cannam@95: /*
cannam@95:  * Copyright (c) 2003, 2007-11 Matteo Frigo
cannam@95:  * Copyright (c) 2003, 2007-11 Massachusetts Institute of Technology
cannam@95:  *
cannam@95:  * This program is free software; you can redistribute it and/or modify
cannam@95:  * it under the terms of the GNU General Public License as published by
cannam@95:  * the Free Software Foundation; either version 2 of the License, or
cannam@95:  * (at your option) any later version.
cannam@95:  *
cannam@95:  * This program is distributed in the hope that it will be useful,
cannam@95:  * but WITHOUT ANY WARRANTY; without even the implied warranty of
cannam@95:  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
cannam@95:  * GNU General Public License for more details.
cannam@95:  *
cannam@95:  * You should have received a copy of the GNU General Public License
cannam@95:  * along with this program; if not, write to the Free Software
cannam@95:  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02110-1301  USA
cannam@95:  *
cannam@95:  */
cannam@95: 
cannam@95: /* This file was automatically generated --- DO NOT EDIT */
cannam@95: /* Generated on Sun Nov 25 07:36:52 EST 2012 */
cannam@95: 
cannam@95: #include "codelet-dft.h"
cannam@95: 
cannam@95: #ifdef HAVE_FMA
cannam@95: 
cannam@95: /* Generated by: ../../../genfft/gen_notw_c.native -fma -reorder-insns -schedule-for-pipeline -simd -compact -variables 4 -pipeline-latency 8 -n 14 -name n1fv_14 -include n1f.h */
cannam@95: 
cannam@95: /*
cannam@95:  * This function contains 74 FP additions, 48 FP multiplications,
cannam@95:  * (or, 32 additions, 6 multiplications, 42 fused multiply/add),
cannam@95:  * 63 stack variables, 6 constants, and 28 memory accesses
cannam@95:  */
cannam@95: #include "n1f.h"
cannam@95: 
cannam@95: static void n1fv_14(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs)
cannam@95: {
cannam@95:      DVK(KP900968867, +0.900968867902419126236102319507445051165919162);
cannam@95:      DVK(KP801937735, +0.801937735804838252472204639014890102331838324);
cannam@95:      DVK(KP974927912, +0.974927912181823607018131682993931217232785801);
cannam@95:      DVK(KP692021471, +0.692021471630095869627814897002069140197260599);
cannam@95:      DVK(KP554958132, +0.554958132087371191422194871006410481067288862);
cannam@95:      DVK(KP356895867, +0.356895867892209443894399510021300583399127187);
cannam@95:      {
cannam@95: 	  INT i;
cannam@95: 	  const R *xi;
cannam@95: 	  R *xo;
cannam@95: 	  xi = ri;
cannam@95: 	  xo = ro;
cannam@95: 	  for (i = v; i > 0; i = i - VL, xi = xi + (VL * ivs), xo = xo + (VL * ovs), MAKE_VOLATILE_STRIDE(28, is), MAKE_VOLATILE_STRIDE(28, os)) {
cannam@95: 	       V TH, T3, TP, Tn, Ta, Ts, TW, TK, TO, Tk, TM, Tg, TL, Td, T1;
cannam@95: 	       V T2;
cannam@95: 	       T1 = LD(&(xi[0]), ivs, &(xi[0]));
cannam@95: 	       T2 = LD(&(xi[WS(is, 7)]), ivs, &(xi[WS(is, 1)]));
cannam@95: 	       {
cannam@95: 		    V Ti, TI, T6, TJ, T9, Tj, Te, Tf, Tb, Tc;
cannam@95: 		    {
cannam@95: 			 V T4, T5, T7, T8, Tl, Tm;
cannam@95: 			 T4 = LD(&(xi[WS(is, 2)]), ivs, &(xi[0]));
cannam@95: 			 T5 = LD(&(xi[WS(is, 9)]), ivs, &(xi[WS(is, 1)]));
cannam@95: 			 T7 = LD(&(xi[WS(is, 12)]), ivs, &(xi[0]));
cannam@95: 			 T8 = LD(&(xi[WS(is, 5)]), ivs, &(xi[WS(is, 1)]));
cannam@95: 			 Tl = LD(&(xi[WS(is, 8)]), ivs, &(xi[0]));
cannam@95: 			 Tm = LD(&(xi[WS(is, 1)]), ivs, &(xi[WS(is, 1)]));
cannam@95: 			 Ti = LD(&(xi[WS(is, 6)]), ivs, &(xi[0]));
cannam@95: 			 TH = VADD(T1, T2);
cannam@95: 			 T3 = VSUB(T1, T2);
cannam@95: 			 TI = VADD(T4, T5);
cannam@95: 			 T6 = VSUB(T4, T5);
cannam@95: 			 TJ = VADD(T7, T8);
cannam@95: 			 T9 = VSUB(T7, T8);
cannam@95: 			 TP = VADD(Tl, Tm);
cannam@95: 			 Tn = VSUB(Tl, Tm);
cannam@95: 			 Tj = LD(&(xi[WS(is, 13)]), ivs, &(xi[WS(is, 1)]));
cannam@95: 			 Te = LD(&(xi[WS(is, 10)]), ivs, &(xi[0]));
cannam@95: 			 Tf = LD(&(xi[WS(is, 3)]), ivs, &(xi[WS(is, 1)]));
cannam@95: 			 Tb = LD(&(xi[WS(is, 4)]), ivs, &(xi[0]));
cannam@95: 			 Tc = LD(&(xi[WS(is, 11)]), ivs, &(xi[WS(is, 1)]));
cannam@95: 		    }
cannam@95: 		    Ta = VADD(T6, T9);
cannam@95: 		    Ts = VSUB(T9, T6);
cannam@95: 		    TW = VSUB(TJ, TI);
cannam@95: 		    TK = VADD(TI, TJ);
cannam@95: 		    TO = VADD(Ti, Tj);
cannam@95: 		    Tk = VSUB(Ti, Tj);
cannam@95: 		    TM = VADD(Te, Tf);
cannam@95: 		    Tg = VSUB(Te, Tf);
cannam@95: 		    TL = VADD(Tb, Tc);
cannam@95: 		    Td = VSUB(Tb, Tc);
cannam@95: 	       }
cannam@95: 	       {
cannam@95: 		    V T18, TB, T13, TY, TG, Tw, T11, Tr, T16, TT, Tz, TE, TU, TQ;
cannam@95: 		    TU = VSUB(TO, TP);
cannam@95: 		    TQ = VADD(TO, TP);
cannam@95: 		    {
cannam@95: 			 V Tt, To, TV, TN;
cannam@95: 			 Tt = VSUB(Tn, Tk);
cannam@95: 			 To = VADD(Tk, Tn);
cannam@95: 			 TV = VSUB(TL, TM);
cannam@95: 			 TN = VADD(TL, TM);
cannam@95: 			 {
cannam@95: 			      V Tu, Th, TZ, T17;
cannam@95: 			      Tu = VSUB(Tg, Td);
cannam@95: 			      Th = VADD(Td, Tg);
cannam@95: 			      TZ = VFNMS(LDK(KP356895867), TK, TQ);
cannam@95: 			      T17 = VFNMS(LDK(KP554958132), TU, TW);
cannam@95: 			      {
cannam@95: 				   V Tp, TA, T14, TR;
cannam@95: 				   Tp = VFNMS(LDK(KP356895867), Ta, To);
cannam@95: 				   TA = VFMA(LDK(KP554958132), Tt, Ts);
cannam@95: 				   ST(&(xo[0]), VADD(TH, VADD(TK, VADD(TN, TQ))), ovs, &(xo[0]));
cannam@95: 				   T14 = VFNMS(LDK(KP356895867), TN, TK);
cannam@95: 				   TR = VFNMS(LDK(KP356895867), TQ, TN);
cannam@95: 				   {
cannam@95: 					V T12, TX, Tx, TC;
cannam@95: 					T12 = VFMA(LDK(KP554958132), TV, TU);
cannam@95: 					TX = VFMA(LDK(KP554958132), TW, TV);
cannam@95: 					ST(&(xo[WS(os, 7)]), VADD(T3, VADD(Ta, VADD(Th, To))), ovs, &(xo[WS(os, 1)]));
cannam@95: 					Tx = VFNMS(LDK(KP356895867), Th, Ta);
cannam@95: 					TC = VFNMS(LDK(KP356895867), To, Th);
cannam@95: 					{
cannam@95: 					     V TF, Tv, T10, Tq;
cannam@95: 					     TF = VFNMS(LDK(KP554958132), Ts, Tu);
cannam@95: 					     Tv = VFMA(LDK(KP554958132), Tu, Tt);
cannam@95: 					     T10 = VFNMS(LDK(KP692021471), TZ, TN);
cannam@95: 					     T18 = VMUL(LDK(KP974927912), VFNMS(LDK(KP801937735), T17, TV));
cannam@95: 					     Tq = VFNMS(LDK(KP692021471), Tp, Th);
cannam@95: 					     TB = VMUL(LDK(KP974927912), VFMA(LDK(KP801937735), TA, Tu));
cannam@95: 					     {
cannam@95: 						  V T15, TS, Ty, TD;
cannam@95: 						  T15 = VFNMS(LDK(KP692021471), T14, TQ);
cannam@95: 						  TS = VFNMS(LDK(KP692021471), TR, TK);
cannam@95: 						  T13 = VMUL(LDK(KP974927912), VFMA(LDK(KP801937735), T12, TW));
cannam@95: 						  TY = VMUL(LDK(KP974927912), VFNMS(LDK(KP801937735), TX, TU));
cannam@95: 						  Ty = VFNMS(LDK(KP692021471), Tx, To);
cannam@95: 						  TD = VFNMS(LDK(KP692021471), TC, Ta);
cannam@95: 						  TG = VMUL(LDK(KP974927912), VFNMS(LDK(KP801937735), TF, Tt));
cannam@95: 						  Tw = VMUL(LDK(KP974927912), VFNMS(LDK(KP801937735), Tv, Ts));
cannam@95: 						  T11 = VFNMS(LDK(KP900968867), T10, TH);
cannam@95: 						  Tr = VFNMS(LDK(KP900968867), Tq, T3);
cannam@95: 						  T16 = VFNMS(LDK(KP900968867), T15, TH);
cannam@95: 						  TT = VFNMS(LDK(KP900968867), TS, TH);
cannam@95: 						  Tz = VFNMS(LDK(KP900968867), Ty, T3);
cannam@95: 						  TE = VFNMS(LDK(KP900968867), TD, T3);
cannam@95: 					     }
cannam@95: 					}
cannam@95: 				   }
cannam@95: 			      }
cannam@95: 			 }
cannam@95: 		    }
cannam@95: 		    ST(&(xo[WS(os, 12)]), VFNMSI(T13, T11), ovs, &(xo[0]));
cannam@95: 		    ST(&(xo[WS(os, 2)]), VFMAI(T13, T11), ovs, &(xo[0]));
cannam@95: 		    ST(&(xo[WS(os, 9)]), VFMAI(Tw, Tr), ovs, &(xo[WS(os, 1)]));
cannam@95: 		    ST(&(xo[WS(os, 5)]), VFNMSI(Tw, Tr), ovs, &(xo[WS(os, 1)]));
cannam@95: 		    ST(&(xo[WS(os, 8)]), VFNMSI(T18, T16), ovs, &(xo[0]));
cannam@95: 		    ST(&(xo[WS(os, 6)]), VFMAI(T18, T16), ovs, &(xo[0]));
cannam@95: 		    ST(&(xo[WS(os, 10)]), VFNMSI(TY, TT), ovs, &(xo[0]));
cannam@95: 		    ST(&(xo[WS(os, 4)]), VFMAI(TY, TT), ovs, &(xo[0]));
cannam@95: 		    ST(&(xo[WS(os, 1)]), VFMAI(TB, Tz), ovs, &(xo[WS(os, 1)]));
cannam@95: 		    ST(&(xo[WS(os, 13)]), VFNMSI(TB, Tz), ovs, &(xo[WS(os, 1)]));
cannam@95: 		    ST(&(xo[WS(os, 3)]), VFMAI(TG, TE), ovs, &(xo[WS(os, 1)]));
cannam@95: 		    ST(&(xo[WS(os, 11)]), VFNMSI(TG, TE), ovs, &(xo[WS(os, 1)]));
cannam@95: 	       }
cannam@95: 	  }
cannam@95:      }
cannam@95:      VLEAVE();
cannam@95: }
cannam@95: 
cannam@95: static const kdft_desc desc = { 14, XSIMD_STRING("n1fv_14"), {32, 6, 42, 0}, &GENUS, 0, 0, 0, 0 };
cannam@95: 
cannam@95: void XSIMD(codelet_n1fv_14) (planner *p) {
cannam@95:      X(kdft_register) (p, n1fv_14, &desc);
cannam@95: }
cannam@95: 
cannam@95: #else				/* HAVE_FMA */
cannam@95: 
cannam@95: /* Generated by: ../../../genfft/gen_notw_c.native -simd -compact -variables 4 -pipeline-latency 8 -n 14 -name n1fv_14 -include n1f.h */
cannam@95: 
cannam@95: /*
cannam@95:  * This function contains 74 FP additions, 36 FP multiplications,
cannam@95:  * (or, 50 additions, 12 multiplications, 24 fused multiply/add),
cannam@95:  * 33 stack variables, 6 constants, and 28 memory accesses
cannam@95:  */
cannam@95: #include "n1f.h"
cannam@95: 
cannam@95: static void n1fv_14(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs)
cannam@95: {
cannam@95:      DVK(KP222520933, +0.222520933956314404288902564496794759466355569);
cannam@95:      DVK(KP900968867, +0.900968867902419126236102319507445051165919162);
cannam@95:      DVK(KP623489801, +0.623489801858733530525004884004239810632274731);
cannam@95:      DVK(KP433883739, +0.433883739117558120475768332848358754609990728);
cannam@95:      DVK(KP781831482, +0.781831482468029808708444526674057750232334519);
cannam@95:      DVK(KP974927912, +0.974927912181823607018131682993931217232785801);
cannam@95:      {
cannam@95: 	  INT i;
cannam@95: 	  const R *xi;
cannam@95: 	  R *xo;
cannam@95: 	  xi = ri;
cannam@95: 	  xo = ro;
cannam@95: 	  for (i = v; i > 0; i = i - VL, xi = xi + (VL * ivs), xo = xo + (VL * ovs), MAKE_VOLATILE_STRIDE(28, is), MAKE_VOLATILE_STRIDE(28, os)) {
cannam@95: 	       V T3, Ty, To, TK, Tr, TE, Ta, TJ, Tq, TB, Th, TL, Ts, TH, T1;
cannam@95: 	       V T2;
cannam@95: 	       T1 = LD(&(xi[0]), ivs, &(xi[0]));
cannam@95: 	       T2 = LD(&(xi[WS(is, 7)]), ivs, &(xi[WS(is, 1)]));
cannam@95: 	       T3 = VSUB(T1, T2);
cannam@95: 	       Ty = VADD(T1, T2);
cannam@95: 	       {
cannam@95: 		    V Tk, TC, Tn, TD;
cannam@95: 		    {
cannam@95: 			 V Ti, Tj, Tl, Tm;
cannam@95: 			 Ti = LD(&(xi[WS(is, 6)]), ivs, &(xi[0]));
cannam@95: 			 Tj = LD(&(xi[WS(is, 13)]), ivs, &(xi[WS(is, 1)]));
cannam@95: 			 Tk = VSUB(Ti, Tj);
cannam@95: 			 TC = VADD(Ti, Tj);
cannam@95: 			 Tl = LD(&(xi[WS(is, 8)]), ivs, &(xi[0]));
cannam@95: 			 Tm = LD(&(xi[WS(is, 1)]), ivs, &(xi[WS(is, 1)]));
cannam@95: 			 Tn = VSUB(Tl, Tm);
cannam@95: 			 TD = VADD(Tl, Tm);
cannam@95: 		    }
cannam@95: 		    To = VADD(Tk, Tn);
cannam@95: 		    TK = VSUB(TC, TD);
cannam@95: 		    Tr = VSUB(Tn, Tk);
cannam@95: 		    TE = VADD(TC, TD);
cannam@95: 	       }
cannam@95: 	       {
cannam@95: 		    V T6, Tz, T9, TA;
cannam@95: 		    {
cannam@95: 			 V T4, T5, T7, T8;
cannam@95: 			 T4 = LD(&(xi[WS(is, 2)]), ivs, &(xi[0]));
cannam@95: 			 T5 = LD(&(xi[WS(is, 9)]), ivs, &(xi[WS(is, 1)]));
cannam@95: 			 T6 = VSUB(T4, T5);
cannam@95: 			 Tz = VADD(T4, T5);
cannam@95: 			 T7 = LD(&(xi[WS(is, 12)]), ivs, &(xi[0]));
cannam@95: 			 T8 = LD(&(xi[WS(is, 5)]), ivs, &(xi[WS(is, 1)]));
cannam@95: 			 T9 = VSUB(T7, T8);
cannam@95: 			 TA = VADD(T7, T8);
cannam@95: 		    }
cannam@95: 		    Ta = VADD(T6, T9);
cannam@95: 		    TJ = VSUB(TA, Tz);
cannam@95: 		    Tq = VSUB(T9, T6);
cannam@95: 		    TB = VADD(Tz, TA);
cannam@95: 	       }
cannam@95: 	       {
cannam@95: 		    V Td, TF, Tg, TG;
cannam@95: 		    {
cannam@95: 			 V Tb, Tc, Te, Tf;
cannam@95: 			 Tb = LD(&(xi[WS(is, 4)]), ivs, &(xi[0]));
cannam@95: 			 Tc = LD(&(xi[WS(is, 11)]), ivs, &(xi[WS(is, 1)]));
cannam@95: 			 Td = VSUB(Tb, Tc);
cannam@95: 			 TF = VADD(Tb, Tc);
cannam@95: 			 Te = LD(&(xi[WS(is, 10)]), ivs, &(xi[0]));
cannam@95: 			 Tf = LD(&(xi[WS(is, 3)]), ivs, &(xi[WS(is, 1)]));
cannam@95: 			 Tg = VSUB(Te, Tf);
cannam@95: 			 TG = VADD(Te, Tf);
cannam@95: 		    }
cannam@95: 		    Th = VADD(Td, Tg);
cannam@95: 		    TL = VSUB(TF, TG);
cannam@95: 		    Ts = VSUB(Tg, Td);
cannam@95: 		    TH = VADD(TF, TG);
cannam@95: 	       }
cannam@95: 	       ST(&(xo[WS(os, 7)]), VADD(T3, VADD(Ta, VADD(Th, To))), ovs, &(xo[WS(os, 1)]));
cannam@95: 	       ST(&(xo[0]), VADD(Ty, VADD(TB, VADD(TH, TE))), ovs, &(xo[0]));
cannam@95: 	       {
cannam@95: 		    V Tt, Tp, TP, TQ;
cannam@95: 		    Tt = VBYI(VFNMS(LDK(KP781831482), Tr, VFNMS(LDK(KP433883739), Ts, VMUL(LDK(KP974927912), Tq))));
cannam@95: 		    Tp = VFMA(LDK(KP623489801), To, VFNMS(LDK(KP900968867), Th, VFNMS(LDK(KP222520933), Ta, T3)));
cannam@95: 		    ST(&(xo[WS(os, 5)]), VSUB(Tp, Tt), ovs, &(xo[WS(os, 1)]));
cannam@95: 		    ST(&(xo[WS(os, 9)]), VADD(Tp, Tt), ovs, &(xo[WS(os, 1)]));
cannam@95: 		    TP = VBYI(VFMA(LDK(KP974927912), TJ, VFMA(LDK(KP433883739), TL, VMUL(LDK(KP781831482), TK))));
cannam@95: 		    TQ = VFMA(LDK(KP623489801), TE, VFNMS(LDK(KP900968867), TH, VFNMS(LDK(KP222520933), TB, Ty)));
cannam@95: 		    ST(&(xo[WS(os, 2)]), VADD(TP, TQ), ovs, &(xo[0]));
cannam@95: 		    ST(&(xo[WS(os, 12)]), VSUB(TQ, TP), ovs, &(xo[0]));
cannam@95: 	       }
cannam@95: 	       {
cannam@95: 		    V Tv, Tu, TM, TI;
cannam@95: 		    Tv = VBYI(VFMA(LDK(KP781831482), Tq, VFMA(LDK(KP974927912), Ts, VMUL(LDK(KP433883739), Tr))));
cannam@95: 		    Tu = VFMA(LDK(KP623489801), Ta, VFNMS(LDK(KP900968867), To, VFNMS(LDK(KP222520933), Th, T3)));
cannam@95: 		    ST(&(xo[WS(os, 13)]), VSUB(Tu, Tv), ovs, &(xo[WS(os, 1)]));
cannam@95: 		    ST(&(xo[WS(os, 1)]), VADD(Tu, Tv), ovs, &(xo[WS(os, 1)]));
cannam@95: 		    TM = VBYI(VFNMS(LDK(KP433883739), TK, VFNMS(LDK(KP974927912), TL, VMUL(LDK(KP781831482), TJ))));
cannam@95: 		    TI = VFMA(LDK(KP623489801), TB, VFNMS(LDK(KP900968867), TE, VFNMS(LDK(KP222520933), TH, Ty)));
cannam@95: 		    ST(&(xo[WS(os, 6)]), VSUB(TI, TM), ovs, &(xo[0]));
cannam@95: 		    ST(&(xo[WS(os, 8)]), VADD(TM, TI), ovs, &(xo[0]));
cannam@95: 	       }
cannam@95: 	       {
cannam@95: 		    V TO, TN, Tx, Tw;
cannam@95: 		    TO = VBYI(VFMA(LDK(KP433883739), TJ, VFNMS(LDK(KP974927912), TK, VMUL(LDK(KP781831482), TL))));
cannam@95: 		    TN = VFMA(LDK(KP623489801), TH, VFNMS(LDK(KP222520933), TE, VFNMS(LDK(KP900968867), TB, Ty)));
cannam@95: 		    ST(&(xo[WS(os, 4)]), VSUB(TN, TO), ovs, &(xo[0]));
cannam@95: 		    ST(&(xo[WS(os, 10)]), VADD(TO, TN), ovs, &(xo[0]));
cannam@95: 		    Tx = VBYI(VFMA(LDK(KP433883739), Tq, VFNMS(LDK(KP781831482), Ts, VMUL(LDK(KP974927912), Tr))));
cannam@95: 		    Tw = VFMA(LDK(KP623489801), Th, VFNMS(LDK(KP222520933), To, VFNMS(LDK(KP900968867), Ta, T3)));
cannam@95: 		    ST(&(xo[WS(os, 11)]), VSUB(Tw, Tx), ovs, &(xo[WS(os, 1)]));
cannam@95: 		    ST(&(xo[WS(os, 3)]), VADD(Tw, Tx), ovs, &(xo[WS(os, 1)]));
cannam@95: 	       }
cannam@95: 	  }
cannam@95:      }
cannam@95:      VLEAVE();
cannam@95: }
cannam@95: 
cannam@95: static const kdft_desc desc = { 14, XSIMD_STRING("n1fv_14"), {50, 12, 24, 0}, &GENUS, 0, 0, 0, 0 };
cannam@95: 
cannam@95: void XSIMD(codelet_n1fv_14) (planner *p) {
cannam@95:      X(kdft_register) (p, n1fv_14, &desc);
cannam@95: }
cannam@95: 
cannam@95: #endif				/* HAVE_FMA */