Chris@82: /*
Chris@82:  * Copyright (c) 2003, 2007-14 Matteo Frigo
Chris@82:  * Copyright (c) 2003, 2007-14 Massachusetts Institute of Technology
Chris@82:  *
Chris@82:  * This program is free software; you can redistribute it and/or modify
Chris@82:  * it under the terms of the GNU General Public License as published by
Chris@82:  * the Free Software Foundation; either version 2 of the License, or
Chris@82:  * (at your option) any later version.
Chris@82:  *
Chris@82:  * This program is distributed in the hope that it will be useful,
Chris@82:  * but WITHOUT ANY WARRANTY; without even the implied warranty of
Chris@82:  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
Chris@82:  * GNU General Public License for more details.
Chris@82:  *
Chris@82:  * You should have received a copy of the GNU General Public License
Chris@82:  * along with this program; if not, write to the Free Software
Chris@82:  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02110-1301  USA
Chris@82:  *
Chris@82:  */
Chris@82: 
Chris@82: /* This file was automatically generated --- DO NOT EDIT */
Chris@82: /* Generated on Thu May 24 08:07:57 EDT 2018 */
Chris@82: 
Chris@82: #include "rdft/codelet-rdft.h"
Chris@82: 
Chris@82: #if defined(ARCH_PREFERS_FMA) || defined(ISA_EXTENSION_PREFERS_FMA)
Chris@82: 
Chris@82: /* Generated by: ../../../genfft/gen_hc2cdft.native -fma -compact -variables 4 -pipeline-latency 4 -sign 1 -n 2 -dif -name hc2cbdft_2 -include rdft/scalar/hc2cb.h */
Chris@82: 
Chris@82: /*
Chris@82:  * This function contains 10 FP additions, 4 FP multiplications,
Chris@82:  * (or, 8 additions, 2 multiplications, 2 fused multiply/add),
Chris@82:  * 15 stack variables, 0 constants, and 8 memory accesses
Chris@82:  */
Chris@82: #include "rdft/scalar/hc2cb.h"
Chris@82: 
Chris@82: static void hc2cbdft_2(R *Rp, R *Ip, R *Rm, R *Im, const R *W, stride rs, INT mb, INT me, INT ms)
Chris@82: {
Chris@82:      {
Chris@82: 	  INT m;
Chris@82: 	  for (m = mb, W = W + ((mb - 1) * 2); m < me; m = m + 1, Rp = Rp + ms, Ip = Ip + ms, Rm = Rm - ms, Im = Im - ms, W = W + 2, MAKE_VOLATILE_STRIDE(8, rs)) {
Chris@82: 	       E T3, Ta, Tc, T9, Td, T4, T8, Tb, Te;
Chris@82: 	       {
Chris@82: 		    E T1, T2, T5, T6, T7;
Chris@82: 		    T1 = Ip[0];
Chris@82: 		    T2 = Im[0];
Chris@82: 		    T3 = T1 - T2;
Chris@82: 		    Ta = T1 + T2;
Chris@82: 		    T5 = Rp[0];
Chris@82: 		    T6 = Rm[0];
Chris@82: 		    T7 = T5 - T6;
Chris@82: 		    Tc = T5 + T6;
Chris@82: 		    T9 = W[1];
Chris@82: 		    Td = T9 * T7;
Chris@82: 		    T4 = W[0];
Chris@82: 		    T8 = T4 * T7;
Chris@82: 	       }
Chris@82: 	       Tb = FNMS(T9, Ta, T8);
Chris@82: 	       Ip[0] = T3 + Tb;
Chris@82: 	       Im[0] = Tb - T3;
Chris@82: 	       Te = FMA(T4, Ta, Td);
Chris@82: 	       Rp[0] = Tc - Te;
Chris@82: 	       Rm[0] = Tc + Te;
Chris@82: 	  }
Chris@82:      }
Chris@82: }
Chris@82: 
Chris@82: static const tw_instr twinstr[] = {
Chris@82:      {TW_FULL, 1, 2},
Chris@82:      {TW_NEXT, 1, 0}
Chris@82: };
Chris@82: 
Chris@82: static const hc2c_desc desc = { 2, "hc2cbdft_2", twinstr, &GENUS, {8, 2, 2, 0} };
Chris@82: 
Chris@82: void X(codelet_hc2cbdft_2) (planner *p) {
Chris@82:      X(khc2c_register) (p, hc2cbdft_2, &desc, HC2C_VIA_DFT);
Chris@82: }
Chris@82: #else
Chris@82: 
Chris@82: /* Generated by: ../../../genfft/gen_hc2cdft.native -compact -variables 4 -pipeline-latency 4 -sign 1 -n 2 -dif -name hc2cbdft_2 -include rdft/scalar/hc2cb.h */
Chris@82: 
Chris@82: /*
Chris@82:  * This function contains 10 FP additions, 4 FP multiplications,
Chris@82:  * (or, 8 additions, 2 multiplications, 2 fused multiply/add),
Chris@82:  * 9 stack variables, 0 constants, and 8 memory accesses
Chris@82:  */
Chris@82: #include "rdft/scalar/hc2cb.h"
Chris@82: 
Chris@82: static void hc2cbdft_2(R *Rp, R *Ip, R *Rm, R *Im, const R *W, stride rs, INT mb, INT me, INT ms)
Chris@82: {
Chris@82:      {
Chris@82: 	  INT m;
Chris@82: 	  for (m = mb, W = W + ((mb - 1) * 2); m < me; m = m + 1, Rp = Rp + ms, Ip = Ip + ms, Rm = Rm - ms, Im = Im - ms, W = W + 2, MAKE_VOLATILE_STRIDE(8, rs)) {
Chris@82: 	       E T3, T9, T7, Tb;
Chris@82: 	       {
Chris@82: 		    E T1, T2, T5, T6;
Chris@82: 		    T1 = Ip[0];
Chris@82: 		    T2 = Im[0];
Chris@82: 		    T3 = T1 - T2;
Chris@82: 		    T9 = T1 + T2;
Chris@82: 		    T5 = Rp[0];
Chris@82: 		    T6 = Rm[0];
Chris@82: 		    T7 = T5 - T6;
Chris@82: 		    Tb = T5 + T6;
Chris@82: 	       }
Chris@82: 	       {
Chris@82: 		    E Ta, Tc, T4, T8;
Chris@82: 		    T4 = W[0];
Chris@82: 		    T8 = W[1];
Chris@82: 		    Ta = FNMS(T8, T9, T4 * T7);
Chris@82: 		    Tc = FMA(T8, T7, T4 * T9);
Chris@82: 		    Ip[0] = T3 + Ta;
Chris@82: 		    Rp[0] = Tb - Tc;
Chris@82: 		    Im[0] = Ta - T3;
Chris@82: 		    Rm[0] = Tb + Tc;
Chris@82: 	       }
Chris@82: 	  }
Chris@82:      }
Chris@82: }
Chris@82: 
Chris@82: static const tw_instr twinstr[] = {
Chris@82:      {TW_FULL, 1, 2},
Chris@82:      {TW_NEXT, 1, 0}
Chris@82: };
Chris@82: 
Chris@82: static const hc2c_desc desc = { 2, "hc2cbdft_2", twinstr, &GENUS, {8, 2, 2, 0} };
Chris@82: 
Chris@82: void X(codelet_hc2cbdft_2) (planner *p) {
Chris@82:      X(khc2c_register) (p, hc2cbdft_2, &desc, HC2C_VIA_DFT);
Chris@82: }
Chris@82: #endif