cannam@167: /*
cannam@167:  * Copyright (c) 2003, 2007-14 Matteo Frigo
cannam@167:  * Copyright (c) 2003, 2007-14 Massachusetts Institute of Technology
cannam@167:  *
cannam@167:  * This program is free software; you can redistribute it and/or modify
cannam@167:  * it under the terms of the GNU General Public License as published by
cannam@167:  * the Free Software Foundation; either version 2 of the License, or
cannam@167:  * (at your option) any later version.
cannam@167:  *
cannam@167:  * This program is distributed in the hope that it will be useful,
cannam@167:  * but WITHOUT ANY WARRANTY; without even the implied warranty of
cannam@167:  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
cannam@167:  * GNU General Public License for more details.
cannam@167:  *
cannam@167:  * You should have received a copy of the GNU General Public License
cannam@167:  * along with this program; if not, write to the Free Software
cannam@167:  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02110-1301  USA
cannam@167:  *
cannam@167:  */
cannam@167: 
cannam@167: /* This file was automatically generated --- DO NOT EDIT */
cannam@167: /* Generated on Thu May 24 08:07:28 EDT 2018 */
cannam@167: 
cannam@167: #include "rdft/codelet-rdft.h"
cannam@167: 
cannam@167: #if defined(ARCH_PREFERS_FMA) || defined(ISA_EXTENSION_PREFERS_FMA)
cannam@167: 
cannam@167: /* Generated by: ../../../genfft/gen_r2cb.native -fma -compact -variables 4 -pipeline-latency 4 -sign 1 -n 12 -name r2cb_12 -include rdft/scalar/r2cb.h */
cannam@167: 
cannam@167: /*
cannam@167:  * This function contains 38 FP additions, 16 FP multiplications,
cannam@167:  * (or, 22 additions, 0 multiplications, 16 fused multiply/add),
cannam@167:  * 25 stack variables, 2 constants, and 24 memory accesses
cannam@167:  */
cannam@167: #include "rdft/scalar/r2cb.h"
cannam@167: 
cannam@167: static void r2cb_12(R *R0, R *R1, R *Cr, R *Ci, stride rs, stride csr, stride csi, INT v, INT ivs, INT ovs)
cannam@167: {
cannam@167:      DK(KP1_732050807, +1.732050807568877293527446341505872366942805254);
cannam@167:      DK(KP2_000000000, +2.000000000000000000000000000000000000000000000);
cannam@167:      {
cannam@167: 	  INT i;
cannam@167: 	  for (i = v; i > 0; i = i - 1, R0 = R0 + ovs, R1 = R1 + ovs, Cr = Cr + ivs, Ci = Ci + ivs, MAKE_VOLATILE_STRIDE(48, rs), MAKE_VOLATILE_STRIDE(48, csr), MAKE_VOLATILE_STRIDE(48, csi)) {
cannam@167: 	       E T8, Tb, Tk, Tz, Tu, Tv, Tn, Ty, T3, Tp, Tf, T6, Tq, Ti;
cannam@167: 	       {
cannam@167: 		    E T9, Ta, Tl, Tm;
cannam@167: 		    T8 = Cr[WS(csr, 3)];
cannam@167: 		    T9 = Cr[WS(csr, 5)];
cannam@167: 		    Ta = Cr[WS(csr, 1)];
cannam@167: 		    Tb = T9 + Ta;
cannam@167: 		    Tk = FNMS(KP2_000000000, T8, Tb);
cannam@167: 		    Tz = T9 - Ta;
cannam@167: 		    Tu = Ci[WS(csi, 3)];
cannam@167: 		    Tl = Ci[WS(csi, 5)];
cannam@167: 		    Tm = Ci[WS(csi, 1)];
cannam@167: 		    Tv = Tl + Tm;
cannam@167: 		    Tn = Tl - Tm;
cannam@167: 		    Ty = FMA(KP2_000000000, Tu, Tv);
cannam@167: 	       }
cannam@167: 	       {
cannam@167: 		    E Te, T1, T2, Td;
cannam@167: 		    Te = Ci[WS(csi, 4)];
cannam@167: 		    T1 = Cr[0];
cannam@167: 		    T2 = Cr[WS(csr, 4)];
cannam@167: 		    Td = T1 - T2;
cannam@167: 		    T3 = FMA(KP2_000000000, T2, T1);
cannam@167: 		    Tp = FNMS(KP1_732050807, Te, Td);
cannam@167: 		    Tf = FMA(KP1_732050807, Te, Td);
cannam@167: 	       }
cannam@167: 	       {
cannam@167: 		    E Th, T4, T5, Tg;
cannam@167: 		    Th = Ci[WS(csi, 2)];
cannam@167: 		    T4 = Cr[WS(csr, 6)];
cannam@167: 		    T5 = Cr[WS(csr, 2)];
cannam@167: 		    Tg = T4 - T5;
cannam@167: 		    T6 = FMA(KP2_000000000, T5, T4);
cannam@167: 		    Tq = FMA(KP1_732050807, Th, Tg);
cannam@167: 		    Ti = FNMS(KP1_732050807, Th, Tg);
cannam@167: 	       }
cannam@167: 	       {
cannam@167: 		    E T7, Tc, Tx, TA;
cannam@167: 		    T7 = T3 + T6;
cannam@167: 		    Tc = T8 + Tb;
cannam@167: 		    R0[WS(rs, 3)] = FNMS(KP2_000000000, Tc, T7);
cannam@167: 		    R0[0] = FMA(KP2_000000000, Tc, T7);
cannam@167: 		    {
cannam@167: 			 E Tj, To, TB, TC;
cannam@167: 			 Tj = Tf + Ti;
cannam@167: 			 To = FMA(KP1_732050807, Tn, Tk);
cannam@167: 			 R0[WS(rs, 1)] = Tj + To;
cannam@167: 			 R0[WS(rs, 4)] = Tj - To;
cannam@167: 			 TB = Tf - Ti;
cannam@167: 			 TC = FNMS(KP1_732050807, Tz, Ty);
cannam@167: 			 R1[WS(rs, 2)] = TB - TC;
cannam@167: 			 R1[WS(rs, 5)] = TB + TC;
cannam@167: 		    }
cannam@167: 		    Tx = Tp - Tq;
cannam@167: 		    TA = FMA(KP1_732050807, Tz, Ty);
cannam@167: 		    R1[0] = Tx - TA;
cannam@167: 		    R1[WS(rs, 3)] = Tx + TA;
cannam@167: 		    {
cannam@167: 			 E Tt, Tw, Tr, Ts;
cannam@167: 			 Tt = T3 - T6;
cannam@167: 			 Tw = Tu - Tv;
cannam@167: 			 R1[WS(rs, 4)] = FNMS(KP2_000000000, Tw, Tt);
cannam@167: 			 R1[WS(rs, 1)] = FMA(KP2_000000000, Tw, Tt);
cannam@167: 			 Tr = Tp + Tq;
cannam@167: 			 Ts = FNMS(KP1_732050807, Tn, Tk);
cannam@167: 			 R0[WS(rs, 5)] = Tr + Ts;
cannam@167: 			 R0[WS(rs, 2)] = Tr - Ts;
cannam@167: 		    }
cannam@167: 	       }
cannam@167: 	  }
cannam@167:      }
cannam@167: }
cannam@167: 
cannam@167: static const kr2c_desc desc = { 12, "r2cb_12", {22, 0, 16, 0}, &GENUS };
cannam@167: 
cannam@167: void X(codelet_r2cb_12) (planner *p) {
cannam@167:      X(kr2c_register) (p, r2cb_12, &desc);
cannam@167: }
cannam@167: 
cannam@167: #else
cannam@167: 
cannam@167: /* Generated by: ../../../genfft/gen_r2cb.native -compact -variables 4 -pipeline-latency 4 -sign 1 -n 12 -name r2cb_12 -include rdft/scalar/r2cb.h */
cannam@167: 
cannam@167: /*
cannam@167:  * This function contains 38 FP additions, 10 FP multiplications,
cannam@167:  * (or, 34 additions, 6 multiplications, 4 fused multiply/add),
cannam@167:  * 25 stack variables, 2 constants, and 24 memory accesses
cannam@167:  */
cannam@167: #include "rdft/scalar/r2cb.h"
cannam@167: 
cannam@167: static void r2cb_12(R *R0, R *R1, R *Cr, R *Ci, stride rs, stride csr, stride csi, INT v, INT ivs, INT ovs)
cannam@167: {
cannam@167:      DK(KP1_732050807, +1.732050807568877293527446341505872366942805254);
cannam@167:      DK(KP2_000000000, +2.000000000000000000000000000000000000000000000);
cannam@167:      {
cannam@167: 	  INT i;
cannam@167: 	  for (i = v; i > 0; i = i - 1, R0 = R0 + ovs, R1 = R1 + ovs, Cr = Cr + ivs, Ci = Ci + ivs, MAKE_VOLATILE_STRIDE(48, rs), MAKE_VOLATILE_STRIDE(48, csr), MAKE_VOLATILE_STRIDE(48, csi)) {
cannam@167: 	       E T8, Tb, Tm, TA, Tw, Tx, Tp, TB, T3, Tr, Tg, T6, Ts, Tk;
cannam@167: 	       {
cannam@167: 		    E T9, Ta, Tn, To;
cannam@167: 		    T8 = Cr[WS(csr, 3)];
cannam@167: 		    T9 = Cr[WS(csr, 5)];
cannam@167: 		    Ta = Cr[WS(csr, 1)];
cannam@167: 		    Tb = T9 + Ta;
cannam@167: 		    Tm = FMS(KP2_000000000, T8, Tb);
cannam@167: 		    TA = KP1_732050807 * (T9 - Ta);
cannam@167: 		    Tw = Ci[WS(csi, 3)];
cannam@167: 		    Tn = Ci[WS(csi, 5)];
cannam@167: 		    To = Ci[WS(csi, 1)];
cannam@167: 		    Tx = Tn + To;
cannam@167: 		    Tp = KP1_732050807 * (Tn - To);
cannam@167: 		    TB = FMA(KP2_000000000, Tw, Tx);
cannam@167: 	       }
cannam@167: 	       {
cannam@167: 		    E Tf, T1, T2, Td, Te;
cannam@167: 		    Te = Ci[WS(csi, 4)];
cannam@167: 		    Tf = KP1_732050807 * Te;
cannam@167: 		    T1 = Cr[0];
cannam@167: 		    T2 = Cr[WS(csr, 4)];
cannam@167: 		    Td = T1 - T2;
cannam@167: 		    T3 = FMA(KP2_000000000, T2, T1);
cannam@167: 		    Tr = Td - Tf;
cannam@167: 		    Tg = Td + Tf;
cannam@167: 	       }
cannam@167: 	       {
cannam@167: 		    E Tj, T4, T5, Th, Ti;
cannam@167: 		    Ti = Ci[WS(csi, 2)];
cannam@167: 		    Tj = KP1_732050807 * Ti;
cannam@167: 		    T4 = Cr[WS(csr, 6)];
cannam@167: 		    T5 = Cr[WS(csr, 2)];
cannam@167: 		    Th = T4 - T5;
cannam@167: 		    T6 = FMA(KP2_000000000, T5, T4);
cannam@167: 		    Ts = Th + Tj;
cannam@167: 		    Tk = Th - Tj;
cannam@167: 	       }
cannam@167: 	       {
cannam@167: 		    E T7, Tc, Tz, TC;
cannam@167: 		    T7 = T3 + T6;
cannam@167: 		    Tc = KP2_000000000 * (T8 + Tb);
cannam@167: 		    R0[WS(rs, 3)] = T7 - Tc;
cannam@167: 		    R0[0] = T7 + Tc;
cannam@167: 		    {
cannam@167: 			 E Tl, Tq, TD, TE;
cannam@167: 			 Tl = Tg + Tk;
cannam@167: 			 Tq = Tm - Tp;
cannam@167: 			 R0[WS(rs, 1)] = Tl - Tq;
cannam@167: 			 R0[WS(rs, 4)] = Tl + Tq;
cannam@167: 			 TD = Tg - Tk;
cannam@167: 			 TE = TB - TA;
cannam@167: 			 R1[WS(rs, 2)] = TD - TE;
cannam@167: 			 R1[WS(rs, 5)] = TD + TE;
cannam@167: 		    }
cannam@167: 		    Tz = Tr - Ts;
cannam@167: 		    TC = TA + TB;
cannam@167: 		    R1[0] = Tz - TC;
cannam@167: 		    R1[WS(rs, 3)] = Tz + TC;
cannam@167: 		    {
cannam@167: 			 E Tv, Ty, Tt, Tu;
cannam@167: 			 Tv = T3 - T6;
cannam@167: 			 Ty = KP2_000000000 * (Tw - Tx);
cannam@167: 			 R1[WS(rs, 4)] = Tv - Ty;
cannam@167: 			 R1[WS(rs, 1)] = Tv + Ty;
cannam@167: 			 Tt = Tr + Ts;
cannam@167: 			 Tu = Tm + Tp;
cannam@167: 			 R0[WS(rs, 5)] = Tt - Tu;
cannam@167: 			 R0[WS(rs, 2)] = Tt + Tu;
cannam@167: 		    }
cannam@167: 	       }
cannam@167: 	  }
cannam@167:      }
cannam@167: }
cannam@167: 
cannam@167: static const kr2c_desc desc = { 12, "r2cb_12", {34, 6, 4, 0}, &GENUS };
cannam@167: 
cannam@167: void X(codelet_r2cb_12) (planner *p) {
cannam@167:      X(kr2c_register) (p, r2cb_12, &desc);
cannam@167: }
cannam@167: 
cannam@167: #endif