Chris@42: /*
Chris@42:  * Copyright (c) 2003, 2007-14 Matteo Frigo
Chris@42:  * Copyright (c) 2003, 2007-14 Massachusetts Institute of Technology
Chris@42:  *
Chris@42:  * This program is free software; you can redistribute it and/or modify
Chris@42:  * it under the terms of the GNU General Public License as published by
Chris@42:  * the Free Software Foundation; either version 2 of the License, or
Chris@42:  * (at your option) any later version.
Chris@42:  *
Chris@42:  * This program is distributed in the hope that it will be useful,
Chris@42:  * but WITHOUT ANY WARRANTY; without even the implied warranty of
Chris@42:  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
Chris@42:  * GNU General Public License for more details.
Chris@42:  *
Chris@42:  * You should have received a copy of the GNU General Public License
Chris@42:  * along with this program; if not, write to the Free Software
Chris@42:  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02110-1301  USA
Chris@42:  *
Chris@42:  */
Chris@42: 
Chris@42: /* This file was automatically generated --- DO NOT EDIT */
Chris@42: /* Generated on Sat Jul 30 16:49:26 EDT 2016 */
Chris@42: 
Chris@42: #include "codelet-rdft.h"
Chris@42: 
Chris@42: #ifdef HAVE_FMA
Chris@42: 
Chris@42: /* Generated by: ../../../genfft/gen_r2cb.native -fma -reorder-insns -schedule-for-pipeline -compact -variables 4 -pipeline-latency 4 -sign 1 -n 9 -name r2cb_9 -include r2cb.h */
Chris@42: 
Chris@42: /*
Chris@42:  * This function contains 32 FP additions, 24 FP multiplications,
Chris@42:  * (or, 8 additions, 0 multiplications, 24 fused multiply/add),
Chris@42:  * 40 stack variables, 12 constants, and 18 memory accesses
Chris@42:  */
Chris@42: #include "r2cb.h"
Chris@42: 
Chris@42: static void r2cb_9(R *R0, R *R1, R *Cr, R *Ci, stride rs, stride csr, stride csi, INT v, INT ivs, INT ovs)
Chris@42: {
Chris@42:      DK(KP1_326827896, +1.326827896337876792410842639271782594433726619);
Chris@42:      DK(KP1_705737063, +1.705737063904886419256501927880148143872040591);
Chris@42:      DK(KP766044443, +0.766044443118978035202392650555416673935832457);
Chris@42:      DK(KP1_532088886, +1.532088886237956070404785301110833347871664914);
Chris@42:      DK(KP984807753, +0.984807753012208059366743024589523013670643252);
Chris@42:      DK(KP1_969615506, +1.969615506024416118733486049179046027341286503);
Chris@42:      DK(KP839099631, +0.839099631177280011763127298123181364687434283);
Chris@42:      DK(KP176326980, +0.176326980708464973471090386868618986121633062);
Chris@42:      DK(KP866025403, +0.866025403784438646763723170752936183471402627);
Chris@42:      DK(KP500000000, +0.500000000000000000000000000000000000000000000);
Chris@42:      DK(KP1_732050807, +1.732050807568877293527446341505872366942805254);
Chris@42:      DK(KP2_000000000, +2.000000000000000000000000000000000000000000000);
Chris@42:      {
Chris@42: 	  INT i;
Chris@42: 	  for (i = v; i > 0; i = i - 1, R0 = R0 + ovs, R1 = R1 + ovs, Cr = Cr + ivs, Ci = Ci + ivs, MAKE_VOLATILE_STRIDE(36, rs), MAKE_VOLATILE_STRIDE(36, csr), MAKE_VOLATILE_STRIDE(36, csi)) {
Chris@42: 	       E T4, Th, T3, Tb, Tp, Tk, T7, Tf, Ti, Ta, T1, T2;
Chris@42: 	       Ta = Ci[WS(csi, 3)];
Chris@42: 	       T1 = Cr[0];
Chris@42: 	       T2 = Cr[WS(csr, 3)];
Chris@42: 	       T4 = Cr[WS(csr, 1)];
Chris@42: 	       Th = Ci[WS(csi, 1)];
Chris@42: 	       {
Chris@42: 		    E T5, T9, T6, Td, Te;
Chris@42: 		    T5 = Cr[WS(csr, 4)];
Chris@42: 		    T9 = T1 - T2;
Chris@42: 		    T3 = FMA(KP2_000000000, T2, T1);
Chris@42: 		    T6 = Cr[WS(csr, 2)];
Chris@42: 		    Td = Ci[WS(csi, 4)];
Chris@42: 		    Te = Ci[WS(csi, 2)];
Chris@42: 		    Tb = FNMS(KP1_732050807, Ta, T9);
Chris@42: 		    Tp = FMA(KP1_732050807, Ta, T9);
Chris@42: 		    Tk = T6 - T5;
Chris@42: 		    T7 = T5 + T6;
Chris@42: 		    Tf = Td + Te;
Chris@42: 		    Ti = Td - Te;
Chris@42: 	       }
Chris@42: 	       {
Chris@42: 		    E Tu, To, Tt, Tn, Tc, T8;
Chris@42: 		    Tc = FNMS(KP500000000, T7, T4);
Chris@42: 		    T8 = T4 + T7;
Chris@42: 		    {
Chris@42: 			 E Tw, Tj, Tr, Tg, Tv;
Chris@42: 			 Tw = Ti + Th;
Chris@42: 			 Tj = FNMS(KP500000000, Ti, Th);
Chris@42: 			 Tr = FMA(KP866025403, Tf, Tc);
Chris@42: 			 Tg = FNMS(KP866025403, Tf, Tc);
Chris@42: 			 Tv = T3 - T8;
Chris@42: 			 R0[0] = FMA(KP2_000000000, T8, T3);
Chris@42: 			 {
Chris@42: 			      E Tq, Tl, Ts, Tm;
Chris@42: 			      Tq = FMA(KP866025403, Tk, Tj);
Chris@42: 			      Tl = FNMS(KP866025403, Tk, Tj);
Chris@42: 			      R0[WS(rs, 3)] = FMA(KP1_732050807, Tw, Tv);
Chris@42: 			      R1[WS(rs, 1)] = FNMS(KP1_732050807, Tw, Tv);
Chris@42: 			      Ts = FNMS(KP176326980, Tr, Tq);
Chris@42: 			      Tu = FMA(KP176326980, Tq, Tr);
Chris@42: 			      Tm = FNMS(KP839099631, Tl, Tg);
Chris@42: 			      To = FMA(KP839099631, Tg, Tl);
Chris@42: 			      R0[WS(rs, 1)] = FNMS(KP1_969615506, Ts, Tp);
Chris@42: 			      Tt = FMA(KP984807753, Ts, Tp);
Chris@42: 			      R1[0] = FMA(KP1_532088886, Tm, Tb);
Chris@42: 			      Tn = FNMS(KP766044443, Tm, Tb);
Chris@42: 			 }
Chris@42: 		    }
Chris@42: 		    R1[WS(rs, 2)] = FNMS(KP1_705737063, Tu, Tt);
Chris@42: 		    R0[WS(rs, 4)] = FMA(KP1_705737063, Tu, Tt);
Chris@42: 		    R0[WS(rs, 2)] = FNMS(KP1_326827896, To, Tn);
Chris@42: 		    R1[WS(rs, 3)] = FMA(KP1_326827896, To, Tn);
Chris@42: 	       }
Chris@42: 	  }
Chris@42:      }
Chris@42: }
Chris@42: 
Chris@42: static const kr2c_desc desc = { 9, "r2cb_9", {8, 0, 24, 0}, &GENUS };
Chris@42: 
Chris@42: void X(codelet_r2cb_9) (planner *p) {
Chris@42:      X(kr2c_register) (p, r2cb_9, &desc);
Chris@42: }
Chris@42: 
Chris@42: #else				/* HAVE_FMA */
Chris@42: 
Chris@42: /* Generated by: ../../../genfft/gen_r2cb.native -compact -variables 4 -pipeline-latency 4 -sign 1 -n 9 -name r2cb_9 -include r2cb.h */
Chris@42: 
Chris@42: /*
Chris@42:  * This function contains 32 FP additions, 18 FP multiplications,
Chris@42:  * (or, 22 additions, 8 multiplications, 10 fused multiply/add),
Chris@42:  * 35 stack variables, 12 constants, and 18 memory accesses
Chris@42:  */
Chris@42: #include "r2cb.h"
Chris@42: 
Chris@42: static void r2cb_9(R *R0, R *R1, R *Cr, R *Ci, stride rs, stride csr, stride csi, INT v, INT ivs, INT ovs)
Chris@42: {
Chris@42:      DK(KP984807753, +0.984807753012208059366743024589523013670643252);
Chris@42:      DK(KP173648177, +0.173648177666930348851716626769314796000375677);
Chris@42:      DK(KP300767466, +0.300767466360870593278543795225003852144476517);
Chris@42:      DK(KP1_705737063, +1.705737063904886419256501927880148143872040591);
Chris@42:      DK(KP642787609, +0.642787609686539326322643409907263432907559884);
Chris@42:      DK(KP766044443, +0.766044443118978035202392650555416673935832457);
Chris@42:      DK(KP1_326827896, +1.326827896337876792410842639271782594433726619);
Chris@42:      DK(KP1_113340798, +1.113340798452838732905825904094046265936583811);
Chris@42:      DK(KP500000000, +0.500000000000000000000000000000000000000000000);
Chris@42:      DK(KP866025403, +0.866025403784438646763723170752936183471402627);
Chris@42:      DK(KP2_000000000, +2.000000000000000000000000000000000000000000000);
Chris@42:      DK(KP1_732050807, +1.732050807568877293527446341505872366942805254);
Chris@42:      {
Chris@42: 	  INT i;
Chris@42: 	  for (i = v; i > 0; i = i - 1, R0 = R0 + ovs, R1 = R1 + ovs, Cr = Cr + ivs, Ci = Ci + ivs, MAKE_VOLATILE_STRIDE(36, rs), MAKE_VOLATILE_STRIDE(36, csr), MAKE_VOLATILE_STRIDE(36, csi)) {
Chris@42: 	       E T3, Tq, Tc, Tk, Tj, T8, Tm, Ts, Th, Tr, Tw, Tx;
Chris@42: 	       {
Chris@42: 		    E Tb, T1, T2, T9, Ta;
Chris@42: 		    Ta = Ci[WS(csi, 3)];
Chris@42: 		    Tb = KP1_732050807 * Ta;
Chris@42: 		    T1 = Cr[0];
Chris@42: 		    T2 = Cr[WS(csr, 3)];
Chris@42: 		    T9 = T1 - T2;
Chris@42: 		    T3 = FMA(KP2_000000000, T2, T1);
Chris@42: 		    Tq = T9 + Tb;
Chris@42: 		    Tc = T9 - Tb;
Chris@42: 	       }
Chris@42: 	       {
Chris@42: 		    E T4, T7, Ti, Tg, Tl, Td;
Chris@42: 		    T4 = Cr[WS(csr, 1)];
Chris@42: 		    Tk = Ci[WS(csi, 1)];
Chris@42: 		    {
Chris@42: 			 E T5, T6, Te, Tf;
Chris@42: 			 T5 = Cr[WS(csr, 4)];
Chris@42: 			 T6 = Cr[WS(csr, 2)];
Chris@42: 			 T7 = T5 + T6;
Chris@42: 			 Ti = KP866025403 * (T5 - T6);
Chris@42: 			 Te = Ci[WS(csi, 4)];
Chris@42: 			 Tf = Ci[WS(csi, 2)];
Chris@42: 			 Tg = KP866025403 * (Te + Tf);
Chris@42: 			 Tj = Tf - Te;
Chris@42: 		    }
Chris@42: 		    T8 = T4 + T7;
Chris@42: 		    Tl = FMA(KP500000000, Tj, Tk);
Chris@42: 		    Tm = Ti + Tl;
Chris@42: 		    Ts = Tl - Ti;
Chris@42: 		    Td = FNMS(KP500000000, T7, T4);
Chris@42: 		    Th = Td - Tg;
Chris@42: 		    Tr = Td + Tg;
Chris@42: 	       }
Chris@42: 	       R0[0] = FMA(KP2_000000000, T8, T3);
Chris@42: 	       Tw = T3 - T8;
Chris@42: 	       Tx = KP1_732050807 * (Tk - Tj);
Chris@42: 	       R1[WS(rs, 1)] = Tw - Tx;
Chris@42: 	       R0[WS(rs, 3)] = Tw + Tx;
Chris@42: 	       {
Chris@42: 		    E Tp, Tn, To, Tv, Tt, Tu;
Chris@42: 		    Tp = FMA(KP1_113340798, Th, KP1_326827896 * Tm);
Chris@42: 		    Tn = FNMS(KP642787609, Tm, KP766044443 * Th);
Chris@42: 		    To = Tc - Tn;
Chris@42: 		    R1[0] = FMA(KP2_000000000, Tn, Tc);
Chris@42: 		    R1[WS(rs, 3)] = To + Tp;
Chris@42: 		    R0[WS(rs, 2)] = To - Tp;
Chris@42: 		    Tv = FMA(KP1_705737063, Tr, KP300767466 * Ts);
Chris@42: 		    Tt = FNMS(KP984807753, Ts, KP173648177 * Tr);
Chris@42: 		    Tu = Tq - Tt;
Chris@42: 		    R0[WS(rs, 1)] = FMA(KP2_000000000, Tt, Tq);
Chris@42: 		    R0[WS(rs, 4)] = Tu + Tv;
Chris@42: 		    R1[WS(rs, 2)] = Tu - Tv;
Chris@42: 	       }
Chris@42: 	  }
Chris@42:      }
Chris@42: }
Chris@42: 
Chris@42: static const kr2c_desc desc = { 9, "r2cb_9", {22, 8, 10, 0}, &GENUS };
Chris@42: 
Chris@42: void X(codelet_r2cb_9) (planner *p) {
Chris@42:      X(kr2c_register) (p, r2cb_9, &desc);
Chris@42: }
Chris@42: 
Chris@42: #endif				/* HAVE_FMA */