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_twiddle_c -standalone -fma -reorder-insns -simd -compact -variables 100000 -include fftw-spu.h -trivial-stores -n 4 -name X(spu_t1fv_4) */
d@0: 
d@0: /*
d@0:  * This function contains 11 FP additions, 8 FP multiplications,
d@0:  * (or, 9 additions, 6 multiplications, 2 fused multiply/add),
d@0:  * 17 stack variables, 0 constants, and 8 memory accesses
d@0:  */
d@0: #include "fftw-spu.h"
d@0: 
d@0: void X(spu_t1fv_4) (R *ri, R *ii, const R *W, stride rs, INT mb, INT me, INT ms) {
d@0:      INT m;
d@0:      R *x;
d@0:      x = ri;
d@0:      for (m = mb, W = W + (mb * ((TWVL / VL) * 6)); m < me; m = m + VL, x = x + (VL * ms), W = W + (TWVL * 6), MAKE_VOLATILE_STRIDE(rs)) {
d@0: 	  V T4, Tc, T9, Td, T1, T3, T2, T6, T8, T5, T7, Tb, Te, Ta, Tf;
d@0: 	  T1 = LD(&(x[0]), ms, &(x[0]));
d@0: 	  T2 = LD(&(x[WS(rs, 2)]), ms, &(x[0]));
d@0: 	  T3 = BYTWJ(&(W[TWVL * 2]), T2);
d@0: 	  T4 = VSUB(T1, T3);
d@0: 	  Tc = VADD(T1, T3);
d@0: 	  T5 = LD(&(x[WS(rs, 1)]), ms, &(x[WS(rs, 1)]));
d@0: 	  T6 = BYTWJ(&(W[0]), T5);
d@0: 	  T7 = LD(&(x[WS(rs, 3)]), ms, &(x[WS(rs, 1)]));
d@0: 	  T8 = BYTWJ(&(W[TWVL * 4]), T7);
d@0: 	  T9 = VSUB(T6, T8);
d@0: 	  Td = VADD(T6, T8);
d@0: 	  Ta = VFNMSI(T9, T4);
d@0: 	  Tb = VFMAI(T9, T4);
d@0: 	  ST(&(x[WS(rs, 1)]), Ta, ms, &(x[WS(rs, 1)]));
d@0: 	  Tf = VADD(Tc, Td);
d@0: 	  Te = VSUB(Tc, Td);
d@0: 	  ST(&(x[0]), Tf, ms, &(x[0]));
d@0: 	  ST(&(x[WS(rs, 3)]), Tb, ms, &(x[WS(rs, 1)]));
d@0: 	  ST(&(x[WS(rs, 2)]), Te, ms, &(x[0]));
d@0:      }
d@0: }