci[WS(csi, 4)];
Chris@10: 		    Tj = Ci[WS(csi, 3)];
Chris@10: 		    Tk = Ti - Tj;
Chris@10: 		    Tx = Ti + Tj;
Chris@10: 		    T7 = Cr[WS(csr, 4)];
Chris@10: 		    T8 = Cr[WS(csr, 3)];
Chris@10: 		    T9 = T7 - T8;
Chris@10: 		    Tf = T7 + T8;
Chris@10: 	       }
Chris@10: 	       R1[WS(rs, 3)] = FMA(KP2_000000000, T6 + T9 + Tc, T3);
Chris@10: 	       R0[0] = FMA(KP2_000000000, Te + Tf + Tg, Td);
Chris@10: 	       {
Chris@10: 		    E Tr, Th, TE, TD;
Chris@10: 		    Tr = FNMS(KP1_949855824, Tn, KP1_563662964 * Tk) - (KP867767478 * Tq);
Chris@10: 		    Th = FMA(KP1_246979603, Tf, Td) + FNMA(KP445041867, Tg, KP1_801937735 * Te);
Chris@10: 		    R0[WS(rs, 2)] = Th - Tr;
Chris@10: 		    R0[WS(rs, 5)] = Th + Tr;
Chris@10: 		    TE = FMA(KP867767478, Tx, KP1_563662964 * Ty) - (KP1_949855824 * Tz);
Chris@10: 		    TD = FMA(KP1_246979603, Tc, T3) + FNMA(KP1_801937735, T9, KP445041867 * T6);
Chris@10: 		    R1[WS(rs, 2)] = TD - TE;
Chris@10: 		    R1[WS(rs, 4)] = TD + TE;
Chris@10: 	       }
Chris@10: 	       {
Chris@10: 		    E Tt, Ts, TA, Tw;
Chris@10: 		    Tt = FMA(KP867767478, Tk, KP1_563662964 * Tn) - (KP1_949855824 * Tq);
Chris@10: 		    Ts = FMA(KP1_246979603, Tg, Td) + FNMA(KP1_801937735, Tf, KP445041867 * Te);
Chris@10: 		    R0[WS(rs, 6)] = Ts - Tt;
Chris@10: 		    R0[WS(rs, 1)] = Ts + Tt;
Chris@10: 		    TA = FNMS(KP1_949855824, Ty, KP1_563662964 * Tx) - (KP867767478 * Tz);
Chris@10: 		    Tw = FMA(KP1_246979603, T9, T3) + FNMA(KP445041867, Tc, KP1_801937735 * T6);
Chris@10: 		    R1[WS(rs, 5)] = Tw - TA;
Chris@10: 		    R1[WS(rs, 1)] = Tw + TA;
Chris@10: 	       }
Chris@10: 	       {
Chris@10: 		    E TC, TB, Tv, Tu;
Chris@10: 		    TC = FMA(KP1_563662964, Tz, KP1_949855824 * Tx) + (KP867767478 * Ty);
Chris@10: 		    TB = FMA(KP1_246979603, T6, T3) + FNMA(KP1_801937735, Tc, KP445041867 * T9);
Chris@10: 		    R1[0] = TB - TC;
Chris@10: 		    R1[WS(rs, 6)] = TB + TC;
Chris@10: 		    Tv = FMA(KP1_563662964, Tq, KP1_949855824 * Tk) + (KP867767478 * Tn);
Chris@10: 		    Tu = FMA(KP1_246979603, Te, Td) + FNMA(KP1_801937735, Tg, KP445041867 * Tf);
Chris@10: 		    R0[WS(rs, 4)] = Tu - Tv;
Chris@10: 		    R0[WS(rs, 3)] = Tu + Tv;
Chris@10: 	       }
Chris@10: 	  }
Chris@10:      }
Chris@10: }
Chris@10: 
Chris@10: static const kr2c_desc desc = { 14, "r2cb_14", {36, 12, 26, 0}, &GENUS };
Chris@10: 
Chris@10: void X(codelet_r2cb_14) (planner *p) {
Chris@10:      X(kr2c_register) (p, r2cb_14, &desc);
Chris@10: }
Chris@10: 
Chris@10: #endif				/* HAVE_FMA */