d@0: /*
d@0:  * Copyright (c) 2003, 2007-8 Matteo Frigo
d@0:  * Copyright (c) 2003, 2007-8 Massachusetts Institute of Technology
d@0:  *
d@0:  * This program is free software; you can redistribute it and/or modify
d@0:  * it under the terms of the GNU General Public License as published by
d@0:  * the Free Software Foundation; either version 2 of the License, or
d@0:  * (at your option) any later version.
d@0:  *
d@0:  * This program is distributed in the hope that it will be useful,
d@0:  * but WITHOUT ANY WARRANTY; without even the implied warranty of
d@0:  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
d@0:  * GNU General Public License for more details.
d@0:  *
d@0:  * You should have received a copy of the GNU General Public License
d@0:  * along with this program; if not, write to the Free Software
d@0:  * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
d@0:  *
d@0:  */
d@0: /* Generated by: ../../genfft/gen_notw_c -standalone -fma -reorder-insns -simd -compact -variables 100000 -with-ostride 2 -include fftw-spu.h -store-multiple 2 -n 8 -name X(spu_n2fv_8) */
d@0: 
d@0: /*
d@0:  * This function contains 26 FP additions, 10 FP multiplications,
d@0:  * (or, 16 additions, 0 multiplications, 10 fused multiply/add),
d@0:  * 38 stack variables, 1 constants, and 20 memory accesses
d@0:  */
d@0: #include "fftw-spu.h"
d@0: 
d@0: void X(spu_n2fv_8) (const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs) {
d@0:      DVK(KP707106781, +0.707106781186547524400844362104849039284835938);
d@0:      INT i;
d@0:      const R *xi;
d@0:      R *xo;
d@0:      xi = ri;
d@0:      xo = ro;
d@0:      for (i = v; i > 0; i = i - VL, xi = xi + (VL * ivs), xo = xo + (VL * ovs), MAKE_VOLATILE_STRIDE(is), MAKE_VOLATILE_STRIDE(os)) {
d@0: 	  V Tj, T3, Tk, Te, Tm, Tn, Tf, Ta, T1, T2, Tc, Td, T6, T9, T4;
d@0: 	  V T5, T7, T8, Th, Ti, Tr, Tl, To, Tu, Ts, Tb, Tg, Tp, Tq, Tt;
d@0: 	  V Tv, Tw, Tx, Ty;
d@0: 	  T1 = LD(&(xi[0]), ivs, &(xi[0]));
d@0: 	  T2 = LD(&(xi[WS(is, 4)]), ivs, &(xi[0]));
d@0: 	  Tj = VADD(T1, T2);
d@0: 	  T3 = VSUB(T1, T2);
d@0: 	  Tc = LD(&(xi[WS(is, 2)]), ivs, &(xi[0]));
d@0: 	  Td = LD(&(xi[WS(is, 6)]), ivs, &(xi[0]));
d@0: 	  Tk = VADD(Tc, Td);
d@0: 	  Te = VSUB(Tc, Td);
d@0: 	  T4 = LD(&(xi[WS(is, 1)]), ivs, &(xi[WS(is, 1)]));
d@0: 	  T5 = LD(&(xi[WS(is, 5)]), ivs, &(xi[WS(is, 1)]));
d@0: 	  Tm = VADD(T4, T5);
d@0: 	  T6 = VSUB(T4, T5);
d@0: 	  T7 = LD(&(xi[WS(is, 7)]), ivs, &(xi[WS(is, 1)]));
d@0: 	  T8 = LD(&(xi[WS(is, 3)]), ivs, &(xi[WS(is, 1)]));
d@0: 	  T9 = VSUB(T7, T8);
d@0: 	  Tn = VADD(T7, T8);
d@0: 	  Tf = VSUB(T9, T6);
d@0: 	  Ta = VADD(T6, T9);
d@0: 	  Th = VFNMS(LDK(KP707106781), Ta, T3);
d@0: 	  Tb = VFMA(LDK(KP707106781), Ta, T3);
d@0: 	  Tg = VFNMS(LDK(KP707106781), Tf, Te);
d@0: 	  Ti = VFMA(LDK(KP707106781), Tf, Te);
d@0: 	  Tr = VFNMSI(Tg, Tb);
d@0: 	  STM2(&(xo[2]), Tr, ovs, &(xo[2]));
d@0: 	  Ts = VFMAI(Tg, Tb);
d@0: 	  STM2(&(xo[14]), Ts, ovs, &(xo[2]));
d@0: 	  Tl = VADD(Tj, Tk);
d@0: 	  Tp = VSUB(Tj, Tk);
d@0: 	  Tq = VSUB(Tn, Tm);
d@0: 	  To = VADD(Tm, Tn);
d@0: 	  Tt = VFNMSI(Tq, Tp);
d@0: 	  STM2(&(xo[12]), Tt, ovs, &(xo[0]));
d@0: 	  STN2(&(xo[12]), Tt, Ts, ovs);
d@0: 	  Tu = VFMAI(Tq, Tp);
d@0: 	  STM2(&(xo[4]), Tu, ovs, &(xo[0]));
d@0: 	  Tv = VFNMSI(Ti, Th);
d@0: 	  STM2(&(xo[10]), Tv, ovs, &(xo[2]));
d@0: 	  Tw = VFMAI(Ti, Th);
d@0: 	  STM2(&(xo[6]), Tw, ovs, &(xo[2]));
d@0: 	  STN2(&(xo[4]), Tu, Tw, ovs);
d@0: 	  Tx = VSUB(Tl, To);
d@0: 	  STM2(&(xo[8]), Tx, ovs, &(xo[0]));
d@0: 	  STN2(&(xo[8]), Tx, Tv, ovs);
d@0: 	  Ty = VADD(Tl, To);
d@0: 	  STM2(&(xo[0]), Ty, ovs, &(xo[0]));
d@0: 	  STN2(&(xo[0]), Ty, Tr, ovs);
d@0:      }
d@0: }