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 07:41:07 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 5 -name r2cb_5 -include r2cb.h */
Chris@10: 
Chris@10: /*
Chris@10:  * This function contains 12 FP additions, 10 FP multiplications,
Chris@10:  * (or, 2 additions, 0 multiplications, 10 fused multiply/add),
Chris@10:  * 18 stack variables, 5 constants, and 10 memory accesses
Chris@10:  */
Chris@10: #include "r2cb.h"
Chris@10: 
Chris@10: static void r2cb_5(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_902113032, +1.902113032590307144232878666758764286811397268);
Chris@10:      DK(KP1_118033988, +1.118033988749894848204586834365638117720309180);
Chris@10:      DK(KP500000000, +0.500000000000000000000000000000000000000000000);
Chris@10:      DK(KP2_000000000, +2.000000000000000000000000000000000000000000000);
Chris@10:      DK(KP618033988, +0.618033988749894848204586834365638117720309180);
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(20, rs), MAKE_VOLATILE_STRIDE(20, csr), MAKE_VOLATILE_STRIDE(20, csi)) {
Chris@10: 	       E T1, T2, T3, Tc, Ta, T8, T9;
Chris@10: 	       T8 = Ci[WS(csi, 1)];
Chris@10: 	       T9 = Ci[WS(csi, 2)];
Chris@10: 	       T1 = Cr[0];
Chris@10: 	       T2 = Cr[WS(csr, 1)];
Chris@10: 	       T3 = Cr[WS(csr, 2)];
Chris@10: 	       Tc = FMS(KP618033988, T8, T9);
Chris@10: 	       Ta = FMA(KP618033988, T9, T8);
Chris@10: 	       {
Chris@10: 		    E T6, T4, T5, T7, Tb;
Chris@10: 		    T6 = T2 - T3;
Chris@10: 		    T4 = T2 + T3;
Chris@10: 		    R0[0] = FMA(KP2_000000000, T4, T1);
Chris@10: 		    T5 = FNMS(KP500000000, T4, T1);
Chris@10: 		    T7 = FMA(KP1_118033988, T6, T5);
Chris@10: 		    Tb = FNMS(KP1_118033988, T6, T5);
Chris@10: 		    R0[WS(rs, 2)] = FMA(KP1_902113032, Ta, T7);
Chris@10: 		    R1[0] = FNMS(KP1_902113032, Ta, T7);
Chris@10: 		    R1[WS(rs, 1)] = FMA(KP1_902113032, Tc, Tb);
Chris@10: 		    R0[WS(rs, 1)] = FNMS(KP1_902113032, Tc, Tb);
Chris@10: 	       }
Chris@10: 	  }
Chris@10:      }
Chris@10: }
Chris@10: 
Chris@10: static const kr2c_desc desc = { 5, "r2cb_5", {2, 0, 10, 0}, &GENUS };
Chris@10: 
Chris@10: void X(codelet_r2cb_5) (planner *p) {
Chris@10:      X(kr2c_register) (p, r2cb_5, &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 5 -name r2cb_5 -include r2cb.h */
Chris@10: 
Chris@10: /*
Chris@10:  * This function contains 12 FP additions, 7 FP multiplications,
Chris@10:  * (or, 8 additions, 3 multiplications, 4 fused multiply/add),
Chris@10:  * 18 stack variables, 5 constants, and 10 memory accesses
Chris@10:  */
Chris@10: #include "r2cb.h"
Chris@10: 
Chris@10: static void r2cb_5(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(KP2_000000000, +2.000000000000000000000000000000000000000000000);
Chris@10:      DK(KP1_118033988, +1.118033988749894848204586834365638117720309180);
Chris@10:      DK(KP500000000, +0.500000000000000000000000000000000000000000000);
Chris@10:      DK(KP1_902113032, +1.902113032590307144232878666758764286811397268);
Chris@10:      DK(KP1_175570504, +1.175570504584946258337411909278145537195304875);
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(20, rs), MAKE_VOLATILE_STRIDE(20, csr), MAKE_VOLATILE_STRIDE(20, csi)) {
Chris@10: 	       E Ta, Tc, T1, T4, T5, T6, Tb, T7;
Chris@10: 	       {
Chris@10: 		    E T8, T9, T2, T3;
Chris@10: 		    T8 = Ci[WS(csi, 1)];
Chris@10: 		    T9 = Ci[WS(csi, 2)];
Chris@10: 		    Ta = FNMS(KP1_902113032, T9, KP1_175570504 * T8);
Chris@10: 		    Tc = FMA(KP1_902113032, T8, KP1_175570504 * T9);
Chris@10: 		    T1 = Cr[0];
Chris@10: 		    T2 = Cr[WS(csr, 1)];
Chris@10: 		    T3 = Cr[WS(csr, 2)];
Chris@10: 		    T4 = T2 + T3;
Chris@10: 		    T5 = FNMS(KP500000000, T4, T1);
Chris@10: 		    T6 = KP1_118033988 * (T2 - T3);
Chris@10: 	       }
Chris@10: 	       R0[0] = FMA(KP2_000000000, T4, T1);
Chris@10: 	       Tb = T6 + T5;
Chris@10: 	       R1[0] = Tb - Tc;
Chris@10: 	       R0[WS(rs, 2)] = Tb + Tc;
Chris@10: 	       T7 = T5 - T6;
Chris@10: 	       R0[WS(rs, 1)] = T7 - Ta;
Chris@10: 	       R1[WS(rs, 1)] = T7 + Ta;
Chris@10: 	  }
Chris@10:      }
Chris@10: }
Chris@10: 
Chris@10: static const kr2c_desc desc = { 5, "r2cb_5", {8, 3, 4, 0}, &GENUS };
Chris@10: 
Chris@10: void X(codelet_r2cb_5) (planner *p) {
Chris@10:      X(kr2c_register) (p, r2cb_5, &desc);
Chris@10: }
Chris@10: 
Chris@10: #endif				/* HAVE_FMA */