自学教程：C++ DK函数代码示例

static void r2cfII_20(R *R0, R *R1, R *Cr, R *Ci, stride rs, stride csr, stride csi, INT v, INT ivs, INT ovs){     DK(KP572061402, +0.572061402817684297600072783580302076536153377);     DK(KP218508012, +0.218508012224410535399650602527877556893735408);     DK(KP309016994, +0.309016994374947424102293417182819058860154590);     DK(KP809016994, +0.809016994374947424102293417182819058860154590);     DK(KP559016994, +0.559016994374947424102293417182819058860154590);     DK(KP951056516, +0.951056516295153572116439333379382143405698634);     DK(KP587785252, +0.587785252292473129168705954639072768597652438);     DK(KP250000000, +0.250000000000000000000000000000000000000000000);     DK(KP176776695, +0.176776695296636881100211090526212259821208984);     DK(KP395284707, +0.395284707521047416499861693054089816714944392);     DK(KP672498511, +0.672498511963957326960058968885748755876783111);     DK(KP415626937, +0.415626937777453428589967464113135184222253485);     DK(KP707106781, +0.707106781186547524400844362104849039284835938);     {	  INT i;	  for (i = v; i > 0; i = i - 1, R0 = R0 + ivs, R1 = R1 + ivs, Cr = Cr + ovs, Ci = Ci + ovs, MAKE_VOLATILE_STRIDE(rs), MAKE_VOLATILE_STRIDE(csr), MAKE_VOLATILE_STRIDE(csi)) {	       E T8, TD, Tm, TN, T9, TC, TY, TE, Te, TF, Tl, TK, T12, TL, Tk;	       E TM, T1, T6, Tq, T1l, T1c, Tp, T1f, T1e, T1d, Ty, TW, T1g, T1m, Tx;	       E Tu;	       T8 = R1[WS(rs, 2)];	       TD = KP707106781 * T8;	       Tm = R1[WS(rs, 7)];	       TN = KP707106781 * Tm;	       {		    E Ta, TA, Td, TB, Tb, Tc;		    T9 = R1[WS(rs, 6)];		    Ta = R1[WS(rs, 8)];		    TA = T9 + Ta;		    Tb = R1[0];		    Tc = R1[WS(rs, 4)];		    Td = Tb + Tc;		    TB = Tb - Tc;		    TC = FMA(KP415626937, TA, KP672498511 * TB);		    TY = FNMS(KP415626937, TB, KP672498511 * TA);		    TE = KP395284707 * (Ta - Td);		    Te = Ta + Td;		    TF = KP176776695 * Te;	       }	       {		    E Tg, TJ, Tj, TI, Th, Ti;		    Tg = R1[WS(rs, 1)];		    Tl = R1[WS(rs, 3)];		    TJ = Tg + Tl;		    Th = R1[WS(rs, 5)];		    Ti = R1[WS(rs, 9)];		    Tj = Th + Ti;		    TI = Th - Ti;		    TK = FNMS(KP415626937, TJ, KP672498511 * TI);		    T12 = FMA(KP415626937, TI, KP672498511 * TJ);		    TL = KP395284707 * (Tg - Tj);		    Tk = Tg + Tj;		    TM = KP176776695 * Tk;	       }	       {		    E T2, T5, T3, T4, T1a, T1b;		    T1 = R0[0];		    T2 = R0[WS(rs, 6)];		    T5 = R0[WS(rs, 8)];		    T3 = R0[WS(rs, 2)];		    T4 = R0[WS(rs, 4)];		    T1a = T4 + T2;		    T1b = T5 + T3;		    T6 = T2 + T3 - (T4 + T5);		    Tq = FMA(KP250000000, T6, T1);		    T1l = FNMS(KP951056516, T1b, KP587785252 * T1a);		    T1c = FMA(KP951056516, T1a, KP587785252 * T1b);		    Tp = KP559016994 * (T5 + T2 - (T4 + T3));	       }	       T1f = R0[WS(rs, 5)];	       {		    E Tv, Tw, Ts, Tt;		    Tv = R0[WS(rs, 9)];		    Tw = R0[WS(rs, 1)];		    Tx = Tv - Tw;		    T1e = Tv + Tw;		    Ts = R0[WS(rs, 3)];		    Tt = R0[WS(rs, 7)];		    Tu = Ts - Tt;		    T1d = Ts + Tt;	       }	       Ty = FMA(KP951056516, Tu, KP587785252 * Tx);	       TW = FNMS(KP951056516, Tx, KP587785252 * Tu);	       T1g = FMA(KP809016994, T1d, KP309016994 * T1e) + T1f;	       T1m = FNMS(KP809016994, T1e, T1f) - (KP309016994 * T1d);	       {		    E T7, T1r, To, T1q, Tf, Tn;		    T7 = T1 - T6;		    T1r = T1e + T1f - T1d;		    Tf = T8 + (T9 - Te);		    Tn = (Tk - Tl) - Tm;		    To = KP707106781 * (Tf + Tn);		    T1q = KP707106781 * (Tf - Tn);		    Cr[WS(csr, 2)] = T7 - To;		    Ci[WS(csi, 2)] = T1q - T1r;		    Cr[WS(csr, 7)] = T7 + To;		    Ci[WS(csi, 7)] = T1q + T1r;	       }	       {//.........这里部分代码省略.........

static void hc2cfdft_8(R *Rp, R *Ip, R *Rm, R *Im, const R *W, stride rs, INT mb, INT me, INT ms){     DK(KP353553390, +0.353553390593273762200422181052424519642417969);     DK(KP500000000, +0.500000000000000000000000000000000000000000000);     {	  INT m;	  for (m = mb, W = W + ((mb - 1) * 14); m < me; m = m + 1, Rp = Rp + ms, Ip = Ip + ms, Rm = Rm - ms, Im = Im - ms, W = W + 14, MAKE_VOLATILE_STRIDE(32, rs)) {	       E Tv, TX, Ts, TY, TE, T1a, TJ, T19, T1l, T1m, T9, T10, Ti, T11, TP;	       E T16, TU, T17, T1i, T1j;	       {		    E Tt, Tu, TD, Tz, TA, TB, Tn, TI, Tr, TG, Tk, To;		    Tt = Ip[0];		    Tu = Im[0];		    TD = Tt + Tu;		    Tz = Rm[0];		    TA = Rp[0];		    TB = Tz - TA;		    {			 E Tl, Tm, Tp, Tq;			 Tl = Ip[WS(rs, 2)];			 Tm = Im[WS(rs, 2)];			 Tn = Tl - Tm;			 TI = Tl + Tm;			 Tp = Rp[WS(rs, 2)];			 Tq = Rm[WS(rs, 2)];			 Tr = Tp + Tq;			 TG = Tp - Tq;		    }		    Tv = Tt - Tu;		    TX = TA + Tz;		    Tk = W[6];		    To = W[7];		    Ts = FNMS(To, Tr, Tk * Tn);		    TY = FMA(Tk, Tr, To * Tn);		    {			 E Ty, TC, TF, TH;			 Ty = W[0];			 TC = W[1];			 TE = FNMS(TC, TD, Ty * TB);			 T1a = FMA(TC, TB, Ty * TD);			 TF = W[8];			 TH = W[9];			 TJ = FMA(TF, TG, TH * TI);			 T19 = FNMS(TH, TG, TF * TI);		    }		    T1l = TJ + TE;		    T1m = T1a - T19;	       }	       {		    E T4, TO, T8, TM, Td, TT, Th, TR;		    {			 E T2, T3, T6, T7;			 T2 = Ip[WS(rs, 1)];			 T3 = Im[WS(rs, 1)];			 T4 = T2 - T3;			 TO = T2 + T3;			 T6 = Rp[WS(rs, 1)];			 T7 = Rm[WS(rs, 1)];			 T8 = T6 + T7;			 TM = T6 - T7;		    }		    {			 E Tb, Tc, Tf, Tg;			 Tb = Ip[WS(rs, 3)];			 Tc = Im[WS(rs, 3)];			 Td = Tb - Tc;			 TT = Tb + Tc;			 Tf = Rp[WS(rs, 3)];			 Tg = Rm[WS(rs, 3)];			 Th = Tf + Tg;			 TR = Tf - Tg;		    }		    {			 E T1, T5, Ta, Te;			 T1 = W[2];			 T5 = W[3];			 T9 = FNMS(T5, T8, T1 * T4);			 T10 = FMA(T1, T8, T5 * T4);			 Ta = W[10];			 Te = W[11];			 Ti = FNMS(Te, Th, Ta * Td);			 T11 = FMA(Ta, Th, Te * Td);			 {			      E TL, TN, TQ, TS;			      TL = W[4];			      TN = W[5];			      TP = FMA(TL, TM, TN * TO);			      T16 = FNMS(TN, TM, TL * TO);			      TQ = W[12];			      TS = W[13];			      TU = FMA(TQ, TR, TS * TT);			      T17 = FNMS(TS, TR, TQ * TT);			 }			 T1i = T17 - T16;			 T1j = TP - TU;		    }	       }	       {		    E T1h, T1t, T1w, T1y, T1o, T1s, T1r, T1x;		    {//.........这里部分代码省略.........

static void hc2r_9(const R *ri, const R *ii, R *O, stride ris, stride iis, stride os, int v, int ivs, int ovs){     DK(KP984807753, +0.984807753012208059366743024589523013670643252);     DK(KP173648177, +0.173648177666930348851716626769314796000375677);     DK(KP300767466, +0.300767466360870593278543795225003852144476517);     DK(KP1_705737063, +1.705737063904886419256501927880148143872040591);     DK(KP642787609, +0.642787609686539326322643409907263432907559884);     DK(KP766044443, +0.766044443118978035202392650555416673935832457);     DK(KP1_326827896, +1.326827896337876792410842639271782594433726619);     DK(KP1_113340798, +1.113340798452838732905825904094046265936583811);     DK(KP500000000, +0.500000000000000000000000000000000000000000000);     DK(KP866025403, +0.866025403784438646763723170752936183471402627);     DK(KP2_000000000, +2.000000000000000000000000000000000000000000000);     DK(KP1_732050807, +1.732050807568877293527446341505872366942805254);     int i;     for (i = v; i > 0; i = i - 1, ri = ri + ivs, ii = ii + ivs, O = O + ovs) {	  E T3, Tq, Tc, Tk, Tj, T8, Tm, Ts, Th, Tr, Tw, Tx;	  {	       E Tb, T1, T2, T9, Ta;	       Ta = ii[WS(iis, 3)];	       Tb = KP1_732050807 * Ta;	       T1 = ri[0];	       T2 = ri[WS(ris, 3)];	       T9 = T1 - T2;	       T3 = FMA(KP2_000000000, T2, T1);	       Tq = T9 + Tb;	       Tc = T9 - Tb;	  }	  {	       E T4, T7, Ti, Tg, Tl, Td;	       T4 = ri[WS(ris, 1)];	       Tk = ii[WS(iis, 1)];	       {		    E T5, T6, Te, Tf;		    T5 = ri[WS(ris, 4)];		    T6 = ri[WS(ris, 2)];		    T7 = T5 + T6;		    Ti = KP866025403 * (T5 - T6);		    Te = ii[WS(iis, 4)];		    Tf = ii[WS(iis, 2)];		    Tg = KP866025403 * (Te + Tf);		    Tj = Tf - Te;	       }	       T8 = T4 + T7;	       Tl = FMA(KP500000000, Tj, Tk);	       Tm = Ti + Tl;	       Ts = Tl - Ti;	       Td = FNMS(KP500000000, T7, T4);	       Th = Td - Tg;	       Tr = Td + Tg;	  }	  O[0] = FMA(KP2_000000000, T8, T3);	  Tw = T3 - T8;	  Tx = KP1_732050807 * (Tk - Tj);	  O[WS(os, 3)] = Tw - Tx;	  O[WS(os, 6)] = Tw + Tx;	  {	       E Tp, Tn, To, Tv, Tt, Tu;	       Tp = FMA(KP1_113340798, Th, KP1_326827896 * Tm);	       Tn = FNMS(KP642787609, Tm, KP766044443 * Th);	       To = Tc - Tn;	       O[WS(os, 1)] = FMA(KP2_000000000, Tn, Tc);	       O[WS(os, 7)] = To + Tp;	       O[WS(os, 4)] = To - Tp;	       Tv = FMA(KP1_705737063, Tr, KP300767466 * Ts);	       Tt = FNMS(KP984807753, Ts, KP173648177 * Tr);	       Tu = Tq - Tt;	       O[WS(os, 2)] = FMA(KP2_000000000, Tt, Tq);	       O[WS(os, 8)] = Tu + Tv;	       O[WS(os, 5)] = Tu - Tv;	  }     }}

static void hb2_16(R *cr, R *ci, const R *W, stride rs, INT mb, INT me, INT ms){     DK(KP382683432, +0.382683432365089771728459984030398866761344562);     DK(KP923879532, +0.923879532511286756128183189396788286822416626);     DK(KP707106781, +0.707106781186547524400844362104849039284835938);     {	  INT m;	  for (m = mb, W = W + ((mb - 1) * 8); m < me; m = m + 1, cr = cr + ms, ci = ci - ms, W = W + 8, MAKE_VOLATILE_STRIDE(32, rs)) {	       E Tv, Ty, T1l, T1n, T1p, T1t, T27, T25, Tz, Tw, TB, T21, T1P, T1H, T1X;	       E T17, T1L, T1N, T1v, T1w, T1x, T1B, T2F, T2T, T2b, T2R, T3j, T3x, T35, T3t;	       {		    E TA, T1J, T15, T1G, Tx, T1K, T16, T1F;		    {			 E T1m, T1s, T1o, T1r;			 Tv = W[0];			 Ty = W[1];			 T1l = W[2];			 T1n = W[3];			 T1m = Tv * T1l;			 T1s = Ty * T1l;			 T1o = Ty * T1n;			 T1r = Tv * T1n;			 T1p = T1m + T1o;			 T1t = T1r - T1s;			 T27 = T1r + T1s;			 T25 = T1m - T1o;			 Tz = W[5];			 TA = Ty * Tz;			 T1J = T1l * Tz;			 T15 = Tv * Tz;			 T1G = T1n * Tz;			 Tw = W[4];			 Tx = Tv * Tw;			 T1K = T1n * Tw;			 T16 = Ty * Tw;			 T1F = T1l * Tw;		    }		    TB = Tx - TA;		    T21 = T1J + T1K;		    T1P = T15 - T16;		    T1H = T1F + T1G;		    T1X = T1F - T1G;		    T17 = T15 + T16;		    T1L = T1J - T1K;		    T1N = Tx + TA;		    T1v = W[6];		    T1w = W[7];		    T1x = FMA(Tv, T1v, Ty * T1w);		    T1B = FNMS(Ty, T1v, Tv * T1w);		    {			 E T2D, T2E, T29, T2a;			 T2D = T25 * Tz;			 T2E = T27 * Tw;			 T2F = T2D + T2E;			 T2T = T2D - T2E;			 T29 = T25 * Tw;			 T2a = T27 * Tz;			 T2b = T29 - T2a;			 T2R = T29 + T2a;		    }		    {			 E T3h, T3i, T33, T34;			 T3h = T1p * Tz;			 T3i = T1t * Tw;			 T3j = T3h + T3i;			 T3x = T3h - T3i;			 T33 = T1p * Tw;			 T34 = T1t * Tz;			 T35 = T33 - T34;			 T3t = T33 + T34;		    }	       }	       {		    E T7, T36, T3k, TC, T1f, T2e, T2I, T1Q, Te, TJ, T1R, T18, T2L, T37, T2l;		    E T3l, Tm, T1T, TT, T1h, T2A, T2N, T3b, T3n, Tt, T1U, T12, T1i, T2t, T2O;		    E T3e, T3o;		    {			 E T3, T2c, T1e, T2d, T6, T2G, T1b, T2H;			 {			      E T1, T2, T1c, T1d;			      T1 = cr[0];			      T2 = ci[WS(rs, 7)];			      T3 = T1 + T2;			      T2c = T1 - T2;			      T1c = ci[WS(rs, 11)];			      T1d = cr[WS(rs, 12)];			      T1e = T1c - T1d;			      T2d = T1c + T1d;			 }			 {			      E T4, T5, T19, T1a;			      T4 = cr[WS(rs, 4)];			      T5 = ci[WS(rs, 3)];			      T6 = T4 + T5;			      T2G = T4 - T5;			      T19 = ci[WS(rs, 15)];			      T1a = cr[WS(rs, 8)];			      T1b = T19 - T1a;			      T2H = T19 + T1a;			 }//.........这里部分代码省略.........

static void q1_3(float *rio, float *iio, const float *W, stride rs, stride vs, INT mb, INT me, INT ms){     DK(KP866025403, +0.866025403784438646763723170752936183471402627);     DK(KP500000000, +0.500000000000000000000000000000000000000000000);     INT m;     for (m = mb, W = W + (mb * 4); m < me; m = m + 1, rio = rio + ms, iio = iio + ms, W = W + 4, MAKE_VOLATILE_STRIDE(rs), MAKE_VOLATILE_STRIDE(vs)) {	  E T1, T4, T6, Tc, Td, Te, T9, Tf, Tl, To, Tq, Tw, Tx, Ty, Tt;	  E Tz, TR, TS, TN, TT, TF, TI, TK, TQ;	  {	       E T2, T3, Tr, Ts;	       T1 = rio[0];	       T2 = rio[WS(rs, 1)];	       T3 = rio[WS(rs, 2)];	       T4 = T2 + T3;	       T6 = FNMS(KP500000000, T4, T1);	       Tc = KP866025403 * (T3 - T2);	       {		    E T7, T8, Tm, Tn;		    Td = iio[0];		    T7 = iio[WS(rs, 1)];		    T8 = iio[WS(rs, 2)];		    Te = T7 + T8;		    T9 = KP866025403 * (T7 - T8);		    Tf = FNMS(KP500000000, Te, Td);		    Tl = rio[WS(vs, 1)];		    Tm = rio[WS(vs, 1) + WS(rs, 1)];		    Tn = rio[WS(vs, 1) + WS(rs, 2)];		    To = Tm + Tn;		    Tq = FNMS(KP500000000, To, Tl);		    Tw = KP866025403 * (Tn - Tm);	       }	       Tx = iio[WS(vs, 1)];	       Tr = iio[WS(vs, 1) + WS(rs, 1)];	       Ts = iio[WS(vs, 1) + WS(rs, 2)];	       Ty = Tr + Ts;	       Tt = KP866025403 * (Tr - Ts);	       Tz = FNMS(KP500000000, Ty, Tx);	       {		    E TL, TM, TG, TH;		    TR = iio[WS(vs, 2)];		    TL = iio[WS(vs, 2) + WS(rs, 1)];		    TM = iio[WS(vs, 2) + WS(rs, 2)];		    TS = TL + TM;		    TN = KP866025403 * (TL - TM);		    TT = FNMS(KP500000000, TS, TR);		    TF = rio[WS(vs, 2)];		    TG = rio[WS(vs, 2) + WS(rs, 1)];		    TH = rio[WS(vs, 2) + WS(rs, 2)];		    TI = TG + TH;		    TK = FNMS(KP500000000, TI, TF);		    TQ = KP866025403 * (TH - TG);	       }	  }	  rio[0] = T1 + T4;	  iio[0] = Td + Te;	  rio[WS(rs, 1)] = Tl + To;	  iio[WS(rs, 1)] = Tx + Ty;	  iio[WS(rs, 2)] = TR + TS;	  rio[WS(rs, 2)] = TF + TI;	  {	       E Ta, Tg, T5, Tb;	       Ta = T6 + T9;	       Tg = Tc + Tf;	       T5 = W[0];	       Tb = W[1];	       rio[WS(vs, 1)] = FMA(T5, Ta, Tb * Tg);	       iio[WS(vs, 1)] = FNMS(Tb, Ta, T5 * Tg);	  }	  {	       E TW, TY, TV, TX;	       TW = TK - TN;	       TY = TT - TQ;	       TV = W[2];	       TX = W[3];	       rio[WS(vs, 2) + WS(rs, 2)] = FMA(TV, TW, TX * TY);	       iio[WS(vs, 2) + WS(rs, 2)] = FNMS(TX, TW, TV * TY);	  }	  {	       E TC, TE, TB, TD;	       TC = Tq - Tt;	       TE = Tz - Tw;	       TB = W[2];	       TD = W[3];	       rio[WS(vs, 2) + WS(rs, 1)] = FMA(TB, TC, TD * TE);	       iio[WS(vs, 2) + WS(rs, 1)] = FNMS(TD, TC, TB * TE);	  }	  {	       E Tu, TA, Tp, Tv;	       Tu = Tq + Tt;	       TA = Tw + Tz;	       Tp = W[0];	       Tv = W[1];	       rio[WS(vs, 1) + WS(rs, 1)] = FMA(Tp, Tu, Tv * TA);	       iio[WS(vs, 1) + WS(rs, 1)] = FNMS(Tv, Tu, Tp * TA);	  }	  {	       E TO, TU, TJ, TP;	       TO = TK + TN;	       TU = TQ + TT;	       TJ = W[0];//.........这里部分代码省略.........

static void hc2r_15(const R *ri, const R *ii, R *O, stride ris, stride iis, stride os, INT v, INT ivs, INT ovs){     DK(KP559016994, +0.559016994374947424102293417182819058860154590);     DK(KP1_902113032, +1.902113032590307144232878666758764286811397268);     DK(KP250000000, +0.250000000000000000000000000000000000000000000);     DK(KP866025403, +0.866025403784438646763723170752936183471402627);     DK(KP1_118033988, +1.118033988749894848204586834365638117720309180);     DK(KP618033988, +0.618033988749894848204586834365638117720309180);     DK(KP500000000, +0.500000000000000000000000000000000000000000000);     DK(KP1_732050807, +1.732050807568877293527446341505872366942805254);     DK(KP2_000000000, +2.000000000000000000000000000000000000000000000);     INT i;     for (i = v; i > 0; i = i - 1, ri = ri + ivs, ii = ii + ivs, O = O + ovs, MAKE_VOLATILE_STRIDE(ris), MAKE_VOLATILE_STRIDE(iis), MAKE_VOLATILE_STRIDE(os)) {	  E TL, Tz, TM, TK;	  {	       E T3, Th, Tt, TD, TI, TH, TY, TC, TZ, Tu, Tm, Tv, Tr, Te, TW;	       E Tg, T1, T2, T12, T10, TV;	       Tg = ii[WS(iis, 5)];	       T1 = ri[0];	       T2 = ri[WS(ris, 5)];	       {		    E T4, TA, T9, TF, T7, Tj, Tc, Tk, TG, Tq, Tf, Tl, TB;		    T4 = ri[WS(ris, 3)];		    TA = ii[WS(iis, 3)];		    T9 = ri[WS(ris, 6)];		    Tf = T1 - T2;		    T3 = FMA(KP2_000000000, T2, T1);		    TF = ii[WS(iis, 6)];		    {			 E Ta, Tb, T5, T6, To, Tp;			 T5 = ri[WS(ris, 7)];			 T6 = ri[WS(ris, 2)];			 Th = FMA(KP1_732050807, Tg, Tf);			 Tt = FNMS(KP1_732050807, Tg, Tf);			 Ta = ri[WS(ris, 4)];			 TD = T5 - T6;			 T7 = T5 + T6;			 Tb = ri[WS(ris, 1)];			 To = ii[WS(iis, 4)];			 Tp = ii[WS(iis, 1)];			 Tj = ii[WS(iis, 7)];			 Tc = Ta + Tb;			 TI = Ta - Tb;			 Tk = ii[WS(iis, 2)];			 TG = Tp - To;			 Tq = To + Tp;		    }		    Tl = Tj - Tk;		    TB = Tj + Tk;		    TH = FNMS(KP500000000, TG, TF);		    TY = TG + TF;		    TC = FMA(KP500000000, TB, TA);		    TZ = TA - TB;		    {			 E Ti, T8, Td, Tn;			 Ti = FNMS(KP2_000000000, T4, T7);			 T8 = T4 + T7;			 Td = T9 + Tc;			 Tn = FNMS(KP2_000000000, T9, Tc);			 Tu = FNMS(KP1_732050807, Tl, Ti);			 Tm = FMA(KP1_732050807, Tl, Ti);			 Tv = FNMS(KP1_732050807, Tq, Tn);			 Tr = FMA(KP1_732050807, Tq, Tn);			 Te = T8 + Td;			 TW = T8 - Td;		    }	       }	       T12 = FMA(KP618033988, TY, TZ);	       T10 = FNMS(KP618033988, TZ, TY);	       TV = FNMS(KP500000000, Te, T3);	       O[0] = FMA(KP2_000000000, Te, T3);	       {		    E TJ, TE, TT, TP, TU, TS, Ty, Tw, Tx;		    {			 E TO, Ts, TQ, TN, TR, T11, TX;			 TO = Tr - Tm;			 Ts = Tm + Tr;			 T11 = FMA(KP1_118033988, TW, TV);			 TX = FNMS(KP1_118033988, TW, TV);			 TQ = FNMS(KP866025403, TI, TH);			 TJ = FMA(KP866025403, TI, TH);			 TN = FMA(KP250000000, Ts, Th);			 O[WS(os, 6)] = FNMS(KP1_902113032, T12, T11);			 O[WS(os, 9)] = FMA(KP1_902113032, T12, T11);			 O[WS(os, 12)] = FMA(KP1_902113032, T10, TX);			 O[WS(os, 3)] = FNMS(KP1_902113032, T10, TX);			 TR = FNMS(KP866025403, TD, TC);			 TE = FMA(KP866025403, TD, TC);			 O[WS(os, 5)] = Th - Ts;			 TT = FMA(KP559016994, TO, TN);			 TP = FNMS(KP559016994, TO, TN);			 TU = FMA(KP618033988, TQ, TR);			 TS = FNMS(KP618033988, TR, TQ);		    }		    Ty = Tv - Tu;		    Tw = Tu + Tv;		    O[WS(os, 14)] = FMA(KP1_902113032, TU, TT);		    O[WS(os, 11)] = FNMS(KP1_902113032, TU, TT);		    O[WS(os, 2)] = FMA(KP1_902113032, TS, TP);		    O[WS(os, 8)] = FNMS(KP1_902113032, TS, TP);//.........这里部分代码省略.........

static void r2cbIII_32(R *R0, R *R1, R *Cr, R *Ci, stride rs, stride csr, stride csi, INT v, INT ivs, INT ovs){     DK(KP1_913880671, +1.913880671464417729871595773960539938965698411);     DK(KP580569354, +0.580569354508924735272384751634790549382952557);     DK(KP942793473, +0.942793473651995297112775251810508755314920638);     DK(KP1_763842528, +1.763842528696710059425513727320776699016885241);     DK(KP1_546020906, +1.546020906725473921621813219516939601942082586);     DK(KP1_268786568, +1.268786568327290996430343226450986741351374190);     DK(KP196034280, +0.196034280659121203988391127777283691722273346);     DK(KP1_990369453, +1.990369453344393772489673906218959843150949737);     DK(KP765366864, +0.765366864730179543456919968060797733522689125);     DK(KP1_847759065, +1.847759065022573512256366378793576573644833252);     DK(KP1_961570560, +1.961570560806460898252364472268478073947867462);     DK(KP390180644, +0.390180644032256535696569736954044481855383236);     DK(KP1_111140466, +1.111140466039204449485661627897065748749874382);     DK(KP1_662939224, +1.662939224605090474157576755235811513477121624);     DK(KP1_414213562, +1.414213562373095048801688724209698078569671875);     DK(KP2_000000000, +2.000000000000000000000000000000000000000000000);     DK(KP382683432, +0.382683432365089771728459984030398866761344562);     DK(KP923879532, +0.923879532511286756128183189396788286822416626);     DK(KP707106781, +0.707106781186547524400844362104849039284835938);     {	  INT i;	  for (i = v; i > 0; i = i - 1, R0 = R0 + ovs, R1 = R1 + ovs, Cr = Cr + ivs, Ci = Ci + ivs, MAKE_VOLATILE_STRIDE(128, rs), MAKE_VOLATILE_STRIDE(128, csr), MAKE_VOLATILE_STRIDE(128, csi)) {	       E T7, T2i, T2F, Tz, T1k, T1I, T1Z, T1x, Te, T22, T2E, T2j, T1f, T1y, TK;	       E T1J, Tm, T2B, TW, T1a, T1C, T1L, T28, T2l, Tt, T2A, T17, T1b, T1F, T1M;	       E T2d, T2m;	       {		    E T3, Tv, T1j, T2h, T6, T1g, Ty, T2g;		    {			 E T1, T2, T1h, T1i;			 T1 = Cr[0];			 T2 = Cr[WS(csr, 15)];			 T3 = T1 + T2;			 Tv = T1 - T2;			 T1h = Ci[0];			 T1i = Ci[WS(csi, 15)];			 T1j = T1h + T1i;			 T2h = T1i - T1h;		    }		    {			 E T4, T5, Tw, Tx;			 T4 = Cr[WS(csr, 8)];			 T5 = Cr[WS(csr, 7)];			 T6 = T4 + T5;			 T1g = T4 - T5;			 Tw = Ci[WS(csi, 8)];			 Tx = Ci[WS(csi, 7)];			 Ty = Tw + Tx;			 T2g = Tw - Tx;		    }		    T7 = T3 + T6;		    T2i = T2g + T2h;		    T2F = T2h - T2g;		    Tz = Tv - Ty;		    T1k = T1g + T1j;		    T1I = T1g - T1j;		    T1Z = T3 - T6;		    T1x = Tv + Ty;	       }	       {		    E Ta, TA, TD, T21, Td, TF, TI, T20;		    {			 E T8, T9, TB, TC;			 T8 = Cr[WS(csr, 4)];			 T9 = Cr[WS(csr, 11)];			 Ta = T8 + T9;			 TA = T8 - T9;			 TB = Ci[WS(csi, 4)];			 TC = Ci[WS(csi, 11)];			 TD = TB + TC;			 T21 = TB - TC;		    }		    {			 E Tb, Tc, TG, TH;			 Tb = Cr[WS(csr, 3)];			 Tc = Cr[WS(csr, 12)];			 Td = Tb + Tc;			 TF = Tb - Tc;			 TG = Ci[WS(csi, 3)];			 TH = Ci[WS(csi, 12)];			 TI = TG + TH;			 T20 = TH - TG;		    }		    Te = Ta + Td;		    T22 = T20 - T21;		    T2E = T21 + T20;		    T2j = Ta - Td;		    {			 E T1d, T1e, TE, TJ;			 T1d = TA + TD;			 T1e = TF + TI;			 T1f = KP707106781 * (T1d - T1e);			 T1y = KP707106781 * (T1d + T1e);			 TE = TA - TD;			 TJ = TF - TI;			 TK = KP707106781 * (TE + TJ);			 T1J = KP707106781 * (TE - TJ);		    }	       }//.........这里部分代码省略.........

static void r2cb_20(R *R0, R *R1, R *Cr, R *Ci, stride rs, stride csr, stride csi, INT v, INT ivs, INT ovs){     DK(KP1_118033988, +1.118033988749894848204586834365638117720309180);     DK(KP500000000, +0.500000000000000000000000000000000000000000000);     DK(KP1_902113032, +1.902113032590307144232878666758764286811397268);     DK(KP1_175570504, +1.175570504584946258337411909278145537195304875);     DK(KP2_000000000, +2.000000000000000000000000000000000000000000000);     INT i;     for (i = v; i > 0; i = i - 1, R0 = R0 + ovs, R1 = R1 + ovs, Cr = Cr + ivs, Ci = Ci + ivs, MAKE_VOLATILE_STRIDE(rs), MAKE_VOLATILE_STRIDE(csr), MAKE_VOLATILE_STRIDE(csi)) {	  E T6, TF, Tm, Tt, TQ, T1n, T1f, T12, T1m, TV, T13, T1c, Td, Tk, Tl;	  E Ty, TD, TE, Tn, To, Tp, TG, TH, TI;	  {	       E T5, Ts, T3, Tq;	       {		    E T4, Tr, T1, T2;		    T4 = Cr[WS(csr, 5)];		    T5 = KP2_000000000 * T4;		    Tr = Ci[WS(csi, 5)];		    Ts = KP2_000000000 * Tr;		    T1 = Cr[0];		    T2 = Cr[WS(csr, 10)];		    T3 = T1 + T2;		    Tq = T1 - T2;	       }	       T6 = T3 - T5;	       TF = Tq - Ts;	       Tm = T3 + T5;	       Tt = Tq + Ts;	  }	  {	       E T9, Tu, TO, T1b, Tc, T1a, Tx, TP, Tg, Tz, TT, T1e, Tj, T1d, TC;	       E TU;	       {		    E T7, T8, TM, TN;		    T7 = Cr[WS(csr, 4)];		    T8 = Cr[WS(csr, 6)];		    T9 = T7 + T8;		    Tu = T7 - T8;		    TM = Ci[WS(csi, 4)];		    TN = Ci[WS(csi, 6)];		    TO = TM - TN;		    T1b = TM + TN;	       }	       {		    E Ta, Tb, Tv, Tw;		    Ta = Cr[WS(csr, 9)];		    Tb = Cr[WS(csr, 1)];		    Tc = Ta + Tb;		    T1a = Ta - Tb;		    Tv = Ci[WS(csi, 9)];		    Tw = Ci[WS(csi, 1)];		    Tx = Tv + Tw;		    TP = Tv - Tw;	       }	       {		    E Te, Tf, TR, TS;		    Te = Cr[WS(csr, 8)];		    Tf = Cr[WS(csr, 2)];		    Tg = Te + Tf;		    Tz = Te - Tf;		    TR = Ci[WS(csi, 8)];		    TS = Ci[WS(csi, 2)];		    TT = TR - TS;		    T1e = TR + TS;	       }	       {		    E Th, Ti, TA, TB;		    Th = Cr[WS(csr, 7)];		    Ti = Cr[WS(csr, 3)];		    Tj = Th + Ti;		    T1d = Th - Ti;		    TA = Ci[WS(csi, 7)];		    TB = Ci[WS(csi, 3)];		    TC = TA + TB;		    TU = TB - TA;	       }	       TQ = TO - TP;	       T1n = T1e - T1d;	       T1f = T1d + T1e;	       T12 = TP + TO;	       T1m = T1b - T1a;	       TV = TT - TU;	       T13 = TU + TT;	       T1c = T1a + T1b;	       Td = T9 - Tc;	       Tk = Tg - Tj;	       Tl = Td + Tk;	       Ty = Tu + Tx;	       TD = Tz - TC;	       TE = Ty + TD;	       Tn = T9 + Tc;	       To = Tg + Tj;	       Tp = Tn + To;	       TG = Tu - Tx;	       TH = Tz + TC;	       TI = TG + TH;	  }	  R0[WS(rs, 5)] = FMA(KP2_000000000, Tl, T6);	  R1[WS(rs, 7)] = FMA(KP2_000000000, TE, Tt);	  R1[WS(rs, 2)] = FMA(KP2_000000000, TI, TF);//.........这里部分代码省略.........

static void hc2cbdft_8(R *Rp, R *Ip, R *Rm, R *Im, const R *W, stride rs, INT mb, INT me, INT ms){     DK(KP707106781, +0.707106781186547524400844362104849039284835938);     INT m;     for (m = mb, W = W + ((mb - 1) * 14); m < me; m = m + 1, Rp = Rp + ms, Ip = Ip + ms, Rm = Rm - ms, Im = Im - ms, W = W + 14, MAKE_VOLATILE_STRIDE(rs)) {	  E T7, T1d, T1h, Tl, TG, T14, T19, TO, Te, TL, T18, T15, TB, T1e, Tw;	  E T1i;	  {	       E T3, TC, Tk, TM, T6, Th, TF, TN;	       {		    E T1, T2, Ti, Tj;		    T1 = Rp[0];		    T2 = Rm[WS(rs, 3)];		    T3 = T1 + T2;		    TC = T1 - T2;		    Ti = Ip[0];		    Tj = Im[WS(rs, 3)];		    Tk = Ti + Tj;		    TM = Ti - Tj;	       }	       {		    E T4, T5, TD, TE;		    T4 = Rp[WS(rs, 2)];		    T5 = Rm[WS(rs, 1)];		    T6 = T4 + T5;		    Th = T4 - T5;		    TD = Ip[WS(rs, 2)];		    TE = Im[WS(rs, 1)];		    TF = TD + TE;		    TN = TD - TE;	       }	       T7 = T3 + T6;	       T1d = Tk - Th;	       T1h = TC + TF;	       Tl = Th + Tk;	       TG = TC - TF;	       T14 = T3 - T6;	       T19 = TM - TN;	       TO = TM + TN;	  }	  {	       E Ta, Tm, Tp, TJ, Td, Tr, Tu, TK;	       {		    E T8, T9, Tn, To;		    T8 = Rp[WS(rs, 1)];		    T9 = Rm[WS(rs, 2)];		    Ta = T8 + T9;		    Tm = T8 - T9;		    Tn = Ip[WS(rs, 1)];		    To = Im[WS(rs, 2)];		    Tp = Tn + To;		    TJ = Tn - To;	       }	       {		    E Tb, Tc, Ts, Tt;		    Tb = Rm[0];		    Tc = Rp[WS(rs, 3)];		    Td = Tb + Tc;		    Tr = Tb - Tc;		    Ts = Im[0];		    Tt = Ip[WS(rs, 3)];		    Tu = Ts + Tt;		    TK = Tt - Ts;	       }	       Te = Ta + Td;	       TL = TJ + TK;	       T18 = Ta - Td;	       T15 = TK - TJ;	       {		    E Tz, TA, Tq, Tv;		    Tz = Tm - Tp;		    TA = Tr - Tu;		    TB = KP707106781 * (Tz + TA);		    T1e = KP707106781 * (Tz - TA);		    Tq = Tm + Tp;		    Tv = Tr + Tu;		    Tw = KP707106781 * (Tq - Tv);		    T1i = KP707106781 * (Tq + Tv);	       }	  }	  {	       E Tf, TP, TI, TQ;	       Tf = T7 + Te;	       TP = TL + TO;	       {		    E Tx, TH, Tg, Ty;		    Tx = Tl + Tw;		    TH = TB + TG;		    Tg = W[0];		    Ty = W[1];		    TI = FMA(Tg, Tx, Ty * TH);		    TQ = FNMS(Ty, Tx, Tg * TH);	       }	       Rp[0] = Tf - TI;	       Ip[0] = TP + TQ;	       Rm[0] = Tf + TI;	       Im[0] = TQ - TP;	  }	  {	       E T1r, T1x, T1w, T1y;//.........这里部分代码省略.........

static void t2_8(R *ri, R *ii, const R *W, stride rs, INT mb, INT me, INT ms){     DK(KP707106781, +0.707106781186547524400844362104849039284835938);     {	  INT m;	  for (m = mb, W = W + (mb * 6); m < me; m = m + 1, ri = ri + ms, ii = ii + ms, W = W + 6, MAKE_VOLATILE_STRIDE(16, rs)) {	       E TS, T1m, TJ, T1l, T1k, Tw, T1w, T1u;	       {		    E T2, T3, Tl, Tn, T5, T4, Tm, Tr, T6;		    T2 = W[0];		    T3 = W[2];		    Tl = W[4];		    Tn = W[5];		    T5 = W[1];		    T4 = T2 * T3;		    Tm = T2 * Tl;		    Tr = T2 * Tn;		    T6 = W[3];		    {			 E T1, T1s, TG, Td, T1r, Tu, TY, Tk, TW, T18, T1d, TD, TH, TA, T13;			 E TE, T14;			 {			      E To, Ts, Tf, T7, T8, Ti, Tb, T9, Tc, TC, Ta, TF, TB, Tg, Th;			      E Tj;			      T1 = ri[0];			      To = FMA(T5, Tn, Tm);			      Ts = FNMS(T5, Tl, Tr);			      Tf = FMA(T5, T6, T4);			      T7 = FNMS(T5, T6, T4);			      Ta = T2 * T6;			      T1s = ii[0];			      T8 = ri[WS(rs, 4)];			      TF = Tf * Tn;			      TB = Tf * Tl;			      Ti = FNMS(T5, T3, Ta);			      Tb = FMA(T5, T3, Ta);			      T9 = T7 * T8;			      Tc = ii[WS(rs, 4)];			      TG = FNMS(Ti, Tl, TF);			      TC = FMA(Ti, Tn, TB);			      {				   E Tp, T1q, Tt, Tq, TX;				   Tp = ri[WS(rs, 6)];				   Td = FMA(Tb, Tc, T9);				   T1q = T7 * Tc;				   Tt = ii[WS(rs, 6)];				   Tq = To * Tp;				   Tg = ri[WS(rs, 2)];				   T1r = FNMS(Tb, T8, T1q);				   TX = To * Tt;				   Tu = FMA(Ts, Tt, Tq);				   Th = Tf * Tg;				   Tj = ii[WS(rs, 2)];				   TY = FNMS(Ts, Tp, TX);			      }			      {				   E TO, TQ, TN, TP, T1a, T1b;				   {					E TK, TM, TL, T19, TV;					TK = ri[WS(rs, 7)];					TM = ii[WS(rs, 7)];					Tk = FMA(Ti, Tj, Th);					TV = Tf * Tj;					TL = Tl * TK;					T19 = Tl * TM;					TO = ri[WS(rs, 3)];					TW = FNMS(Ti, Tg, TV);					TQ = ii[WS(rs, 3)];					TN = FMA(Tn, TM, TL);					TP = T3 * TO;					T1a = FNMS(Tn, TK, T19);					T1b = T3 * TQ;				   }				   {					E Tx, Tz, Ty, T12, T1c, TR;					Tx = ri[WS(rs, 1)];					TR = FMA(T6, TQ, TP);					Tz = ii[WS(rs, 1)];					T1c = FNMS(T6, TO, T1b);					Ty = T2 * Tx;					T18 = TN - TR;					TS = TN + TR;					T12 = T2 * Tz;					T1d = T1a - T1c;					T1m = T1a + T1c;					TD = ri[WS(rs, 5)];					TH = ii[WS(rs, 5)];					TA = FMA(T5, Tz, Ty);					T13 = FNMS(T5, Tx, T12);					TE = TC * TD;					T14 = TC * TH;				   }			      }			 }			 {			      E Te, T1p, T1t, Tv;			      {				   E T1g, T10, T1z, T1B, T1A, T1j, T1C, T1f;				   {					E T1x, T11, T16, T1y;//.........这里部分代码省略.........

static void t1_8(float *ri, float *ii, const float *W, stride rs, INT mb, INT me, INT ms){     DK(KP707106781, +0.707106781186547524400844362104849039284835938);     INT m;     for (m = mb, W = W + (mb * 14); m < me; m = m + 1, ri = ri + ms, ii = ii + ms, W = W + 14, MAKE_VOLATILE_STRIDE(rs)) {	  E T1g, T1f, T1e, Tm, T1q, T1o, T1p, TN, T1h, T1i;	  {	       E T1, T1m, T1l, T7, TS, Tk, TQ, Te, To, Tr, T17, TM, T12, Tu, TW;	       E Tp, Tx, Tt, Tq, Tw;	       {		    E T3, T6, T2, T5;		    T1 = ri[0];		    T1m = ii[0];		    T3 = ri[WS(rs, 4)];		    T6 = ii[WS(rs, 4)];		    T2 = W[6];		    T5 = W[7];		    {			 E Ta, Td, T9, Tc;			 {			      E Tg, Tj, Ti, TR, Th, T1k, T4, Tf;			      Tg = ri[WS(rs, 6)];			      Tj = ii[WS(rs, 6)];			      T1k = T2 * T6;			      T4 = T2 * T3;			      Tf = W[10];			      Ti = W[11];			      T1l = FNMS(T5, T3, T1k);			      T7 = FMA(T5, T6, T4);			      TR = Tf * Tj;			      Th = Tf * Tg;			      Ta = ri[WS(rs, 2)];			      Td = ii[WS(rs, 2)];			      TS = FNMS(Ti, Tg, TR);			      Tk = FMA(Ti, Tj, Th);			      T9 = W[2];			      Tc = W[3];			 }			 {			      E TB, TE, TH, T13, TC, TK, TG, TD, TJ, TP, Tb, TA, Tn;			      TB = ri[WS(rs, 7)];			      TE = ii[WS(rs, 7)];			      TP = T9 * Td;			      Tb = T9 * Ta;			      TA = W[12];			      TH = ri[WS(rs, 3)];			      TQ = FNMS(Tc, Ta, TP);			      Te = FMA(Tc, Td, Tb);			      T13 = TA * TE;			      TC = TA * TB;			      TK = ii[WS(rs, 3)];			      TG = W[4];			      TD = W[13];			      TJ = W[5];			      {				   E T14, TF, T16, TL, T15, TI;				   To = ri[WS(rs, 1)];				   T15 = TG * TK;				   TI = TG * TH;				   T14 = FNMS(TD, TB, T13);				   TF = FMA(TD, TE, TC);				   T16 = FNMS(TJ, TH, T15);				   TL = FMA(TJ, TK, TI);				   Tr = ii[WS(rs, 1)];				   Tn = W[0];				   T17 = T14 - T16;				   T1g = T14 + T16;				   TM = TF + TL;				   T12 = TF - TL;			      }			      Tu = ri[WS(rs, 5)];			      TW = Tn * Tr;			      Tp = Tn * To;			      Tx = ii[WS(rs, 5)];			      Tt = W[8];			      Tq = W[1];			      Tw = W[9];			 }		    }	       }	       {		    E T8, T1j, T1n, Tz, T1a, TU, Tl, T1b, T1c, T1v, T1t, T1w, T19, T1u, T1d;		    {			 E T1r, T10, TV, T1s, T11, T18;			 {			      E TO, TX, Ts, TZ, Ty, TT, TY, Tv;			      T8 = T1 + T7;			      TO = T1 - T7;			      TY = Tt * Tx;			      Tv = Tt * Tu;			      TX = FNMS(Tq, To, TW);			      Ts = FMA(Tq, Tr, Tp);			      TZ = FNMS(Tw, Tu, TY);			      Ty = FMA(Tw, Tx, Tv);			      TT = TQ - TS;			      T1j = TQ + TS;			      T1n = T1l + T1m;			      T1r = T1m - T1l;			      T10 = TX - TZ;			      T1f = TX + TZ;//.........这里部分代码省略.........

static void hc2cfdft2_16(R *Rp, R *Ip, R *Rm, R *Im, const R *W, stride rs, INT mb, INT me, INT ms){     DK(KP923879532, +0.923879532511286756128183189396788286822416626);     DK(KP414213562, +0.414213562373095048801688724209698078569671875);     DK(KP707106781, +0.707106781186547524400844362104849039284835938);     DK(KP500000000, +0.500000000000000000000000000000000000000000000);     {	  INT m;	  for (m = mb, W = W + ((mb - 1) * 8); m < me; m = m + 1, Rp = Rp + ms, Ip = Ip + ms, Rm = Rm - ms, Im = Im - ms, W = W + 8, MAKE_VOLATILE_STRIDE(64, rs)) {	       E T4p, T4o, T4n, T4s;	       {		    E T1, T2, Tw, Ty, Th, T3, Tx, TE, Ti, TK, Tj, T4, T5;		    T1 = W[0];		    T2 = W[2];		    Tw = W[6];		    Ty = W[7];		    Th = W[4];		    T3 = T1 * T2;		    Tx = T1 * Tw;		    TE = T1 * Ty;		    Ti = T1 * Th;		    TK = T2 * Th;		    Tj = W[5];		    T4 = W[1];		    T5 = W[3];		    {			 E T1v, T2q, T1s, T2s, T38, T3T, T1Y, T3P, T17, T1h, T2x, T2v, T33, T3Q, T3S;			 E T1N, Tv, T3A, T2E, T3B, T3L, T2c, T3I, T2S, TW, T3E, T3J, T2n, T3D, T2J;			 E T3M, T2X;			 {			      E TF, Tk, Tz, TL, T6, TR, Tq, Tc, T2h, T25, T2k, T29, T1G, T1M, T2P;			      E T2R;			      {				   E T18, TY, T1d, T13, T1H, T1A, T1K, T1E, T37, T1R, T35, T1X;				   {					E T1j, T1o, T1W, T1p, T1m, T1Q, T1U, T1q;					{					     E T1k, T1l, T1S, T1T;					     {						  E T1t, T28, T24, T1D, T1z, T1u, TQ, Tp, Tb;						  T1t = Ip[0];						  TQ = T2 * Tj;						  Tp = T1 * Tj;						  TF = FNMS(T4, Tw, TE);						  T1j = FMA(T4, Tj, Ti);						  Tk = FNMS(T4, Tj, Ti);						  Tz = FMA(T4, Ty, Tx);						  T18 = FNMS(T5, Tj, TK);						  TL = FMA(T5, Tj, TK);						  TY = FNMS(T4, T5, T3);						  T6 = FMA(T4, T5, T3);						  Tb = T1 * T5;						  TR = FNMS(T5, Th, TQ);						  T1d = FMA(T5, Th, TQ);						  Tq = FMA(T4, Th, Tp);						  T1o = FNMS(T4, Th, Tp);						  T28 = T6 * Tj;						  T24 = T6 * Th;						  T1D = TY * Tj;						  T1z = TY * Th;						  Tc = FNMS(T4, T2, Tb);						  T13 = FMA(T4, T2, Tb);						  T1u = Im[0];						  T1k = Ip[WS(rs, 4)];						  T2h = FMA(Tc, Tj, T24);						  T25 = FNMS(Tc, Tj, T24);						  T2k = FNMS(Tc, Th, T28);						  T29 = FMA(Tc, Th, T28);						  T1H = FNMS(T13, Tj, T1z);						  T1A = FMA(T13, Tj, T1z);						  T1K = FMA(T13, Th, T1D);						  T1E = FNMS(T13, Th, T1D);						  T1W = T1t + T1u;						  T1v = T1t - T1u;						  T1l = Im[WS(rs, 4)];					     }					     T1S = Rm[0];					     T1T = Rp[0];					     T1p = Rp[WS(rs, 4)];					     T1m = T1k - T1l;					     T1Q = T1k + T1l;					     T2q = T1T + T1S;					     T1U = T1S - T1T;					     T1q = Rm[WS(rs, 4)];					}					{					     E T36, T1V, T1O, T1r, T1n, T1P, T34, T2r;					     T36 = T4 * T1U;					     T1V = T1 * T1U;					     T1O = T1q - T1p;					     T1r = T1p + T1q;					     T1n = T1j * T1m;					     T37 = FMA(T1, T1W, T36);					     T2r = T1j * T1r;					     T1P = Th * T1O;					     T34 = Tj * T1O;					     T1s = FNMS(T1o, T1r, T1n);					     T2s = FMA(T1o, T1m, T2r);					     T1R = FNMS(Tj, T1Q, T1P);					     T35 = FMA(Th, T1Q, T34);//.........这里部分代码省略.........

static void hc2cfdft_4(R *Rp, R *Ip, R *Rm, R *Im, const R *W, stride rs, INT mb, INT me, INT ms){     DK(KP500000000, +0.500000000000000000000000000000000000000000000);     {	  INT m;	  for (m = mb, W = W + ((mb - 1) * 6); m < me; m = m + 1, Rp = Rp + ms, Ip = Ip + ms, Rm = Rm - ms, Im = Im - ms, W = W + 6, MAKE_VOLATILE_STRIDE(16, rs)) {	       E Tc, Tr, Tk, Tx, T9, Ts, Tp, Tw;	       {		    E Ta, Tb, Tj, Tf, Tg, Th, Te, Ti;		    Ta = Ip[0];		    Tb = Im[0];		    Tj = Ta + Tb;		    Tf = Rm[0];		    Tg = Rp[0];		    Th = Tf - Tg;		    Tc = Ta - Tb;		    Tr = Tg + Tf;		    Te = W[0];		    Ti = W[1];		    Tk = FNMS(Ti, Tj, Te * Th);		    Tx = FMA(Ti, Th, Te * Tj);	       }	       {		    E T4, To, T8, Tm;		    {			 E T2, T3, T6, T7;			 T2 = Ip[WS(rs, 1)];			 T3 = Im[WS(rs, 1)];			 T4 = T2 - T3;			 To = T2 + T3;			 T6 = Rp[WS(rs, 1)];			 T7 = Rm[WS(rs, 1)];			 T8 = T6 + T7;			 Tm = T6 - T7;		    }		    {			 E T1, T5, Tl, Tn;			 T1 = W[2];			 T5 = W[3];			 T9 = FNMS(T5, T8, T1 * T4);			 Ts = FMA(T1, T8, T5 * T4);			 Tl = W[4];			 Tn = W[5];			 Tp = FMA(Tl, Tm, Tn * To);			 Tw = FNMS(Tn, Tm, Tl * To);		    }	       }	       {		    E Td, Tq, Tz, TA;		    Td = T9 + Tc;		    Tq = Tk - Tp;		    Ip[0] = KP500000000 * (Td + Tq);		    Im[WS(rs, 1)] = KP500000000 * (Tq - Td);		    Tz = Tr + Ts;		    TA = Tw + Tx;		    Rm[WS(rs, 1)] = KP500000000 * (Tz - TA);		    Rp[0] = KP500000000 * (Tz + TA);	       }	       {		    E Tt, Tu, Tv, Ty;		    Tt = Tr - Ts;		    Tu = Tp + Tk;		    Rm[0] = KP500000000 * (Tt - Tu);		    Rp[WS(rs, 1)] = KP500000000 * (Tt + Tu);		    Tv = Tc - T9;		    Ty = Tw - Tx;		    Ip[WS(rs, 1)] = KP500000000 * (Tv + Ty);		    Im[0] = KP500000000 * (Ty - Tv);	       }	  }     }}

static void r2hc_11(const R *I, R *ro, R *io, stride is, stride ros, stride ios, INT v, INT ivs, INT ovs){     DK(KP959492973, +0.959492973614497389890368057066327699062454848);     DK(KP876768831, +0.876768831002589333891339807079336796764054852);     DK(KP918985947, +0.918985947228994779780736114132655398124909697);     DK(KP989821441, +0.989821441880932732376092037776718787376519372);     DK(KP778434453, +0.778434453334651800608337670740821884709317477);     DK(KP830830026, +0.830830026003772851058548298459246407048009821);     DK(KP715370323, +0.715370323453429719112414662767260662417897278);     DK(KP634356270, +0.634356270682424498893150776899916060542806975);     DK(KP342584725, +0.342584725681637509502641509861112333758894680);     DK(KP521108558, +0.521108558113202722944698153526659300680427422);     INT i;     for (i = v; i > 0; i = i - 1, I = I + ivs, ro = ro + ovs, io = io + ovs, MAKE_VOLATILE_STRIDE(is), MAKE_VOLATILE_STRIDE(ros), MAKE_VOLATILE_STRIDE(ios)) {	  E T1, Tg, TF, TB, TI, TL, Tz, TA;	  {	       E T4, TC, TE, T7, TD, Ta, TS, TG, TJ, Td, TP, TM, Ty, Tq, Th;	       E Tt, Tl;	       T1 = I[0];	       {		    E Tb, Tc, Tx, Tp;		    {			 E T2, T3, Te, Tf;			 T2 = I[WS(is, 1)];			 T3 = I[WS(is, 10)];			 Te = I[WS(is, 5)];			 Tf = I[WS(is, 6)];			 {			      E T5, T6, T8, T9;			      T5 = I[WS(is, 2)];			      T4 = T2 + T3;			      TC = T3 - T2;			      Tg = Te + Tf;			      TE = Tf - Te;			      T6 = I[WS(is, 9)];			      T8 = I[WS(is, 3)];			      T9 = I[WS(is, 8)];			      Tb = I[WS(is, 4)];			      T7 = T5 + T6;			      TD = T5 - T6;			      Ta = T8 + T9;			      TF = T9 - T8;			      Tc = I[WS(is, 7)];			 }		    }		    TS = FMA(KP521108558, TC, TD);		    TG = FMA(KP521108558, TF, TE);		    TJ = FMA(KP521108558, TE, TC);		    Td = Tb + Tc;		    TB = Tb - Tc;		    Tx = FNMS(KP342584725, Ta, T7);		    Tp = FNMS(KP342584725, T4, Ta);		    TP = FNMS(KP521108558, TB, TF);		    TM = FNMS(KP521108558, TD, TB);		    Ty = FNMS(KP634356270, Tx, Td);		    Tq = FNMS(KP634356270, Tp, Tg);		    Th = FNMS(KP342584725, Tg, Td);		    Tt = FNMS(KP342584725, Td, T4);		    Tl = FNMS(KP342584725, T7, Tg);	       }	       {		    E Tu, Ts, TN, Tv;		    {			 E Tm, TU, Tj, Ti, TT;			 TT = FMA(KP715370323, TS, TF);			 Ti = FNMS(KP634356270, Th, Ta);			 Tu = FNMS(KP634356270, Tt, T7);			 Tm = FNMS(KP634356270, Tl, T4);			 TU = FMA(KP830830026, TT, TB);			 Tj = FNMS(KP778434453, Ti, T7);			 {			      E Tk, TR, To, Tn, TQ, Tr;			      TQ = FMA(KP715370323, TP, TC);			      Tn = FNMS(KP778434453, Tm, Ta);			      io[WS(ios, 5)] = KP989821441 * (FMA(KP918985947, TU, TE));			      Tk = FNMS(KP876768831, Tj, T4);			      TR = FNMS(KP830830026, TQ, TE);			      To = FNMS(KP876768831, Tn, Td);			      Tr = FNMS(KP778434453, Tq, Td);			      ro[WS(ros, 5)] = FNMS(KP959492973, Tk, T1);			      io[WS(ios, 4)] = KP989821441 * (FNMS(KP918985947, TR, TD));			      ro[WS(ros, 4)] = FNMS(KP959492973, To, T1);			      Ts = FNMS(KP876768831, Tr, T7);			 }		    }		    TN = FNMS(KP715370323, TM, TE);		    Tv = FNMS(KP778434453, Tu, Tg);		    ro[0] = T1 + T4 + T7 + Ta + Td + Tg;		    ro[WS(ros, 3)] = FNMS(KP959492973, Ts, T1);		    {			 E TO, Tw, TH, TK;			 TO = FNMS(KP830830026, TN, TF);			 Tw = FNMS(KP876768831, Tv, Ta);			 TH = FMA(KP715370323, TG, TD);			 TK = FNMS(KP715370323, TJ, TB);			 io[WS(ios, 3)] = KP989821441 * (FNMS(KP918985947, TO, TC));			 ro[WS(ros, 2)] = FNMS(KP959492973, Tw, T1);			 TI = FNMS(KP830830026, TH, TC);			 TL = FMA(KP830830026, TK, TD);			 Tz = FNMS(KP778434453, Ty, T4);//.........这里部分代码省略.........

static void hf_25(R *cr, R *ci, const R *W, stride rs, INT mb, INT me, INT ms){     DK(KP949179823, +0.949179823508441261575555465843363271711583843);     DK(KP860541664, +0.860541664367944677098261680920518816412804187);     DK(KP621716863, +0.621716863012209892444754556304102309693593202);     DK(KP614372930, +0.614372930789563808870829930444362096004872855);     DK(KP557913902, +0.557913902031834264187699648465567037992437152);     DK(KP249506682, +0.249506682107067890488084201715862638334226305);     DK(KP560319534, +0.560319534973832390111614715371676131169633784);     DK(KP681693190, +0.681693190061530575150324149145440022633095390);     DK(KP906616052, +0.906616052148196230441134447086066874408359177);     DK(KP968479752, +0.968479752739016373193524836781420152702090879);     DK(KP845997307, +0.845997307939530944175097360758058292389769300);     DK(KP998026728, +0.998026728428271561952336806863450553336905220);     DK(KP994076283, +0.994076283785401014123185814696322018529298887);     DK(KP734762448, +0.734762448793050413546343770063151342619912334);     DK(KP772036680, +0.772036680810363904029489473607579825330539880);     DK(KP062914667, +0.062914667253649757225485955897349402364686947);     DK(KP833417178, +0.833417178328688677408962550243238843138996060);     DK(KP921177326, +0.921177326965143320250447435415066029359282231);     DK(KP541454447, +0.541454447536312777046285590082819509052033189);     DK(KP803003575, +0.803003575438660414833440593570376004635464850);     DK(KP943557151, +0.943557151597354104399655195398983005179443399);     DK(KP554608978, +0.554608978404018097464974850792216217022558774);     DK(KP242145790, +0.242145790282157779872542093866183953459003101);     DK(KP559154169, +0.559154169276087864842202529084232643714075927);     DK(KP683113946, +0.683113946453479238701949862233725244439656928);     DK(KP248028675, +0.248028675328619457762448260696444630363259177);     DK(KP968583161, +0.968583161128631119490168375464735813836012403);     DK(KP525970792, +0.525970792408939708442463226536226366643874659);     DK(KP726211448, +0.726211448929902658173535992263577167607493062);     DK(KP904730450, +0.904730450839922351881287709692877908104763647);     DK(KP831864738, +0.831864738706457140726048799369896829771167132);     DK(KP871714437, +0.871714437527667770979999223229522602943903653);     DK(KP549754652, +0.549754652192770074288023275540779861653779767);     DK(KP992114701, +0.992114701314477831049793042785778521453036709);     DK(KP939062505, +0.939062505817492352556001843133229685779824606);     DK(KP256756360, +0.256756360367726783319498520922669048172391148);     DK(KP851038619, +0.851038619207379630836264138867114231259902550);     DK(KP912575812, +0.912575812670962425556968549836277086778922727);     DK(KP912018591, +0.912018591466481957908415381764119056233607330);     DK(KP634619297, +0.634619297544148100711287640319130485732531031);     DK(KP470564281, +0.470564281212251493087595091036643380879947982);     DK(KP827271945, +0.827271945972475634034355757144307982555673741);     DK(KP126329378, +0.126329378446108174786050455341811215027378105);     DK(KP951056516, +0.951056516295153572116439333379382143405698634);     DK(KP559016994, +0.559016994374947424102293417182819058860154590);     DK(KP250000000, +0.250000000000000000000000000000000000000000000);     DK(KP618033988, +0.618033988749894848204586834365638117720309180);     {	  INT m;	  for (m = mb, W = W + ((mb - 1) * 48); m < me; m = m + 1, cr = cr + ms, ci = ci - ms, W = W + 48, MAKE_VOLATILE_STRIDE(50, rs)) {	       E T7i, T6o, T6m, T7o, T7m, T7h, T6n, T6f, T7j, T7n;	       {		    E T6W, T5G, T3Y, T3M, T7q, T70, T6V, T7P, Tt, T3L, T5T, T45, T5Q, T4c, T3G;		    E T2G, T5P, T49, T5S, T42, T65, T4H, T68, T4A, T2Z, T11, T67, T4x, T64, T4E;		    E T5Y, T4W, T61, T4P, T3d, T1z, T60, T4M, T5X, T4T, T3g, T1G, T3q, T4q, T4j;		    E T26, T3i, T1M, T3k, T1S;		    {			 E T3u, T2e, T3E, T44, T4b, T2E, T3w, T2k, T3y, T2q;			 {			      E T1, T6R, T3P, T7, T3W, Tq, T9, Tc, Tb, T3U, Tk, T3Q, Ta;			      {				   E T3, T6, T2, T5;				   T1 = cr[0];				   T6R = ci[0];				   T3 = cr[WS(rs, 5)];				   T6 = ci[WS(rs, 5)];				   T2 = W[8];				   T5 = W[9];				   {					E Tm, Tp, To, T3V, Tn, T3O, T4, Tl;					Tm = cr[WS(rs, 15)];					Tp = ci[WS(rs, 15)];					T3O = T2 * T6;					T4 = T2 * T3;					Tl = W[28];					To = W[29];					T3P = FNMS(T5, T3, T3O);					T7 = FMA(T5, T6, T4);					T3V = Tl * Tp;					Tn = Tl * Tm;					{					     E Tg, Tj, Tf, Ti, T3T, Th, T8;					     Tg = cr[WS(rs, 10)];					     Tj = ci[WS(rs, 10)];					     T3W = FNMS(To, Tm, T3V);					     Tq = FMA(To, Tp, Tn);					     Tf = W[18];					     Ti = W[19];					     T9 = cr[WS(rs, 20)];					     Tc = ci[WS(rs, 20)];					     T3T = Tf * Tj;					     Th = Tf * Tg;					     T8 = W[38];					     Tb = W[39];					     T3U = FNMS(Ti, Tg, T3T);					     Tk = FMA(Ti, Tj, Th);					     T3Q = T8 * Tc;					     Ta = T8 * T9;//.........这里部分代码省略.........

static void hb_5(R *cr, R *ci, const R *W, stride rs, INT mb, INT me, INT ms){     DK(KP250000000, +0.250000000000000000000000000000000000000000000);     DK(KP587785252, +0.587785252292473129168705954639072768597652438);     DK(KP951056516, +0.951056516295153572116439333379382143405698634);     DK(KP559016994, +0.559016994374947424102293417182819058860154590);     INT m;     for (m = mb, W = W + ((mb - 1) * 8); m < me; m = m + 1, cr = cr + ms, ci = ci - ms, W = W + 8, MAKE_VOLATILE_STRIDE(rs)) {	  E T1, Tj, TG, Ts, T8, Ti, T9, Tn, TD, Tu, Tg, Tt;	  {	       E T4, Tq, T7, Tr;	       T1 = cr[0];	       {		    E T2, T3, T5, T6;		    T2 = cr[WS(rs, 1)];		    T3 = ci[0];		    T4 = T2 + T3;		    Tq = T2 - T3;		    T5 = cr[WS(rs, 2)];		    T6 = ci[WS(rs, 1)];		    T7 = T5 + T6;		    Tr = T5 - T6;	       }	       Tj = KP559016994 * (T4 - T7);	       TG = FMA(KP951056516, Tq, KP587785252 * Tr);	       Ts = FNMS(KP951056516, Tr, KP587785252 * Tq);	       T8 = T4 + T7;	       Ti = FNMS(KP250000000, T8, T1);	  }	  {	       E Tc, Tl, Tf, Tm;	       T9 = ci[WS(rs, 4)];	       {		    E Ta, Tb, Td, Te;		    Ta = ci[WS(rs, 3)];		    Tb = cr[WS(rs, 4)];		    Tc = Ta - Tb;		    Tl = Ta + Tb;		    Td = ci[WS(rs, 2)];		    Te = cr[WS(rs, 3)];		    Tf = Td - Te;		    Tm = Td + Te;	       }	       Tn = FNMS(KP951056516, Tm, KP587785252 * Tl);	       TD = FMA(KP951056516, Tl, KP587785252 * Tm);	       Tu = KP559016994 * (Tc - Tf);	       Tg = Tc + Tf;	       Tt = FNMS(KP250000000, Tg, T9);	  }	  cr[0] = T1 + T8;	  ci[0] = T9 + Tg;	  {	       E To, Ty, Tw, TA, Tk, Tv;	       Tk = Ti - Tj;	       To = Tk - Tn;	       Ty = Tk + Tn;	       Tv = Tt - Tu;	       Tw = Ts + Tv;	       TA = Tv - Ts;	       {		    E Th, Tp, Tx, Tz;		    Th = W[2];		    Tp = W[3];		    cr[WS(rs, 2)] = FNMS(Tp, Tw, Th * To);		    ci[WS(rs, 2)] = FMA(Th, Tw, Tp * To);		    Tx = W[4];		    Tz = W[5];		    cr[WS(rs, 3)] = FNMS(Tz, TA, Tx * Ty);		    ci[WS(rs, 3)] = FMA(Tx, TA, Tz * Ty);	       }	  }	  {	       E TE, TK, TI, TM, TC, TH;	       TC = Tj + Ti;	       TE = TC - TD;	       TK = TC + TD;	       TH = Tu + Tt;	       TI = TG + TH;	       TM = TH - TG;	       {		    E TB, TF, TJ, TL;		    TB = W[0];		    TF = W[1];		    cr[WS(rs, 1)] = FNMS(TF, TI, TB * TE);		    ci[WS(rs, 1)] = FMA(TB, TI, TF * TE);		    TJ = W[6];		    TL = W[7];		    cr[WS(rs, 4)] = FNMS(TL, TM, TJ * TK);		    ci[WS(rs, 4)] = FMA(TJ, TM, TL * TK);	       }	  }     }}

static void hc2cb_16(float *Rp, float *Ip, float *Rm, float *Im, const float *W, stride rs, INT mb, INT me, INT ms){     DK(KP382683432, +0.382683432365089771728459984030398866761344562);     DK(KP923879532, +0.923879532511286756128183189396788286822416626);     DK(KP707106781, +0.707106781186547524400844362104849039284835938);     INT m;     for (m = mb, W = W + ((mb - 1) * 30); m < me; m = m + 1, Rp = Rp + ms, Ip = Ip + ms, Rm = Rm - ms, Im = Im - ms, W = W + 30, MAKE_VOLATILE_STRIDE(rs)) {	  E T7, T2K, T2W, Tw, T17, T1S, T2k, T1w, Te, TD, T1x, T10, T2n, T2L, T1Z;	  E T2X, Tm, T1z, TN, T19, T2e, T2p, T2P, T2Z, Tt, T1A, TW, T1a, T27, T2q;	  E T2S, T30;	  {	       E T3, T1Q, T13, T2j, T6, T2i, T16, T1R;	       {		    E T1, T2, T11, T12;		    T1 = Rp[0];		    T2 = Rm[WS(rs, 7)];		    T3 = T1 + T2;		    T1Q = T1 - T2;		    T11 = Ip[0];		    T12 = Im[WS(rs, 7)];		    T13 = T11 - T12;		    T2j = T11 + T12;	       }	       {		    E T4, T5, T14, T15;		    T4 = Rp[WS(rs, 4)];		    T5 = Rm[WS(rs, 3)];		    T6 = T4 + T5;		    T2i = T4 - T5;		    T14 = Ip[WS(rs, 4)];		    T15 = Im[WS(rs, 3)];		    T16 = T14 - T15;		    T1R = T14 + T15;	       }	       T7 = T3 + T6;	       T2K = T1Q + T1R;	       T2W = T2j - T2i;	       Tw = T3 - T6;	       T17 = T13 - T16;	       T1S = T1Q - T1R;	       T2k = T2i + T2j;	       T1w = T13 + T16;	  }	  {	       E Ta, T1T, TC, T1U, Td, T1W, Tz, T1X;	       {		    E T8, T9, TA, TB;		    T8 = Rp[WS(rs, 2)];		    T9 = Rm[WS(rs, 5)];		    Ta = T8 + T9;		    T1T = T8 - T9;		    TA = Ip[WS(rs, 2)];		    TB = Im[WS(rs, 5)];		    TC = TA - TB;		    T1U = TA + TB;	       }	       {		    E Tb, Tc, Tx, Ty;		    Tb = Rm[WS(rs, 1)];		    Tc = Rp[WS(rs, 6)];		    Td = Tb + Tc;		    T1W = Tb - Tc;		    Tx = Ip[WS(rs, 6)];		    Ty = Im[WS(rs, 1)];		    Tz = Tx - Ty;		    T1X = Tx + Ty;	       }	       Te = Ta + Td;	       TD = Tz - TC;	       T1x = TC + Tz;	       T10 = Ta - Td;	       {		    E T2l, T2m, T1V, T1Y;		    T2l = T1T + T1U;		    T2m = T1W + T1X;		    T2n = KP707106781 * (T2l - T2m);		    T2L = KP707106781 * (T2l + T2m);		    T1V = T1T - T1U;		    T1Y = T1W - T1X;		    T1Z = KP707106781 * (T1V + T1Y);		    T2X = KP707106781 * (T1V - T1Y);	       }	  }	  {	       E Ti, T2b, TI, T29, Tl, T28, TL, T2c, TF, TM;	       {		    E Tg, Th, TG, TH;		    Tg = Rp[WS(rs, 1)];		    Th = Rm[WS(rs, 6)];		    Ti = Tg + Th;		    T2b = Tg - Th;		    TG = Ip[WS(rs, 1)];		    TH = Im[WS(rs, 6)];		    TI = TG - TH;		    T29 = TG + TH;	       }	       {		    E Tj, Tk, TJ, TK;		    Tj = Rp[WS(rs, 5)];		    Tk = Rm[WS(rs, 2)];//.........这里部分代码省略.........