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 12 -name X(spu_t1fv_12) */
d@0: 
d@0: /*
d@0:  * This function contains 59 FP additions, 42 FP multiplications,
d@0:  * (or, 41 additions, 24 multiplications, 18 fused multiply/add),
d@0:  * 75 stack variables, 2 constants, and 24 memory accesses
d@0:  */
d@0: #include "fftw-spu.h"
d@0: 
d@0: void X(spu_t1fv_12) (R *ri, R *ii, const R *W, stride rs, INT mb, INT me, INT ms) {
d@0:      DVK(KP866025403, +0.866025403784438646763723170752936183471402627);
d@0:      DVK(KP500000000, +0.500000000000000000000000000000000000000000000);
d@0:      INT m;
d@0:      R *x;
d@0:      x = ri;
d@0:      for (m = mb, W = W + (mb * ((TWVL / VL) * 22)); m < me; m = m + VL, x = x + (VL * ms), W = W + (TWVL * 22), MAKE_VOLATILE_STRIDE(rs)) {
d@0: 	  V TY, T7, Tf, TZ, TE, TQ, T11, Tw, TA, T12, Tr, TP, T1, T3, Td;
d@0: 	  V Tb, T9, T5, T2, Tc, Ta, T8, T4, TC, T6, TD, Te, Ti, Tk, Ty;
d@0: 	  V Tu, Tp, Tn, Th, Tj, Tx, Tt, To, Tm, Tl, Tv, Tq, Tz, T16, T17;
d@0: 	  V T14, T15, T10, T13, T18, T19, TM, TN, TI, Ts, TF, TJ, Tg, TB, TH;
d@0: 	  V TK, TG, TL, TR, TV, TO, TU, TT, TW, TS, TX;
d@0: 	  T1 = LD(&(x[0]), ms, &(x[0]));
d@0: 	  T2 = LD(&(x[WS(rs, 4)]), ms, &(x[0]));
d@0: 	  T3 = BYTWJ(&(W[TWVL * 6]), T2);
d@0: 	  Tc = LD(&(x[WS(rs, 2)]), ms, &(x[0]));
d@0: 	  Td = BYTWJ(&(W[TWVL * 2]), Tc);
d@0: 	  Ta = LD(&(x[WS(rs, 10)]), ms, &(x[0]));
d@0: 	  Tb = BYTWJ(&(W[TWVL * 18]), Ta);
d@0: 	  T8 = LD(&(x[WS(rs, 6)]), ms, &(x[0]));
d@0: 	  T9 = BYTWJ(&(W[TWVL * 10]), T8);
d@0: 	  T4 = LD(&(x[WS(rs, 8)]), ms, &(x[0]));
d@0: 	  T5 = BYTWJ(&(W[TWVL * 14]), T4);
d@0: 	  T6 = VADD(T3, T5);
d@0: 	  TC = VSUB(T5, T3);
d@0: 	  TY = VADD(T1, T6);
d@0: 	  T7 = VFNMS(LDK(KP500000000), T6, T1);
d@0: 	  TD = VSUB(Td, Tb);
d@0: 	  Te = VADD(Tb, Td);
d@0: 	  Tf = VFNMS(LDK(KP500000000), Te, T9);
d@0: 	  TZ = VADD(T9, Te);
d@0: 	  TE = VSUB(TC, TD);
d@0: 	  TQ = VADD(TC, TD);
d@0: 	  Th = LD(&(x[WS(rs, 11)]), ms, &(x[WS(rs, 1)]));
d@0: 	  Ti = BYTWJ(&(W[TWVL * 20]), Th);
d@0: 	  Tj = LD(&(x[WS(rs, 7)]), ms, &(x[WS(rs, 1)]));
d@0: 	  Tk = BYTWJ(&(W[TWVL * 12]), Tj);
d@0: 	  Tx = LD(&(x[WS(rs, 9)]), ms, &(x[WS(rs, 1)]));
d@0: 	  Ty = BYTWJ(&(W[TWVL * 16]), Tx);
d@0: 	  Tt = LD(&(x[WS(rs, 3)]), ms, &(x[WS(rs, 1)]));
d@0: 	  Tu = BYTWJ(&(W[TWVL * 4]), Tt);
d@0: 	  To = LD(&(x[WS(rs, 5)]), ms, &(x[WS(rs, 1)]));
d@0: 	  Tp = BYTWJ(&(W[TWVL * 8]), To);
d@0: 	  Tm = LD(&(x[WS(rs, 1)]), ms, &(x[WS(rs, 1)]));
d@0: 	  Tn = BYTWJ(&(W[0]), Tm);
d@0: 	  Tv = VADD(Tk, Ti);
d@0: 	  Tl = VSUB(Ti, Tk);
d@0: 	  T11 = VADD(Tu, Tv);
d@0: 	  Tw = VFNMS(LDK(KP500000000), Tv, Tu);
d@0: 	  Tq = VSUB(Tn, Tp);
d@0: 	  Tz = VADD(Tn, Tp);
d@0: 	  TA = VFNMS(LDK(KP500000000), Tz, Ty);
d@0: 	  T12 = VADD(Ty, Tz);
d@0: 	  Tr = VADD(Tl, Tq);
d@0: 	  TP = VSUB(Tl, Tq);
d@0: 	  T10 = VSUB(TY, TZ);
d@0: 	  T16 = VADD(TY, TZ);
d@0: 	  T17 = VADD(T11, T12);
d@0: 	  T13 = VSUB(T11, T12);
d@0: 	  T14 = VFNMSI(T13, T10);
d@0: 	  T15 = VFMAI(T13, T10);
d@0: 	  ST(&(x[WS(rs, 9)]), T14, ms, &(x[WS(rs, 1)]));
d@0: 	  ST(&(x[WS(rs, 3)]), T15, ms, &(x[WS(rs, 1)]));
d@0: 	  T18 = VSUB(T16, T17);
d@0: 	  T19 = VADD(T16, T17);
d@0: 	  ST(&(x[WS(rs, 6)]), T18, ms, &(x[0]));
d@0: 	  ST(&(x[0]), T19, ms, &(x[0]));
d@0: 	  Tg = VSUB(T7, Tf);
d@0: 	  TM = VADD(T7, Tf);
d@0: 	  TI = VFNMS(LDK(KP866025403), Tr, Tg);
d@0: 	  Ts = VFMA(LDK(KP866025403), Tr, Tg);
d@0: 	  TN = VADD(Tw, TA);
d@0: 	  TB = VSUB(Tw, TA);
d@0: 	  TF = VFNMS(LDK(KP866025403), TE, TB);
d@0: 	  TJ = VFMA(LDK(KP866025403), TE, TB);
d@0: 	  TG = VFNMSI(TF, Ts);
d@0: 	  TH = VFMAI(TF, Ts);
d@0: 	  ST(&(x[WS(rs, 1)]), TG, ms, &(x[WS(rs, 1)]));
d@0: 	  TL = VFMAI(TJ, TI);
d@0: 	  TK = VFNMSI(TJ, TI);
d@0: 	  ST(&(x[WS(rs, 7)]), TL, ms, &(x[WS(rs, 1)]));
d@0: 	  ST(&(x[WS(rs, 11)]), TH, ms, &(x[WS(rs, 1)]));
d@0: 	  ST(&(x[WS(rs, 5)]), TK, ms, &(x[WS(rs, 1)]));
d@0: 	  TR = VMUL(LDK(KP866025403), VSUB(TP, TQ));
d@0: 	  TV = VMUL(LDK(KP866025403), VADD(TQ, TP));
d@0: 	  TO = VSUB(TM, TN);
d@0: 	  TU = VADD(TM, TN);
d@0: 	  TS = VFMAI(TR, TO);
d@0: 	  TT = VFNMSI(TR, TO);
d@0: 	  ST(&(x[WS(rs, 2)]), TS, ms, &(x[0]));
d@0: 	  TX = VFNMSI(TV, TU);
d@0: 	  TW = VFMAI(TV, TU);
d@0: 	  ST(&(x[WS(rs, 8)]), TX, ms, &(x[0]));
d@0: 	  ST(&(x[WS(rs, 10)]), TT, ms, &(x[0]));
d@0: 	  ST(&(x[WS(rs, 4)]), TW, ms, &(x[0]));
d@0:      }
d@0: }