static void r2cf_64(R *R0, R *R1, R *Cr, R *Ci, stride rs, stride csr, stride csi, INT v, INT ivs, INT ovs)
{
     DK(KP773010453, +0.773010453362736960810906609758469800971041293);
     DK(KP995184726, +0.995184726672196886244836953109479921575474869);
     DK(KP098491403, +0.098491403357164253077197521291327432293052451);
     DK(KP820678790, +0.820678790828660330972281985331011598767386482);
     DK(KP956940335, +0.956940335732208864935797886980269969482849206);
     DK(KP881921264, +0.881921264348355029712756863660388349508442621);
     DK(KP534511135, +0.534511135950791641089685961295362908582039528);
     DK(KP303346683, +0.303346683607342391675883946941299872384187453);
     DK(KP980785280, +0.980785280403230449126182236134239036973933731);
     DK(KP198912367, +0.198912367379658006911597622644676228597850501);
     DK(KP831469612, +0.831469612302545237078788377617905756738560812);
     DK(KP668178637, +0.668178637919298919997757686523080761552472251);
     DK(KP923879532, +0.923879532511286756128183189396788286822416626);
     DK(KP414213562, +0.414213562373095048801688724209698078569671875);
     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 T5n, T5o;
	  {
	       E T11, T2j, T4P, T5P, T3D, T5p, T3d, Tf, T1k, T1H, T5D, T4l, T5A, T4a, T3i;
	       E T2U, T1R, T2e, T5K, T4G, T5H, T4v, T3l, T31, T5s, T42, T5t, T3Z, T2n, T1b;
	       E T3f, TZ, T5v, T3T, T5w, T3Q, T2m, T18, T3e, TK, T3K, T5Q, T4S, T5q, T14;
	       E T2k, T3p, Tu, T4w, T1U, T5E, T4h, T5B, T4o, T3j, T2X, T1I, T1z, T1Z, T4A;
	       E T24, T4x, T1X, T20;
	       {
		    E TN, T3V, TS, TX, T3X, TQ, T40, TT;
		    {
			 E T1g, T46, T1B, T1G, T47, T1j, T4j, T1C;
			 {
			      E T4, T3z, T3, T3B, Td, T5, T8, T9;
			      {
				   E T1, T2, Tb, Tc;
				   T1 = R0[0];
				   T2 = R0[WS(rs, 16)];
				   Tb = R0[WS(rs, 28)];
				   Tc = R0[WS(rs, 12)];
				   T4 = R0[WS(rs, 8)];
				   T3z = T1 - T2;
				   T3 = T1 + T2;
				   T3B = Tb - Tc;
				   Td = Tb + Tc;
				   T5 = R0[WS(rs, 24)];
				   T8 = R0[WS(rs, 4)];
				   T9 = R0[WS(rs, 20)];
			      }
			      {
				   E T1E, T1F, T1h, T1i;
				   {
					E T1e, T4N, T6, T3A, Ta, T1f;
					T1e = R1[0];
					T4N = T4 - T5;
					T6 = T4 + T5;
					T3A = T8 - T9;
					Ta = T8 + T9;
					T1f = R1[WS(rs, 16)];
					{
					     E T7, T3C, T4O, Te;
					     T11 = T3 - T6;
					     T7 = T3 + T6;
					     T3C = T3A + T3B;
					     T4O = T3B - T3A;
					     T2j = Td - Ta;
					     Te = Ta + Td;
					     T4P = FNMS(KP707106781, T4O, T4N);
					     T5P = FMA(KP707106781, T4O, T4N);
					     T3D = FMA(KP707106781, T3C, T3z);
					     T5p = FNMS(KP707106781, T3C, T3z);
					     T3d = T7 - Te;
					     Tf = T7 + Te;
					     T1g = T1e + T1f;
					     T46 = T1e - T1f;
					}
				   }
				   T1E = R1[WS(rs, 4)];
				   T1F = R1[WS(rs, 20)];
				   T1h = R1[WS(rs, 8)];
				   T1i = R1[WS(rs, 24)];
				   T1B = R1[WS(rs, 28)];
				   T1G = T1E + T1F;
				   T47 = T1E - T1F;
				   T1j = T1h + T1i;
				   T4j = T1h - T1i;
				   T1C = R1[WS(rs, 12)];
			      }
			 }
			 {
			      E T1N, T4r, T28, T2d, T4s, T1Q, T4E, T29;
			      {
				   E T2b, T2c, T1O, T1P;
				   {
					E T2S, T48, T1D, T1L, T1M, T4k, T49, T2T;
					T1L = R1[WS(rs, 31)];
					T1M = R1[WS(rs, 15)];
					T2S = T1g + T1j;
					T1k = T1g - T1j;
					T48 = T1B - T1C;
					T1D = T1B + T1C;
					T1N = T1L + T1M;
					T4r = T1L - T1M;
					T4k = T47 - T48;
					T49 = T47 + T48;
					T2T = T1G + T1D;
					T1H = T1D - T1G;
					T5D = FNMS(KP707106781, T4k, T4j);
					T4l = FMA(KP707106781, T4k, T4j);
					T5A = FNMS(KP707106781, T49, T46);
					T4a = FMA(KP707106781, T49, T46);
					T3i = T2S - T2T;
					T2U = T2S + T2T;
					T2b = R1[WS(rs, 3)];
					T2c = R1[WS(rs, 19)];
				   }
				   T1O = R1[WS(rs, 7)];
				   T1P = R1[WS(rs, 23)];
				   T28 = R1[WS(rs, 27)];
				   T2d = T2b + T2c;
				   T4s = T2b - T2c;
				   T1Q = T1O + T1P;
				   T4E = T1P - T1O;
				   T29 = R1[WS(rs, 11)];
			      }
			      {
				   E TV, TW, TO, TP;
				   {
					E T2Z, T4t, T2a, TL, TM, T4F, T4u, T30;
					TL = R0[WS(rs, 31)];
					TM = R0[WS(rs, 15)];
					T2Z = T1N + T1Q;
					T1R = T1N - T1Q;
					T4t = T28 - T29;
					T2a = T28 + T29;
					TN = TL + TM;
					T3V = TL - TM;
					T4F = T4t - T4s;
					T4u = T4s + T4t;
					T30 = T2d + T2a;
					T2e = T2a - T2d;
					T5K = FNMS(KP707106781, T4F, T4E);
					T4G = FMA(KP707106781, T4F, T4E);
					T5H = FNMS(KP707106781, T4u, T4r);
					T4v = FMA(KP707106781, T4u, T4r);
					T3l = T2Z - T30;
					T31 = T2Z + T30;
					TV = R0[WS(rs, 27)];
					TW = R0[WS(rs, 11)];
				   }
				   TO = R0[WS(rs, 7)];
				   TP = R0[WS(rs, 23)];
				   TS = R0[WS(rs, 3)];
				   TX = TV + TW;
				   T3X = TV - TW;
				   TQ = TO + TP;
				   T40 = TO - TP;
				   TT = R0[WS(rs, 19)];
			      }
			 }
		    }
		    {
			 E Ti, T3E, Tn, Ts, T3I, Tl, T3F, To;
			 {
			      E Ty, T3M, TD, TI, T3O, TB, T3R, TE;
			      {
				   E TG, TH, Tz, TA;
				   {
					E T19, TR, T3W, TU, Tw, Tx;
					Tw = R0[WS(rs, 1)];
					Tx = R0[WS(rs, 17)];
					T19 = TN - TQ;
					TR = TN + TQ;
					T3W = TS - TT;
					TU = TS + TT;
					Ty = Tw + Tx;
					T3M = Tw - Tx;
					{
					     E T41, T3Y, T1a, TY;
					     T41 = T3W - T3X;
					     T3Y = T3W + T3X;
					     T1a = TX - TU;
					     TY = TU + TX;
					     T5s = FNMS(KP707106781, T41, T40);
					     T42 = FMA(KP707106781, T41, T40);
					     T5t = FNMS(KP707106781, T3Y, T3V);
					     T3Z = FMA(KP707106781, T3Y, T3V);
					     T2n = FMA(KP414213562, T19, T1a);
					     T1b = FNMS(KP414213562, T1a, T19);
					     T3f = TR - TY;
					     TZ = TR + TY;
					     TG = R0[WS(rs, 29)];
					     TH = R0[WS(rs, 13)];
					}
				   }
				   Tz = R0[WS(rs, 9)];
				   TA = R0[WS(rs, 25)];
				   TD = R0[WS(rs, 5)];
				   TI = TG + TH;
				   T3O = TG - TH;
				   TB = Tz + TA;
				   T3R = Tz - TA;
				   TE = R0[WS(rs, 21)];
			      }
			      {
				   E Tq, Tr, Tj, Tk;
				   {
					E T16, TC, T3N, TF, Tg, Th;
					Tg = R0[WS(rs, 2)];
					Th = R0[WS(rs, 18)];
					T16 = Ty - TB;
					TC = Ty + TB;
					T3N = TD - TE;
					TF = TD + TE;
					Ti = Tg + Th;
					T3E = Tg - Th;
					{
					     E T3S, T3P, T17, TJ;
					     T3S = T3N - T3O;
					     T3P = T3N + T3O;
					     T17 = TI - TF;
					     TJ = TF + TI;
					     T5v = FNMS(KP707106781, T3S, T3R);
					     T3T = FMA(KP707106781, T3S, T3R);
					     T5w = FNMS(KP707106781, T3P, T3M);
					     T3Q = FMA(KP707106781, T3P, T3M);
					     T2m = FNMS(KP414213562, T16, T17);
					     T18 = FMA(KP414213562, T17, T16);
					     T3e = TC - TJ;
					     TK = TC + TJ;
					     Tq = R0[WS(rs, 6)];
					     Tr = R0[WS(rs, 22)];
					}
				   }
				   Tj = R0[WS(rs, 10)];
				   Tk = R0[WS(rs, 26)];
				   Tn = R0[WS(rs, 30)];
				   Ts = Tq + Tr;
				   T3I = Tq - Tr;
				   Tl = Tj + Tk;
				   T3F = Tj - Tk;
				   To = R0[WS(rs, 14)];
			      }
			 }
			 {
			      E T1n, T4b, T1s, T4f, T1x, T4c, T1q, T1t;
			      {
				   E T1v, T1w, T1o, T1p;
				   {
					E T1l, T4Q, T3G, Tm, T12, Tp, T3H, T1m;
					T1l = R1[WS(rs, 2)];
					T4Q = FMA(KP414213562, T3E, T3F);
					T3G = FNMS(KP414213562, T3F, T3E);
					Tm = Ti + Tl;
					T12 = Ti - Tl;
					Tp = Tn + To;
					T3H = Tn - To;
					T1m = R1[WS(rs, 18)];
					T1v = R1[WS(rs, 6)];
					{
					     E T4R, T3J, Tt, T13;
					     T4R = FNMS(KP414213562, T3H, T3I);
					     T3J = FMA(KP414213562, T3I, T3H);
					     Tt = Tp + Ts;
					     T13 = Tp - Ts;
					     T1n = T1l + T1m;
					     T4b = T1l - T1m;
					     T3K = T3G + T3J;
					     T5Q = T3J - T3G;
					     T4S = T4Q + T4R;
					     T5q = T4Q - T4R;
					     T14 = T12 + T13;
					     T2k = T13 - T12;
					     T3p = Tt - Tm;
					     Tu = Tm + Tt;
					     T1w = R1[WS(rs, 22)];
					}
				   }
				   T1o = R1[WS(rs, 10)];
				   T1p = R1[WS(rs, 26)];
				   T1s = R1[WS(rs, 30)];
				   T4f = T1v - T1w;
				   T1x = T1v + T1w;
				   T4c = T1o - T1p;
				   T1q = T1o + T1p;
				   T1t = R1[WS(rs, 14)];
			      }
			      {
				   E T22, T23, T1V, T1W;
				   {
					E T1S, T4d, T4m, T2V, T1r, T4e, T1u, T1T;
					T1S = R1[WS(rs, 1)];
					T4d = FNMS(KP414213562, T4c, T4b);
					T4m = FMA(KP414213562, T4b, T4c);
					T2V = T1n + T1q;
					T1r = T1n - T1q;
					T4e = T1s - T1t;
					T1u = T1s + T1t;
					T1T = R1[WS(rs, 17)];
					T22 = R1[WS(rs, 5)];
					{
					     E T4g, T4n, T2W, T1y;
					     T4g = FMA(KP414213562, T4f, T4e);
					     T4n = FNMS(KP414213562, T4e, T4f);
					     T2W = T1u + T1x;
					     T1y = T1u - T1x;
					     T4w = T1S - T1T;
					     T1U = T1S + T1T;
					     T5E = T4g - T4d;
					     T4h = T4d + T4g;
					     T5B = T4m - T4n;
					     T4o = T4m + T4n;
					     T3j = T2W - T2V;
					     T2X = T2V + T2W;
					     T1I = T1y - T1r;
					     T1z = T1r + T1y;
					     T23 = R1[WS(rs, 21)];
					}
				   }
				   T1V = R1[WS(rs, 9)];
				   T1W = R1[WS(rs, 25)];
				   T1Z = R1[WS(rs, 29)];
				   T4A = T23 - T22;
				   T24 = T22 + T23;
				   T4x = T1W - T1V;
				   T1X = T1V + T1W;
				   T20 = R1[WS(rs, 13)];
			      }
			 }
		    }
	       }
	       {
		    E T4C, T5L, T4J, T5I, T26, T2f, T3q, T3h, T3w, T3s, T3o, T3r, T3t;
		    {
			 E T2R, T37, T2Y, T3a, T39, T3m, T3b, T35, Tv, T10, T34, T3c, T3x, T3y;
			 {
			      E T4y, T4H, T32, T1Y, T4z, T21;
			      T2R = Tf - Tu;
			      Tv = Tf + Tu;
			      T4y = FMA(KP414213562, T4x, T4w);
			      T4H = FNMS(KP414213562, T4w, T4x);
			      T32 = T1U + T1X;
			      T1Y = T1U - T1X;
			      T4z = T1Z - T20;
			      T21 = T1Z + T20;
			      T10 = TK + TZ;
			      T37 = TZ - TK;
			      T2Y = T2U - T2X;
			      T3a = T2U + T2X;
			      {
				   E T4B, T4I, T33, T25;
				   T4B = FNMS(KP414213562, T4A, T4z);
				   T4I = FMA(KP414213562, T4z, T4A);
				   T33 = T21 + T24;
				   T25 = T21 - T24;
				   T39 = Tv + T10;
				   T4C = T4y + T4B;
				   T5L = T4B - T4y;
				   T4J = T4H + T4I;
				   T5I = T4I - T4H;
				   T34 = T32 + T33;
				   T3m = T33 - T32;
				   T26 = T1Y + T25;
				   T2f = T25 - T1Y;
			      }
			 }
			 Cr[WS(csr, 16)] = Tv - T10;
			 T3b = T31 + T34;
			 T35 = T31 - T34;
			 Ci[WS(csi, 16)] = T3b - T3a;
			 T3c = T3a + T3b;
			 {
			      E T3k, T3u, T3v, T3n, T36, T38, T3g;
			      T3g = T3e + T3f;
			      T3q = T3f - T3e;
			      Cr[0] = T39 + T3c;
			      Cr[WS(csr, 32)] = T39 - T3c;
			      T36 = T2Y + T35;
			      T38 = T35 - T2Y;
			      T3x = FNMS(KP707106781, T3g, T3d);
			      T3h = FMA(KP707106781, T3g, T3d);
			      Ci[WS(csi, 8)] = FMA(KP707106781, T38, T37);
			      Ci[WS(csi, 24)] = FMS(KP707106781, T38, T37);
			      Cr[WS(csr, 8)] = FMA(KP707106781, T36, T2R);
			      Cr[WS(csr, 24)] = FNMS(KP707106781, T36, T2R);
			      T3k = FMA(KP414213562, T3j, T3i);
			      T3u = FNMS(KP414213562, T3i, T3j);
			      T3v = FMA(KP414213562, T3l, T3m);
			      T3n = FNMS(KP414213562, T3m, T3l);
			      T3y = T3v - T3u;
			      T3w = T3u + T3v;
			      T3s = T3n - T3k;
			      T3o = T3k + T3n;
			 }
			 Cr[WS(csr, 12)] = FMA(KP923879532, T3y, T3x);
			 Cr[WS(csr, 20)] = FNMS(KP923879532, T3y, T3x);
		    }
		    Cr[WS(csr, 4)] = FMA(KP923879532, T3o, T3h);
		    Cr[WS(csr, 28)] = FNMS(KP923879532, T3o, T3h);
		    T3r = FNMS(KP707106781, T3q, T3p);
		    T3t = FMA(KP707106781, T3q, T3p);
		    {
			 E T27, T2g, T2v, T1d, T2r, T2p, T2s, T1K, T6l, T6m;
			 {
			      E T15, T2o, T2P, T2z, T2l, T1c, T1A, T1J, T2D, T2L, T2J, T2M, T2C, T2E, T2N;
			      E T2F;
			      {
				   E T2H, T2I, T2x, T2y, T2A, T2B;
				   T15 = FMA(KP707106781, T14, T11);
				   T2x = FNMS(KP707106781, T14, T11);
				   T2y = T2n - T2m;
				   T2o = T2m + T2n;
				   Ci[WS(csi, 4)] = FMA(KP923879532, T3w, T3t);
				   Ci[WS(csi, 28)] = FMS(KP923879532, T3w, T3t);
				   Ci[WS(csi, 20)] = FMA(KP923879532, T3s, T3r);
				   Ci[WS(csi, 12)] = FMS(KP923879532, T3s, T3r);
				   T2P = FNMS(KP923879532, T2y, T2x);
				   T2z = FMA(KP923879532, T2y, T2x);
				   T2l = FMA(KP707106781, T2k, T2j);
				   T2H = FNMS(KP707106781, T2k, T2j);
				   T2I = T1b - T18;
				   T1c = T18 + T1b;
				   T1A = FMA(KP707106781, T1z, T1k);
				   T2A = FNMS(KP707106781, T1z, T1k);
				   T2B = FNMS(KP707106781, T1I, T1H);
				   T1J = FMA(KP707106781, T1I, T1H);
				   T27 = FMA(KP707106781, T26, T1R);
				   T2D = FNMS(KP707106781, T26, T1R);
				   T2L = FNMS(KP923879532, T2I, T2H);
				   T2J = FMA(KP923879532, T2I, T2H);
				   T2M = FMA(KP668178637, T2A, T2B);
				   T2C = FNMS(KP668178637, T2B, T2A);
				   T2E = FNMS(KP707106781, T2f, T2e);
				   T2g = FMA(KP707106781, T2f, T2e);
			      }
			      T2N = FNMS(KP668178637, T2D, T2E);
			      T2F = FMA(KP668178637, T2E, T2D);
			      T2v = FNMS(KP923879532, T1c, T15);
			      T1d = FMA(KP923879532, T1c, T15);
			      {
				   E T2Q, T2O, T2K, T2G;
				   T2Q = T2M - T2N;
				   T2O = T2M + T2N;
				   T2K = T2F - T2C;
				   T2G = T2C + T2F;
				   Cr[WS(csr, 10)] = FMA(KP831469612, T2Q, T2P);
				   Cr[WS(csr, 22)] = FNMS(KP831469612, T2Q, T2P);
				   Ci[WS(csi, 26)] = FNMS(KP831469612, T2O, T2L);
				   Ci[WS(csi, 6)] = -(FMA(KP831469612, T2O, T2L));
				   Ci[WS(csi, 22)] = FMS(KP831469612, T2K, T2J);
				   Ci[WS(csi, 10)] = FMA(KP831469612, T2K, T2J);
				   Cr[WS(csr, 6)] = FMA(KP831469612, T2G, T2z);
				   Cr[WS(csr, 26)] = FNMS(KP831469612, T2G, T2z);
			      }
			      T2r = FMA(KP923879532, T2o, T2l);
			      T2p = FNMS(KP923879532, T2o, T2l);
			      T2s = FNMS(KP198912367, T1A, T1J);
			      T1K = FMA(KP198912367, T1J, T1A);
			 }
			 {
			      E T63, T5r, T5R, T6d, T5J, T5M, T6e, T5y, T6j, T6b, T66, T67, T64, T5U, T5Z;
			      E T5G;
			      {
				   E T5S, T5u, T5x, T5T, T2t, T2h;
				   T63 = FMA(KP923879532, T5q, T5p);
				   T5r = FNMS(KP923879532, T5q, T5p);
				   T5R = FNMS(KP923879532, T5Q, T5P);
				   T6d = FMA(KP923879532, T5Q, T5P);
				   T2t = FMA(KP198912367, T27, T2g);
				   T2h = FNMS(KP198912367, T2g, T27);
				   T5S = FNMS(KP668178637, T5s, T5t);
				   T5u = FMA(KP668178637, T5t, T5s);
				   {
					E T2w, T2u, T2q, T2i;
					T2w = T2t - T2s;
					T2u = T2s + T2t;
					T2q = T2h - T1K;
					T2i = T1K + T2h;
					Cr[WS(csr, 14)] = FMA(KP980785280, T2w, T2v);
					Cr[WS(csr, 18)] = FNMS(KP980785280, T2w, T2v);
					Ci[WS(csi, 30)] = FMS(KP980785280, T2u, T2r);
					Ci[WS(csi, 2)] = FMA(KP980785280, T2u, T2r);
					Ci[WS(csi, 18)] = FMA(KP980785280, T2q, T2p);
					Ci[WS(csi, 14)] = FMS(KP980785280, T2q, T2p);
					Cr[WS(csr, 2)] = FMA(KP980785280, T2i, T1d);
					Cr[WS(csr, 30)] = FNMS(KP980785280, T2i, T1d);
					T5x = FNMS(KP668178637, T5w, T5v);
					T5T = FMA(KP668178637, T5v, T5w);
				   }
				   {
					E T69, T6a, T5C, T5F;
					T5J = FNMS(KP923879532, T5I, T5H);
					T69 = FMA(KP923879532, T5I, T5H);
					T6a = FNMS(KP923879532, T5L, T5K);
					T5M = FMA(KP923879532, T5L, T5K);
					T6e = T5x + T5u;
					T5y = T5u - T5x;
					T6j = FNMS(KP303346683, T69, T6a);
					T6b = FMA(KP303346683, T6a, T69);
					T66 = FMA(KP923879532, T5B, T5A);
					T5C = FNMS(KP923879532, T5B, T5A);
					T5F = FNMS(KP923879532, T5E, T5D);
					T67 = FMA(KP923879532, T5E, T5D);
					T64 = T5T + T5S;
					T5U = T5S - T5T;
					T5Z = FMA(KP534511135, T5C, T5F);
					T5G = FNMS(KP534511135, T5F, T5C);
				   }
			      }
			      {
				   E T61, T6i, T68, T62;
				   {
					E T5z, T5Y, T5N, T5X, T5V, T60, T5W, T5O;
					T61 = FNMS(KP831469612, T5y, T5r);
					T5z = FMA(KP831469612, T5y, T5r);
					T6i = FNMS(KP303346683, T66, T67);
					T68 = FMA(KP303346683, T67, T66);
					T5Y = FMA(KP534511135, T5J, T5M);
					T5N = FNMS(KP534511135, T5M, T5J);
					T5X = FNMS(KP831469612, T5U, T5R);
					T5V = FMA(KP831469612, T5U, T5R);
					T60 = T5Y - T5Z;
					T62 = T5Z + T5Y;
					T5W = T5N - T5G;
					T5O = T5G + T5N;
					Ci[WS(csi, 27)] = FMA(KP881921264, T60, T5X);
					Ci[WS(csi, 5)] = FMS(KP881921264, T60, T5X);
					Cr[WS(csr, 5)] = FMA(KP881921264, T5O, T5z);
					Cr[WS(csr, 27)] = FNMS(KP881921264, T5O, T5z);
					Ci[WS(csi, 21)] = FMS(KP881921264, T5W, T5V);
					Ci[WS(csi, 11)] = FMA(KP881921264, T5W, T5V);
				   }
				   {
					E T6g, T6f, T6h, T6k, T65, T6c;
					T6l = FNMS(KP831469612, T64, T63);
					T65 = FMA(KP831469612, T64, T63);
					T6c = T68 + T6b;
					T6g = T6b - T68;
					T6f = FNMS(KP831469612, T6e, T6d);
					T6h = FMA(KP831469612, T6e, T6d);
					Cr[WS(csr, 11)] = FMA(KP881921264, T62, T61);
					Cr[WS(csr, 21)] = FNMS(KP881921264, T62, T61);
					Cr[WS(csr, 3)] = FMA(KP956940335, T6c, T65);
					Cr[WS(csr, 29)] = FNMS(KP956940335, T6c, T65);
					T6k = T6i - T6j;
					T6m = T6i + T6j;
					Ci[WS(csi, 29)] = FMS(KP956940335, T6k, T6h);
					Ci[WS(csi, 3)] = FMA(KP956940335, T6k, T6h);
					Ci[WS(csi, 19)] = FMA(KP956940335, T6g, T6f);
					Ci[WS(csi, 13)] = FMS(KP956940335, T6g, T6f);
				   }
			      }
			 }
			 {
			      E T55, T3L, T4T, T5f, T4D, T4K, T5g, T44, T5l, T5d, T58, T59, T56, T4W, T51;
			      E T4q;
			      {
				   E T4U, T3U, T43, T4V;
				   T55 = FNMS(KP923879532, T3K, T3D);
				   T3L = FMA(KP923879532, T3K, T3D);
				   T4T = FMA(KP923879532, T4S, T4P);
				   T5f = FNMS(KP923879532, T4S, T4P);
				   Cr[WS(csr, 13)] = FNMS(KP956940335, T6m, T6l);
				   Cr[WS(csr, 19)] = FMA(KP956940335, T6m, T6l);
				   T4U = FMA(KP198912367, T3Q, T3T);
				   T3U = FNMS(KP198912367, T3T, T3Q);
				   T43 = FMA(KP198912367, T42, T3Z);
				   T4V = FNMS(KP198912367, T3Z, T42);
				   {
					E T5b, T5c, T4i, T4p;
					T4D = FMA(KP923879532, T4C, T4v);
					T5b = FNMS(KP923879532, T4C, T4v);
					T5c = FNMS(KP923879532, T4J, T4G);
					T4K = FMA(KP923879532, T4J, T4G);
					T5g = T43 - T3U;
					T44 = T3U + T43;
					T5l = FNMS(KP820678790, T5b, T5c);
					T5d = FMA(KP820678790, T5c, T5b);
					T58 = FNMS(KP923879532, T4h, T4a);
					T4i = FMA(KP923879532, T4h, T4a);
					T4p = FMA(KP923879532, T4o, T4l);
					T59 = FNMS(KP923879532, T4o, T4l);
					T56 = T4U - T4V;
					T4W = T4U + T4V;
					T51 = FMA(KP098491403, T4i, T4p);
					T4q = FNMS(KP098491403, T4p, T4i);
				   }
			      }
			      {
				   E T53, T5k, T5a, T54;
				   {
					E T45, T50, T4L, T4Z, T4X, T52, T4Y, T4M;
					T53 = FNMS(KP980785280, T44, T3L);
					T45 = FMA(KP980785280, T44, T3L);
					T5k = FNMS(KP820678790, T58, T59);
					T5a = FMA(KP820678790, T59, T58);
					T50 = FMA(KP098491403, T4D, T4K);
					T4L = FNMS(KP098491403, T4K, T4D);
					T4Z = FMA(KP980785280, T4W, T4T);
					T4X = FNMS(KP980785280, T4W, T4T);
					T52 = T50 - T51;
					T54 = T51 + T50;
					T4Y = T4L - T4q;
					T4M = T4q + T4L;
					Ci[WS(csi, 31)] = FMA(KP995184726, T52, T4Z);
					Ci[WS(csi, 1)] = FMS(KP995184726, T52, T4Z);
					Cr[WS(csr, 1)] = FMA(KP995184726, T4M, T45);
					Cr[WS(csr, 31)] = FNMS(KP995184726, T4M, T45);
					Ci[WS(csi, 17)] = FMS(KP995184726, T4Y, T4X);
					Ci[WS(csi, 15)] = FMA(KP995184726, T4Y, T4X);
				   }
				   {
					E T5i, T5h, T5j, T5m, T57, T5e;
					T5n = FNMS(KP980785280, T56, T55);
					T57 = FMA(KP980785280, T56, T55);
					T5e = T5a + T5d;
					T5i = T5d - T5a;
					T5h = FNMS(KP980785280, T5g, T5f);
					T5j = FMA(KP980785280, T5g, T5f);
					Cr[WS(csr, 15)] = FMA(KP995184726, T54, T53);
					Cr[WS(csr, 17)] = FNMS(KP995184726, T54, T53);
					Cr[WS(csr, 7)] = FMA(KP773010453, T5e, T57);
					Cr[WS(csr, 25)] = FNMS(KP773010453, T5e, T57);
					T5m = T5k - T5l;
					T5o = T5k + T5l;
					Ci[WS(csi, 25)] = FMS(KP773010453, T5m, T5j);
					Ci[WS(csi, 7)] = FMA(KP773010453, T5m, T5j);
					Ci[WS(csi, 23)] = FMA(KP773010453, T5i, T5h);
					Ci[WS(csi, 9)] = FMS(KP773010453, T5i, T5h);
				   }
			      }
			 }
		    }
	       }
	  }
	  Cr[WS(csr, 9)] = FNMS(KP773010453, T5o, T5n);
	  Cr[WS(csr, 23)] = FMA(KP773010453, T5o, T5n);
     }
}
/* Included from initramfs et al code */
STATIC int INIT gunzip(unsigned char *buf, long len,
		       long (*fill)(void*, unsigned long),
		       long (*flush)(void*, unsigned long),
		       unsigned char *out_buf,
		       long *pos,
		       void(*error)(char *x)) {
	u8 *zbuf;
	struct z_stream_s *strm;
	int rc;
	size_t out_len;

	rc = -1;
	if (flush) {
		out_len = 0x8000; /* 32 K */
		out_buf = malloc(out_len);
	} else {
		out_len = 0x7fffffff; /* no limit */
	}
	if (!out_buf) {
		error("Out of memory while allocating output buffer");
		goto gunzip_nomem1;
	}

	if (buf)
		zbuf = buf;
	else {
		zbuf = malloc(GZIP_IOBUF_SIZE);
		len = 0;
	}
	if (!zbuf) {
		error("Out of memory while allocating input buffer");
		goto gunzip_nomem2;
	}

	strm = malloc(sizeof(*strm));
	if (strm == NULL) {
		error("Out of memory while allocating z_stream");
		goto gunzip_nomem3;
	}

	strm->workspace = malloc(flush ? zlib_inflate_workspacesize() :
				 sizeof(struct inflate_state));
	if (strm->workspace == NULL) {
		error("Out of memory while allocating workspace");
		goto gunzip_nomem4;
	}

	if (!fill)
		fill = nofill;

	if (len == 0)
		len = fill(zbuf, GZIP_IOBUF_SIZE);

	/* verify the gzip header */
	if (len < 10 ||
	   zbuf[0] != 0x1f || zbuf[1] != 0x8b || zbuf[2] != 0x08) {
		if (pos)
			*pos = 0;
		error("Not a gzip file");
		goto gunzip_5;
	}

	/* skip over gzip header (1f,8b,08... 10 bytes total +
	 * possible asciz filename)
	 */
	strm->next_in = zbuf + 10;
	strm->avail_in = len - 10;
	/* skip over asciz filename */
	if (zbuf[3] & 0x8) {
		do {
			/*
			 * If the filename doesn't fit into the buffer,
			 * the file is very probably corrupt. Don't try
			 * to read more data.
			 */
			if (strm->avail_in == 0) {
				error("header error");
				goto gunzip_5;
			}
			--strm->avail_in;
		} while (*strm->next_in++);
	}

	strm->next_out = out_buf;
	strm->avail_out = out_len;

	rc = zlib_inflateInit2(strm, -MAX_WBITS);

	if (!flush) {
		WS(strm)->inflate_state.wsize = 0;
		WS(strm)->inflate_state.window = NULL;
	}

	while (rc == Z_OK) {
		if (strm->avail_in == 0) {
			/* TODO: handle case where both pos and fill are set */
			len = fill(zbuf, GZIP_IOBUF_SIZE);
			if (len < 0) {
				rc = -1;
				error("read error");
				break;
			}
			strm->next_in = zbuf;
			strm->avail_in = len;
		}
		rc = zlib_inflate(strm, 0);

		/* Write any data generated */
		if (flush && strm->next_out > out_buf) {
			long l = strm->next_out - out_buf;
			if (l != flush(out_buf, l)) {
				rc = -1;
				error("write error");
				break;
			}
			strm->next_out = out_buf;
			strm->avail_out = out_len;
		}

		/* after Z_FINISH, only Z_STREAM_END is "we unpacked it all" */
		if (rc == Z_STREAM_END) {
			rc = 0;
			break;
		} else if (rc != Z_OK) {
			error("uncompression error");
			rc = -1;
		}
	}

	zlib_inflateEnd(strm);
	if (pos)
		/* add + 8 to skip over trailer */
		*pos = strm->next_in - zbuf+8;

gunzip_5:
	free(strm->workspace);
gunzip_nomem4:
	free(strm);
gunzip_nomem3:
	if (!buf)
		free(zbuf);
gunzip_nomem2:
	if (flush)
		free(out_buf);
gunzip_nomem1:
	return rc; /* returns Z_OK (0) if successful */
}
static void r2cbIII_20(float *R0, float *R1, float *Cr, float *Ci, stride rs, stride csr, stride csi, INT v, INT ivs, INT ovs)
{
     DK(KP1_414213562, +1.414213562373095048801688724209698078569671875);
     DK(KP951056516, +0.951056516295153572116439333379382143405698634);
     DK(KP559016994, +0.559016994374947424102293417182819058860154590);
     DK(KP2_000000000, +2.000000000000000000000000000000000000000000000);
     DK(KP250000000, +0.250000000000000000000000000000000000000000000);
     DK(KP618033988, +0.618033988749894848204586834365638117720309180);
     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 TZ, TD, TW, Tw, Tt, TF, T1f, T1b;
	  {
	       E T1l, Tk, T9, Tj, Ta, TV, TI, Ts, TU, T1t, T11, Tx, T13, TC, T1a;
	       E T1i, Th, Tv, Ty;
	       {
		    E TQ, TS, Tr, Tm, Tn;
		    {
			 E T1, T5, T6, T2, T3, T7, TY;
			 T1 = Cr[WS(csr, 2)];
			 T5 = Cr[WS(csr, 9)];
			 T6 = Cr[WS(csr, 5)];
			 T2 = Cr[WS(csr, 6)];
			 T3 = Cr[WS(csr, 1)];
			 TQ = Ci[WS(csi, 2)];
			 T7 = T5 + T6;
			 TY = T5 - T6;
			 {
			      E T4, TX, T8, Tp, Tq;
			      T4 = T2 + T3;
			      TX = T2 - T3;
			      Tp = Ci[WS(csi, 5)];
			      Tq = Ci[WS(csi, 9)];
			      T1l = FNMS(KP618033988, TX, TY);
			      TZ = FMA(KP618033988, TY, TX);
			      Tk = T4 - T7;
			      T8 = T4 + T7;
			      TS = Tp + Tq;
			      Tr = Tp - Tq;
			      T9 = T1 + T8;
			      Tj = FNMS(KP250000000, T8, T1);
			      Tm = Ci[WS(csi, 6)];
			      Tn = Ci[WS(csi, 1)];
			 }
		    }
		    {
			 E Tb, T19, Tg, Tc;
			 Ta = Cr[WS(csr, 7)];
			 {
			      E Te, Tf, To, TR, TT;
			      Te = Cr[0];
			      Tf = Cr[WS(csr, 4)];
			      To = Tm + Tn;
			      TR = Tm - Tn;
			      Tb = Cr[WS(csr, 3)];
			      T19 = Te - Tf;
			      Tg = Te + Tf;
			      TT = TR - TS;
			      TV = TR + TS;
			      TI = FNMS(KP618033988, To, Tr);
			      Ts = FMA(KP618033988, Tr, To);
			      TU = FNMS(KP250000000, TT, TQ);
			      T1t = TT + TQ;
			      Tc = Cr[WS(csr, 8)];
			 }
			 T11 = Ci[WS(csi, 7)];
			 {
			      E TA, TB, Td, T18;
			      TA = Ci[WS(csi, 4)];
			      TB = Ci[0];
			      Td = Tb + Tc;
			      T18 = Tb - Tc;
			      Tx = Ci[WS(csi, 3)];
			      T13 = TB + TA;
			      TC = TA - TB;
			      T1a = FMA(KP618033988, T19, T18);
			      T1i = FNMS(KP618033988, T18, T19);
			      Th = Td + Tg;
			      Tv = Td - Tg;
			      Ty = Ci[WS(csi, 8)];
			 }
		    }
	       }
	       {
		    E Tu, T1w, T16, TL, T15, T1u;
		    {
			 E Ti, T12, Tz, T14;
			 Tu = FNMS(KP250000000, Th, Ta);
			 Ti = Ta + Th;
			 T12 = Tx - Ty;
			 Tz = Tx + Ty;
			 T1w = T9 - Ti;
			 T14 = T12 - T13;
			 T16 = T12 + T13;
			 TL = FNMS(KP618033988, Tz, TC);
			 TD = FMA(KP618033988, TC, Tz);
			 T15 = FNMS(KP250000000, T14, T11);
			 T1u = T14 + T11;
			 R0[0] = KP2_000000000 * (T9 + Ti);
		    }
		    {
			 E Tl, TJ, TN, T1q, T1m, TK, T1h, T17, TH, T1k, T1v;
			 Tl = FMA(KP559016994, Tk, Tj);
			 TH = FNMS(KP559016994, Tk, Tj);
			 T1k = FNMS(KP559016994, TV, TU);
			 TW = FMA(KP559016994, TV, TU);
			 R0[WS(rs, 5)] = KP2_000000000 * (T1u - T1t);
			 T1v = T1t + T1u;
			 TJ = FNMS(KP951056516, TI, TH);
			 TN = FMA(KP951056516, TI, TH);
			 T1q = FMA(KP951056516, T1l, T1k);
			 T1m = FNMS(KP951056516, T1l, T1k);
			 R1[WS(rs, 7)] = KP1_414213562 * (T1w + T1v);
			 R1[WS(rs, 2)] = KP1_414213562 * (T1v - T1w);
			 Tw = FMA(KP559016994, Tv, Tu);
			 TK = FNMS(KP559016994, Tv, Tu);
			 T1h = FNMS(KP559016994, T16, T15);
			 T17 = FMA(KP559016994, T16, T15);
			 {
			      E TM, TO, T1j, T1r;
			      TM = FMA(KP951056516, TL, TK);
			      TO = FNMS(KP951056516, TL, TK);
			      T1j = FMA(KP951056516, T1i, T1h);
			      T1r = FNMS(KP951056516, T1i, T1h);
			      Tt = FNMS(KP951056516, Ts, Tl);
			      TF = FMA(KP951056516, Ts, Tl);
			      {
				   E T1n, T1p, T1s, T1o;
				   T1n = TN - TO;
				   R0[WS(rs, 6)] = -(KP2_000000000 * (TN + TO));
				   T1p = TM - TJ;
				   R0[WS(rs, 4)] = KP2_000000000 * (TJ + TM);
				   T1s = T1q + T1r;
				   R0[WS(rs, 9)] = KP2_000000000 * (T1r - T1q);
				   T1o = T1m + T1j;
				   R0[WS(rs, 1)] = KP2_000000000 * (T1j - T1m);
				   R1[WS(rs, 6)] = KP1_414213562 * (T1p + T1s);
				   R1[WS(rs, 1)] = KP1_414213562 * (T1p - T1s);
				   R1[WS(rs, 3)] = KP1_414213562 * (T1n + T1o);
				   R1[WS(rs, 8)] = KP1_414213562 * (T1n - T1o);
				   T1f = FMA(KP951056516, T1a, T17);
				   T1b = FNMS(KP951056516, T1a, T17);
			      }
			 }
		    }
	       }
	  }
	  {
	       E TE, TG, T10, T1e;
	       TE = FMA(KP951056516, TD, Tw);
	       TG = FNMS(KP951056516, TD, Tw);
	       T10 = FMA(KP951056516, TZ, TW);
	       T1e = FNMS(KP951056516, TZ, TW);
	       {
		    E T1d, TP, T1g, T1c;
		    T1d = TF - TG;
		    R0[WS(rs, 2)] = -(KP2_000000000 * (TF + TG));
		    TP = Tt - TE;
		    R0[WS(rs, 8)] = KP2_000000000 * (Tt + TE);
		    T1g = T1e + T1f;
		    R0[WS(rs, 7)] = KP2_000000000 * (T1e - T1f);
		    T1c = T10 + T1b;
		    R0[WS(rs, 3)] = KP2_000000000 * (T10 - T1b);
		    R1[WS(rs, 9)] = -(KP1_414213562 * (T1d + T1g));
		    R1[WS(rs, 4)] = KP1_414213562 * (T1d - T1g);
		    R1[WS(rs, 5)] = -(KP1_414213562 * (TP + T1c));
		    R1[0] = KP1_414213562 * (TP - T1c);
	       }
	  }
     }
}
Exemple #4
0
static void hc2cb_4(R *Rp, R *Ip, R *Rm, R *Im, const R *W, stride rs, INT mb, INT me, INT ms)
{
     {
	  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(rs)) {
	       E Th, Ta, T7, Ti, T9;
	       {
		    E Tq, Td, T3, Tg, Tu, Tm, T6, Tp;
		    {
			 E Tk, T4, Tl, T5;
			 {
			      E Tb, Tc, T1, T2, Te, Tf;
			      Tb = Ip[0];
			      Tc = Im[WS(rs, 1)];
			      T1 = Rp[0];
			      T2 = Rm[WS(rs, 1)];
			      Te = Ip[WS(rs, 1)];
			      Tq = Tb + Tc;
			      Td = Tb - Tc;
			      Tf = Im[0];
			      Tk = T1 - T2;
			      T3 = T1 + T2;
			      T4 = Rp[WS(rs, 1)];
			      Tg = Te - Tf;
			      Tl = Te + Tf;
			      T5 = Rm[0];
			 }
			 Tu = Tk + Tl;
			 Tm = Tk - Tl;
			 T6 = T4 + T5;
			 Tp = T4 - T5;
		    }
		    Rm[0] = Td + Tg;
		    {
			 E Tx, Tr, T8, Tn, Ts, To, Tj;
			 Tj = W[0];
			 Tx = Tq - Tp;
			 Tr = Tp + Tq;
			 Rp[0] = T3 + T6;
			 T8 = T3 - T6;
			 Tn = Tj * Tm;
			 Ts = Tj * Tr;
			 To = W[1];
			 {
			      E Tt, Tw, Ty, Tv;
			      Tt = W[4];
			      Tw = W[5];
			      Th = Td - Tg;
			      Im[0] = FMA(To, Tm, Ts);
			      Ip[0] = FNMS(To, Tr, Tn);
			      Ty = Tt * Tx;
			      Tv = Tt * Tu;
			      Ta = W[3];
			      T7 = W[2];
			      Im[WS(rs, 1)] = FMA(Tw, Tu, Ty);
			      Ip[WS(rs, 1)] = FNMS(Tw, Tx, Tv);
			      Ti = Ta * T8;
			      T9 = T7 * T8;
			 }
		    }
	       }
	       Rm[WS(rs, 1)] = FMA(T7, Th, Ti);
	       Rp[WS(rs, 1)] = FNMS(Ta, Th, T9);
	  }
     }
}
Exemple #5
0
static const R *hb_6(R *rio, R *iio, const R *W, stride ios, INT m, INT dist)
{
     DK(KP866025403, +0.866025403784438646763723170752936183471402627);
     DK(KP500000000, +0.500000000000000000000000000000000000000000000);
     INT i;
     for (i = m - 2; i > 0; i = i - 2, rio = rio + dist, iio = iio - dist, W = W + 10, MAKE_VOLATILE_STRIDE(ios)) {
	  E Tk, Tt, Tz, Tv, Tb, TA, Tm, Tl, Tu;
	  {
	       E Tp, TR, Tj, Tq, TN, TV, T3, TC, T7, T8, T4, T5;
	       {
		    E TJ, TK, Ti, Tf, TL, Td, Te, T1, T2, TM;
		    {
			 E Tn, To, Tg, Th;
			 Tn = iio[0];
			 To = rio[WS(ios, 3)];
			 Tg = iio[-WS(ios, 2)];
			 Th = rio[WS(ios, 5)];
			 Td = iio[-WS(ios, 1)];
			 TJ = Tn + To;
			 Tp = Tn - To;
			 Te = rio[WS(ios, 4)];
			 TK = Tg + Th;
			 Ti = Tg - Th;
		    }
		    Tf = Td - Te;
		    TL = Te + Td;
		    T1 = rio[0];
		    T2 = iio[-WS(ios, 3)];
		    TR = TL + TK;
		    TM = TK - TL;
		    Tj = Tf - Ti;
		    Tq = Tf + Ti;
		    TN = TJ + TM;
		    TV = FNMS(KP500000000, TM, TJ);
		    T3 = T1 + T2;
		    TC = T1 - T2;
		    T7 = iio[-WS(ios, 4)];
		    T8 = rio[WS(ios, 1)];
		    T4 = rio[WS(ios, 2)];
		    T5 = iio[-WS(ios, 5)];
	       }
	       {
		    E TZ, T12, Ts, TG, T10, TX, Tc, TS, T13, T11, TP, TU, T14;
		    iio[-WS(ios, 5)] = Tp + Tq;
		    {
			 E T9, TE, T6, TD;
			 T9 = T7 + T8;
			 TE = T7 - T8;
			 T6 = T4 + T5;
			 TD = T4 - T5;
			 TZ = W[8];
			 T12 = W[9];
			 {
			      E TW, TF, Ta, TQ;
			      TW = TD - TE;
			      TF = TD + TE;
			      Ts = T6 - T9;
			      Ta = T6 + T9;
			      TG = TC + TF;
			      TQ = FNMS(KP500000000, TF, TC);
			      T10 = FNMS(KP866025403, TW, TV);
			      TX = FMA(KP866025403, TW, TV);
			      rio[0] = T3 + Ta;
			      Tc = FNMS(KP500000000, Ta, T3);
			      TS = FNMS(KP866025403, TR, TQ);
			      T13 = FMA(KP866025403, TR, TQ);
			      T11 = TZ * T10;
			 }
		    }
		    T14 = TZ * T13;
		    iio[0] = FMA(T12, T13, T11);
		    TP = W[0];
		    TU = W[1];
		    rio[WS(ios, 5)] = FNMS(T12, T10, T14);
		    {
			 E TB, TI, TH, TY, TT;
			 TY = TP * TX;
			 TT = TP * TS;
			 TB = W[4];
			 iio[-WS(ios, 4)] = FMA(TU, TS, TY);
			 rio[WS(ios, 1)] = FNMS(TU, TX, TT);
			 TI = W[5];
			 TH = TB * TG;
			 {
			      E Ty, Tx, Tr, Tw, TO;
			      Tr = FNMS(KP500000000, Tq, Tp);
			      Tw = FNMS(KP866025403, Tj, Tc);
			      Tk = FMA(KP866025403, Tj, Tc);
			      TO = TI * TG;
			      rio[WS(ios, 3)] = FNMS(TI, TN, TH);
			      Tt = FMA(KP866025403, Ts, Tr);
			      Tz = FNMS(KP866025403, Ts, Tr);
			      iio[-WS(ios, 2)] = FMA(TB, TN, TO);
			      Tv = W[2];
			      Ty = W[3];
			      Tb = W[6];
			      Tx = Tv * Tw;
			      TA = Ty * Tw;
			      Tm = W[7];
			      Tl = Tb * Tk;
			      rio[WS(ios, 2)] = FNMS(Ty, Tz, Tx);
			 }
		    }
	       }
	  }
	  iio[-WS(ios, 3)] = FMA(Tv, Tz, TA);
	  Tu = Tm * Tk;
	  rio[WS(ios, 4)] = FNMS(Tm, Tt, Tl);
	  iio[-WS(ios, 1)] = FMA(Tb, Tt, Tu);
     }
     return W;
}
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;
		    {
			 E T1o, T1q, T1n, T1p;
			 T1o = T14 - T15;
			 T1q = T19 - T18;
			 T1n = W[10];
			 T1p = W[11];
			 T1r = FNMS(T1p, T1q, T1n * T1o);
			 T1x = FMA(T1p, T1o, T1n * T1q);
		    }
		    {
			 E T1t, T1v, T1s, T1u;
			 T1t = T1d - T1e;
			 T1v = T1i + T1h;
			 T1s = W[12];
			 T1u = W[13];
			 T1w = FMA(T1s, T1t, T1u * T1v);
			 T1y = FNMS(T1u, T1t, T1s * T1v);
		    }
		    Rp[WS(rs, 3)] = T1r - T1w;
		    Ip[WS(rs, 3)] = T1x + T1y;
		    Rm[WS(rs, 3)] = T1r + T1w;
		    Im[WS(rs, 3)] = T1y - T1x;
	       }
	       {
		    E TV, T11, T10, T12;
		    {
			 E TS, TU, TR, TT;
			 TS = T7 - Te;
			 TU = TO - TL;
			 TR = W[6];
			 TT = W[7];
			 TV = FNMS(TT, TU, TR * TS);
			 T11 = FMA(TT, TS, TR * TU);
		    }
		    {
			 E TX, TZ, TW, TY;
			 TX = Tl - Tw;
			 TZ = TG - TB;
			 TW = W[8];
			 TY = W[9];
			 T10 = FMA(TW, TX, TY * TZ);
			 T12 = FNMS(TY, TX, TW * TZ);
		    }
		    Rp[WS(rs, 2)] = TV - T10;
		    Ip[WS(rs, 2)] = T11 + T12;
		    Rm[WS(rs, 2)] = TV + T10;
		    Im[WS(rs, 2)] = T12 - T11;
	       }
	       {
		    E T1b, T1l, T1k, T1m;
		    {
			 E T16, T1a, T13, T17;
			 T16 = T14 + T15;
			 T1a = T18 + T19;
			 T13 = W[2];
			 T17 = W[3];
			 T1b = FNMS(T17, T1a, T13 * T16);
			 T1l = FMA(T17, T16, T13 * T1a);
		    }
		    {
			 E T1f, T1j, T1c, T1g;
			 T1f = T1d + T1e;
			 T1j = T1h - T1i;
			 T1c = W[4];
			 T1g = W[5];
			 T1k = FMA(T1c, T1f, T1g * T1j);
			 T1m = FNMS(T1g, T1f, T1c * T1j);
		    }
		    Rp[WS(rs, 1)] = T1b - T1k;
		    Ip[WS(rs, 1)] = T1l + T1m;
		    Rm[WS(rs, 1)] = T1b + T1k;
		    Im[WS(rs, 1)] = T1m - T1l;
	       }
	  }
     }
}
Exemple #7
0
static void q1_2(float *rio, float *iio, const float *W, stride rs, stride vs, INT mb, INT me, INT ms)
{
     INT m;
     for (m = mb, W = W + (mb * 2); m < me; m = m + 1, rio = rio + ms, iio = iio + ms, W = W + 2, MAKE_VOLATILE_STRIDE(rs), MAKE_VOLATILE_STRIDE(vs)) {
	  E T9, T6, T5;
	  {
	       E T1, T2, T7, T8, Tb, T4, Tc, Th, Ti, Te, Tj, Td, Tg;
	       T1 = rio[0];
	       T2 = rio[WS(rs, 1)];
	       T7 = iio[0];
	       T8 = iio[WS(rs, 1)];
	       Tb = rio[WS(vs, 1)];
	       T4 = T1 - T2;
	       Tc = rio[WS(vs, 1) + WS(rs, 1)];
	       T9 = T7 - T8;
	       Th = iio[WS(vs, 1)];
	       Ti = iio[WS(vs, 1) + WS(rs, 1)];
	       Te = Tb - Tc;
	       rio[0] = T1 + T2;
	       iio[0] = T7 + T8;
	       Tj = Th - Ti;
	       rio[WS(rs, 1)] = Tb + Tc;
	       iio[WS(rs, 1)] = Th + Ti;
	       Td = W[0];
	       Tg = W[1];
	       {
		    E T3, Tk, Tf, Ta;
		    T3 = W[0];
		    T6 = W[1];
		    Tk = Td * Tj;
		    Tf = Td * Te;
		    Ta = T3 * T9;
		    T5 = T3 * T4;
		    iio[WS(vs, 1) + WS(rs, 1)] = FNMS(Tg, Te, Tk);
		    rio[WS(vs, 1) + WS(rs, 1)] = FMA(Tg, Tj, Tf);
		    iio[WS(vs, 1)] = FNMS(T6, T4, Ta);
	       }
	  }
	  rio[WS(vs, 1)] = FMA(T6, T9, T5);
     }
}
Exemple #8
0
static void t2fv_16(R *ri, R *ii, const R *W, stride rs, INT mb, INT me, INT ms)
{
     DVK(KP923879532, +0.923879532511286756128183189396788286822416626);
     DVK(KP382683432, +0.382683432365089771728459984030398866761344562);
     DVK(KP707106781, +0.707106781186547524400844362104849039284835938);
     {
	  INT m;
	  R *x;
	  x = ri;
	  for (m = mb, W = W + (mb * ((TWVL / VL) * 30)); m < me; m = m + VL, x = x + (VL * ms), W = W + (TWVL * 30), MAKE_VOLATILE_STRIDE(16, rs)) {
	       V TJ, T10, TD, T11, T1b, T1c, Ty, TK, T16, T17, T18, Tb, TN, T13, T14;
	       V T15, Tm, TM, TG, TI, TH;
	       TG = LD(&(x[0]), ms, &(x[0]));
	       TH = LD(&(x[WS(rs, 8)]), ms, &(x[0]));
	       TI = BYTWJ(&(W[TWVL * 14]), TH);
	       TJ = VSUB(TG, TI);
	       T10 = VADD(TG, TI);
	       {
		    V TA, TC, Tz, TB;
		    Tz = LD(&(x[WS(rs, 4)]), ms, &(x[0]));
		    TA = BYTWJ(&(W[TWVL * 6]), Tz);
		    TB = LD(&(x[WS(rs, 12)]), ms, &(x[0]));
		    TC = BYTWJ(&(W[TWVL * 22]), TB);
		    TD = VSUB(TA, TC);
		    T11 = VADD(TA, TC);
	       }
	       {
		    V Tp, Tw, Tr, Tu, Ts, Tx;
		    {
			 V To, Tv, Tq, Tt;
			 To = LD(&(x[WS(rs, 14)]), ms, &(x[0]));
			 Tp = BYTWJ(&(W[TWVL * 26]), To);
			 Tv = LD(&(x[WS(rs, 10)]), ms, &(x[0]));
			 Tw = BYTWJ(&(W[TWVL * 18]), Tv);
			 Tq = LD(&(x[WS(rs, 6)]), ms, &(x[0]));
			 Tr = BYTWJ(&(W[TWVL * 10]), Tq);
			 Tt = LD(&(x[WS(rs, 2)]), ms, &(x[0]));
			 Tu = BYTWJ(&(W[TWVL * 2]), Tt);
		    }
		    T1b = VADD(Tp, Tr);
		    T1c = VADD(Tu, Tw);
		    Ts = VSUB(Tp, Tr);
		    Tx = VSUB(Tu, Tw);
		    Ty = VMUL(LDK(KP707106781), VSUB(Ts, Tx));
		    TK = VMUL(LDK(KP707106781), VADD(Tx, Ts));
	       }
	       {
		    V T2, T9, T4, T7, T5, Ta;
		    {
			 V T1, T8, T3, T6;
			 T1 = LD(&(x[WS(rs, 15)]), ms, &(x[WS(rs, 1)]));
			 T2 = BYTWJ(&(W[TWVL * 28]), T1);
			 T8 = LD(&(x[WS(rs, 11)]), ms, &(x[WS(rs, 1)]));
			 T9 = BYTWJ(&(W[TWVL * 20]), T8);
			 T3 = LD(&(x[WS(rs, 7)]), ms, &(x[WS(rs, 1)]));
			 T4 = BYTWJ(&(W[TWVL * 12]), T3);
			 T6 = LD(&(x[WS(rs, 3)]), ms, &(x[WS(rs, 1)]));
			 T7 = BYTWJ(&(W[TWVL * 4]), T6);
		    }
		    T16 = VADD(T2, T4);
		    T17 = VADD(T7, T9);
		    T18 = VSUB(T16, T17);
		    T5 = VSUB(T2, T4);
		    Ta = VSUB(T7, T9);
		    Tb = VFNMS(LDK(KP923879532), Ta, VMUL(LDK(KP382683432), T5));
		    TN = VFMA(LDK(KP923879532), T5, VMUL(LDK(KP382683432), Ta));
	       }
	       {
		    V Td, Tk, Tf, Ti, Tg, Tl;
		    {
			 V Tc, Tj, Te, Th;
			 Tc = LD(&(x[WS(rs, 1)]), ms, &(x[WS(rs, 1)]));
			 Td = BYTWJ(&(W[0]), Tc);
			 Tj = LD(&(x[WS(rs, 13)]), ms, &(x[WS(rs, 1)]));
			 Tk = BYTWJ(&(W[TWVL * 24]), Tj);
			 Te = LD(&(x[WS(rs, 9)]), ms, &(x[WS(rs, 1)]));
			 Tf = BYTWJ(&(W[TWVL * 16]), Te);
			 Th = LD(&(x[WS(rs, 5)]), ms, &(x[WS(rs, 1)]));
			 Ti = BYTWJ(&(W[TWVL * 8]), Th);
		    }
		    T13 = VADD(Td, Tf);
		    T14 = VADD(Ti, Tk);
		    T15 = VSUB(T13, T14);
		    Tg = VSUB(Td, Tf);
		    Tl = VSUB(Ti, Tk);
		    Tm = VFMA(LDK(KP382683432), Tg, VMUL(LDK(KP923879532), Tl));
		    TM = VFNMS(LDK(KP382683432), Tl, VMUL(LDK(KP923879532), Tg));
	       }
	       {
		    V T1a, T1g, T1f, T1h;
		    {
			 V T12, T19, T1d, T1e;
			 T12 = VSUB(T10, T11);
			 T19 = VMUL(LDK(KP707106781), VADD(T15, T18));
			 T1a = VADD(T12, T19);
			 T1g = VSUB(T12, T19);
			 T1d = VSUB(T1b, T1c);
			 T1e = VMUL(LDK(KP707106781), VSUB(T18, T15));
			 T1f = VBYI(VADD(T1d, T1e));
			 T1h = VBYI(VSUB(T1e, T1d));
		    }
		    ST(&(x[WS(rs, 14)]), VSUB(T1a, T1f), ms, &(x[0]));
		    ST(&(x[WS(rs, 6)]), VADD(T1g, T1h), ms, &(x[0]));
		    ST(&(x[WS(rs, 2)]), VADD(T1a, T1f), ms, &(x[0]));
		    ST(&(x[WS(rs, 10)]), VSUB(T1g, T1h), ms, &(x[0]));
	       }
	       {
		    V T1k, T1o, T1n, T1p;
		    {
			 V T1i, T1j, T1l, T1m;
			 T1i = VADD(T10, T11);
			 T1j = VADD(T1c, T1b);
			 T1k = VADD(T1i, T1j);
			 T1o = VSUB(T1i, T1j);
			 T1l = VADD(T13, T14);
			 T1m = VADD(T16, T17);
			 T1n = VADD(T1l, T1m);
			 T1p = VBYI(VSUB(T1m, T1l));
		    }
		    ST(&(x[WS(rs, 8)]), VSUB(T1k, T1n), ms, &(x[0]));
		    ST(&(x[WS(rs, 4)]), VADD(T1o, T1p), ms, &(x[0]));
		    ST(&(x[0]), VADD(T1k, T1n), ms, &(x[0]));
		    ST(&(x[WS(rs, 12)]), VSUB(T1o, T1p), ms, &(x[0]));
	       }
	       {
		    V TF, TQ, TP, TR;
		    {
			 V Tn, TE, TL, TO;
			 Tn = VSUB(Tb, Tm);
			 TE = VSUB(Ty, TD);
			 TF = VBYI(VSUB(Tn, TE));
			 TQ = VBYI(VADD(TE, Tn));
			 TL = VADD(TJ, TK);
			 TO = VADD(TM, TN);
			 TP = VSUB(TL, TO);
			 TR = VADD(TL, TO);
		    }
		    ST(&(x[WS(rs, 7)]), VADD(TF, TP), ms, &(x[WS(rs, 1)]));
		    ST(&(x[WS(rs, 15)]), VSUB(TR, TQ), ms, &(x[WS(rs, 1)]));
		    ST(&(x[WS(rs, 9)]), VSUB(TP, TF), ms, &(x[WS(rs, 1)]));
		    ST(&(x[WS(rs, 1)]), VADD(TQ, TR), ms, &(x[WS(rs, 1)]));
	       }
	       {
		    V TU, TY, TX, TZ;
		    {
			 V TS, TT, TV, TW;
			 TS = VSUB(TJ, TK);
			 TT = VADD(Tm, Tb);
			 TU = VADD(TS, TT);
			 TY = VSUB(TS, TT);
			 TV = VADD(TD, Ty);
			 TW = VSUB(TN, TM);
			 TX = VBYI(VADD(TV, TW));
			 TZ = VBYI(VSUB(TW, TV));
		    }
		    ST(&(x[WS(rs, 13)]), VSUB(TU, TX), ms, &(x[WS(rs, 1)]));
		    ST(&(x[WS(rs, 5)]), VADD(TY, TZ), ms, &(x[WS(rs, 1)]));
		    ST(&(x[WS(rs, 3)]), VADD(TU, TX), ms, &(x[WS(rs, 1)]));
		    ST(&(x[WS(rs, 11)]), VSUB(TY, TZ), ms, &(x[WS(rs, 1)]));
	       }
	  }
     }
     VLEAVE();
}
Exemple #9
0
static void t2fv_16(R *ri, R *ii, const R *W, stride rs, INT mb, INT me, INT ms)
{
     DVK(KP923879532, +0.923879532511286756128183189396788286822416626);
     DVK(KP414213562, +0.414213562373095048801688724209698078569671875);
     DVK(KP707106781, +0.707106781186547524400844362104849039284835938);
     {
	  INT m;
	  R *x;
	  x = ri;
	  for (m = mb, W = W + (mb * ((TWVL / VL) * 30)); m < me; m = m + VL, x = x + (VL * ms), W = W + (TWVL * 30), MAKE_VOLATILE_STRIDE(16, rs)) {
	       V TO, Ta, TJ, TP, T14, Tq, T1i, T10, T1b, T1l, T13, T1c, TR, Tl, T15;
	       V Tv;
	       {
		    V Tc, TW, T4, T19, T9, TD, TI, Tj, TZ, T1a, Te, Th, Tn, Tr, Tu;
		    V Tp;
		    {
			 V T1, T2, T5, T7;
			 T1 = LD(&(x[0]), ms, &(x[0]));
			 T2 = LD(&(x[WS(rs, 8)]), ms, &(x[0]));
			 T5 = LD(&(x[WS(rs, 4)]), ms, &(x[0]));
			 T7 = LD(&(x[WS(rs, 12)]), ms, &(x[0]));
			 {
			      V Tz, TG, TB, TE;
			      Tz = LD(&(x[WS(rs, 14)]), ms, &(x[0]));
			      TG = LD(&(x[WS(rs, 10)]), ms, &(x[0]));
			      TB = LD(&(x[WS(rs, 6)]), ms, &(x[0]));
			      TE = LD(&(x[WS(rs, 2)]), ms, &(x[0]));
			      {
				   V Ti, TY, TX, Td, Tg, Tm, Tt, To;
				   {
					V T3, T6, T8, TA, TH, TC, TF, Tb;
					Tb = LD(&(x[WS(rs, 1)]), ms, &(x[WS(rs, 1)]));
					T3 = BYTWJ(&(W[TWVL * 14]), T2);
					T6 = BYTWJ(&(W[TWVL * 6]), T5);
					T8 = BYTWJ(&(W[TWVL * 22]), T7);
					TA = BYTWJ(&(W[TWVL * 26]), Tz);
					TH = BYTWJ(&(W[TWVL * 18]), TG);
					TC = BYTWJ(&(W[TWVL * 10]), TB);
					TF = BYTWJ(&(W[TWVL * 2]), TE);
					Tc = BYTWJ(&(W[0]), Tb);
					TW = VSUB(T1, T3);
					T4 = VADD(T1, T3);
					T19 = VSUB(T6, T8);
					T9 = VADD(T6, T8);
					Ti = LD(&(x[WS(rs, 13)]), ms, &(x[WS(rs, 1)]));
					TD = VADD(TA, TC);
					TY = VSUB(TA, TC);
					TI = VADD(TF, TH);
					TX = VSUB(TF, TH);
				   }
				   Td = LD(&(x[WS(rs, 9)]), ms, &(x[WS(rs, 1)]));
				   Tg = LD(&(x[WS(rs, 5)]), ms, &(x[WS(rs, 1)]));
				   Tm = LD(&(x[WS(rs, 15)]), ms, &(x[WS(rs, 1)]));
				   Tj = BYTWJ(&(W[TWVL * 24]), Ti);
				   Tt = LD(&(x[WS(rs, 11)]), ms, &(x[WS(rs, 1)]));
				   To = LD(&(x[WS(rs, 7)]), ms, &(x[WS(rs, 1)]));
				   TZ = VADD(TX, TY);
				   T1a = VSUB(TY, TX);
				   Te = BYTWJ(&(W[TWVL * 16]), Td);
				   Th = BYTWJ(&(W[TWVL * 8]), Tg);
				   Tn = BYTWJ(&(W[TWVL * 28]), Tm);
				   Tr = LD(&(x[WS(rs, 3)]), ms, &(x[WS(rs, 1)]));
				   Tu = BYTWJ(&(W[TWVL * 20]), Tt);
				   Tp = BYTWJ(&(W[TWVL * 12]), To);
			      }
			 }
		    }
		    {
			 V Tf, T11, Tk, T12, Ts;
			 TO = VADD(T4, T9);
			 Ta = VSUB(T4, T9);
			 TJ = VSUB(TD, TI);
			 TP = VADD(TI, TD);
			 Tf = VADD(Tc, Te);
			 T11 = VSUB(Tc, Te);
			 Tk = VADD(Th, Tj);
			 T12 = VSUB(Th, Tj);
			 Ts = BYTWJ(&(W[TWVL * 4]), Tr);
			 T14 = VSUB(Tn, Tp);
			 Tq = VADD(Tn, Tp);
			 T1i = VFNMS(LDK(KP707106781), TZ, TW);
			 T10 = VFMA(LDK(KP707106781), TZ, TW);
			 T1b = VFNMS(LDK(KP707106781), T1a, T19);
			 T1l = VFMA(LDK(KP707106781), T1a, T19);
			 T13 = VFNMS(LDK(KP414213562), T12, T11);
			 T1c = VFMA(LDK(KP414213562), T11, T12);
			 TR = VADD(Tf, Tk);
			 Tl = VSUB(Tf, Tk);
			 T15 = VSUB(Tu, Ts);
			 Tv = VADD(Ts, Tu);
		    }
	       }
	       {
		    V T1d, T16, TS, Tw, TU, TQ;
		    T1d = VFMA(LDK(KP414213562), T14, T15);
		    T16 = VFNMS(LDK(KP414213562), T15, T14);
		    TS = VADD(Tq, Tv);
		    Tw = VSUB(Tq, Tv);
		    TU = VSUB(TO, TP);
		    TQ = VADD(TO, TP);
		    {
			 V T1e, T1j, T17, T1m;
			 T1e = VSUB(T1c, T1d);
			 T1j = VADD(T1c, T1d);
			 T17 = VADD(T13, T16);
			 T1m = VSUB(T16, T13);
			 {
			      V TV, TT, TK, Tx;
			      TV = VSUB(TS, TR);
			      TT = VADD(TR, TS);
			      TK = VSUB(Tw, Tl);
			      Tx = VADD(Tl, Tw);
			      {
				   V T1h, T1f, T1o, T1k;
				   T1h = VFMA(LDK(KP923879532), T1e, T1b);
				   T1f = VFNMS(LDK(KP923879532), T1e, T1b);
				   T1o = VFMA(LDK(KP923879532), T1j, T1i);
				   T1k = VFNMS(LDK(KP923879532), T1j, T1i);
				   {
					V T1g, T18, T1p, T1n;
					T1g = VFMA(LDK(KP923879532), T17, T10);
					T18 = VFNMS(LDK(KP923879532), T17, T10);
					T1p = VFMA(LDK(KP923879532), T1m, T1l);
					T1n = VFNMS(LDK(KP923879532), T1m, T1l);
					ST(&(x[WS(rs, 12)]), VFNMSI(TV, TU), ms, &(x[0]));
					ST(&(x[WS(rs, 4)]), VFMAI(TV, TU), ms, &(x[0]));
					ST(&(x[0]), VADD(TQ, TT), ms, &(x[0]));
					ST(&(x[WS(rs, 8)]), VSUB(TQ, TT), ms, &(x[0]));
					{
					     V TN, TL, TM, Ty;
					     TN = VFMA(LDK(KP707106781), TK, TJ);
					     TL = VFNMS(LDK(KP707106781), TK, TJ);
					     TM = VFMA(LDK(KP707106781), Tx, Ta);
					     Ty = VFNMS(LDK(KP707106781), Tx, Ta);
					     ST(&(x[WS(rs, 1)]), VFNMSI(T1h, T1g), ms, &(x[WS(rs, 1)]));
					     ST(&(x[WS(rs, 15)]), VFMAI(T1h, T1g), ms, &(x[WS(rs, 1)]));
					     ST(&(x[WS(rs, 7)]), VFMAI(T1f, T18), ms, &(x[WS(rs, 1)]));
					     ST(&(x[WS(rs, 9)]), VFNMSI(T1f, T18), ms, &(x[WS(rs, 1)]));
					     ST(&(x[WS(rs, 3)]), VFMAI(T1p, T1o), ms, &(x[WS(rs, 1)]));
					     ST(&(x[WS(rs, 13)]), VFNMSI(T1p, T1o), ms, &(x[WS(rs, 1)]));
					     ST(&(x[WS(rs, 11)]), VFMAI(T1n, T1k), ms, &(x[WS(rs, 1)]));
					     ST(&(x[WS(rs, 5)]), VFNMSI(T1n, T1k), ms, &(x[WS(rs, 1)]));
					     ST(&(x[WS(rs, 14)]), VFNMSI(TN, TM), ms, &(x[0]));
					     ST(&(x[WS(rs, 2)]), VFMAI(TN, TM), ms, &(x[0]));
					     ST(&(x[WS(rs, 10)]), VFMAI(TL, Ty), ms, &(x[0]));
					     ST(&(x[WS(rs, 6)]), VFNMSI(TL, Ty), ms, &(x[0]));
					}
				   }
			      }
			 }
		    }
	       }
	  }
     }
     VLEAVE();
}
static void r2cb_13(float *R0, float *R1, float *Cr, float *Ci, stride rs, stride csr, stride csi, INT v, INT ivs, INT ovs)
{
     DK(KP1_007074065, +1.007074065727533254493747707736933954186697125);
     DK(KP227708958, +0.227708958111581597949308691735310621069285120);
     DK(KP531932498, +0.531932498429674575175042127684371897596660533);
     DK(KP774781170, +0.774781170935234584261351932853525703557550433);
     DK(KP265966249, +0.265966249214837287587521063842185948798330267);
     DK(KP516520780, +0.516520780623489722840901288569017135705033622);
     DK(KP151805972, +0.151805972074387731966205794490207080712856746);
     DK(KP503537032, +0.503537032863766627246873853868466977093348562);
     DK(KP166666666, +0.166666666666666666666666666666666666666666667);
     DK(KP600925212, +0.600925212577331548853203544578415991041882762);
     DK(KP500000000, +0.500000000000000000000000000000000000000000000);
     DK(KP256247671, +0.256247671582936600958684654061725059144125175);
     DK(KP156891391, +0.156891391051584611046832726756003269660212636);
     DK(KP348277202, +0.348277202304271810011321589858529485233929352);
     DK(KP1_150281458, +1.150281458948006242736771094910906776922003215);
     DK(KP300238635, +0.300238635966332641462884626667381504676006424);
     DK(KP011599105, +0.011599105605768290721655456654083252189827041);
     DK(KP1_732050807, +1.732050807568877293527446341505872366942805254);
     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 TG, TS, TR, T15, TJ, TT, T1, Tm, Tc, Td, Tg, Tj, Tk, Tn, To;
	  E Tp;
	  {
	       E Ts, Tv, Tw, TE, TC, TB, Tz, TD, TA, TF;
	       {
		    E Tt, Tu, Tx, Ty;
		    Ts = Ci[WS(csi, 1)];
		    Tt = Ci[WS(csi, 3)];
		    Tu = Ci[WS(csi, 4)];
		    Tv = Tt - Tu;
		    Tw = FMS(KP2_000000000, Ts, Tv);
		    TE = KP1_732050807 * (Tt + Tu);
		    TC = Ci[WS(csi, 5)];
		    Tx = Ci[WS(csi, 6)];
		    Ty = Ci[WS(csi, 2)];
		    TB = Tx + Ty;
		    Tz = KP1_732050807 * (Tx - Ty);
		    TD = FNMS(KP2_000000000, TC, TB);
	       }
	       TA = Tw + Tz;
	       TF = TD - TE;
	       TG = FMA(KP011599105, TA, KP300238635 * TF);
	       TS = FNMS(KP011599105, TF, KP300238635 * TA);
	       {
		    E TP, TQ, TH, TI;
		    TP = Ts + Tv;
		    TQ = TB + TC;
		    TR = FNMS(KP348277202, TQ, KP1_150281458 * TP);
		    T15 = FMA(KP348277202, TP, KP1_150281458 * TQ);
		    TH = Tw - Tz;
		    TI = TE + TD;
		    TJ = FMA(KP156891391, TH, KP256247671 * TI);
		    TT = FNMS(KP256247671, TH, KP156891391 * TI);
	       }
	  }
	  {
	       E Tb, Ti, Tf, T6, Th, Te;
	       T1 = Cr[0];
	       {
		    E T7, T8, T9, Ta;
		    T7 = Cr[WS(csr, 5)];
		    T8 = Cr[WS(csr, 2)];
		    T9 = Cr[WS(csr, 6)];
		    Ta = T8 + T9;
		    Tb = T7 + Ta;
		    Ti = FNMS(KP500000000, Ta, T7);
		    Tf = T8 - T9;
	       }
	       {
		    E T2, T3, T4, T5;
		    T2 = Cr[WS(csr, 1)];
		    T3 = Cr[WS(csr, 3)];
		    T4 = Cr[WS(csr, 4)];
		    T5 = T3 + T4;
		    T6 = T2 + T5;
		    Th = FNMS(KP500000000, T5, T2);
		    Te = T3 - T4;
	       }
	       Tm = KP600925212 * (T6 - Tb);
	       Tc = T6 + Tb;
	       Td = FNMS(KP166666666, Tc, T1);
	       Tg = Te + Tf;
	       Tj = Th + Ti;
	       Tk = FMA(KP503537032, Tg, KP151805972 * Tj);
	       Tn = Th - Ti;
	       To = Te - Tf;
	       Tp = FNMS(KP265966249, To, KP516520780 * Tn);
	  }
	  R0[0] = FMA(KP2_000000000, Tc, T1);
	  {
	       E TK, T1b, TV, T12, T16, T18, TO, T1a, Tr, T17, T11, T13;
	       {
		    E TU, T14, TM, TN;
		    TK = KP1_732050807 * (TG + TJ);
		    T1b = KP1_732050807 * (TS - TT);
		    TU = TS + TT;
		    TV = TR - TU;
		    T12 = FMA(KP2_000000000, TU, TR);
		    T14 = TG - TJ;
		    T16 = FMS(KP2_000000000, T14, T15);
		    T18 = T14 + T15;
		    TM = FMA(KP774781170, To, KP531932498 * Tn);
		    TN = FNMS(KP1_007074065, Tj, KP227708958 * Tg);
		    TO = TM - TN;
		    T1a = TM + TN;
		    {
			 E Tl, Tq, TZ, T10;
			 Tl = Td - Tk;
			 Tq = Tm - Tp;
			 Tr = Tl - Tq;
			 T17 = Tq + Tl;
			 TZ = FMA(KP2_000000000, Tk, Td);
			 T10 = FMA(KP2_000000000, Tp, Tm);
			 T11 = TZ - T10;
			 T13 = T10 + TZ;
		    }
	       }
	       R1[WS(rs, 2)] = T11 - T12;
	       R0[WS(rs, 6)] = T13 - T16;
	       R1[0] = T13 + T16;
	       R0[WS(rs, 4)] = T11 + T12;
	       {
		    E TL, TW, T19, T1c;
		    TL = Tr - TK;
		    TW = TO - TV;
		    R1[WS(rs, 3)] = TL - TW;
		    R0[WS(rs, 1)] = TL + TW;
		    T19 = T17 - T18;
		    T1c = T1a + T1b;
		    R1[WS(rs, 1)] = T19 - T1c;
		    R1[WS(rs, 4)] = T1c + T19;
	       }
	       {
		    E T1d, T1e, TX, TY;
		    T1d = T1a - T1b;
		    T1e = T17 + T18;
		    R0[WS(rs, 2)] = T1d + T1e;
		    R0[WS(rs, 5)] = T1e - T1d;
		    TX = Tr + TK;
		    TY = TO + TV;
		    R0[WS(rs, 3)] = TX - TY;
		    R1[WS(rs, 5)] = TX + TY;
	       }
	  }
     }
}
static void r2cb_13(float *R0, float *R1, float *Cr, float *Ci, stride rs, stride csr, stride csi, INT v, INT ivs, INT ovs)
{
     DK(KP968287244, +0.968287244361984016049539446938120421179794516);
     DK(KP875502302, +0.875502302409147941146295545768755143177842006);
     DK(KP1_150281458, +1.150281458948006242736771094910906776922003215);
     DK(KP1_040057143, +1.040057143777729238234261000998465604986476278);
     DK(KP1_200954543, +1.200954543865330565851538506669526018704025697);
     DK(KP769338817, +0.769338817572980603471413688209101117038278899);
     DK(KP600925212, +0.600925212577331548853203544578415991041882762);
     DK(KP1_033041561, +1.033041561246979445681802577138034271410067244);
     DK(KP1_007074065, +1.007074065727533254493747707736933954186697125);
     DK(KP503537032, +0.503537032863766627246873853868466977093348562);
     DK(KP581704778, +0.581704778510515730456870384989698884939833902);
     DK(KP859542535, +0.859542535098774820163672132761689612766401925);
     DK(KP166666666, +0.166666666666666666666666666666666666666666667);
     DK(KP2_000000000, +2.000000000000000000000000000000000000000000000);
     DK(KP301479260, +0.301479260047709873958013540496673347309208464);
     DK(KP226109445, +0.226109445035782405468510155372505010481906348);
     DK(KP686558370, +0.686558370781754340655719594850823015421401653);
     DK(KP514918778, +0.514918778086315755491789696138117261566051239);
     DK(KP957805992, +0.957805992594665126462521754605754580515587217);
     DK(KP522026385, +0.522026385161275033714027226654165028300441940);
     DK(KP853480001, +0.853480001859823990758994934970528322872359049);
     DK(KP038632954, +0.038632954644348171955506895830342264440241080);
     DK(KP612264650, +0.612264650376756543746494474777125408779395514);
     DK(KP302775637, +0.302775637731994646559610633735247973125648287);
     DK(KP866025403, +0.866025403784438646763723170752936183471402627);
     DK(KP500000000, +0.500000000000000000000000000000000000000000000);
     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 TW, T14, TS, TO, T18, T1e, TY, TX, TQ, Tq, TP, Tl, T1d, Tr;
	  {
	       E T1, TN, T16, TJ, TV, TG, TU, Tf, T2, T3, Tb, Ti, T4;
	       {
		    E Ts, TB, Tx, Ty, Tv, TE, Tt, Tu, Tz, TC;
		    Ts = Ci[WS(csi, 5)];
		    Tt = Ci[WS(csi, 2)];
		    Tu = Ci[WS(csi, 6)];
		    TB = Ci[WS(csi, 1)];
		    Tx = Ci[WS(csi, 3)];
		    Ty = Ci[WS(csi, 4)];
		    Tv = Tt + Tu;
		    TE = Tu - Tt;
		    T1 = Cr[0];
		    Tz = Tx + Ty;
		    TC = Tx - Ty;
		    {
			 E TL, Tw, T7, Ta;
			 TL = Ts + Tv;
			 Tw = FNMS(KP500000000, Tv, Ts);
			 T7 = Cr[WS(csr, 5)];
			 {
			      E TD, TM, TA, TH;
			      TD = FNMS(KP500000000, TC, TB);
			      TM = TB + TC;
			      TA = FMA(KP866025403, Tz, Tw);
			      TH = FNMS(KP866025403, Tz, Tw);
			      TN = FMA(KP302775637, TM, TL);
			      T16 = FNMS(KP302775637, TL, TM);
			      {
				   E TF, TI, T8, T9;
				   TF = FMA(KP866025403, TE, TD);
				   TI = FNMS(KP866025403, TE, TD);
				   T8 = Cr[WS(csr, 2)];
				   T9 = Cr[WS(csr, 6)];
				   TJ = FNMS(KP612264650, TI, TH);
				   TV = FMA(KP612264650, TH, TI);
				   TG = FNMS(KP038632954, TF, TA);
				   TU = FMA(KP038632954, TA, TF);
				   Tf = T8 - T9;
				   Ta = T8 + T9;
			      }
			 }
			 T2 = Cr[WS(csr, 1)];
			 T3 = Cr[WS(csr, 3)];
			 Tb = T7 + Ta;
			 Ti = FMS(KP500000000, Ta, T7);
			 T4 = Cr[WS(csr, 4)];
		    }
	       }
	       {
		    E T17, TK, T5, Te, Tk, Td;
		    TW = FMA(KP853480001, TV, TU);
		    T17 = FNMS(KP853480001, TV, TU);
		    TK = FNMS(KP853480001, TJ, TG);
		    T14 = FMA(KP853480001, TJ, TG);
		    T5 = T3 + T4;
		    Te = T3 - T4;
		    {
			 E Tn, Tg, Th, T6;
			 TS = FNMS(KP522026385, TK, TN);
			 TO = FMA(KP957805992, TN, TK);
			 Tn = Te - Tf;
			 Tg = Te + Tf;
			 Th = FNMS(KP500000000, T5, T2);
			 T6 = T2 + T5;
			 T18 = FNMS(KP522026385, T17, T16);
			 T1e = FMA(KP957805992, T16, T17);
			 {
			      E Tm, Tj, Tc, Tp, To;
			      Tm = Th + Ti;
			      Tj = Th - Ti;
			      Tc = T6 + Tb;
			      Tp = T6 - Tb;
			      To = FNMS(KP514918778, Tn, Tm);
			      TY = FMA(KP686558370, Tm, Tn);
			      TX = FNMS(KP226109445, Tg, Tj);
			      Tk = FMA(KP301479260, Tj, Tg);
			      R0[0] = FMA(KP2_000000000, Tc, T1);
			      Td = FNMS(KP166666666, Tc, T1);
			      TQ = FNMS(KP859542535, To, Tp);
			      Tq = FMA(KP581704778, Tp, To);
			 }
		    }
		    TP = FNMS(KP503537032, Tk, Td);
		    Tl = FMA(KP1_007074065, Tk, Td);
	       }
	  }
	  T1d = FNMS(KP1_033041561, Tq, Tl);
	  Tr = FMA(KP1_033041561, Tq, Tl);
	  {
	       E T13, TR, T19, TZ;
	       T13 = FNMS(KP600925212, TQ, TP);
	       TR = FMA(KP600925212, TQ, TP);
	       T19 = FMA(KP769338817, TY, TX);
	       TZ = FNMS(KP769338817, TY, TX);
	       R0[WS(rs, 4)] = FMA(KP1_200954543, T1e, T1d);
	       R1[WS(rs, 2)] = FNMS(KP1_200954543, T1e, T1d);
	       R0[WS(rs, 6)] = FMA(KP1_200954543, TO, Tr);
	       R1[0] = FNMS(KP1_200954543, TO, Tr);
	       {
		    E T1b, T15, T11, TT;
		    T1b = FNMS(KP1_040057143, T14, T13);
		    T15 = FMA(KP1_040057143, T14, T13);
		    T11 = FMA(KP1_150281458, TS, TR);
		    TT = FNMS(KP1_150281458, TS, TR);
		    {
			 E T1c, T1a, T12, T10;
			 T1c = FMA(KP875502302, T19, T18);
			 T1a = FNMS(KP875502302, T19, T18);
			 T12 = FMA(KP968287244, TZ, TW);
			 T10 = FNMS(KP968287244, TZ, TW);
			 R1[WS(rs, 5)] = FMA(KP1_150281458, T1c, T1b);
			 R0[WS(rs, 3)] = FNMS(KP1_150281458, T1c, T1b);
			 R1[WS(rs, 3)] = FMA(KP1_150281458, T1a, T15);
			 R0[WS(rs, 1)] = FNMS(KP1_150281458, T1a, T15);
			 R0[WS(rs, 5)] = FMA(KP1_040057143, T12, T11);
			 R0[WS(rs, 2)] = FNMS(KP1_040057143, T12, T11);
			 R1[WS(rs, 4)] = FMA(KP1_040057143, T10, TT);
			 R1[WS(rs, 1)] = FNMS(KP1_040057143, T10, TT);
		    }
	       }
	  }
     }
}
Exemple #12
0
static void t1bv_5(R *ri, R *ii, const R *W, stride rs, INT mb, INT me, INT ms)
{
    DVK(KP559016994, +0.559016994374947424102293417182819058860154590);
    DVK(KP250000000, +0.250000000000000000000000000000000000000000000);
    DVK(KP618033988, +0.618033988749894848204586834365638117720309180);
    DVK(KP951056516, +0.951056516295153572116439333379382143405698634);
    INT m;
    R *x;
    x = ii;
    for (m = mb, W = W + (mb * ((TWVL / VL) * 8)); m < me; m = m + VL, x = x + (VL * ms), W = W + (TWVL * 8), MAKE_VOLATILE_STRIDE(rs)) {
        V T1, T2, T9, T4, T7;
        T1 = LD(&(x[0]), ms, &(x[0]));
        T2 = LD(&(x[WS(rs, 1)]), ms, &(x[WS(rs, 1)]));
        T9 = LD(&(x[WS(rs, 3)]), ms, &(x[WS(rs, 1)]));
        T4 = LD(&(x[WS(rs, 4)]), ms, &(x[0]));
        T7 = LD(&(x[WS(rs, 2)]), ms, &(x[0]));
        {
            V T3, Ta, T5, T8;
            T3 = BYTW(&(W[0]), T2);
            Ta = BYTW(&(W[TWVL * 4]), T9);
            T5 = BYTW(&(W[TWVL * 6]), T4);
            T8 = BYTW(&(W[TWVL * 2]), T7);
            {
                V T6, Tg, Tb, Th;
                T6 = VADD(T3, T5);
                Tg = VSUB(T3, T5);
                Tb = VADD(T8, Ta);
                Th = VSUB(T8, Ta);
                {
                    V Te, Tc, Tk, Ti, Td, Tj, Tf;
                    Te = VSUB(T6, Tb);
                    Tc = VADD(T6, Tb);
                    Tk = VMUL(LDK(KP951056516), VFNMS(LDK(KP618033988), Tg, Th));
                    Ti = VMUL(LDK(KP951056516), VFMA(LDK(KP618033988), Th, Tg));
                    Td = VFNMS(LDK(KP250000000), Tc, T1);
                    ST(&(x[0]), VADD(T1, Tc), ms, &(x[0]));
                    Tj = VFNMS(LDK(KP559016994), Te, Td);
                    Tf = VFMA(LDK(KP559016994), Te, Td);
                    ST(&(x[WS(rs, 2)]), VFNMSI(Tk, Tj), ms, &(x[0]));
                    ST(&(x[WS(rs, 3)]), VFMAI(Tk, Tj), ms, &(x[WS(rs, 1)]));
                    ST(&(x[WS(rs, 4)]), VFNMSI(Ti, Tf), ms, &(x[0]));
                    ST(&(x[WS(rs, 1)]), VFMAI(Ti, Tf), ms, &(x[WS(rs, 1)]));
                }
            }
        }
    }
}
Exemple #13
0
static void t1bv_5(R *ri, R *ii, const R *W, stride rs, INT mb, INT me, INT ms)
{
    DVK(KP250000000, +0.250000000000000000000000000000000000000000000);
    DVK(KP559016994, +0.559016994374947424102293417182819058860154590);
    DVK(KP587785252, +0.587785252292473129168705954639072768597652438);
    DVK(KP951056516, +0.951056516295153572116439333379382143405698634);
    INT m;
    R *x;
    x = ii;
    for (m = mb, W = W + (mb * ((TWVL / VL) * 8)); m < me; m = m + VL, x = x + (VL * ms), W = W + (TWVL * 8), MAKE_VOLATILE_STRIDE(rs)) {
        V Tf, T5, Ta, Tc, Td, Tg;
        Tf = LD(&(x[0]), ms, &(x[0]));
        {
            V T2, T9, T4, T7;
            {
                V T1, T8, T3, T6;
                T1 = LD(&(x[WS(rs, 1)]), ms, &(x[WS(rs, 1)]));
                T2 = BYTW(&(W[0]), T1);
                T8 = LD(&(x[WS(rs, 3)]), ms, &(x[WS(rs, 1)]));
                T9 = BYTW(&(W[TWVL * 4]), T8);
                T3 = LD(&(x[WS(rs, 4)]), ms, &(x[0]));
                T4 = BYTW(&(W[TWVL * 6]), T3);
                T6 = LD(&(x[WS(rs, 2)]), ms, &(x[0]));
                T7 = BYTW(&(W[TWVL * 2]), T6);
            }
            T5 = VSUB(T2, T4);
            Ta = VSUB(T7, T9);
            Tc = VADD(T2, T4);
            Td = VADD(T7, T9);
            Tg = VADD(Tc, Td);
        }
        ST(&(x[0]), VADD(Tf, Tg), ms, &(x[0]));
        {
            V Tb, Tj, Ti, Tk, Te, Th;
            Tb = VBYI(VFMA(LDK(KP951056516), T5, VMUL(LDK(KP587785252), Ta)));
            Tj = VBYI(VFNMS(LDK(KP951056516), Ta, VMUL(LDK(KP587785252), T5)));
            Te = VMUL(LDK(KP559016994), VSUB(Tc, Td));
            Th = VFNMS(LDK(KP250000000), Tg, Tf);
            Ti = VADD(Te, Th);
            Tk = VSUB(Th, Te);
            ST(&(x[WS(rs, 1)]), VADD(Tb, Ti), ms, &(x[WS(rs, 1)]));
            ST(&(x[WS(rs, 3)]), VSUB(Tk, Tj), ms, &(x[WS(rs, 1)]));
            ST(&(x[WS(rs, 4)]), VSUB(Ti, Tb), ms, &(x[0]));
            ST(&(x[WS(rs, 2)]), VADD(Tj, Tk), ms, &(x[0]));
        }
    }
}
static void r2cf_64(R *R0, R *R1, R *Cr, R *Ci, stride rs, stride csr, stride csi, INT v, INT ivs, INT ovs)
{
     DK(KP773010453, +0.773010453362736960810906609758469800971041293);
     DK(KP634393284, +0.634393284163645498215171613225493370675687095);
     DK(KP098017140, +0.098017140329560601994195563888641845861136673);
     DK(KP995184726, +0.995184726672196886244836953109479921575474869);
     DK(KP290284677, +0.290284677254462367636192375817395274691476278);
     DK(KP956940335, +0.956940335732208864935797886980269969482849206);
     DK(KP471396736, +0.471396736825997648556387625905254377657460319);
     DK(KP881921264, +0.881921264348355029712756863660388349508442621);
     DK(KP195090322, +0.195090322016128267848284868477022240927691618);
     DK(KP980785280, +0.980785280403230449126182236134239036973933731);
     DK(KP555570233, +0.555570233019602224742830813948532874374937191);
     DK(KP831469612, +0.831469612302545237078788377617905756738560812);
     DK(KP382683432, +0.382683432365089771728459984030398866761344562);
     DK(KP923879532, +0.923879532511286756128183189396788286822416626);
     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 T4l, T5a, T15, T3n, T2T, T3Q, T7, Te, Tf, T4A, T4L, T1X, T3B, T23, T3y;
	  E T5I, T66, T4R, T52, T2j, T3F, T2H, T3I, T5P, T69, T1i, T3t, T1l, T3u, TZ;
	  E T63, T4v, T58, T1r, T3r, T1u, T3q, TK, T62, T4s, T57, Tm, Tt, Tu, T4o;
	  E T5b, T1c, T3R, T2Q, T3o, T1M, T3z, T5L, T67, T26, T3C, T4H, T4M, T2y, T3J;
	  E T5S, T6a, T2C, T3G, T4Y, T53;
	  {
	       E T3, T11, Td, T13, T6, T2S, Ta, T12, T14, T2R;
	       {
		    E T1, T2, Tb, Tc;
		    T1 = R0[0];
		    T2 = R0[WS(rs, 16)];
		    T3 = T1 + T2;
		    T11 = T1 - T2;
		    Tb = R0[WS(rs, 28)];
		    Tc = R0[WS(rs, 12)];
		    Td = Tb + Tc;
		    T13 = Tb - Tc;
	       }
	       {
		    E T4, T5, T8, T9;
		    T4 = R0[WS(rs, 8)];
		    T5 = R0[WS(rs, 24)];
		    T6 = T4 + T5;
		    T2S = T4 - T5;
		    T8 = R0[WS(rs, 4)];
		    T9 = R0[WS(rs, 20)];
		    Ta = T8 + T9;
		    T12 = T8 - T9;
	       }
	       T4l = T3 - T6;
	       T5a = Td - Ta;
	       T14 = KP707106781 * (T12 + T13);
	       T15 = T11 + T14;
	       T3n = T11 - T14;
	       T2R = KP707106781 * (T13 - T12);
	       T2T = T2R - T2S;
	       T3Q = T2S + T2R;
	       T7 = T3 + T6;
	       Te = Ta + Td;
	       Tf = T7 + Te;
	  }
	  {
	       E T1P, T4J, T21, T4y, T1S, T4K, T1W, T4z;
	       {
		    E T1N, T1O, T1Z, T20;
		    T1N = R1[WS(rs, 28)];
		    T1O = R1[WS(rs, 12)];
		    T1P = T1N - T1O;
		    T4J = T1N + T1O;
		    T1Z = R1[0];
		    T20 = R1[WS(rs, 16)];
		    T21 = T1Z - T20;
		    T4y = T1Z + T20;
	       }
	       {
		    E T1Q, T1R, T1U, T1V;
		    T1Q = R1[WS(rs, 4)];
		    T1R = R1[WS(rs, 20)];
		    T1S = T1Q - T1R;
		    T4K = T1Q + T1R;
		    T1U = R1[WS(rs, 8)];
		    T1V = R1[WS(rs, 24)];
		    T1W = T1U - T1V;
		    T4z = T1U + T1V;
	       }
	       T4A = T4y - T4z;
	       T4L = T4J - T4K;
	       {
		    E T1T, T22, T5G, T5H;
		    T1T = KP707106781 * (T1P - T1S);
		    T1X = T1T - T1W;
		    T3B = T1W + T1T;
		    T22 = KP707106781 * (T1S + T1P);
		    T23 = T21 + T22;
		    T3y = T21 - T22;
		    T5G = T4y + T4z;
		    T5H = T4K + T4J;
		    T5I = T5G + T5H;
		    T66 = T5G - T5H;
	       }
	  }
	  {
	       E T2b, T4P, T2G, T4Q, T2e, T51, T2h, T50;
	       {
		    E T29, T2a, T2E, T2F;
		    T29 = R1[WS(rs, 31)];
		    T2a = R1[WS(rs, 15)];
		    T2b = T29 - T2a;
		    T4P = T29 + T2a;
		    T2E = R1[WS(rs, 7)];
		    T2F = R1[WS(rs, 23)];
		    T2G = T2E - T2F;
		    T4Q = T2E + T2F;
	       }
	       {
		    E T2c, T2d, T2f, T2g;
		    T2c = R1[WS(rs, 3)];
		    T2d = R1[WS(rs, 19)];
		    T2e = T2c - T2d;
		    T51 = T2c + T2d;
		    T2f = R1[WS(rs, 27)];
		    T2g = R1[WS(rs, 11)];
		    T2h = T2f - T2g;
		    T50 = T2f + T2g;
	       }
	       T4R = T4P - T4Q;
	       T52 = T50 - T51;
	       {
		    E T2i, T2D, T5N, T5O;
		    T2i = KP707106781 * (T2e + T2h);
		    T2j = T2b + T2i;
		    T3F = T2b - T2i;
		    T2D = KP707106781 * (T2h - T2e);
		    T2H = T2D - T2G;
		    T3I = T2G + T2D;
		    T5N = T4P + T4Q;
		    T5O = T51 + T50;
		    T5P = T5N + T5O;
		    T69 = T5N - T5O;
	       }
	  }
	  {
	       E TN, T1e, TX, T1g, TQ, T1k, TU, T1f, T1h, T1j;
	       {
		    E TL, TM, TV, TW;
		    TL = R0[WS(rs, 31)];
		    TM = R0[WS(rs, 15)];
		    TN = TL + TM;
		    T1e = TL - TM;
		    TV = R0[WS(rs, 27)];
		    TW = R0[WS(rs, 11)];
		    TX = TV + TW;
		    T1g = TV - TW;
	       }
	       {
		    E TO, TP, TS, TT;
		    TO = R0[WS(rs, 7)];
		    TP = R0[WS(rs, 23)];
		    TQ = TO + TP;
		    T1k = TO - TP;
		    TS = R0[WS(rs, 3)];
		    TT = R0[WS(rs, 19)];
		    TU = TS + TT;
		    T1f = TS - TT;
	       }
	       T1h = KP707106781 * (T1f + T1g);
	       T1i = T1e + T1h;
	       T3t = T1e - T1h;
	       T1j = KP707106781 * (T1g - T1f);
	       T1l = T1j - T1k;
	       T3u = T1k + T1j;
	       {
		    E TR, TY, T4t, T4u;
		    TR = TN + TQ;
		    TY = TU + TX;
		    TZ = TR + TY;
		    T63 = TR - TY;
		    T4t = TN - TQ;
		    T4u = TX - TU;
		    T4v = FNMS(KP382683432, T4u, KP923879532 * T4t);
		    T58 = FMA(KP382683432, T4t, KP923879532 * T4u);
	       }
	  }
	  {
	       E Ty, T1s, TI, T1n, TB, T1q, TF, T1o, T1p, T1t;
	       {
		    E Tw, Tx, TG, TH;
		    Tw = R0[WS(rs, 1)];
		    Tx = R0[WS(rs, 17)];
		    Ty = Tw + Tx;
		    T1s = Tw - Tx;
		    TG = R0[WS(rs, 29)];
		    TH = R0[WS(rs, 13)];
		    TI = TG + TH;
		    T1n = TG - TH;
	       }
	       {
		    E Tz, TA, TD, TE;
		    Tz = R0[WS(rs, 9)];
		    TA = R0[WS(rs, 25)];
		    TB = Tz + TA;
		    T1q = Tz - TA;
		    TD = R0[WS(rs, 5)];
		    TE = R0[WS(rs, 21)];
		    TF = TD + TE;
		    T1o = TD - TE;
	       }
	       T1p = KP707106781 * (T1n - T1o);
	       T1r = T1p - T1q;
	       T3r = T1q + T1p;
	       T1t = KP707106781 * (T1o + T1n);
	       T1u = T1s + T1t;
	       T3q = T1s - T1t;
	       {
		    E TC, TJ, T4q, T4r;
		    TC = Ty + TB;
		    TJ = TF + TI;
		    TK = TC + TJ;
		    T62 = TC - TJ;
		    T4q = Ty - TB;
		    T4r = TI - TF;
		    T4s = FMA(KP923879532, T4q, KP382683432 * T4r);
		    T57 = FNMS(KP382683432, T4q, KP923879532 * T4r);
	       }
	  }
	  {
	       E Ti, T16, Ts, T1a, Tl, T17, Tp, T19, T4m, T4n;
	       {
		    E Tg, Th, Tq, Tr;
		    Tg = R0[WS(rs, 2)];
		    Th = R0[WS(rs, 18)];
		    Ti = Tg + Th;
		    T16 = Tg - Th;
		    Tq = R0[WS(rs, 6)];
		    Tr = R0[WS(rs, 22)];
		    Ts = Tq + Tr;
		    T1a = Tq - Tr;
	       }
	       {
		    E Tj, Tk, Tn, To;
		    Tj = R0[WS(rs, 10)];
		    Tk = R0[WS(rs, 26)];
		    Tl = Tj + Tk;
		    T17 = Tj - Tk;
		    Tn = R0[WS(rs, 30)];
		    To = R0[WS(rs, 14)];
		    Tp = Tn + To;
		    T19 = Tn - To;
	       }
	       Tm = Ti + Tl;
	       Tt = Tp + Ts;
	       Tu = Tm + Tt;
	       T4m = Ti - Tl;
	       T4n = Tp - Ts;
	       T4o = KP707106781 * (T4m + T4n);
	       T5b = KP707106781 * (T4n - T4m);
	       {
		    E T18, T1b, T2O, T2P;
		    T18 = FNMS(KP382683432, T17, KP923879532 * T16);
		    T1b = FMA(KP923879532, T19, KP382683432 * T1a);
		    T1c = T18 + T1b;
		    T3R = T1b - T18;
		    T2O = FNMS(KP923879532, T1a, KP382683432 * T19);
		    T2P = FMA(KP382683432, T16, KP923879532 * T17);
		    T2Q = T2O - T2P;
		    T3o = T2P + T2O;
	       }
	  }
	  {
	       E T1A, T4E, T1K, T4C, T1D, T4F, T1H, T4B;
	       {
		    E T1y, T1z, T1I, T1J;
		    T1y = R1[WS(rs, 30)];
		    T1z = R1[WS(rs, 14)];
		    T1A = T1y - T1z;
		    T4E = T1y + T1z;
		    T1I = R1[WS(rs, 10)];
		    T1J = R1[WS(rs, 26)];
		    T1K = T1I - T1J;
		    T4C = T1I + T1J;
	       }
	       {
		    E T1B, T1C, T1F, T1G;
		    T1B = R1[WS(rs, 6)];
		    T1C = R1[WS(rs, 22)];
		    T1D = T1B - T1C;
		    T4F = T1B + T1C;
		    T1F = R1[WS(rs, 2)];
		    T1G = R1[WS(rs, 18)];
		    T1H = T1F - T1G;
		    T4B = T1F + T1G;
	       }
	       {
		    E T1E, T1L, T5J, T5K;
		    T1E = FNMS(KP923879532, T1D, KP382683432 * T1A);
		    T1L = FMA(KP382683432, T1H, KP923879532 * T1K);
		    T1M = T1E - T1L;
		    T3z = T1L + T1E;
		    T5J = T4B + T4C;
		    T5K = T4E + T4F;
		    T5L = T5J + T5K;
		    T67 = T5K - T5J;
	       }
	       {
		    E T24, T25, T4D, T4G;
		    T24 = FNMS(KP382683432, T1K, KP923879532 * T1H);
		    T25 = FMA(KP923879532, T1A, KP382683432 * T1D);
		    T26 = T24 + T25;
		    T3C = T25 - T24;
		    T4D = T4B - T4C;
		    T4G = T4E - T4F;
		    T4H = KP707106781 * (T4D + T4G);
		    T4M = KP707106781 * (T4G - T4D);
	       }
	  }
	  {
	       E T2m, T4S, T2w, T4W, T2p, T4T, T2t, T4V;
	       {
		    E T2k, T2l, T2u, T2v;
		    T2k = R1[WS(rs, 1)];
		    T2l = R1[WS(rs, 17)];
		    T2m = T2k - T2l;
		    T4S = T2k + T2l;
		    T2u = R1[WS(rs, 5)];
		    T2v = R1[WS(rs, 21)];
		    T2w = T2u - T2v;
		    T4W = T2u + T2v;
	       }
	       {
		    E T2n, T2o, T2r, T2s;
		    T2n = R1[WS(rs, 9)];
		    T2o = R1[WS(rs, 25)];
		    T2p = T2n - T2o;
		    T4T = T2n + T2o;
		    T2r = R1[WS(rs, 29)];
		    T2s = R1[WS(rs, 13)];
		    T2t = T2r - T2s;
		    T4V = T2r + T2s;
	       }
	       {
		    E T2q, T2x, T5Q, T5R;
		    T2q = FNMS(KP382683432, T2p, KP923879532 * T2m);
		    T2x = FMA(KP923879532, T2t, KP382683432 * T2w);
		    T2y = T2q + T2x;
		    T3J = T2x - T2q;
		    T5Q = T4S + T4T;
		    T5R = T4V + T4W;
		    T5S = T5Q + T5R;
		    T6a = T5R - T5Q;
	       }
	       {
		    E T2A, T2B, T4U, T4X;
		    T2A = FNMS(KP923879532, T2w, KP382683432 * T2t);
		    T2B = FMA(KP382683432, T2m, KP923879532 * T2p);
		    T2C = T2A - T2B;
		    T3G = T2B + T2A;
		    T4U = T4S - T4T;
		    T4X = T4V - T4W;
		    T4Y = KP707106781 * (T4U + T4X);
		    T53 = KP707106781 * (T4X - T4U);
	       }
	  }
	  {
	       E Tv, T10, T5X, T5Y, T5Z, T60;
	       Tv = Tf + Tu;
	       T10 = TK + TZ;
	       T5X = Tv + T10;
	       T5Y = T5I + T5L;
	       T5Z = T5P + T5S;
	       T60 = T5Y + T5Z;
	       Cr[WS(csr, 16)] = Tv - T10;
	       Ci[WS(csi, 16)] = T5Z - T5Y;
	       Cr[WS(csr, 32)] = T5X - T60;
	       Cr[0] = T5X + T60;
	  }
	  {
	       E T5F, T5V, T5U, T5W, T5M, T5T;
	       T5F = Tf - Tu;
	       T5V = TZ - TK;
	       T5M = T5I - T5L;
	       T5T = T5P - T5S;
	       T5U = KP707106781 * (T5M + T5T);
	       T5W = KP707106781 * (T5T - T5M);
	       Cr[WS(csr, 24)] = T5F - T5U;
	       Ci[WS(csi, 24)] = T5W - T5V;
	       Cr[WS(csr, 8)] = T5F + T5U;
	       Ci[WS(csi, 8)] = T5V + T5W;
	  }
	  {
	       E T65, T6l, T6k, T6m, T6c, T6g, T6f, T6h;
	       {
		    E T61, T64, T6i, T6j;
		    T61 = T7 - Te;
		    T64 = KP707106781 * (T62 + T63);
		    T65 = T61 + T64;
		    T6l = T61 - T64;
		    T6i = FNMS(KP382683432, T66, KP923879532 * T67);
		    T6j = FMA(KP382683432, T69, KP923879532 * T6a);
		    T6k = T6i + T6j;
		    T6m = T6j - T6i;
	       }
	       {
		    E T68, T6b, T6d, T6e;
		    T68 = FMA(KP923879532, T66, KP382683432 * T67);
		    T6b = FNMS(KP382683432, T6a, KP923879532 * T69);
		    T6c = T68 + T6b;
		    T6g = T6b - T68;
		    T6d = KP707106781 * (T63 - T62);
		    T6e = Tt - Tm;
		    T6f = T6d - T6e;
		    T6h = T6e + T6d;
	       }
	       Cr[WS(csr, 28)] = T65 - T6c;
	       Ci[WS(csi, 28)] = T6k - T6h;
	       Cr[WS(csr, 4)] = T65 + T6c;
	       Ci[WS(csi, 4)] = T6h + T6k;
	       Ci[WS(csi, 12)] = T6f + T6g;
	       Cr[WS(csr, 12)] = T6l + T6m;
	       Ci[WS(csi, 20)] = T6g - T6f;
	       Cr[WS(csr, 20)] = T6l - T6m;
	  }
	  {
	       E T5n, T5D, T5x, T5z, T5q, T5A, T5t, T5B;
	       {
		    E T5l, T5m, T5v, T5w;
		    T5l = T4l - T4o;
		    T5m = T58 - T57;
		    T5n = T5l + T5m;
		    T5D = T5l - T5m;
		    T5v = T4v - T4s;
		    T5w = T5b - T5a;
		    T5x = T5v - T5w;
		    T5z = T5w + T5v;
	       }
	       {
		    E T5o, T5p, T5r, T5s;
		    T5o = T4A - T4H;
		    T5p = T4M - T4L;
		    T5q = FMA(KP831469612, T5o, KP555570233 * T5p);
		    T5A = FNMS(KP555570233, T5o, KP831469612 * T5p);
		    T5r = T4R - T4Y;
		    T5s = T53 - T52;
		    T5t = FNMS(KP555570233, T5s, KP831469612 * T5r);
		    T5B = FMA(KP555570233, T5r, KP831469612 * T5s);
	       }
	       {
		    E T5u, T5C, T5y, T5E;
		    T5u = T5q + T5t;
		    Cr[WS(csr, 26)] = T5n - T5u;
		    Cr[WS(csr, 6)] = T5n + T5u;
		    T5C = T5A + T5B;
		    Ci[WS(csi, 6)] = T5z + T5C;
		    Ci[WS(csi, 26)] = T5C - T5z;
		    T5y = T5t - T5q;
		    Ci[WS(csi, 10)] = T5x + T5y;
		    Ci[WS(csi, 22)] = T5y - T5x;
		    T5E = T5B - T5A;
		    Cr[WS(csr, 22)] = T5D - T5E;
		    Cr[WS(csr, 10)] = T5D + T5E;
	       }
	  }
	  {
	       E T4x, T5j, T5d, T5f, T4O, T5g, T55, T5h;
	       {
		    E T4p, T4w, T59, T5c;
		    T4p = T4l + T4o;
		    T4w = T4s + T4v;
		    T4x = T4p + T4w;
		    T5j = T4p - T4w;
		    T59 = T57 + T58;
		    T5c = T5a + T5b;
		    T5d = T59 - T5c;
		    T5f = T5c + T59;
	       }
	       {
		    E T4I, T4N, T4Z, T54;
		    T4I = T4A + T4H;
		    T4N = T4L + T4M;
		    T4O = FMA(KP980785280, T4I, KP195090322 * T4N);
		    T5g = FNMS(KP195090322, T4I, KP980785280 * T4N);
		    T4Z = T4R + T4Y;
		    T54 = T52 + T53;
		    T55 = FNMS(KP195090322, T54, KP980785280 * T4Z);
		    T5h = FMA(KP195090322, T4Z, KP980785280 * T54);
	       }
	       {
		    E T56, T5i, T5e, T5k;
		    T56 = T4O + T55;
		    Cr[WS(csr, 30)] = T4x - T56;
		    Cr[WS(csr, 2)] = T4x + T56;
		    T5i = T5g + T5h;
		    Ci[WS(csi, 2)] = T5f + T5i;
		    Ci[WS(csi, 30)] = T5i - T5f;
		    T5e = T55 - T4O;
		    Ci[WS(csi, 14)] = T5d + T5e;
		    Ci[WS(csi, 18)] = T5e - T5d;
		    T5k = T5h - T5g;
		    Cr[WS(csr, 18)] = T5j - T5k;
		    Cr[WS(csr, 14)] = T5j + T5k;
	       }
	  }
	  {
	       E T3p, T41, T4c, T3S, T3w, T4b, T49, T4h, T3P, T42, T3E, T3W, T46, T4g, T3L;
	       E T3X;
	       {
		    E T3s, T3v, T3A, T3D;
		    T3p = T3n + T3o;
		    T41 = T3n - T3o;
		    T4c = T3R - T3Q;
		    T3S = T3Q + T3R;
		    T3s = FMA(KP831469612, T3q, KP555570233 * T3r);
		    T3v = FNMS(KP555570233, T3u, KP831469612 * T3t);
		    T3w = T3s + T3v;
		    T4b = T3v - T3s;
		    {
			 E T47, T48, T3N, T3O;
			 T47 = T3F - T3G;
			 T48 = T3J - T3I;
			 T49 = FNMS(KP471396736, T48, KP881921264 * T47);
			 T4h = FMA(KP471396736, T47, KP881921264 * T48);
			 T3N = FNMS(KP555570233, T3q, KP831469612 * T3r);
			 T3O = FMA(KP555570233, T3t, KP831469612 * T3u);
			 T3P = T3N + T3O;
			 T42 = T3O - T3N;
		    }
		    T3A = T3y + T3z;
		    T3D = T3B + T3C;
		    T3E = FMA(KP956940335, T3A, KP290284677 * T3D);
		    T3W = FNMS(KP290284677, T3A, KP956940335 * T3D);
		    {
			 E T44, T45, T3H, T3K;
			 T44 = T3y - T3z;
			 T45 = T3C - T3B;
			 T46 = FMA(KP881921264, T44, KP471396736 * T45);
			 T4g = FNMS(KP471396736, T44, KP881921264 * T45);
			 T3H = T3F + T3G;
			 T3K = T3I + T3J;
			 T3L = FNMS(KP290284677, T3K, KP956940335 * T3H);
			 T3X = FMA(KP290284677, T3H, KP956940335 * T3K);
		    }
	       }
	       {
		    E T3x, T3M, T3V, T3Y;
		    T3x = T3p + T3w;
		    T3M = T3E + T3L;
		    Cr[WS(csr, 29)] = T3x - T3M;
		    Cr[WS(csr, 3)] = T3x + T3M;
		    T3V = T3S + T3P;
		    T3Y = T3W + T3X;
		    Ci[WS(csi, 3)] = T3V + T3Y;
		    Ci[WS(csi, 29)] = T3Y - T3V;
	       }
	       {
		    E T3T, T3U, T3Z, T40;
		    T3T = T3P - T3S;
		    T3U = T3L - T3E;
		    Ci[WS(csi, 13)] = T3T + T3U;
		    Ci[WS(csi, 19)] = T3U - T3T;
		    T3Z = T3p - T3w;
		    T40 = T3X - T3W;
		    Cr[WS(csr, 19)] = T3Z - T40;
		    Cr[WS(csr, 13)] = T3Z + T40;
	       }
	       {
		    E T43, T4a, T4f, T4i;
		    T43 = T41 + T42;
		    T4a = T46 + T49;
		    Cr[WS(csr, 27)] = T43 - T4a;
		    Cr[WS(csr, 5)] = T43 + T4a;
		    T4f = T4c + T4b;
		    T4i = T4g + T4h;
		    Ci[WS(csi, 5)] = T4f + T4i;
		    Ci[WS(csi, 27)] = T4i - T4f;
	       }
	       {
		    E T4d, T4e, T4j, T4k;
		    T4d = T4b - T4c;
		    T4e = T49 - T46;
		    Ci[WS(csi, 11)] = T4d + T4e;
		    Ci[WS(csi, 21)] = T4e - T4d;
		    T4j = T41 - T42;
		    T4k = T4h - T4g;
		    Cr[WS(csr, 21)] = T4j - T4k;
		    Cr[WS(csr, 11)] = T4j + T4k;
	       }
	  }
	  {
	       E T1d, T33, T3e, T2U, T1w, T3d, T3b, T3j, T2N, T34, T28, T2Y, T38, T3i, T2J;
	       E T2Z;
	       {
		    E T1m, T1v, T1Y, T27;
		    T1d = T15 - T1c;
		    T33 = T15 + T1c;
		    T3e = T2T + T2Q;
		    T2U = T2Q - T2T;
		    T1m = FMA(KP195090322, T1i, KP980785280 * T1l);
		    T1v = FNMS(KP195090322, T1u, KP980785280 * T1r);
		    T1w = T1m - T1v;
		    T3d = T1v + T1m;
		    {
			 E T39, T3a, T2L, T2M;
			 T39 = T2j + T2y;
			 T3a = T2H + T2C;
			 T3b = FNMS(KP098017140, T3a, KP995184726 * T39);
			 T3j = FMA(KP995184726, T3a, KP098017140 * T39);
			 T2L = FNMS(KP195090322, T1l, KP980785280 * T1i);
			 T2M = FMA(KP980785280, T1u, KP195090322 * T1r);
			 T2N = T2L - T2M;
			 T34 = T2M + T2L;
		    }
		    T1Y = T1M - T1X;
		    T27 = T23 - T26;
		    T28 = FMA(KP634393284, T1Y, KP773010453 * T27);
		    T2Y = FNMS(KP634393284, T27, KP773010453 * T1Y);
		    {
			 E T36, T37, T2z, T2I;
			 T36 = T1X + T1M;
			 T37 = T23 + T26;
			 T38 = FMA(KP098017140, T36, KP995184726 * T37);
			 T3i = FNMS(KP098017140, T37, KP995184726 * T36);
			 T2z = T2j - T2y;
			 T2I = T2C - T2H;
			 T2J = FNMS(KP634393284, T2I, KP773010453 * T2z);
			 T2Z = FMA(KP773010453, T2I, KP634393284 * T2z);
		    }
	       }
	       {
		    E T1x, T2K, T2X, T30;
		    T1x = T1d + T1w;
		    T2K = T28 + T2J;
		    Cr[WS(csr, 25)] = T1x - T2K;
		    Cr[WS(csr, 7)] = T1x + T2K;
		    T2X = T2U + T2N;
		    T30 = T2Y + T2Z;
		    Ci[WS(csi, 7)] = T2X + T30;
		    Ci[WS(csi, 25)] = T30 - T2X;
	       }
	       {
		    E T2V, T2W, T31, T32;
		    T2V = T2N - T2U;
		    T2W = T2J - T28;
		    Ci[WS(csi, 9)] = T2V + T2W;
		    Ci[WS(csi, 23)] = T2W - T2V;
		    T31 = T1d - T1w;
		    T32 = T2Z - T2Y;
		    Cr[WS(csr, 23)] = T31 - T32;
		    Cr[WS(csr, 9)] = T31 + T32;
	       }
	       {
		    E T35, T3c, T3h, T3k;
		    T35 = T33 + T34;
		    T3c = T38 + T3b;
		    Cr[WS(csr, 31)] = T35 - T3c;
		    Cr[WS(csr, 1)] = T35 + T3c;
		    T3h = T3e + T3d;
		    T3k = T3i + T3j;
		    Ci[WS(csi, 1)] = T3h + T3k;
		    Ci[WS(csi, 31)] = T3k - T3h;
	       }
	       {
		    E T3f, T3g, T3l, T3m;
		    T3f = T3d - T3e;
		    T3g = T3b - T38;
		    Ci[WS(csi, 15)] = T3f + T3g;
		    Ci[WS(csi, 17)] = T3g - T3f;
		    T3l = T33 - T34;
		    T3m = T3j - T3i;
		    Cr[WS(csr, 17)] = T3l - T3m;
		    Cr[WS(csr, 15)] = T3l + T3m;
	       }
	  }
     }
}
static void r2cb_14(R *R0, R *R1, R *Cr, R *Ci, stride rs, stride csr, stride csi, INT v, INT ivs, INT ovs)
{
     DK(KP1_801937735, +1.801937735804838252472204639014890102331838324);
     DK(KP445041867, +0.445041867912628808577805128993589518932711138);
     DK(KP1_246979603, +1.246979603717467061050009768008479621264549462);
     DK(KP867767478, +0.867767478235116240951536665696717509219981456);
     DK(KP1_949855824, +1.949855824363647214036263365987862434465571601);
     DK(KP1_563662964, +1.563662964936059617416889053348115500464669037);
     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(56, rs), MAKE_VOLATILE_STRIDE(56, csr), MAKE_VOLATILE_STRIDE(56, csi)) {
	       E T3, Td, T6, Te, Tq, Tz, Tn, Ty, Tc, Tg, Tk, Tx, T9, Tf, T1;
	       E T2;
	       T1 = Cr[0];
	       T2 = Cr[WS(csr, 7)];
	       T3 = T1 - T2;
	       Td = T1 + T2;
	       {
		    E T4, T5, To, Tp;
		    T4 = Cr[WS(csr, 2)];
		    T5 = Cr[WS(csr, 5)];
		    T6 = T4 - T5;
		    Te = T4 + T5;
		    To = Ci[WS(csi, 2)];
		    Tp = Ci[WS(csi, 5)];
		    Tq = To - Tp;
		    Tz = To + Tp;
	       }
	       {
		    E Tl, Tm, Ta, Tb;
		    Tl = Ci[WS(csi, 6)];
		    Tm = Ci[WS(csi, 1)];
		    Tn = Tl - Tm;
		    Ty = Tl + Tm;
		    Ta = Cr[WS(csr, 6)];
		    Tb = Cr[WS(csr, 1)];
		    Tc = Ta - Tb;
		    Tg = Ta + Tb;
	       }
	       {
		    E Ti, Tj, T7, T8;
		    Ti = Ci[WS(csi, 4)];
		    Tj = Ci[WS(csi, 3)];
		    Tk = Ti - Tj;
		    Tx = Ti + Tj;
		    T7 = Cr[WS(csr, 4)];
		    T8 = Cr[WS(csr, 3)];
		    T9 = T7 - T8;
		    Tf = T7 + T8;
	       }
	       R1[WS(rs, 3)] = FMA(KP2_000000000, T6 + T9 + Tc, T3);
	       R0[0] = FMA(KP2_000000000, Te + Tf + Tg, Td);
	       {
		    E Tr, Th, TE, TD;
		    Tr = FNMS(KP1_949855824, Tn, KP1_563662964 * Tk) - (KP867767478 * Tq);
		    Th = FMA(KP1_246979603, Tf, Td) + FNMA(KP445041867, Tg, KP1_801937735 * Te);
		    R0[WS(rs, 2)] = Th - Tr;
		    R0[WS(rs, 5)] = Th + Tr;
		    TE = FMA(KP867767478, Tx, KP1_563662964 * Ty) - (KP1_949855824 * Tz);
		    TD = FMA(KP1_246979603, Tc, T3) + FNMA(KP1_801937735, T9, KP445041867 * T6);
		    R1[WS(rs, 2)] = TD - TE;
		    R1[WS(rs, 4)] = TD + TE;
	       }
	       {
		    E Tt, Ts, TA, Tw;
		    Tt = FMA(KP867767478, Tk, KP1_563662964 * Tn) - (KP1_949855824 * Tq);
		    Ts = FMA(KP1_246979603, Tg, Td) + FNMA(KP1_801937735, Tf, KP445041867 * Te);
		    R0[WS(rs, 6)] = Ts - Tt;
		    R0[WS(rs, 1)] = Ts + Tt;
		    TA = FNMS(KP1_949855824, Ty, KP1_563662964 * Tx) - (KP867767478 * Tz);
		    Tw = FMA(KP1_246979603, T9, T3) + FNMA(KP445041867, Tc, KP1_801937735 * T6);
		    R1[WS(rs, 5)] = Tw - TA;
		    R1[WS(rs, 1)] = Tw + TA;
	       }
	       {
		    E TC, TB, Tv, Tu;
		    TC = FMA(KP1_563662964, Tz, KP1_949855824 * Tx) + (KP867767478 * Ty);
		    TB = FMA(KP1_246979603, T6, T3) + FNMA(KP1_801937735, Tc, KP445041867 * T9);
		    R1[0] = TB - TC;
		    R1[WS(rs, 6)] = TB + TC;
		    Tv = FMA(KP1_563662964, Tq, KP1_949855824 * Tk) + (KP867767478 * Tn);
		    Tu = FMA(KP1_246979603, Te, Td) + FNMA(KP1_801937735, Tg, KP445041867 * Tf);
		    R0[WS(rs, 4)] = Tu - Tv;
		    R0[WS(rs, 3)] = Tu + Tv;
	       }
	  }
     }
}
Exemple #16
0
static void n1fv_13(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs)
{
     DVK(KP2_000000000, +2.000000000000000000000000000000000000000000000);
     DVK(KP083333333, +0.083333333333333333333333333333333333333333333);
     DVK(KP075902986, +0.075902986037193865983102897245103540356428373);
     DVK(KP251768516, +0.251768516431883313623436926934233488546674281);
     DVK(KP132983124, +0.132983124607418643793760531921092974399165133);
     DVK(KP258260390, +0.258260390311744861420450644284508567852516811);
     DVK(KP1_732050807, +1.732050807568877293527446341505872366942805254);
     DVK(KP300238635, +0.300238635966332641462884626667381504676006424);
     DVK(KP011599105, +0.011599105605768290721655456654083252189827041);
     DVK(KP156891391, +0.156891391051584611046832726756003269660212636);
     DVK(KP256247671, +0.256247671582936600958684654061725059144125175);
     DVK(KP174138601, +0.174138601152135905005660794929264742616964676);
     DVK(KP575140729, +0.575140729474003121368385547455453388461001608);
     DVK(KP503537032, +0.503537032863766627246873853868466977093348562);
     DVK(KP113854479, +0.113854479055790798974654345867655310534642560);
     DVK(KP265966249, +0.265966249214837287587521063842185948798330267);
     DVK(KP387390585, +0.387390585467617292130675966426762851778775217);
     DVK(KP300462606, +0.300462606288665774426601772289207995520941381);
     DVK(KP866025403, +0.866025403784438646763723170752936183471402627);
     DVK(KP500000000, +0.500000000000000000000000000000000000000000000);
     {
	  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(26, is), MAKE_VOLATILE_STRIDE(26, os)) {
	       V TW, Tb, Tm, Tu, TC, TR, TX, TK, TU, Tz, TB, TN, TT;
	       TW = LD(&(xi[0]), ivs, &(xi[0]));
	       {
		    V T3, TH, Tl, Tw, Tp, Tg, Tv, To, T6, Tr, T9, Ts, Ta, TI, T1;
		    V T2, Tq, Tt;
		    T1 = LD(&(xi[WS(is, 8)]), ivs, &(xi[0]));
		    T2 = LD(&(xi[WS(is, 5)]), ivs, &(xi[WS(is, 1)]));
		    T3 = VSUB(T1, T2);
		    TH = VADD(T1, T2);
		    {
			 V Th, Ti, Tj, Tk;
			 Th = LD(&(xi[WS(is, 12)]), ivs, &(xi[0]));
			 Ti = LD(&(xi[WS(is, 10)]), ivs, &(xi[0]));
			 Tj = LD(&(xi[WS(is, 4)]), ivs, &(xi[0]));
			 Tk = VADD(Ti, Tj);
			 Tl = VADD(Th, Tk);
			 Tw = VSUB(Ti, Tj);
			 Tp = VFNMS(LDK(KP500000000), Tk, Th);
		    }
		    {
			 V Tc, Td, Te, Tf;
			 Tc = LD(&(xi[WS(is, 1)]), ivs, &(xi[WS(is, 1)]));
			 Td = LD(&(xi[WS(is, 3)]), ivs, &(xi[WS(is, 1)]));
			 Te = LD(&(xi[WS(is, 9)]), ivs, &(xi[WS(is, 1)]));
			 Tf = VADD(Td, Te);
			 Tg = VADD(Tc, Tf);
			 Tv = VSUB(Td, Te);
			 To = VFNMS(LDK(KP500000000), Tf, Tc);
		    }
		    {
			 V T4, T5, T7, T8;
			 T4 = LD(&(xi[WS(is, 11)]), ivs, &(xi[WS(is, 1)]));
			 T5 = LD(&(xi[WS(is, 6)]), ivs, &(xi[0]));
			 T6 = VSUB(T4, T5);
			 Tr = VADD(T4, T5);
			 T7 = LD(&(xi[WS(is, 7)]), ivs, &(xi[WS(is, 1)]));
			 T8 = LD(&(xi[WS(is, 2)]), ivs, &(xi[0]));
			 T9 = VSUB(T7, T8);
			 Ts = VADD(T7, T8);
		    }
		    Ta = VADD(T6, T9);
		    TI = VADD(Tr, Ts);
		    Tb = VADD(T3, Ta);
		    Tm = VSUB(Tg, Tl);
		    Tq = VSUB(To, Tp);
		    Tt = VMUL(LDK(KP866025403), VSUB(Tr, Ts));
		    Tu = VADD(Tq, Tt);
		    TC = VSUB(Tq, Tt);
		    {
			 V TP, TQ, TG, TJ;
			 TP = VADD(Tg, Tl);
			 TQ = VADD(TH, TI);
			 TR = VMUL(LDK(KP300462606), VSUB(TP, TQ));
			 TX = VADD(TP, TQ);
			 TG = VADD(To, Tp);
			 TJ = VFNMS(LDK(KP500000000), TI, TH);
			 TK = VSUB(TG, TJ);
			 TU = VADD(TG, TJ);
		    }
		    {
			 V Tx, Ty, TL, TM;
			 Tx = VMUL(LDK(KP866025403), VSUB(Tv, Tw));
			 Ty = VFNMS(LDK(KP500000000), Ta, T3);
			 Tz = VSUB(Tx, Ty);
			 TB = VADD(Tx, Ty);
			 TL = VADD(Tv, Tw);
			 TM = VSUB(T6, T9);
			 TN = VSUB(TL, TM);
			 TT = VADD(TL, TM);
		    }
	       }
	       ST(&(xo[0]), VADD(TW, TX), ovs, &(xo[0]));
	       {
		    V T19, T1n, T14, T13, T1f, T1k, Tn, TE, T1e, T1j, TS, T1m, TZ, T1c, TA;
		    V TD;
		    {
			 V T17, T18, T11, T12;
			 T17 = VFMA(LDK(KP387390585), TN, VMUL(LDK(KP265966249), TK));
			 T18 = VFNMS(LDK(KP503537032), TU, VMUL(LDK(KP113854479), TT));
			 T19 = VSUB(T17, T18);
			 T1n = VADD(T17, T18);
			 T14 = VFMA(LDK(KP575140729), Tm, VMUL(LDK(KP174138601), Tb));
			 T11 = VFNMS(LDK(KP156891391), TB, VMUL(LDK(KP256247671), TC));
			 T12 = VFMA(LDK(KP011599105), Tz, VMUL(LDK(KP300238635), Tu));
			 T13 = VSUB(T11, T12);
			 T1f = VADD(T14, T13);
			 T1k = VMUL(LDK(KP1_732050807), VADD(T11, T12));
		    }
		    Tn = VFNMS(LDK(KP174138601), Tm, VMUL(LDK(KP575140729), Tb));
		    TA = VFNMS(LDK(KP300238635), Tz, VMUL(LDK(KP011599105), Tu));
		    TD = VFMA(LDK(KP256247671), TB, VMUL(LDK(KP156891391), TC));
		    TE = VSUB(TA, TD);
		    T1e = VMUL(LDK(KP1_732050807), VADD(TD, TA));
		    T1j = VSUB(Tn, TE);
		    {
			 V TO, T1b, TV, TY, T1a;
			 TO = VFNMS(LDK(KP132983124), TN, VMUL(LDK(KP258260390), TK));
			 T1b = VSUB(TR, TO);
			 TV = VFMA(LDK(KP251768516), TT, VMUL(LDK(KP075902986), TU));
			 TY = VFNMS(LDK(KP083333333), TX, TW);
			 T1a = VSUB(TY, TV);
			 TS = VFMA(LDK(KP2_000000000), TO, TR);
			 T1m = VADD(T1b, T1a);
			 TZ = VFMA(LDK(KP2_000000000), TV, TY);
			 T1c = VSUB(T1a, T1b);
		    }
		    {
			 V TF, T10, T1l, T1o;
			 TF = VBYI(VFMA(LDK(KP2_000000000), TE, Tn));
			 T10 = VADD(TS, TZ);
			 ST(&(xo[WS(os, 1)]), VADD(TF, T10), ovs, &(xo[WS(os, 1)]));
			 ST(&(xo[WS(os, 12)]), VSUB(T10, TF), ovs, &(xo[0]));
			 {
			      V T15, T16, T1p, T1q;
			      T15 = VBYI(VFMS(LDK(KP2_000000000), T13, T14));
			      T16 = VSUB(TZ, TS);
			      ST(&(xo[WS(os, 5)]), VADD(T15, T16), ovs, &(xo[WS(os, 1)]));
			      ST(&(xo[WS(os, 8)]), VSUB(T16, T15), ovs, &(xo[0]));
			      T1p = VADD(T1n, T1m);
			      T1q = VBYI(VADD(T1j, T1k));
			      ST(&(xo[WS(os, 4)]), VSUB(T1p, T1q), ovs, &(xo[0]));
			      ST(&(xo[WS(os, 9)]), VADD(T1q, T1p), ovs, &(xo[WS(os, 1)]));
			 }
			 T1l = VBYI(VSUB(T1j, T1k));
			 T1o = VSUB(T1m, T1n);
			 ST(&(xo[WS(os, 3)]), VADD(T1l, T1o), ovs, &(xo[WS(os, 1)]));
			 ST(&(xo[WS(os, 10)]), VSUB(T1o, T1l), ovs, &(xo[0]));
			 {
			      V T1h, T1i, T1d, T1g;
			      T1h = VBYI(VSUB(T1e, T1f));
			      T1i = VSUB(T1c, T19);
			      ST(&(xo[WS(os, 6)]), VADD(T1h, T1i), ovs, &(xo[0]));
			      ST(&(xo[WS(os, 7)]), VSUB(T1i, T1h), ovs, &(xo[WS(os, 1)]));
			      T1d = VADD(T19, T1c);
			      T1g = VBYI(VADD(T1e, T1f));
			      ST(&(xo[WS(os, 2)]), VSUB(T1d, T1g), ovs, &(xo[0]));
			      ST(&(xo[WS(os, 11)]), VADD(T1g, T1d), ovs, &(xo[WS(os, 1)]));
			 }
		    }
	       }
	  }
     }
     VLEAVE();
}
static void r2cb_14(R *R0, R *R1, R *Cr, R *Ci, stride rs, stride csr, stride csi, INT v, INT ivs, INT ovs)
{
     DK(KP1_949855824, +1.949855824363647214036263365987862434465571601);
     DK(KP1_801937735, +1.801937735804838252472204639014890102331838324);
     DK(KP692021471, +0.692021471630095869627814897002069140197260599);
     DK(KP2_000000000, +2.000000000000000000000000000000000000000000000);
     DK(KP356895867, +0.356895867892209443894399510021300583399127187);
     DK(KP801937735, +0.801937735804838252472204639014890102331838324);
     DK(KP554958132, +0.554958132087371191422194871006410481067288862);
     {
	  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(56, rs), MAKE_VOLATILE_STRIDE(56, csr), MAKE_VOLATILE_STRIDE(56, csi)) {
	       E Te, TO, TT, TG, TJ, TD, TR, TE;
	       {
		    E T3, TK, To, TM, Tu, TL, Tr, TS, TA, TN, TX, TF, Tv, T7, Tf;
		    E T6, Th, Tc, T8, T1, T2;
		    T1 = Cr[0];
		    T2 = Cr[WS(csr, 7)];
		    {
			 E Ts, Tt, Tp, Tq, Tm, Tn;
			 Tm = Ci[WS(csi, 4)];
			 Tn = Ci[WS(csi, 3)];
			 Ts = Ci[WS(csi, 6)];
			 Te = T1 + T2;
			 T3 = T1 - T2;
			 TK = Tm + Tn;
			 To = Tm - Tn;
			 Tt = Ci[WS(csi, 1)];
			 Tp = Ci[WS(csi, 2)];
			 Tq = Ci[WS(csi, 5)];
			 {
			      E T4, T5, Ta, Tb;
			      T4 = Cr[WS(csr, 2)];
			      TM = Ts + Tt;
			      Tu = Ts - Tt;
			      TL = Tp + Tq;
			      Tr = Tp - Tq;
			      TS = FMA(KP554958132, TK, TM);
			      TA = FMA(KP554958132, To, Tu);
			      TN = FMA(KP554958132, TM, TL);
			      TX = FNMS(KP554958132, TL, TK);
			      TF = FNMS(KP554958132, Tr, To);
			      Tv = FMA(KP554958132, Tu, Tr);
			      T5 = Cr[WS(csr, 5)];
			      Ta = Cr[WS(csr, 6)];
			      Tb = Cr[WS(csr, 1)];
			      T7 = Cr[WS(csr, 4)];
			      Tf = T4 + T5;
			      T6 = T4 - T5;
			      Th = Ta + Tb;
			      Tc = Ta - Tb;
			      T8 = Cr[WS(csr, 3)];
			 }
		    }
		    {
			 E Tw, Tx, TP, Tg, T9, TY, TC, TI, TQ;
			 Tw = FMA(KP801937735, Tv, To);
			 Tx = FNMS(KP356895867, Tf, Th);
			 TP = FNMS(KP356895867, T6, Tc);
			 Tg = T7 + T8;
			 T9 = T7 - T8;
			 TY = FNMS(KP801937735, TX, TM);
			 {
			      E TB, TH, TV, Ty, Tl, Ti, TW, Tz;
			      TB = FNMS(KP801937735, TA, Tr);
			      Ti = Tf + Tg + Th;
			      TC = FNMS(KP356895867, Th, Tg);
			      {
				   E Tj, Td, TU, Tk;
				   Tj = FNMS(KP356895867, Tg, Tf);
				   Td = T6 + T9 + Tc;
				   TH = FNMS(KP356895867, T9, T6);
				   TU = FNMS(KP356895867, Tc, T9);
				   R0[0] = FMA(KP2_000000000, Ti, Te);
				   Tk = FNMS(KP692021471, Tj, Th);
				   R1[WS(rs, 3)] = FMA(KP2_000000000, Td, T3);
				   TV = FNMS(KP692021471, TU, T6);
				   Ty = FNMS(KP692021471, Tx, Tg);
				   Tl = FNMS(KP1_801937735, Tk, Te);
			      }
			      TO = FMA(KP801937735, TN, TK);
			      TW = FNMS(KP1_801937735, TV, T3);
			      Tz = FNMS(KP1_801937735, Ty, Te);
			      R0[WS(rs, 3)] = FMA(KP1_949855824, Tw, Tl);
			      R0[WS(rs, 4)] = FNMS(KP1_949855824, Tw, Tl);
			      R1[WS(rs, 5)] = FMA(KP1_949855824, TY, TW);
			      R1[WS(rs, 1)] = FNMS(KP1_949855824, TY, TW);
			      R0[WS(rs, 6)] = FMA(KP1_949855824, TB, Tz);
			      R0[WS(rs, 1)] = FNMS(KP1_949855824, TB, Tz);
			      TI = FNMS(KP692021471, TH, Tc);
			 }
			 TT = FNMS(KP801937735, TS, TL);
			 TQ = FNMS(KP692021471, TP, T9);
			 TG = FNMS(KP801937735, TF, Tu);
			 TJ = FNMS(KP1_801937735, TI, T3);
			 TD = FNMS(KP692021471, TC, Tf);
			 TR = FNMS(KP1_801937735, TQ, T3);
		    }
	       }
	       R1[WS(rs, 6)] = FMA(KP1_949855824, TO, TJ);
	       R1[0] = FNMS(KP1_949855824, TO, TJ);
	       TE = FNMS(KP1_801937735, TD, Te);
	       R1[WS(rs, 2)] = FMA(KP1_949855824, TT, TR);
	       R1[WS(rs, 4)] = FNMS(KP1_949855824, TT, TR);
	       R0[WS(rs, 2)] = FMA(KP1_949855824, TG, TE);
	       R0[WS(rs, 5)] = FNMS(KP1_949855824, TG, TE);
	  }
     }
}
Exemple #18
0
static void n1fv_13(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs)
{
     DVK(KP904176221, +0.904176221990848204433795481776887926501523162);
     DVK(KP575140729, +0.575140729474003121368385547455453388461001608);
     DVK(KP300462606, +0.300462606288665774426601772289207995520941381);
     DVK(KP516520780, +0.516520780623489722840901288569017135705033622);
     DVK(KP522026385, +0.522026385161275033714027226654165028300441940);
     DVK(KP957805992, +0.957805992594665126462521754605754580515587217);
     DVK(KP600477271, +0.600477271932665282925769253334763009352012849);
     DVK(KP251768516, +0.251768516431883313623436926934233488546674281);
     DVK(KP503537032, +0.503537032863766627246873853868466977093348562);
     DVK(KP769338817, +0.769338817572980603471413688209101117038278899);
     DVK(KP859542535, +0.859542535098774820163672132761689612766401925);
     DVK(KP581704778, +0.581704778510515730456870384989698884939833902);
     DVK(KP853480001, +0.853480001859823990758994934970528322872359049);
     DVK(KP083333333, +0.083333333333333333333333333333333333333333333);
     DVK(KP226109445, +0.226109445035782405468510155372505010481906348);
     DVK(KP301479260, +0.301479260047709873958013540496673347309208464);
     DVK(KP686558370, +0.686558370781754340655719594850823015421401653);
     DVK(KP514918778, +0.514918778086315755491789696138117261566051239);
     DVK(KP038632954, +0.038632954644348171955506895830342264440241080);
     DVK(KP612264650, +0.612264650376756543746494474777125408779395514);
     DVK(KP302775637, +0.302775637731994646559610633735247973125648287);
     DVK(KP866025403, +0.866025403784438646763723170752936183471402627);
     DVK(KP500000000, +0.500000000000000000000000000000000000000000000);
     {
	  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(26, is), MAKE_VOLATILE_STRIDE(26, os)) {
	       V T1, T7, T2, Tg, Tf, TN, Th, Tq, Ta, Tj, T5, Tr, Tk;
	       T1 = LD(&(xi[0]), ivs, &(xi[0]));
	       {
		    V Td, Te, T8, T9, T3, T4;
		    Td = LD(&(xi[WS(is, 8)]), ivs, &(xi[0]));
		    Te = LD(&(xi[WS(is, 5)]), ivs, &(xi[WS(is, 1)]));
		    T7 = LD(&(xi[WS(is, 12)]), ivs, &(xi[0]));
		    T8 = LD(&(xi[WS(is, 10)]), ivs, &(xi[0]));
		    T9 = LD(&(xi[WS(is, 4)]), ivs, &(xi[0]));
		    T2 = LD(&(xi[WS(is, 1)]), ivs, &(xi[WS(is, 1)]));
		    T3 = LD(&(xi[WS(is, 3)]), ivs, &(xi[WS(is, 1)]));
		    T4 = LD(&(xi[WS(is, 9)]), ivs, &(xi[WS(is, 1)]));
		    Tg = LD(&(xi[WS(is, 11)]), ivs, &(xi[WS(is, 1)]));
		    Tf = VADD(Td, Te);
		    TN = VSUB(Td, Te);
		    Th = LD(&(xi[WS(is, 6)]), ivs, &(xi[0]));
		    Tq = VSUB(T8, T9);
		    Ta = VADD(T8, T9);
		    Tj = LD(&(xi[WS(is, 7)]), ivs, &(xi[WS(is, 1)]));
		    T5 = VADD(T3, T4);
		    Tr = VSUB(T4, T3);
		    Tk = LD(&(xi[WS(is, 2)]), ivs, &(xi[0]));
	       }
	       {
		    V Tt, Ti, Ty, Tb, Ts, TQ, Tx, T6, Tu, Tl;
		    Tt = VSUB(Tg, Th);
		    Ti = VADD(Tg, Th);
		    Ty = VFMS(LDK(KP500000000), Ta, T7);
		    Tb = VADD(T7, Ta);
		    Ts = VSUB(Tq, Tr);
		    TQ = VADD(Tr, Tq);
		    Tx = VFNMS(LDK(KP500000000), T5, T2);
		    T6 = VADD(T2, T5);
		    Tu = VSUB(Tj, Tk);
		    Tl = VADD(Tj, Tk);
		    {
			 V TK, Tz, Tc, TX, Tv, TO, TL, Tm;
			 TK = VADD(Tx, Ty);
			 Tz = VSUB(Tx, Ty);
			 Tc = VADD(T6, Tb);
			 TX = VSUB(T6, Tb);
			 Tv = VSUB(Tt, Tu);
			 TO = VADD(Tt, Tu);
			 TL = VSUB(Ti, Tl);
			 Tm = VADD(Ti, Tl);
			 {
			      V TF, Tw, TP, TY, TT, TM, TA, Tn;
			      TF = VSUB(Ts, Tv);
			      Tw = VADD(Ts, Tv);
			      TP = VFNMS(LDK(KP500000000), TO, TN);
			      TY = VADD(TN, TO);
			      TT = VFNMS(LDK(KP866025403), TL, TK);
			      TM = VFMA(LDK(KP866025403), TL, TK);
			      TA = VFNMS(LDK(KP500000000), Tm, Tf);
			      Tn = VADD(Tf, Tm);
			      {
				   V T1f, T1n, TI, T18, T1k, T1c, TD, T17, T10, T1m, T16, T1e, TU, TR;
				   TU = VFNMS(LDK(KP866025403), TQ, TP);
				   TR = VFMA(LDK(KP866025403), TQ, TP);
				   {
					V TZ, T15, TE, TB;
					TZ = VFMA(LDK(KP302775637), TY, TX);
					T15 = VFNMS(LDK(KP302775637), TX, TY);
					TE = VSUB(Tz, TA);
					TB = VADD(Tz, TA);
					{
					     V TH, To, TV, T13;
					     TH = VSUB(Tc, Tn);
					     To = VADD(Tc, Tn);
					     TV = VFNMS(LDK(KP612264650), TU, TT);
					     T13 = VFMA(LDK(KP612264650), TT, TU);
					     {
						  V TS, T12, TG, T1b;
						  TS = VFNMS(LDK(KP038632954), TR, TM);
						  T12 = VFMA(LDK(KP038632954), TM, TR);
						  TG = VFNMS(LDK(KP514918778), TF, TE);
						  T1b = VFMA(LDK(KP686558370), TE, TF);
						  {
						       V TC, T1a, Tp, TW, T14;
						       TC = VFMA(LDK(KP301479260), TB, Tw);
						       T1a = VFNMS(LDK(KP226109445), Tw, TB);
						       Tp = VFNMS(LDK(KP083333333), To, T1);
						       ST(&(xo[0]), VADD(T1, To), ovs, &(xo[0]));
						       T1f = VFMA(LDK(KP853480001), TV, TS);
						       TW = VFNMS(LDK(KP853480001), TV, TS);
						       T1n = VFMA(LDK(KP853480001), T13, T12);
						       T14 = VFNMS(LDK(KP853480001), T13, T12);
						       TI = VFMA(LDK(KP581704778), TH, TG);
						       T18 = VFNMS(LDK(KP859542535), TG, TH);
						       T1k = VFMA(LDK(KP769338817), T1b, T1a);
						       T1c = VFNMS(LDK(KP769338817), T1b, T1a);
						       TD = VFMA(LDK(KP503537032), TC, Tp);
						       T17 = VFNMS(LDK(KP251768516), TC, Tp);
						       T10 = VMUL(LDK(KP600477271), VFMA(LDK(KP957805992), TZ, TW));
						       T1m = VFNMS(LDK(KP522026385), TW, TZ);
						       T16 = VMUL(LDK(KP600477271), VFMA(LDK(KP957805992), T15, T14));
						       T1e = VFNMS(LDK(KP522026385), T14, T15);
						  }
					     }
					}
				   }
				   {
					V T1o, T1q, T1g, T1i, T1d, T1h, T1l, T1p;
					{
					     V T11, TJ, T19, T1j;
					     T11 = VFMA(LDK(KP516520780), TI, TD);
					     TJ = VFNMS(LDK(KP516520780), TI, TD);
					     T19 = VFMA(LDK(KP300462606), T18, T17);
					     T1j = VFNMS(LDK(KP300462606), T18, T17);
					     T1o = VMUL(LDK(KP575140729), VFNMS(LDK(KP904176221), T1n, T1m));
					     T1q = VMUL(LDK(KP575140729), VFMA(LDK(KP904176221), T1n, T1m));
					     T1g = VMUL(LDK(KP575140729), VFMA(LDK(KP904176221), T1f, T1e));
					     T1i = VMUL(LDK(KP575140729), VFNMS(LDK(KP904176221), T1f, T1e));
					     ST(&(xo[WS(os, 12)]), VFNMSI(T16, T11), ovs, &(xo[0]));
					     ST(&(xo[WS(os, 1)]), VFMAI(T16, T11), ovs, &(xo[WS(os, 1)]));
					     ST(&(xo[WS(os, 8)]), VFMAI(T10, TJ), ovs, &(xo[0]));
					     ST(&(xo[WS(os, 5)]), VFNMSI(T10, TJ), ovs, &(xo[WS(os, 1)]));
					     T1d = VFNMS(LDK(KP503537032), T1c, T19);
					     T1h = VFMA(LDK(KP503537032), T1c, T19);
					     T1l = VFNMS(LDK(KP503537032), T1k, T1j);
					     T1p = VFMA(LDK(KP503537032), T1k, T1j);
					}
					ST(&(xo[WS(os, 9)]), VFMAI(T1g, T1d), ovs, &(xo[WS(os, 1)]));
					ST(&(xo[WS(os, 4)]), VFNMSI(T1g, T1d), ovs, &(xo[0]));
					ST(&(xo[WS(os, 10)]), VFNMSI(T1i, T1h), ovs, &(xo[0]));
					ST(&(xo[WS(os, 3)]), VFMAI(T1i, T1h), ovs, &(xo[WS(os, 1)]));
					ST(&(xo[WS(os, 7)]), VFMAI(T1o, T1l), ovs, &(xo[WS(os, 1)]));
					ST(&(xo[WS(os, 6)]), VFNMSI(T1o, T1l), ovs, &(xo[0]));
					ST(&(xo[WS(os, 11)]), VFMAI(T1q, T1p), ovs, &(xo[WS(os, 1)]));
					ST(&(xo[WS(os, 2)]), VFNMSI(T1q, T1p), ovs, &(xo[0]));
				   }
			      }
			 }
		    }
	       }
	  }
     }
     VLEAVE();
}
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 T1m, T1r, T1i, T1u, T1o, T1v, T1n, T1w, T1s;
	       {
		    E T1k, Tl, T1p, TE, TP, T1g, TM, T1b, T1f, T1a, TU, Tf, T1l, TH, Tw;
		    E T1q;
		    {
			 E TA, T3, TN, Tk, Th, T6, TO, TD, Tb, Tm, Ta, TK, Tp, Tc, Ts;
			 E Tt;
			 {
			      E T4, T5, TB, TC;
			      {
				   E T1, T2, Ti, Tj;
				   T1 = Rp[0];
				   T2 = Rm[WS(rs, 3)];
				   Ti = Ip[0];
				   Tj = Im[WS(rs, 3)];
				   T4 = Rp[WS(rs, 2)];
				   TA = T1 - T2;
				   T3 = T1 + T2;
				   TN = Ti - Tj;
				   Tk = Ti + Tj;
				   T5 = Rm[WS(rs, 1)];
				   TB = Ip[WS(rs, 2)];
				   TC = Im[WS(rs, 1)];
			      }
			      {
				   E T8, T9, Tn, To;
				   T8 = Rp[WS(rs, 1)];
				   Th = T4 - T5;
				   T6 = T4 + T5;
				   TO = TB - TC;
				   TD = TB + TC;
				   T9 = Rm[WS(rs, 2)];
				   Tn = Ip[WS(rs, 1)];
				   To = Im[WS(rs, 2)];
				   Tb = Rm[0];
				   Tm = T8 - T9;
				   Ta = T8 + T9;
				   TK = Tn - To;
				   Tp = Tn + To;
				   Tc = Rp[WS(rs, 3)];
				   Ts = Im[0];
				   Tt = Ip[WS(rs, 3)];
			      }
			 }
			 {
			      E Tr, Td, Tu, TL, Te, T7;
			      T1k = Tk - Th;
			      Tl = Th + Tk;
			      Tr = Tb - Tc;
			      Td = Tb + Tc;
			      TL = Tt - Ts;
			      Tu = Ts + Tt;
			      T1p = TA + TD;
			      TE = TA - TD;
			      TP = TN + TO;
			      T1g = TN - TO;
			      TM = TK + TL;
			      T1b = TL - TK;
			      T1f = Ta - Td;
			      Te = Ta + Td;
			      T1a = T3 - T6;
			      T7 = T3 + T6;
			      {
				   E Tq, TF, TG, Tv;
				   Tq = Tm + Tp;
				   TF = Tm - Tp;
				   TG = Tr - Tu;
				   Tv = Tr + Tu;
				   TU = T7 - Te;
				   Tf = T7 + Te;
				   T1l = TF - TG;
				   TH = TF + TG;
				   Tw = Tq - Tv;
				   T1q = Tq + Tv;
			      }
			 }
		    }
		    {
			 E TX, T10, T1c, T13, T1h, T1E, T1H, T1C, T1K, T1G, T1L, T1F;
			 {
			      E TQ, Tx, T1y, TI, Tg, Tz;
			      TX = TP - TM;
			      TQ = TM + TP;
			      Tx = FMA(KP707106781, Tw, Tl);
			      T10 = FNMS(KP707106781, Tw, Tl);
			      T1c = T1a + T1b;
			      T1y = T1a - T1b;
			      T13 = FNMS(KP707106781, TH, TE);
			      TI = FMA(KP707106781, TH, TE);
			      Tg = W[0];
			      Tz = W[1];
			      {
				   E T1B, T1A, T1x, T1J, T1z, T1D;
				   {
					E TR, Ty, TS, TJ;
					T1B = T1g - T1f;
					T1h = T1f + T1g;
					T1A = W[11];
					TR = Tg * TI;
					Ty = Tg * Tx;
					T1x = W[10];
					T1J = T1A * T1y;
					TS = FNMS(Tz, Tx, TR);
					TJ = FMA(Tz, TI, Ty);
					T1z = T1x * T1y;
					T1m = FMA(KP707106781, T1l, T1k);
					T1E = FNMS(KP707106781, T1l, T1k);
					Im[0] = TS - TQ;
					Ip[0] = TQ + TS;
					Rm[0] = Tf + TJ;
					Rp[0] = Tf - TJ;
					T1H = FMA(KP707106781, T1q, T1p);
					T1r = FNMS(KP707106781, T1q, T1p);
					T1D = W[12];
				   }
				   T1C = FNMS(T1A, T1B, T1z);
				   T1K = FMA(T1x, T1B, T1J);
				   T1G = W[13];
				   T1L = T1D * T1H;
				   T1F = T1D * T1E;
			      }
			 }
			 {
			      E TY, T16, T12, T17, T11;
			      {
				   E TW, TT, T15, TV, TZ, T1M, T1I;
				   TW = W[7];
				   T1M = FNMS(T1G, T1E, T1L);
				   T1I = FMA(T1G, T1H, T1F);
				   TT = W[6];
				   T15 = TW * TU;
				   Im[WS(rs, 3)] = T1M - T1K;
				   Ip[WS(rs, 3)] = T1K + T1M;
				   Rm[WS(rs, 3)] = T1C + T1I;
				   Rp[WS(rs, 3)] = T1C - T1I;
				   TV = TT * TU;
				   TZ = W[8];
				   TY = FNMS(TW, TX, TV);
				   T16 = FMA(TT, TX, T15);
				   T12 = W[9];
				   T17 = TZ * T13;
				   T11 = TZ * T10;
			      }
			      {
				   E T1e, T19, T1t, T1d, T1j, T18, T14;
				   T1e = W[3];
				   T18 = FNMS(T12, T10, T17);
				   T14 = FMA(T12, T13, T11);
				   T19 = W[2];
				   T1t = T1e * T1c;
				   Im[WS(rs, 2)] = T18 - T16;
				   Ip[WS(rs, 2)] = T16 + T18;
				   Rm[WS(rs, 2)] = TY + T14;
				   Rp[WS(rs, 2)] = TY - T14;
				   T1d = T19 * T1c;
				   T1j = W[4];
				   T1i = FNMS(T1e, T1h, T1d);
				   T1u = FMA(T19, T1h, T1t);
				   T1o = W[5];
				   T1v = T1j * T1r;
				   T1n = T1j * T1m;
			      }
			 }
		    }
	       }
	       T1w = FNMS(T1o, T1m, T1v);
	       T1s = FMA(T1o, T1r, T1n);
	       Im[WS(rs, 1)] = T1w - T1u;
	       Ip[WS(rs, 1)] = T1u + T1w;
	       Rm[WS(rs, 1)] = T1i + T1s;
	       Rp[WS(rs, 1)] = T1i - T1s;
	  }
     }
}
Exemple #20
0
static const R *hb_10(R *rio, R *iio, const R *W, stride ios, int m, int dist)
{
     DK(KP250000000, +0.250000000000000000000000000000000000000000000);
     DK(KP951056516, +0.951056516295153572116439333379382143405698634);
     DK(KP587785252, +0.587785252292473129168705954639072768597652438);
     DK(KP559016994, +0.559016994374947424102293417182819058860154590);
     int i;
     for (i = m - 2; i > 0; i = i - 2, rio = rio + dist, iio = iio - dist, W = W + 18) {
	  E T3, Tk, Tw, T1w, TO, TP, T1E, T1D, Tr, TX, Ti, T1l, TZ, T10, T1s;
	  E T1p, T1z, T1B, TL, TS;
	  {
	       E T1, T2, Tu, Tv;
	       T1 = rio[0];
	       T2 = iio[-WS(ios, 5)];
	       T3 = T1 + T2;
	       Tk = T1 - T2;
	       Tu = iio[0];
	       Tv = rio[WS(ios, 5)];
	       Tw = Tu + Tv;
	       T1w = Tu - Tv;
	  }
	  {
	       E T6, Tl, Tg, Tp, T9, Tm, Td, To;
	       {
		    E T4, T5, Te, Tf;
		    T4 = rio[WS(ios, 2)];
		    T5 = iio[-WS(ios, 7)];
		    T6 = T4 + T5;
		    Tl = T4 - T5;
		    Te = iio[-WS(ios, 6)];
		    Tf = rio[WS(ios, 1)];
		    Tg = Te + Tf;
		    Tp = Te - Tf;
	       }
	       {
		    E T7, T8, Tb, Tc;
		    T7 = iio[-WS(ios, 8)];
		    T8 = rio[WS(ios, 3)];
		    T9 = T7 + T8;
		    Tm = T7 - T8;
		    Tb = rio[WS(ios, 4)];
		    Tc = iio[-WS(ios, 9)];
		    Td = Tb + Tc;
		    To = Tb - Tc;
	       }
	       TO = Tl - Tm;
	       TP = To - Tp;
	       T1E = Td - Tg;
	       T1D = T6 - T9;
	       {
		    E Tn, Tq, Ta, Th;
		    Tn = Tl + Tm;
		    Tq = To + Tp;
		    Tr = Tn + Tq;
		    TX = KP559016994 * (Tn - Tq);
		    Ta = T6 + T9;
		    Th = Td + Tg;
		    Ti = Ta + Th;
		    T1l = KP559016994 * (Ta - Th);
	       }
	  }
	  {
	       E Tz, T1n, TJ, T1r, TC, T1o, TG, T1q;
	       {
		    E Tx, Ty, TH, TI;
		    Tx = iio[-WS(ios, 2)];
		    Ty = rio[WS(ios, 7)];
		    Tz = Tx + Ty;
		    T1n = Tx - Ty;
		    TH = rio[WS(ios, 6)];
		    TI = iio[-WS(ios, 1)];
		    TJ = TH + TI;
		    T1r = TI - TH;
	       }
	       {
		    E TA, TB, TE, TF;
		    TA = rio[WS(ios, 8)];
		    TB = iio[-WS(ios, 3)];
		    TC = TA + TB;
		    T1o = TB - TA;
		    TE = iio[-WS(ios, 4)];
		    TF = rio[WS(ios, 9)];
		    TG = TE + TF;
		    T1q = TE - TF;
	       }
	       TZ = Tz + TC;
	       T10 = TG + TJ;
	       T1s = T1q - T1r;
	       T1p = T1n - T1o;
	       {
		    E T1x, T1y, TD, TK;
		    T1x = T1n + T1o;
		    T1y = T1q + T1r;
		    T1z = T1x + T1y;
		    T1B = KP559016994 * (T1x - T1y);
		    TD = Tz - TC;
		    TK = TG - TJ;
		    TL = TD + TK;
		    TS = KP559016994 * (TD - TK);
	       }
	  }
	  rio[0] = T3 + Ti;
	  iio[-WS(ios, 9)] = T1w + T1z;
	  {
	       E Ts, TM, Tj, Tt;
	       Ts = Tk + Tr;
	       TM = Tw + TL;
	       Tj = W[8];
	       Tt = W[9];
	       rio[WS(ios, 5)] = FNMS(Tt, TM, Tj * Ts);
	       iio[-WS(ios, 4)] = FMA(Tt, Ts, Tj * TM);
	  }
	  {
	       E T1t, T1F, T1Q, T1N, T1C, T1R, T1m, T1M, T1A, T1k;
	       T1t = FNMS(KP951056516, T1s, KP587785252 * T1p);
	       T1F = FNMS(KP951056516, T1E, KP587785252 * T1D);
	       T1Q = FMA(KP951056516, T1D, KP587785252 * T1E);
	       T1N = FMA(KP951056516, T1p, KP587785252 * T1s);
	       T1A = FNMS(KP250000000, T1z, T1w);
	       T1C = T1A - T1B;
	       T1R = T1B + T1A;
	       T1k = FNMS(KP250000000, Ti, T3);
	       T1m = T1k - T1l;
	       T1M = T1l + T1k;
	       {
		    E T1u, T1G, T1j, T1v;
		    T1u = T1m + T1t;
		    T1G = T1C - T1F;
		    T1j = W[14];
		    T1v = W[15];
		    rio[WS(ios, 8)] = FNMS(T1v, T1G, T1j * T1u);
		    iio[-WS(ios, 1)] = FMA(T1v, T1u, T1j * T1G);
	       }
	       {
		    E T1U, T1W, T1T, T1V;
		    T1U = T1M + T1N;
		    T1W = T1R - T1Q;
		    T1T = W[6];
		    T1V = W[7];
		    rio[WS(ios, 4)] = FNMS(T1V, T1W, T1T * T1U);
		    iio[-WS(ios, 5)] = FMA(T1V, T1U, T1T * T1W);
	       }
	       {
		    E T1I, T1K, T1H, T1J;
		    T1I = T1m - T1t;
		    T1K = T1F + T1C;
		    T1H = W[2];
		    T1J = W[3];
		    rio[WS(ios, 2)] = FNMS(T1J, T1K, T1H * T1I);
		    iio[-WS(ios, 7)] = FMA(T1J, T1I, T1H * T1K);
	       }
	       {
		    E T1O, T1S, T1L, T1P;
		    T1O = T1M - T1N;
		    T1S = T1Q + T1R;
		    T1L = W[10];
		    T1P = W[11];
		    rio[WS(ios, 6)] = FNMS(T1P, T1S, T1L * T1O);
		    iio[-WS(ios, 3)] = FMA(T1P, T1O, T1L * T1S);
	       }
	  }
	  {
	       E TQ, T11, T1c, T19, TY, T18, TT, T1d, TW, TR;
	       TQ = FNMS(KP951056516, TP, KP587785252 * TO);
	       T11 = FNMS(KP951056516, T10, KP587785252 * TZ);
	       T1c = FMA(KP951056516, TO, KP587785252 * TP);
	       T19 = FMA(KP951056516, TZ, KP587785252 * T10);
	       TW = FNMS(KP250000000, Tr, Tk);
	       TY = TW - TX;
	       T18 = TX + TW;
	       TR = FNMS(KP250000000, TL, Tw);
	       TT = TR - TS;
	       T1d = TS + TR;
	       {
		    E TU, T12, TN, TV;
		    TU = TQ + TT;
		    T12 = TY - T11;
		    TN = W[12];
		    TV = W[13];
		    iio[-WS(ios, 2)] = FMA(TN, TU, TV * T12);
		    rio[WS(ios, 7)] = FNMS(TV, TU, TN * T12);
	       }
	       {
		    E T1g, T1i, T1f, T1h;
		    T1g = T1d - T1c;
		    T1i = T18 + T19;
		    T1f = W[16];
		    T1h = W[17];
		    iio[0] = FMA(T1f, T1g, T1h * T1i);
		    rio[WS(ios, 9)] = FNMS(T1h, T1g, T1f * T1i);
	       }
	       {
		    E T14, T16, T13, T15;
		    T14 = TY + T11;
		    T16 = TT - TQ;
		    T13 = W[4];
		    T15 = W[5];
		    rio[WS(ios, 3)] = FNMS(T15, T16, T13 * T14);
		    iio[-WS(ios, 6)] = FMA(T13, T16, T15 * T14);
	       }
	       {
		    E T1a, T1e, T17, T1b;
		    T1a = T18 - T19;
		    T1e = T1c + T1d;
		    T17 = W[0];
		    T1b = W[1];
		    rio[WS(ios, 1)] = FNMS(T1b, T1e, T17 * T1a);
		    iio[-WS(ios, 8)] = FMA(T17, T1e, T1b * T1a);
	       }
	  }
     }
     return W;
}
Exemple #21
0
static void q1_2(float *rio, float *iio, const float *W, stride rs, stride vs, INT mb, INT me, INT ms)
{
     INT m;
     for (m = mb, W = W + (mb * 2); m < me; m = m + 1, rio = rio + ms, iio = iio + ms, W = W + 2, MAKE_VOLATILE_STRIDE(rs), MAKE_VOLATILE_STRIDE(vs)) {
	  E T1, T2, T4, T6, T7, T8, T9, Ta, Tc, Te, Tf, Tg;
	  T1 = rio[0];
	  T2 = rio[WS(rs, 1)];
	  T4 = T1 - T2;
	  T6 = iio[0];
	  T7 = iio[WS(rs, 1)];
	  T8 = T6 - T7;
	  T9 = rio[WS(vs, 1)];
	  Ta = rio[WS(vs, 1) + WS(rs, 1)];
	  Tc = T9 - Ta;
	  Te = iio[WS(vs, 1)];
	  Tf = iio[WS(vs, 1) + WS(rs, 1)];
	  Tg = Te - Tf;
	  rio[0] = T1 + T2;
	  iio[0] = T6 + T7;
	  rio[WS(rs, 1)] = T9 + Ta;
	  iio[WS(rs, 1)] = Te + Tf;
	  {
	       E Tb, Td, T3, T5;
	       Tb = W[0];
	       Td = W[1];
	       rio[WS(vs, 1) + WS(rs, 1)] = FMA(Tb, Tc, Td * Tg);
	       iio[WS(vs, 1) + WS(rs, 1)] = FNMS(Td, Tc, Tb * Tg);
	       T3 = W[0];
	       T5 = W[1];
	       rio[WS(vs, 1)] = FMA(T3, T4, T5 * T8);
	       iio[WS(vs, 1)] = FNMS(T5, T4, T3 * T8);
	  }
     }
}
Exemple #22
0
static void hb_20(R *cr, R *ci, const R *W, stride rs, INT mb, INT me, INT ms)
{
     DK(KP951056516, +0.951056516295153572116439333379382143405698634);
     DK(KP559016994, +0.559016994374947424102293417182819058860154590);
     DK(KP618033988, +0.618033988749894848204586834365638117720309180);
     DK(KP250000000, +0.250000000000000000000000000000000000000000000);
     {
	  INT m;
	  for (m = mb, W = W + ((mb - 1) * 38); m < me; m = m + 1, cr = cr + ms, ci = ci - ms, W = W + 38, MAKE_VOLATILE_STRIDE(rs)) {
	       E T1T, T1Q, T1P;
	       {
		    E T2W, T4e, T7, TE, T3z, T4z, T1t, T2l, T3a, T3G, T13, T33, T3H, T1i, T2g;
		    E T4H, T4G, T2d, T1B, T4u, T4B, T4A, T4r, T1A, T2s, T3l, T2t, T3s, T2o, T2q;
		    E T1w, T1y, TC, T29, T3E, T3C, T4n, T4l, TN, TL;
		    {
			 E T4, T2U, T3, T2V, T1s, T5, T1n, T1o;
			 {
			      E T1, T2, T1q, T1r;
			      T1 = cr[0];
			      T2 = ci[WS(rs, 9)];
			      T1q = ci[WS(rs, 14)];
			      T1r = cr[WS(rs, 15)];
			      T4 = cr[WS(rs, 5)];
			      T2U = T1 - T2;
			      T3 = T1 + T2;
			      T2V = T1q + T1r;
			      T1s = T1q - T1r;
			      T5 = ci[WS(rs, 4)];
			      T1n = ci[WS(rs, 19)];
			      T1o = cr[WS(rs, 10)];
			 }
			 {
			      E T3y, T6, T3x, T1p;
			      T2W = T2U + T2V;
			      T4e = T2U - T2V;
			      T3y = T4 - T5;
			      T6 = T4 + T5;
			      T3x = T1n + T1o;
			      T1p = T1n - T1o;
			      T7 = T3 + T6;
			      TE = T3 - T6;
			      T3z = T3x - T3y;
			      T4z = T3y + T3x;
			      T1t = T1p - T1s;
			      T2l = T1p + T1s;
			 }
		    }
		    {
			 E T2Z, T4f, Te, TF, T3o, T4p, T1a, T2b, TJ, TA, T4t, T3k, T4j, T39, T2f;
			 E T12, T32, T4g, Tl, TG, T3r, T4q, T1h, T2c, T36, T4i, Tt, TI, T3h, T4s;
			 E TV, T2e;
			 {
			      E Tb, T2X, Ta, T2Y, T19, Tc, T14, T15;
			      {
				   E T8, T9, T17, T18;
				   T8 = cr[WS(rs, 4)];
				   T9 = ci[WS(rs, 5)];
				   T17 = ci[WS(rs, 10)];
				   T18 = cr[WS(rs, 19)];
				   Tb = cr[WS(rs, 9)];
				   T2X = T8 - T9;
				   Ta = T8 + T9;
				   T2Y = T17 + T18;
				   T19 = T17 - T18;
				   Tc = ci[0];
				   T14 = ci[WS(rs, 15)];
				   T15 = cr[WS(rs, 14)];
			      }
			      {
				   E T3n, Td, T3m, T16;
				   T2Z = T2X + T2Y;
				   T4f = T2X - T2Y;
				   T3n = Tb - Tc;
				   Td = Tb + Tc;
				   T3m = T14 + T15;
				   T16 = T14 - T15;
				   Te = Ta + Td;
				   TF = Ta - Td;
				   T3o = T3m - T3n;
				   T4p = T3n + T3m;
				   T1a = T16 - T19;
				   T2b = T16 + T19;
			      }
			 }
			 {
			      E TW, T37, Tw, T3i, Tz, TX, TZ, T10;
			      {
				   E Tu, Tv, Tx, Ty;
				   Tu = ci[WS(rs, 7)];
				   Tv = cr[WS(rs, 2)];
				   Tx = ci[WS(rs, 2)];
				   Ty = cr[WS(rs, 7)];
				   TW = ci[WS(rs, 17)];
				   T37 = Tu - Tv;
				   Tw = Tu + Tv;
				   T3i = Tx - Ty;
				   Tz = Tx + Ty;
				   TX = cr[WS(rs, 12)];
				   TZ = ci[WS(rs, 12)];
				   T10 = cr[WS(rs, 17)];
			      }
			      {
				   E TY, T38, T11, T3j;
				   TJ = Tw - Tz;
				   TA = Tw + Tz;
				   T3j = TW + TX;
				   TY = TW - TX;
				   T38 = TZ + T10;
				   T11 = TZ - T10;
				   T4t = T3i - T3j;
				   T3k = T3i + T3j;
				   T4j = T37 + T38;
				   T39 = T37 - T38;
				   T2f = TY + T11;
				   T12 = TY - T11;
			      }
			 }
			 {
			      E Ti, T30, Th, T31, T1g, Tj, T1b, T1c;
			      {
				   E Tf, Tg, T1e, T1f;
				   Tf = ci[WS(rs, 3)];
				   Tg = cr[WS(rs, 6)];
				   T1e = ci[WS(rs, 18)];
				   T1f = cr[WS(rs, 11)];
				   Ti = cr[WS(rs, 1)];
				   T30 = Tf - Tg;
				   Th = Tf + Tg;
				   T31 = T1e + T1f;
				   T1g = T1e - T1f;
				   Tj = ci[WS(rs, 8)];
				   T1b = ci[WS(rs, 13)];
				   T1c = cr[WS(rs, 16)];
			      }
			      {
				   E T3p, Tk, T3q, T1d;
				   T32 = T30 + T31;
				   T4g = T30 - T31;
				   T3p = Ti - Tj;
				   Tk = Ti + Tj;
				   T3q = T1b + T1c;
				   T1d = T1b - T1c;
				   Tl = Th + Tk;
				   TG = Th - Tk;
				   T3r = T3p + T3q;
				   T4q = T3p - T3q;
				   T1h = T1d - T1g;
				   T2c = T1d + T1g;
			      }
			 }
			 {
			      E Tq, T34, Tp, T35, TU, Tr, TP, TQ;
			      {
				   E Tn, To, TS, TT;
				   Tn = cr[WS(rs, 8)];
				   To = ci[WS(rs, 1)];
				   TS = ci[WS(rs, 16)];
				   TT = cr[WS(rs, 13)];
				   Tq = ci[WS(rs, 6)];
				   T34 = Tn - To;
				   Tp = Tn + To;
				   T35 = TS + TT;
				   TU = TS - TT;
				   Tr = cr[WS(rs, 3)];
				   TP = ci[WS(rs, 11)];
				   TQ = cr[WS(rs, 18)];
			      }
			      {
				   E T3g, Ts, T3f, TR;
				   T36 = T34 - T35;
				   T4i = T34 + T35;
				   T3g = Tq - Tr;
				   Ts = Tq + Tr;
				   T3f = TP + TQ;
				   TR = TP - TQ;
				   Tt = Tp + Ts;
				   TI = Tp - Ts;
				   T3h = T3f - T3g;
				   T4s = T3g + T3f;
				   TV = TR - TU;
				   T2e = TR + TU;
			      }
			 }
			 {
			      E T1v, T1u, T2n, T4k, T4h, T2m, TH, TK;
			      T3a = T36 + T39;
			      T3G = T36 - T39;
			      T13 = TV - T12;
			      T1v = TV + T12;
			      T33 = T2Z + T32;
			      T3H = T2Z - T32;
			      T1i = T1a - T1h;
			      T1u = T1a + T1h;
			      T2n = T2e + T2f;
			      T2g = T2e - T2f;
			      T4H = T4i - T4j;
			      T4k = T4i + T4j;
			      T4h = T4f + T4g;
			      T4G = T4f - T4g;
			      T2d = T2b - T2c;
			      T2m = T2b + T2c;
			      TH = TF + TG;
			      T1B = TF - TG;
			      T4u = T4s - T4t;
			      T4B = T4s + T4t;
			      T4A = T4p + T4q;
			      T4r = T4p - T4q;
			      T1A = TI - TJ;
			      TK = TI + TJ;
			      {
				   E Tm, T3B, TB, T3A;
				   Tm = Te + Tl;
				   T2s = Te - Tl;
				   T3l = T3h + T3k;
				   T3B = T3h - T3k;
				   TB = Tt + TA;
				   T2t = Tt - TA;
				   T3s = T3o + T3r;
				   T3A = T3o - T3r;
				   T2o = T2m + T2n;
				   T2q = T2m - T2n;
				   T1w = T1u + T1v;
				   T1y = T1u - T1v;
				   TC = Tm + TB;
				   T29 = Tm - TB;
				   T3E = T3A - T3B;
				   T3C = T3A + T3B;
				   T4n = T4h - T4k;
				   T4l = T4h + T4k;
				   TN = TH - TK;
				   TL = TH + TK;
			      }
			 }
		    }
		    {
			 E T3d, T3b, T4E, T1x, TM, T4m, T58, T5b, T4D, T5a, T5c, T59, T4C;
			 cr[0] = T7 + TC;
			 T3d = T33 - T3a;
			 T3b = T33 + T3a;
			 T4E = T4A - T4B;
			 T4C = T4A + T4B;
			 ci[0] = T2l + T2o;
			 {
			      E T25, T22, T21, T24, T23, T26, T57;
			      T1x = FNMS(KP250000000, T1w, T1t);
			      T25 = T1t + T1w;
			      T22 = TE + TL;
			      TM = FNMS(KP250000000, TL, TE);
			      T21 = W[18];
			      T24 = W[19];
			      T4m = FNMS(KP250000000, T4l, T4e);
			      T58 = T4e + T4l;
			      T5b = T4z + T4C;
			      T4D = FNMS(KP250000000, T4C, T4z);
			      T23 = T21 * T22;
			      T26 = T24 * T22;
			      T57 = W[8];
			      T5a = W[9];
			      cr[WS(rs, 10)] = FNMS(T24, T25, T23);
			      ci[WS(rs, 10)] = FMA(T21, T25, T26);
			      T5c = T57 * T5b;
			      T59 = T57 * T58;
			 }
			 {
			      E T3U, T3Z, T3W, T40, T3V;
			      {
				   E T3c, T48, T4b, T3D, T47, T4a;
				   T3c = FNMS(KP250000000, T3b, T2W);
				   T48 = T2W + T3b;
				   T4b = T3z + T3C;
				   T3D = FNMS(KP250000000, T3C, T3z);
				   ci[WS(rs, 5)] = FMA(T5a, T58, T5c);
				   cr[WS(rs, 5)] = FNMS(T5a, T5b, T59);
				   T47 = W[28];
				   T4a = W[29];
				   {
					E T3I, T3Y, T42, T3u, T3M, T3X, T3F;
					{
					     E T3T, T3t, T4c, T49, T3e, T3S;
					     T3T = FMA(KP618033988, T3l, T3s);
					     T3t = FNMS(KP618033988, T3s, T3l);
					     T4c = T47 * T4b;
					     T49 = T47 * T48;
					     T3I = FNMS(KP618033988, T3H, T3G);
					     T3Y = FMA(KP618033988, T3G, T3H);
					     ci[WS(rs, 15)] = FMA(T4a, T48, T4c);
					     cr[WS(rs, 15)] = FNMS(T4a, T4b, T49);
					     T3e = FNMS(KP559016994, T3d, T3c);
					     T3S = FMA(KP559016994, T3d, T3c);
					     T42 = FMA(KP951056516, T3T, T3S);
					     T3U = FNMS(KP951056516, T3T, T3S);
					     T3u = FNMS(KP951056516, T3t, T3e);
					     T3M = FMA(KP951056516, T3t, T3e);
					     T3X = FMA(KP559016994, T3E, T3D);
					     T3F = FNMS(KP559016994, T3E, T3D);
					}
					{
					     E T3P, T45, T44, T46, T43;
					     {
						  E T3w, T3J, T3v, T3K, T2T, T41;
						  T2T = W[4];
						  T3w = W[5];
						  T3J = FMA(KP951056516, T3I, T3F);
						  T3P = FNMS(KP951056516, T3I, T3F);
						  T45 = FNMS(KP951056516, T3Y, T3X);
						  T3Z = FMA(KP951056516, T3Y, T3X);
						  T3v = T2T * T3u;
						  T3K = T2T * T3J;
						  T41 = W[36];
						  T44 = W[37];
						  cr[WS(rs, 3)] = FNMS(T3w, T3J, T3v);
						  ci[WS(rs, 3)] = FMA(T3w, T3u, T3K);
						  T46 = T41 * T45;
						  T43 = T41 * T42;
					     }
					     {
						  E T3O, T3Q, T3N, T3L, T3R;
						  T3L = W[12];
						  T3O = W[13];
						  ci[WS(rs, 19)] = FMA(T44, T42, T46);
						  cr[WS(rs, 19)] = FNMS(T44, T45, T43);
						  T3Q = T3L * T3P;
						  T3N = T3L * T3M;
						  T3R = W[20];
						  T3W = W[21];
						  ci[WS(rs, 7)] = FMA(T3O, T3M, T3Q);
						  cr[WS(rs, 7)] = FNMS(T3O, T3P, T3N);
						  T40 = T3R * T3Z;
						  T3V = T3R * T3U;
					     }
					}
				   }
			      }
			      {
				   E T4U, T4Z, T4W, T50, T4V, T2L, T2I, T2H;
				   {
					E T4T, T4v, T4I, T4Y, T4o, T4S;
					T4T = FNMS(KP618033988, T4r, T4u);
					T4v = FMA(KP618033988, T4u, T4r);
					ci[WS(rs, 11)] = FMA(T3W, T3U, T40);
					cr[WS(rs, 11)] = FNMS(T3W, T3Z, T3V);
					T4I = FMA(KP618033988, T4H, T4G);
					T4Y = FNMS(KP618033988, T4G, T4H);
					T4o = FMA(KP559016994, T4n, T4m);
					T4S = FNMS(KP559016994, T4n, T4m);
					{
					     E T52, T4M, T55, T4P, T54, T56, T53;
					     {
						  E T4d, T4w, T4J, T4x, T4y, T4X, T4F, T51, T4K;
						  T4d = W[0];
						  T4X = FNMS(KP559016994, T4E, T4D);
						  T4F = FMA(KP559016994, T4E, T4D);
						  T4U = FNMS(KP951056516, T4T, T4S);
						  T52 = FMA(KP951056516, T4T, T4S);
						  T4M = FMA(KP951056516, T4v, T4o);
						  T4w = FNMS(KP951056516, T4v, T4o);
						  T4Z = FMA(KP951056516, T4Y, T4X);
						  T55 = FNMS(KP951056516, T4Y, T4X);
						  T4P = FNMS(KP951056516, T4I, T4F);
						  T4J = FMA(KP951056516, T4I, T4F);
						  T4x = T4d * T4w;
						  T4y = W[1];
						  T51 = W[32];
						  T4K = T4d * T4J;
						  T54 = W[33];
						  cr[WS(rs, 1)] = FNMS(T4y, T4J, T4x);
						  T56 = T51 * T55;
						  T53 = T51 * T52;
						  ci[WS(rs, 1)] = FMA(T4y, T4w, T4K);
					     }
					     {
						  E T4O, T4Q, T4N, T4L, T4R;
						  T4L = W[16];
						  ci[WS(rs, 17)] = FMA(T54, T52, T56);
						  cr[WS(rs, 17)] = FNMS(T54, T55, T53);
						  T4O = W[17];
						  T4Q = T4L * T4P;
						  T4N = T4L * T4M;
						  T4R = W[24];
						  T4W = W[25];
						  ci[WS(rs, 9)] = FMA(T4O, T4M, T4Q);
						  cr[WS(rs, 9)] = FNMS(T4O, T4P, T4N);
						  T50 = T4R * T4Z;
						  T4V = T4R * T4U;
					     }
					}
				   }
				   {
					E T2K, T2u, T2F, T2h, T28, T2J, T2r, T2p;
					T2K = FNMS(KP618033988, T2s, T2t);
					T2u = FMA(KP618033988, T2t, T2s);
					ci[WS(rs, 13)] = FMA(T4W, T4U, T50);
					cr[WS(rs, 13)] = FNMS(T4W, T4Z, T4V);
					T2p = FNMS(KP250000000, T2o, T2l);
					T2F = FNMS(KP618033988, T2d, T2g);
					T2h = FMA(KP618033988, T2g, T2d);
					T28 = FNMS(KP250000000, TC, T7);
					T2J = FNMS(KP559016994, T2q, T2p);
					T2r = FMA(KP559016994, T2q, T2p);
					{
					     E T2B, T2G, T2y, T2R, T2Q, T2P, T2A, T2x;
					     {
						  E T2k, T2v, T27, T2O, T2i, T2a, T2E;
						  T2k = W[7];
						  T2a = FMA(KP559016994, T29, T28);
						  T2E = FNMS(KP559016994, T29, T28);
						  T2B = FMA(KP951056516, T2u, T2r);
						  T2v = FNMS(KP951056516, T2u, T2r);
						  T27 = W[6];
						  T2O = FMA(KP951056516, T2F, T2E);
						  T2G = FNMS(KP951056516, T2F, T2E);
						  T2i = FMA(KP951056516, T2h, T2a);
						  T2y = FNMS(KP951056516, T2h, T2a);
						  {
						       E T2N, T2j, T2w, T2S;
						       T2L = FMA(KP951056516, T2K, T2J);
						       T2R = FNMS(KP951056516, T2K, T2J);
						       T2Q = W[23];
						       T2N = W[22];
						       T2j = T27 * T2i;
						       T2w = T2k * T2i;
						       T2S = T2Q * T2O;
						       T2P = T2N * T2O;
						       cr[WS(rs, 4)] = FNMS(T2k, T2v, T2j);
						       ci[WS(rs, 4)] = FMA(T27, T2v, T2w);
						       ci[WS(rs, 12)] = FMA(T2N, T2R, T2S);
						  }
					     }
					     cr[WS(rs, 12)] = FNMS(T2Q, T2R, T2P);
					     T2A = W[31];
					     T2x = W[30];
					     {
						  E T2D, T2M, T2C, T2z;
						  T2I = W[15];
						  T2C = T2A * T2y;
						  T2z = T2x * T2y;
						  T2D = W[14];
						  T2M = T2I * T2G;
						  ci[WS(rs, 16)] = FMA(T2x, T2B, T2C);
						  cr[WS(rs, 16)] = FNMS(T2A, T2B, T2z);
						  T2H = T2D * T2G;
						  ci[WS(rs, 8)] = FMA(T2D, T2L, T2M);
					     }
					}
				   }
				   {
					E T1S, T1C, T1j, T1N, T1z, T1R;
					T1S = FMA(KP618033988, T1A, T1B);
					T1C = FNMS(KP618033988, T1B, T1A);
					cr[WS(rs, 8)] = FNMS(T2I, T2L, T2H);
					T1j = FNMS(KP618033988, T1i, T13);
					T1N = FMA(KP618033988, T13, T1i);
					T1z = FNMS(KP559016994, T1y, T1x);
					T1R = FMA(KP559016994, T1y, T1x);
					{
					     E T1J, T1O, T1G, T1Z, T1Y, T1X, T1I, T1F;
					     {
						  E T1m, T1D, TD, T1W, T1k, T1M, TO;
						  T1m = W[3];
						  T1M = FMA(KP559016994, TN, TM);
						  TO = FNMS(KP559016994, TN, TM);
						  T1D = FNMS(KP951056516, T1C, T1z);
						  T1J = FMA(KP951056516, T1C, T1z);
						  TD = W[2];
						  T1O = FNMS(KP951056516, T1N, T1M);
						  T1W = FMA(KP951056516, T1N, T1M);
						  T1G = FNMS(KP951056516, T1j, TO);
						  T1k = FMA(KP951056516, T1j, TO);
						  {
						       E T1V, T1l, T1E, T20;
						       T1Z = FNMS(KP951056516, T1S, T1R);
						       T1T = FMA(KP951056516, T1S, T1R);
						       T1Y = W[27];
						       T1V = W[26];
						       T1l = TD * T1k;
						       T1E = T1m * T1k;
						       T20 = T1Y * T1W;
						       T1X = T1V * T1W;
						       cr[WS(rs, 2)] = FNMS(T1m, T1D, T1l);
						       ci[WS(rs, 2)] = FMA(TD, T1D, T1E);
						       ci[WS(rs, 14)] = FMA(T1V, T1Z, T20);
						  }
					     }
					     cr[WS(rs, 14)] = FNMS(T1Y, T1Z, T1X);
					     T1I = W[35];
					     T1F = W[34];
					     {
						  E T1L, T1U, T1K, T1H;
						  T1Q = W[11];
						  T1K = T1I * T1G;
						  T1H = T1F * T1G;
						  T1L = W[10];
						  T1U = T1Q * T1O;
						  ci[WS(rs, 18)] = FMA(T1F, T1J, T1K);
						  cr[WS(rs, 18)] = FNMS(T1I, T1J, T1H);
						  T1P = T1L * T1O;
						  ci[WS(rs, 6)] = FMA(T1L, T1T, T1U);
					     }
					}
				   }
			      }
			 }
		    }
	       }
	       cr[WS(rs, 6)] = FNMS(T1Q, T1T, T1P);
	  }
     }
}
Exemple #23
0
static const R *hb_6(R *rio, R *iio, const R *W, stride ios, INT m, INT dist)
{
     DK(KP500000000, +0.500000000000000000000000000000000000000000000);
     DK(KP866025403, +0.866025403784438646763723170752936183471402627);
     INT i;
     for (i = m - 2; i > 0; i = i - 2, rio = rio + dist, iio = iio - dist, W = W + 10, MAKE_VOLATILE_STRIDE(ios)) {
	  E T3, Ty, Tp, TE, Ta, TO, Tm, TB, Tj, TL, Tq, TH;
	  {
	       E T1, T2, Tn, To;
	       T1 = rio[0];
	       T2 = iio[-WS(ios, 3)];
	       T3 = T1 + T2;
	       Ty = T1 - T2;
	       Tn = iio[0];
	       To = rio[WS(ios, 3)];
	       Tp = Tn - To;
	       TE = Tn + To;
	  }
	  {
	       E T6, Tz, T9, TA;
	       {
		    E T4, T5, T7, T8;
		    T4 = rio[WS(ios, 2)];
		    T5 = iio[-WS(ios, 5)];
		    T6 = T4 + T5;
		    Tz = T4 - T5;
		    T7 = iio[-WS(ios, 4)];
		    T8 = rio[WS(ios, 1)];
		    T9 = T7 + T8;
		    TA = T7 - T8;
	       }
	       Ta = T6 + T9;
	       TO = KP866025403 * (Tz - TA);
	       Tm = KP866025403 * (T6 - T9);
	       TB = Tz + TA;
	  }
	  {
	       E Tf, TF, Ti, TG;
	       {
		    E Td, Te, Tg, Th;
		    Td = iio[-WS(ios, 1)];
		    Te = rio[WS(ios, 4)];
		    Tf = Td - Te;
		    TF = Te + Td;
		    Tg = iio[-WS(ios, 2)];
		    Th = rio[WS(ios, 5)];
		    Ti = Tg - Th;
		    TG = Tg + Th;
	       }
	       Tj = KP866025403 * (Tf - Ti);
	       TL = KP866025403 * (TF + TG);
	       Tq = Tf + Ti;
	       TH = TF - TG;
	  }
	  rio[0] = T3 + Ta;
	  iio[-WS(ios, 5)] = Tp + Tq;
	  {
	       E TC, TI, Tx, TD;
	       TC = Ty + TB;
	       TI = TE - TH;
	       Tx = W[4];
	       TD = W[5];
	       rio[WS(ios, 3)] = FNMS(TD, TI, Tx * TC);
	       iio[-WS(ios, 2)] = FMA(TD, TC, Tx * TI);
	  }
	  {
	       E Tk, Tu, Ts, Tw, Tc, Tr;
	       Tc = FNMS(KP500000000, Ta, T3);
	       Tk = Tc + Tj;
	       Tu = Tc - Tj;
	       Tr = FNMS(KP500000000, Tq, Tp);
	       Ts = Tm + Tr;
	       Tw = Tr - Tm;
	       {
		    E Tb, Tl, Tt, Tv;
		    Tb = W[6];
		    Tl = W[7];
		    rio[WS(ios, 4)] = FNMS(Tl, Ts, Tb * Tk);
		    iio[-WS(ios, 1)] = FMA(Tl, Tk, Tb * Ts);
		    Tt = W[2];
		    Tv = W[3];
		    rio[WS(ios, 2)] = FNMS(Tv, Tw, Tt * Tu);
		    iio[-WS(ios, 3)] = FMA(Tv, Tu, Tt * Tw);
	       }
	  }
	  {
	       E TM, TU, TQ, TS, TK, TP;
	       TK = FNMS(KP500000000, TB, Ty);
	       TM = TK - TL;
	       TU = TK + TL;
	       TP = FMA(KP500000000, TH, TE);
	       TQ = TO + TP;
	       TS = TP - TO;
	       {
		    E TJ, TN, TR, TT;
		    TJ = W[0];
		    TN = W[1];
		    rio[WS(ios, 1)] = FNMS(TN, TQ, TJ * TM);
		    iio[-WS(ios, 4)] = FMA(TJ, TQ, TN * TM);
		    TR = W[8];
		    TT = W[9];
		    iio[0] = FMA(TR, TS, TT * TU);
		    rio[WS(ios, 5)] = FNMS(TT, TS, TR * TU);
	       }
	  }
     }
     return W;
}
Exemple #24
0
static void hb_20(R *cr, R *ci, const R *W, stride rs, INT mb, INT me, INT ms)
{
     DK(KP250000000, +0.250000000000000000000000000000000000000000000);
     DK(KP559016994, +0.559016994374947424102293417182819058860154590);
     DK(KP587785252, +0.587785252292473129168705954639072768597652438);
     DK(KP951056516, +0.951056516295153572116439333379382143405698634);
     {
	  INT m;
	  for (m = mb, W = W + ((mb - 1) * 38); m < me; m = m + 1, cr = cr + ms, ci = ci - ms, W = W + 38, MAKE_VOLATILE_STRIDE(rs)) {
	       E T7, T3T, T49, TE, T1v, T2T, T3g, T2d, T13, T3n, T3o, T1i, T26, T4e, T4d;
	       E T23, T1n, T42, T3Z, T1m, T2h, T2I, T2i, T2P, T30, T37, T38, Tm, TB, TC;
	       E T46, T47, T4a, T2a, T2b, T2e, T1w, T1x, T1y, T3O, T3R, T3U, T3h, T3i, T3j;
	       E TH, TK, TL;
	       {
		    E T3, T2R, T1u, T2S, T6, T3f, T1r, T3e;
		    {
			 E T1, T2, T1s, T1t;
			 T1 = cr[0];
			 T2 = ci[WS(rs, 9)];
			 T3 = T1 + T2;
			 T2R = T1 - T2;
			 T1s = ci[WS(rs, 14)];
			 T1t = cr[WS(rs, 15)];
			 T1u = T1s - T1t;
			 T2S = T1s + T1t;
		    }
		    {
			 E T4, T5, T1p, T1q;
			 T4 = cr[WS(rs, 5)];
			 T5 = ci[WS(rs, 4)];
			 T6 = T4 + T5;
			 T3f = T4 - T5;
			 T1p = ci[WS(rs, 19)];
			 T1q = cr[WS(rs, 10)];
			 T1r = T1p - T1q;
			 T3e = T1p + T1q;
		    }
		    T7 = T3 + T6;
		    T3T = T2R - T2S;
		    T49 = T3f + T3e;
		    TE = T3 - T6;
		    T1v = T1r - T1u;
		    T2T = T2R + T2S;
		    T3g = T3e - T3f;
		    T2d = T1r + T1u;
	       }
	       {
		    E Te, T3M, T3X, TF, TV, T2E, T2W, T21, TA, T3Q, T41, TJ, T1h, T2O, T36;
		    E T25, Tl, T3N, T3Y, TG, T12, T2H, T2Z, T22, Tt, T3P, T40, TI, T1a, T2L;
		    E T33, T24;
		    {
			 E Ta, T2U, TU, T2V, Td, T2D, TR, T2C;
			 {
			      E T8, T9, TS, TT;
			      T8 = cr[WS(rs, 4)];
			      T9 = ci[WS(rs, 5)];
			      Ta = T8 + T9;
			      T2U = T8 - T9;
			      TS = ci[WS(rs, 10)];
			      TT = cr[WS(rs, 19)];
			      TU = TS - TT;
			      T2V = TS + TT;
			 }
			 {
			      E Tb, Tc, TP, TQ;
			      Tb = cr[WS(rs, 9)];
			      Tc = ci[0];
			      Td = Tb + Tc;
			      T2D = Tb - Tc;
			      TP = ci[WS(rs, 15)];
			      TQ = cr[WS(rs, 14)];
			      TR = TP - TQ;
			      T2C = TP + TQ;
			 }
			 Te = Ta + Td;
			 T3M = T2U - T2V;
			 T3X = T2D + T2C;
			 TF = Ta - Td;
			 TV = TR - TU;
			 T2E = T2C - T2D;
			 T2W = T2U + T2V;
			 T21 = TR + TU;
		    }
		    {
			 E Tw, T34, Tz, T2M, T1d, T2N, T1g, T35;
			 {
			      E Tu, Tv, Tx, Ty;
			      Tu = ci[WS(rs, 7)];
			      Tv = cr[WS(rs, 2)];
			      Tw = Tu + Tv;
			      T34 = Tu - Tv;
			      Tx = ci[WS(rs, 2)];
			      Ty = cr[WS(rs, 7)];
			      Tz = Tx + Ty;
			      T2M = Tx - Ty;
			 }
			 {
			      E T1b, T1c, T1e, T1f;
			      T1b = ci[WS(rs, 17)];
			      T1c = cr[WS(rs, 12)];
			      T1d = T1b - T1c;
			      T2N = T1b + T1c;
			      T1e = ci[WS(rs, 12)];
			      T1f = cr[WS(rs, 17)];
			      T1g = T1e - T1f;
			      T35 = T1e + T1f;
			 }
			 TA = Tw + Tz;
			 T3Q = T34 + T35;
			 T41 = T2M - T2N;
			 TJ = Tw - Tz;
			 T1h = T1d - T1g;
			 T2O = T2M + T2N;
			 T36 = T34 - T35;
			 T25 = T1d + T1g;
		    }
		    {
			 E Th, T2X, T11, T2Y, Tk, T2F, TY, T2G;
			 {
			      E Tf, Tg, TZ, T10;
			      Tf = ci[WS(rs, 3)];
			      Tg = cr[WS(rs, 6)];
			      Th = Tf + Tg;
			      T2X = Tf - Tg;
			      TZ = ci[WS(rs, 18)];
			      T10 = cr[WS(rs, 11)];
			      T11 = TZ - T10;
			      T2Y = TZ + T10;
			 }
			 {
			      E Ti, Tj, TW, TX;
			      Ti = cr[WS(rs, 1)];
			      Tj = ci[WS(rs, 8)];
			      Tk = Ti + Tj;
			      T2F = Ti - Tj;
			      TW = ci[WS(rs, 13)];
			      TX = cr[WS(rs, 16)];
			      TY = TW - TX;
			      T2G = TW + TX;
			 }
			 Tl = Th + Tk;
			 T3N = T2X - T2Y;
			 T3Y = T2F - T2G;
			 TG = Th - Tk;
			 T12 = TY - T11;
			 T2H = T2F + T2G;
			 T2Z = T2X + T2Y;
			 T22 = TY + T11;
		    }
		    {
			 E Tp, T31, T19, T32, Ts, T2K, T16, T2J;
			 {
			      E Tn, To, T17, T18;
			      Tn = cr[WS(rs, 8)];
			      To = ci[WS(rs, 1)];
			      Tp = Tn + To;
			      T31 = Tn - To;
			      T17 = ci[WS(rs, 16)];
			      T18 = cr[WS(rs, 13)];
			      T19 = T17 - T18;
			      T32 = T17 + T18;
			 }
			 {
			      E Tq, Tr, T14, T15;
			      Tq = ci[WS(rs, 6)];
			      Tr = cr[WS(rs, 3)];
			      Ts = Tq + Tr;
			      T2K = Tq - Tr;
			      T14 = ci[WS(rs, 11)];
			      T15 = cr[WS(rs, 18)];
			      T16 = T14 - T15;
			      T2J = T14 + T15;
			 }
			 Tt = Tp + Ts;
			 T3P = T31 + T32;
			 T40 = T2K + T2J;
			 TI = Tp - Ts;
			 T1a = T16 - T19;
			 T2L = T2J - T2K;
			 T33 = T31 - T32;
			 T24 = T16 + T19;
		    }
		    T13 = TV - T12;
		    T3n = T2W - T2Z;
		    T3o = T33 - T36;
		    T1i = T1a - T1h;
		    T26 = T24 - T25;
		    T4e = T3P - T3Q;
		    T4d = T3M - T3N;
		    T23 = T21 - T22;
		    T1n = TI - TJ;
		    T42 = T40 - T41;
		    T3Z = T3X - T3Y;
		    T1m = TF - TG;
		    T2h = Te - Tl;
		    T2I = T2E + T2H;
		    T2i = Tt - TA;
		    T2P = T2L + T2O;
		    T30 = T2W + T2Z;
		    T37 = T33 + T36;
		    T38 = T30 + T37;
		    Tm = Te + Tl;
		    TB = Tt + TA;
		    TC = Tm + TB;
		    T46 = T3X + T3Y;
		    T47 = T40 + T41;
		    T4a = T46 + T47;
		    T2a = T21 + T22;
		    T2b = T24 + T25;
		    T2e = T2a + T2b;
		    T1w = TV + T12;
		    T1x = T1a + T1h;
		    T1y = T1w + T1x;
		    T3O = T3M + T3N;
		    T3R = T3P + T3Q;
		    T3U = T3O + T3R;
		    T3h = T2E - T2H;
		    T3i = T2L - T2O;
		    T3j = T3h + T3i;
		    TH = TF + TG;
		    TK = TI + TJ;
		    TL = TH + TK;
	       }
	       cr[0] = T7 + TC;
	       ci[0] = T2d + T2e;
	       {
		    E T1U, T1W, T1T, T1V;
		    T1U = TE + TL;
		    T1W = T1v + T1y;
		    T1T = W[18];
		    T1V = W[19];
		    cr[WS(rs, 10)] = FNMS(T1V, T1W, T1T * T1U);
		    ci[WS(rs, 10)] = FMA(T1V, T1U, T1T * T1W);
	       }
	       {
		    E T4y, T4A, T4x, T4z;
		    T4y = T3T + T3U;
		    T4A = T49 + T4a;
		    T4x = W[8];
		    T4z = W[9];
		    cr[WS(rs, 5)] = FNMS(T4z, T4A, T4x * T4y);
		    ci[WS(rs, 5)] = FMA(T4x, T4A, T4z * T4y);
	       }
	       {
		    E T3I, T3K, T3H, T3J;
		    T3I = T2T + T38;
		    T3K = T3g + T3j;
		    T3H = W[28];
		    T3J = W[29];
		    cr[WS(rs, 15)] = FNMS(T3J, T3K, T3H * T3I);
		    ci[WS(rs, 15)] = FMA(T3H, T3K, T3J * T3I);
	       }
	       {
		    E T27, T2j, T2v, T2r, T2g, T2u, T20, T2q;
		    T27 = FMA(KP951056516, T23, KP587785252 * T26);
		    T2j = FMA(KP951056516, T2h, KP587785252 * T2i);
		    T2v = FNMS(KP951056516, T2i, KP587785252 * T2h);
		    T2r = FNMS(KP951056516, T26, KP587785252 * T23);
		    {
			 E T2c, T2f, T1Y, T1Z;
			 T2c = KP559016994 * (T2a - T2b);
			 T2f = FNMS(KP250000000, T2e, T2d);
			 T2g = T2c + T2f;
			 T2u = T2f - T2c;
			 T1Y = KP559016994 * (Tm - TB);
			 T1Z = FNMS(KP250000000, TC, T7);
			 T20 = T1Y + T1Z;
			 T2q = T1Z - T1Y;
		    }
		    {
			 E T28, T2k, T1X, T29;
			 T28 = T20 + T27;
			 T2k = T2g - T2j;
			 T1X = W[6];
			 T29 = W[7];
			 cr[WS(rs, 4)] = FNMS(T29, T2k, T1X * T28);
			 ci[WS(rs, 4)] = FMA(T29, T28, T1X * T2k);
		    }
		    {
			 E T2y, T2A, T2x, T2z;
			 T2y = T2q - T2r;
			 T2A = T2v + T2u;
			 T2x = W[22];
			 T2z = W[23];
			 cr[WS(rs, 12)] = FNMS(T2z, T2A, T2x * T2y);
			 ci[WS(rs, 12)] = FMA(T2z, T2y, T2x * T2A);
		    }
		    {
			 E T2m, T2o, T2l, T2n;
			 T2m = T20 - T27;
			 T2o = T2j + T2g;
			 T2l = W[30];
			 T2n = W[31];
			 cr[WS(rs, 16)] = FNMS(T2n, T2o, T2l * T2m);
			 ci[WS(rs, 16)] = FMA(T2n, T2m, T2l * T2o);
		    }
		    {
			 E T2s, T2w, T2p, T2t;
			 T2s = T2q + T2r;
			 T2w = T2u - T2v;
			 T2p = W[14];
			 T2t = W[15];
			 cr[WS(rs, 8)] = FNMS(T2t, T2w, T2p * T2s);
			 ci[WS(rs, 8)] = FMA(T2t, T2s, T2p * T2w);
		    }
	       }
	       {
		    E T43, T4f, T4r, T4m, T4c, T4q, T3W, T4n;
		    T43 = FMA(KP951056516, T3Z, KP587785252 * T42);
		    T4f = FMA(KP951056516, T4d, KP587785252 * T4e);
		    T4r = FNMS(KP951056516, T4e, KP587785252 * T4d);
		    T4m = FNMS(KP951056516, T42, KP587785252 * T3Z);
		    {
			 E T48, T4b, T3S, T3V;
			 T48 = KP559016994 * (T46 - T47);
			 T4b = FNMS(KP250000000, T4a, T49);
			 T4c = T48 + T4b;
			 T4q = T4b - T48;
			 T3S = KP559016994 * (T3O - T3R);
			 T3V = FNMS(KP250000000, T3U, T3T);
			 T3W = T3S + T3V;
			 T4n = T3V - T3S;
		    }
		    {
			 E T44, T4g, T3L, T45;
			 T44 = T3W - T43;
			 T4g = T4c + T4f;
			 T3L = W[0];
			 T45 = W[1];
			 cr[WS(rs, 1)] = FNMS(T45, T4g, T3L * T44);
			 ci[WS(rs, 1)] = FMA(T3L, T4g, T45 * T44);
		    }
		    {
			 E T4u, T4w, T4t, T4v;
			 T4u = T4n - T4m;
			 T4w = T4q + T4r;
			 T4t = W[32];
			 T4v = W[33];
			 cr[WS(rs, 17)] = FNMS(T4v, T4w, T4t * T4u);
			 ci[WS(rs, 17)] = FMA(T4t, T4w, T4v * T4u);
		    }
		    {
			 E T4i, T4k, T4h, T4j;
			 T4i = T43 + T3W;
			 T4k = T4c - T4f;
			 T4h = W[16];
			 T4j = W[17];
			 cr[WS(rs, 9)] = FNMS(T4j, T4k, T4h * T4i);
			 ci[WS(rs, 9)] = FMA(T4h, T4k, T4j * T4i);
		    }
		    {
			 E T4o, T4s, T4l, T4p;
			 T4o = T4m + T4n;
			 T4s = T4q - T4r;
			 T4l = W[24];
			 T4p = W[25];
			 cr[WS(rs, 13)] = FNMS(T4p, T4s, T4l * T4o);
			 ci[WS(rs, 13)] = FMA(T4l, T4s, T4p * T4o);
		    }
	       }
	       {
		    E T1j, T1o, T1M, T1J, T1B, T1N, TO, T1I;
		    T1j = FNMS(KP951056516, T1i, KP587785252 * T13);
		    T1o = FNMS(KP951056516, T1n, KP587785252 * T1m);
		    T1M = FMA(KP951056516, T1m, KP587785252 * T1n);
		    T1J = FMA(KP951056516, T13, KP587785252 * T1i);
		    {
			 E T1z, T1A, TM, TN;
			 T1z = FNMS(KP250000000, T1y, T1v);
			 T1A = KP559016994 * (T1w - T1x);
			 T1B = T1z - T1A;
			 T1N = T1A + T1z;
			 TM = FNMS(KP250000000, TL, TE);
			 TN = KP559016994 * (TH - TK);
			 TO = TM - TN;
			 T1I = TN + TM;
		    }
		    {
			 E T1k, T1C, TD, T1l;
			 T1k = TO - T1j;
			 T1C = T1o + T1B;
			 TD = W[2];
			 T1l = W[3];
			 cr[WS(rs, 2)] = FNMS(T1l, T1C, TD * T1k);
			 ci[WS(rs, 2)] = FMA(T1l, T1k, TD * T1C);
		    }
		    {
			 E T1Q, T1S, T1P, T1R;
			 T1Q = T1I + T1J;
			 T1S = T1N - T1M;
			 T1P = W[26];
			 T1R = W[27];
			 cr[WS(rs, 14)] = FNMS(T1R, T1S, T1P * T1Q);
			 ci[WS(rs, 14)] = FMA(T1R, T1Q, T1P * T1S);
		    }
		    {
			 E T1E, T1G, T1D, T1F;
			 T1E = TO + T1j;
			 T1G = T1B - T1o;
			 T1D = W[34];
			 T1F = W[35];
			 cr[WS(rs, 18)] = FNMS(T1F, T1G, T1D * T1E);
			 ci[WS(rs, 18)] = FMA(T1F, T1E, T1D * T1G);
		    }
		    {
			 E T1K, T1O, T1H, T1L;
			 T1K = T1I - T1J;
			 T1O = T1M + T1N;
			 T1H = W[10];
			 T1L = W[11];
			 cr[WS(rs, 6)] = FNMS(T1L, T1O, T1H * T1K);
			 ci[WS(rs, 6)] = FMA(T1L, T1K, T1H * T1O);
		    }
	       }
	       {
		    E T2Q, T3p, T3B, T3x, T3m, T3A, T3b, T3w;
		    T2Q = FNMS(KP951056516, T2P, KP587785252 * T2I);
		    T3p = FNMS(KP951056516, T3o, KP587785252 * T3n);
		    T3B = FMA(KP951056516, T3n, KP587785252 * T3o);
		    T3x = FMA(KP951056516, T2I, KP587785252 * T2P);
		    {
			 E T3k, T3l, T39, T3a;
			 T3k = FNMS(KP250000000, T3j, T3g);
			 T3l = KP559016994 * (T3h - T3i);
			 T3m = T3k - T3l;
			 T3A = T3l + T3k;
			 T39 = FNMS(KP250000000, T38, T2T);
			 T3a = KP559016994 * (T30 - T37);
			 T3b = T39 - T3a;
			 T3w = T3a + T39;
		    }
		    {
			 E T3c, T3q, T2B, T3d;
			 T3c = T2Q + T3b;
			 T3q = T3m - T3p;
			 T2B = W[4];
			 T3d = W[5];
			 cr[WS(rs, 3)] = FNMS(T3d, T3q, T2B * T3c);
			 ci[WS(rs, 3)] = FMA(T2B, T3q, T3d * T3c);
		    }
		    {
			 E T3E, T3G, T3D, T3F;
			 T3E = T3x + T3w;
			 T3G = T3A - T3B;
			 T3D = W[36];
			 T3F = W[37];
			 cr[WS(rs, 19)] = FNMS(T3F, T3G, T3D * T3E);
			 ci[WS(rs, 19)] = FMA(T3D, T3G, T3F * T3E);
		    }
		    {
			 E T3s, T3u, T3r, T3t;
			 T3s = T3b - T2Q;
			 T3u = T3m + T3p;
			 T3r = W[12];
			 T3t = W[13];
			 cr[WS(rs, 7)] = FNMS(T3t, T3u, T3r * T3s);
			 ci[WS(rs, 7)] = FMA(T3r, T3u, T3t * T3s);
		    }
		    {
			 E T3y, T3C, T3v, T3z;
			 T3y = T3w - T3x;
			 T3C = T3A + T3B;
			 T3v = W[20];
			 T3z = W[21];
			 cr[WS(rs, 11)] = FNMS(T3z, T3C, T3v * T3y);
			 ci[WS(rs, 11)] = FMA(T3v, T3C, T3z * T3y);
		    }
	       }
	  }
     }
}
static void q1bv_2(R *ri, R *ii, const R *W, stride rs, stride vs, INT mb, INT me, INT ms)
{
     {
	  INT m;
	  R *x;
	  x = ii;
	  for (m = mb, W = W + (mb * ((TWVL / VL) * 2)); m < me; m = m + VL, x = x + (VL * ms), W = W + (TWVL * 2), MAKE_VOLATILE_STRIDE(rs), MAKE_VOLATILE_STRIDE(vs)) {
	       V T1, T2, T3, T4, T5, T6;
	       T1 = LD(&(x[0]), ms, &(x[0]));
	       T2 = LD(&(x[WS(rs, 1)]), ms, &(x[WS(rs, 1)]));
	       T3 = BYTW(&(W[0]), VSUB(T1, T2));
	       T4 = LD(&(x[WS(vs, 1)]), ms, &(x[WS(vs, 1)]));
	       T5 = LD(&(x[WS(vs, 1) + WS(rs, 1)]), ms, &(x[WS(vs, 1) + WS(rs, 1)]));
	       T6 = BYTW(&(W[0]), VSUB(T4, T5));
	       ST(&(x[WS(vs, 1)]), T3, ms, &(x[WS(vs, 1)]));
	       ST(&(x[WS(vs, 1) + WS(rs, 1)]), T6, ms, &(x[WS(vs, 1) + WS(rs, 1)]));
	       ST(&(x[0]), VADD(T1, T2), ms, &(x[0]));
	       ST(&(x[WS(rs, 1)]), VADD(T4, T5), ms, &(x[WS(rs, 1)]));
	  }
     }
     VLEAVE();
}
Exemple #26
0
static void n1_9(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs)
{
     DK(KP939692620, +0.939692620785908384054109277324731469936208134);
     DK(KP342020143, +0.342020143325668733044099614682259580763083368);
     DK(KP984807753, +0.984807753012208059366743024589523013670643252);
     DK(KP173648177, +0.173648177666930348851716626769314796000375677);
     DK(KP642787609, +0.642787609686539326322643409907263432907559884);
     DK(KP766044443, +0.766044443118978035202392650555416673935832457);
     DK(KP500000000, +0.500000000000000000000000000000000000000000000);
     DK(KP866025403, +0.866025403784438646763723170752936183471402627);
     INT i;
     for (i = v; i > 0; i = i - 1, ri = ri + ivs, ii = ii + ivs, ro = ro + ovs, io = io + ovs, MAKE_VOLATILE_STRIDE(is), MAKE_VOLATILE_STRIDE(os)) {
	  E T5, TO, Th, Tk, T1g, TR, Ta, T1c, Tq, TW, Tv, TX, Tf, T1d, TB;
	  E T10, TG, TZ;
	  {
	       E T1, T2, T3, T4;
	       T1 = ri[0];
	       T2 = ri[WS(is, 3)];
	       T3 = ri[WS(is, 6)];
	       T4 = T2 + T3;
	       T5 = T1 + T4;
	       TO = KP866025403 * (T3 - T2);
	       Th = FNMS(KP500000000, T4, T1);
	  }
	  {
	       E TP, Ti, Tj, TQ;
	       TP = ii[0];
	       Ti = ii[WS(is, 3)];
	       Tj = ii[WS(is, 6)];
	       TQ = Ti + Tj;
	       Tk = KP866025403 * (Ti - Tj);
	       T1g = TP + TQ;
	       TR = FNMS(KP500000000, TQ, TP);
	  }
	  {
	       E T6, Ts, T9, Tr, Tp, Tt, Tm, Tu;
	       T6 = ri[WS(is, 1)];
	       Ts = ii[WS(is, 1)];
	       {
		    E T7, T8, Tn, To;
		    T7 = ri[WS(is, 4)];
		    T8 = ri[WS(is, 7)];
		    T9 = T7 + T8;
		    Tr = KP866025403 * (T8 - T7);
		    Tn = ii[WS(is, 4)];
		    To = ii[WS(is, 7)];
		    Tp = KP866025403 * (Tn - To);
		    Tt = Tn + To;
	       }
	       Ta = T6 + T9;
	       T1c = Ts + Tt;
	       Tm = FNMS(KP500000000, T9, T6);
	       Tq = Tm + Tp;
	       TW = Tm - Tp;
	       Tu = FNMS(KP500000000, Tt, Ts);
	       Tv = Tr + Tu;
	       TX = Tu - Tr;
	  }
	  {
	       E Tb, TD, Te, TC, TA, TE, Tx, TF;
	       Tb = ri[WS(is, 2)];
	       TD = ii[WS(is, 2)];
	       {
		    E Tc, Td, Ty, Tz;
		    Tc = ri[WS(is, 5)];
		    Td = ri[WS(is, 8)];
		    Te = Tc + Td;
		    TC = KP866025403 * (Td - Tc);
		    Ty = ii[WS(is, 5)];
		    Tz = ii[WS(is, 8)];
		    TA = KP866025403 * (Ty - Tz);
		    TE = Ty + Tz;
	       }
	       Tf = Tb + Te;
	       T1d = TD + TE;
	       Tx = FNMS(KP500000000, Te, Tb);
	       TB = Tx + TA;
	       T10 = Tx - TA;
	       TF = FNMS(KP500000000, TE, TD);
	       TG = TC + TF;
	       TZ = TF - TC;
	  }
	  {
	       E T1e, Tg, T1b, T1f, T1h, T1i;
	       T1e = KP866025403 * (T1c - T1d);
	       Tg = Ta + Tf;
	       T1b = FNMS(KP500000000, Tg, T5);
	       ro[0] = T5 + Tg;
	       ro[WS(os, 3)] = T1b + T1e;
	       ro[WS(os, 6)] = T1b - T1e;
	       T1f = KP866025403 * (Tf - Ta);
	       T1h = T1c + T1d;
	       T1i = FNMS(KP500000000, T1h, T1g);
	       io[WS(os, 3)] = T1f + T1i;
	       io[0] = T1g + T1h;
	       io[WS(os, 6)] = T1i - T1f;
	  }
	  {
	       E Tl, TS, TI, TN, TM, TT, TJ, TU;
	       Tl = Th + Tk;
	       TS = TO + TR;
	       {
		    E Tw, TH, TK, TL;
		    Tw = FMA(KP766044443, Tq, KP642787609 * Tv);
		    TH = FMA(KP173648177, TB, KP984807753 * TG);
		    TI = Tw + TH;
		    TN = KP866025403 * (TH - Tw);
		    TK = FNMS(KP642787609, Tq, KP766044443 * Tv);
		    TL = FNMS(KP984807753, TB, KP173648177 * TG);
		    TM = KP866025403 * (TK - TL);
		    TT = TK + TL;
	       }
	       ro[WS(os, 1)] = Tl + TI;
	       io[WS(os, 1)] = TS + TT;
	       TJ = FNMS(KP500000000, TI, Tl);
	       ro[WS(os, 7)] = TJ - TM;
	       ro[WS(os, 4)] = TJ + TM;
	       TU = FNMS(KP500000000, TT, TS);
	       io[WS(os, 4)] = TN + TU;
	       io[WS(os, 7)] = TU - TN;
	  }
	  {
	       E TV, T14, T12, T13, T17, T1a, T18, T19;
	       TV = Th - Tk;
	       T14 = TR - TO;
	       {
		    E TY, T11, T15, T16;
		    TY = FMA(KP173648177, TW, KP984807753 * TX);
		    T11 = FNMS(KP939692620, T10, KP342020143 * TZ);
		    T12 = TY + T11;
		    T13 = KP866025403 * (T11 - TY);
		    T15 = FNMS(KP984807753, TW, KP173648177 * TX);
		    T16 = FMA(KP342020143, T10, KP939692620 * TZ);
		    T17 = T15 - T16;
		    T1a = KP866025403 * (T15 + T16);
	       }
	       ro[WS(os, 2)] = TV + T12;
	       io[WS(os, 2)] = T14 + T17;
	       T18 = FNMS(KP500000000, T17, T14);
	       io[WS(os, 5)] = T13 + T18;
	       io[WS(os, 8)] = T18 - T13;
	       T19 = FNMS(KP500000000, T12, TV);
	       ro[WS(os, 8)] = T19 - T1a;
	       ro[WS(os, 5)] = T19 + T1a;
	  }
     }
}
static void r2cbIII_20(float *R0, float *R1, float *Cr, float *Ci, stride rs, stride csr, stride csi, INT v, INT ivs, INT ovs)
{
     DK(KP1_414213562, +1.414213562373095048801688724209698078569671875);
     DK(KP2_000000000, +2.000000000000000000000000000000000000000000000);
     DK(KP250000000, +0.250000000000000000000000000000000000000000000);
     DK(KP951056516, +0.951056516295153572116439333379382143405698634);
     DK(KP587785252, +0.587785252292473129168705954639072768597652438);
     DK(KP559016994, +0.559016994374947424102293417182819058860154590);
     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 T1, Tj, T1k, T13, T8, Tk, T17, Ts, T16, TI, T18, T19, Ta, Tu, T1i;
	  E TS, Th, Tv, TX, TD, TV, TL, TW, TY;
	  {
	       E T7, T12, T4, T11;
	       T1 = Cr[WS(csr, 2)];
	       {
		    E T5, T6, T2, T3;
		    T5 = Cr[WS(csr, 9)];
		    T6 = Cr[WS(csr, 5)];
		    T7 = T5 + T6;
		    T12 = T5 - T6;
		    T2 = Cr[WS(csr, 6)];
		    T3 = Cr[WS(csr, 1)];
		    T4 = T2 + T3;
		    T11 = T2 - T3;
	       }
	       Tj = KP559016994 * (T4 - T7);
	       T1k = FNMS(KP951056516, T12, KP587785252 * T11);
	       T13 = FMA(KP951056516, T11, KP587785252 * T12);
	       T8 = T4 + T7;
	       Tk = FNMS(KP250000000, T8, T1);
	  }
	  {
	       E Tr, T15, To, T14;
	       T17 = Ci[WS(csi, 2)];
	       {
		    E Tp, Tq, Tm, Tn;
		    Tp = Ci[WS(csi, 5)];
		    Tq = Ci[WS(csi, 9)];
		    Tr = Tp - Tq;
		    T15 = Tp + Tq;
		    Tm = Ci[WS(csi, 6)];
		    Tn = Ci[WS(csi, 1)];
		    To = Tm + Tn;
		    T14 = Tm - Tn;
	       }
	       Ts = FMA(KP951056516, To, KP587785252 * Tr);
	       T16 = KP559016994 * (T14 + T15);
	       TI = FNMS(KP951056516, Tr, KP587785252 * To);
	       T18 = T14 - T15;
	       T19 = FNMS(KP250000000, T18, T17);
	  }
	  {
	       E Tg, TR, Td, TQ;
	       Ta = Cr[WS(csr, 7)];
	       {
		    E Te, Tf, Tb, Tc;
		    Te = Cr[0];
		    Tf = Cr[WS(csr, 4)];
		    Tg = Te + Tf;
		    TR = Te - Tf;
		    Tb = Cr[WS(csr, 3)];
		    Tc = Cr[WS(csr, 8)];
		    Td = Tb + Tc;
		    TQ = Tb - Tc;
	       }
	       Tu = KP559016994 * (Td - Tg);
	       T1i = FNMS(KP951056516, TR, KP587785252 * TQ);
	       TS = FMA(KP951056516, TQ, KP587785252 * TR);
	       Th = Td + Tg;
	       Tv = FNMS(KP250000000, Th, Ta);
	  }
	  {
	       E TC, TU, Tz, TT;
	       TX = Ci[WS(csi, 7)];
	       {
		    E TA, TB, Tx, Ty;
		    TA = Ci[WS(csi, 4)];
		    TB = Ci[0];
		    TC = TA - TB;
		    TU = TB + TA;
		    Tx = Ci[WS(csi, 3)];
		    Ty = Ci[WS(csi, 8)];
		    Tz = Tx + Ty;
		    TT = Ty - Tx;
	       }
	       TD = FMA(KP951056516, Tz, KP587785252 * TC);
	       TV = KP559016994 * (TT - TU);
	       TL = FNMS(KP587785252, Tz, KP951056516 * TC);
	       TW = TT + TU;
	       TY = FMA(KP250000000, TW, TX);
	  }
	  {
	       E T9, Ti, T1w, T1t, T1u, T1v;
	       T9 = T1 + T8;
	       Ti = Ta + Th;
	       T1w = T9 - Ti;
	       T1t = T18 + T17;
	       T1u = TX - TW;
	       T1v = T1t + T1u;
	       R0[0] = KP2_000000000 * (T9 + Ti);
	       R0[WS(rs, 5)] = KP2_000000000 * (T1u - T1t);
	       R1[WS(rs, 2)] = KP1_414213562 * (T1v - T1w);
	       R1[WS(rs, 7)] = KP1_414213562 * (T1w + T1v);
	  }
	  {
	       E TJ, TO, T1m, T1q, TM, TN, T1j, T1r;
	       {
		    E TH, T1l, TK, T1h;
		    TH = Tk - Tj;
		    TJ = TH + TI;
		    TO = TH - TI;
		    T1l = T19 - T16;
		    T1m = T1k + T1l;
		    T1q = T1l - T1k;
		    TK = Tv - Tu;
		    TM = TK + TL;
		    TN = TL - TK;
		    T1h = TV + TY;
		    T1j = T1h - T1i;
		    T1r = T1i + T1h;
	       }
	       R0[WS(rs, 4)] = KP2_000000000 * (TJ + TM);
	       R0[WS(rs, 6)] = KP2_000000000 * (TN - TO);
	       R0[WS(rs, 9)] = KP2_000000000 * (T1r - T1q);
	       R0[WS(rs, 1)] = KP2_000000000 * (T1j - T1m);
	       {
		    E T1p, T1s, T1n, T1o;
		    T1p = TM - TJ;
		    T1s = T1q + T1r;
		    R1[WS(rs, 1)] = KP1_414213562 * (T1p - T1s);
		    R1[WS(rs, 6)] = KP1_414213562 * (T1p + T1s);
		    T1n = TO + TN;
		    T1o = T1m + T1j;
		    R1[WS(rs, 8)] = KP1_414213562 * (T1n - T1o);
		    R1[WS(rs, 3)] = KP1_414213562 * (T1n + T1o);
	       }
	  }
	  {
	       E Tt, TG, T1b, T1f, TE, TF, T10, T1e;
	       {
		    E Tl, T1a, Tw, TZ;
		    Tl = Tj + Tk;
		    Tt = Tl - Ts;
		    TG = Tl + Ts;
		    T1a = T16 + T19;
		    T1b = T13 + T1a;
		    T1f = T1a - T13;
		    Tw = Tu + Tv;
		    TE = Tw + TD;
		    TF = TD - Tw;
		    TZ = TV - TY;
		    T10 = TS + TZ;
		    T1e = TZ - TS;
	       }
	       R0[WS(rs, 8)] = KP2_000000000 * (Tt + TE);
	       R0[WS(rs, 2)] = KP2_000000000 * (TF - TG);
	       R0[WS(rs, 7)] = KP2_000000000 * (T1f + T1e);
	       R0[WS(rs, 3)] = KP2_000000000 * (T1b + T10);
	       {
		    E T1d, T1g, TP, T1c;
		    T1d = TG + TF;
		    T1g = T1e - T1f;
		    R1[WS(rs, 4)] = KP1_414213562 * (T1d + T1g);
		    R1[WS(rs, 9)] = KP1_414213562 * (T1g - T1d);
		    TP = Tt - TE;
		    T1c = T10 - T1b;
		    R1[0] = KP1_414213562 * (TP + T1c);
		    R1[WS(rs, 5)] = KP1_414213562 * (T1c - TP);
	       }
	  }
     }
}
Exemple #28
0
static void n1_9(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs)
{
     DK(KP954188894, +0.954188894138671133499268364187245676532219158);
     DK(KP363970234, +0.363970234266202361351047882776834043890471784);
     DK(KP852868531, +0.852868531952443209628250963940074071936020296);
     DK(KP984807753, +0.984807753012208059366743024589523013670643252);
     DK(KP492403876, +0.492403876506104029683371512294761506835321626);
     DK(KP777861913, +0.777861913430206160028177977318626690410586096);
     DK(KP839099631, +0.839099631177280011763127298123181364687434283);
     DK(KP176326980, +0.176326980708464973471090386868618986121633062);
     DK(KP866025403, +0.866025403784438646763723170752936183471402627);
     DK(KP500000000, +0.500000000000000000000000000000000000000000000);
     INT i;
     for (i = v; i > 0; i = i - 1, ri = ri + ivs, ii = ii + ivs, ro = ro + ovs, io = io + ovs, MAKE_VOLATILE_STRIDE(is), MAKE_VOLATILE_STRIDE(os)) {
	  E T17, TV, T14, TY, T11, T15;
	  {
	       E Tm, TM, TL, T5, Tl, T1f, Tb, Tt, Ta, T1c, TI, TX, TF, TW, Tc;
	       E Td, Tp, Tq;
	       {
		    E T1, Th, Ti, Tj, T4, T2, T3;
		    T1 = ri[0];
		    T2 = ri[WS(is, 3)];
		    T3 = ri[WS(is, 6)];
		    Th = ii[0];
		    Ti = ii[WS(is, 3)];
		    Tj = ii[WS(is, 6)];
		    T4 = T2 + T3;
		    Tm = T3 - T2;
		    {
			 E T6, Tz, T7, T8, TA, TB, Tk;
			 T6 = ri[WS(is, 1)];
			 TM = Ti - Tj;
			 Tk = Ti + Tj;
			 TL = FNMS(KP500000000, T4, T1);
			 T5 = T1 + T4;
			 Tz = ii[WS(is, 1)];
			 Tl = FNMS(KP500000000, Tk, Th);
			 T1f = Th + Tk;
			 T7 = ri[WS(is, 4)];
			 T8 = ri[WS(is, 7)];
			 TA = ii[WS(is, 4)];
			 TB = ii[WS(is, 7)];
			 {
			      E TE, T9, TH, TC, TG, TD;
			      Tb = ri[WS(is, 2)];
			      TE = T7 - T8;
			      T9 = T7 + T8;
			      TH = TB - TA;
			      TC = TA + TB;
			      Tt = ii[WS(is, 2)];
			      Ta = T6 + T9;
			      TG = FNMS(KP500000000, T9, T6);
			      T1c = Tz + TC;
			      TD = FNMS(KP500000000, TC, Tz);
			      TI = FNMS(KP866025403, TH, TG);
			      TX = FMA(KP866025403, TH, TG);
			      TF = FNMS(KP866025403, TE, TD);
			      TW = FMA(KP866025403, TE, TD);
			      Tc = ri[WS(is, 5)];
			      Td = ri[WS(is, 8)];
			      Tp = ii[WS(is, 5)];
			      Tq = ii[WS(is, 8)];
			 }
		    }
	       }
	       {
		    E Tn, TN, TZ, T10, TO, Ty, TJ, TP;
		    {
			 E Tw, Te, Tu, Tr;
			 T17 = FNMS(KP866025403, Tm, Tl);
			 Tn = FMA(KP866025403, Tm, Tl);
			 Tw = Td - Tc;
			 Te = Tc + Td;
			 Tu = Tp + Tq;
			 Tr = Tp - Tq;
			 TN = FMA(KP866025403, TM, TL);
			 TV = FNMS(KP866025403, TM, TL);
			 {
			      E Tf, To, T1d, Tv;
			      Tf = Tb + Te;
			      To = FNMS(KP500000000, Te, Tb);
			      T1d = Tt + Tu;
			      Tv = FNMS(KP500000000, Tu, Tt);
			      {
				   E Ts, Tg, T1i, Tx;
				   Ts = FMA(KP866025403, Tr, To);
				   TZ = FNMS(KP866025403, Tr, To);
				   Tg = Ta + Tf;
				   T1i = Tf - Ta;
				   Tx = FMA(KP866025403, Tw, Tv);
				   T10 = FNMS(KP866025403, Tw, Tv);
				   {
					E T1e, T1g, T1b, T1h;
					T1e = T1c - T1d;
					T1g = T1c + T1d;
					ro[0] = T5 + Tg;
					T1b = FNMS(KP500000000, Tg, T5);
					io[0] = T1f + T1g;
					T1h = FNMS(KP500000000, T1g, T1f);
					TO = FMA(KP176326980, Ts, Tx);
					Ty = FNMS(KP176326980, Tx, Ts);
					ro[WS(os, 6)] = FNMS(KP866025403, T1e, T1b);
					ro[WS(os, 3)] = FMA(KP866025403, T1e, T1b);
					io[WS(os, 6)] = FNMS(KP866025403, T1i, T1h);
					io[WS(os, 3)] = FMA(KP866025403, T1i, T1h);
					TJ = FNMS(KP839099631, TI, TF);
					TP = FMA(KP839099631, TF, TI);
				   }
			      }
			 }
		    }
		    {
			 E TS, TK, TU, TQ, TT, TR;
			 TS = FMA(KP777861913, TJ, Ty);
			 TK = FNMS(KP777861913, TJ, Ty);
			 TU = FNMS(KP777861913, TP, TO);
			 TQ = FMA(KP777861913, TP, TO);
			 TT = FMA(KP492403876, TK, Tn);
			 io[WS(os, 1)] = FNMS(KP984807753, TK, Tn);
			 TR = FNMS(KP492403876, TQ, TN);
			 ro[WS(os, 1)] = FMA(KP984807753, TQ, TN);
			 io[WS(os, 4)] = FMA(KP852868531, TU, TT);
			 io[WS(os, 7)] = FNMS(KP852868531, TU, TT);
			 ro[WS(os, 7)] = FNMS(KP852868531, TS, TR);
			 ro[WS(os, 4)] = FMA(KP852868531, TS, TR);
			 T14 = FNMS(KP176326980, TW, TX);
			 TY = FMA(KP176326980, TX, TW);
			 T11 = FNMS(KP363970234, T10, TZ);
			 T15 = FMA(KP363970234, TZ, T10);
		    }
	       }
	  }
	  {
	       E T12, T1a, T16, T18, T13, T19;
	       T12 = FNMS(KP954188894, T11, TY);
	       T1a = FMA(KP954188894, T11, TY);
	       T16 = FNMS(KP954188894, T15, T14);
	       T18 = FMA(KP954188894, T15, T14);
	       T13 = FNMS(KP492403876, T12, TV);
	       ro[WS(os, 2)] = FMA(KP984807753, T12, TV);
	       T19 = FMA(KP492403876, T18, T17);
	       io[WS(os, 2)] = FNMS(KP984807753, T18, T17);
	       ro[WS(os, 8)] = FMA(KP852868531, T16, T13);
	       ro[WS(os, 5)] = FNMS(KP852868531, T16, T13);
	       io[WS(os, 8)] = FMA(KP852868531, T1a, T19);
	       io[WS(os, 5)] = FNMS(KP852868531, T1a, T19);
	  }
     }
}
Exemple #29
0
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 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 = VSUB(Tj, Tk);
		    Tl = VADD(Tj, Tk);
		    {
			 V Tq, To, Ta, Tf;
			 Tq = VSUB(Tn, Tm);
			 To = VADD(Tm, Tn);
			 Ta = VADD(T6, T9);
			 Tf = VSUB(T9, T6);
			 {
			      V Tr, Ts, Tt, Tu, Tg, Ti, Tb, Th;
			      Tr = VADD(Tl, To);
			      STM2(&(xo[0]), Tr, ovs, &(xo[0]));
			      Ts = VSUB(Tl, To);
			      STM2(&(xo[8]), Ts, ovs, &(xo[0]));
			      Tt = VFMAI(Tq, Tp);
			      STM2(&(xo[4]), Tt, ovs, &(xo[0]));
			      Tu = VFNMSI(Tq, Tp);
			      STM2(&(xo[12]), Tu, ovs, &(xo[0]));
			      Tg = VFNMS(LDK(KP707106781), Tf, Te);
			      Ti = VFMA(LDK(KP707106781), Tf, Te);
			      Tb = VFMA(LDK(KP707106781), Ta, T3);
			      Th = VFNMS(LDK(KP707106781), Ta, T3);
			      {
				   V Tv, Tw, Tx, Ty;
				   Tv = VFMAI(Ti, Th);
				   STM2(&(xo[6]), Tv, ovs, &(xo[2]));
				   STN2(&(xo[4]), Tt, Tv, ovs);
				   Tw = VFNMSI(Ti, Th);
				   STM2(&(xo[10]), Tw, ovs, &(xo[2]));
				   STN2(&(xo[8]), Ts, Tw, ovs);
				   Tx = VFMAI(Tg, Tb);
				   STM2(&(xo[14]), Tx, ovs, &(xo[2]));
				   STN2(&(xo[12]), Tu, Tx, ovs);
				   Ty = VFNMSI(Tg, Tb);
				   STM2(&(xo[2]), Ty, ovs, &(xo[2]));
				   STN2(&(xo[0]), Tr, Ty, ovs);
			      }
			 }
		    }
	       }
	  }
     }
     VLEAVE();
}
static void hc2cb2_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) * 6); m < me; m = m + 1, Rp = Rp + ms, Ip = Ip + ms, Rm = Rm - ms, Im = Im - ms, W = W + 6, MAKE_VOLATILE_STRIDE(rs)) {
	  E Tf, Ti, TK, Tq, TH, TT, TX, TW, TY, TU, TI;
	  {
	       E Tg, Tl, Tp, Th, T1n, T1t, Tj;
	       Tf = W[0];
	       Tg = W[2];
	       Tl = W[4];
	       Tp = W[5];
	       Ti = W[1];
	       Th = Tf * Tg;
	       T1n = Tf * Tl;
	       T1t = Tf * Tp;
	       Tj = W[3];
	       {
		    E T1o, T1u, Tk, T1b, To, T1e, T13, TP, T1p, T7, T1h, T1v, TZ, Tv, T1i;
		    E TB, TA, TQ, Te, T1w, TE, T1j;
		    {
			 E Tr, T3, Ts, T1f, TO, TL, T6, Tt;
			 {
			      E TM, TN, T4, T5;
			      {
				   E T1, Tn, T2, TJ, Tm;
				   T1 = Rp[0];
				   T1o = FMA(Ti, Tp, T1n);
				   T1u = FNMS(Ti, Tl, T1t);
				   Tk = FMA(Ti, Tj, Th);
				   T1b = FNMS(Ti, Tj, Th);
				   Tn = Tf * Tj;
				   T2 = Rm[WS(rs, 3)];
				   TM = Ip[0];
				   TJ = Tk * Tp;
				   Tm = Tk * Tl;
				   To = FNMS(Ti, Tg, Tn);
				   T1e = FMA(Ti, Tg, Tn);
				   Tr = T1 - T2;
				   T3 = T1 + T2;
				   TK = FNMS(To, Tl, TJ);
				   Tq = FMA(To, Tp, Tm);
				   TN = Im[WS(rs, 3)];
			      }
			      T4 = Rp[WS(rs, 2)];
			      T5 = Rm[WS(rs, 1)];
			      Ts = Ip[WS(rs, 2)];
			      T1f = TM - TN;
			      TO = TM + TN;
			      TL = T4 - T5;
			      T6 = T4 + T5;
			      Tt = Im[WS(rs, 1)];
			 }
			 {
			      E Tw, Ta, TC, Tz, Td, TD;
			      {
				   E Tx, Ty, Tb, Tc;
				   {
					E T8, T1g, Tu, T9;
					T8 = Rp[WS(rs, 1)];
					T13 = TO - TL;
					TP = TL + TO;
					T1p = T3 - T6;
					T7 = T3 + T6;
					T1g = Ts - Tt;
					Tu = Ts + Tt;
					T9 = Rm[WS(rs, 2)];
					Tx = Ip[WS(rs, 1)];
					T1h = T1f + T1g;
					T1v = T1f - T1g;
					TZ = Tr + Tu;
					Tv = Tr - Tu;
					Tw = T8 - T9;
					Ta = T8 + T9;
					Ty = Im[WS(rs, 2)];
				   }
				   Tb = Rm[0];
				   Tc = Rp[WS(rs, 3)];
				   TC = Ip[WS(rs, 3)];
				   T1i = Tx - Ty;
				   Tz = Tx + Ty;
				   TB = Tb - Tc;
				   Td = Tb + Tc;
				   TD = Im[0];
			      }
			      TA = Tw - Tz;
			      TQ = Tw + Tz;
			      Te = Ta + Td;
			      T1w = Ta - Td;
			      TE = TC + TD;
			      T1j = TC - TD;
			 }
		    }
		    {
			 E T1x, T1k, T1r, TG, TS, T19, T15, T17, T11, T16, T12;
			 {
			      E T1B, T1z, T10, T1A, T1C;
			      T1x = T1v - T1w;
			      T1B = T1w + T1v;
			      Rp[0] = T7 + Te;
			      {
				   E T1q, TR, TF, T14;
				   T1k = T1i + T1j;
				   T1q = T1j - T1i;
				   TR = TB + TE;
				   TF = TB - TE;
				   T1r = T1p - T1q;
				   T1z = T1p + T1q;
				   Rm[0] = T1h + T1k;
				   TG = TA + TF;
				   T14 = TA - TF;
				   TS = TQ - TR;
				   T10 = TQ + TR;
				   T1A = Tk * T1z;
				   T19 = FNMS(KP707106781, T14, T13);
				   T15 = FMA(KP707106781, T14, T13);
				   T1C = Tk * T1B;
			      }
			      T17 = FMA(KP707106781, T10, TZ);
			      T11 = FNMS(KP707106781, T10, TZ);
			      Rp[WS(rs, 1)] = FNMS(To, T1B, T1A);
			      T16 = Tg * T15;
			      Rm[WS(rs, 1)] = FMA(To, T1z, T1C);
			 }
			 T12 = Tg * T11;
			 {
			      E T1l, T1a, T1c, T18;
			      Im[WS(rs, 1)] = FMA(Tj, T11, T16);
			      Ip[WS(rs, 1)] = FNMS(Tj, T15, T12);
			      T18 = Tl * T17;
			      T1l = T1h - T1k;
			      T1a = Tl * T19;
			      T1c = T7 - Te;
			      Ip[WS(rs, 3)] = FNMS(Tp, T19, T18);
			      {
				   E T1s, T1m, T1d, T1y, TV;
				   Im[WS(rs, 3)] = FMA(Tp, T17, T1a);
				   T1m = T1e * T1c;
				   T1d = T1b * T1c;
				   T1s = T1o * T1r;
				   Rm[WS(rs, 2)] = FMA(T1b, T1l, T1m);
				   Rp[WS(rs, 2)] = FNMS(T1e, T1l, T1d);
				   Rp[WS(rs, 3)] = FNMS(T1u, T1x, T1s);
				   T1y = T1o * T1x;
				   TV = FMA(KP707106781, TG, Tv);
				   TH = FNMS(KP707106781, TG, Tv);
				   TT = FNMS(KP707106781, TS, TP);
				   TX = FMA(KP707106781, TS, TP);
				   Rm[WS(rs, 3)] = FMA(T1u, T1r, T1y);
				   TW = Tf * TV;
				   TY = Ti * TV;
			      }
			 }
		    }
	       }
	  }
	  Ip[0] = FNMS(Ti, TX, TW);
	  Im[0] = FMA(Tf, TX, TY);
	  TU = TK * TH;
	  TI = Tq * TH;
	  Im[WS(rs, 2)] = FMA(Tq, TT, TU);
	  Ip[WS(rs, 2)] = FNMS(TK, TT, TI);
     }
}