/* * Copyright (c) 2003, 2007-8 Matteo Frigo * Copyright (c) 2003, 2007-8 Massachusetts Institute of Technology * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by * the Free Software Foundation; either version 2 of the License, or * (at your option) any later version. * * This program is distributed in the hope that it will be useful, * but WITHOUT ANY WARRANTY; without even the implied warranty of * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the * GNU General Public License for more details. * * You should have received a copy of the GNU General Public License * along with this program; if not, write to the Free Software * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA * */ /* 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 14 -name X(spu_n2fv_14) */ /* * This function contains 74 FP additions, 48 FP multiplications, * (or, 32 additions, 6 multiplications, 42 fused multiply/add), * 93 stack variables, 6 constants, and 35 memory accesses */ #include "fftw-spu.h" void X(spu_n2fv_14) (const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs) { DVK(KP801937735, +0.801937735804838252472204639014890102331838324); DVK(KP974927912, +0.974927912181823607018131682993931217232785801); DVK(KP554958132, +0.554958132087371191422194871006410481067288862); DVK(KP900968867, +0.900968867902419126236102319507445051165919162); DVK(KP692021471, +0.692021471630095869627814897002069140197260599); DVK(KP356895867, +0.356895867892209443894399510021300583399127187); INT i; const R *xi; R *xo; xi = ri; xo = ro; for (i = v; i > 0; i = i - VL, xi = xi + (VL * ivs), xo = xo + (VL * ovs), MAKE_VOLATILE_STRIDE(is), MAKE_VOLATILE_STRIDE(os)) { V TH, T3, Ta, Ts, TV, TN, TK, TW, Tt, To, Th, Tu, TU, TQ, Tp; V TC, Tx, TR, T14, TZ, T1, T2, TI, T6, TJ, T9, TP, Tn, TO, Tk; V TM, Tg, TL, Td, T4, T5, T7, T8, Tl, Tm, Ti, Tj, Te, Tf, Tb; V Tc, T1a, T1b, T19, T1c, Tr, Tw, Tq, Tv, T16, T18, T15, T17, T1d, T1e; V T1f, T1i, Tz, TB, Ty, TA, T1g, TT, TY, TS, TX, T1h, T1j, T11, T13; V T10, T12, T1k, TE, TG, TD, TF, T1l, T1m; T1 = LD(&(xi[0]), ivs, &(xi[0])); T2 = LD(&(xi[WS(is, 7)]), ivs, &(xi[WS(is, 1)])); TH = VADD(T1, T2); T3 = VSUB(T1, T2); T4 = LD(&(xi[WS(is, 2)]), ivs, &(xi[0])); T5 = LD(&(xi[WS(is, 9)]), ivs, &(xi[WS(is, 1)])); TI = VADD(T4, T5); T6 = VSUB(T4, T5); T7 = LD(&(xi[WS(is, 12)]), ivs, &(xi[0])); T8 = LD(&(xi[WS(is, 5)]), ivs, &(xi[WS(is, 1)])); TJ = VADD(T7, T8); T9 = VSUB(T7, T8); Tl = LD(&(xi[WS(is, 8)]), ivs, &(xi[0])); Tm = LD(&(xi[WS(is, 1)]), ivs, &(xi[WS(is, 1)])); TP = VADD(Tl, Tm); Tn = VSUB(Tl, Tm); Ti = LD(&(xi[WS(is, 6)]), ivs, &(xi[0])); Tj = LD(&(xi[WS(is, 13)]), ivs, &(xi[WS(is, 1)])); TO = VADD(Ti, Tj); Tk = VSUB(Ti, Tj); Te = LD(&(xi[WS(is, 10)]), ivs, &(xi[0])); Tf = LD(&(xi[WS(is, 3)]), ivs, &(xi[WS(is, 1)])); TM = VADD(Te, Tf); Tg = VSUB(Te, Tf); Tb = LD(&(xi[WS(is, 4)]), ivs, &(xi[0])); Tc = LD(&(xi[WS(is, 11)]), ivs, &(xi[WS(is, 1)])); TL = VADD(Tb, Tc); Td = VSUB(Tb, Tc); Ta = VADD(T6, T9); Ts = VSUB(T9, T6); TV = VSUB(TL, TM); TN = VADD(TL, TM); TK = VADD(TI, TJ); TW = VSUB(TJ, TI); Tt = VSUB(Tn, Tk); To = VADD(Tk, Tn); Th = VADD(Td, Tg); Tu = VSUB(Tg, Td); TU = VSUB(TO, TP); TQ = VADD(TO, TP); Tp = VFNMS(LDK(KP356895867), Ta, To); TC = VFNMS(LDK(KP356895867), To, Th); Tx = VFNMS(LDK(KP356895867), Th, Ta); TR = VFNMS(LDK(KP356895867), TQ, TN); T14 = VFNMS(LDK(KP356895867), TN, TK); TZ = VFNMS(LDK(KP356895867), TK, TQ); T19 = VADD(T3, VADD(Ta, VADD(Th, To))); STM2(&(xo[14]), T19, ovs, &(xo[2])); T1a = VADD(TH, VADD(TK, VADD(TN, TQ))); STM2(&(xo[0]), T1a, ovs, &(xo[0])); Tq = VFNMS(LDK(KP692021471), Tp, Th); Tr = VFNMS(LDK(KP900968867), Tq, T3); Tv = VFMA(LDK(KP554958132), Tu, Tt); Tw = VMUL(LDK(KP974927912), VFNMS(LDK(KP801937735), Tv, Ts)); T1b = VFNMSI(Tw, Tr); STM2(&(xo[10]), T1b, ovs, &(xo[2])); T1c = VFMAI(Tw, Tr); STM2(&(xo[18]), T1c, ovs, &(xo[2])); T15 = VFNMS(LDK(KP692021471), T14, TQ); T16 = VFNMS(LDK(KP900968867), T15, TH); T17 = VFNMS(LDK(KP554958132), TU, TW); T18 = VMUL(LDK(KP974927912), VFNMS(LDK(KP801937735), T17, TV)); T1d = VFMAI(T18, T16); STM2(&(xo[12]), T1d, ovs, &(xo[0])); STN2(&(xo[12]), T1d, T19, ovs); T1e = VFNMSI(T18, T16); STM2(&(xo[16]), T1e, ovs, &(xo[0])); STN2(&(xo[16]), T1e, T1c, ovs); Ty = VFNMS(LDK(KP692021471), Tx, To); Tz = VFNMS(LDK(KP900968867), Ty, T3); TA = VFMA(LDK(KP554958132), Tt, Ts); TB = VMUL(LDK(KP974927912), VFMA(LDK(KP801937735), TA, Tu)); T1f = VFNMSI(TB, Tz); STM2(&(xo[26]), T1f, ovs, &(xo[2])); T1g = VFMAI(TB, Tz); STM2(&(xo[2]), T1g, ovs, &(xo[2])); STN2(&(xo[0]), T1a, T1g, ovs); TS = VFNMS(LDK(KP692021471), TR, TK); TT = VFNMS(LDK(KP900968867), TS, TH); TX = VFMA(LDK(KP554958132), TW, TV); TY = VMUL(LDK(KP974927912), VFNMS(LDK(KP801937735), TX, TU)); T1h = VFMAI(TY, TT); STM2(&(xo[8]), T1h, ovs, &(xo[0])); STN2(&(xo[8]), T1h, T1b, ovs); T1i = VFNMSI(TY, TT); STM2(&(xo[20]), T1i, ovs, &(xo[0])); T10 = VFNMS(LDK(KP692021471), TZ, TN); T11 = VFNMS(LDK(KP900968867), T10, TH); T12 = VFMA(LDK(KP554958132), TV, TU); T13 = VMUL(LDK(KP974927912), VFMA(LDK(KP801937735), T12, TW)); T1j = VFMAI(T13, T11); STM2(&(xo[4]), T1j, ovs, &(xo[0])); T1k = VFNMSI(T13, T11); STM2(&(xo[24]), T1k, ovs, &(xo[0])); STN2(&(xo[24]), T1k, T1f, ovs); TD = VFNMS(LDK(KP692021471), TC, Ta); TE = VFNMS(LDK(KP900968867), TD, T3); TF = VFNMS(LDK(KP554958132), Ts, Tu); TG = VMUL(LDK(KP974927912), VFNMS(LDK(KP801937735), TF, Tt)); T1l = VFNMSI(TG, TE); STM2(&(xo[22]), T1l, ovs, &(xo[2])); STN2(&(xo[20]), T1i, T1l, ovs); T1m = VFMAI(TG, TE); STM2(&(xo[6]), T1m, ovs, &(xo[2])); STN2(&(xo[4]), T1j, T1m, ovs); } }