static void n2fv_4(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs) { 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 T1, T2, T4, T5; T1 = LD(&(xi[0]), ivs, &(xi[0])); T2 = LD(&(xi[WS(is, 2)]), ivs, &(xi[0])); T4 = LD(&(xi[WS(is, 1)]), ivs, &(xi[WS(is, 1)])); T5 = LD(&(xi[WS(is, 3)]), ivs, &(xi[WS(is, 1)])); { V T3, T7, T6, T8; T3 = VSUB(T1, T2); T7 = VADD(T1, T2); T6 = VSUB(T4, T5); T8 = VADD(T4, T5); { V T9, Ta, Tb, Tc; T9 = VSUB(T7, T8); STM2(&(xo[4]), T9, ovs, &(xo[0])); Ta = VADD(T7, T8); STM2(&(xo[0]), Ta, ovs, &(xo[0])); Tb = VFMAI(T6, T3); STM2(&(xo[6]), Tb, ovs, &(xo[2])); STN2(&(xo[4]), T9, Tb, ovs); Tc = VFNMSI(T6, T3); STM2(&(xo[2]), Tc, ovs, &(xo[2])); STN2(&(xo[0]), Ta, Tc, ovs); } } } }
static void n2bv_2(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs) { INT i; const R *xi; R *xo; xi = ii; xo = io; 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 T1, T2, T3, T4; T1 = LD(&(xi[0]), ivs, &(xi[0])); T2 = LD(&(xi[WS(is, 1)]), ivs, &(xi[WS(is, 1)])); T3 = VSUB(T1, T2); STM2(&(xo[2]), T3, ovs, &(xo[2])); T4 = VADD(T1, T2); STM2(&(xo[0]), T4, ovs, &(xo[0])); STN2(&(xo[0]), T4, T3, ovs); } }
static void n2bv_6(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs) { DVK(KP500000000, +0.500000000000000000000000000000000000000000000); DVK(KP866025403, +0.866025403784438646763723170752936183471402627); INT i; const R *xi; R *xo; xi = ii; xo = io; 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 T1, T2, T4, T5, T7, T8; T1 = LD(&(xi[0]), ivs, &(xi[0])); T2 = LD(&(xi[WS(is, 3)]), ivs, &(xi[WS(is, 1)])); T4 = LD(&(xi[WS(is, 2)]), ivs, &(xi[0])); T5 = LD(&(xi[WS(is, 5)]), ivs, &(xi[WS(is, 1)])); T7 = LD(&(xi[WS(is, 4)]), ivs, &(xi[0])); T8 = LD(&(xi[WS(is, 1)]), ivs, &(xi[WS(is, 1)])); { V T3, Td, T6, Te, T9, Tf; T3 = VSUB(T1, T2); Td = VADD(T1, T2); T6 = VSUB(T4, T5); Te = VADD(T4, T5); T9 = VSUB(T7, T8); Tf = VADD(T7, T8); { V Tg, Ti, Ta, Tc; Tg = VADD(Te, Tf); Ti = VMUL(LDK(KP866025403), VSUB(Te, Tf)); Ta = VADD(T6, T9); Tc = VMUL(LDK(KP866025403), VSUB(T6, T9)); { V Th, Tj, Tb, Tk; Th = VFNMS(LDK(KP500000000), Tg, Td); Tj = VADD(Td, Tg); STM2(&(xo[0]), Tj, ovs, &(xo[0])); Tb = VFNMS(LDK(KP500000000), Ta, T3); Tk = VADD(T3, Ta); STM2(&(xo[6]), Tk, ovs, &(xo[2])); { V Tl, Tm, Tn, To; Tl = VFMAI(Ti, Th); STM2(&(xo[8]), Tl, ovs, &(xo[0])); Tm = VFNMSI(Ti, Th); STM2(&(xo[4]), Tm, ovs, &(xo[0])); STN2(&(xo[4]), Tm, Tk, ovs); Tn = VFNMSI(Tc, Tb); STM2(&(xo[10]), Tn, ovs, &(xo[2])); STN2(&(xo[8]), Tl, Tn, ovs); To = VFMAI(Tc, Tb); STM2(&(xo[2]), To, ovs, &(xo[2])); STN2(&(xo[0]), Tj, To, ovs); } } } } } }
static void n2bv_10(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs) { DVK(KP250000000, +0.250000000000000000000000000000000000000000000); DVK(KP559016994, +0.559016994374947424102293417182819058860154590); DVK(KP587785252, +0.587785252292473129168705954639072768597652438); DVK(KP951056516, +0.951056516295153572116439333379382143405698634); INT i; const R *xi; R *xo; xi = ii; xo = io; 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 Tl, Ty, T7, Te, Tw, Tt, Tz, TA, TB, Tg, Th, Tm, Tj, Tk; Tj = LD(&(xi[0]), ivs, &(xi[0])); Tk = LD(&(xi[WS(is, 5)]), ivs, &(xi[WS(is, 1)])); Tl = VSUB(Tj, Tk); Ty = VADD(Tj, Tk); { V T3, Tr, Td, Tv, T6, Ts, Ta, Tu; { V T1, T2, Tb, Tc; T1 = LD(&(xi[WS(is, 2)]), ivs, &(xi[0])); T2 = LD(&(xi[WS(is, 7)]), ivs, &(xi[WS(is, 1)])); T3 = VSUB(T1, T2); Tr = VADD(T1, T2); Tb = LD(&(xi[WS(is, 6)]), ivs, &(xi[0])); Tc = LD(&(xi[WS(is, 1)]), ivs, &(xi[WS(is, 1)])); Td = VSUB(Tb, Tc); Tv = VADD(Tb, Tc); } { V T4, T5, T8, T9; T4 = LD(&(xi[WS(is, 8)]), ivs, &(xi[0])); T5 = LD(&(xi[WS(is, 3)]), ivs, &(xi[WS(is, 1)])); T6 = VSUB(T4, T5); Ts = VADD(T4, T5); T8 = LD(&(xi[WS(is, 4)]), ivs, &(xi[0])); T9 = LD(&(xi[WS(is, 9)]), ivs, &(xi[WS(is, 1)])); Ta = VSUB(T8, T9); Tu = VADD(T8, T9); } T7 = VSUB(T3, T6); Te = VSUB(Ta, Td); Tw = VSUB(Tu, Tv); Tt = VSUB(Tr, Ts); Tz = VADD(Tr, Ts); TA = VADD(Tu, Tv); TB = VADD(Tz, TA); Tg = VADD(T3, T6); Th = VADD(Ta, Td); Tm = VADD(Tg, Th); } { V TH, TI, TK, TL, TM; TH = VADD(Tl, Tm); STM2(&(xo[10]), TH, ovs, &(xo[2])); TI = VADD(Ty, TB); STM2(&(xo[0]), TI, ovs, &(xo[0])); { V Tf, Tq, To, Tp, Ti, Tn, TJ; Tf = VBYI(VFMA(LDK(KP951056516), T7, VMUL(LDK(KP587785252), Te))); Tq = VBYI(VFNMS(LDK(KP951056516), Te, VMUL(LDK(KP587785252), T7))); Ti = VMUL(LDK(KP559016994), VSUB(Tg, Th)); Tn = VFNMS(LDK(KP250000000), Tm, Tl); To = VADD(Ti, Tn); Tp = VSUB(Tn, Ti); TJ = VADD(Tf, To); STM2(&(xo[2]), TJ, ovs, &(xo[2])); STN2(&(xo[0]), TI, TJ, ovs); TK = VADD(Tq, Tp); STM2(&(xo[14]), TK, ovs, &(xo[2])); TL = VSUB(To, Tf); STM2(&(xo[18]), TL, ovs, &(xo[2])); TM = VSUB(Tp, Tq); STM2(&(xo[6]), TM, ovs, &(xo[2])); } { V Tx, TG, TE, TF, TC, TD; Tx = VBYI(VFNMS(LDK(KP951056516), Tw, VMUL(LDK(KP587785252), Tt))); TG = VBYI(VFMA(LDK(KP951056516), Tt, VMUL(LDK(KP587785252), Tw))); TC = VFNMS(LDK(KP250000000), TB, Ty); TD = VMUL(LDK(KP559016994), VSUB(Tz, TA)); TE = VSUB(TC, TD); TF = VADD(TD, TC); { V TN, TO, TP, TQ; TN = VADD(Tx, TE); STM2(&(xo[4]), TN, ovs, &(xo[0])); STN2(&(xo[4]), TN, TM, ovs); TO = VADD(TG, TF); STM2(&(xo[12]), TO, ovs, &(xo[0])); STN2(&(xo[12]), TO, TK, ovs); TP = VSUB(TE, Tx); STM2(&(xo[16]), TP, ovs, &(xo[0])); STN2(&(xo[16]), TP, TL, ovs); TQ = VSUB(TF, TG); STM2(&(xo[8]), TQ, ovs, &(xo[0])); STN2(&(xo[8]), TQ, TH, ovs); } } } } }
static void n2bv_12(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs) { DVK(KP866025403, +0.866025403784438646763723170752936183471402627); DVK(KP500000000, +0.500000000000000000000000000000000000000000000); { INT i; const R *xi; R *xo; xi = ii; xo = io; 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 T5, Ta, TG, TF, Ty, Tm, Ti, Tp, TJ, TI, Tx, Ts; { V T1, T6, T4, Tk, T9, Tl; T1 = LD(&(xi[0]), ivs, &(xi[0])); T6 = LD(&(xi[WS(is, 6)]), ivs, &(xi[0])); { V T2, T3, T7, T8; T2 = LD(&(xi[WS(is, 4)]), ivs, &(xi[0])); T3 = LD(&(xi[WS(is, 8)]), ivs, &(xi[0])); T4 = VADD(T2, T3); Tk = VSUB(T2, T3); T7 = LD(&(xi[WS(is, 10)]), ivs, &(xi[0])); T8 = LD(&(xi[WS(is, 2)]), ivs, &(xi[0])); T9 = VADD(T7, T8); Tl = VSUB(T7, T8); } T5 = VFNMS(LDK(KP500000000), T4, T1); Ta = VFNMS(LDK(KP500000000), T9, T6); TG = VADD(T6, T9); TF = VADD(T1, T4); Ty = VADD(Tk, Tl); Tm = VMUL(LDK(KP866025403), VSUB(Tk, Tl)); } { V Tn, Tq, Te, To, Th, Tr; Tn = LD(&(xi[WS(is, 3)]), ivs, &(xi[WS(is, 1)])); Tq = LD(&(xi[WS(is, 9)]), ivs, &(xi[WS(is, 1)])); { V Tc, Td, Tf, Tg; Tc = LD(&(xi[WS(is, 7)]), ivs, &(xi[WS(is, 1)])); Td = LD(&(xi[WS(is, 11)]), ivs, &(xi[WS(is, 1)])); Te = VSUB(Tc, Td); To = VADD(Tc, Td); Tf = LD(&(xi[WS(is, 1)]), ivs, &(xi[WS(is, 1)])); Tg = LD(&(xi[WS(is, 5)]), ivs, &(xi[WS(is, 1)])); Th = VSUB(Tf, Tg); Tr = VADD(Tf, Tg); } Ti = VMUL(LDK(KP866025403), VSUB(Te, Th)); Tp = VFNMS(LDK(KP500000000), To, Tn); TJ = VADD(Tq, Tr); TI = VADD(Tn, To); Tx = VADD(Te, Th); Ts = VFNMS(LDK(KP500000000), Tr, Tq); } { V TN, TO, TP, TQ, TR, TS; { V TH, TK, TL, TM; TH = VSUB(TF, TG); TK = VBYI(VSUB(TI, TJ)); TN = VSUB(TH, TK); STM2(&(xo[6]), TN, ovs, &(xo[2])); TO = VADD(TH, TK); STM2(&(xo[18]), TO, ovs, &(xo[2])); TL = VADD(TF, TG); TM = VADD(TI, TJ); TP = VSUB(TL, TM); STM2(&(xo[12]), TP, ovs, &(xo[0])); TQ = VADD(TL, TM); STM2(&(xo[0]), TQ, ovs, &(xo[0])); } { V Tj, Tv, Tu, Tw, Tb, Tt, TT, TU; Tb = VSUB(T5, Ta); Tj = VSUB(Tb, Ti); Tv = VADD(Tb, Ti); Tt = VSUB(Tp, Ts); Tu = VBYI(VADD(Tm, Tt)); Tw = VBYI(VSUB(Tt, Tm)); TR = VSUB(Tj, Tu); STM2(&(xo[22]), TR, ovs, &(xo[2])); TS = VADD(Tv, Tw); STM2(&(xo[10]), TS, ovs, &(xo[2])); TT = VADD(Tj, Tu); STM2(&(xo[2]), TT, ovs, &(xo[2])); STN2(&(xo[0]), TQ, TT, ovs); TU = VSUB(Tv, Tw); STM2(&(xo[14]), TU, ovs, &(xo[2])); STN2(&(xo[12]), TP, TU, ovs); } { V Tz, TD, TC, TE, TA, TB; Tz = VBYI(VMUL(LDK(KP866025403), VSUB(Tx, Ty))); TD = VBYI(VMUL(LDK(KP866025403), VADD(Ty, Tx))); TA = VADD(T5, Ta); TB = VADD(Tp, Ts); TC = VSUB(TA, TB); TE = VADD(TA, TB); { V TV, TW, TX, TY; TV = VADD(Tz, TC); STM2(&(xo[4]), TV, ovs, &(xo[0])); STN2(&(xo[4]), TV, TN, ovs); TW = VSUB(TE, TD); STM2(&(xo[16]), TW, ovs, &(xo[0])); STN2(&(xo[16]), TW, TO, ovs); TX = VSUB(TC, Tz); STM2(&(xo[20]), TX, ovs, &(xo[0])); STN2(&(xo[20]), TX, TR, ovs); TY = VADD(TD, TE); STM2(&(xo[8]), TY, ovs, &(xo[0])); STN2(&(xo[8]), TY, TS, ovs); } } } } } VLEAVE(); }
static void n2bv_14(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs) { DVK(KP900968867, +0.900968867902419126236102319507445051165919162); DVK(KP801937735, +0.801937735804838252472204639014890102331838324); DVK(KP974927912, +0.974927912181823607018131682993931217232785801); DVK(KP692021471, +0.692021471630095869627814897002069140197260599); DVK(KP554958132, +0.554958132087371191422194871006410481067288862); DVK(KP356895867, +0.356895867892209443894399510021300583399127187); INT i; const R *xi; R *xo; xi = ii; xo = io; 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, TP, Tn, Ta, Tu, TU, TK, TO, Tk, TM, Tg, TL, Td, T1; V T2; T1 = LD(&(xi[0]), ivs, &(xi[0])); T2 = LD(&(xi[WS(is, 7)]), ivs, &(xi[WS(is, 1)])); { V Ti, TI, T6, TJ, T9, Tj, Te, Tf, Tb, Tc; { V T4, T5, T7, T8, Tl, Tm; T4 = LD(&(xi[WS(is, 2)]), ivs, &(xi[0])); T5 = LD(&(xi[WS(is, 9)]), ivs, &(xi[WS(is, 1)])); T7 = LD(&(xi[WS(is, 12)]), ivs, &(xi[0])); T8 = LD(&(xi[WS(is, 5)]), ivs, &(xi[WS(is, 1)])); Tl = LD(&(xi[WS(is, 8)]), ivs, &(xi[0])); Tm = LD(&(xi[WS(is, 1)]), ivs, &(xi[WS(is, 1)])); Ti = LD(&(xi[WS(is, 6)]), ivs, &(xi[0])); TH = VADD(T1, T2); T3 = VSUB(T1, T2); TI = VADD(T4, T5); T6 = VSUB(T4, T5); TJ = VADD(T7, T8); T9 = VSUB(T7, T8); TP = VADD(Tl, Tm); Tn = VSUB(Tl, Tm); Tj = LD(&(xi[WS(is, 13)]), ivs, &(xi[WS(is, 1)])); Te = LD(&(xi[WS(is, 10)]), ivs, &(xi[0])); Tf = LD(&(xi[WS(is, 3)]), ivs, &(xi[WS(is, 1)])); Tb = LD(&(xi[WS(is, 4)]), ivs, &(xi[0])); Tc = LD(&(xi[WS(is, 11)]), ivs, &(xi[WS(is, 1)])); } Ta = VADD(T6, T9); Tu = VSUB(T6, T9); TU = VSUB(TI, TJ); TK = VADD(TI, TJ); TO = VADD(Ti, Tj); Tk = VSUB(Ti, Tj); TM = VADD(Te, Tf); Tg = VSUB(Te, Tf); TL = VADD(Tb, Tc); Td = VSUB(Tb, Tc); } { V T19, T1a, T13, TG, TY, T18, TB, Tw, TT, Tz, T11, T16, TE, Tr, TV; V TQ; TV = VSUB(TP, TO); TQ = VADD(TO, TP); { V Ts, To, TW, TN; Ts = VSUB(Tk, Tn); To = VADD(Tk, Tn); TW = VSUB(TM, TL); TN = VADD(TL, TM); { V Tt, Th, TR, T12; Tt = VSUB(Td, Tg); Th = VADD(Td, Tg); TR = VFNMS(LDK(KP356895867), TK, TQ); T12 = VFNMS(LDK(KP554958132), TV, TU); { V Tx, TF, TZ, T14; Tx = VFNMS(LDK(KP356895867), Ta, To); TF = VFMA(LDK(KP554958132), Ts, Tu); T19 = VADD(TH, VADD(TK, VADD(TN, TQ))); STM2(&(xo[0]), T19, ovs, &(xo[0])); TZ = VFNMS(LDK(KP356895867), TN, TK); T14 = VFNMS(LDK(KP356895867), TQ, TN); { V TX, T17, TC, Tp; TX = VFMA(LDK(KP554958132), TW, TV); T17 = VFMA(LDK(KP554958132), TU, TW); T1a = VADD(T3, VADD(Ta, VADD(Th, To))); STM2(&(xo[14]), T1a, ovs, &(xo[2])); TC = VFNMS(LDK(KP356895867), Th, Ta); Tp = VFNMS(LDK(KP356895867), To, Th); { V TA, Tv, TS, Ty; TA = VFMA(LDK(KP554958132), Tt, Ts); Tv = VFNMS(LDK(KP554958132), Tu, Tt); TS = VFNMS(LDK(KP692021471), TR, TN); T13 = VMUL(LDK(KP974927912), VFNMS(LDK(KP801937735), T12, TW)); Ty = VFNMS(LDK(KP692021471), Tx, Th); TG = VMUL(LDK(KP974927912), VFMA(LDK(KP801937735), TF, Tt)); { V T10, T15, TD, Tq; T10 = VFNMS(LDK(KP692021471), TZ, TQ); T15 = VFNMS(LDK(KP692021471), T14, TK); TY = VMUL(LDK(KP974927912), VFMA(LDK(KP801937735), TX, TU)); T18 = VMUL(LDK(KP974927912), VFNMS(LDK(KP801937735), T17, TV)); TD = VFNMS(LDK(KP692021471), TC, To); Tq = VFNMS(LDK(KP692021471), Tp, Ta); TB = VMUL(LDK(KP974927912), VFNMS(LDK(KP801937735), TA, Tu)); Tw = VMUL(LDK(KP974927912), VFNMS(LDK(KP801937735), Tv, Ts)); TT = VFNMS(LDK(KP900968867), TS, TH); Tz = VFNMS(LDK(KP900968867), Ty, T3); T11 = VFNMS(LDK(KP900968867), T10, TH); T16 = VFNMS(LDK(KP900968867), T15, TH); TE = VFNMS(LDK(KP900968867), TD, T3); Tr = VFNMS(LDK(KP900968867), Tq, T3); } } } } } } { V T1b, T1c, T1d, T1e; T1b = VFMAI(TY, TT); STM2(&(xo[4]), T1b, ovs, &(xo[0])); T1c = VFNMSI(TY, TT); STM2(&(xo[24]), T1c, ovs, &(xo[0])); T1d = VFMAI(TB, Tz); STM2(&(xo[18]), T1d, ovs, &(xo[2])); T1e = VFNMSI(TB, Tz); STM2(&(xo[10]), T1e, ovs, &(xo[2])); { V T1f, T1g, T1h, T1i; T1f = VFMAI(T13, T11); STM2(&(xo[12]), T1f, ovs, &(xo[0])); STN2(&(xo[12]), T1f, T1a, ovs); T1g = VFNMSI(T13, T11); STM2(&(xo[16]), T1g, ovs, &(xo[0])); STN2(&(xo[16]), T1g, T1d, ovs); T1h = VFMAI(T18, T16); STM2(&(xo[8]), T1h, ovs, &(xo[0])); STN2(&(xo[8]), T1h, T1e, ovs); T1i = VFNMSI(T18, T16); STM2(&(xo[20]), T1i, ovs, &(xo[0])); { V T1j, T1k, T1l, T1m; T1j = VFNMSI(TG, TE); STM2(&(xo[26]), T1j, ovs, &(xo[2])); STN2(&(xo[24]), T1c, T1j, ovs); T1k = VFMAI(TG, TE); STM2(&(xo[2]), T1k, ovs, &(xo[2])); STN2(&(xo[0]), T19, T1k, ovs); T1l = VFNMSI(Tw, Tr); STM2(&(xo[22]), T1l, ovs, &(xo[2])); STN2(&(xo[20]), T1i, T1l, ovs); T1m = VFMAI(Tw, Tr); STM2(&(xo[6]), T1m, ovs, &(xo[2])); STN2(&(xo[4]), T1b, T1m, ovs); } } } } } }
static void n2bv_14(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs) { DVK(KP900968867, +0.900968867902419126236102319507445051165919162); DVK(KP222520933, +0.222520933956314404288902564496794759466355569); DVK(KP623489801, +0.623489801858733530525004884004239810632274731); DVK(KP781831482, +0.781831482468029808708444526674057750232334519); DVK(KP974927912, +0.974927912181823607018131682993931217232785801); DVK(KP433883739, +0.433883739117558120475768332848358754609990728); INT i; const R *xi; R *xo; xi = ii; xo = io; 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 Tp, Ty, Tl, TL, Tq, TE, T7, TJ, Ts, TB, Te, TK, Tr, TH, Tn; V To; Tn = LD(&(xi[0]), ivs, &(xi[0])); To = LD(&(xi[WS(is, 7)]), ivs, &(xi[WS(is, 1)])); Tp = VSUB(Tn, To); Ty = VADD(Tn, To); { V Th, TC, Tk, TD; { V Tf, Tg, Ti, Tj; Tf = LD(&(xi[WS(is, 4)]), ivs, &(xi[0])); Tg = LD(&(xi[WS(is, 11)]), ivs, &(xi[WS(is, 1)])); Th = VSUB(Tf, Tg); TC = VADD(Tf, Tg); Ti = LD(&(xi[WS(is, 10)]), ivs, &(xi[0])); Tj = LD(&(xi[WS(is, 3)]), ivs, &(xi[WS(is, 1)])); Tk = VSUB(Ti, Tj); TD = VADD(Ti, Tj); } Tl = VSUB(Th, Tk); TL = VSUB(TD, TC); Tq = VADD(Th, Tk); TE = VADD(TC, TD); } { V T3, Tz, T6, TA; { V T1, T2, T4, T5; T1 = LD(&(xi[WS(is, 2)]), ivs, &(xi[0])); T2 = LD(&(xi[WS(is, 9)]), ivs, &(xi[WS(is, 1)])); T3 = VSUB(T1, T2); Tz = VADD(T1, T2); T4 = LD(&(xi[WS(is, 12)]), ivs, &(xi[0])); T5 = LD(&(xi[WS(is, 5)]), ivs, &(xi[WS(is, 1)])); T6 = VSUB(T4, T5); TA = VADD(T4, T5); } T7 = VSUB(T3, T6); TJ = VSUB(Tz, TA); Ts = VADD(T3, T6); TB = VADD(Tz, TA); } { V Ta, TF, Td, TG; { V T8, T9, Tb, Tc; T8 = LD(&(xi[WS(is, 6)]), ivs, &(xi[0])); T9 = LD(&(xi[WS(is, 13)]), ivs, &(xi[WS(is, 1)])); Ta = VSUB(T8, T9); TF = VADD(T8, T9); Tb = LD(&(xi[WS(is, 8)]), ivs, &(xi[0])); Tc = LD(&(xi[WS(is, 1)]), ivs, &(xi[WS(is, 1)])); Td = VSUB(Tb, Tc); TG = VADD(Tb, Tc); } Te = VSUB(Ta, Td); TK = VSUB(TG, TF); Tr = VADD(Ta, Td); TH = VADD(TF, TG); } { V TR, TS, TU, TV; TR = VADD(Tp, VADD(Ts, VADD(Tq, Tr))); STM2(&(xo[14]), TR, ovs, &(xo[2])); TS = VADD(Ty, VADD(TB, VADD(TE, TH))); STM2(&(xo[0]), TS, ovs, &(xo[0])); { V TT, Tm, Tt, TQ, TP, TW; Tm = VBYI(VFMA(LDK(KP433883739), T7, VFNMS(LDK(KP781831482), Tl, VMUL(LDK(KP974927912), Te)))); Tt = VFMA(LDK(KP623489801), Tq, VFNMS(LDK(KP222520933), Tr, VFNMS(LDK(KP900968867), Ts, Tp))); TT = VADD(Tm, Tt); STM2(&(xo[6]), TT, ovs, &(xo[2])); TU = VSUB(Tt, Tm); STM2(&(xo[22]), TU, ovs, &(xo[2])); TQ = VBYI(VFMA(LDK(KP974927912), TJ, VFMA(LDK(KP433883739), TL, VMUL(LDK(KP781831482), TK)))); TP = VFMA(LDK(KP623489801), TH, VFNMS(LDK(KP900968867), TE, VFNMS(LDK(KP222520933), TB, Ty))); TV = VSUB(TP, TQ); STM2(&(xo[24]), TV, ovs, &(xo[0])); TW = VADD(TP, TQ); STM2(&(xo[4]), TW, ovs, &(xo[0])); STN2(&(xo[4]), TW, TT, ovs); } { V T10, TM, TI, TZ; { V Tu, Tv, TX, TY; Tu = VBYI(VFMA(LDK(KP781831482), T7, VFMA(LDK(KP974927912), Tl, VMUL(LDK(KP433883739), Te)))); Tv = VFMA(LDK(KP623489801), Ts, VFNMS(LDK(KP900968867), Tr, VFNMS(LDK(KP222520933), Tq, Tp))); TX = VADD(Tu, Tv); STM2(&(xo[2]), TX, ovs, &(xo[2])); STN2(&(xo[0]), TS, TX, ovs); TY = VSUB(Tv, Tu); STM2(&(xo[26]), TY, ovs, &(xo[2])); STN2(&(xo[24]), TV, TY, ovs); } TM = VBYI(VFNMS(LDK(KP433883739), TK, VFNMS(LDK(KP974927912), TL, VMUL(LDK(KP781831482), TJ)))); TI = VFMA(LDK(KP623489801), TB, VFNMS(LDK(KP900968867), TH, VFNMS(LDK(KP222520933), TE, Ty))); TZ = VSUB(TI, TM); STM2(&(xo[12]), TZ, ovs, &(xo[0])); STN2(&(xo[12]), TZ, TR, ovs); T10 = VADD(TI, TM); STM2(&(xo[16]), T10, ovs, &(xo[0])); { V T11, TO, TN, T12; TO = VBYI(VFMA(LDK(KP433883739), TJ, VFNMS(LDK(KP974927912), TK, VMUL(LDK(KP781831482), TL)))); TN = VFMA(LDK(KP623489801), TE, VFNMS(LDK(KP222520933), TH, VFNMS(LDK(KP900968867), TB, Ty))); T11 = VSUB(TN, TO); STM2(&(xo[8]), T11, ovs, &(xo[0])); T12 = VADD(TN, TO); STM2(&(xo[20]), T12, ovs, &(xo[0])); STN2(&(xo[20]), T12, TU, ovs); { V Tx, Tw, T13, T14; Tx = VBYI(VFNMS(LDK(KP781831482), Te, VFNMS(LDK(KP433883739), Tl, VMUL(LDK(KP974927912), T7)))); Tw = VFMA(LDK(KP623489801), Tr, VFNMS(LDK(KP900968867), Tq, VFNMS(LDK(KP222520933), Ts, Tp))); T13 = VSUB(Tw, Tx); STM2(&(xo[10]), T13, ovs, &(xo[2])); STN2(&(xo[8]), T11, T13, ovs); T14 = VADD(Tx, Tw); STM2(&(xo[18]), T14, ovs, &(xo[2])); STN2(&(xo[16]), T10, T14, ovs); } } } } } }
static void 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(KP414213562, +0.414213562373095048801688724209698078569671875); DVK(KP707106781, +0.707106781186547524400844362104849039284835938); 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 T7, Tu, TF, TB, T13, TL, TO, TX, TC, Te, TP, Th, TQ, Tk, TW; V T16; { V TH, TU, Tz, Tf, TK, TV, TA, TM, Ta, TN, Td, Tg, Ti, Tj; { V T1, T2, T4, T5, To, Tp, Tr, Ts; T1 = LD(&(xi[0]), ivs, &(xi[0])); T2 = LD(&(xi[WS(is, 8)]), ivs, &(xi[0])); T4 = LD(&(xi[WS(is, 4)]), ivs, &(xi[0])); T5 = LD(&(xi[WS(is, 12)]), ivs, &(xi[0])); To = LD(&(xi[WS(is, 14)]), ivs, &(xi[0])); Tp = LD(&(xi[WS(is, 6)]), ivs, &(xi[0])); Tr = LD(&(xi[WS(is, 2)]), ivs, &(xi[0])); Ts = LD(&(xi[WS(is, 10)]), ivs, &(xi[0])); { V T8, TJ, Tq, TI, Tt, T9, Tb, Tc, T3, T6; T8 = LD(&(xi[WS(is, 1)]), ivs, &(xi[WS(is, 1)])); TH = VSUB(T1, T2); T3 = VADD(T1, T2); TU = VSUB(T4, T5); T6 = VADD(T4, T5); TJ = VSUB(To, Tp); Tq = VADD(To, Tp); TI = VSUB(Tr, Ts); Tt = VADD(Tr, Ts); T9 = LD(&(xi[WS(is, 9)]), ivs, &(xi[WS(is, 1)])); Tb = LD(&(xi[WS(is, 5)]), ivs, &(xi[WS(is, 1)])); Tc = LD(&(xi[WS(is, 13)]), ivs, &(xi[WS(is, 1)])); T7 = VSUB(T3, T6); Tz = VADD(T3, T6); Tf = LD(&(xi[WS(is, 15)]), ivs, &(xi[WS(is, 1)])); TK = VADD(TI, TJ); TV = VSUB(TJ, TI); TA = VADD(Tt, Tq); Tu = VSUB(Tq, Tt); TM = VSUB(T8, T9); Ta = VADD(T8, T9); TN = VSUB(Tb, Tc); Td = VADD(Tb, Tc); Tg = LD(&(xi[WS(is, 7)]), ivs, &(xi[WS(is, 1)])); Ti = LD(&(xi[WS(is, 3)]), ivs, &(xi[WS(is, 1)])); Tj = LD(&(xi[WS(is, 11)]), ivs, &(xi[WS(is, 1)])); } } TF = VSUB(Tz, TA); TB = VADD(Tz, TA); T13 = VFNMS(LDK(KP707106781), TK, TH); TL = VFMA(LDK(KP707106781), TK, TH); TO = VFNMS(LDK(KP414213562), TN, TM); TX = VFMA(LDK(KP414213562), TM, TN); TC = VADD(Ta, Td); Te = VSUB(Ta, Td); TP = VSUB(Tf, Tg); Th = VADD(Tf, Tg); TQ = VSUB(Tj, Ti); Tk = VADD(Ti, Tj); TW = VFNMS(LDK(KP707106781), TV, TU); T16 = VFMA(LDK(KP707106781), TV, TU); } { V TY, TR, Tl, TD; TY = VFMA(LDK(KP414213562), TP, TQ); TR = VFNMS(LDK(KP414213562), TQ, TP); Tl = VSUB(Th, Tk); TD = VADD(Th, Tk); { V TS, T17, TZ, T14; TS = VADD(TO, TR); T17 = VSUB(TR, TO); TZ = VSUB(TX, TY); T14 = VADD(TX, TY); { V TE, TG, Tm, Tv; TE = VADD(TC, TD); TG = VSUB(TD, TC); Tm = VADD(Te, Tl); Tv = VSUB(Tl, Te); { V T18, T1a, TT, T11; T18 = VFNMS(LDK(KP923879532), T17, T16); T1a = VFMA(LDK(KP923879532), T17, T16); TT = VFNMS(LDK(KP923879532), TS, TL); T11 = VFMA(LDK(KP923879532), TS, TL); { V T15, T19, T10, T12; T15 = VFNMS(LDK(KP923879532), T14, T13); T19 = VFMA(LDK(KP923879532), T14, T13); T10 = VFNMS(LDK(KP923879532), TZ, TW); T12 = VFMA(LDK(KP923879532), TZ, TW); { V T1b, T1c, T1d, T1e; T1b = VFMAI(TG, TF); STM2(&(xo[8]), T1b, ovs, &(xo[0])); T1c = VFNMSI(TG, TF); STM2(&(xo[24]), T1c, ovs, &(xo[0])); T1d = VADD(TB, TE); STM2(&(xo[0]), T1d, ovs, &(xo[0])); T1e = VSUB(TB, TE); STM2(&(xo[16]), T1e, ovs, &(xo[0])); { V Tw, Ty, Tn, Tx; Tw = VFNMS(LDK(KP707106781), Tv, Tu); Ty = VFMA(LDK(KP707106781), Tv, Tu); Tn = VFNMS(LDK(KP707106781), Tm, T7); Tx = VFMA(LDK(KP707106781), Tm, T7); { V T1f, T1g, T1h, T1i; T1f = VFMAI(T1a, T19); STM2(&(xo[6]), T1f, ovs, &(xo[2])); T1g = VFNMSI(T1a, T19); STM2(&(xo[26]), T1g, ovs, &(xo[2])); STN2(&(xo[24]), T1c, T1g, ovs); T1h = VFMAI(T18, T15); STM2(&(xo[22]), T1h, ovs, &(xo[2])); T1i = VFNMSI(T18, T15); STM2(&(xo[10]), T1i, ovs, &(xo[2])); STN2(&(xo[8]), T1b, T1i, ovs); { V T1j, T1k, T1l, T1m; T1j = VFNMSI(T12, T11); STM2(&(xo[2]), T1j, ovs, &(xo[2])); STN2(&(xo[0]), T1d, T1j, ovs); T1k = VFMAI(T12, T11); STM2(&(xo[30]), T1k, ovs, &(xo[2])); T1l = VFMAI(T10, TT); STM2(&(xo[14]), T1l, ovs, &(xo[2])); T1m = VFNMSI(T10, TT); STM2(&(xo[18]), T1m, ovs, &(xo[2])); STN2(&(xo[16]), T1e, T1m, ovs); { V T1n, T1o, T1p, T1q; T1n = VFNMSI(Ty, Tx); STM2(&(xo[28]), T1n, ovs, &(xo[0])); STN2(&(xo[28]), T1n, T1k, ovs); T1o = VFMAI(Ty, Tx); STM2(&(xo[4]), T1o, ovs, &(xo[0])); STN2(&(xo[4]), T1o, T1f, ovs); T1p = VFMAI(Tw, Tn); STM2(&(xo[20]), T1p, ovs, &(xo[0])); STN2(&(xo[20]), T1p, T1h, ovs); T1q = VFNMSI(Tw, Tn); STM2(&(xo[12]), T1q, ovs, &(xo[0])); STN2(&(xo[12]), T1q, T1l, ovs); } } } } } } } } } } } }
static void 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(KP382683432, +0.382683432365089771728459984030398866761344562); DVK(KP707106781, +0.707106781186547524400844362104849039284835938); 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 Tp, T13, Tu, TN, Tm, T14, Tv, TY, T7, T17, Ty, TT, Te, T16, Tx; V TQ; { V Tn, To, TM, Ts, Tt, TL; Tn = LD(&(xi[WS(is, 4)]), ivs, &(xi[0])); To = LD(&(xi[WS(is, 12)]), ivs, &(xi[0])); TM = VADD(Tn, To); Ts = LD(&(xi[0]), ivs, &(xi[0])); Tt = LD(&(xi[WS(is, 8)]), ivs, &(xi[0])); TL = VADD(Ts, Tt); Tp = VSUB(Tn, To); T13 = VADD(TL, TM); Tu = VSUB(Ts, Tt); TN = VSUB(TL, TM); } { V Ti, TW, Tl, TX; { V Tg, Th, Tj, Tk; Tg = LD(&(xi[WS(is, 14)]), ivs, &(xi[0])); Th = LD(&(xi[WS(is, 6)]), ivs, &(xi[0])); Ti = VSUB(Tg, Th); TW = VADD(Tg, Th); Tj = LD(&(xi[WS(is, 2)]), ivs, &(xi[0])); Tk = LD(&(xi[WS(is, 10)]), ivs, &(xi[0])); Tl = VSUB(Tj, Tk); TX = VADD(Tj, Tk); } Tm = VMUL(LDK(KP707106781), VSUB(Ti, Tl)); T14 = VADD(TX, TW); Tv = VMUL(LDK(KP707106781), VADD(Tl, Ti)); TY = VSUB(TW, TX); } { V T3, TR, T6, TS; { V T1, T2, T4, T5; T1 = LD(&(xi[WS(is, 15)]), ivs, &(xi[WS(is, 1)])); T2 = LD(&(xi[WS(is, 7)]), ivs, &(xi[WS(is, 1)])); T3 = VSUB(T1, T2); TR = VADD(T1, T2); T4 = LD(&(xi[WS(is, 3)]), ivs, &(xi[WS(is, 1)])); T5 = LD(&(xi[WS(is, 11)]), ivs, &(xi[WS(is, 1)])); T6 = VSUB(T4, T5); TS = VADD(T4, T5); } T7 = VFNMS(LDK(KP923879532), T6, VMUL(LDK(KP382683432), T3)); T17 = VADD(TR, TS); Ty = VFMA(LDK(KP923879532), T3, VMUL(LDK(KP382683432), T6)); TT = VSUB(TR, TS); } { V Ta, TO, Td, TP; { V T8, T9, Tb, Tc; T8 = LD(&(xi[WS(is, 1)]), ivs, &(xi[WS(is, 1)])); T9 = LD(&(xi[WS(is, 9)]), ivs, &(xi[WS(is, 1)])); Ta = VSUB(T8, T9); TO = 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)])); Td = VSUB(Tb, Tc); TP = VADD(Tb, Tc); } Te = VFMA(LDK(KP382683432), Ta, VMUL(LDK(KP923879532), Td)); T16 = VADD(TO, TP); Tx = VFNMS(LDK(KP382683432), Td, VMUL(LDK(KP923879532), Ta)); TQ = VSUB(TO, TP); } { V T1b, T1c, T1d, T1e; { V T15, T18, T19, T1a; T15 = VADD(T13, T14); T18 = VADD(T16, T17); T1b = VSUB(T15, T18); STM2(&(xo[16]), T1b, ovs, &(xo[0])); T1c = VADD(T15, T18); STM2(&(xo[0]), T1c, ovs, &(xo[0])); T19 = VSUB(T13, T14); T1a = VBYI(VSUB(T17, T16)); T1d = VSUB(T19, T1a); STM2(&(xo[24]), T1d, ovs, &(xo[0])); T1e = VADD(T19, T1a); STM2(&(xo[8]), T1e, ovs, &(xo[0])); } { V T1f, T1g, T1h, T1i; { V TV, T11, T10, T12, TU, TZ; TU = VMUL(LDK(KP707106781), VADD(TQ, TT)); TV = VADD(TN, TU); T11 = VSUB(TN, TU); TZ = VMUL(LDK(KP707106781), VSUB(TT, TQ)); T10 = VBYI(VADD(TY, TZ)); T12 = VBYI(VSUB(TZ, TY)); T1f = VSUB(TV, T10); STM2(&(xo[28]), T1f, ovs, &(xo[0])); T1g = VADD(T11, T12); STM2(&(xo[12]), T1g, ovs, &(xo[0])); T1h = VADD(TV, T10); STM2(&(xo[4]), T1h, ovs, &(xo[0])); T1i = VSUB(T11, T12); STM2(&(xo[20]), T1i, ovs, &(xo[0])); } { V Tr, TB, TA, TC; { V Tf, Tq, Tw, Tz; Tf = VSUB(T7, Te); Tq = VSUB(Tm, Tp); Tr = VBYI(VSUB(Tf, Tq)); TB = VBYI(VADD(Tq, Tf)); Tw = VADD(Tu, Tv); Tz = VADD(Tx, Ty); TA = VSUB(Tw, Tz); TC = VADD(Tw, Tz); } { V T1j, T1k, T1l, T1m; T1j = VADD(Tr, TA); STM2(&(xo[14]), T1j, ovs, &(xo[2])); STN2(&(xo[12]), T1g, T1j, ovs); T1k = VSUB(TC, TB); STM2(&(xo[30]), T1k, ovs, &(xo[2])); STN2(&(xo[28]), T1f, T1k, ovs); T1l = VSUB(TA, Tr); STM2(&(xo[18]), T1l, ovs, &(xo[2])); STN2(&(xo[16]), T1b, T1l, ovs); T1m = VADD(TB, TC); STM2(&(xo[2]), T1m, ovs, &(xo[2])); STN2(&(xo[0]), T1c, T1m, ovs); } } { V TF, TJ, TI, TK; { V TD, TE, TG, TH; TD = VSUB(Tu, Tv); TE = VADD(Te, T7); TF = VADD(TD, TE); TJ = VSUB(TD, TE); TG = VADD(Tp, Tm); TH = VSUB(Ty, Tx); TI = VBYI(VADD(TG, TH)); TK = VBYI(VSUB(TH, TG)); } { V T1n, T1o, T1p, T1q; T1n = VSUB(TF, TI); STM2(&(xo[26]), T1n, ovs, &(xo[2])); STN2(&(xo[24]), T1d, T1n, ovs); T1o = VADD(TJ, TK); STM2(&(xo[10]), T1o, ovs, &(xo[2])); STN2(&(xo[8]), T1e, T1o, ovs); T1p = VADD(TF, TI); STM2(&(xo[6]), T1p, ovs, &(xo[2])); STN2(&(xo[4]), T1h, T1p, ovs); T1q = VSUB(TJ, TK); STM2(&(xo[22]), T1q, ovs, &(xo[2])); STN2(&(xo[20]), T1i, T1q, ovs); } } } } } }
static void n2bv_32(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs) { DVK(KP555570233, +0.555570233019602224742830813948532874374937191); DVK(KP831469612, +0.831469612302545237078788377617905756738560812); DVK(KP195090322, +0.195090322016128267848284868477022240927691618); DVK(KP980785280, +0.980785280403230449126182236134239036973933731); DVK(KP382683432, +0.382683432365089771728459984030398866761344562); DVK(KP923879532, +0.923879532511286756128183189396788286822416626); DVK(KP707106781, +0.707106781186547524400844362104849039284835938); { INT i; const R *xi; R *xo; xi = ii; xo = io; for (i = v; i > 0; i = i - VL, xi = xi + (VL * ivs), xo = xo + (VL * ovs), MAKE_VOLATILE_STRIDE(64, is), MAKE_VOLATILE_STRIDE(64, os)) { V T2f, T2k, T2N, T2M, T19, T1B, Tb, T1p, TT, T1v, TY, T1w, T2E, T2F, T2G; V T24, T2o, TC, T1s, TH, T1t, T2B, T2C, T2D, T1X, T2n, T2I, T2J, Tq, T1A; V T14, T1q, T2c, T2l; { V T3, T2i, T18, T2j, T6, T2d, T9, T2e, T15, Ta; { V T1, T2, T16, T17; T1 = LD(&(xi[0]), ivs, &(xi[0])); T2 = LD(&(xi[WS(is, 16)]), ivs, &(xi[0])); T3 = VSUB(T1, T2); T2i = VADD(T1, T2); T16 = LD(&(xi[WS(is, 8)]), ivs, &(xi[0])); T17 = LD(&(xi[WS(is, 24)]), ivs, &(xi[0])); T18 = VSUB(T16, T17); T2j = VADD(T16, T17); } { V T4, T5, T7, T8; T4 = LD(&(xi[WS(is, 4)]), ivs, &(xi[0])); T5 = LD(&(xi[WS(is, 20)]), ivs, &(xi[0])); T6 = VSUB(T4, T5); T2d = VADD(T4, T5); T7 = LD(&(xi[WS(is, 28)]), ivs, &(xi[0])); T8 = LD(&(xi[WS(is, 12)]), ivs, &(xi[0])); T9 = VSUB(T7, T8); T2e = VADD(T7, T8); } T2f = VSUB(T2d, T2e); T2k = VSUB(T2i, T2j); T2N = VADD(T2d, T2e); T2M = VADD(T2i, T2j); T15 = VMUL(LDK(KP707106781), VSUB(T6, T9)); T19 = VSUB(T15, T18); T1B = VADD(T18, T15); Ta = VMUL(LDK(KP707106781), VADD(T6, T9)); Tb = VSUB(T3, Ta); T1p = VADD(T3, Ta); } { V TL, T21, TW, T1Y, TO, T22, TS, T1Z; { V TJ, TK, TU, TV; TJ = LD(&(xi[WS(is, 3)]), ivs, &(xi[WS(is, 1)])); TK = LD(&(xi[WS(is, 19)]), ivs, &(xi[WS(is, 1)])); TL = VSUB(TJ, TK); T21 = VADD(TJ, TK); TU = LD(&(xi[WS(is, 31)]), ivs, &(xi[WS(is, 1)])); TV = LD(&(xi[WS(is, 15)]), ivs, &(xi[WS(is, 1)])); TW = VSUB(TU, TV); T1Y = VADD(TU, TV); } { V TM, TN, TQ, TR; TM = LD(&(xi[WS(is, 27)]), ivs, &(xi[WS(is, 1)])); TN = LD(&(xi[WS(is, 11)]), ivs, &(xi[WS(is, 1)])); TO = VSUB(TM, TN); T22 = VADD(TM, TN); TQ = LD(&(xi[WS(is, 7)]), ivs, &(xi[WS(is, 1)])); TR = LD(&(xi[WS(is, 23)]), ivs, &(xi[WS(is, 1)])); TS = VSUB(TQ, TR); T1Z = VADD(TQ, TR); } { V TP, TX, T20, T23; TP = VMUL(LDK(KP707106781), VSUB(TL, TO)); TT = VSUB(TP, TS); T1v = VADD(TS, TP); TX = VMUL(LDK(KP707106781), VADD(TL, TO)); TY = VSUB(TW, TX); T1w = VADD(TW, TX); T2E = VADD(T1Y, T1Z); T2F = VADD(T21, T22); T2G = VSUB(T2E, T2F); T20 = VSUB(T1Y, T1Z); T23 = VSUB(T21, T22); T24 = VFMA(LDK(KP923879532), T20, VMUL(LDK(KP382683432), T23)); T2o = VFNMS(LDK(KP382683432), T20, VMUL(LDK(KP923879532), T23)); } } { V Tu, T1U, TF, T1R, Tx, T1V, TB, T1S; { V Ts, Tt, TD, TE; Ts = LD(&(xi[WS(is, 5)]), ivs, &(xi[WS(is, 1)])); Tt = LD(&(xi[WS(is, 21)]), ivs, &(xi[WS(is, 1)])); Tu = VSUB(Ts, Tt); T1U = VADD(Ts, Tt); TD = LD(&(xi[WS(is, 1)]), ivs, &(xi[WS(is, 1)])); TE = LD(&(xi[WS(is, 17)]), ivs, &(xi[WS(is, 1)])); TF = VSUB(TD, TE); T1R = VADD(TD, TE); } { V Tv, Tw, Tz, TA; Tv = LD(&(xi[WS(is, 29)]), ivs, &(xi[WS(is, 1)])); Tw = LD(&(xi[WS(is, 13)]), ivs, &(xi[WS(is, 1)])); Tx = VSUB(Tv, Tw); T1V = VADD(Tv, Tw); Tz = LD(&(xi[WS(is, 9)]), ivs, &(xi[WS(is, 1)])); TA = LD(&(xi[WS(is, 25)]), ivs, &(xi[WS(is, 1)])); TB = VSUB(Tz, TA); T1S = VADD(Tz, TA); } { V Ty, TG, T1T, T1W; Ty = VMUL(LDK(KP707106781), VSUB(Tu, Tx)); TC = VSUB(Ty, TB); T1s = VADD(TB, Ty); TG = VMUL(LDK(KP707106781), VADD(Tu, Tx)); TH = VSUB(TF, TG); T1t = VADD(TF, TG); T2B = VADD(T1R, T1S); T2C = VADD(T1U, T1V); T2D = VSUB(T2B, T2C); T1T = VSUB(T1R, T1S); T1W = VSUB(T1U, T1V); T1X = VFNMS(LDK(KP382683432), T1W, VMUL(LDK(KP923879532), T1T)); T2n = VFMA(LDK(KP382683432), T1T, VMUL(LDK(KP923879532), T1W)); } } { V Te, T26, To, T29, Th, T27, Tl, T2a, Ti, Tp; { V Tc, Td, Tm, Tn; Tc = LD(&(xi[WS(is, 2)]), ivs, &(xi[0])); Td = LD(&(xi[WS(is, 18)]), ivs, &(xi[0])); Te = VSUB(Tc, Td); T26 = VADD(Tc, Td); Tm = LD(&(xi[WS(is, 30)]), ivs, &(xi[0])); Tn = LD(&(xi[WS(is, 14)]), ivs, &(xi[0])); To = VSUB(Tm, Tn); T29 = VADD(Tm, Tn); } { V Tf, Tg, Tj, Tk; Tf = LD(&(xi[WS(is, 10)]), ivs, &(xi[0])); Tg = LD(&(xi[WS(is, 26)]), ivs, &(xi[0])); Th = VSUB(Tf, Tg); T27 = VADD(Tf, Tg); Tj = LD(&(xi[WS(is, 6)]), ivs, &(xi[0])); Tk = LD(&(xi[WS(is, 22)]), ivs, &(xi[0])); Tl = VSUB(Tj, Tk); T2a = VADD(Tj, Tk); } T2I = VADD(T26, T27); T2J = VADD(T29, T2a); Ti = VFMA(LDK(KP382683432), Te, VMUL(LDK(KP923879532), Th)); Tp = VFNMS(LDK(KP382683432), To, VMUL(LDK(KP923879532), Tl)); Tq = VSUB(Ti, Tp); T1A = VADD(Ti, Tp); { V T12, T13, T28, T2b; T12 = VFNMS(LDK(KP382683432), Th, VMUL(LDK(KP923879532), Te)); T13 = VFMA(LDK(KP923879532), To, VMUL(LDK(KP382683432), Tl)); T14 = VSUB(T12, T13); T1q = VADD(T12, T13); T28 = VSUB(T26, T27); T2b = VSUB(T29, T2a); T2c = VMUL(LDK(KP707106781), VSUB(T28, T2b)); T2l = VMUL(LDK(KP707106781), VADD(T28, T2b)); } } { V T31, T32, T33, T34, T35, T36, T37, T38, T39, T3a, T3b, T3c; { V T2L, T2R, T2Q, T2S; { V T2H, T2K, T2O, T2P; T2H = VMUL(LDK(KP707106781), VSUB(T2D, T2G)); T2K = VSUB(T2I, T2J); T2L = VBYI(VSUB(T2H, T2K)); T2R = VBYI(VADD(T2K, T2H)); T2O = VSUB(T2M, T2N); T2P = VMUL(LDK(KP707106781), VADD(T2D, T2G)); T2Q = VSUB(T2O, T2P); T2S = VADD(T2O, T2P); } T31 = VADD(T2L, T2Q); STM2(&(xo[24]), T31, ovs, &(xo[0])); T32 = VSUB(T2S, T2R); STM2(&(xo[56]), T32, ovs, &(xo[0])); T33 = VSUB(T2Q, T2L); STM2(&(xo[40]), T33, ovs, &(xo[0])); T34 = VADD(T2R, T2S); STM2(&(xo[8]), T34, ovs, &(xo[0])); } { V T2h, T2r, T2q, T2s; { V T25, T2g, T2m, T2p; T25 = VSUB(T1X, T24); T2g = VSUB(T2c, T2f); T2h = VBYI(VSUB(T25, T2g)); T2r = VBYI(VADD(T2g, T25)); T2m = VSUB(T2k, T2l); T2p = VSUB(T2n, T2o); T2q = VSUB(T2m, T2p); T2s = VADD(T2m, T2p); } T35 = VADD(T2h, T2q); STM2(&(xo[20]), T35, ovs, &(xo[0])); T36 = VSUB(T2s, T2r); STM2(&(xo[52]), T36, ovs, &(xo[0])); T37 = VSUB(T2q, T2h); STM2(&(xo[44]), T37, ovs, &(xo[0])); T38 = VADD(T2r, T2s); STM2(&(xo[12]), T38, ovs, &(xo[0])); } { V T2V, T2Z, T2Y, T30; { V T2T, T2U, T2W, T2X; T2T = VADD(T2M, T2N); T2U = VADD(T2I, T2J); T2V = VSUB(T2T, T2U); T2Z = VADD(T2T, T2U); T2W = VADD(T2B, T2C); T2X = VADD(T2E, T2F); T2Y = VBYI(VSUB(T2W, T2X)); T30 = VADD(T2W, T2X); } T39 = VSUB(T2V, T2Y); STM2(&(xo[48]), T39, ovs, &(xo[0])); T3a = VADD(T2Z, T30); STM2(&(xo[0]), T3a, ovs, &(xo[0])); T3b = VADD(T2V, T2Y); STM2(&(xo[16]), T3b, ovs, &(xo[0])); T3c = VSUB(T2Z, T30); STM2(&(xo[32]), T3c, ovs, &(xo[0])); } { V T3d, T3e, T3f, T3g; { V T2v, T2z, T2y, T2A; { V T2t, T2u, T2w, T2x; T2t = VADD(T2k, T2l); T2u = VADD(T1X, T24); T2v = VADD(T2t, T2u); T2z = VSUB(T2t, T2u); T2w = VADD(T2f, T2c); T2x = VADD(T2n, T2o); T2y = VBYI(VADD(T2w, T2x)); T2A = VBYI(VSUB(T2x, T2w)); } T3d = VSUB(T2v, T2y); STM2(&(xo[60]), T3d, ovs, &(xo[0])); T3e = VADD(T2z, T2A); STM2(&(xo[28]), T3e, ovs, &(xo[0])); T3f = VADD(T2v, T2y); STM2(&(xo[4]), T3f, ovs, &(xo[0])); T3g = VSUB(T2z, T2A); STM2(&(xo[36]), T3g, ovs, &(xo[0])); } { V T1r, T1C, T1M, T1K, T1F, T1N, T1y, T1J; T1r = VSUB(T1p, T1q); T1C = VSUB(T1A, T1B); T1M = VADD(T1p, T1q); T1K = VADD(T1B, T1A); { V T1D, T1E, T1u, T1x; T1D = VFNMS(LDK(KP195090322), T1s, VMUL(LDK(KP980785280), T1t)); T1E = VFMA(LDK(KP195090322), T1v, VMUL(LDK(KP980785280), T1w)); T1F = VSUB(T1D, T1E); T1N = VADD(T1D, T1E); T1u = VFMA(LDK(KP980785280), T1s, VMUL(LDK(KP195090322), T1t)); T1x = VFNMS(LDK(KP195090322), T1w, VMUL(LDK(KP980785280), T1v)); T1y = VSUB(T1u, T1x); T1J = VADD(T1u, T1x); } { V T1z, T1G, T3h, T3i; T1z = VADD(T1r, T1y); T1G = VBYI(VADD(T1C, T1F)); T3h = VSUB(T1z, T1G); STM2(&(xo[50]), T3h, ovs, &(xo[2])); STN2(&(xo[48]), T39, T3h, ovs); T3i = VADD(T1z, T1G); STM2(&(xo[14]), T3i, ovs, &(xo[2])); STN2(&(xo[12]), T38, T3i, ovs); } { V T1P, T1Q, T3j, T3k; T1P = VBYI(VADD(T1K, T1J)); T1Q = VADD(T1M, T1N); T3j = VADD(T1P, T1Q); STM2(&(xo[2]), T3j, ovs, &(xo[2])); STN2(&(xo[0]), T3a, T3j, ovs); T3k = VSUB(T1Q, T1P); STM2(&(xo[62]), T3k, ovs, &(xo[2])); STN2(&(xo[60]), T3d, T3k, ovs); } { V T1H, T1I, T3l, T3m; T1H = VSUB(T1r, T1y); T1I = VBYI(VSUB(T1F, T1C)); T3l = VSUB(T1H, T1I); STM2(&(xo[46]), T3l, ovs, &(xo[2])); STN2(&(xo[44]), T37, T3l, ovs); T3m = VADD(T1H, T1I); STM2(&(xo[18]), T3m, ovs, &(xo[2])); STN2(&(xo[16]), T3b, T3m, ovs); } { V T1L, T1O, T3n, T3o; T1L = VBYI(VSUB(T1J, T1K)); T1O = VSUB(T1M, T1N); T3n = VADD(T1L, T1O); STM2(&(xo[30]), T3n, ovs, &(xo[2])); STN2(&(xo[28]), T3e, T3n, ovs); T3o = VSUB(T1O, T1L); STM2(&(xo[34]), T3o, ovs, &(xo[2])); STN2(&(xo[32]), T3c, T3o, ovs); } } { V Tr, T1a, T1k, T1i, T1d, T1l, T10, T1h; Tr = VSUB(Tb, Tq); T1a = VSUB(T14, T19); T1k = VADD(Tb, Tq); T1i = VADD(T19, T14); { V T1b, T1c, TI, TZ; T1b = VFNMS(LDK(KP555570233), TC, VMUL(LDK(KP831469612), TH)); T1c = VFMA(LDK(KP555570233), TT, VMUL(LDK(KP831469612), TY)); T1d = VSUB(T1b, T1c); T1l = VADD(T1b, T1c); TI = VFMA(LDK(KP831469612), TC, VMUL(LDK(KP555570233), TH)); TZ = VFNMS(LDK(KP555570233), TY, VMUL(LDK(KP831469612), TT)); T10 = VSUB(TI, TZ); T1h = VADD(TI, TZ); } { V T11, T1e, T3p, T3q; T11 = VADD(Tr, T10); T1e = VBYI(VADD(T1a, T1d)); T3p = VSUB(T11, T1e); STM2(&(xo[54]), T3p, ovs, &(xo[2])); STN2(&(xo[52]), T36, T3p, ovs); T3q = VADD(T11, T1e); STM2(&(xo[10]), T3q, ovs, &(xo[2])); STN2(&(xo[8]), T34, T3q, ovs); } { V T1n, T1o, T3r, T3s; T1n = VBYI(VADD(T1i, T1h)); T1o = VADD(T1k, T1l); T3r = VADD(T1n, T1o); STM2(&(xo[6]), T3r, ovs, &(xo[2])); STN2(&(xo[4]), T3f, T3r, ovs); T3s = VSUB(T1o, T1n); STM2(&(xo[58]), T3s, ovs, &(xo[2])); STN2(&(xo[56]), T32, T3s, ovs); } { V T1f, T1g, T3t, T3u; T1f = VSUB(Tr, T10); T1g = VBYI(VSUB(T1d, T1a)); T3t = VSUB(T1f, T1g); STM2(&(xo[42]), T3t, ovs, &(xo[2])); STN2(&(xo[40]), T33, T3t, ovs); T3u = VADD(T1f, T1g); STM2(&(xo[22]), T3u, ovs, &(xo[2])); STN2(&(xo[20]), T35, T3u, ovs); } { V T1j, T1m, T3v, T3w; T1j = VBYI(VSUB(T1h, T1i)); T1m = VSUB(T1k, T1l); T3v = VADD(T1j, T1m); STM2(&(xo[26]), T3v, ovs, &(xo[2])); STN2(&(xo[24]), T31, T3v, ovs); T3w = VSUB(T1m, T1j); STM2(&(xo[38]), T3w, ovs, &(xo[2])); STN2(&(xo[36]), T3g, T3w, ovs); } } } } } } VLEAVE(); }
static void n2fv_20(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs) { DVK(KP587785252, +0.587785252292473129168705954639072768597652438); DVK(KP951056516, +0.951056516295153572116439333379382143405698634); DVK(KP250000000, +0.250000000000000000000000000000000000000000000); DVK(KP559016994, +0.559016994374947424102293417182819058860154590); { 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(40, is), MAKE_VOLATILE_STRIDE(40, os)) { V T3, T1B, Tm, T1i, TG, TN, TO, TH, T13, T16, T1k, T1u, T1v, T1z, T1r; V T1s, T1y, T1a, T1d, T1j, Ti, TD, TB, TL; { V T1, T2, T1g, Tk, Tl, T1h; T1 = LD(&(xi[0]), ivs, &(xi[0])); T2 = LD(&(xi[WS(is, 10)]), ivs, &(xi[0])); T1g = VADD(T1, T2); Tk = LD(&(xi[WS(is, 5)]), ivs, &(xi[WS(is, 1)])); Tl = LD(&(xi[WS(is, 15)]), ivs, &(xi[WS(is, 1)])); T1h = VADD(Tk, Tl); T3 = VSUB(T1, T2); T1B = VADD(T1g, T1h); Tm = VSUB(Tk, Tl); T1i = VSUB(T1g, T1h); } { V T6, T18, Tw, T12, Tz, T15, T9, T1b, Td, T11, Tp, T19, Ts, T1c, Tg; V T14; { V T4, T5, Tu, Tv; T4 = LD(&(xi[WS(is, 4)]), ivs, &(xi[0])); T5 = LD(&(xi[WS(is, 14)]), ivs, &(xi[0])); T6 = VSUB(T4, T5); T18 = VADD(T4, T5); Tu = LD(&(xi[WS(is, 13)]), ivs, &(xi[WS(is, 1)])); Tv = LD(&(xi[WS(is, 3)]), ivs, &(xi[WS(is, 1)])); Tw = VSUB(Tu, Tv); T12 = VADD(Tu, Tv); } { V Tx, Ty, T7, T8; Tx = LD(&(xi[WS(is, 17)]), ivs, &(xi[WS(is, 1)])); Ty = LD(&(xi[WS(is, 7)]), ivs, &(xi[WS(is, 1)])); Tz = VSUB(Tx, Ty); T15 = VADD(Tx, Ty); T7 = LD(&(xi[WS(is, 16)]), ivs, &(xi[0])); T8 = LD(&(xi[WS(is, 6)]), ivs, &(xi[0])); T9 = VSUB(T7, T8); T1b = VADD(T7, T8); } { V Tb, Tc, Tn, To; Tb = LD(&(xi[WS(is, 8)]), ivs, &(xi[0])); Tc = LD(&(xi[WS(is, 18)]), ivs, &(xi[0])); Td = VSUB(Tb, Tc); T11 = VADD(Tb, Tc); Tn = LD(&(xi[WS(is, 9)]), ivs, &(xi[WS(is, 1)])); To = LD(&(xi[WS(is, 19)]), ivs, &(xi[WS(is, 1)])); Tp = VSUB(Tn, To); T19 = VADD(Tn, To); } { V Tq, Tr, Te, Tf; Tq = LD(&(xi[WS(is, 1)]), ivs, &(xi[WS(is, 1)])); Tr = LD(&(xi[WS(is, 11)]), ivs, &(xi[WS(is, 1)])); Ts = VSUB(Tq, Tr); T1c = VADD(Tq, Tr); Te = LD(&(xi[WS(is, 12)]), ivs, &(xi[0])); Tf = LD(&(xi[WS(is, 2)]), ivs, &(xi[0])); Tg = VSUB(Te, Tf); T14 = VADD(Te, Tf); } TG = VSUB(Ts, Tp); TN = VSUB(T6, T9); TO = VSUB(Td, Tg); TH = VSUB(Tz, Tw); T13 = VSUB(T11, T12); T16 = VSUB(T14, T15); T1k = VADD(T13, T16); T1u = VADD(T11, T12); T1v = VADD(T14, T15); T1z = VADD(T1u, T1v); T1r = VADD(T18, T19); T1s = VADD(T1b, T1c); T1y = VADD(T1r, T1s); T1a = VSUB(T18, T19); T1d = VSUB(T1b, T1c); T1j = VADD(T1a, T1d); { V Ta, Th, Tt, TA; Ta = VADD(T6, T9); Th = VADD(Td, Tg); Ti = VADD(Ta, Th); TD = VMUL(LDK(KP559016994), VSUB(Ta, Th)); Tt = VADD(Tp, Ts); TA = VADD(Tw, Tz); TB = VADD(Tt, TA); TL = VMUL(LDK(KP559016994), VSUB(TA, Tt)); } } { V T1I, T1J, T1K, T1L, T1N, T1H, Tj, TC; Tj = VADD(T3, Ti); TC = VBYI(VADD(Tm, TB)); T1H = VSUB(Tj, TC); STM2(&(xo[10]), T1H, ovs, &(xo[2])); T1I = VADD(Tj, TC); STM2(&(xo[30]), T1I, ovs, &(xo[2])); { V T1A, T1C, T1D, T1x, T1G, T1t, T1w, T1F, T1E, T1M; T1A = VMUL(LDK(KP559016994), VSUB(T1y, T1z)); T1C = VADD(T1y, T1z); T1D = VFNMS(LDK(KP250000000), T1C, T1B); T1t = VSUB(T1r, T1s); T1w = VSUB(T1u, T1v); T1x = VBYI(VFMA(LDK(KP951056516), T1t, VMUL(LDK(KP587785252), T1w))); T1G = VBYI(VFNMS(LDK(KP587785252), T1t, VMUL(LDK(KP951056516), T1w))); T1J = VADD(T1B, T1C); STM2(&(xo[0]), T1J, ovs, &(xo[0])); T1F = VSUB(T1D, T1A); T1K = VSUB(T1F, T1G); STM2(&(xo[16]), T1K, ovs, &(xo[0])); T1L = VADD(T1G, T1F); STM2(&(xo[24]), T1L, ovs, &(xo[0])); T1E = VADD(T1A, T1D); T1M = VADD(T1x, T1E); STM2(&(xo[8]), T1M, ovs, &(xo[0])); STN2(&(xo[8]), T1M, T1H, ovs); T1N = VSUB(T1E, T1x); STM2(&(xo[32]), T1N, ovs, &(xo[0])); } { V T1O, T1P, T1R, T1S; { V T1n, T1l, T1m, T1f, T1q, T17, T1e, T1p, T1Q, T1o; T1n = VMUL(LDK(KP559016994), VSUB(T1j, T1k)); T1l = VADD(T1j, T1k); T1m = VFNMS(LDK(KP250000000), T1l, T1i); T17 = VSUB(T13, T16); T1e = VSUB(T1a, T1d); T1f = VBYI(VFNMS(LDK(KP587785252), T1e, VMUL(LDK(KP951056516), T17))); T1q = VBYI(VFMA(LDK(KP951056516), T1e, VMUL(LDK(KP587785252), T17))); T1O = VADD(T1i, T1l); STM2(&(xo[20]), T1O, ovs, &(xo[0])); T1p = VADD(T1n, T1m); T1P = VSUB(T1p, T1q); STM2(&(xo[12]), T1P, ovs, &(xo[0])); T1Q = VADD(T1q, T1p); STM2(&(xo[28]), T1Q, ovs, &(xo[0])); STN2(&(xo[28]), T1Q, T1I, ovs); T1o = VSUB(T1m, T1n); T1R = VADD(T1f, T1o); STM2(&(xo[4]), T1R, ovs, &(xo[0])); T1S = VSUB(T1o, T1f); STM2(&(xo[36]), T1S, ovs, &(xo[0])); } { V TI, TP, TX, TU, TM, TW, TF, TT, TK, TE; TI = VFMA(LDK(KP951056516), TG, VMUL(LDK(KP587785252), TH)); TP = VFMA(LDK(KP951056516), TN, VMUL(LDK(KP587785252), TO)); TX = VFNMS(LDK(KP587785252), TN, VMUL(LDK(KP951056516), TO)); TU = VFNMS(LDK(KP587785252), TG, VMUL(LDK(KP951056516), TH)); TK = VFMS(LDK(KP250000000), TB, Tm); TM = VADD(TK, TL); TW = VSUB(TL, TK); TE = VFNMS(LDK(KP250000000), Ti, T3); TF = VADD(TD, TE); TT = VSUB(TE, TD); { V TJ, TQ, T1T, T1U; TJ = VADD(TF, TI); TQ = VBYI(VSUB(TM, TP)); T1T = VSUB(TJ, TQ); STM2(&(xo[38]), T1T, ovs, &(xo[2])); STN2(&(xo[36]), T1S, T1T, ovs); T1U = VADD(TJ, TQ); STM2(&(xo[2]), T1U, ovs, &(xo[2])); STN2(&(xo[0]), T1J, T1U, ovs); } { V TZ, T10, T1V, T1W; TZ = VADD(TT, TU); T10 = VBYI(VADD(TX, TW)); T1V = VSUB(TZ, T10); STM2(&(xo[26]), T1V, ovs, &(xo[2])); STN2(&(xo[24]), T1L, T1V, ovs); T1W = VADD(TZ, T10); STM2(&(xo[14]), T1W, ovs, &(xo[2])); STN2(&(xo[12]), T1P, T1W, ovs); } { V TR, TS, T1X, T1Y; TR = VSUB(TF, TI); TS = VBYI(VADD(TP, TM)); T1X = VSUB(TR, TS); STM2(&(xo[22]), T1X, ovs, &(xo[2])); STN2(&(xo[20]), T1O, T1X, ovs); T1Y = VADD(TR, TS); STM2(&(xo[18]), T1Y, ovs, &(xo[2])); STN2(&(xo[16]), T1K, T1Y, ovs); } { V TV, TY, T1Z, T20; TV = VSUB(TT, TU); TY = VBYI(VSUB(TW, TX)); T1Z = VSUB(TV, TY); STM2(&(xo[34]), T1Z, ovs, &(xo[2])); STN2(&(xo[32]), T1N, T1Z, ovs); T20 = VADD(TV, TY); STM2(&(xo[6]), T20, ovs, &(xo[2])); STN2(&(xo[4]), T1R, T20, ovs); } } } } } } VLEAVE(); }
static void n2bv_8(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs) { DVK(KP707106781, +0.707106781186547524400844362104849039284835938); INT i; const R *xi; R *xo; xi = ii; xo = io; 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 T1, T2, Tc, Td, T4, T5, T7, T8; T1 = LD(&(xi[0]), ivs, &(xi[0])); T2 = LD(&(xi[WS(is, 4)]), ivs, &(xi[0])); Tc = LD(&(xi[WS(is, 2)]), ivs, &(xi[0])); Td = LD(&(xi[WS(is, 6)]), ivs, &(xi[0])); T4 = LD(&(xi[WS(is, 1)]), ivs, &(xi[WS(is, 1)])); T5 = LD(&(xi[WS(is, 5)]), ivs, &(xi[WS(is, 1)])); T7 = LD(&(xi[WS(is, 7)]), ivs, &(xi[WS(is, 1)])); T8 = LD(&(xi[WS(is, 3)]), ivs, &(xi[WS(is, 1)])); { V T3, Tj, Te, Tk, T6, Tm, T9, Tn, Tp, Tl; T3 = VSUB(T1, T2); Tj = VADD(T1, T2); Te = VSUB(Tc, Td); Tk = VADD(Tc, Td); T6 = VSUB(T4, T5); Tm = VADD(T4, T5); T9 = VSUB(T7, T8); Tn = VADD(T7, T8); Tp = VADD(Tj, Tk); Tl = VSUB(Tj, Tk); { V Tq, To, Ta, Tf; Tq = VADD(Tm, Tn); To = VSUB(Tm, Tn); Ta = VADD(T6, T9); Tf = VSUB(T6, T9); { V Tr, Ts, Tt, Tu, Tg, Ti, Tb, Th; Tr = VFMAI(To, Tl); STM2(&(xo[4]), Tr, ovs, &(xo[0])); Ts = VFNMSI(To, Tl); STM2(&(xo[12]), Ts, ovs, &(xo[0])); Tt = VADD(Tp, Tq); STM2(&(xo[0]), Tt, ovs, &(xo[0])); Tu = VSUB(Tp, Tq); STM2(&(xo[8]), Tu, ovs, &(xo[0])); Tg = VFNMS(LDK(KP707106781), Tf, Te); Ti = VFMA(LDK(KP707106781), Tf, Te); Tb = VFNMS(LDK(KP707106781), Ta, T3); Th = VFMA(LDK(KP707106781), Ta, T3); { V Tv, Tw, Tx, Ty; Tv = VFNMSI(Ti, Th); STM2(&(xo[14]), Tv, ovs, &(xo[2])); STN2(&(xo[12]), Ts, Tv, ovs); Tw = VFMAI(Ti, Th); STM2(&(xo[2]), Tw, ovs, &(xo[2])); STN2(&(xo[0]), Tt, Tw, ovs); Tx = VFMAI(Tg, Tb); STM2(&(xo[10]), Tx, ovs, &(xo[2])); STN2(&(xo[8]), Tu, Tx, ovs); Ty = VFNMSI(Tg, Tb); STM2(&(xo[6]), Ty, ovs, &(xo[2])); STN2(&(xo[4]), Tr, Ty, ovs); } } } } } }
static void n2fv_8(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs) { DVK(KP707106781, +0.707106781186547524400844362104849039284835938); { 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 T3, Tj, Tf, Tk, Ta, Tn, Tc, Tm, Ts, Tu; { V T1, T2, Td, Te; T1 = LD(&(xi[0]), ivs, &(xi[0])); T2 = LD(&(xi[WS(is, 4)]), ivs, &(xi[0])); T3 = VSUB(T1, T2); Tj = VADD(T1, T2); Td = LD(&(xi[WS(is, 2)]), ivs, &(xi[0])); Te = LD(&(xi[WS(is, 6)]), ivs, &(xi[0])); Tf = VSUB(Td, Te); Tk = VADD(Td, Te); { V T4, T5, T6, T7, T8, T9; T4 = LD(&(xi[WS(is, 1)]), ivs, &(xi[WS(is, 1)])); T5 = LD(&(xi[WS(is, 5)]), ivs, &(xi[WS(is, 1)])); T6 = VSUB(T4, T5); T7 = LD(&(xi[WS(is, 7)]), ivs, &(xi[WS(is, 1)])); T8 = LD(&(xi[WS(is, 3)]), ivs, &(xi[WS(is, 1)])); T9 = VSUB(T7, T8); Ta = VMUL(LDK(KP707106781), VADD(T6, T9)); Tn = VADD(T7, T8); Tc = VMUL(LDK(KP707106781), VSUB(T9, T6)); Tm = VADD(T4, T5); } } { V Tr, Tb, Tg, Tp, Tq, Tt; Tb = VADD(T3, Ta); Tg = VBYI(VSUB(Tc, Tf)); Tr = VSUB(Tb, Tg); STM2(&(xo[14]), Tr, ovs, &(xo[2])); Ts = VADD(Tb, Tg); STM2(&(xo[2]), Ts, ovs, &(xo[2])); Tp = VSUB(Tj, Tk); Tq = VBYI(VSUB(Tn, Tm)); Tt = VSUB(Tp, Tq); STM2(&(xo[12]), Tt, ovs, &(xo[0])); STN2(&(xo[12]), Tt, Tr, ovs); Tu = VADD(Tp, Tq); STM2(&(xo[4]), Tu, ovs, &(xo[0])); } { V Tv, Th, Ti, Tw; Th = VSUB(T3, Ta); Ti = VBYI(VADD(Tf, Tc)); Tv = VSUB(Th, Ti); STM2(&(xo[10]), Tv, ovs, &(xo[2])); Tw = VADD(Th, Ti); STM2(&(xo[6]), Tw, ovs, &(xo[2])); STN2(&(xo[4]), Tu, Tw, ovs); { V Tl, To, Tx, Ty; Tl = VADD(Tj, Tk); To = VADD(Tm, Tn); Tx = VSUB(Tl, To); STM2(&(xo[8]), Tx, ovs, &(xo[0])); STN2(&(xo[8]), Tx, Tv, ovs); Ty = VADD(Tl, To); STM2(&(xo[0]), Ty, ovs, &(xo[0])); STN2(&(xo[0]), Ty, Ts, ovs); } } } } VLEAVE(); }
static void n2bv_10(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs) { DVK(KP559016994, +0.559016994374947424102293417182819058860154590); DVK(KP250000000, +0.250000000000000000000000000000000000000000000); DVK(KP618033988, +0.618033988749894848204586834365638117720309180); DVK(KP951056516, +0.951056516295153572116439333379382143405698634); INT i; const R *xi; R *xo; xi = ii; xo = io; 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 Tb, Tr, T3, Ts, T6, Tw, Tg, Tt, T9, Tc, T1, T2; T1 = LD(&(xi[0]), ivs, &(xi[0])); T2 = LD(&(xi[WS(is, 5)]), ivs, &(xi[WS(is, 1)])); { V T4, T5, Te, Tf, T7, T8; T4 = LD(&(xi[WS(is, 2)]), ivs, &(xi[0])); T5 = LD(&(xi[WS(is, 7)]), ivs, &(xi[WS(is, 1)])); Te = LD(&(xi[WS(is, 6)]), ivs, &(xi[0])); Tf = LD(&(xi[WS(is, 1)]), ivs, &(xi[WS(is, 1)])); T7 = LD(&(xi[WS(is, 8)]), ivs, &(xi[0])); T8 = LD(&(xi[WS(is, 3)]), ivs, &(xi[WS(is, 1)])); Tb = LD(&(xi[WS(is, 4)]), ivs, &(xi[0])); Tr = VADD(T1, T2); T3 = VSUB(T1, T2); Ts = VADD(T4, T5); T6 = VSUB(T4, T5); Tw = VADD(Te, Tf); Tg = VSUB(Te, Tf); Tt = VADD(T7, T8); T9 = VSUB(T7, T8); Tc = LD(&(xi[WS(is, 9)]), ivs, &(xi[WS(is, 1)])); } { V TD, Tu, Tm, Ta, Td, Tv; TD = VSUB(Ts, Tt); Tu = VADD(Ts, Tt); Tm = VSUB(T6, T9); Ta = VADD(T6, T9); Td = VSUB(Tb, Tc); Tv = VADD(Tb, Tc); { V TC, Tx, Tn, Th; TC = VSUB(Tv, Tw); Tx = VADD(Tv, Tw); Tn = VSUB(Td, Tg); Th = VADD(Td, Tg); { V Ty, TA, TE, TG, Ti, Tk, To, Tq; Ty = VADD(Tu, Tx); TA = VSUB(Tu, Tx); TE = VMUL(LDK(KP951056516), VFNMS(LDK(KP618033988), TD, TC)); TG = VMUL(LDK(KP951056516), VFMA(LDK(KP618033988), TC, TD)); Ti = VADD(Ta, Th); Tk = VSUB(Ta, Th); To = VMUL(LDK(KP951056516), VFMA(LDK(KP618033988), Tn, Tm)); Tq = VMUL(LDK(KP951056516), VFNMS(LDK(KP618033988), Tm, Tn)); { V Tz, TH, Tj, TI; Tz = VFNMS(LDK(KP250000000), Ty, Tr); TH = VADD(Tr, Ty); STM2(&(xo[0]), TH, ovs, &(xo[0])); Tj = VFNMS(LDK(KP250000000), Ti, T3); TI = VADD(T3, Ti); STM2(&(xo[10]), TI, ovs, &(xo[2])); { V TB, TF, Tl, Tp; TB = VFNMS(LDK(KP559016994), TA, Tz); TF = VFMA(LDK(KP559016994), TA, Tz); Tl = VFMA(LDK(KP559016994), Tk, Tj); Tp = VFNMS(LDK(KP559016994), Tk, Tj); { V TJ, TK, TL, TM; TJ = VFNMSI(TG, TF); STM2(&(xo[8]), TJ, ovs, &(xo[0])); STN2(&(xo[8]), TJ, TI, ovs); TK = VFMAI(TG, TF); STM2(&(xo[12]), TK, ovs, &(xo[0])); TL = VFMAI(TE, TB); STM2(&(xo[16]), TL, ovs, &(xo[0])); TM = VFNMSI(TE, TB); STM2(&(xo[4]), TM, ovs, &(xo[0])); { V TN, TO, TP, TQ; TN = VFMAI(Tq, Tp); STM2(&(xo[6]), TN, ovs, &(xo[2])); STN2(&(xo[4]), TM, TN, ovs); TO = VFNMSI(Tq, Tp); STM2(&(xo[14]), TO, ovs, &(xo[2])); STN2(&(xo[12]), TK, TO, ovs); TP = VFNMSI(To, Tl); STM2(&(xo[18]), TP, ovs, &(xo[2])); STN2(&(xo[16]), TL, TP, ovs); TQ = VFMAI(To, Tl); STM2(&(xo[2]), TQ, ovs, &(xo[2])); STN2(&(xo[0]), TH, TQ, ovs); } } } } } } } } }
static void n2fv_20(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs) { DVK(KP559016994, +0.559016994374947424102293417182819058860154590); DVK(KP618033988, +0.618033988749894848204586834365638117720309180); DVK(KP951056516, +0.951056516295153572116439333379382143405698634); DVK(KP250000000, +0.250000000000000000000000000000000000000000000); { 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(40, is), MAKE_VOLATILE_STRIDE(40, os)) { V T1H, T1I, TU, TI, TP, TX, T1M, T1N, T1O, T1P, T1R, T1S, TM, TW, TT; V TF; { V T3, Tm, T1r, T13, Ta, TN, TH, TA, TG, Tt, Th, TO, T1u, T1C, T1n; V T1a, T1m, T1h, T1x, T1D, TE, Ti; { V T1, T2, Tk, Tl; T1 = LD(&(xi[0]), ivs, &(xi[0])); T2 = LD(&(xi[WS(is, 10)]), ivs, &(xi[0])); Tk = LD(&(xi[WS(is, 5)]), ivs, &(xi[WS(is, 1)])); Tl = LD(&(xi[WS(is, 15)]), ivs, &(xi[WS(is, 1)])); { V T14, T6, T1c, Tw, Tn, T1f, Tz, T17, T9, To, Tq, T1b, Td, Tr, Te; V Tf, T15, Tp; { V Tx, Ty, T7, T8, Tb, Tc; { V T4, T5, Tu, Tv, T11, T12; T4 = LD(&(xi[WS(is, 4)]), ivs, &(xi[0])); T5 = LD(&(xi[WS(is, 14)]), ivs, &(xi[0])); Tu = LD(&(xi[WS(is, 13)]), ivs, &(xi[WS(is, 1)])); Tv = LD(&(xi[WS(is, 3)]), ivs, &(xi[WS(is, 1)])); Tx = LD(&(xi[WS(is, 17)]), ivs, &(xi[WS(is, 1)])); T3 = VSUB(T1, T2); T11 = VADD(T1, T2); Tm = VSUB(Tk, Tl); T12 = VADD(Tk, Tl); T14 = VADD(T4, T5); T6 = VSUB(T4, T5); T1c = VADD(Tu, Tv); Tw = VSUB(Tu, Tv); Ty = LD(&(xi[WS(is, 7)]), ivs, &(xi[WS(is, 1)])); T7 = LD(&(xi[WS(is, 16)]), ivs, &(xi[0])); T8 = LD(&(xi[WS(is, 6)]), ivs, &(xi[0])); T1r = VADD(T11, T12); T13 = VSUB(T11, T12); } Tb = LD(&(xi[WS(is, 8)]), ivs, &(xi[0])); Tc = LD(&(xi[WS(is, 18)]), ivs, &(xi[0])); Tn = LD(&(xi[WS(is, 9)]), ivs, &(xi[WS(is, 1)])); T1f = VADD(Tx, Ty); Tz = VSUB(Tx, Ty); T17 = VADD(T7, T8); T9 = VSUB(T7, T8); To = LD(&(xi[WS(is, 19)]), ivs, &(xi[WS(is, 1)])); Tq = LD(&(xi[WS(is, 1)]), ivs, &(xi[WS(is, 1)])); T1b = VADD(Tb, Tc); Td = VSUB(Tb, Tc); Tr = LD(&(xi[WS(is, 11)]), ivs, &(xi[WS(is, 1)])); Te = LD(&(xi[WS(is, 12)]), ivs, &(xi[0])); Tf = LD(&(xi[WS(is, 2)]), ivs, &(xi[0])); } Ta = VADD(T6, T9); TN = VSUB(T6, T9); T15 = VADD(Tn, To); Tp = VSUB(Tn, To); TH = VSUB(Tz, Tw); TA = VADD(Tw, Tz); { V T1d, T1v, T18, Ts, T1e, Tg, T16, T1s; T1d = VSUB(T1b, T1c); T1v = VADD(T1b, T1c); T18 = VADD(Tq, Tr); Ts = VSUB(Tq, Tr); T1e = VADD(Te, Tf); Tg = VSUB(Te, Tf); T16 = VSUB(T14, T15); T1s = VADD(T14, T15); { V T1t, T19, T1w, T1g; T1t = VADD(T17, T18); T19 = VSUB(T17, T18); TG = VSUB(Ts, Tp); Tt = VADD(Tp, Ts); T1w = VADD(T1e, T1f); T1g = VSUB(T1e, T1f); Th = VADD(Td, Tg); TO = VSUB(Td, Tg); T1u = VADD(T1s, T1t); T1C = VSUB(T1s, T1t); T1n = VSUB(T16, T19); T1a = VADD(T16, T19); T1m = VSUB(T1d, T1g); T1h = VADD(T1d, T1g); T1x = VADD(T1v, T1w); T1D = VSUB(T1v, T1w); } } } } TE = VSUB(Ta, Th); Ti = VADD(Ta, Th); { V TL, T1k, T1A, Tj, TD, T1E, T1G, TK, TC, T1j, T1z, T1i, T1y, TB; TL = VSUB(TA, Tt); TB = VADD(Tt, TA); T1i = VADD(T1a, T1h); T1k = VSUB(T1a, T1h); T1y = VADD(T1u, T1x); T1A = VSUB(T1u, T1x); Tj = VADD(T3, Ti); TD = VFNMS(LDK(KP250000000), Ti, T3); T1E = VMUL(LDK(KP951056516), VFMA(LDK(KP618033988), T1D, T1C)); T1G = VMUL(LDK(KP951056516), VFNMS(LDK(KP618033988), T1C, T1D)); TK = VFNMS(LDK(KP250000000), TB, Tm); TC = VADD(Tm, TB); T1j = VFNMS(LDK(KP250000000), T1i, T13); T1H = VADD(T1r, T1y); STM2(&(xo[0]), T1H, ovs, &(xo[0])); T1z = VFNMS(LDK(KP250000000), T1y, T1r); T1I = VADD(T13, T1i); STM2(&(xo[20]), T1I, ovs, &(xo[0])); { V T1J, T1K, T1p, T1l, T1o, T1q, T1F, T1B, T1L, T1Q; TU = VFNMS(LDK(KP618033988), TG, TH); TI = VFMA(LDK(KP618033988), TH, TG); TP = VFMA(LDK(KP618033988), TO, TN); TX = VFNMS(LDK(KP618033988), TN, TO); T1J = VFMAI(TC, Tj); STM2(&(xo[30]), T1J, ovs, &(xo[2])); T1K = VFNMSI(TC, Tj); STM2(&(xo[10]), T1K, ovs, &(xo[2])); T1p = VFMA(LDK(KP559016994), T1k, T1j); T1l = VFNMS(LDK(KP559016994), T1k, T1j); T1o = VMUL(LDK(KP951056516), VFNMS(LDK(KP618033988), T1n, T1m)); T1q = VMUL(LDK(KP951056516), VFMA(LDK(KP618033988), T1m, T1n)); T1F = VFNMS(LDK(KP559016994), T1A, T1z); T1B = VFMA(LDK(KP559016994), T1A, T1z); T1L = VFMAI(T1q, T1p); STM2(&(xo[28]), T1L, ovs, &(xo[0])); STN2(&(xo[28]), T1L, T1J, ovs); T1M = VFNMSI(T1q, T1p); STM2(&(xo[12]), T1M, ovs, &(xo[0])); T1N = VFNMSI(T1o, T1l); STM2(&(xo[36]), T1N, ovs, &(xo[0])); T1O = VFMAI(T1o, T1l); STM2(&(xo[4]), T1O, ovs, &(xo[0])); T1P = VFNMSI(T1E, T1B); STM2(&(xo[32]), T1P, ovs, &(xo[0])); T1Q = VFMAI(T1E, T1B); STM2(&(xo[8]), T1Q, ovs, &(xo[0])); STN2(&(xo[8]), T1Q, T1K, ovs); T1R = VFMAI(T1G, T1F); STM2(&(xo[24]), T1R, ovs, &(xo[0])); T1S = VFNMSI(T1G, T1F); STM2(&(xo[16]), T1S, ovs, &(xo[0])); TM = VFNMS(LDK(KP559016994), TL, TK); TW = VFMA(LDK(KP559016994), TL, TK); TT = VFNMS(LDK(KP559016994), TE, TD); TF = VFMA(LDK(KP559016994), TE, TD); } } } { V T10, TY, TQ, TS, TJ, TR, TZ, TV; T10 = VFMA(LDK(KP951056516), TX, TW); TY = VFNMS(LDK(KP951056516), TX, TW); TQ = VFMA(LDK(KP951056516), TP, TM); TS = VFNMS(LDK(KP951056516), TP, TM); TJ = VFMA(LDK(KP951056516), TI, TF); TR = VFNMS(LDK(KP951056516), TI, TF); TZ = VFMA(LDK(KP951056516), TU, TT); TV = VFNMS(LDK(KP951056516), TU, TT); { V T1T, T1U, T1V, T1W; T1T = VFMAI(TS, TR); STM2(&(xo[22]), T1T, ovs, &(xo[2])); STN2(&(xo[20]), T1I, T1T, ovs); T1U = VFNMSI(TS, TR); STM2(&(xo[18]), T1U, ovs, &(xo[2])); STN2(&(xo[16]), T1S, T1U, ovs); T1V = VFMAI(TQ, TJ); STM2(&(xo[38]), T1V, ovs, &(xo[2])); STN2(&(xo[36]), T1N, T1V, ovs); T1W = VFNMSI(TQ, TJ); STM2(&(xo[2]), T1W, ovs, &(xo[2])); STN2(&(xo[0]), T1H, T1W, ovs); { V T1X, T1Y, T1Z, T20; T1X = VFMAI(TY, TV); STM2(&(xo[6]), T1X, ovs, &(xo[2])); STN2(&(xo[4]), T1O, T1X, ovs); T1Y = VFNMSI(TY, TV); STM2(&(xo[34]), T1Y, ovs, &(xo[2])); STN2(&(xo[32]), T1P, T1Y, ovs); T1Z = VFMAI(T10, TZ); STM2(&(xo[14]), T1Z, ovs, &(xo[2])); STN2(&(xo[12]), T1M, T1Z, ovs); T20 = VFNMSI(T10, TZ); STM2(&(xo[26]), T20, ovs, &(xo[2])); STN2(&(xo[24]), T1R, T20, ovs); } } } } } VLEAVE(); }
static void n2bv_12(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs) { DVK(KP866025403, +0.866025403784438646763723170752936183471402627); DVK(KP500000000, +0.500000000000000000000000000000000000000000000); INT i; const R *xi; R *xo; xi = ii; xo = io; 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 T1, T6, Tc, Th, Td, Te, Ti, Tz, T4, TA, T9, Tj, Tf, Tw; { V T2, T3, T7, T8; T1 = LD(&(xi[0]), ivs, &(xi[0])); T6 = LD(&(xi[WS(is, 6)]), ivs, &(xi[0])); T2 = LD(&(xi[WS(is, 4)]), ivs, &(xi[0])); T3 = LD(&(xi[WS(is, 8)]), ivs, &(xi[0])); T7 = LD(&(xi[WS(is, 10)]), ivs, &(xi[0])); T8 = LD(&(xi[WS(is, 2)]), ivs, &(xi[0])); Tc = LD(&(xi[WS(is, 3)]), ivs, &(xi[WS(is, 1)])); Th = LD(&(xi[WS(is, 9)]), ivs, &(xi[WS(is, 1)])); Td = LD(&(xi[WS(is, 7)]), ivs, &(xi[WS(is, 1)])); Te = LD(&(xi[WS(is, 11)]), ivs, &(xi[WS(is, 1)])); Ti = LD(&(xi[WS(is, 1)]), ivs, &(xi[WS(is, 1)])); Tz = VSUB(T2, T3); T4 = VADD(T2, T3); TA = VSUB(T7, T8); T9 = VADD(T7, T8); Tj = LD(&(xi[WS(is, 5)]), ivs, &(xi[WS(is, 1)])); } Tf = VADD(Td, Te); Tw = VSUB(Td, Te); { V T5, Tp, TJ, TB, Ta, Tq, Tk, Tx, Tg, Ts; T5 = VADD(T1, T4); Tp = VFNMS(LDK(KP500000000), T4, T1); TJ = VSUB(Tz, TA); TB = VADD(Tz, TA); Ta = VADD(T6, T9); Tq = VFNMS(LDK(KP500000000), T9, T6); Tk = VADD(Ti, Tj); Tx = VSUB(Tj, Ti); Tg = VADD(Tc, Tf); Ts = VFNMS(LDK(KP500000000), Tf, Tc); { V Tr, TF, Tb, Tn, TG, Ty, Tl, Tt; Tr = VADD(Tp, Tq); TF = VSUB(Tp, Tq); Tb = VSUB(T5, Ta); Tn = VADD(T5, Ta); TG = VADD(Tw, Tx); Ty = VSUB(Tw, Tx); Tl = VADD(Th, Tk); Tt = VFNMS(LDK(KP500000000), Tk, Th); { V TC, TE, TH, TL, Tu, TI, Tm, To; TC = VMUL(LDK(KP866025403), VSUB(Ty, TB)); TE = VMUL(LDK(KP866025403), VADD(TB, Ty)); TH = VFNMS(LDK(KP866025403), TG, TF); TL = VFMA(LDK(KP866025403), TG, TF); Tu = VADD(Ts, Tt); TI = VSUB(Ts, Tt); Tm = VSUB(Tg, Tl); To = VADD(Tg, Tl); { V TK, TM, Tv, TD; TK = VFMA(LDK(KP866025403), TJ, TI); TM = VFNMS(LDK(KP866025403), TJ, TI); Tv = VSUB(Tr, Tu); TD = VADD(Tr, Tu); { V TN, TO, TP, TQ; TN = VADD(Tn, To); STM2(&(xo[0]), TN, ovs, &(xo[0])); TO = VSUB(Tn, To); STM2(&(xo[12]), TO, ovs, &(xo[0])); TP = VFMAI(Tm, Tb); STM2(&(xo[18]), TP, ovs, &(xo[2])); TQ = VFNMSI(Tm, Tb); STM2(&(xo[6]), TQ, ovs, &(xo[2])); { V TR, TS, TT, TU; TR = VFMAI(TM, TL); STM2(&(xo[10]), TR, ovs, &(xo[2])); TS = VFNMSI(TM, TL); STM2(&(xo[14]), TS, ovs, &(xo[2])); STN2(&(xo[12]), TO, TS, ovs); TT = VFNMSI(TK, TH); STM2(&(xo[22]), TT, ovs, &(xo[2])); TU = VFMAI(TK, TH); STM2(&(xo[2]), TU, ovs, &(xo[2])); STN2(&(xo[0]), TN, TU, ovs); { V TV, TW, TX, TY; TV = VFNMSI(TE, TD); STM2(&(xo[16]), TV, ovs, &(xo[0])); STN2(&(xo[16]), TV, TP, ovs); TW = VFMAI(TE, TD); STM2(&(xo[8]), TW, ovs, &(xo[0])); STN2(&(xo[8]), TW, TR, ovs); TX = VFMAI(TC, Tv); STM2(&(xo[4]), TX, ovs, &(xo[0])); STN2(&(xo[4]), TX, TQ, ovs); TY = VFNMSI(TC, Tv); STM2(&(xo[20]), TY, ovs, &(xo[0])); STN2(&(xo[20]), TY, TT, ovs); } } } } } } } } }
static void n2bv_32(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs) { DVK(KP980785280, +0.980785280403230449126182236134239036973933731); DVK(KP831469612, +0.831469612302545237078788377617905756738560812); DVK(KP198912367, +0.198912367379658006911597622644676228597850501); DVK(KP668178637, +0.668178637919298919997757686523080761552472251); DVK(KP923879532, +0.923879532511286756128183189396788286822416626); DVK(KP414213562, +0.414213562373095048801688724209698078569671875); DVK(KP707106781, +0.707106781186547524400844362104849039284835938); { INT i; const R *xi; R *xo; xi = ii; xo = io; for (i = v; i > 0; i = i - VL, xi = xi + (VL * ivs), xo = xo + (VL * ovs), MAKE_VOLATILE_STRIDE(64, is), MAKE_VOLATILE_STRIDE(64, os)) { V T31, T32, T33, T34, T35, T36, T37, T38, T39, T3a, T3b, T3c, T1h, Tr, T3d; V T3e, T3f, T3g, T1a, T1k, TI, T1b, T1L, T1P, T1I, T1G, T1O, T1Q, T1H, T1z; V T1c, TZ; { V T2x, T1T, T2K, T1W, T1p, Tb, T1A, T16, Tu, TF, T2O, T2H, T2b, T2t, TY; V T1w, TT, T1v, T20, T2C, Tj, Te, T2e, To, T2i, T23, T2D, TB, TG, Th; V T2f, Tk; { V TL, TW, TP, TQ, T2F, T27, T28, TO; { V T1, T2, T12, T13, T4, T5, T7, T8; T1 = LD(&(xi[0]), ivs, &(xi[0])); T2 = LD(&(xi[WS(is, 16)]), ivs, &(xi[0])); T12 = LD(&(xi[WS(is, 8)]), ivs, &(xi[0])); T13 = LD(&(xi[WS(is, 24)]), ivs, &(xi[0])); T4 = LD(&(xi[WS(is, 4)]), ivs, &(xi[0])); T5 = LD(&(xi[WS(is, 20)]), ivs, &(xi[0])); T7 = LD(&(xi[WS(is, 28)]), ivs, &(xi[0])); T8 = LD(&(xi[WS(is, 12)]), ivs, &(xi[0])); { V TM, T25, T26, TN; { V TJ, T3, T14, T1U, T6, T1V, T9, TK, TU, TV, T1R, T1S, Ta, T15; TJ = LD(&(xi[WS(is, 31)]), ivs, &(xi[WS(is, 1)])); T1R = VADD(T1, T2); T3 = VSUB(T1, T2); T1S = VADD(T12, T13); T14 = VSUB(T12, T13); T1U = VADD(T4, T5); T6 = VSUB(T4, T5); T1V = VADD(T7, T8); T9 = VSUB(T7, T8); TK = LD(&(xi[WS(is, 15)]), ivs, &(xi[WS(is, 1)])); TU = LD(&(xi[WS(is, 23)]), ivs, &(xi[WS(is, 1)])); T2x = VSUB(T1R, T1S); T1T = VADD(T1R, T1S); TV = LD(&(xi[WS(is, 7)]), ivs, &(xi[WS(is, 1)])); TM = LD(&(xi[WS(is, 3)]), ivs, &(xi[WS(is, 1)])); T2K = VSUB(T1U, T1V); T1W = VADD(T1U, T1V); Ta = VADD(T6, T9); T15 = VSUB(T6, T9); T25 = VADD(TJ, TK); TL = VSUB(TJ, TK); T26 = VADD(TV, TU); TW = VSUB(TU, TV); TN = LD(&(xi[WS(is, 19)]), ivs, &(xi[WS(is, 1)])); TP = LD(&(xi[WS(is, 27)]), ivs, &(xi[WS(is, 1)])); T1p = VFNMS(LDK(KP707106781), Ta, T3); Tb = VFMA(LDK(KP707106781), Ta, T3); T1A = VFNMS(LDK(KP707106781), T15, T14); T16 = VFMA(LDK(KP707106781), T15, T14); TQ = LD(&(xi[WS(is, 11)]), ivs, &(xi[WS(is, 1)])); } T2F = VSUB(T25, T26); T27 = VADD(T25, T26); T28 = VADD(TM, TN); TO = VSUB(TM, TN); } } { V Ty, T21, Tx, Tz, T1Y, T1Z; { V Ts, Tt, TD, T29, TR, TE, Tv, Tw; Ts = LD(&(xi[WS(is, 1)]), ivs, &(xi[WS(is, 1)])); Tt = LD(&(xi[WS(is, 17)]), ivs, &(xi[WS(is, 1)])); TD = LD(&(xi[WS(is, 9)]), ivs, &(xi[WS(is, 1)])); T29 = VADD(TP, TQ); TR = VSUB(TP, TQ); TE = LD(&(xi[WS(is, 25)]), ivs, &(xi[WS(is, 1)])); Tv = LD(&(xi[WS(is, 5)]), ivs, &(xi[WS(is, 1)])); Tw = LD(&(xi[WS(is, 21)]), ivs, &(xi[WS(is, 1)])); Ty = LD(&(xi[WS(is, 29)]), ivs, &(xi[WS(is, 1)])); T1Y = VADD(Ts, Tt); Tu = VSUB(Ts, Tt); { V T2G, T2a, TX, TS; T2G = VSUB(T29, T28); T2a = VADD(T28, T29); TX = VSUB(TR, TO); TS = VADD(TO, TR); T1Z = VADD(TD, TE); TF = VSUB(TD, TE); T21 = VADD(Tv, Tw); Tx = VSUB(Tv, Tw); T2O = VFMA(LDK(KP414213562), T2F, T2G); T2H = VFNMS(LDK(KP414213562), T2G, T2F); T2b = VSUB(T27, T2a); T2t = VADD(T27, T2a); TY = VFMA(LDK(KP707106781), TX, TW); T1w = VFNMS(LDK(KP707106781), TX, TW); TT = VFMA(LDK(KP707106781), TS, TL); T1v = VFNMS(LDK(KP707106781), TS, TL); Tz = LD(&(xi[WS(is, 13)]), ivs, &(xi[WS(is, 1)])); } } T20 = VADD(T1Y, T1Z); T2C = VSUB(T1Y, T1Z); { V Tc, Td, Tm, Tn; Tc = LD(&(xi[WS(is, 2)]), ivs, &(xi[0])); Td = LD(&(xi[WS(is, 18)]), ivs, &(xi[0])); Tm = LD(&(xi[WS(is, 22)]), ivs, &(xi[0])); Tn = LD(&(xi[WS(is, 6)]), ivs, &(xi[0])); { V Tf, TA, T22, Tg; Tf = LD(&(xi[WS(is, 10)]), ivs, &(xi[0])); TA = VSUB(Ty, Tz); T22 = VADD(Ty, Tz); Tg = LD(&(xi[WS(is, 26)]), ivs, &(xi[0])); Tj = LD(&(xi[WS(is, 30)]), ivs, &(xi[0])); Te = VSUB(Tc, Td); T2e = VADD(Tc, Td); To = VSUB(Tm, Tn); T2i = VADD(Tn, Tm); T23 = VADD(T21, T22); T2D = VSUB(T21, T22); TB = VADD(Tx, TA); TG = VSUB(Tx, TA); Th = VSUB(Tf, Tg); T2f = VADD(Tf, Tg); Tk = LD(&(xi[WS(is, 14)]), ivs, &(xi[0])); } } } } { V T1t, TH, T1s, TC, T2P, T2U, T2n, T2d, T2w, T2u, T1q, T19, T1B, Tq, T2W; V T2M, T2B, T2T, T2v, T2r, T2o, T2m, T2X, T2I; { V T1X, T2p, T2E, T2N, T2s, T2y, T2g, T17, Ti, T2h, Tl, T2c, T2l, T24; T1X = VSUB(T1T, T1W); T2p = VADD(T1T, T1W); T2E = VFNMS(LDK(KP414213562), T2D, T2C); T2N = VFMA(LDK(KP414213562), T2C, T2D); T2s = VADD(T20, T23); T24 = VSUB(T20, T23); T1t = VFNMS(LDK(KP707106781), TG, TF); TH = VFMA(LDK(KP707106781), TG, TF); T1s = VFNMS(LDK(KP707106781), TB, Tu); TC = VFMA(LDK(KP707106781), TB, Tu); T2y = VSUB(T2e, T2f); T2g = VADD(T2e, T2f); T17 = VFMA(LDK(KP414213562), Te, Th); Ti = VFNMS(LDK(KP414213562), Th, Te); T2h = VADD(Tj, Tk); Tl = VSUB(Tj, Tk); T2c = VADD(T24, T2b); T2l = VSUB(T24, T2b); { V T2L, T2A, T2q, T2k; T2P = VSUB(T2N, T2O); T2U = VADD(T2N, T2O); { V T2z, T2j, T18, Tp; T2z = VSUB(T2h, T2i); T2j = VADD(T2h, T2i); T18 = VFMA(LDK(KP414213562), Tl, To); Tp = VFNMS(LDK(KP414213562), To, Tl); T2n = VFMA(LDK(KP707106781), T2c, T1X); T2d = VFNMS(LDK(KP707106781), T2c, T1X); T2w = VADD(T2s, T2t); T2u = VSUB(T2s, T2t); T2L = VSUB(T2y, T2z); T2A = VADD(T2y, T2z); T2q = VADD(T2g, T2j); T2k = VSUB(T2g, T2j); T1q = VADD(T17, T18); T19 = VSUB(T17, T18); T1B = VSUB(Ti, Tp); Tq = VADD(Ti, Tp); } T2W = VFNMS(LDK(KP707106781), T2L, T2K); T2M = VFMA(LDK(KP707106781), T2L, T2K); T2B = VFMA(LDK(KP707106781), T2A, T2x); T2T = VFNMS(LDK(KP707106781), T2A, T2x); T2v = VADD(T2p, T2q); T2r = VSUB(T2p, T2q); T2o = VFMA(LDK(KP707106781), T2l, T2k); T2m = VFNMS(LDK(KP707106781), T2l, T2k); T2X = VSUB(T2E, T2H); T2I = VADD(T2E, T2H); } } { V T2V, T2Z, T2Y, T30, T2R, T2J; T2V = VFNMS(LDK(KP923879532), T2U, T2T); T2Z = VFMA(LDK(KP923879532), T2U, T2T); T31 = VSUB(T2v, T2w); STM2(&(xo[32]), T31, ovs, &(xo[0])); T32 = VADD(T2v, T2w); STM2(&(xo[0]), T32, ovs, &(xo[0])); T33 = VFMAI(T2u, T2r); STM2(&(xo[16]), T33, ovs, &(xo[0])); T34 = VFNMSI(T2u, T2r); STM2(&(xo[48]), T34, ovs, &(xo[0])); T35 = VFMAI(T2o, T2n); STM2(&(xo[8]), T35, ovs, &(xo[0])); T36 = VFNMSI(T2o, T2n); STM2(&(xo[56]), T36, ovs, &(xo[0])); T37 = VFMAI(T2m, T2d); STM2(&(xo[40]), T37, ovs, &(xo[0])); T38 = VFNMSI(T2m, T2d); STM2(&(xo[24]), T38, ovs, &(xo[0])); T2Y = VFMA(LDK(KP923879532), T2X, T2W); T30 = VFNMS(LDK(KP923879532), T2X, T2W); T2R = VFMA(LDK(KP923879532), T2I, T2B); T2J = VFNMS(LDK(KP923879532), T2I, T2B); { V T1J, T1r, T1C, T1M, T2S, T2Q, T1u, T1D, T1E, T1x; T1J = VFNMS(LDK(KP923879532), T1q, T1p); T1r = VFMA(LDK(KP923879532), T1q, T1p); T1C = VFNMS(LDK(KP923879532), T1B, T1A); T1M = VFMA(LDK(KP923879532), T1B, T1A); T39 = VFNMSI(T30, T2Z); STM2(&(xo[12]), T39, ovs, &(xo[0])); T3a = VFMAI(T30, T2Z); STM2(&(xo[52]), T3a, ovs, &(xo[0])); T3b = VFNMSI(T2Y, T2V); STM2(&(xo[44]), T3b, ovs, &(xo[0])); T3c = VFMAI(T2Y, T2V); STM2(&(xo[20]), T3c, ovs, &(xo[0])); T2S = VFMA(LDK(KP923879532), T2P, T2M); T2Q = VFNMS(LDK(KP923879532), T2P, T2M); T1u = VFMA(LDK(KP668178637), T1t, T1s); T1D = VFNMS(LDK(KP668178637), T1s, T1t); T1E = VFNMS(LDK(KP668178637), T1v, T1w); T1x = VFMA(LDK(KP668178637), T1w, T1v); { V T1K, T1F, T1N, T1y; T1h = VFNMS(LDK(KP923879532), Tq, Tb); Tr = VFMA(LDK(KP923879532), Tq, Tb); T3d = VFNMSI(T2S, T2R); STM2(&(xo[60]), T3d, ovs, &(xo[0])); T3e = VFMAI(T2S, T2R); STM2(&(xo[4]), T3e, ovs, &(xo[0])); T3f = VFMAI(T2Q, T2J); STM2(&(xo[36]), T3f, ovs, &(xo[0])); T3g = VFNMSI(T2Q, T2J); STM2(&(xo[28]), T3g, ovs, &(xo[0])); T1K = VADD(T1D, T1E); T1F = VSUB(T1D, T1E); T1N = VSUB(T1u, T1x); T1y = VADD(T1u, T1x); T1a = VFMA(LDK(KP923879532), T19, T16); T1k = VFNMS(LDK(KP923879532), T19, T16); TI = VFNMS(LDK(KP198912367), TH, TC); T1b = VFMA(LDK(KP198912367), TC, TH); T1L = VFMA(LDK(KP831469612), T1K, T1J); T1P = VFNMS(LDK(KP831469612), T1K, T1J); T1I = VFMA(LDK(KP831469612), T1F, T1C); T1G = VFNMS(LDK(KP831469612), T1F, T1C); T1O = VFNMS(LDK(KP831469612), T1N, T1M); T1Q = VFMA(LDK(KP831469612), T1N, T1M); T1H = VFMA(LDK(KP831469612), T1y, T1r); T1z = VFNMS(LDK(KP831469612), T1y, T1r); T1c = VFMA(LDK(KP198912367), TT, TY); TZ = VFNMS(LDK(KP198912367), TY, TT); } } } } } { V T1d, T1i, T10, T1l; { V T3h, T3i, T3j, T3k; T3h = VFMAI(T1O, T1L); STM2(&(xo[42]), T3h, ovs, &(xo[2])); STN2(&(xo[40]), T37, T3h, ovs); T3i = VFNMSI(T1O, T1L); STM2(&(xo[22]), T3i, ovs, &(xo[2])); STN2(&(xo[20]), T3c, T3i, ovs); T3j = VFNMSI(T1Q, T1P); STM2(&(xo[54]), T3j, ovs, &(xo[2])); STN2(&(xo[52]), T3a, T3j, ovs); T3k = VFMAI(T1Q, T1P); STM2(&(xo[10]), T3k, ovs, &(xo[2])); STN2(&(xo[8]), T35, T3k, ovs); { V T3l, T3m, T3n, T3o; T3l = VFMAI(T1I, T1H); STM2(&(xo[58]), T3l, ovs, &(xo[2])); STN2(&(xo[56]), T36, T3l, ovs); T3m = VFNMSI(T1I, T1H); STM2(&(xo[6]), T3m, ovs, &(xo[2])); STN2(&(xo[4]), T3e, T3m, ovs); T3n = VFMAI(T1G, T1z); STM2(&(xo[26]), T3n, ovs, &(xo[2])); STN2(&(xo[24]), T38, T3n, ovs); T3o = VFNMSI(T1G, T1z); STM2(&(xo[38]), T3o, ovs, &(xo[2])); STN2(&(xo[36]), T3f, T3o, ovs); T1d = VSUB(T1b, T1c); T1i = VADD(T1b, T1c); T10 = VADD(TI, TZ); T1l = VSUB(TI, TZ); } } { V T1n, T1j, T1e, T1g, T1o, T1m, T11, T1f; T1n = VFMA(LDK(KP980785280), T1i, T1h); T1j = VFNMS(LDK(KP980785280), T1i, T1h); T1e = VFNMS(LDK(KP980785280), T1d, T1a); T1g = VFMA(LDK(KP980785280), T1d, T1a); T1o = VFNMS(LDK(KP980785280), T1l, T1k); T1m = VFMA(LDK(KP980785280), T1l, T1k); T11 = VFNMS(LDK(KP980785280), T10, Tr); T1f = VFMA(LDK(KP980785280), T10, Tr); { V T3p, T3q, T3r, T3s; T3p = VFNMSI(T1m, T1j); STM2(&(xo[46]), T3p, ovs, &(xo[2])); STN2(&(xo[44]), T3b, T3p, ovs); T3q = VFMAI(T1m, T1j); STM2(&(xo[18]), T3q, ovs, &(xo[2])); STN2(&(xo[16]), T33, T3q, ovs); T3r = VFMAI(T1o, T1n); STM2(&(xo[50]), T3r, ovs, &(xo[2])); STN2(&(xo[48]), T34, T3r, ovs); T3s = VFNMSI(T1o, T1n); STM2(&(xo[14]), T3s, ovs, &(xo[2])); STN2(&(xo[12]), T39, T3s, ovs); { V T3t, T3u, T3v, T3w; T3t = VFMAI(T1g, T1f); STM2(&(xo[2]), T3t, ovs, &(xo[2])); STN2(&(xo[0]), T32, T3t, ovs); T3u = VFNMSI(T1g, T1f); STM2(&(xo[62]), T3u, ovs, &(xo[2])); STN2(&(xo[60]), T3d, T3u, ovs); T3v = VFMAI(T1e, T11); STM2(&(xo[34]), T3v, ovs, &(xo[2])); STN2(&(xo[32]), T31, T3v, ovs); T3w = VFNMSI(T1e, T11); STM2(&(xo[30]), T3w, ovs, &(xo[2])); STN2(&(xo[28]), T3g, T3w, ovs); } } } } } } VLEAVE(); }