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 07:35:42 EST 2012 */
Chris@10: 
Chris@10: #include "codelet-dft.h"
Chris@10: 
Chris@10: #ifdef HAVE_FMA
Chris@10: 
Chris@10: /* Generated by: ../../../genfft/gen_notw.native -fma -reorder-insns -schedule-for-pipeline -compact -variables 4 -pipeline-latency 4 -n 3 -name n1_3 -include n.h */
Chris@10: 
Chris@10: /*
Chris@10:  * This function contains 12 FP additions, 6 FP multiplications,
Chris@10:  * (or, 6 additions, 0 multiplications, 6 fused multiply/add),
Chris@10:  * 15 stack variables, 2 constants, and 12 memory accesses
Chris@10:  */
Chris@10: #include "n.h"
Chris@10: 
Chris@10: static void n1_3(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs)
Chris@10: {
Chris@10:      DK(KP866025403, +0.866025403784438646763723170752936183471402627);
Chris@10:      DK(KP500000000, +0.500000000000000000000000000000000000000000000);
Chris@10:      {
Chris@10: 	  INT i;
Chris@10: 	  for (i = v; i > 0; i = i - 1, ri = ri + ivs, ii = ii + ivs, ro = ro + ovs, io = io + ovs, MAKE_VOLATILE_STRIDE(12, is), MAKE_VOLATILE_STRIDE(12, os)) {
Chris@10: 	       E T1, T9, T2, T3, T6, T7;
Chris@10: 	       T1 = ri[0];
Chris@10: 	       T9 = ii[0];
Chris@10: 	       T2 = ri[WS(is, 1)];
Chris@10: 	       T3 = ri[WS(is, 2)];
Chris@10: 	       T6 = ii[WS(is, 1)];
Chris@10: 	       T7 = ii[WS(is, 2)];
Chris@10: 	       {
Chris@10: 		    E T4, Tc, T8, Ta, T5, Tb;
Chris@10: 		    T4 = T2 + T3;
Chris@10: 		    Tc = T3 - T2;
Chris@10: 		    T8 = T6 - T7;
Chris@10: 		    Ta = T6 + T7;
Chris@10: 		    T5 = FNMS(KP500000000, T4, T1);
Chris@10: 		    ro[0] = T1 + T4;
Chris@10: 		    Tb = FNMS(KP500000000, Ta, T9);
Chris@10: 		    io[0] = T9 + Ta;
Chris@10: 		    ro[WS(os, 1)] = FMA(KP866025403, T8, T5);
Chris@10: 		    ro[WS(os, 2)] = FNMS(KP866025403, T8, T5);
Chris@10: 		    io[WS(os, 2)] = FNMS(KP866025403, Tc, Tb);
Chris@10: 		    io[WS(os, 1)] = FMA(KP866025403, Tc, Tb);
Chris@10: 	       }
Chris@10: 	  }
Chris@10:      }
Chris@10: }
Chris@10: 
Chris@10: static const kdft_desc desc = { 3, "n1_3", {6, 0, 6, 0}, &GENUS, 0, 0, 0, 0 };
Chris@10: 
Chris@10: void X(codelet_n1_3) (planner *p) {
Chris@10:      X(kdft_register) (p, n1_3, &desc);
Chris@10: }
Chris@10: 
Chris@10: #else				/* HAVE_FMA */
Chris@10: 
Chris@10: /* Generated by: ../../../genfft/gen_notw.native -compact -variables 4 -pipeline-latency 4 -n 3 -name n1_3 -include n.h */
Chris@10: 
Chris@10: /*
Chris@10:  * This function contains 12 FP additions, 4 FP multiplications,
Chris@10:  * (or, 10 additions, 2 multiplications, 2 fused multiply/add),
Chris@10:  * 15 stack variables, 2 constants, and 12 memory accesses
Chris@10:  */
Chris@10: #include "n.h"
Chris@10: 
Chris@10: static void n1_3(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs)
Chris@10: {
Chris@10:      DK(KP500000000, +0.500000000000000000000000000000000000000000000);
Chris@10:      DK(KP866025403, +0.866025403784438646763723170752936183471402627);
Chris@10:      {
Chris@10: 	  INT i;
Chris@10: 	  for (i = v; i > 0; i = i - 1, ri = ri + ivs, ii = ii + ivs, ro = ro + ovs, io = io + ovs, MAKE_VOLATILE_STRIDE(12, is), MAKE_VOLATILE_STRIDE(12, os)) {
Chris@10: 	       E T1, Ta, T4, T9, T8, Tb, T5, Tc;
Chris@10: 	       T1 = ri[0];
Chris@10: 	       Ta = ii[0];
Chris@10: 	       {
Chris@10: 		    E T2, T3, T6, T7;
Chris@10: 		    T2 = ri[WS(is, 1)];
Chris@10: 		    T3 = ri[WS(is, 2)];
Chris@10: 		    T4 = T2 + T3;
Chris@10: 		    T9 = KP866025403 * (T3 - T2);
Chris@10: 		    T6 = ii[WS(is, 1)];
Chris@10: 		    T7 = ii[WS(is, 2)];
Chris@10: 		    T8 = KP866025403 * (T6 - T7);
Chris@10: 		    Tb = T6 + T7;
Chris@10: 	       }
Chris@10: 	       ro[0] = T1 + T4;
Chris@10: 	       io[0] = Ta + Tb;
Chris@10: 	       T5 = FNMS(KP500000000, T4, T1);
Chris@10: 	       ro[WS(os, 2)] = T5 - T8;
Chris@10: 	       ro[WS(os, 1)] = T5 + T8;
Chris@10: 	       Tc = FNMS(KP500000000, Tb, Ta);
Chris@10: 	       io[WS(os, 1)] = T9 + Tc;
Chris@10: 	       io[WS(os, 2)] = Tc - T9;
Chris@10: 	  }
Chris@10:      }
Chris@10: }
Chris@10: 
Chris@10: static const kdft_desc desc = { 3, "n1_3", {10, 2, 2, 0}, &GENUS, 0, 0, 0, 0 };
Chris@10: 
Chris@10: void X(codelet_n1_3) (planner *p) {
Chris@10:      X(kdft_register) (p, n1_3, &desc);
Chris@10: }
Chris@10: 
Chris@10: #endif				/* HAVE_FMA */