/* * 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 16 -name X(spu_n2fv_16) */ /* * This function contains 72 FP additions, 34 FP multiplications, * (or, 38 additions, 0 multiplications, 34 fused multiply/add), * 94 stack variables, 3 constants, and 40 memory accesses */ #include "fftw-spu.h" void X(spu_n2fv_16) (const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs) { DVK(KP923879532, +0.923879532511286756128183189396788286822416626); DVK(KP707106781, +0.707106781186547524400844362104849039284835938); DVK(KP414213562, +0.414213562373095048801688724209698078569671875); 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, TU, Tz, T7, TA, Tu, TV, TK, TC, Te, TX, TO, TD, Tl, TY; V TR, T3, T6, T1, T2, T4, T5, TJ, Tq, TI, Tt, To, Tp, Tr, Ts; V TM, Ta, TN, Td, T8, T9, Tb, Tc, TP, Th, TQ, Tk, Tf, Tg, Ti; V Tj, T1b, T1c, T1d, T1e, TF, TG, TB, TE, T1f, T1g, T1h, T1i, Tn, Tx; V Ty, Tw, Tv, Tm, T13, T17, T16, T14, T11, TT, T12, T10, TL, TS, TW; V TZ, T1j, T1k, T1l, T1m, T19, T15, T1a, T18, T1n, T1o, T1p, T1q; T1 = LD(&(xi[0]), ivs, &(xi[0])); T2 = LD(&(xi[WS(is, 8)]), ivs, &(xi[0])); TH = VSUB(T1, T2); T3 = VADD(T1, T2); T4 = LD(&(xi[WS(is, 4)]), ivs, &(xi[0])); T5 = LD(&(xi[WS(is, 12)]), ivs, &(xi[0])); T6 = VADD(T4, T5); TU = VSUB(T4, T5); Tz = VADD(T3, T6); T7 = VSUB(T3, T6); To = LD(&(xi[WS(is, 14)]), ivs, &(xi[0])); Tp = LD(&(xi[WS(is, 6)]), ivs, &(xi[0])); TJ = VSUB(To, Tp); Tq = VADD(To, Tp); Tr = LD(&(xi[WS(is, 2)]), ivs, &(xi[0])); Ts = LD(&(xi[WS(is, 10)]), ivs, &(xi[0])); TI = VSUB(Tr, Ts); Tt = VADD(Tr, Ts); TA = VADD(Tt, Tq); Tu = VSUB(Tq, Tt); TV = VSUB(TJ, TI); TK = VADD(TI, TJ); T8 = LD(&(xi[WS(is, 1)]), ivs, &(xi[WS(is, 1)])); T9 = LD(&(xi[WS(is, 9)]), ivs, &(xi[WS(is, 1)])); TM = VSUB(T8, T9); Ta = VADD(T8, T9); Tb = LD(&(xi[WS(is, 5)]), ivs, &(xi[WS(is, 1)])); Tc = LD(&(xi[WS(is, 13)]), ivs, &(xi[WS(is, 1)])); TN = VSUB(Tb, Tc); Td = VADD(Tb, Tc); TC = VADD(Ta, Td); Te = VSUB(Ta, Td); TX = VFMA(LDK(KP414213562), TM, TN); TO = VFNMS(LDK(KP414213562), TN, TM); Tf = LD(&(xi[WS(is, 15)]), ivs, &(xi[WS(is, 1)])); Tg = LD(&(xi[WS(is, 7)]), ivs, &(xi[WS(is, 1)])); TP = VSUB(Tf, Tg); Th = VADD(Tf, Tg); Ti = LD(&(xi[WS(is, 3)]), ivs, &(xi[WS(is, 1)])); Tj = LD(&(xi[WS(is, 11)]), ivs, &(xi[WS(is, 1)])); TQ = VSUB(Tj, Ti); Tk = VADD(Ti, Tj); TD = VADD(Th, Tk); Tl = VSUB(Th, Tk); TY = VFMA(LDK(KP414213562), TP, TQ); TR = VFNMS(LDK(KP414213562), TQ, TP); TF = VSUB(Tz, TA); TB = VADD(Tz, TA); TE = VADD(TC, TD); TG = VSUB(TD, TC); T1b = VSUB(TB, TE); STM2(&(xo[16]), T1b, ovs, &(xo[0])); T1c = VADD(TB, TE); STM2(&(xo[0]), T1c, ovs, &(xo[0])); T1d = VFNMSI(TG, TF); STM2(&(xo[24]), T1d, ovs, &(xo[0])); T1e = VFMAI(TG, TF); STM2(&(xo[8]), T1e, ovs, &(xo[0])); Tv = VSUB(Tl, Te); Tm = VADD(Te, Tl); Tn = VFNMS(LDK(KP707106781), Tm, T7); Tx = VFMA(LDK(KP707106781), Tm, T7); Ty = VFMA(LDK(KP707106781), Tv, Tu); Tw = VFNMS(LDK(KP707106781), Tv, Tu); T1f = VFNMSI(Tw, Tn); STM2(&(xo[12]), T1f, ovs, &(xo[0])); T1g = VFMAI(Ty, Tx); STM2(&(xo[4]), T1g, ovs, &(xo[0])); T1h = VFMAI(Tw, Tn); STM2(&(xo[20]), T1h, ovs, &(xo[0])); T1i = VFNMSI(Ty, Tx); STM2(&(xo[28]), T1i, ovs, &(xo[0])); T13 = VFNMS(LDK(KP707106781), TK, TH); TL = VFMA(LDK(KP707106781), TK, TH); TS = VADD(TO, TR); T17 = VSUB(TR, TO); T11 = VFMA(LDK(KP923879532), TS, TL); TT = VFNMS(LDK(KP923879532), TS, TL); T16 = VFMA(LDK(KP707106781), TV, TU); TW = VFNMS(LDK(KP707106781), TV, TU); TZ = VSUB(TX, TY); T14 = VADD(TX, TY); T12 = VFMA(LDK(KP923879532), TZ, TW); T10 = VFNMS(LDK(KP923879532), TZ, TW); T1j = VFNMSI(T10, TT); STM2(&(xo[18]), T1j, ovs, &(xo[2])); STN2(&(xo[16]), T1b, T1j, ovs); T1k = VFMAI(T12, T11); STM2(&(xo[30]), T1k, ovs, &(xo[2])); STN2(&(xo[28]), T1i, T1k, ovs); T1l = VFMAI(T10, TT); STM2(&(xo[14]), T1l, ovs, &(xo[2])); STN2(&(xo[12]), T1f, T1l, ovs); T1m = VFNMSI(T12, T11); STM2(&(xo[2]), T1m, ovs, &(xo[2])); STN2(&(xo[0]), T1c, T1m, ovs); T19 = VFMA(LDK(KP923879532), T14, T13); T15 = VFNMS(LDK(KP923879532), T14, T13); T1a = VFMA(LDK(KP923879532), T17, T16); T18 = VFNMS(LDK(KP923879532), T17, T16); T1n = VFNMSI(T18, T15); STM2(&(xo[10]), T1n, ovs, &(xo[2])); STN2(&(xo[8]), T1e, T1n, ovs); T1o = VFNMSI(T1a, T19); STM2(&(xo[26]), T1o, ovs, &(xo[2])); STN2(&(xo[24]), T1d, T1o, ovs); T1p = VFMAI(T18, T15); STM2(&(xo[22]), T1p, ovs, &(xo[2])); STN2(&(xo[20]), T1h, T1p, ovs); T1q = VFMAI(T1a, T19); STM2(&(xo[6]), T1q, ovs, &(xo[2])); STN2(&(xo[4]), T1g, T1q, ovs); } }