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 07:39:49 EST 2012 */
cannam@95: 
cannam@95: #include "codelet-rdft.h"
cannam@95: 
cannam@95: #ifdef HAVE_FMA
cannam@95: 
cannam@95: /* Generated by: ../../../genfft/gen_hc2hc.native -fma -reorder-insns -schedule-for-pipeline -compact -variables 4 -pipeline-latency 4 -n 2 -dit -name hf_2 -include hf.h */
cannam@95: 
cannam@95: /*
cannam@95:  * This function contains 6 FP additions, 4 FP multiplications,
cannam@95:  * (or, 4 additions, 2 multiplications, 2 fused multiply/add),
cannam@95:  * 11 stack variables, 0 constants, and 8 memory accesses
cannam@95:  */
cannam@95: #include "hf.h"
cannam@95: 
cannam@95: static void hf_2(R *cr, R *ci, const R *W, stride rs, INT mb, INT me, INT ms)
cannam@95: {
cannam@95:      {
cannam@95: 	  INT m;
cannam@95: 	  for (m = mb, W = W + ((mb - 1) * 2); m < me; m = m + 1, cr = cr + ms, ci = ci - ms, W = W + 2, MAKE_VOLATILE_STRIDE(4, rs)) {
cannam@95: 	       E T1, Ta, T3, T6, T2, T5;
cannam@95: 	       T1 = cr[0];
cannam@95: 	       Ta = ci[0];
cannam@95: 	       T3 = cr[WS(rs, 1)];
cannam@95: 	       T6 = ci[WS(rs, 1)];
cannam@95: 	       T2 = W[0];
cannam@95: 	       T5 = W[1];
cannam@95: 	       {
cannam@95: 		    E T8, T4, T9, T7;
cannam@95: 		    T8 = T2 * T6;
cannam@95: 		    T4 = T2 * T3;
cannam@95: 		    T9 = FNMS(T5, T3, T8);
cannam@95: 		    T7 = FMA(T5, T6, T4);
cannam@95: 		    ci[WS(rs, 1)] = T9 + Ta;
cannam@95: 		    cr[WS(rs, 1)] = T9 - Ta;
cannam@95: 		    cr[0] = T1 + T7;
cannam@95: 		    ci[0] = T1 - T7;
cannam@95: 	       }
cannam@95: 	  }
cannam@95:      }
cannam@95: }
cannam@95: 
cannam@95: static const tw_instr twinstr[] = {
cannam@95:      {TW_FULL, 1, 2},
cannam@95:      {TW_NEXT, 1, 0}
cannam@95: };
cannam@95: 
cannam@95: static const hc2hc_desc desc = { 2, "hf_2", twinstr, &GENUS, {4, 2, 2, 0} };
cannam@95: 
cannam@95: void X(codelet_hf_2) (planner *p) {
cannam@95:      X(khc2hc_register) (p, hf_2, &desc);
cannam@95: }
cannam@95: #else				/* HAVE_FMA */
cannam@95: 
cannam@95: /* Generated by: ../../../genfft/gen_hc2hc.native -compact -variables 4 -pipeline-latency 4 -n 2 -dit -name hf_2 -include hf.h */
cannam@95: 
cannam@95: /*
cannam@95:  * This function contains 6 FP additions, 4 FP multiplications,
cannam@95:  * (or, 4 additions, 2 multiplications, 2 fused multiply/add),
cannam@95:  * 9 stack variables, 0 constants, and 8 memory accesses
cannam@95:  */
cannam@95: #include "hf.h"
cannam@95: 
cannam@95: static void hf_2(R *cr, R *ci, const R *W, stride rs, INT mb, INT me, INT ms)
cannam@95: {
cannam@95:      {
cannam@95: 	  INT m;
cannam@95: 	  for (m = mb, W = W + ((mb - 1) * 2); m < me; m = m + 1, cr = cr + ms, ci = ci - ms, W = W + 2, MAKE_VOLATILE_STRIDE(4, rs)) {
cannam@95: 	       E T1, T8, T6, T7;
cannam@95: 	       T1 = cr[0];
cannam@95: 	       T8 = ci[0];
cannam@95: 	       {
cannam@95: 		    E T3, T5, T2, T4;
cannam@95: 		    T3 = cr[WS(rs, 1)];
cannam@95: 		    T5 = ci[WS(rs, 1)];
cannam@95: 		    T2 = W[0];
cannam@95: 		    T4 = W[1];
cannam@95: 		    T6 = FMA(T2, T3, T4 * T5);
cannam@95: 		    T7 = FNMS(T4, T3, T2 * T5);
cannam@95: 	       }
cannam@95: 	       ci[0] = T1 - T6;
cannam@95: 	       cr[0] = T1 + T6;
cannam@95: 	       cr[WS(rs, 1)] = T7 - T8;
cannam@95: 	       ci[WS(rs, 1)] = T7 + T8;
cannam@95: 	  }
cannam@95:      }
cannam@95: }
cannam@95: 
cannam@95: static const tw_instr twinstr[] = {
cannam@95:      {TW_FULL, 1, 2},
cannam@95:      {TW_NEXT, 1, 0}
cannam@95: };
cannam@95: 
cannam@95: static const hc2hc_desc desc = { 2, "hf_2", twinstr, &GENUS, {4, 2, 2, 0} };
cannam@95: 
cannam@95: void X(codelet_hf_2) (planner *p) {
cannam@95:      X(khc2hc_register) (p, hf_2, &desc);
cannam@95: }
cannam@95: #endif				/* HAVE_FMA */