static void q1fv_4(R *ri, R *ii, const R *W, stride rs, stride vs, INT mb, INT me, INT ms)
{
     {
	  INT m;
	  R *x;
	  x = ri;
	  for (m = mb, W = W + (mb * ((TWVL / VL) * 6)); m < me; m = m + VL, x = x + (VL * ms), W = W + (TWVL * 6), MAKE_VOLATILE_STRIDE(8, rs), MAKE_VOLATILE_STRIDE(8, vs)) {
	       V Tb, Tm, Tx, TI;
	       {
		    V Tc, T9, T3, TG, TA, TH, TD, Ta, T6, Td, Tn, To, Tq, Tr, Tf;
		    V Tg;
		    {
			 V T1, T2, Ty, Tz, TB, TC, T4, T5;
			 T1 = LD(&(x[0]), ms, &(x[0]));
			 T2 = LD(&(x[WS(rs, 2)]), ms, &(x[0]));
			 Ty = LD(&(x[WS(vs, 3)]), ms, &(x[WS(vs, 3)]));
			 Tz = LD(&(x[WS(vs, 3) + WS(rs, 2)]), ms, &(x[WS(vs, 3)]));
			 TB = LD(&(x[WS(vs, 3) + WS(rs, 1)]), ms, &(x[WS(vs, 3) + WS(rs, 1)]));
			 TC = LD(&(x[WS(vs, 3) + WS(rs, 3)]), ms, &(x[WS(vs, 3) + WS(rs, 1)]));
			 T4 = LD(&(x[WS(rs, 1)]), ms, &(x[WS(rs, 1)]));
			 T5 = LD(&(x[WS(rs, 3)]), ms, &(x[WS(rs, 1)]));
			 Tc = LD(&(x[WS(vs, 1)]), ms, &(x[WS(vs, 1)]));
			 T9 = VADD(T1, T2);
			 T3 = VSUB(T1, T2);
			 TG = VADD(Ty, Tz);
			 TA = VSUB(Ty, Tz);
			 TH = VADD(TB, TC);
			 TD = VSUB(TB, TC);
			 Ta = VADD(T4, T5);
			 T6 = VSUB(T4, T5);
			 Td = LD(&(x[WS(vs, 1) + WS(rs, 2)]), ms, &(x[WS(vs, 1)]));
			 Tn = LD(&(x[WS(vs, 2)]), ms, &(x[WS(vs, 2)]));
			 To = LD(&(x[WS(vs, 2) + WS(rs, 2)]), ms, &(x[WS(vs, 2)]));
			 Tq = LD(&(x[WS(vs, 2) + WS(rs, 1)]), ms, &(x[WS(vs, 2) + WS(rs, 1)]));
			 Tr = LD(&(x[WS(vs, 2) + WS(rs, 3)]), ms, &(x[WS(vs, 2) + WS(rs, 1)]));
			 Tf = LD(&(x[WS(vs, 1) + WS(rs, 1)]), ms, &(x[WS(vs, 1) + WS(rs, 1)]));
			 Tg = LD(&(x[WS(vs, 1) + WS(rs, 3)]), ms, &(x[WS(vs, 1) + WS(rs, 1)]));
		    }
		    {
			 V Tk, Te, Tv, Tp, Tw, Ts, Tl, Th, T7, TE, Tu, TF;
			 ST(&(x[0]), VADD(T9, Ta), ms, &(x[0]));
			 Tk = VADD(Tc, Td);
			 Te = VSUB(Tc, Td);
			 Tv = VADD(Tn, To);
			 Tp = VSUB(Tn, To);
			 Tw = VADD(Tq, Tr);
			 Ts = VSUB(Tq, Tr);
			 Tl = VADD(Tf, Tg);
			 Th = VSUB(Tf, Tg);
			 ST(&(x[WS(rs, 3)]), VADD(TG, TH), ms, &(x[WS(rs, 1)]));
			 T7 = BYTWJ(&(W[0]), VFNMSI(T6, T3));
			 TE = BYTWJ(&(W[0]), VFNMSI(TD, TA));
			 {
			      V Tt, Ti, Tj, T8;
			      T8 = BYTWJ(&(W[TWVL * 4]), VFMAI(T6, T3));
			      ST(&(x[WS(rs, 2)]), VADD(Tv, Tw), ms, &(x[0]));
			      Tt = BYTWJ(&(W[0]), VFNMSI(Ts, Tp));
			      ST(&(x[WS(rs, 1)]), VADD(Tk, Tl), ms, &(x[WS(rs, 1)]));
			      Ti = BYTWJ(&(W[0]), VFNMSI(Th, Te));
			      Tj = BYTWJ(&(W[TWVL * 4]), VFMAI(Th, Te));
			      ST(&(x[WS(vs, 1)]), T7, ms, &(x[WS(vs, 1)]));
			      ST(&(x[WS(vs, 1) + WS(rs, 3)]), TE, ms, &(x[WS(vs, 1) + WS(rs, 1)]));
			      ST(&(x[WS(vs, 3)]), T8, ms, &(x[WS(vs, 3)]));
			      Tu = BYTWJ(&(W[TWVL * 4]), VFMAI(Ts, Tp));
			      ST(&(x[WS(vs, 1) + WS(rs, 2)]), Tt, ms, &(x[WS(vs, 1)]));
			      TF = BYTWJ(&(W[TWVL * 4]), VFMAI(TD, TA));
			      ST(&(x[WS(vs, 1) + WS(rs, 1)]), Ti, ms, &(x[WS(vs, 1) + WS(rs, 1)]));
			      ST(&(x[WS(vs, 3) + WS(rs, 1)]), Tj, ms, &(x[WS(vs, 3) + WS(rs, 1)]));
			 }
			 Tb = BYTWJ(&(W[TWVL * 2]), VSUB(T9, Ta));
			 Tm = BYTWJ(&(W[TWVL * 2]), VSUB(Tk, Tl));
			 Tx = BYTWJ(&(W[TWVL * 2]), VSUB(Tv, Tw));
			 ST(&(x[WS(vs, 3) + WS(rs, 2)]), Tu, ms, &(x[WS(vs, 3)]));
			 TI = BYTWJ(&(W[TWVL * 2]), VSUB(TG, TH));
			 ST(&(x[WS(vs, 3) + WS(rs, 3)]), TF, ms, &(x[WS(vs, 3) + WS(rs, 1)]));
		    }
	       }
	       ST(&(x[WS(vs, 2)]), Tb, ms, &(x[WS(vs, 2)]));
	       ST(&(x[WS(vs, 2) + WS(rs, 1)]), Tm, ms, &(x[WS(vs, 2) + WS(rs, 1)]));
	       ST(&(x[WS(vs, 2) + WS(rs, 2)]), Tx, ms, &(x[WS(vs, 2)]));
	       ST(&(x[WS(vs, 2) + WS(rs, 3)]), TI, ms, &(x[WS(vs, 2) + WS(rs, 1)]));
	  }
     }
     VLEAVE();
}
static void n2bv_20(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs)
{
     DVK(KP559016994, +0.559016994374947424102293417182819058860154590);
     DVK(KP618033988, +0.618033988749894848204586834365638117720309180);
     DVK(KP951056516, +0.951056516295153572116439333379382143405698634);
     DVK(KP250000000, +0.250000000000000000000000000000000000000000000);
     {
	  INT i;
	  const R *xi;
	  R *xo;
	  xi = ii;
	  xo = io;
	  for (i = v; i > 0; i = i - VL, xi = xi + (VL * ivs), xo = xo + (VL * ovs), MAKE_VOLATILE_STRIDE(40, is), MAKE_VOLATILE_STRIDE(40, os)) {
	       V T1H, T1I, TS, TA, TN, TV, T1M, T1N, T1O, T1P, T1R, T1S, TK, TU, TR;
	       V Tl;
	       {
		    V T3, TE, T1r, T13, Ta, TL, Tz, TG, Ts, TF, Th, TM, T1u, T1C, T1n;
		    V T1a, T1m, T1h, T1x, T1D, Tk, Ti;
		    {
			 V T1, T2, TC, TD;
			 T1 = LD(&(xi[0]), ivs, &(xi[0]));
			 T2 = LD(&(xi[WS(is, 10)]), ivs, &(xi[0]));
			 TC = LD(&(xi[WS(is, 5)]), ivs, &(xi[WS(is, 1)]));
			 TD = LD(&(xi[WS(is, 15)]), ivs, &(xi[WS(is, 1)]));
			 {
			      V T14, T6, T1c, Tv, Tm, T1f, Ty, T17, T9, Tn, Tp, T1b, Td, Tq, Te;
			      V Tf, T15, To;
			      {
				   V Tw, Tx, T7, T8, Tb, Tc;
				   {
					V T4, T5, Tt, Tu, T11, T12;
					T4 = LD(&(xi[WS(is, 4)]), ivs, &(xi[0]));
					T5 = LD(&(xi[WS(is, 14)]), ivs, &(xi[0]));
					Tt = LD(&(xi[WS(is, 13)]), ivs, &(xi[WS(is, 1)]));
					Tu = LD(&(xi[WS(is, 3)]), ivs, &(xi[WS(is, 1)]));
					Tw = LD(&(xi[WS(is, 17)]), ivs, &(xi[WS(is, 1)]));
					T3 = VSUB(T1, T2);
					T11 = VADD(T1, T2);
					TE = VSUB(TC, TD);
					T12 = VADD(TC, TD);
					T14 = VADD(T4, T5);
					T6 = VSUB(T4, T5);
					T1c = VADD(Tt, Tu);
					Tv = VSUB(Tt, Tu);
					Tx = LD(&(xi[WS(is, 7)]), ivs, &(xi[WS(is, 1)]));
					T7 = LD(&(xi[WS(is, 16)]), ivs, &(xi[0]));
					T8 = LD(&(xi[WS(is, 6)]), ivs, &(xi[0]));
					T1r = VADD(T11, T12);
					T13 = VSUB(T11, T12);
				   }
				   Tb = LD(&(xi[WS(is, 8)]), ivs, &(xi[0]));
				   Tc = LD(&(xi[WS(is, 18)]), ivs, &(xi[0]));
				   Tm = LD(&(xi[WS(is, 9)]), ivs, &(xi[WS(is, 1)]));
				   T1f = VADD(Tw, Tx);
				   Ty = VSUB(Tw, Tx);
				   T17 = VADD(T7, T8);
				   T9 = VSUB(T7, T8);
				   Tn = LD(&(xi[WS(is, 19)]), ivs, &(xi[WS(is, 1)]));
				   Tp = LD(&(xi[WS(is, 1)]), ivs, &(xi[WS(is, 1)]));
				   T1b = VADD(Tb, Tc);
				   Td = VSUB(Tb, Tc);
				   Tq = LD(&(xi[WS(is, 11)]), ivs, &(xi[WS(is, 1)]));
				   Te = LD(&(xi[WS(is, 12)]), ivs, &(xi[0]));
				   Tf = LD(&(xi[WS(is, 2)]), ivs, &(xi[0]));
			      }
			      Ta = VADD(T6, T9);
			      TL = VSUB(T6, T9);
			      T15 = VADD(Tm, Tn);
			      To = VSUB(Tm, Tn);
			      Tz = VSUB(Tv, Ty);
			      TG = VADD(Tv, Ty);
			      {
				   V T1d, T1v, T18, Tr, T1e, Tg, T16, T1s;
				   T1d = VSUB(T1b, T1c);
				   T1v = VADD(T1b, T1c);
				   T18 = VADD(Tp, Tq);
				   Tr = VSUB(Tp, Tq);
				   T1e = VADD(Te, Tf);
				   Tg = VSUB(Te, Tf);
				   T16 = VSUB(T14, T15);
				   T1s = VADD(T14, T15);
				   {
					V T1t, T19, T1w, T1g;
					T1t = VADD(T17, T18);
					T19 = VSUB(T17, T18);
					Ts = VSUB(To, Tr);
					TF = VADD(To, Tr);
					T1w = VADD(T1e, T1f);
					T1g = VSUB(T1e, T1f);
					Th = VADD(Td, Tg);
					TM = VSUB(Td, Tg);
					T1u = VADD(T1s, T1t);
					T1C = VSUB(T1s, T1t);
					T1n = VSUB(T16, T19);
					T1a = VADD(T16, T19);
					T1m = VSUB(T1d, T1g);
					T1h = VADD(T1d, T1g);
					T1x = VADD(T1v, T1w);
					T1D = VSUB(T1v, T1w);
				   }
			      }
			 }
		    }
		    Tk = VSUB(Ta, Th);
		    Ti = VADD(Ta, Th);
		    {
			 V TJ, T1k, T1A, TZ, Tj, T1E, T1G, TI, T10, T1j, T1z, T1i, T1y, TH;
			 TJ = VSUB(TF, TG);
			 TH = VADD(TF, TG);
			 T1i = VADD(T1a, T1h);
			 T1k = VSUB(T1a, T1h);
			 T1y = VADD(T1u, T1x);
			 T1A = VSUB(T1u, T1x);
			 TZ = VADD(T3, Ti);
			 Tj = VFNMS(LDK(KP250000000), Ti, T3);
			 T1E = VMUL(LDK(KP951056516), VFMA(LDK(KP618033988), T1D, T1C));
			 T1G = VMUL(LDK(KP951056516), VFNMS(LDK(KP618033988), T1C, T1D));
			 TI = VFNMS(LDK(KP250000000), TH, TE);
			 T10 = VADD(TE, TH);
			 T1j = VFNMS(LDK(KP250000000), T1i, T13);
			 T1H = VADD(T1r, T1y);
			 STM2(&(xo[0]), T1H, ovs, &(xo[0]));
			 T1z = VFNMS(LDK(KP250000000), T1y, T1r);
			 T1I = VADD(T13, T1i);
			 STM2(&(xo[20]), T1I, ovs, &(xo[0]));
			 {
			      V T1J, T1K, T1p, T1l, T1o, T1q, T1F, T1B, T1L, T1Q;
			      TS = VFNMS(LDK(KP618033988), Ts, Tz);
			      TA = VFMA(LDK(KP618033988), Tz, Ts);
			      TN = VFMA(LDK(KP618033988), TM, TL);
			      TV = VFNMS(LDK(KP618033988), TL, TM);
			      T1J = VFMAI(T10, TZ);
			      STM2(&(xo[10]), T1J, ovs, &(xo[2]));
			      T1K = VFNMSI(T10, TZ);
			      STM2(&(xo[30]), T1K, ovs, &(xo[2]));
			      T1p = VFMA(LDK(KP559016994), T1k, T1j);
			      T1l = VFNMS(LDK(KP559016994), T1k, T1j);
			      T1o = VMUL(LDK(KP951056516), VFNMS(LDK(KP618033988), T1n, T1m));
			      T1q = VMUL(LDK(KP951056516), VFMA(LDK(KP618033988), T1m, T1n));
			      T1F = VFNMS(LDK(KP559016994), T1A, T1z);
			      T1B = VFMA(LDK(KP559016994), T1A, T1z);
			      T1L = VFNMSI(T1q, T1p);
			      STM2(&(xo[28]), T1L, ovs, &(xo[0]));
			      STN2(&(xo[28]), T1L, T1K, ovs);
			      T1M = VFMAI(T1q, T1p);
			      STM2(&(xo[12]), T1M, ovs, &(xo[0]));
			      T1N = VFMAI(T1o, T1l);
			      STM2(&(xo[36]), T1N, ovs, &(xo[0]));
			      T1O = VFNMSI(T1o, T1l);
			      STM2(&(xo[4]), T1O, ovs, &(xo[0]));
			      T1P = VFMAI(T1E, T1B);
			      STM2(&(xo[32]), T1P, ovs, &(xo[0]));
			      T1Q = VFNMSI(T1E, T1B);
			      STM2(&(xo[8]), T1Q, ovs, &(xo[0]));
			      STN2(&(xo[8]), T1Q, T1J, ovs);
			      T1R = VFNMSI(T1G, T1F);
			      STM2(&(xo[24]), T1R, ovs, &(xo[0]));
			      T1S = VFMAI(T1G, T1F);
			      STM2(&(xo[16]), T1S, ovs, &(xo[0]));
			      TK = VFMA(LDK(KP559016994), TJ, TI);
			      TU = VFNMS(LDK(KP559016994), TJ, TI);
			      TR = VFNMS(LDK(KP559016994), Tk, Tj);
			      Tl = VFMA(LDK(KP559016994), Tk, Tj);
			 }
		    }
	       }
	       {
		    V TY, TW, TO, TQ, TB, TP, TX, TT;
		    TY = VFMA(LDK(KP951056516), TV, TU);
		    TW = VFNMS(LDK(KP951056516), TV, TU);
		    TO = VFMA(LDK(KP951056516), TN, TK);
		    TQ = VFNMS(LDK(KP951056516), TN, TK);
		    TB = VFNMS(LDK(KP951056516), TA, Tl);
		    TP = VFMA(LDK(KP951056516), TA, Tl);
		    TX = VFNMS(LDK(KP951056516), TS, TR);
		    TT = VFMA(LDK(KP951056516), TS, TR);
		    {
			 V T1T, T1U, T1V, T1W;
			 T1T = VFMAI(TQ, TP);
			 STM2(&(xo[18]), T1T, ovs, &(xo[2]));
			 STN2(&(xo[16]), T1S, T1T, ovs);
			 T1U = VFNMSI(TQ, TP);
			 STM2(&(xo[22]), T1U, ovs, &(xo[2]));
			 STN2(&(xo[20]), T1I, T1U, ovs);
			 T1V = VFMAI(TO, TB);
			 STM2(&(xo[2]), T1V, ovs, &(xo[2]));
			 STN2(&(xo[0]), T1H, T1V, ovs);
			 T1W = VFNMSI(TO, TB);
			 STM2(&(xo[38]), T1W, ovs, &(xo[2]));
			 STN2(&(xo[36]), T1N, T1W, ovs);
			 {
			      V T1X, T1Y, T1Z, T20;
			      T1X = VFMAI(TW, TT);
			      STM2(&(xo[34]), T1X, ovs, &(xo[2]));
			      STN2(&(xo[32]), T1P, T1X, ovs);
			      T1Y = VFNMSI(TW, TT);
			      STM2(&(xo[6]), T1Y, ovs, &(xo[2]));
			      STN2(&(xo[4]), T1O, T1Y, ovs);
			      T1Z = VFMAI(TY, TX);
			      STM2(&(xo[26]), T1Z, ovs, &(xo[2]));
			      STN2(&(xo[24]), T1R, T1Z, ovs);
			      T20 = VFNMSI(TY, TX);
			      STM2(&(xo[14]), T20, ovs, &(xo[2]));
			      STN2(&(xo[12]), T1M, T20, ovs);
			 }
		    }
	       }
	  }
     }
     VLEAVE();
}
Example #3
0
static void n1fv_10(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs)
{
     DVK(KP559016994, +0.559016994374947424102293417182819058860154590);
     DVK(KP250000000, +0.250000000000000000000000000000000000000000000);
     DVK(KP618033988, +0.618033988749894848204586834365638117720309180);
     DVK(KP951056516, +0.951056516295153572116439333379382143405698634);
     {
	  INT i;
	  const R *xi;
	  R *xo;
	  xi = ri;
	  xo = ro;
	  for (i = v; i > 0; i = i - VL, xi = xi + (VL * ivs), xo = xo + (VL * ovs), MAKE_VOLATILE_STRIDE(20, is), MAKE_VOLATILE_STRIDE(20, os)) {
	       V Tb, Tr, T3, Ts, T6, Tw, Tg, Tt, T9, Tc, T1, T2;
	       T1 = LD(&(xi[0]), ivs, &(xi[0]));
	       T2 = LD(&(xi[WS(is, 5)]), ivs, &(xi[WS(is, 1)]));
	       {
		    V T4, T5, Te, Tf, T7, T8;
		    T4 = LD(&(xi[WS(is, 2)]), ivs, &(xi[0]));
		    T5 = LD(&(xi[WS(is, 7)]), ivs, &(xi[WS(is, 1)]));
		    Te = LD(&(xi[WS(is, 6)]), ivs, &(xi[0]));
		    Tf = LD(&(xi[WS(is, 1)]), ivs, &(xi[WS(is, 1)]));
		    T7 = LD(&(xi[WS(is, 8)]), ivs, &(xi[0]));
		    T8 = LD(&(xi[WS(is, 3)]), ivs, &(xi[WS(is, 1)]));
		    Tb = LD(&(xi[WS(is, 4)]), ivs, &(xi[0]));
		    Tr = VADD(T1, T2);
		    T3 = VSUB(T1, T2);
		    Ts = VADD(T4, T5);
		    T6 = VSUB(T4, T5);
		    Tw = VADD(Te, Tf);
		    Tg = VSUB(Te, Tf);
		    Tt = VADD(T7, T8);
		    T9 = VSUB(T7, T8);
		    Tc = LD(&(xi[WS(is, 9)]), ivs, &(xi[WS(is, 1)]));
	       }
	       {
		    V TD, Tu, Tm, Ta, Td, Tv;
		    TD = VSUB(Ts, Tt);
		    Tu = VADD(Ts, Tt);
		    Tm = VSUB(T6, T9);
		    Ta = VADD(T6, T9);
		    Td = VSUB(Tb, Tc);
		    Tv = VADD(Tb, Tc);
		    {
			 V TC, Tx, Tn, Th;
			 TC = VSUB(Tv, Tw);
			 Tx = VADD(Tv, Tw);
			 Tn = VSUB(Td, Tg);
			 Th = VADD(Td, Tg);
			 {
			      V Ty, TA, TE, TG, Ti, Tk, To, Tq, Tz, Tj;
			      Ty = VADD(Tu, Tx);
			      TA = VSUB(Tu, Tx);
			      TE = VMUL(LDK(KP951056516), VFNMS(LDK(KP618033988), TD, TC));
			      TG = VMUL(LDK(KP951056516), VFMA(LDK(KP618033988), TC, TD));
			      Ti = VADD(Ta, Th);
			      Tk = VSUB(Ta, Th);
			      To = VMUL(LDK(KP951056516), VFMA(LDK(KP618033988), Tn, Tm));
			      Tq = VMUL(LDK(KP951056516), VFNMS(LDK(KP618033988), Tm, Tn));
			      Tz = VFNMS(LDK(KP250000000), Ty, Tr);
			      ST(&(xo[0]), VADD(Tr, Ty), ovs, &(xo[0]));
			      Tj = VFNMS(LDK(KP250000000), Ti, T3);
			      ST(&(xo[WS(os, 5)]), VADD(T3, Ti), ovs, &(xo[WS(os, 1)]));
			      {
				   V TB, TF, Tl, Tp;
				   TB = VFNMS(LDK(KP559016994), TA, Tz);
				   TF = VFMA(LDK(KP559016994), TA, Tz);
				   Tl = VFMA(LDK(KP559016994), Tk, Tj);
				   Tp = VFNMS(LDK(KP559016994), Tk, Tj);
				   ST(&(xo[WS(os, 4)]), VFMAI(TG, TF), ovs, &(xo[0]));
				   ST(&(xo[WS(os, 6)]), VFNMSI(TG, TF), ovs, &(xo[0]));
				   ST(&(xo[WS(os, 8)]), VFNMSI(TE, TB), ovs, &(xo[0]));
				   ST(&(xo[WS(os, 2)]), VFMAI(TE, TB), ovs, &(xo[0]));
				   ST(&(xo[WS(os, 3)]), VFNMSI(Tq, Tp), ovs, &(xo[WS(os, 1)]));
				   ST(&(xo[WS(os, 7)]), VFMAI(Tq, Tp), ovs, &(xo[WS(os, 1)]));
				   ST(&(xo[WS(os, 9)]), VFMAI(To, Tl), ovs, &(xo[WS(os, 1)]));
				   ST(&(xo[WS(os, 1)]), VFNMSI(To, Tl), ovs, &(xo[WS(os, 1)]));
			      }
			 }
		    }
	       }
	  }
     }
     VLEAVE();
}
Example #4
0
static void n1bv_12(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs)
{
    DVK(KP866025403, +0.866025403784438646763723170752936183471402627);
    DVK(KP500000000, +0.500000000000000000000000000000000000000000000);
    {
        INT i;
        const R *xi;
        R *xo;
        xi = ii;
        xo = io;
        for (i = v; i > 0; i = i - VL, xi = xi + (VL * ivs), xo = xo + (VL * ovs), MAKE_VOLATILE_STRIDE(24, is), MAKE_VOLATILE_STRIDE(24, os)) {
            V T1, T6, Tc, Th, Td, Te, Ti, Tz, T4, TA, T9, Tj, Tf, Tw;
            {
                V T2, T3, T7, T8;
                T1 = LD(&(xi[0]), ivs, &(xi[0]));
                T6 = LD(&(xi[WS(is, 6)]), ivs, &(xi[0]));
                T2 = LD(&(xi[WS(is, 4)]), ivs, &(xi[0]));
                T3 = LD(&(xi[WS(is, 8)]), ivs, &(xi[0]));
                T7 = LD(&(xi[WS(is, 10)]), ivs, &(xi[0]));
                T8 = LD(&(xi[WS(is, 2)]), ivs, &(xi[0]));
                Tc = LD(&(xi[WS(is, 3)]), ivs, &(xi[WS(is, 1)]));
                Th = LD(&(xi[WS(is, 9)]), ivs, &(xi[WS(is, 1)]));
                Td = LD(&(xi[WS(is, 7)]), ivs, &(xi[WS(is, 1)]));
                Te = LD(&(xi[WS(is, 11)]), ivs, &(xi[WS(is, 1)]));
                Ti = LD(&(xi[WS(is, 1)]), ivs, &(xi[WS(is, 1)]));
                Tz = VSUB(T2, T3);
                T4 = VADD(T2, T3);
                TA = VSUB(T7, T8);
                T9 = VADD(T7, T8);
                Tj = LD(&(xi[WS(is, 5)]), ivs, &(xi[WS(is, 1)]));
            }
            Tf = VADD(Td, Te);
            Tw = VSUB(Td, Te);
            {
                V T5, Tp, TJ, TB, Ta, Tq, Tk, Tx, Tg, Ts;
                T5 = VADD(T1, T4);
                Tp = VFNMS(LDK(KP500000000), T4, T1);
                TJ = VSUB(Tz, TA);
                TB = VADD(Tz, TA);
                Ta = VADD(T6, T9);
                Tq = VFNMS(LDK(KP500000000), T9, T6);
                Tk = VADD(Ti, Tj);
                Tx = VSUB(Tj, Ti);
                Tg = VADD(Tc, Tf);
                Ts = VFNMS(LDK(KP500000000), Tf, Tc);
                {
                    V Tr, TF, Tb, Tn, TG, Ty, Tl, Tt;
                    Tr = VADD(Tp, Tq);
                    TF = VSUB(Tp, Tq);
                    Tb = VSUB(T5, Ta);
                    Tn = VADD(T5, Ta);
                    TG = VADD(Tw, Tx);
                    Ty = VSUB(Tw, Tx);
                    Tl = VADD(Th, Tk);
                    Tt = VFNMS(LDK(KP500000000), Tk, Th);
                    {
                        V TC, TE, TH, TL, Tu, TI, Tm, To;
                        TC = VMUL(LDK(KP866025403), VSUB(Ty, TB));
                        TE = VMUL(LDK(KP866025403), VADD(TB, Ty));
                        TH = VFNMS(LDK(KP866025403), TG, TF);
                        TL = VFMA(LDK(KP866025403), TG, TF);
                        Tu = VADD(Ts, Tt);
                        TI = VSUB(Ts, Tt);
                        Tm = VSUB(Tg, Tl);
                        To = VADD(Tg, Tl);
                        {
                            V TK, TM, Tv, TD;
                            TK = VFMA(LDK(KP866025403), TJ, TI);
                            TM = VFNMS(LDK(KP866025403), TJ, TI);
                            Tv = VSUB(Tr, Tu);
                            TD = VADD(Tr, Tu);
                            ST(&(xo[0]), VADD(Tn, To), ovs, &(xo[0]));
                            ST(&(xo[WS(os, 6)]), VSUB(Tn, To), ovs, &(xo[0]));
                            ST(&(xo[WS(os, 9)]), VFMAI(Tm, Tb), ovs, &(xo[WS(os, 1)]));
                            ST(&(xo[WS(os, 3)]), VFNMSI(Tm, Tb), ovs, &(xo[WS(os, 1)]));
                            ST(&(xo[WS(os, 5)]), VFMAI(TM, TL), ovs, &(xo[WS(os, 1)]));
                            ST(&(xo[WS(os, 7)]), VFNMSI(TM, TL), ovs, &(xo[WS(os, 1)]));
                            ST(&(xo[WS(os, 11)]), VFNMSI(TK, TH), ovs, &(xo[WS(os, 1)]));
                            ST(&(xo[WS(os, 1)]), VFMAI(TK, TH), ovs, &(xo[WS(os, 1)]));
                            ST(&(xo[WS(os, 8)]), VFNMSI(TE, TD), ovs, &(xo[0]));
                            ST(&(xo[WS(os, 4)]), VFMAI(TE, TD), ovs, &(xo[0]));
                            ST(&(xo[WS(os, 2)]), VFMAI(TC, Tv), ovs, &(xo[0]));
                            ST(&(xo[WS(os, 10)]), VFNMSI(TC, Tv), ovs, &(xo[0]));
                        }
                    }
                }
            }
        }
    }
    VLEAVE();
}
static void t1bv_12(R *ri, R *ii, const R *W, stride rs, INT mb, INT me, INT ms)
{
     DVK(KP866025403, +0.866025403784438646763723170752936183471402627);
     DVK(KP500000000, +0.500000000000000000000000000000000000000000000);
     INT m;
     R *x;
     x = ii;
     for (m = mb, W = W + (mb * ((TWVL / VL) * 22)); m < me; m = m + VL, x = x + (VL * ms), W = W + (TWVL * 22), MAKE_VOLATILE_STRIDE(rs)) {
	  V TI, Ti, TA, T7, Tm, TE, Tw, Tk, Tf, TB, TU, TM;
	  {
	       V T9, TK, Tj, TL, Te;
	       {
		    V T1, T4, T2, Tp, Tt, Tr;
		    T1 = LD(&(x[0]), ms, &(x[0]));
		    T4 = LD(&(x[WS(rs, 8)]), ms, &(x[0]));
		    T2 = LD(&(x[WS(rs, 4)]), ms, &(x[0]));
		    Tp = LD(&(x[WS(rs, 9)]), ms, &(x[WS(rs, 1)]));
		    Tt = LD(&(x[WS(rs, 5)]), ms, &(x[WS(rs, 1)]));
		    Tr = LD(&(x[WS(rs, 1)]), ms, &(x[WS(rs, 1)]));
		    {
			 V T5, T3, Tq, Tu, Ts, Td, Tb, T8, Tc, Ta;
			 T8 = LD(&(x[WS(rs, 6)]), ms, &(x[0]));
			 Tc = LD(&(x[WS(rs, 2)]), ms, &(x[0]));
			 Ta = LD(&(x[WS(rs, 10)]), ms, &(x[0]));
			 T5 = BYTW(&(W[TWVL * 14]), T4);
			 T3 = BYTW(&(W[TWVL * 6]), T2);
			 Tq = BYTW(&(W[TWVL * 16]), Tp);
			 Tu = BYTW(&(W[TWVL * 8]), Tt);
			 Ts = BYTW(&(W[0]), Tr);
			 T9 = BYTW(&(W[TWVL * 10]), T8);
			 Td = BYTW(&(W[TWVL * 2]), Tc);
			 Tb = BYTW(&(W[TWVL * 18]), Ta);
			 {
			      V Th, T6, Tl, Tv;
			      Th = LD(&(x[WS(rs, 3)]), ms, &(x[WS(rs, 1)]));
			      TK = VSUB(T3, T5);
			      T6 = VADD(T3, T5);
			      Tl = LD(&(x[WS(rs, 11)]), ms, &(x[WS(rs, 1)]));
			      Tv = VADD(Ts, Tu);
			      TI = VSUB(Tu, Ts);
			      Tj = LD(&(x[WS(rs, 7)]), ms, &(x[WS(rs, 1)]));
			      TL = VSUB(Tb, Td);
			      Te = VADD(Tb, Td);
			      Ti = BYTW(&(W[TWVL * 4]), Th);
			      TA = VFNMS(LDK(KP500000000), T6, T1);
			      T7 = VADD(T1, T6);
			      Tm = BYTW(&(W[TWVL * 20]), Tl);
			      TE = VFNMS(LDK(KP500000000), Tv, Tq);
			      Tw = VADD(Tq, Tv);
			 }
		    }
	       }
	       Tk = BYTW(&(W[TWVL * 12]), Tj);
	       Tf = VADD(T9, Te);
	       TB = VFNMS(LDK(KP500000000), Te, T9);
	       TU = VSUB(TK, TL);
	       TM = VADD(TK, TL);
	  }
	  {
	       V Tn, TH, TC, TQ, Ty, Tg;
	       Tn = VADD(Tk, Tm);
	       TH = VSUB(Tk, Tm);
	       TC = VADD(TA, TB);
	       TQ = VSUB(TA, TB);
	       Ty = VADD(T7, Tf);
	       Tg = VSUB(T7, Tf);
	       {
		    V To, TD, TJ, TR;
		    To = VADD(Ti, Tn);
		    TD = VFNMS(LDK(KP500000000), Tn, Ti);
		    TJ = VSUB(TH, TI);
		    TR = VADD(TH, TI);
		    {
			 V TP, TN, TW, TS, TO, TG, TX, TV;
			 {
			      V Tz, Tx, TF, TT;
			      Tz = VADD(To, Tw);
			      Tx = VSUB(To, Tw);
			      TF = VADD(TD, TE);
			      TT = VSUB(TD, TE);
			      TP = VMUL(LDK(KP866025403), VADD(TM, TJ));
			      TN = VMUL(LDK(KP866025403), VSUB(TJ, TM));
			      TW = VFMA(LDK(KP866025403), TR, TQ);
			      TS = VFNMS(LDK(KP866025403), TR, TQ);
			      ST(&(x[WS(rs, 6)]), VSUB(Ty, Tz), ms, &(x[0]));
			      ST(&(x[0]), VADD(Ty, Tz), ms, &(x[0]));
			      ST(&(x[WS(rs, 9)]), VFMAI(Tx, Tg), ms, &(x[WS(rs, 1)]));
			      ST(&(x[WS(rs, 3)]), VFNMSI(Tx, Tg), ms, &(x[WS(rs, 1)]));
			      TO = VADD(TC, TF);
			      TG = VSUB(TC, TF);
			      TX = VFNMS(LDK(KP866025403), TU, TT);
			      TV = VFMA(LDK(KP866025403), TU, TT);
			 }
			 ST(&(x[WS(rs, 8)]), VFNMSI(TP, TO), ms, &(x[0]));
			 ST(&(x[WS(rs, 4)]), VFMAI(TP, TO), ms, &(x[0]));
			 ST(&(x[WS(rs, 2)]), VFMAI(TN, TG), ms, &(x[0]));
			 ST(&(x[WS(rs, 10)]), VFNMSI(TN, TG), ms, &(x[0]));
			 ST(&(x[WS(rs, 5)]), VFMAI(TX, TW), ms, &(x[WS(rs, 1)]));
			 ST(&(x[WS(rs, 7)]), VFNMSI(TX, TW), ms, &(x[WS(rs, 1)]));
			 ST(&(x[WS(rs, 11)]), VFNMSI(TV, TS), ms, &(x[WS(rs, 1)]));
			 ST(&(x[WS(rs, 1)]), VFMAI(TV, TS), ms, &(x[WS(rs, 1)]));
		    }
	       }
	  }
     }
}
Example #6
0
static void t3bv_10(R *ri, R *ii, const R *W, stride rs, INT mb, INT me, INT ms)
{
     DVK(KP587785252, +0.587785252292473129168705954639072768597652438);
     DVK(KP951056516, +0.951056516295153572116439333379382143405698634);
     DVK(KP250000000, +0.250000000000000000000000000000000000000000000);
     DVK(KP559016994, +0.559016994374947424102293417182819058860154590);
     INT m;
     R *x;
     x = ii;
     for (m = mb, W = W + (mb * ((TWVL / VL) * 6)); m < me; m = m + VL, x = x + (VL * ms), W = W + (TWVL * 6), MAKE_VOLATILE_STRIDE(rs)) {
	  V T1, T2, T3, Ti, T6, T7, TA, Tb, To;
	  T1 = LDW(&(W[0]));
	  T2 = LDW(&(W[TWVL * 2]));
	  T3 = VZMULJ(T1, T2);
	  Ti = VZMUL(T1, T2);
	  T6 = LDW(&(W[TWVL * 4]));
	  T7 = VZMULJ(T3, T6);
	  TA = VZMULJ(Ti, T6);
	  Tb = VZMULJ(T1, T6);
	  To = VZMULJ(T2, T6);
	  {
	       V TD, TQ, Tn, Tt, Tx, TM, TN, TS, Ta, Tg, Tw, TJ, TK, TR, Tz;
	       V TC, TB;
	       Tz = LD(&(x[0]), ms, &(x[0]));
	       TB = LD(&(x[WS(rs, 5)]), ms, &(x[WS(rs, 1)]));
	       TC = VZMUL(TA, TB);
	       TD = VSUB(Tz, TC);
	       TQ = VADD(Tz, TC);
	       {
		    V Tk, Ts, Tm, Tq;
		    {
			 V Tj, Tr, Tl, Tp;
			 Tj = LD(&(x[WS(rs, 4)]), ms, &(x[0]));
			 Tk = VZMUL(Ti, Tj);
			 Tr = LD(&(x[WS(rs, 1)]), ms, &(x[WS(rs, 1)]));
			 Ts = VZMUL(T1, Tr);
			 Tl = LD(&(x[WS(rs, 9)]), ms, &(x[WS(rs, 1)]));
			 Tm = VZMUL(T6, Tl);
			 Tp = LD(&(x[WS(rs, 6)]), ms, &(x[0]));
			 Tq = VZMUL(To, Tp);
		    }
		    Tn = VSUB(Tk, Tm);
		    Tt = VSUB(Tq, Ts);
		    Tx = VADD(Tn, Tt);
		    TM = VADD(Tk, Tm);
		    TN = VADD(Tq, Ts);
		    TS = VADD(TM, TN);
	       }
	       {
		    V T5, Tf, T9, Td;
		    {
			 V T4, Te, T8, Tc;
			 T4 = LD(&(x[WS(rs, 2)]), ms, &(x[0]));
			 T5 = VZMUL(T3, T4);
			 Te = LD(&(x[WS(rs, 3)]), ms, &(x[WS(rs, 1)]));
			 Tf = VZMUL(T2, Te);
			 T8 = LD(&(x[WS(rs, 7)]), ms, &(x[WS(rs, 1)]));
			 T9 = VZMUL(T7, T8);
			 Tc = LD(&(x[WS(rs, 8)]), ms, &(x[0]));
			 Td = VZMUL(Tb, Tc);
		    }
		    Ta = VSUB(T5, T9);
		    Tg = VSUB(Td, Tf);
		    Tw = VADD(Ta, Tg);
		    TJ = VADD(T5, T9);
		    TK = VADD(Td, Tf);
		    TR = VADD(TJ, TK);
	       }
	       {
		    V Ty, TE, TF, Tv, TI, Th, Tu, TH, TG;
		    Ty = VMUL(LDK(KP559016994), VSUB(Tw, Tx));
		    TE = VADD(Tw, Tx);
		    TF = VFNMS(LDK(KP250000000), TE, TD);
		    Th = VSUB(Ta, Tg);
		    Tu = VSUB(Tn, Tt);
		    Tv = VBYI(VFMA(LDK(KP951056516), Th, VMUL(LDK(KP587785252), Tu)));
		    TI = VBYI(VFNMS(LDK(KP951056516), Tu, VMUL(LDK(KP587785252), Th)));
		    ST(&(x[WS(rs, 5)]), VADD(TD, TE), ms, &(x[WS(rs, 1)]));
		    TH = VSUB(TF, Ty);
		    ST(&(x[WS(rs, 3)]), VSUB(TH, TI), ms, &(x[WS(rs, 1)]));
		    ST(&(x[WS(rs, 7)]), VADD(TI, TH), ms, &(x[WS(rs, 1)]));
		    TG = VADD(Ty, TF);
		    ST(&(x[WS(rs, 1)]), VADD(Tv, TG), ms, &(x[WS(rs, 1)]));
		    ST(&(x[WS(rs, 9)]), VSUB(TG, Tv), ms, &(x[WS(rs, 1)]));
	       }
	       {
		    V TV, TT, TU, TP, TY, TL, TO, TX, TW;
		    TV = VMUL(LDK(KP559016994), VSUB(TR, TS));
		    TT = VADD(TR, TS);
		    TU = VFNMS(LDK(KP250000000), TT, TQ);
		    TL = VSUB(TJ, TK);
		    TO = VSUB(TM, TN);
		    TP = VBYI(VFNMS(LDK(KP951056516), TO, VMUL(LDK(KP587785252), TL)));
		    TY = VBYI(VFMA(LDK(KP951056516), TL, VMUL(LDK(KP587785252), TO)));
		    ST(&(x[0]), VADD(TQ, TT), ms, &(x[0]));
		    TX = VADD(TV, TU);
		    ST(&(x[WS(rs, 4)]), VSUB(TX, TY), ms, &(x[0]));
		    ST(&(x[WS(rs, 6)]), VADD(TY, TX), ms, &(x[0]));
		    TW = VSUB(TU, TV);
		    ST(&(x[WS(rs, 2)]), VADD(TP, TW), ms, &(x[0]));
		    ST(&(x[WS(rs, 8)]), VSUB(TW, TP), ms, &(x[0]));
	       }
	  }
     }
}
Example #7
0
static void t1fuv_7(R *ri, R *ii, const R *W, stride rs, INT mb, INT me, INT ms)
{
     DVK(KP900968867, +0.900968867902419126236102319507445051165919162);
     DVK(KP801937735, +0.801937735804838252472204639014890102331838324);
     DVK(KP974927912, +0.974927912181823607018131682993931217232785801);
     DVK(KP692021471, +0.692021471630095869627814897002069140197260599);
     DVK(KP554958132, +0.554958132087371191422194871006410481067288862);
     DVK(KP356895867, +0.356895867892209443894399510021300583399127187);
     INT m;
     R *x;
     x = ri;
     for (m = mb, W = W + (mb * ((TWVL / VL) * 12)); m < me; m = m + VL, x = x + (VL * ms), W = W + (TWVL * 12), MAKE_VOLATILE_STRIDE(rs)) {
	  V T1, T2, T4, Te, Tc, T9, T7;
	  T1 = LD(&(x[0]), ms, &(x[0]));
	  T2 = LD(&(x[WS(rs, 1)]), ms, &(x[WS(rs, 1)]));
	  T4 = LD(&(x[WS(rs, 6)]), ms, &(x[0]));
	  Te = LD(&(x[WS(rs, 4)]), ms, &(x[0]));
	  Tc = LD(&(x[WS(rs, 3)]), ms, &(x[WS(rs, 1)]));
	  T9 = LD(&(x[WS(rs, 5)]), ms, &(x[WS(rs, 1)]));
	  T7 = LD(&(x[WS(rs, 2)]), ms, &(x[0]));
	  {
	       V T3, T5, Tf, Td, Ta, T8;
	       T3 = BYTWJ(&(W[0]), T2);
	       T5 = BYTWJ(&(W[TWVL * 10]), T4);
	       Tf = BYTWJ(&(W[TWVL * 6]), Te);
	       Td = BYTWJ(&(W[TWVL * 4]), Tc);
	       Ta = BYTWJ(&(W[TWVL * 8]), T9);
	       T8 = BYTWJ(&(W[TWVL * 2]), T7);
	       {
		    V T6, Tk, Tg, Tl, Tb, Tm;
		    T6 = VADD(T3, T5);
		    Tk = VSUB(T5, T3);
		    Tg = VADD(Td, Tf);
		    Tl = VSUB(Tf, Td);
		    Tb = VADD(T8, Ta);
		    Tm = VSUB(Ta, T8);
		    {
			 V Th, Ts, Tp, Tu, Tn, Tx, Ti, Tt;
			 Th = VFNMS(LDK(KP356895867), T6, Tg);
			 Ts = VFMA(LDK(KP554958132), Tl, Tk);
			 ST(&(x[0]), VADD(T1, VADD(T6, VADD(Tb, Tg))), ms, &(x[0]));
			 Tp = VFNMS(LDK(KP356895867), Tb, T6);
			 Tu = VFNMS(LDK(KP356895867), Tg, Tb);
			 Tn = VFMA(LDK(KP554958132), Tm, Tl);
			 Tx = VFNMS(LDK(KP554958132), Tk, Tm);
			 Ti = VFNMS(LDK(KP692021471), Th, Tb);
			 Tt = VMUL(LDK(KP974927912), VFMA(LDK(KP801937735), Ts, Tm));
			 {
			      V Tq, Tv, To, Ty, Tj, Tr, Tw;
			      Tq = VFNMS(LDK(KP692021471), Tp, Tg);
			      Tv = VFNMS(LDK(KP692021471), Tu, T6);
			      To = VMUL(LDK(KP974927912), VFNMS(LDK(KP801937735), Tn, Tk));
			      Ty = VMUL(LDK(KP974927912), VFNMS(LDK(KP801937735), Tx, Tl));
			      Tj = VFNMS(LDK(KP900968867), Ti, T1);
			      Tr = VFNMS(LDK(KP900968867), Tq, T1);
			      Tw = VFNMS(LDK(KP900968867), Tv, T1);
			      ST(&(x[WS(rs, 2)]), VFMAI(To, Tj), ms, &(x[0]));
			      ST(&(x[WS(rs, 5)]), VFNMSI(To, Tj), ms, &(x[WS(rs, 1)]));
			      ST(&(x[WS(rs, 1)]), VFMAI(Tt, Tr), ms, &(x[WS(rs, 1)]));
			      ST(&(x[WS(rs, 6)]), VFNMSI(Tt, Tr), ms, &(x[0]));
			      ST(&(x[WS(rs, 3)]), VFMAI(Ty, Tw), ms, &(x[WS(rs, 1)]));
			      ST(&(x[WS(rs, 4)]), VFNMSI(Ty, Tw), ms, &(x[0]));
			 }
		    }
	       }
	  }
     }
}
static void n1fv_9(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs)
{
     DVK(KP939692620, +0.939692620785908384054109277324731469936208134);
     DVK(KP826351822, +0.826351822333069651148283373230685203999624323);
     DVK(KP879385241, +0.879385241571816768108218554649462939872416269);
     DVK(KP984807753, +0.984807753012208059366743024589523013670643252);
     DVK(KP666666666, +0.666666666666666666666666666666666666666666667);
     DVK(KP852868531, +0.852868531952443209628250963940074071936020296);
     DVK(KP907603734, +0.907603734547952313649323976213898122064543220);
     DVK(KP420276625, +0.420276625461206169731530603237061658838781920);
     DVK(KP673648177, +0.673648177666930348851716626769314796000375677);
     DVK(KP898197570, +0.898197570222573798468955502359086394667167570);
     DVK(KP347296355, +0.347296355333860697703433253538629592000751354);
     DVK(KP866025403, +0.866025403784438646763723170752936183471402627);
     DVK(KP439692620, +0.439692620785908384054109277324731469936208134);
     DVK(KP203604859, +0.203604859554852403062088995281827210665664861);
     DVK(KP152703644, +0.152703644666139302296566746461370407999248646);
     DVK(KP586256827, +0.586256827714544512072145703099641959914944179);
     DVK(KP968908795, +0.968908795874236621082202410917456709164223497);
     DVK(KP726681596, +0.726681596905677465811651808188092531873167623);
     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(is), MAKE_VOLATILE_STRIDE(os)) {
	  V T1, T2, T3, T6, Tb, T7, T8, Tc, Td, Tv, T4;
	  T1 = LD(&(xi[0]), ivs, &(xi[0]));
	  T2 = LD(&(xi[WS(is, 3)]), ivs, &(xi[WS(is, 1)]));
	  T3 = LD(&(xi[WS(is, 6)]), ivs, &(xi[0]));
	  T6 = LD(&(xi[WS(is, 1)]), ivs, &(xi[WS(is, 1)]));
	  Tb = LD(&(xi[WS(is, 2)]), ivs, &(xi[0]));
	  T7 = LD(&(xi[WS(is, 4)]), ivs, &(xi[0]));
	  T8 = LD(&(xi[WS(is, 7)]), ivs, &(xi[WS(is, 1)]));
	  Tc = LD(&(xi[WS(is, 5)]), ivs, &(xi[WS(is, 1)]));
	  Td = LD(&(xi[WS(is, 8)]), ivs, &(xi[0]));
	  Tv = VSUB(T3, T2);
	  T4 = VADD(T2, T3);
	  {
	       V Tl, T9, Tm, Te, Tj, T5;
	       Tl = VSUB(T7, T8);
	       T9 = VADD(T7, T8);
	       Tm = VSUB(Td, Tc);
	       Te = VADD(Tc, Td);
	       Tj = VFNMS(LDK(KP500000000), T4, T1);
	       T5 = VADD(T1, T4);
	       {
		    V Tn, Ta, Tk, Tf;
		    Tn = VFNMS(LDK(KP500000000), T9, T6);
		    Ta = VADD(T6, T9);
		    Tk = VFNMS(LDK(KP500000000), Te, Tb);
		    Tf = VADD(Tb, Te);
		    {
			 V Ty, TC, To, TB, Tx, Ts, Tg, Ti;
			 Ty = VFNMS(LDK(KP726681596), Tl, Tn);
			 TC = VFMA(LDK(KP968908795), Tn, Tl);
			 To = VFNMS(LDK(KP586256827), Tn, Tm);
			 TB = VFNMS(LDK(KP152703644), Tm, Tk);
			 Tx = VFMA(LDK(KP203604859), Tk, Tm);
			 Ts = VFNMS(LDK(KP439692620), Tl, Tk);
			 Tg = VADD(Ta, Tf);
			 Ti = VMUL(LDK(KP866025403), VSUB(Tf, Ta));
			 {
			      V Tz, TI, TF, TD, Tt, Th, Tq, Tp;
			      Tp = VFNMS(LDK(KP347296355), To, Tl);
			      Tz = VFMA(LDK(KP898197570), Ty, Tx);
			      TI = VFNMS(LDK(KP898197570), Ty, Tx);
			      TF = VFNMS(LDK(KP673648177), TC, TB);
			      TD = VFMA(LDK(KP673648177), TC, TB);
			      Tt = VFNMS(LDK(KP420276625), Ts, Tm);
			      ST(&(xo[0]), VADD(T5, Tg), ovs, &(xo[0]));
			      Th = VFNMS(LDK(KP500000000), Tg, T5);
			      Tq = VFNMS(LDK(KP907603734), Tp, Tk);
			      {
				   V TA, TJ, TE, TG, Tu, Tr, TK, TH, Tw;
				   TA = VFMA(LDK(KP852868531), Tz, Tj);
				   TJ = VFMA(LDK(KP666666666), TD, TI);
				   TE = VMUL(LDK(KP984807753), VFNMS(LDK(KP879385241), Tv, TD));
				   TG = VFNMS(LDK(KP500000000), Tz, TF);
				   Tu = VFNMS(LDK(KP826351822), Tt, Tn);
				   ST(&(xo[WS(os, 6)]), VFNMSI(Ti, Th), ovs, &(xo[0]));
				   ST(&(xo[WS(os, 3)]), VFMAI(Ti, Th), ovs, &(xo[WS(os, 1)]));
				   Tr = VFNMS(LDK(KP939692620), Tq, Tj);
				   TK = VMUL(LDK(KP866025403), VFMA(LDK(KP852868531), TJ, Tv));
				   ST(&(xo[WS(os, 8)]), VFMAI(TE, TA), ovs, &(xo[0]));
				   ST(&(xo[WS(os, 1)]), VFNMSI(TE, TA), ovs, &(xo[WS(os, 1)]));
				   TH = VFMA(LDK(KP852868531), TG, Tj);
				   Tw = VMUL(LDK(KP984807753), VFMA(LDK(KP879385241), Tv, Tu));
				   ST(&(xo[WS(os, 4)]), VFMAI(TK, TH), ovs, &(xo[0]));
				   ST(&(xo[WS(os, 5)]), VFNMSI(TK, TH), ovs, &(xo[WS(os, 1)]));
				   ST(&(xo[WS(os, 7)]), VFMAI(Tw, Tr), ovs, &(xo[WS(os, 1)]));
				   ST(&(xo[WS(os, 2)]), VFNMSI(Tw, Tr), ovs, &(xo[0]));
			      }
			 }
		    }
	       }
	  }
     }
}
Example #9
0
RETf INC( float &x, const __m128 y ) { __m128 t=ADD(LD(x),y); return STR(x,t); }
Example #10
0
static void t1sv_4(R *ri, R *ii, const R *W, stride rs, INT mb, INT me, INT ms)
{
     {
	  INT m;
	  for (m = mb, W = W + (mb * 6); m < me; m = m + (2 * VL), ri = ri + ((2 * VL) * ms), ii = ii + ((2 * VL) * ms), W = W + ((2 * VL) * 6), MAKE_VOLATILE_STRIDE(rs)) {
	       V T1, Tv, T3, T6, T5, Ta, Td, Tc, Tg, Tj, Tt, T4, Tf, Ti, Tn;
	       V Tb, T2, T9;
	       T1 = LD(&(ri[0]), ms, &(ri[0]));
	       Tv = LD(&(ii[0]), ms, &(ii[0]));
	       T3 = LD(&(ri[WS(rs, 2)]), ms, &(ri[0]));
	       T6 = LD(&(ii[WS(rs, 2)]), ms, &(ii[0]));
	       T2 = LDW(&(W[TWVL * 2]));
	       T5 = LDW(&(W[TWVL * 3]));
	       Ta = LD(&(ri[WS(rs, 1)]), ms, &(ri[WS(rs, 1)]));
	       Td = LD(&(ii[WS(rs, 1)]), ms, &(ii[WS(rs, 1)]));
	       T9 = LDW(&(W[0]));
	       Tc = LDW(&(W[TWVL * 1]));
	       Tg = LD(&(ri[WS(rs, 3)]), ms, &(ri[WS(rs, 1)]));
	       Tj = LD(&(ii[WS(rs, 3)]), ms, &(ii[WS(rs, 1)]));
	       Tt = VMUL(T2, T6);
	       T4 = VMUL(T2, T3);
	       Tf = LDW(&(W[TWVL * 4]));
	       Ti = LDW(&(W[TWVL * 5]));
	       Tn = VMUL(T9, Td);
	       Tb = VMUL(T9, Ta);
	       {
		    V Tu, T7, Tp, Th, To, Te;
		    Tu = VFNMS(T5, T3, Tt);
		    T7 = VFMA(T5, T6, T4);
		    Tp = VMUL(Tf, Tj);
		    Th = VMUL(Tf, Tg);
		    To = VFNMS(Tc, Ta, Tn);
		    Te = VFMA(Tc, Td, Tb);
		    {
			 V Tw, Tx, T8, Tm, Tq, Tk;
			 Tw = VADD(Tu, Tv);
			 Tx = VSUB(Tv, Tu);
			 T8 = VADD(T1, T7);
			 Tm = VSUB(T1, T7);
			 Tq = VFNMS(Ti, Tg, Tp);
			 Tk = VFMA(Ti, Tj, Th);
			 {
			      V Ts, Tr, Tl, Ty;
			      Ts = VADD(To, Tq);
			      Tr = VSUB(To, Tq);
			      Tl = VADD(Te, Tk);
			      Ty = VSUB(Te, Tk);
			      ST(&(ri[WS(rs, 1)]), VADD(Tm, Tr), ms, &(ri[WS(rs, 1)]));
			      ST(&(ri[WS(rs, 3)]), VSUB(Tm, Tr), ms, &(ri[WS(rs, 1)]));
			      ST(&(ii[WS(rs, 2)]), VSUB(Tw, Ts), ms, &(ii[0]));
			      ST(&(ii[0]), VADD(Ts, Tw), ms, &(ii[0]));
			      ST(&(ii[WS(rs, 3)]), VADD(Ty, Tx), ms, &(ii[WS(rs, 1)]));
			      ST(&(ii[WS(rs, 1)]), VSUB(Tx, Ty), ms, &(ii[WS(rs, 1)]));
			      ST(&(ri[0]), VADD(T8, Tl), ms, &(ri[0]));
			      ST(&(ri[WS(rs, 2)]), VSUB(T8, Tl), ms, &(ri[0]));
			 }
		    }
	       }
	  }
     }
     VLEAVE();
}
static void n1fv_9(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs)
{
     DVK(KP342020143, +0.342020143325668733044099614682259580763083368);
     DVK(KP813797681, +0.813797681349373692844693217248393223289101568);
     DVK(KP939692620, +0.939692620785908384054109277324731469936208134);
     DVK(KP296198132, +0.296198132726023843175338011893050938967728390);
     DVK(KP642787609, +0.642787609686539326322643409907263432907559884);
     DVK(KP663413948, +0.663413948168938396205421319635891297216863310);
     DVK(KP556670399, +0.556670399226419366452912952047023132968291906);
     DVK(KP766044443, +0.766044443118978035202392650555416673935832457);
     DVK(KP984807753, +0.984807753012208059366743024589523013670643252);
     DVK(KP150383733, +0.150383733180435296639271897612501926072238258);
     DVK(KP852868531, +0.852868531952443209628250963940074071936020296);
     DVK(KP173648177, +0.173648177666930348851716626769314796000375677);
     DVK(KP500000000, +0.500000000000000000000000000000000000000000000);
     DVK(KP866025403, +0.866025403784438646763723170752936183471402627);
     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 T5, Ts, Tj, To, Tf, Tn, Tp, Tu, Tl, Ta, Tk, Tm, Tt;
	  {
	       V T1, T2, T3, T4;
	       T1 = LD(&(xi[0]), ivs, &(xi[0]));
	       T2 = LD(&(xi[WS(is, 3)]), ivs, &(xi[WS(is, 1)]));
	       T3 = LD(&(xi[WS(is, 6)]), ivs, &(xi[0]));
	       T4 = VADD(T2, T3);
	       T5 = VADD(T1, T4);
	       Ts = VMUL(LDK(KP866025403), VSUB(T3, T2));
	       Tj = VFNMS(LDK(KP500000000), T4, T1);
	  }
	  {
	       V Tb, Te, Tc, Td;
	       Tb = LD(&(xi[WS(is, 2)]), ivs, &(xi[0]));
	       Tc = LD(&(xi[WS(is, 5)]), ivs, &(xi[WS(is, 1)]));
	       Td = LD(&(xi[WS(is, 8)]), ivs, &(xi[0]));
	       Te = VADD(Tc, Td);
	       To = VSUB(Td, Tc);
	       Tf = VADD(Tb, Te);
	       Tn = VFNMS(LDK(KP500000000), Te, Tb);
	       Tp = VFMA(LDK(KP173648177), Tn, VMUL(LDK(KP852868531), To));
	       Tu = VFNMS(LDK(KP984807753), Tn, VMUL(LDK(KP150383733), To));
	  }
	  {
	       V T6, T9, T7, T8;
	       T6 = LD(&(xi[WS(is, 1)]), ivs, &(xi[WS(is, 1)]));
	       T7 = LD(&(xi[WS(is, 4)]), ivs, &(xi[0]));
	       T8 = LD(&(xi[WS(is, 7)]), ivs, &(xi[WS(is, 1)]));
	       T9 = VADD(T7, T8);
	       Tl = VSUB(T8, T7);
	       Ta = VADD(T6, T9);
	       Tk = VFNMS(LDK(KP500000000), T9, T6);
	       Tm = VFMA(LDK(KP766044443), Tk, VMUL(LDK(KP556670399), Tl));
	       Tt = VFNMS(LDK(KP642787609), Tk, VMUL(LDK(KP663413948), Tl));
	  }
	  {
	       V Ti, Tg, Th, Tz, TA;
	       Ti = VBYI(VMUL(LDK(KP866025403), VSUB(Tf, Ta)));
	       Tg = VADD(Ta, Tf);
	       Th = VFNMS(LDK(KP500000000), Tg, T5);
	       ST(&(xo[0]), VADD(T5, Tg), ovs, &(xo[0]));
	       ST(&(xo[WS(os, 3)]), VADD(Th, Ti), ovs, &(xo[WS(os, 1)]));
	       ST(&(xo[WS(os, 6)]), VSUB(Th, Ti), ovs, &(xo[0]));
	       Tz = VFMA(LDK(KP173648177), Tk, VFNMS(LDK(KP296198132), To, VFNMS(LDK(KP939692620), Tn, VFNMS(LDK(KP852868531), Tl, Tj))));
	       TA = VBYI(VSUB(VFNMS(LDK(KP342020143), Tn, VFNMS(LDK(KP150383733), Tl, VFNMS(LDK(KP984807753), Tk, VMUL(LDK(KP813797681), To)))), Ts));
	       ST(&(xo[WS(os, 7)]), VSUB(Tz, TA), ovs, &(xo[WS(os, 1)]));
	       ST(&(xo[WS(os, 2)]), VADD(Tz, TA), ovs, &(xo[0]));
	       {
		    V Tr, Tx, Tw, Ty, Tq, Tv;
		    Tq = VADD(Tm, Tp);
		    Tr = VADD(Tj, Tq);
		    Tx = VFMA(LDK(KP866025403), VSUB(Tt, Tu), VFNMS(LDK(KP500000000), Tq, Tj));
		    Tv = VADD(Tt, Tu);
		    Tw = VBYI(VADD(Ts, Tv));
		    Ty = VBYI(VADD(Ts, VFNMS(LDK(KP500000000), Tv, VMUL(LDK(KP866025403), VSUB(Tp, Tm)))));
		    ST(&(xo[WS(os, 8)]), VSUB(Tr, Tw), ovs, &(xo[0]));
		    ST(&(xo[WS(os, 4)]), VADD(Tx, Ty), ovs, &(xo[0]));
		    ST(&(xo[WS(os, 1)]), VADD(Tw, Tr), ovs, &(xo[WS(os, 1)]));
		    ST(&(xo[WS(os, 5)]), VSUB(Tx, Ty), ovs, &(xo[WS(os, 1)]));
	       }
	  }
     }
}
Example #12
0
static void t1sv_4(R *ri, R *ii, const R *W, stride rs, INT mb, INT me, INT ms)
{
     {
	  INT m;
	  for (m = mb, W = W + (mb * 6); m < me; m = m + (2 * VL), ri = ri + ((2 * VL) * ms), ii = ii + ((2 * VL) * ms), W = W + ((2 * VL) * 6), MAKE_VOLATILE_STRIDE(rs)) {
	       V T1, Tp, T6, To, Tc, Tk, Th, Tl;
	       T1 = LD(&(ri[0]), ms, &(ri[0]));
	       Tp = LD(&(ii[0]), ms, &(ii[0]));
	       {
		    V T3, T5, T2, T4;
		    T3 = LD(&(ri[WS(rs, 2)]), ms, &(ri[0]));
		    T5 = LD(&(ii[WS(rs, 2)]), ms, &(ii[0]));
		    T2 = LDW(&(W[TWVL * 2]));
		    T4 = LDW(&(W[TWVL * 3]));
		    T6 = VFMA(T2, T3, VMUL(T4, T5));
		    To = VFNMS(T4, T3, VMUL(T2, T5));
	       }
	       {
		    V T9, Tb, T8, Ta;
		    T9 = LD(&(ri[WS(rs, 1)]), ms, &(ri[WS(rs, 1)]));
		    Tb = LD(&(ii[WS(rs, 1)]), ms, &(ii[WS(rs, 1)]));
		    T8 = LDW(&(W[0]));
		    Ta = LDW(&(W[TWVL * 1]));
		    Tc = VFMA(T8, T9, VMUL(Ta, Tb));
		    Tk = VFNMS(Ta, T9, VMUL(T8, Tb));
	       }
	       {
		    V Te, Tg, Td, Tf;
		    Te = LD(&(ri[WS(rs, 3)]), ms, &(ri[WS(rs, 1)]));
		    Tg = LD(&(ii[WS(rs, 3)]), ms, &(ii[WS(rs, 1)]));
		    Td = LDW(&(W[TWVL * 4]));
		    Tf = LDW(&(W[TWVL * 5]));
		    Th = VFMA(Td, Te, VMUL(Tf, Tg));
		    Tl = VFNMS(Tf, Te, VMUL(Td, Tg));
	       }
	       {
		    V T7, Ti, Tn, Tq;
		    T7 = VADD(T1, T6);
		    Ti = VADD(Tc, Th);
		    ST(&(ri[WS(rs, 2)]), VSUB(T7, Ti), ms, &(ri[0]));
		    ST(&(ri[0]), VADD(T7, Ti), ms, &(ri[0]));
		    Tn = VADD(Tk, Tl);
		    Tq = VADD(To, Tp);
		    ST(&(ii[0]), VADD(Tn, Tq), ms, &(ii[0]));
		    ST(&(ii[WS(rs, 2)]), VSUB(Tq, Tn), ms, &(ii[0]));
	       }
	       {
		    V Tj, Tm, Tr, Ts;
		    Tj = VSUB(T1, T6);
		    Tm = VSUB(Tk, Tl);
		    ST(&(ri[WS(rs, 3)]), VSUB(Tj, Tm), ms, &(ri[WS(rs, 1)]));
		    ST(&(ri[WS(rs, 1)]), VADD(Tj, Tm), ms, &(ri[WS(rs, 1)]));
		    Tr = VSUB(Tp, To);
		    Ts = VSUB(Tc, Th);
		    ST(&(ii[WS(rs, 1)]), VSUB(Tr, Ts), ms, &(ii[WS(rs, 1)]));
		    ST(&(ii[WS(rs, 3)]), VADD(Ts, Tr), ms, &(ii[WS(rs, 1)]));
	       }
	  }
     }
     VLEAVE();
}
Example #13
0
SprxPatch emulator_api_patches[] =
{
	// Read umd patches
	{ psp_read, STDU(SP, 0xFF90, SP), &condition_psp_iso },
	{ psp_read+4, MFLR(R0), &condition_psp_iso },
	{ psp_read+8, STD(R0, 0x80, SP), &condition_psp_iso },
	{ psp_read+0x0C, MR(R8, R7), &condition_psp_iso },
	{ psp_read+0x10, MR(R7, R6), &condition_psp_iso },
	{ psp_read+0x14, MR(R6, R5), &condition_psp_iso },
	{ psp_read+0x18, MR(R5, R4), &condition_psp_iso },
	{ psp_read+0x1C, MR(R4, R3), &condition_psp_iso },
	{ psp_read+0x20, LI(R3, SYSCALL8_OPCODE_READ_PSP_UMD), &condition_psp_iso },
	{ psp_read+0x24, LI(R11, 8), &condition_psp_iso },
	{ psp_read+0x28, SC, &condition_psp_iso },
	{ psp_read+0x2C, LD(R0, 0x80, SP), &condition_psp_iso },
	{ psp_read+0x30, MTLR(R0), &condition_psp_iso },
	{ psp_read+0x34, ADDI(SP, SP, 0x70), &condition_psp_iso },
	{ psp_read+0x38, BLR, &condition_psp_iso },

	// Read header patches
	{ psp_read+0x3C, STDU(SP, 0xFF90, SP), &condition_psp_iso },
	{ psp_read+0x40, MFLR(R0), &condition_psp_iso },
	{ psp_read+0x44, STD(R0, 0x80, SP), &condition_psp_iso },
	{ psp_read+0x48, MR(R7, R6), &condition_psp_iso },
	{ psp_read+0x4C, MR(R6, R5), &condition_psp_iso },
	{ psp_read+0x50, MR(R5, R4), &condition_psp_iso },
	{ psp_read+0x54, MR(R4, R3), &condition_psp_iso },
	{ psp_read+0x58, LI(R3, SYSCALL8_OPCODE_READ_PSP_HEADER), &condition_psp_iso },
	{ psp_read+0x5C, LI(R11, 8), &condition_psp_iso },
	{ psp_read+0x60, SC, &condition_psp_iso },
static void t2sv_4(R *ri, R *ii, const R *W, stride rs, INT mb, INT me, INT ms)
{
     {
	  INT m;
	  for (m = mb, W = W + (mb * 4); m < me; m = m + (2 * VL), ri = ri + ((2 * VL) * ms), ii = ii + ((2 * VL) * ms), W = W + ((2 * VL) * 4), MAKE_VOLATILE_STRIDE(8, rs)) {
	       V T2, T6, T3, T5, T1, Tx, T8, Tc, Tf, Ta, T4, Th, Tj, Tl;
	       T2 = LDW(&(W[0]));
	       T6 = LDW(&(W[TWVL * 3]));
	       T3 = LDW(&(W[TWVL * 2]));
	       T5 = LDW(&(W[TWVL * 1]));
	       T1 = LD(&(ri[0]), ms, &(ri[0]));
	       Tx = LD(&(ii[0]), ms, &(ii[0]));
	       T8 = LD(&(ri[WS(rs, 2)]), ms, &(ri[0]));
	       Tc = LD(&(ii[WS(rs, 2)]), ms, &(ii[0]));
	       Tf = LD(&(ri[WS(rs, 1)]), ms, &(ri[WS(rs, 1)]));
	       Ta = VMUL(T2, T6);
	       T4 = VMUL(T2, T3);
	       Th = LD(&(ii[WS(rs, 1)]), ms, &(ii[WS(rs, 1)]));
	       Tj = LD(&(ri[WS(rs, 3)]), ms, &(ri[WS(rs, 1)]));
	       Tl = LD(&(ii[WS(rs, 3)]), ms, &(ii[WS(rs, 1)]));
	       {
		    V Tg, Tb, T7, Tp, Tk, Tr, Ti;
		    Tg = VMUL(T2, Tf);
		    Tb = VFNMS(T5, T3, Ta);
		    T7 = VFMA(T5, T6, T4);
		    Tp = VMUL(T2, Th);
		    Tk = VMUL(T3, Tj);
		    Tr = VMUL(T3, Tl);
		    Ti = VFMA(T5, Th, Tg);
		    {
			 V Tv, T9, Tq, Tm, Ts, Tw, Td;
			 Tv = VMUL(T7, Tc);
			 T9 = VMUL(T7, T8);
			 Tq = VFNMS(T5, Tf, Tp);
			 Tm = VFMA(T6, Tl, Tk);
			 Ts = VFNMS(T6, Tj, Tr);
			 Tw = VFNMS(Tb, T8, Tv);
			 Td = VFMA(Tb, Tc, T9);
			 {
			      V Tn, TA, Tu, Tt;
			      Tn = VADD(Ti, Tm);
			      TA = VSUB(Ti, Tm);
			      Tu = VADD(Tq, Ts);
			      Tt = VSUB(Tq, Ts);
			      {
				   V Ty, Tz, Te, To;
				   Ty = VADD(Tw, Tx);
				   Tz = VSUB(Tx, Tw);
				   Te = VADD(T1, Td);
				   To = VSUB(T1, Td);
				   ST(&(ii[WS(rs, 3)]), VADD(TA, Tz), ms, &(ii[WS(rs, 1)]));
				   ST(&(ii[WS(rs, 1)]), VSUB(Tz, TA), ms, &(ii[WS(rs, 1)]));
				   ST(&(ii[WS(rs, 2)]), VSUB(Ty, Tu), ms, &(ii[0]));
				   ST(&(ii[0]), VADD(Tu, Ty), ms, &(ii[0]));
				   ST(&(ri[WS(rs, 1)]), VADD(To, Tt), ms, &(ri[WS(rs, 1)]));
				   ST(&(ri[WS(rs, 3)]), VSUB(To, Tt), ms, &(ri[WS(rs, 1)]));
				   ST(&(ri[0]), VADD(Te, Tn), ms, &(ri[0]));
				   ST(&(ri[WS(rs, 2)]), VSUB(Te, Tn), ms, &(ri[0]));
			      }
			 }
		    }
	       }
	  }
     }
     VLEAVE();
}
static void n2fv_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;
     BEGIN_SIMD();
     for (i = v; i > 0; i = i - VL, xi = xi + (VL * ivs), xo = xo + (VL * ovs)) {
	  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[2]), VADD(TF, T10), ovs, &(xo[2]));
		    ST(&(xo[24]), 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[10]), VADD(T15, T16), ovs, &(xo[2]));
			 ST(&(xo[16]), VSUB(T16, T15), ovs, &(xo[0]));
			 T1p = VADD(T1n, T1m);
			 T1q = VBYI(VADD(T1j, T1k));
			 ST(&(xo[8]), VSUB(T1p, T1q), ovs, &(xo[0]));
			 ST(&(xo[18]), VADD(T1q, T1p), ovs, &(xo[2]));
		    }
		    T1l = VBYI(VSUB(T1j, T1k));
		    T1o = VSUB(T1m, T1n);
		    ST(&(xo[6]), VADD(T1l, T1o), ovs, &(xo[2]));
		    ST(&(xo[20]), VSUB(T1o, T1l), ovs, &(xo[0]));
		    {
			 V T1h, T1i, T1d, T1g;
			 T1h = VBYI(VSUB(T1e, T1f));
			 T1i = VSUB(T1c, T19);
			 ST(&(xo[12]), VADD(T1h, T1i), ovs, &(xo[0]));
			 ST(&(xo[14]), VSUB(T1i, T1h), ovs, &(xo[2]));
			 T1d = VADD(T19, T1c);
			 T1g = VBYI(VADD(T1e, T1f));
			 ST(&(xo[4]), VSUB(T1d, T1g), ovs, &(xo[0]));
			 ST(&(xo[22]), VADD(T1g, T1d), ovs, &(xo[2]));
		    }
	       }
	  }
     }
     END_SIMD();
}
Example #16
0
RETf DEC( float &x, const __m128 y ) { __m128 t=SUB(LD(x),y); return STR(x,t); }
Example #17
0
static void n1bv_7(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs)
{
     DVK(KP900968867, +0.900968867902419126236102319507445051165919162);
     DVK(KP692021471, +0.692021471630095869627814897002069140197260599);
     DVK(KP801937735, +0.801937735804838252472204639014890102331838324);
     DVK(KP974927912, +0.974927912181823607018131682993931217232785801);
     DVK(KP356895867, +0.356895867892209443894399510021300583399127187);
     DVK(KP554958132, +0.554958132087371191422194871006410481067288862);
     {
	  INT i;
	  const R *xi;
	  R *xo;
	  xi = ii;
	  xo = io;
	  for (i = v; i > 0; i = i - VL, xi = xi + (VL * ivs), xo = xo + (VL * ovs), MAKE_VOLATILE_STRIDE(is), MAKE_VOLATILE_STRIDE(os)) {
	       V T1, T2, T3, T8, T9, T5, T6;
	       T1 = LD(&(xi[0]), ivs, &(xi[0]));
	       T2 = LD(&(xi[WS(is, 1)]), ivs, &(xi[WS(is, 1)]));
	       T3 = LD(&(xi[WS(is, 6)]), ivs, &(xi[0]));
	       T8 = LD(&(xi[WS(is, 3)]), ivs, &(xi[WS(is, 1)]));
	       T9 = LD(&(xi[WS(is, 4)]), ivs, &(xi[0]));
	       T5 = LD(&(xi[WS(is, 2)]), ivs, &(xi[0]));
	       T6 = LD(&(xi[WS(is, 5)]), ivs, &(xi[WS(is, 1)]));
	       {
		    V Tg, T4, Te, Ta, Tf, T7;
		    Tg = VSUB(T2, T3);
		    T4 = VADD(T2, T3);
		    Te = VSUB(T8, T9);
		    Ta = VADD(T8, T9);
		    Tf = VSUB(T5, T6);
		    T7 = VADD(T5, T6);
		    {
			 V Tr, Tj, Tm, Th, To, Tb;
			 Tr = VFMA(LDK(KP554958132), Te, Tg);
			 Tj = VFNMS(LDK(KP356895867), T4, Ta);
			 Tm = VFMA(LDK(KP554958132), Tf, Te);
			 Th = VFNMS(LDK(KP554958132), Tg, Tf);
			 ST(&(xo[0]), VADD(T1, VADD(T4, VADD(T7, Ta))), ovs, &(xo[0]));
			 To = VFNMS(LDK(KP356895867), T7, T4);
			 Tb = VFNMS(LDK(KP356895867), Ta, T7);
			 {
			      V Ts, Tk, Tn, Ti;
			      Ts = VMUL(LDK(KP974927912), VFMA(LDK(KP801937735), Tr, Tf));
			      Tk = VFNMS(LDK(KP692021471), Tj, T7);
			      Tn = VMUL(LDK(KP974927912), VFNMS(LDK(KP801937735), Tm, Tg));
			      Ti = VMUL(LDK(KP974927912), VFNMS(LDK(KP801937735), Th, Te));
			      {
				   V Tp, Tc, Tl, Tq, Td;
				   Tp = VFNMS(LDK(KP692021471), To, Ta);
				   Tc = VFNMS(LDK(KP692021471), Tb, T4);
				   Tl = VFNMS(LDK(KP900968867), Tk, T1);
				   Tq = VFNMS(LDK(KP900968867), Tp, T1);
				   Td = VFNMS(LDK(KP900968867), Tc, T1);
				   ST(&(xo[WS(os, 5)]), VFNMSI(Tn, Tl), ovs, &(xo[WS(os, 1)]));
				   ST(&(xo[WS(os, 2)]), VFMAI(Tn, Tl), ovs, &(xo[0]));
				   ST(&(xo[WS(os, 6)]), VFNMSI(Ts, Tq), ovs, &(xo[0]));
				   ST(&(xo[WS(os, 1)]), VFMAI(Ts, Tq), ovs, &(xo[WS(os, 1)]));
				   ST(&(xo[WS(os, 4)]), VFNMSI(Ti, Td), ovs, &(xo[0]));
				   ST(&(xo[WS(os, 3)]), VFMAI(Ti, Td), ovs, &(xo[WS(os, 1)]));
			      }
			 }
		    }
	       }
	  }
     }
     VLEAVE();
}
Example #18
0
static void t1bv_16(R *ri, R *ii, const R *W, stride rs, INT mb, INT me, INT ms)
{
     DVK(KP382683432, +0.382683432365089771728459984030398866761344562);
     DVK(KP923879532, +0.923879532511286756128183189396788286822416626);
     DVK(KP707106781, +0.707106781186547524400844362104849039284835938);
     {
	  INT m;
	  R *x;
	  x = ii;
	  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(rs)) {
	       V TJ, T1b, TD, T1c, T17, T18, Ty, TK, T10, T11, T12, Tb, TM, T13, T14;
	       V T15, Tm, TN, TG, TI, TH;
	       TG = LD(&(x[0]), ms, &(x[0]));
	       TH = LD(&(x[WS(rs, 8)]), ms, &(x[0]));
	       TI = BYTW(&(W[TWVL * 14]), TH);
	       TJ = VSUB(TG, TI);
	       T1b = VADD(TG, TI);
	       {
		    V TA, TC, Tz, TB;
		    Tz = LD(&(x[WS(rs, 4)]), ms, &(x[0]));
		    TA = BYTW(&(W[TWVL * 6]), Tz);
		    TB = LD(&(x[WS(rs, 12)]), ms, &(x[0]));
		    TC = BYTW(&(W[TWVL * 22]), TB);
		    TD = VSUB(TA, TC);
		    T1c = VADD(TA, TC);
	       }
	       {
		    V Tp, Tw, Tr, Tu, Ts, Tx;
		    {
			 V To, Tv, Tq, Tt;
			 To = LD(&(x[WS(rs, 2)]), ms, &(x[0]));
			 Tp = BYTW(&(W[TWVL * 2]), To);
			 Tv = LD(&(x[WS(rs, 6)]), ms, &(x[0]));
			 Tw = BYTW(&(W[TWVL * 10]), Tv);
			 Tq = LD(&(x[WS(rs, 10)]), ms, &(x[0]));
			 Tr = BYTW(&(W[TWVL * 18]), Tq);
			 Tt = LD(&(x[WS(rs, 14)]), ms, &(x[0]));
			 Tu = BYTW(&(W[TWVL * 26]), Tt);
		    }
		    T17 = VADD(Tp, Tr);
		    T18 = VADD(Tu, Tw);
		    Ts = VSUB(Tp, Tr);
		    Tx = VSUB(Tu, Tw);
		    Ty = VMUL(LDK(KP707106781), VSUB(Ts, Tx));
		    TK = VMUL(LDK(KP707106781), VADD(Ts, Tx));
	       }
	       {
		    V T2, T9, T4, T7, T5, Ta;
		    {
			 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, 13)]), ms, &(x[WS(rs, 1)]));
			 T9 = BYTW(&(W[TWVL * 24]), T8);
			 T3 = LD(&(x[WS(rs, 9)]), ms, &(x[WS(rs, 1)]));
			 T4 = BYTW(&(W[TWVL * 16]), T3);
			 T6 = LD(&(x[WS(rs, 5)]), ms, &(x[WS(rs, 1)]));
			 T7 = BYTW(&(W[TWVL * 8]), T6);
		    }
		    T10 = VADD(T2, T4);
		    T11 = VADD(T7, T9);
		    T12 = VSUB(T10, T11);
		    T5 = VSUB(T2, T4);
		    Ta = VSUB(T7, T9);
		    Tb = VFNMS(LDK(KP382683432), Ta, VMUL(LDK(KP923879532), T5));
		    TM = VFMA(LDK(KP382683432), T5, VMUL(LDK(KP923879532), Ta));
	       }
	       {
		    V Td, Tk, Tf, Ti, Tg, Tl;
		    {
			 V Tc, Tj, Te, Th;
			 Tc = LD(&(x[WS(rs, 15)]), ms, &(x[WS(rs, 1)]));
			 Td = BYTW(&(W[TWVL * 28]), Tc);
			 Tj = LD(&(x[WS(rs, 11)]), ms, &(x[WS(rs, 1)]));
			 Tk = BYTW(&(W[TWVL * 20]), Tj);
			 Te = LD(&(x[WS(rs, 7)]), ms, &(x[WS(rs, 1)]));
			 Tf = BYTW(&(W[TWVL * 12]), Te);
			 Th = LD(&(x[WS(rs, 3)]), ms, &(x[WS(rs, 1)]));
			 Ti = BYTW(&(W[TWVL * 4]), Th);
		    }
		    T13 = VADD(Td, Tf);
		    T14 = VADD(Ti, Tk);
		    T15 = VSUB(T13, T14);
		    Tg = VSUB(Td, Tf);
		    Tl = VSUB(Ti, Tk);
		    Tm = VFMA(LDK(KP923879532), Tg, VMUL(LDK(KP382683432), Tl));
		    TN = VFNMS(LDK(KP382683432), Tg, VMUL(LDK(KP923879532), Tl));
	       }
	       {
		    V T1a, T1g, T1f, T1h;
		    {
			 V T16, T19, T1d, T1e;
			 T16 = VMUL(LDK(KP707106781), VSUB(T12, T15));
			 T19 = VSUB(T17, T18);
			 T1a = VBYI(VSUB(T16, T19));
			 T1g = VBYI(VADD(T19, T16));
			 T1d = VSUB(T1b, T1c);
			 T1e = VMUL(LDK(KP707106781), VADD(T12, T15));
			 T1f = VSUB(T1d, T1e);
			 T1h = VADD(T1d, T1e);
		    }
		    ST(&(x[WS(rs, 6)]), VADD(T1a, T1f), ms, &(x[0]));
		    ST(&(x[WS(rs, 14)]), VSUB(T1h, T1g), ms, &(x[0]));
		    ST(&(x[WS(rs, 10)]), VSUB(T1f, T1a), ms, &(x[0]));
		    ST(&(x[WS(rs, 2)]), VADD(T1g, T1h), ms, &(x[0]));
	       }
	       {
		    V T1k, T1o, T1n, T1p;
		    {
			 V T1i, T1j, T1l, T1m;
			 T1i = VADD(T1b, T1c);
			 T1j = VADD(T17, T18);
			 T1k = VSUB(T1i, T1j);
			 T1o = VADD(T1i, T1j);
			 T1l = VADD(T10, T11);
			 T1m = VADD(T13, T14);
			 T1n = VBYI(VSUB(T1l, T1m));
			 T1p = VADD(T1l, T1m);
		    }
		    ST(&(x[WS(rs, 12)]), VSUB(T1k, T1n), ms, &(x[0]));
		    ST(&(x[0]), VADD(T1o, T1p), ms, &(x[0]));
		    ST(&(x[WS(rs, 4)]), VADD(T1k, T1n), ms, &(x[0]));
		    ST(&(x[WS(rs, 8)]), 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 = VSUB(TJ, TK);
			 TO = VSUB(TM, TN);
			 TP = VSUB(TL, TO);
			 TR = VADD(TL, TO);
		    }
		    ST(&(x[WS(rs, 5)]), VADD(TF, TP), ms, &(x[WS(rs, 1)]));
		    ST(&(x[WS(rs, 13)]), VSUB(TR, TQ), ms, &(x[WS(rs, 1)]));
		    ST(&(x[WS(rs, 11)]), VSUB(TP, TF), ms, &(x[WS(rs, 1)]));
		    ST(&(x[WS(rs, 3)]), VADD(TQ, TR), ms, &(x[WS(rs, 1)]));
	       }
	       {
		    V TU, TY, TX, TZ;
		    {
			 V TS, TT, TV, TW;
			 TS = VADD(TJ, TK);
			 TT = VADD(Tb, Tm);
			 TU = VADD(TS, TT);
			 TY = VSUB(TS, TT);
			 TV = VADD(TD, Ty);
			 TW = VADD(TM, TN);
			 TX = VBYI(VADD(TV, TW));
			 TZ = VBYI(VSUB(TW, TV));
		    }
		    ST(&(x[WS(rs, 15)]), VSUB(TU, TX), ms, &(x[WS(rs, 1)]));
		    ST(&(x[WS(rs, 7)]), VADD(TY, TZ), ms, &(x[WS(rs, 1)]));
		    ST(&(x[WS(rs, 1)]), VADD(TU, TX), ms, &(x[WS(rs, 1)]));
		    ST(&(x[WS(rs, 9)]), VSUB(TY, TZ), ms, &(x[WS(rs, 1)]));
	       }
	  }
     }
     VLEAVE();
}
Example #19
0
static void t3bv_10(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) * 6)); m < me; m = m + VL, x = x + (VL * ms), W = W + (TWVL * 6), MAKE_VOLATILE_STRIDE(rs)) {
	  V T1, T7, Th, Tx, Tr, Td, Tp, T6, Tv, Tc, Te, Ti, Tl, T2, T3;
	  V T5;
	  T2 = LDW(&(W[0]));
	  T3 = LDW(&(W[TWVL * 2]));
	  T5 = LDW(&(W[TWVL * 4]));
	  T1 = LD(&(x[0]), ms, &(x[0]));
	  T7 = LD(&(x[WS(rs, 5)]), ms, &(x[WS(rs, 1)]));
	  {
	       V To, Tw, Tq, Tu, Ta, T4, Tt, Tk, Tb;
	       To = LD(&(x[WS(rs, 4)]), ms, &(x[0]));
	       Tw = LD(&(x[WS(rs, 1)]), ms, &(x[WS(rs, 1)]));
	       Tq = LD(&(x[WS(rs, 9)]), ms, &(x[WS(rs, 1)]));
	       Tu = LD(&(x[WS(rs, 6)]), ms, &(x[0]));
	       Ta = VZMULJ(T2, T3);
	       T4 = VZMUL(T2, T3);
	       Th = VZMULJ(T2, T5);
	       Tt = VZMULJ(T3, T5);
	       Tb = LD(&(x[WS(rs, 2)]), ms, &(x[0]));
	       Tx = VZMUL(T2, Tw);
	       Tr = VZMUL(T5, Tq);
	       Tk = LD(&(x[WS(rs, 3)]), ms, &(x[WS(rs, 1)]));
	       Td = VZMULJ(Ta, T5);
	       Tp = VZMUL(T4, To);
	       T6 = VZMULJ(T4, T5);
	       Tv = VZMUL(Tt, Tu);
	       Tc = VZMUL(Ta, Tb);
	       Te = LD(&(x[WS(rs, 7)]), ms, &(x[WS(rs, 1)]));
	       Ti = LD(&(x[WS(rs, 8)]), ms, &(x[0]));
	       Tl = VZMUL(T3, Tk);
	  }
	  {
	       V TN, Ts, T8, Ty, TO, Tf, Tj;
	       TN = VADD(Tp, Tr);
	       Ts = VSUB(Tp, Tr);
	       T8 = VZMUL(T6, T7);
	       Ty = VSUB(Tv, Tx);
	       TO = VADD(Tv, Tx);
	       Tf = VZMUL(Td, Te);
	       Tj = VZMUL(Th, Ti);
	       {
		    V T9, TJ, TP, TU, Tz, TF, Tg, TK, Tm, TL;
		    T9 = VSUB(T1, T8);
		    TJ = VADD(T1, T8);
		    TP = VADD(TN, TO);
		    TU = VSUB(TN, TO);
		    Tz = VADD(Ts, Ty);
		    TF = VSUB(Ts, Ty);
		    Tg = VSUB(Tc, Tf);
		    TK = VADD(Tc, Tf);
		    Tm = VSUB(Tj, Tl);
		    TL = VADD(Tj, Tl);
		    {
			 V TM, TV, Tn, TE;
			 TM = VADD(TK, TL);
			 TV = VSUB(TK, TL);
			 Tn = VADD(Tg, Tm);
			 TE = VSUB(Tg, Tm);
			 {
			      V TW, TY, TS, TQ, TG, TI, TC, TA, TR, TB;
			      TW = VMUL(LDK(KP951056516), VFNMS(LDK(KP618033988), TV, TU));
			      TY = VMUL(LDK(KP951056516), VFMA(LDK(KP618033988), TU, TV));
			      TS = VSUB(TM, TP);
			      TQ = VADD(TM, TP);
			      TG = VMUL(LDK(KP951056516), VFMA(LDK(KP618033988), TF, TE));
			      TI = VMUL(LDK(KP951056516), VFNMS(LDK(KP618033988), TE, TF));
			      TC = VSUB(Tn, Tz);
			      TA = VADD(Tn, Tz);
			      ST(&(x[0]), VADD(TJ, TQ), ms, &(x[0]));
			      TR = VFNMS(LDK(KP250000000), TQ, TJ);
			      ST(&(x[WS(rs, 5)]), VADD(T9, TA), ms, &(x[WS(rs, 1)]));
			      TB = VFNMS(LDK(KP250000000), TA, T9);
			      {
				   V TX, TT, TH, TD;
				   TX = VFMA(LDK(KP559016994), TS, TR);
				   TT = VFNMS(LDK(KP559016994), TS, TR);
				   TH = VFNMS(LDK(KP559016994), TC, TB);
				   TD = VFMA(LDK(KP559016994), TC, TB);
				   ST(&(x[WS(rs, 8)]), VFMAI(TW, TT), ms, &(x[0]));
				   ST(&(x[WS(rs, 2)]), VFNMSI(TW, TT), ms, &(x[0]));
				   ST(&(x[WS(rs, 6)]), VFMAI(TY, TX), ms, &(x[0]));
				   ST(&(x[WS(rs, 4)]), VFNMSI(TY, TX), ms, &(x[0]));
				   ST(&(x[WS(rs, 9)]), VFNMSI(TG, TD), ms, &(x[WS(rs, 1)]));
				   ST(&(x[WS(rs, 1)]), VFMAI(TG, TD), ms, &(x[WS(rs, 1)]));
				   ST(&(x[WS(rs, 7)]), VFNMSI(TI, TH), ms, &(x[WS(rs, 1)]));
				   ST(&(x[WS(rs, 3)]), VFMAI(TI, TH), ms, &(x[WS(rs, 1)]));
			      }
			 }
		    }
	       }
	  }
     }
}
Example #20
0
static void t1bv_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 = ii;
	  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(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, 2)]), ms, &(x[0]));
			      TG = LD(&(x[WS(rs, 6)]), ms, &(x[0]));
			      TB = LD(&(x[WS(rs, 10)]), ms, &(x[0]));
			      TE = LD(&(x[WS(rs, 14)]), ms, &(x[0]));
			      {
				   V Ti, TX, TY, 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 = BYTW(&(W[TWVL * 14]), T2);
					T6 = BYTW(&(W[TWVL * 6]), T5);
					T8 = BYTW(&(W[TWVL * 22]), T7);
					TA = BYTW(&(W[TWVL * 2]), Tz);
					TH = BYTW(&(W[TWVL * 10]), TG);
					TC = BYTW(&(W[TWVL * 18]), TB);
					TF = BYTW(&(W[TWVL * 26]), TE);
					Tc = BYTW(&(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);
					TX = VSUB(TA, TC);
					TI = VADD(TF, TH);
					TY = 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 = BYTW(&(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(TX, TY);
				   Te = BYTW(&(W[TWVL * 16]), Td);
				   Th = BYTW(&(W[TWVL * 8]), Tg);
				   Tn = BYTW(&(W[TWVL * 28]), Tm);
				   Tr = LD(&(x[WS(rs, 3)]), ms, &(x[WS(rs, 1)]));
				   Tu = BYTW(&(W[TWVL * 20]), Tt);
				   Tp = BYTW(&(W[TWVL * 12]), To);
			      }
			 }
		    }
		    {
			 V Tf, T11, Tk, T12, Ts;
			 TO = VADD(T4, T9);
			 Ta = VSUB(T4, T9);
			 TJ = VSUB(TD, TI);
			 TP = VADD(TD, TI);
			 Tf = VADD(Tc, Te);
			 T11 = VSUB(Tc, Te);
			 Tk = VADD(Th, Tj);
			 T12 = VSUB(Th, Tj);
			 Ts = BYTW(&(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 = VFMA(LDK(KP707106781), T1a, T19);
			 T1l = VFNMS(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 = VADD(TO, TP);
		    TQ = VSUB(TO, TP);
		    {
			 V T1e, T1j, T17, T1m;
			 T1e = VSUB(T1c, T1d);
			 T1j = VADD(T1c, T1d);
			 T17 = VADD(T13, T16);
			 T1m = VSUB(T13, T16);
			 {
			      V TV, TT, TK, Tx;
			      TV = VADD(TR, TS);
			      TT = VSUB(TR, TS);
			      TK = VSUB(Tl, Tw);
			      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 = VFNMS(LDK(KP923879532), T1m, T1l);
					T1n = VFMA(LDK(KP923879532), T1m, T1l);
					ST(&(x[WS(rs, 8)]), VSUB(TU, TV), ms, &(x[0]));
					ST(&(x[0]), VADD(TU, TV), ms, &(x[0]));
					ST(&(x[WS(rs, 4)]), VFMAI(TT, TQ), ms, &(x[0]));
					ST(&(x[WS(rs, 12)]), VFNMSI(TT, TQ), 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, 15)]), VFNMSI(T1h, T1g), ms, &(x[WS(rs, 1)]));
					     ST(&(x[WS(rs, 1)]), VFMAI(T1h, T1g), ms, &(x[WS(rs, 1)]));
					     ST(&(x[WS(rs, 9)]), VFMAI(T1f, T18), ms, &(x[WS(rs, 1)]));
					     ST(&(x[WS(rs, 7)]), VFNMSI(T1f, T18), ms, &(x[WS(rs, 1)]));
					     ST(&(x[WS(rs, 3)]), VFNMSI(T1p, T1o), ms, &(x[WS(rs, 1)]));
					     ST(&(x[WS(rs, 13)]), VFMAI(T1p, T1o), ms, &(x[WS(rs, 1)]));
					     ST(&(x[WS(rs, 11)]), VFNMSI(T1n, T1k), ms, &(x[WS(rs, 1)]));
					     ST(&(x[WS(rs, 5)]), VFMAI(T1n, T1k), ms, &(x[WS(rs, 1)]));
					     ST(&(x[WS(rs, 2)]), VFMAI(TN, TM), ms, &(x[0]));
					     ST(&(x[WS(rs, 14)]), VFNMSI(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();
}
Example #21
0
static void n1bv_12(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs)
{
    DVK(KP866025403, +0.866025403784438646763723170752936183471402627);
    DVK(KP500000000, +0.500000000000000000000000000000000000000000000);
    {
        INT i;
        const R *xi;
        R *xo;
        xi = ii;
        xo = io;
        for (i = v; i > 0; i = i - VL, xi = xi + (VL * ivs), xo = xo + (VL * ovs), MAKE_VOLATILE_STRIDE(24, is), MAKE_VOLATILE_STRIDE(24, os)) {
            V T5, Ta, TG, TF, Ty, Tm, Ti, Tp, TJ, TI, Tx, Ts;
            {
                V T1, T6, T4, Tk, T9, Tl;
                T1 = LD(&(xi[0]), ivs, &(xi[0]));
                T6 = LD(&(xi[WS(is, 6)]), ivs, &(xi[0]));
                {
                    V T2, T3, T7, T8;
                    T2 = LD(&(xi[WS(is, 4)]), ivs, &(xi[0]));
                    T3 = LD(&(xi[WS(is, 8)]), ivs, &(xi[0]));
                    T4 = VADD(T2, T3);
                    Tk = VSUB(T2, T3);
                    T7 = LD(&(xi[WS(is, 10)]), ivs, &(xi[0]));
                    T8 = LD(&(xi[WS(is, 2)]), ivs, &(xi[0]));
                    T9 = VADD(T7, T8);
                    Tl = VSUB(T7, T8);
                }
                T5 = VFNMS(LDK(KP500000000), T4, T1);
                Ta = VFNMS(LDK(KP500000000), T9, T6);
                TG = VADD(T6, T9);
                TF = VADD(T1, T4);
                Ty = VADD(Tk, Tl);
                Tm = VMUL(LDK(KP866025403), VSUB(Tk, Tl));
            }
            {
                V Tn, Tq, Te, To, Th, Tr;
                Tn = LD(&(xi[WS(is, 3)]), ivs, &(xi[WS(is, 1)]));
                Tq = LD(&(xi[WS(is, 9)]), ivs, &(xi[WS(is, 1)]));
                {
                    V Tc, Td, Tf, Tg;
                    Tc = LD(&(xi[WS(is, 7)]), ivs, &(xi[WS(is, 1)]));
                    Td = LD(&(xi[WS(is, 11)]), ivs, &(xi[WS(is, 1)]));
                    Te = VSUB(Tc, Td);
                    To = VADD(Tc, Td);
                    Tf = LD(&(xi[WS(is, 1)]), ivs, &(xi[WS(is, 1)]));
                    Tg = LD(&(xi[WS(is, 5)]), ivs, &(xi[WS(is, 1)]));
                    Th = VSUB(Tf, Tg);
                    Tr = VADD(Tf, Tg);
                }
                Ti = VMUL(LDK(KP866025403), VSUB(Te, Th));
                Tp = VFNMS(LDK(KP500000000), To, Tn);
                TJ = VADD(Tq, Tr);
                TI = VADD(Tn, To);
                Tx = VADD(Te, Th);
                Ts = VFNMS(LDK(KP500000000), Tr, Tq);
            }
            {
                V TH, TK, TL, TM;
                TH = VSUB(TF, TG);
                TK = VBYI(VSUB(TI, TJ));
                ST(&(xo[WS(os, 3)]), VSUB(TH, TK), ovs, &(xo[WS(os, 1)]));
                ST(&(xo[WS(os, 9)]), VADD(TH, TK), ovs, &(xo[WS(os, 1)]));
                TL = VADD(TF, TG);
                TM = VADD(TI, TJ);
                ST(&(xo[WS(os, 6)]), VSUB(TL, TM), ovs, &(xo[0]));
                ST(&(xo[0]), VADD(TL, TM), ovs, &(xo[0]));
            }
            {
                V Tj, Tv, Tu, Tw, Tb, Tt;
                Tb = VSUB(T5, Ta);
                Tj = VSUB(Tb, Ti);
                Tv = VADD(Tb, Ti);
                Tt = VSUB(Tp, Ts);
                Tu = VBYI(VADD(Tm, Tt));
                Tw = VBYI(VSUB(Tt, Tm));
                ST(&(xo[WS(os, 11)]), VSUB(Tj, Tu), ovs, &(xo[WS(os, 1)]));
                ST(&(xo[WS(os, 5)]), VADD(Tv, Tw), ovs, &(xo[WS(os, 1)]));
                ST(&(xo[WS(os, 1)]), VADD(Tj, Tu), ovs, &(xo[WS(os, 1)]));
                ST(&(xo[WS(os, 7)]), VSUB(Tv, Tw), ovs, &(xo[WS(os, 1)]));
            }
            {
                V Tz, TD, TC, TE, TA, TB;
                Tz = VBYI(VMUL(LDK(KP866025403), VSUB(Tx, Ty)));
                TD = VBYI(VMUL(LDK(KP866025403), VADD(Ty, Tx)));
                TA = VADD(T5, Ta);
                TB = VADD(Tp, Ts);
                TC = VSUB(TA, TB);
                TE = VADD(TA, TB);
                ST(&(xo[WS(os, 2)]), VADD(Tz, TC), ovs, &(xo[0]));
                ST(&(xo[WS(os, 8)]), VSUB(TE, TD), ovs, &(xo[0]));
                ST(&(xo[WS(os, 10)]), VSUB(TC, Tz), ovs, &(xo[0]));
                ST(&(xo[WS(os, 4)]), VADD(TD, TE), ovs, &(xo[0]));
            }
        }
    }
    VLEAVE();
}
Example #22
0
static void n1fv_16(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs)
{
     DVK(KP923879532, +0.923879532511286756128183189396788286822416626);
     DVK(KP382683432, +0.382683432365089771728459984030398866761344562);
     DVK(KP707106781, +0.707106781186547524400844362104849039284835938);
     {
	  INT i;
	  const R *xi;
	  R *xo;
	  xi = ri;
	  xo = ro;
	  for (i = v; i > 0; i = i - VL, xi = xi + (VL * ivs), xo = xo + (VL * ovs), MAKE_VOLATILE_STRIDE(32, is), MAKE_VOLATILE_STRIDE(32, os)) {
	       V Tp, T13, Tu, TN, Tm, T14, Tv, TY, T7, T17, Ty, TT, Te, T16, Tx;
	       V TQ;
	       {
		    V Tn, To, TM, Ts, Tt, TL;
		    Tn = LD(&(xi[WS(is, 4)]), ivs, &(xi[0]));
		    To = LD(&(xi[WS(is, 12)]), ivs, &(xi[0]));
		    TM = VADD(Tn, To);
		    Ts = LD(&(xi[0]), ivs, &(xi[0]));
		    Tt = LD(&(xi[WS(is, 8)]), ivs, &(xi[0]));
		    TL = VADD(Ts, Tt);
		    Tp = VSUB(Tn, To);
		    T13 = VADD(TL, TM);
		    Tu = VSUB(Ts, Tt);
		    TN = VSUB(TL, TM);
	       }
	       {
		    V Ti, TW, Tl, TX;
		    {
			 V Tg, Th, Tj, Tk;
			 Tg = LD(&(xi[WS(is, 14)]), ivs, &(xi[0]));
			 Th = LD(&(xi[WS(is, 6)]), ivs, &(xi[0]));
			 Ti = VSUB(Tg, Th);
			 TW = VADD(Tg, Th);
			 Tj = LD(&(xi[WS(is, 2)]), ivs, &(xi[0]));
			 Tk = LD(&(xi[WS(is, 10)]), ivs, &(xi[0]));
			 Tl = VSUB(Tj, Tk);
			 TX = VADD(Tj, Tk);
		    }
		    Tm = VMUL(LDK(KP707106781), VSUB(Ti, Tl));
		    T14 = VADD(TX, TW);
		    Tv = VMUL(LDK(KP707106781), VADD(Tl, Ti));
		    TY = VSUB(TW, TX);
	       }
	       {
		    V T3, TR, T6, TS;
		    {
			 V T1, T2, T4, T5;
			 T1 = LD(&(xi[WS(is, 15)]), ivs, &(xi[WS(is, 1)]));
			 T2 = LD(&(xi[WS(is, 7)]), ivs, &(xi[WS(is, 1)]));
			 T3 = VSUB(T1, T2);
			 TR = VADD(T1, T2);
			 T4 = LD(&(xi[WS(is, 3)]), ivs, &(xi[WS(is, 1)]));
			 T5 = LD(&(xi[WS(is, 11)]), ivs, &(xi[WS(is, 1)]));
			 T6 = VSUB(T4, T5);
			 TS = VADD(T4, T5);
		    }
		    T7 = VFNMS(LDK(KP923879532), T6, VMUL(LDK(KP382683432), T3));
		    T17 = VADD(TR, TS);
		    Ty = VFMA(LDK(KP923879532), T3, VMUL(LDK(KP382683432), T6));
		    TT = VSUB(TR, TS);
	       }
	       {
		    V Ta, TO, Td, TP;
		    {
			 V T8, T9, Tb, Tc;
			 T8 = LD(&(xi[WS(is, 1)]), ivs, &(xi[WS(is, 1)]));
			 T9 = LD(&(xi[WS(is, 9)]), ivs, &(xi[WS(is, 1)]));
			 Ta = VSUB(T8, T9);
			 TO = VADD(T8, T9);
			 Tb = LD(&(xi[WS(is, 5)]), ivs, &(xi[WS(is, 1)]));
			 Tc = LD(&(xi[WS(is, 13)]), ivs, &(xi[WS(is, 1)]));
			 Td = VSUB(Tb, Tc);
			 TP = VADD(Tb, Tc);
		    }
		    Te = VFMA(LDK(KP382683432), Ta, VMUL(LDK(KP923879532), Td));
		    T16 = VADD(TO, TP);
		    Tx = VFNMS(LDK(KP382683432), Td, VMUL(LDK(KP923879532), Ta));
		    TQ = VSUB(TO, TP);
	       }
	       {
		    V T15, T18, T19, T1a;
		    T15 = VADD(T13, T14);
		    T18 = VADD(T16, T17);
		    ST(&(xo[WS(os, 8)]), VSUB(T15, T18), ovs, &(xo[0]));
		    ST(&(xo[0]), VADD(T15, T18), ovs, &(xo[0]));
		    T19 = VSUB(T13, T14);
		    T1a = VBYI(VSUB(T17, T16));
		    ST(&(xo[WS(os, 12)]), VSUB(T19, T1a), ovs, &(xo[0]));
		    ST(&(xo[WS(os, 4)]), VADD(T19, T1a), ovs, &(xo[0]));
	       }
	       {
		    V TV, T11, T10, T12, TU, TZ;
		    TU = VMUL(LDK(KP707106781), VADD(TQ, TT));
		    TV = VADD(TN, TU);
		    T11 = VSUB(TN, TU);
		    TZ = VMUL(LDK(KP707106781), VSUB(TT, TQ));
		    T10 = VBYI(VADD(TY, TZ));
		    T12 = VBYI(VSUB(TZ, TY));
		    ST(&(xo[WS(os, 14)]), VSUB(TV, T10), ovs, &(xo[0]));
		    ST(&(xo[WS(os, 6)]), VADD(T11, T12), ovs, &(xo[0]));
		    ST(&(xo[WS(os, 2)]), VADD(TV, T10), ovs, &(xo[0]));
		    ST(&(xo[WS(os, 10)]), VSUB(T11, T12), ovs, &(xo[0]));
	       }
	       {
		    V Tr, TB, TA, TC;
		    {
			 V Tf, Tq, Tw, Tz;
			 Tf = VSUB(T7, Te);
			 Tq = VSUB(Tm, Tp);
			 Tr = VBYI(VSUB(Tf, Tq));
			 TB = VBYI(VADD(Tq, Tf));
			 Tw = VADD(Tu, Tv);
			 Tz = VADD(Tx, Ty);
			 TA = VSUB(Tw, Tz);
			 TC = VADD(Tw, Tz);
		    }
		    ST(&(xo[WS(os, 7)]), VADD(Tr, TA), ovs, &(xo[WS(os, 1)]));
		    ST(&(xo[WS(os, 15)]), VSUB(TC, TB), ovs, &(xo[WS(os, 1)]));
		    ST(&(xo[WS(os, 9)]), VSUB(TA, Tr), ovs, &(xo[WS(os, 1)]));
		    ST(&(xo[WS(os, 1)]), VADD(TB, TC), ovs, &(xo[WS(os, 1)]));
	       }
	       {
		    V TF, TJ, TI, TK;
		    {
			 V TD, TE, TG, TH;
			 TD = VSUB(Tu, Tv);
			 TE = VADD(Te, T7);
			 TF = VADD(TD, TE);
			 TJ = VSUB(TD, TE);
			 TG = VADD(Tp, Tm);
			 TH = VSUB(Ty, Tx);
			 TI = VBYI(VADD(TG, TH));
			 TK = VBYI(VSUB(TH, TG));
		    }
		    ST(&(xo[WS(os, 13)]), VSUB(TF, TI), ovs, &(xo[WS(os, 1)]));
		    ST(&(xo[WS(os, 5)]), VADD(TJ, TK), ovs, &(xo[WS(os, 1)]));
		    ST(&(xo[WS(os, 3)]), VADD(TF, TI), ovs, &(xo[WS(os, 1)]));
		    ST(&(xo[WS(os, 11)]), VSUB(TJ, TK), ovs, &(xo[WS(os, 1)]));
	       }
	  }
     }
     VLEAVE();
}
static void t1bv_12(R *ri, R *ii, const R *W, stride rs, INT mb, INT me, INT ms)
{
     DVK(KP866025403, +0.866025403784438646763723170752936183471402627);
     DVK(KP500000000, +0.500000000000000000000000000000000000000000000);
     INT m;
     R *x;
     x = ii;
     for (m = mb, W = W + (mb * ((TWVL / VL) * 22)); m < me; m = m + VL, x = x + (VL * ms), W = W + (TWVL * 22), MAKE_VOLATILE_STRIDE(rs)) {
	  V T1, Tt, T6, T7, TB, Tq, TC, TD, T9, Tu, Te, Tf, Tx, Tl, Ty;
	  V Tz;
	  {
	       V T5, T3, T4, T2;
	       T1 = LD(&(x[0]), ms, &(x[0]));
	       T4 = LD(&(x[WS(rs, 8)]), ms, &(x[0]));
	       T5 = BYTW(&(W[TWVL * 14]), T4);
	       T2 = LD(&(x[WS(rs, 4)]), ms, &(x[0]));
	       T3 = BYTW(&(W[TWVL * 6]), T2);
	       Tt = VSUB(T3, T5);
	       T6 = VADD(T3, T5);
	       T7 = VFNMS(LDK(KP500000000), T6, T1);
	  }
	  {
	       V Tn, Tp, Tm, TA, To;
	       Tm = LD(&(x[WS(rs, 1)]), ms, &(x[WS(rs, 1)]));
	       Tn = BYTW(&(W[0]), Tm);
	       TA = LD(&(x[WS(rs, 9)]), ms, &(x[WS(rs, 1)]));
	       TB = BYTW(&(W[TWVL * 16]), TA);
	       To = LD(&(x[WS(rs, 5)]), ms, &(x[WS(rs, 1)]));
	       Tp = BYTW(&(W[TWVL * 8]), To);
	       Tq = VSUB(Tn, Tp);
	       TC = VADD(Tn, Tp);
	       TD = VFNMS(LDK(KP500000000), TC, TB);
	  }
	  {
	       V Td, Tb, T8, Tc, Ta;
	       T8 = LD(&(x[WS(rs, 6)]), ms, &(x[0]));
	       T9 = BYTW(&(W[TWVL * 10]), T8);
	       Tc = LD(&(x[WS(rs, 2)]), ms, &(x[0]));
	       Td = BYTW(&(W[TWVL * 2]), Tc);
	       Ta = LD(&(x[WS(rs, 10)]), ms, &(x[0]));
	       Tb = BYTW(&(W[TWVL * 18]), Ta);
	       Tu = VSUB(Tb, Td);
	       Te = VADD(Tb, Td);
	       Tf = VFNMS(LDK(KP500000000), Te, T9);
	  }
	  {
	       V Ti, Tk, Th, Tw, Tj;
	       Th = LD(&(x[WS(rs, 7)]), ms, &(x[WS(rs, 1)]));
	       Ti = BYTW(&(W[TWVL * 12]), Th);
	       Tw = LD(&(x[WS(rs, 3)]), ms, &(x[WS(rs, 1)]));
	       Tx = BYTW(&(W[TWVL * 4]), Tw);
	       Tj = LD(&(x[WS(rs, 11)]), ms, &(x[WS(rs, 1)]));
	       Tk = BYTW(&(W[TWVL * 20]), Tj);
	       Tl = VSUB(Ti, Tk);
	       Ty = VADD(Ti, Tk);
	       Tz = VFNMS(LDK(KP500000000), Ty, Tx);
	  }
	  {
	       V Ts, TG, TF, TH;
	       {
		    V Tg, Tr, Tv, TE;
		    Tg = VSUB(T7, Tf);
		    Tr = VMUL(LDK(KP866025403), VSUB(Tl, Tq));
		    Ts = VSUB(Tg, Tr);
		    TG = VADD(Tg, Tr);
		    Tv = VMUL(LDK(KP866025403), VSUB(Tt, Tu));
		    TE = VSUB(Tz, TD);
		    TF = VBYI(VADD(Tv, TE));
		    TH = VBYI(VSUB(TE, Tv));
	       }
	       ST(&(x[WS(rs, 11)]), VSUB(Ts, TF), ms, &(x[WS(rs, 1)]));
	       ST(&(x[WS(rs, 5)]), VADD(TG, TH), ms, &(x[WS(rs, 1)]));
	       ST(&(x[WS(rs, 1)]), VADD(Ts, TF), ms, &(x[WS(rs, 1)]));
	       ST(&(x[WS(rs, 7)]), VSUB(TG, TH), ms, &(x[WS(rs, 1)]));
	  }
	  {
	       V TS, TW, TV, TX;
	       {
		    V TQ, TR, TT, TU;
		    TQ = VADD(T1, T6);
		    TR = VADD(T9, Te);
		    TS = VSUB(TQ, TR);
		    TW = VADD(TQ, TR);
		    TT = VADD(Tx, Ty);
		    TU = VADD(TB, TC);
		    TV = VBYI(VSUB(TT, TU));
		    TX = VADD(TT, TU);
	       }
	       ST(&(x[WS(rs, 3)]), VSUB(TS, TV), ms, &(x[WS(rs, 1)]));
	       ST(&(x[0]), VADD(TW, TX), ms, &(x[0]));
	       ST(&(x[WS(rs, 9)]), VADD(TS, TV), ms, &(x[WS(rs, 1)]));
	       ST(&(x[WS(rs, 6)]), VSUB(TW, TX), ms, &(x[0]));
	  }
	  {
	       V TK, TO, TN, TP;
	       {
		    V TI, TJ, TL, TM;
		    TI = VADD(Tl, Tq);
		    TJ = VADD(Tt, Tu);
		    TK = VBYI(VMUL(LDK(KP866025403), VSUB(TI, TJ)));
		    TO = VBYI(VMUL(LDK(KP866025403), VADD(TJ, TI)));
		    TL = VADD(T7, Tf);
		    TM = VADD(Tz, TD);
		    TN = VSUB(TL, TM);
		    TP = VADD(TL, TM);
	       }
	       ST(&(x[WS(rs, 2)]), VADD(TK, TN), ms, &(x[0]));
	       ST(&(x[WS(rs, 8)]), VSUB(TP, TO), ms, &(x[0]));
	       ST(&(x[WS(rs, 10)]), VSUB(TN, TK), ms, &(x[0]));
	       ST(&(x[WS(rs, 4)]), VADD(TO, TP), ms, &(x[0]));
	  }
     }
}
Example #24
0
static void n1fv_16(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs)
{
     DVK(KP923879532, +0.923879532511286756128183189396788286822416626);
     DVK(KP707106781, +0.707106781186547524400844362104849039284835938);
     DVK(KP414213562, +0.414213562373095048801688724209698078569671875);
     {
	  INT i;
	  const R *xi;
	  R *xo;
	  xi = ri;
	  xo = ro;
	  for (i = v; i > 0; i = i - VL, xi = xi + (VL * ivs), xo = xo + (VL * ovs), MAKE_VOLATILE_STRIDE(32, is), MAKE_VOLATILE_STRIDE(32, os)) {
	       V T7, TU, Tz, TH, Tu, TV, TA, TK, Te, TX, TC, TO, Tl, TY, TD;
	       V TR;
	       {
		    V T1, T2, T3, T4, T5, T6;
		    T1 = LD(&(xi[0]), ivs, &(xi[0]));
		    T2 = LD(&(xi[WS(is, 8)]), ivs, &(xi[0]));
		    T3 = VADD(T1, T2);
		    T4 = LD(&(xi[WS(is, 4)]), ivs, &(xi[0]));
		    T5 = LD(&(xi[WS(is, 12)]), ivs, &(xi[0]));
		    T6 = VADD(T4, T5);
		    T7 = VSUB(T3, T6);
		    TU = VSUB(T4, T5);
		    Tz = VADD(T3, T6);
		    TH = VSUB(T1, T2);
	       }
	       {
		    V Tq, TJ, Tt, TI;
		    {
			 V To, Tp, Tr, Ts;
			 To = LD(&(xi[WS(is, 14)]), ivs, &(xi[0]));
			 Tp = LD(&(xi[WS(is, 6)]), ivs, &(xi[0]));
			 Tq = VADD(To, Tp);
			 TJ = VSUB(To, Tp);
			 Tr = LD(&(xi[WS(is, 2)]), ivs, &(xi[0]));
			 Ts = LD(&(xi[WS(is, 10)]), ivs, &(xi[0]));
			 Tt = VADD(Tr, Ts);
			 TI = VSUB(Tr, Ts);
		    }
		    Tu = VSUB(Tq, Tt);
		    TV = VSUB(TJ, TI);
		    TA = VADD(Tt, Tq);
		    TK = VADD(TI, TJ);
	       }
	       {
		    V Ta, TM, Td, TN;
		    {
			 V T8, T9, Tb, Tc;
			 T8 = LD(&(xi[WS(is, 1)]), ivs, &(xi[WS(is, 1)]));
			 T9 = LD(&(xi[WS(is, 9)]), ivs, &(xi[WS(is, 1)]));
			 Ta = VADD(T8, T9);
			 TM = VSUB(T8, T9);
			 Tb = LD(&(xi[WS(is, 5)]), ivs, &(xi[WS(is, 1)]));
			 Tc = LD(&(xi[WS(is, 13)]), ivs, &(xi[WS(is, 1)]));
			 Td = VADD(Tb, Tc);
			 TN = VSUB(Tb, Tc);
		    }
		    Te = VSUB(Ta, Td);
		    TX = VFMA(LDK(KP414213562), TM, TN);
		    TC = VADD(Ta, Td);
		    TO = VFNMS(LDK(KP414213562), TN, TM);
	       }
	       {
		    V Th, TP, Tk, TQ;
		    {
			 V Tf, Tg, Ti, Tj;
			 Tf = LD(&(xi[WS(is, 15)]), ivs, &(xi[WS(is, 1)]));
			 Tg = LD(&(xi[WS(is, 7)]), ivs, &(xi[WS(is, 1)]));
			 Th = VADD(Tf, Tg);
			 TP = VSUB(Tf, Tg);
			 Ti = LD(&(xi[WS(is, 3)]), ivs, &(xi[WS(is, 1)]));
			 Tj = LD(&(xi[WS(is, 11)]), ivs, &(xi[WS(is, 1)]));
			 Tk = VADD(Ti, Tj);
			 TQ = VSUB(Tj, Ti);
		    }
		    Tl = VSUB(Th, Tk);
		    TY = VFMA(LDK(KP414213562), TP, TQ);
		    TD = VADD(Th, Tk);
		    TR = VFNMS(LDK(KP414213562), TQ, TP);
	       }
	       {
		    V TB, TE, TF, TG;
		    TB = VADD(Tz, TA);
		    TE = VADD(TC, TD);
		    ST(&(xo[WS(os, 8)]), VSUB(TB, TE), ovs, &(xo[0]));
		    ST(&(xo[0]), VADD(TB, TE), ovs, &(xo[0]));
		    TF = VSUB(Tz, TA);
		    TG = VSUB(TD, TC);
		    ST(&(xo[WS(os, 12)]), VFNMSI(TG, TF), ovs, &(xo[0]));
		    ST(&(xo[WS(os, 4)]), VFMAI(TG, TF), ovs, &(xo[0]));
	       }
	       {
		    V Tn, Tx, Tw, Ty, Tm, Tv;
		    Tm = VADD(Te, Tl);
		    Tn = VFNMS(LDK(KP707106781), Tm, T7);
		    Tx = VFMA(LDK(KP707106781), Tm, T7);
		    Tv = VSUB(Tl, Te);
		    Tw = VFNMS(LDK(KP707106781), Tv, Tu);
		    Ty = VFMA(LDK(KP707106781), Tv, Tu);
		    ST(&(xo[WS(os, 6)]), VFNMSI(Tw, Tn), ovs, &(xo[0]));
		    ST(&(xo[WS(os, 2)]), VFMAI(Ty, Tx), ovs, &(xo[0]));
		    ST(&(xo[WS(os, 10)]), VFMAI(Tw, Tn), ovs, &(xo[0]));
		    ST(&(xo[WS(os, 14)]), VFNMSI(Ty, Tx), ovs, &(xo[0]));
	       }
	       {
		    V TT, T11, T10, T12;
		    {
			 V TL, TS, TW, TZ;
			 TL = VFMA(LDK(KP707106781), TK, TH);
			 TS = VADD(TO, TR);
			 TT = VFNMS(LDK(KP923879532), TS, TL);
			 T11 = VFMA(LDK(KP923879532), TS, TL);
			 TW = VFNMS(LDK(KP707106781), TV, TU);
			 TZ = VSUB(TX, TY);
			 T10 = VFNMS(LDK(KP923879532), TZ, TW);
			 T12 = VFMA(LDK(KP923879532), TZ, TW);
		    }
		    ST(&(xo[WS(os, 9)]), VFNMSI(T10, TT), ovs, &(xo[WS(os, 1)]));
		    ST(&(xo[WS(os, 15)]), VFMAI(T12, T11), ovs, &(xo[WS(os, 1)]));
		    ST(&(xo[WS(os, 7)]), VFMAI(T10, TT), ovs, &(xo[WS(os, 1)]));
		    ST(&(xo[WS(os, 1)]), VFNMSI(T12, T11), ovs, &(xo[WS(os, 1)]));
	       }
	       {
		    V T15, T19, T18, T1a;
		    {
			 V T13, T14, T16, T17;
			 T13 = VFNMS(LDK(KP707106781), TK, TH);
			 T14 = VADD(TX, TY);
			 T15 = VFNMS(LDK(KP923879532), T14, T13);
			 T19 = VFMA(LDK(KP923879532), T14, T13);
			 T16 = VFMA(LDK(KP707106781), TV, TU);
			 T17 = VSUB(TR, TO);
			 T18 = VFNMS(LDK(KP923879532), T17, T16);
			 T1a = VFMA(LDK(KP923879532), T17, T16);
		    }
		    ST(&(xo[WS(os, 5)]), VFNMSI(T18, T15), ovs, &(xo[WS(os, 1)]));
		    ST(&(xo[WS(os, 13)]), VFNMSI(T1a, T19), ovs, &(xo[WS(os, 1)]));
		    ST(&(xo[WS(os, 11)]), VFMAI(T18, T15), ovs, &(xo[WS(os, 1)]));
		    ST(&(xo[WS(os, 3)]), VFMAI(T1a, T19), ovs, &(xo[WS(os, 1)]));
	       }
	  }
     }
     VLEAVE();
}
static void n2bv_20(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs)
{
     DVK(KP587785252, +0.587785252292473129168705954639072768597652438);
     DVK(KP951056516, +0.951056516295153572116439333379382143405698634);
     DVK(KP250000000, +0.250000000000000000000000000000000000000000000);
     DVK(KP559016994, +0.559016994374947424102293417182819058860154590);
     {
	  INT i;
	  const R *xi;
	  R *xo;
	  xi = ii;
	  xo = io;
	  for (i = v; i > 0; i = i - VL, xi = xi + (VL * ivs), xo = xo + (VL * ovs), MAKE_VOLATILE_STRIDE(40, is), MAKE_VOLATILE_STRIDE(40, os)) {
	       V T3, T1y, TH, T1i, Ts, TL, TM, Tz, T13, T16, T1j, T1u, T1v, T1w, T1r;
	       V T1s, T1t, T1a, T1d, T1k, Ti, Tk, TE, TI;
	       {
		    V T1, T2, T1g, TF, TG, T1h;
		    T1 = LD(&(xi[0]), ivs, &(xi[0]));
		    T2 = LD(&(xi[WS(is, 10)]), ivs, &(xi[0]));
		    T1g = VADD(T1, T2);
		    TF = LD(&(xi[WS(is, 5)]), ivs, &(xi[WS(is, 1)]));
		    TG = LD(&(xi[WS(is, 15)]), ivs, &(xi[WS(is, 1)]));
		    T1h = VADD(TF, TG);
		    T3 = VSUB(T1, T2);
		    T1y = VADD(T1g, T1h);
		    TH = VSUB(TF, TG);
		    T1i = VSUB(T1g, T1h);
	       }
	       {
		    V T6, T11, Tv, T19, Ty, T1c, T9, T14, Td, T18, To, T12, Tr, T15, Tg;
		    V T1b;
		    {
			 V T4, T5, Tt, Tu;
			 T4 = LD(&(xi[WS(is, 4)]), ivs, &(xi[0]));
			 T5 = LD(&(xi[WS(is, 14)]), ivs, &(xi[0]));
			 T6 = VSUB(T4, T5);
			 T11 = VADD(T4, T5);
			 Tt = LD(&(xi[WS(is, 13)]), ivs, &(xi[WS(is, 1)]));
			 Tu = LD(&(xi[WS(is, 3)]), ivs, &(xi[WS(is, 1)]));
			 Tv = VSUB(Tt, Tu);
			 T19 = VADD(Tt, Tu);
		    }
		    {
			 V Tw, Tx, T7, T8;
			 Tw = LD(&(xi[WS(is, 17)]), ivs, &(xi[WS(is, 1)]));
			 Tx = LD(&(xi[WS(is, 7)]), ivs, &(xi[WS(is, 1)]));
			 Ty = VSUB(Tw, Tx);
			 T1c = VADD(Tw, Tx);
			 T7 = LD(&(xi[WS(is, 16)]), ivs, &(xi[0]));
			 T8 = LD(&(xi[WS(is, 6)]), ivs, &(xi[0]));
			 T9 = VSUB(T7, T8);
			 T14 = VADD(T7, T8);
		    }
		    {
			 V Tb, Tc, Tm, Tn;
			 Tb = LD(&(xi[WS(is, 8)]), ivs, &(xi[0]));
			 Tc = LD(&(xi[WS(is, 18)]), ivs, &(xi[0]));
			 Td = VSUB(Tb, Tc);
			 T18 = VADD(Tb, Tc);
			 Tm = LD(&(xi[WS(is, 9)]), ivs, &(xi[WS(is, 1)]));
			 Tn = LD(&(xi[WS(is, 19)]), ivs, &(xi[WS(is, 1)]));
			 To = VSUB(Tm, Tn);
			 T12 = VADD(Tm, Tn);
		    }
		    {
			 V Tp, Tq, Te, Tf;
			 Tp = LD(&(xi[WS(is, 1)]), ivs, &(xi[WS(is, 1)]));
			 Tq = LD(&(xi[WS(is, 11)]), ivs, &(xi[WS(is, 1)]));
			 Tr = VSUB(Tp, Tq);
			 T15 = VADD(Tp, Tq);
			 Te = LD(&(xi[WS(is, 12)]), ivs, &(xi[0]));
			 Tf = LD(&(xi[WS(is, 2)]), ivs, &(xi[0]));
			 Tg = VSUB(Te, Tf);
			 T1b = VADD(Te, Tf);
		    }
		    Ts = VSUB(To, Tr);
		    TL = VSUB(T6, T9);
		    TM = VSUB(Td, Tg);
		    Tz = VSUB(Tv, Ty);
		    T13 = VSUB(T11, T12);
		    T16 = VSUB(T14, T15);
		    T1j = VADD(T13, T16);
		    T1u = VADD(T18, T19);
		    T1v = VADD(T1b, T1c);
		    T1w = VADD(T1u, T1v);
		    T1r = VADD(T11, T12);
		    T1s = VADD(T14, T15);
		    T1t = VADD(T1r, T1s);
		    T1a = VSUB(T18, T19);
		    T1d = VSUB(T1b, T1c);
		    T1k = VADD(T1a, T1d);
		    {
			 V Ta, Th, TC, TD;
			 Ta = VADD(T6, T9);
			 Th = VADD(Td, Tg);
			 Ti = VADD(Ta, Th);
			 Tk = VMUL(LDK(KP559016994), VSUB(Ta, Th));
			 TC = VADD(To, Tr);
			 TD = VADD(Tv, Ty);
			 TE = VMUL(LDK(KP559016994), VSUB(TC, TD));
			 TI = VADD(TC, TD);
		    }
	       }
	       {
		    V T1H, T1J, T1K, T1L, T1N, T1I, TZ, T10;
		    TZ = VADD(T3, Ti);
		    T10 = VBYI(VADD(TH, TI));
		    T1H = VSUB(TZ, T10);
		    STM2(&(xo[30]), T1H, ovs, &(xo[2]));
		    T1I = VADD(TZ, T10);
		    STM2(&(xo[10]), T1I, ovs, &(xo[2]));
		    {
			 V T1x, T1z, T1A, T1E, T1G, T1C, T1D, T1F, T1B, T1M;
			 T1x = VMUL(LDK(KP559016994), VSUB(T1t, T1w));
			 T1z = VADD(T1t, T1w);
			 T1A = VFNMS(LDK(KP250000000), T1z, T1y);
			 T1C = VSUB(T1r, T1s);
			 T1D = VSUB(T1u, T1v);
			 T1E = VBYI(VFMA(LDK(KP951056516), T1C, VMUL(LDK(KP587785252), T1D)));
			 T1G = VBYI(VFNMS(LDK(KP951056516), T1D, VMUL(LDK(KP587785252), T1C)));
			 T1J = VADD(T1y, T1z);
			 STM2(&(xo[0]), T1J, ovs, &(xo[0]));
			 T1F = VSUB(T1A, T1x);
			 T1K = VSUB(T1F, T1G);
			 STM2(&(xo[16]), T1K, ovs, &(xo[0]));
			 T1L = VADD(T1G, T1F);
			 STM2(&(xo[24]), T1L, ovs, &(xo[0]));
			 T1B = VADD(T1x, T1A);
			 T1M = VSUB(T1B, T1E);
			 STM2(&(xo[8]), T1M, ovs, &(xo[0]));
			 STN2(&(xo[8]), T1M, T1I, ovs);
			 T1N = VADD(T1E, T1B);
			 STM2(&(xo[32]), T1N, ovs, &(xo[0]));
		    }
		    {
			 V T1O, T1P, T1R, T1S;
			 {
			      V T1n, T1l, T1m, T1f, T1p, T17, T1e, T1q, T1Q, T1o;
			      T1n = VMUL(LDK(KP559016994), VSUB(T1j, T1k));
			      T1l = VADD(T1j, T1k);
			      T1m = VFNMS(LDK(KP250000000), T1l, T1i);
			      T17 = VSUB(T13, T16);
			      T1e = VSUB(T1a, T1d);
			      T1f = VBYI(VFNMS(LDK(KP951056516), T1e, VMUL(LDK(KP587785252), T17)));
			      T1p = VBYI(VFMA(LDK(KP951056516), T17, VMUL(LDK(KP587785252), T1e)));
			      T1O = VADD(T1i, T1l);
			      STM2(&(xo[20]), T1O, ovs, &(xo[0]));
			      T1q = VADD(T1n, T1m);
			      T1P = VADD(T1p, T1q);
			      STM2(&(xo[12]), T1P, ovs, &(xo[0]));
			      T1Q = VSUB(T1q, T1p);
			      STM2(&(xo[28]), T1Q, ovs, &(xo[0]));
			      STN2(&(xo[28]), T1Q, T1H, ovs);
			      T1o = VSUB(T1m, T1n);
			      T1R = VADD(T1f, T1o);
			      STM2(&(xo[4]), T1R, ovs, &(xo[0]));
			      T1S = VSUB(T1o, T1f);
			      STM2(&(xo[36]), T1S, ovs, &(xo[0]));
			 }
			 {
			      V TA, TN, TU, TS, TK, TV, Tl, TR, TJ, Tj;
			      TA = VFNMS(LDK(KP951056516), Tz, VMUL(LDK(KP587785252), Ts));
			      TN = VFNMS(LDK(KP951056516), TM, VMUL(LDK(KP587785252), TL));
			      TU = VFMA(LDK(KP951056516), TL, VMUL(LDK(KP587785252), TM));
			      TS = VFMA(LDK(KP951056516), Ts, VMUL(LDK(KP587785252), Tz));
			      TJ = VFNMS(LDK(KP250000000), TI, TH);
			      TK = VSUB(TE, TJ);
			      TV = VADD(TE, TJ);
			      Tj = VFNMS(LDK(KP250000000), Ti, T3);
			      Tl = VSUB(Tj, Tk);
			      TR = VADD(Tk, Tj);
			      {
				   V TB, TO, T1T, T1U;
				   TB = VSUB(Tl, TA);
				   TO = VBYI(VSUB(TK, TN));
				   T1T = VSUB(TB, TO);
				   STM2(&(xo[34]), T1T, ovs, &(xo[2]));
				   STN2(&(xo[32]), T1N, T1T, ovs);
				   T1U = VADD(TB, TO);
				   STM2(&(xo[6]), T1U, ovs, &(xo[2]));
				   STN2(&(xo[4]), T1R, T1U, ovs);
			      }
			      {
				   V TX, TY, T1V, T1W;
				   TX = VADD(TR, TS);
				   TY = VBYI(VSUB(TV, TU));
				   T1V = VSUB(TX, TY);
				   STM2(&(xo[22]), T1V, ovs, &(xo[2]));
				   STN2(&(xo[20]), T1O, T1V, ovs);
				   T1W = VADD(TX, TY);
				   STM2(&(xo[18]), T1W, ovs, &(xo[2]));
				   STN2(&(xo[16]), T1K, T1W, ovs);
			      }
			      {
				   V TP, TQ, T1X, T1Y;
				   TP = VADD(Tl, TA);
				   TQ = VBYI(VADD(TN, TK));
				   T1X = VSUB(TP, TQ);
				   STM2(&(xo[26]), T1X, ovs, &(xo[2]));
				   STN2(&(xo[24]), T1L, T1X, ovs);
				   T1Y = VADD(TP, TQ);
				   STM2(&(xo[14]), T1Y, ovs, &(xo[2]));
				   STN2(&(xo[12]), T1P, T1Y, ovs);
			      }
			      {
				   V TT, TW, T1Z, T20;
				   TT = VSUB(TR, TS);
				   TW = VBYI(VADD(TU, TV));
				   T1Z = VSUB(TT, TW);
				   STM2(&(xo[38]), T1Z, ovs, &(xo[2]));
				   STN2(&(xo[36]), T1S, T1Z, ovs);
				   T20 = VADD(TT, TW);
				   STM2(&(xo[2]), T20, ovs, &(xo[2]));
				   STN2(&(xo[0]), T1J, T20, ovs);
			      }
			 }
		    }
	       }
	  }
     }
     VLEAVE();
}
static void n1bv_14(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs)
{
     DVK(KP900968867, +0.900968867902419126236102319507445051165919162);
     DVK(KP222520933, +0.222520933956314404288902564496794759466355569);
     DVK(KP623489801, +0.623489801858733530525004884004239810632274731);
     DVK(KP781831482, +0.781831482468029808708444526674057750232334519);
     DVK(KP974927912, +0.974927912181823607018131682993931217232785801);
     DVK(KP433883739, +0.433883739117558120475768332848358754609990728);
     {
	  INT i;
	  const R *xi;
	  R *xo;
	  xi = ii;
	  xo = io;
	  for (i = v; i > 0; i = i - VL, xi = xi + (VL * ivs), xo = xo + (VL * ovs), MAKE_VOLATILE_STRIDE(is), MAKE_VOLATILE_STRIDE(os)) {
	       V Tp, Ty, Tl, TL, Tq, TE, T7, TJ, Ts, TB, Te, TK, Tr, TH, Tn;
	       V To;
	       Tn = LD(&(xi[0]), ivs, &(xi[0]));
	       To = LD(&(xi[WS(is, 7)]), ivs, &(xi[WS(is, 1)]));
	       Tp = VSUB(Tn, To);
	       Ty = VADD(Tn, To);
	       {
		    V Th, TC, Tk, TD;
		    {
			 V Tf, Tg, Ti, Tj;
			 Tf = LD(&(xi[WS(is, 4)]), ivs, &(xi[0]));
			 Tg = LD(&(xi[WS(is, 11)]), ivs, &(xi[WS(is, 1)]));
			 Th = VSUB(Tf, Tg);
			 TC = VADD(Tf, Tg);
			 Ti = LD(&(xi[WS(is, 10)]), ivs, &(xi[0]));
			 Tj = LD(&(xi[WS(is, 3)]), ivs, &(xi[WS(is, 1)]));
			 Tk = VSUB(Ti, Tj);
			 TD = VADD(Ti, Tj);
		    }
		    Tl = VSUB(Th, Tk);
		    TL = VSUB(TD, TC);
		    Tq = VADD(Th, Tk);
		    TE = VADD(TC, TD);
	       }
	       {
		    V T3, Tz, T6, TA;
		    {
			 V T1, T2, T4, T5;
			 T1 = LD(&(xi[WS(is, 2)]), ivs, &(xi[0]));
			 T2 = LD(&(xi[WS(is, 9)]), ivs, &(xi[WS(is, 1)]));
			 T3 = VSUB(T1, T2);
			 Tz = VADD(T1, T2);
			 T4 = LD(&(xi[WS(is, 12)]), ivs, &(xi[0]));
			 T5 = LD(&(xi[WS(is, 5)]), ivs, &(xi[WS(is, 1)]));
			 T6 = VSUB(T4, T5);
			 TA = VADD(T4, T5);
		    }
		    T7 = VSUB(T3, T6);
		    TJ = VSUB(Tz, TA);
		    Ts = VADD(T3, T6);
		    TB = VADD(Tz, TA);
	       }
	       {
		    V Ta, TF, Td, TG;
		    {
			 V T8, T9, Tb, Tc;
			 T8 = LD(&(xi[WS(is, 6)]), ivs, &(xi[0]));
			 T9 = LD(&(xi[WS(is, 13)]), ivs, &(xi[WS(is, 1)]));
			 Ta = VSUB(T8, T9);
			 TF = VADD(T8, T9);
			 Tb = LD(&(xi[WS(is, 8)]), ivs, &(xi[0]));
			 Tc = LD(&(xi[WS(is, 1)]), ivs, &(xi[WS(is, 1)]));
			 Td = VSUB(Tb, Tc);
			 TG = VADD(Tb, Tc);
		    }
		    Te = VSUB(Ta, Td);
		    TK = VSUB(TG, TF);
		    Tr = VADD(Ta, Td);
		    TH = VADD(TF, TG);
	       }
	       ST(&(xo[WS(os, 7)]), VADD(Tp, VADD(Ts, VADD(Tq, Tr))), ovs, &(xo[WS(os, 1)]));
	       ST(&(xo[0]), VADD(Ty, VADD(TB, VADD(TE, TH))), ovs, &(xo[0]));
	       {
		    V Tm, Tt, TQ, TP;
		    Tm = VBYI(VFMA(LDK(KP433883739), T7, VFNMS(LDK(KP781831482), Tl, VMUL(LDK(KP974927912), Te))));
		    Tt = VFMA(LDK(KP623489801), Tq, VFNMS(LDK(KP222520933), Tr, VFNMS(LDK(KP900968867), Ts, Tp)));
		    ST(&(xo[WS(os, 3)]), VADD(Tm, Tt), ovs, &(xo[WS(os, 1)]));
		    ST(&(xo[WS(os, 11)]), VSUB(Tt, Tm), ovs, &(xo[WS(os, 1)]));
		    TQ = VBYI(VFMA(LDK(KP974927912), TJ, VFMA(LDK(KP433883739), TL, VMUL(LDK(KP781831482), TK))));
		    TP = VFMA(LDK(KP623489801), TH, VFNMS(LDK(KP900968867), TE, VFNMS(LDK(KP222520933), TB, Ty)));
		    ST(&(xo[WS(os, 12)]), VSUB(TP, TQ), ovs, &(xo[0]));
		    ST(&(xo[WS(os, 2)]), VADD(TP, TQ), ovs, &(xo[0]));
	       }
	       {
		    V Tu, Tv, TM, TI;
		    Tu = VBYI(VFMA(LDK(KP781831482), T7, VFMA(LDK(KP974927912), Tl, VMUL(LDK(KP433883739), Te))));
		    Tv = VFMA(LDK(KP623489801), Ts, VFNMS(LDK(KP900968867), Tr, VFNMS(LDK(KP222520933), Tq, Tp)));
		    ST(&(xo[WS(os, 1)]), VADD(Tu, Tv), ovs, &(xo[WS(os, 1)]));
		    ST(&(xo[WS(os, 13)]), VSUB(Tv, Tu), ovs, &(xo[WS(os, 1)]));
		    TM = VBYI(VFNMS(LDK(KP433883739), TK, VFNMS(LDK(KP974927912), TL, VMUL(LDK(KP781831482), TJ))));
		    TI = VFMA(LDK(KP623489801), TB, VFNMS(LDK(KP900968867), TH, VFNMS(LDK(KP222520933), TE, Ty)));
		    ST(&(xo[WS(os, 6)]), VSUB(TI, TM), ovs, &(xo[0]));
		    ST(&(xo[WS(os, 8)]), VADD(TI, TM), ovs, &(xo[0]));
	       }
	       {
		    V TO, TN, Tx, Tw;
		    TO = VBYI(VFMA(LDK(KP433883739), TJ, VFNMS(LDK(KP974927912), TK, VMUL(LDK(KP781831482), TL))));
		    TN = VFMA(LDK(KP623489801), TE, VFNMS(LDK(KP222520933), TH, VFNMS(LDK(KP900968867), TB, Ty)));
		    ST(&(xo[WS(os, 4)]), VSUB(TN, TO), ovs, &(xo[0]));
		    ST(&(xo[WS(os, 10)]), VADD(TN, TO), ovs, &(xo[0]));
		    Tx = VBYI(VFNMS(LDK(KP781831482), Te, VFNMS(LDK(KP433883739), Tl, VMUL(LDK(KP974927912), T7))));
		    Tw = VFMA(LDK(KP623489801), Tr, VFNMS(LDK(KP900968867), Tq, VFNMS(LDK(KP222520933), Ts, Tp)));
		    ST(&(xo[WS(os, 5)]), VSUB(Tw, Tx), ovs, &(xo[WS(os, 1)]));
		    ST(&(xo[WS(os, 9)]), VADD(Tx, Tw), ovs, &(xo[WS(os, 1)]));
	       }
	  }
     }
     VLEAVE();
}
Example #27
0
static void n1fv_10(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs)
{
     DVK(KP250000000, +0.250000000000000000000000000000000000000000000);
     DVK(KP559016994, +0.559016994374947424102293417182819058860154590);
     DVK(KP587785252, +0.587785252292473129168705954639072768597652438);
     DVK(KP951056516, +0.951056516295153572116439333379382143405698634);
     {
	  INT i;
	  const R *xi;
	  R *xo;
	  xi = ri;
	  xo = ro;
	  for (i = v; i > 0; i = i - VL, xi = xi + (VL * ivs), xo = xo + (VL * ovs), MAKE_VOLATILE_STRIDE(20, is), MAKE_VOLATILE_STRIDE(20, os)) {
	       V Ti, Ty, Tm, Tn, Tw, Tt, Tz, TA, TB, T7, Te, Tj, Tg, Th;
	       Tg = LD(&(xi[0]), ivs, &(xi[0]));
	       Th = LD(&(xi[WS(is, 5)]), ivs, &(xi[WS(is, 1)]));
	       Ti = VSUB(Tg, Th);
	       Ty = VADD(Tg, Th);
	       {
		    V T3, Tu, Td, Ts, T6, Tv, Ta, Tr;
		    {
			 V T1, T2, Tb, Tc;
			 T1 = LD(&(xi[WS(is, 2)]), ivs, &(xi[0]));
			 T2 = LD(&(xi[WS(is, 7)]), ivs, &(xi[WS(is, 1)]));
			 T3 = VSUB(T1, T2);
			 Tu = VADD(T1, T2);
			 Tb = LD(&(xi[WS(is, 6)]), ivs, &(xi[0]));
			 Tc = LD(&(xi[WS(is, 1)]), ivs, &(xi[WS(is, 1)]));
			 Td = VSUB(Tb, Tc);
			 Ts = VADD(Tb, Tc);
		    }
		    {
			 V T4, T5, T8, T9;
			 T4 = LD(&(xi[WS(is, 8)]), ivs, &(xi[0]));
			 T5 = LD(&(xi[WS(is, 3)]), ivs, &(xi[WS(is, 1)]));
			 T6 = VSUB(T4, T5);
			 Tv = VADD(T4, T5);
			 T8 = LD(&(xi[WS(is, 4)]), ivs, &(xi[0]));
			 T9 = LD(&(xi[WS(is, 9)]), ivs, &(xi[WS(is, 1)]));
			 Ta = VSUB(T8, T9);
			 Tr = VADD(T8, T9);
		    }
		    Tm = VSUB(T3, T6);
		    Tn = VSUB(Ta, Td);
		    Tw = VSUB(Tu, Tv);
		    Tt = VSUB(Tr, Ts);
		    Tz = VADD(Tu, Tv);
		    TA = VADD(Tr, Ts);
		    TB = VADD(Tz, TA);
		    T7 = VADD(T3, T6);
		    Te = VADD(Ta, Td);
		    Tj = VADD(T7, Te);
	       }
	       ST(&(xo[WS(os, 5)]), VADD(Ti, Tj), ovs, &(xo[WS(os, 1)]));
	       ST(&(xo[0]), VADD(Ty, TB), ovs, &(xo[0]));
	       {
		    V To, Tq, Tl, Tp, Tf, Tk;
		    To = VBYI(VFMA(LDK(KP951056516), Tm, VMUL(LDK(KP587785252), Tn)));
		    Tq = VBYI(VFNMS(LDK(KP587785252), Tm, VMUL(LDK(KP951056516), Tn)));
		    Tf = VMUL(LDK(KP559016994), VSUB(T7, Te));
		    Tk = VFNMS(LDK(KP250000000), Tj, Ti);
		    Tl = VADD(Tf, Tk);
		    Tp = VSUB(Tk, Tf);
		    ST(&(xo[WS(os, 1)]), VSUB(Tl, To), ovs, &(xo[WS(os, 1)]));
		    ST(&(xo[WS(os, 7)]), VADD(Tq, Tp), ovs, &(xo[WS(os, 1)]));
		    ST(&(xo[WS(os, 9)]), VADD(To, Tl), ovs, &(xo[WS(os, 1)]));
		    ST(&(xo[WS(os, 3)]), VSUB(Tp, Tq), ovs, &(xo[WS(os, 1)]));
	       }
	       {
		    V Tx, TF, TE, TG, TC, TD;
		    Tx = VBYI(VFNMS(LDK(KP587785252), Tw, VMUL(LDK(KP951056516), Tt)));
		    TF = VBYI(VFMA(LDK(KP951056516), Tw, VMUL(LDK(KP587785252), Tt)));
		    TC = VFNMS(LDK(KP250000000), TB, Ty);
		    TD = VMUL(LDK(KP559016994), VSUB(Tz, TA));
		    TE = VSUB(TC, TD);
		    TG = VADD(TD, TC);
		    ST(&(xo[WS(os, 2)]), VADD(Tx, TE), ovs, &(xo[0]));
		    ST(&(xo[WS(os, 6)]), VSUB(TG, TF), ovs, &(xo[0]));
		    ST(&(xo[WS(os, 8)]), VSUB(TE, Tx), ovs, &(xo[0]));
		    ST(&(xo[WS(os, 4)]), VADD(TF, TG), ovs, &(xo[0]));
	       }
	  }
     }
     VLEAVE();
}
static void n1bv_14(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs)
{
     DVK(KP900968867, +0.900968867902419126236102319507445051165919162);
     DVK(KP801937735, +0.801937735804838252472204639014890102331838324);
     DVK(KP974927912, +0.974927912181823607018131682993931217232785801);
     DVK(KP692021471, +0.692021471630095869627814897002069140197260599);
     DVK(KP554958132, +0.554958132087371191422194871006410481067288862);
     DVK(KP356895867, +0.356895867892209443894399510021300583399127187);
     {
	  INT i;
	  const R *xi;
	  R *xo;
	  xi = ii;
	  xo = io;
	  for (i = v; i > 0; i = i - VL, xi = xi + (VL * ivs), xo = xo + (VL * ovs), MAKE_VOLATILE_STRIDE(is), MAKE_VOLATILE_STRIDE(os)) {
	       V TH, T3, TP, Tn, Ta, Tu, TU, TK, TO, Tk, TM, Tg, TL, Td, T1;
	       V T2;
	       T1 = LD(&(xi[0]), ivs, &(xi[0]));
	       T2 = LD(&(xi[WS(is, 7)]), ivs, &(xi[WS(is, 1)]));
	       {
		    V Ti, TI, T6, TJ, T9, Tj, Te, Tf, Tb, Tc;
		    {
			 V T4, T5, T7, T8, Tl, Tm;
			 T4 = LD(&(xi[WS(is, 2)]), ivs, &(xi[0]));
			 T5 = LD(&(xi[WS(is, 9)]), ivs, &(xi[WS(is, 1)]));
			 T7 = LD(&(xi[WS(is, 12)]), ivs, &(xi[0]));
			 T8 = LD(&(xi[WS(is, 5)]), ivs, &(xi[WS(is, 1)]));
			 Tl = LD(&(xi[WS(is, 8)]), ivs, &(xi[0]));
			 Tm = LD(&(xi[WS(is, 1)]), ivs, &(xi[WS(is, 1)]));
			 Ti = LD(&(xi[WS(is, 6)]), ivs, &(xi[0]));
			 TH = VADD(T1, T2);
			 T3 = VSUB(T1, T2);
			 TI = VADD(T4, T5);
			 T6 = VSUB(T4, T5);
			 TJ = VADD(T7, T8);
			 T9 = VSUB(T7, T8);
			 TP = VADD(Tl, Tm);
			 Tn = VSUB(Tl, Tm);
			 Tj = LD(&(xi[WS(is, 13)]), ivs, &(xi[WS(is, 1)]));
			 Te = LD(&(xi[WS(is, 10)]), ivs, &(xi[0]));
			 Tf = LD(&(xi[WS(is, 3)]), ivs, &(xi[WS(is, 1)]));
			 Tb = LD(&(xi[WS(is, 4)]), ivs, &(xi[0]));
			 Tc = LD(&(xi[WS(is, 11)]), ivs, &(xi[WS(is, 1)]));
		    }
		    Ta = VADD(T6, T9);
		    Tu = VSUB(T6, T9);
		    TU = VSUB(TI, TJ);
		    TK = VADD(TI, TJ);
		    TO = VADD(Ti, Tj);
		    Tk = VSUB(Ti, Tj);
		    TM = VADD(Te, Tf);
		    Tg = VSUB(Te, Tf);
		    TL = VADD(Tb, Tc);
		    Td = VSUB(Tb, Tc);
	       }
	       {
		    V T13, TG, TY, T18, TB, Tw, TT, Tz, T11, T16, TE, Tr, TV, TQ;
		    TV = VSUB(TP, TO);
		    TQ = VADD(TO, TP);
		    {
			 V Ts, To, TW, TN;
			 Ts = VSUB(Tk, Tn);
			 To = VADD(Tk, Tn);
			 TW = VSUB(TM, TL);
			 TN = VADD(TL, TM);
			 {
			      V Tt, Th, TR, T12;
			      Tt = VSUB(Td, Tg);
			      Th = VADD(Td, Tg);
			      TR = VFNMS(LDK(KP356895867), TK, TQ);
			      T12 = VFNMS(LDK(KP554958132), TV, TU);
			      {
				   V Tx, TF, TZ, T14;
				   Tx = VFNMS(LDK(KP356895867), Ta, To);
				   TF = VFMA(LDK(KP554958132), Ts, Tu);
				   ST(&(xo[0]), VADD(TH, VADD(TK, VADD(TN, TQ))), ovs, &(xo[0]));
				   TZ = VFNMS(LDK(KP356895867), TN, TK);
				   T14 = VFNMS(LDK(KP356895867), TQ, TN);
				   {
					V TX, T17, TC, Tp;
					TX = VFMA(LDK(KP554958132), TW, TV);
					T17 = VFMA(LDK(KP554958132), TU, TW);
					ST(&(xo[WS(os, 7)]), VADD(T3, VADD(Ta, VADD(Th, To))), ovs, &(xo[WS(os, 1)]));
					TC = VFNMS(LDK(KP356895867), Th, Ta);
					Tp = VFNMS(LDK(KP356895867), To, Th);
					{
					     V TA, Tv, TS, Ty;
					     TA = VFMA(LDK(KP554958132), Tt, Ts);
					     Tv = VFNMS(LDK(KP554958132), Tu, Tt);
					     TS = VFNMS(LDK(KP692021471), TR, TN);
					     T13 = VMUL(LDK(KP974927912), VFNMS(LDK(KP801937735), T12, TW));
					     Ty = VFNMS(LDK(KP692021471), Tx, Th);
					     TG = VMUL(LDK(KP974927912), VFMA(LDK(KP801937735), TF, Tt));
					     {
						  V T10, T15, TD, Tq;
						  T10 = VFNMS(LDK(KP692021471), TZ, TQ);
						  T15 = VFNMS(LDK(KP692021471), T14, TK);
						  TY = VMUL(LDK(KP974927912), VFMA(LDK(KP801937735), TX, TU));
						  T18 = VMUL(LDK(KP974927912), VFNMS(LDK(KP801937735), T17, TV));
						  TD = VFNMS(LDK(KP692021471), TC, To);
						  Tq = VFNMS(LDK(KP692021471), Tp, Ta);
						  TB = VMUL(LDK(KP974927912), VFNMS(LDK(KP801937735), TA, Tu));
						  Tw = VMUL(LDK(KP974927912), VFNMS(LDK(KP801937735), Tv, Ts));
						  TT = VFNMS(LDK(KP900968867), TS, TH);
						  Tz = VFNMS(LDK(KP900968867), Ty, T3);
						  T11 = VFNMS(LDK(KP900968867), T10, TH);
						  T16 = VFNMS(LDK(KP900968867), T15, TH);
						  TE = VFNMS(LDK(KP900968867), TD, T3);
						  Tr = VFNMS(LDK(KP900968867), Tq, T3);
					     }
					}
				   }
			      }
			 }
		    }
		    ST(&(xo[WS(os, 2)]), VFMAI(TY, TT), ovs, &(xo[0]));
		    ST(&(xo[WS(os, 12)]), VFNMSI(TY, TT), ovs, &(xo[0]));
		    ST(&(xo[WS(os, 9)]), VFMAI(TB, Tz), ovs, &(xo[WS(os, 1)]));
		    ST(&(xo[WS(os, 5)]), VFNMSI(TB, Tz), ovs, &(xo[WS(os, 1)]));
		    ST(&(xo[WS(os, 6)]), VFMAI(T13, T11), ovs, &(xo[0]));
		    ST(&(xo[WS(os, 8)]), VFNMSI(T13, T11), ovs, &(xo[0]));
		    ST(&(xo[WS(os, 4)]), VFMAI(T18, T16), ovs, &(xo[0]));
		    ST(&(xo[WS(os, 10)]), VFNMSI(T18, T16), ovs, &(xo[0]));
		    ST(&(xo[WS(os, 13)]), VFNMSI(TG, TE), ovs, &(xo[WS(os, 1)]));
		    ST(&(xo[WS(os, 1)]), VFMAI(TG, TE), ovs, &(xo[WS(os, 1)]));
		    ST(&(xo[WS(os, 11)]), VFNMSI(Tw, Tr), ovs, &(xo[WS(os, 1)]));
		    ST(&(xo[WS(os, 3)]), VFMAI(Tw, Tr), ovs, &(xo[WS(os, 1)]));
	       }
	  }
     }
     VLEAVE();
}
Example #29
0
static void t1fv_15(R *ri, R *ii, const R *W, stride rs, INT mb, INT me, INT ms)
{
     DVK(KP216506350, +0.216506350946109661690930792688234045867850657);
     DVK(KP484122918, +0.484122918275927110647408174972799951354115213);
     DVK(KP866025403, +0.866025403784438646763723170752936183471402627);
     DVK(KP509036960, +0.509036960455127183450980863393907648510733164);
     DVK(KP823639103, +0.823639103546331925877420039278190003029660514);
     DVK(KP587785252, +0.587785252292473129168705954639072768597652438);
     DVK(KP951056516, +0.951056516295153572116439333379382143405698634);
     DVK(KP250000000, +0.250000000000000000000000000000000000000000000);
     DVK(KP559016994, +0.559016994374947424102293417182819058860154590);
     DVK(KP500000000, +0.500000000000000000000000000000000000000000000);
     INT m;
     R *x;
     x = ri;
     for (m = mb, W = W + (mb * ((TWVL / VL) * 28)); m < me; m = m + VL, x = x + (VL * ms), W = W + (TWVL * 28), MAKE_VOLATILE_STRIDE(rs)) {
	  V T1e, T7, TP, T12, T15, Tf, Tn, To, T1b, T1c, T1f, TQ, TR, TS, Tw;
	  V TE, TF, TT, TU, TV;
	  {
	       V T1, T5, T3, T4, T2, T6;
	       T1 = LD(&(x[0]), ms, &(x[0]));
	       T4 = LD(&(x[WS(rs, 10)]), ms, &(x[0]));
	       T5 = BYTWJ(&(W[TWVL * 18]), T4);
	       T2 = LD(&(x[WS(rs, 5)]), ms, &(x[WS(rs, 1)]));
	       T3 = BYTWJ(&(W[TWVL * 8]), T2);
	       T1e = VSUB(T5, T3);
	       T6 = VADD(T3, T5);
	       T7 = VADD(T1, T6);
	       TP = VFNMS(LDK(KP500000000), T6, T1);
	  }
	  {
	       V T9, Tq, Ty, Th, Te, T13, Tv, T10, TD, T11, Tm, T14;
	       {
		    V T8, Tp, Tx, Tg;
		    T8 = LD(&(x[WS(rs, 3)]), ms, &(x[WS(rs, 1)]));
		    T9 = BYTWJ(&(W[TWVL * 4]), T8);
		    Tp = LD(&(x[WS(rs, 6)]), ms, &(x[0]));
		    Tq = BYTWJ(&(W[TWVL * 10]), Tp);
		    Tx = LD(&(x[WS(rs, 9)]), ms, &(x[WS(rs, 1)]));
		    Ty = BYTWJ(&(W[TWVL * 16]), Tx);
		    Tg = LD(&(x[WS(rs, 12)]), ms, &(x[0]));
		    Th = BYTWJ(&(W[TWVL * 22]), Tg);
	       }
	       {
		    V Tb, Td, Ta, Tc;
		    Ta = LD(&(x[WS(rs, 8)]), ms, &(x[0]));
		    Tb = BYTWJ(&(W[TWVL * 14]), Ta);
		    Tc = LD(&(x[WS(rs, 13)]), ms, &(x[WS(rs, 1)]));
		    Td = BYTWJ(&(W[TWVL * 24]), Tc);
		    Te = VADD(Tb, Td);
		    T13 = VSUB(Td, Tb);
	       }
	       {
		    V Ts, Tu, Tr, Tt;
		    Tr = LD(&(x[WS(rs, 11)]), ms, &(x[WS(rs, 1)]));
		    Ts = BYTWJ(&(W[TWVL * 20]), Tr);
		    Tt = LD(&(x[WS(rs, 1)]), ms, &(x[WS(rs, 1)]));
		    Tu = BYTWJ(&(W[0]), Tt);
		    Tv = VADD(Ts, Tu);
		    T10 = VSUB(Tu, Ts);
	       }
	       {
		    V TA, TC, Tz, TB;
		    Tz = LD(&(x[WS(rs, 14)]), ms, &(x[0]));
		    TA = BYTWJ(&(W[TWVL * 26]), Tz);
		    TB = LD(&(x[WS(rs, 4)]), ms, &(x[0]));
		    TC = BYTWJ(&(W[TWVL * 6]), TB);
		    TD = VADD(TA, TC);
		    T11 = VSUB(TC, TA);
	       }
	       {
		    V Tj, Tl, Ti, Tk;
		    Ti = LD(&(x[WS(rs, 2)]), ms, &(x[0]));
		    Tj = BYTWJ(&(W[TWVL * 2]), Ti);
		    Tk = LD(&(x[WS(rs, 7)]), ms, &(x[WS(rs, 1)]));
		    Tl = BYTWJ(&(W[TWVL * 12]), Tk);
		    Tm = VADD(Tj, Tl);
		    T14 = VSUB(Tl, Tj);
	       }
	       T12 = VSUB(T10, T11);
	       T15 = VSUB(T13, T14);
	       Tf = VADD(T9, Te);
	       Tn = VADD(Th, Tm);
	       To = VADD(Tf, Tn);
	       T1b = VADD(T13, T14);
	       T1c = VADD(T10, T11);
	       T1f = VADD(T1b, T1c);
	       TQ = VFNMS(LDK(KP500000000), Te, T9);
	       TR = VFNMS(LDK(KP500000000), Tm, Th);
	       TS = VADD(TQ, TR);
	       Tw = VADD(Tq, Tv);
	       TE = VADD(Ty, TD);
	       TF = VADD(Tw, TE);
	       TT = VFNMS(LDK(KP500000000), Tv, Tq);
	       TU = VFNMS(LDK(KP500000000), TD, Ty);
	       TV = VADD(TT, TU);
	  }
	  {
	       V TI, TG, TH, TM, TO, TK, TL, TN, TJ;
	       TI = VMUL(LDK(KP559016994), VSUB(To, TF));
	       TG = VADD(To, TF);
	       TH = VFNMS(LDK(KP250000000), TG, T7);
	       TK = VSUB(Tw, TE);
	       TL = VSUB(Tf, Tn);
	       TM = VBYI(VFNMS(LDK(KP587785252), TL, VMUL(LDK(KP951056516), TK)));
	       TO = VBYI(VFMA(LDK(KP951056516), TL, VMUL(LDK(KP587785252), TK)));
	       ST(&(x[0]), VADD(T7, TG), ms, &(x[0]));
	       TN = VADD(TI, TH);
	       ST(&(x[WS(rs, 6)]), VSUB(TN, TO), ms, &(x[0]));
	       ST(&(x[WS(rs, 9)]), VADD(TO, TN), ms, &(x[WS(rs, 1)]));
	       TJ = VSUB(TH, TI);
	       ST(&(x[WS(rs, 3)]), VSUB(TJ, TM), ms, &(x[WS(rs, 1)]));
	       ST(&(x[WS(rs, 12)]), VADD(TM, TJ), ms, &(x[0]));
	  }
	  {
	       V T16, T1m, T1u, T1h, T1o, T1a, T1p, TZ, T1t, T1l, T1d, T1g;
	       T16 = VFNMS(LDK(KP509036960), T15, VMUL(LDK(KP823639103), T12));
	       T1m = VFMA(LDK(KP823639103), T15, VMUL(LDK(KP509036960), T12));
	       T1u = VBYI(VMUL(LDK(KP866025403), VADD(T1e, T1f)));
	       T1d = VMUL(LDK(KP484122918), VSUB(T1b, T1c));
	       T1g = VFNMS(LDK(KP216506350), T1f, VMUL(LDK(KP866025403), T1e));
	       T1h = VSUB(T1d, T1g);
	       T1o = VADD(T1d, T1g);
	       {
		    V T18, T19, TY, TW, TX;
		    T18 = VSUB(TT, TU);
		    T19 = VSUB(TQ, TR);
		    T1a = VFNMS(LDK(KP587785252), T19, VMUL(LDK(KP951056516), T18));
		    T1p = VFMA(LDK(KP951056516), T19, VMUL(LDK(KP587785252), T18));
		    TY = VMUL(LDK(KP559016994), VSUB(TS, TV));
		    TW = VADD(TS, TV);
		    TX = VFNMS(LDK(KP250000000), TW, TP);
		    TZ = VSUB(TX, TY);
		    T1t = VADD(TP, TW);
		    T1l = VADD(TY, TX);
	       }
	       {
		    V T17, T1i, T1r, T1s;
		    ST(&(x[WS(rs, 5)]), VSUB(T1t, T1u), ms, &(x[WS(rs, 1)]));
		    ST(&(x[WS(rs, 10)]), VADD(T1t, T1u), ms, &(x[0]));
		    T17 = VSUB(TZ, T16);
		    T1i = VBYI(VSUB(T1a, T1h));
		    ST(&(x[WS(rs, 8)]), VSUB(T17, T1i), ms, &(x[0]));
		    ST(&(x[WS(rs, 7)]), VADD(T17, T1i), ms, &(x[WS(rs, 1)]));
		    T1r = VSUB(T1l, T1m);
		    T1s = VBYI(VADD(T1p, T1o));
		    ST(&(x[WS(rs, 11)]), VSUB(T1r, T1s), ms, &(x[WS(rs, 1)]));
		    ST(&(x[WS(rs, 4)]), VADD(T1r, T1s), ms, &(x[0]));
		    {
			 V T1n, T1q, T1j, T1k;
			 T1n = VADD(T1l, T1m);
			 T1q = VBYI(VSUB(T1o, T1p));
			 ST(&(x[WS(rs, 14)]), VSUB(T1n, T1q), ms, &(x[0]));
			 ST(&(x[WS(rs, 1)]), VADD(T1n, T1q), ms, &(x[WS(rs, 1)]));
			 T1j = VADD(TZ, T16);
			 T1k = VBYI(VADD(T1a, T1h));
			 ST(&(x[WS(rs, 13)]), VSUB(T1j, T1k), ms, &(x[WS(rs, 1)]));
			 ST(&(x[WS(rs, 2)]), VADD(T1j, T1k), ms, &(x[0]));
		    }
	       }
	  }
     }
}
Example #30
0
static void hc2cbdftv_16(R *Rp, R *Ip, R *Rm, R *Im, 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;
	  for (m = mb, W = W + ((mb - 1) * ((TWVL / VL) * 30)); m < me; m = m + VL, Rp = Rp + (VL * ms), Ip = Ip + (VL * ms), Rm = Rm - (VL * ms), Im = Im - (VL * ms), W = W + (TWVL * 30), MAKE_VOLATILE_STRIDE(64, rs)) {
	       V T8, Tv, TE, T1t, TP, T1w, T10, T1p, Tn, Tw, T13, T1q, TL, T1x, TS;
	       V T1u;
	       {
		    V T4, TA, Tu, TC, T7, TN, Tr, TB, T2, T3, Ts, Tt, T5, T6, Tp;
		    V Tq, TD, TO, TY, TZ, Tb, TF, Tl, TJ, Te, TG, Ti, TI, T9, Ta;
		    V Tj, Tk, Tc, Td, Tg, Th, Tf, Tm, T11, T12, TH, TK, TQ, TR;
		    T2 = LD(&(Rp[0]), ms, &(Rp[0]));
		    T3 = LD(&(Rm[WS(rs, 7)]), -ms, &(Rm[WS(rs, 1)]));
		    T4 = VFMACONJ(T3, T2);
		    TA = VFNMSCONJ(T3, T2);
		    Ts = LD(&(Rp[WS(rs, 6)]), ms, &(Rp[0]));
		    Tt = LD(&(Rm[WS(rs, 1)]), -ms, &(Rm[WS(rs, 1)]));
		    Tu = VFMACONJ(Tt, Ts);
		    TC = VFMSCONJ(Tt, Ts);
		    T5 = LD(&(Rp[WS(rs, 4)]), ms, &(Rp[0]));
		    T6 = LD(&(Rm[WS(rs, 3)]), -ms, &(Rm[WS(rs, 1)]));
		    T7 = VFMACONJ(T6, T5);
		    TN = VFNMSCONJ(T6, T5);
		    Tp = LD(&(Rp[WS(rs, 2)]), ms, &(Rp[0]));
		    Tq = LD(&(Rm[WS(rs, 5)]), -ms, &(Rm[WS(rs, 1)]));
		    Tr = VFMACONJ(Tq, Tp);
		    TB = VFNMSCONJ(Tq, Tp);
		    T8 = VSUB(T4, T7);
		    Tv = VSUB(Tr, Tu);
		    TD = VADD(TB, TC);
		    TE = VFMA(LDK(KP707106781), TD, TA);
		    T1t = VFNMS(LDK(KP707106781), TD, TA);
		    TO = VSUB(TB, TC);
		    TP = VFMA(LDK(KP707106781), TO, TN);
		    T1w = VFNMS(LDK(KP707106781), TO, TN);
		    TY = VADD(T4, T7);
		    TZ = VADD(Tr, Tu);
		    T10 = VADD(TY, TZ);
		    T1p = VSUB(TY, TZ);
		    T9 = LD(&(Rp[WS(rs, 1)]), ms, &(Rp[WS(rs, 1)]));
		    Ta = LD(&(Rm[WS(rs, 6)]), -ms, &(Rm[0]));
		    Tb = VFMACONJ(Ta, T9);
		    TF = VFNMSCONJ(Ta, T9);
		    Tj = LD(&(Rp[WS(rs, 3)]), ms, &(Rp[WS(rs, 1)]));
		    Tk = LD(&(Rm[WS(rs, 4)]), -ms, &(Rm[0]));
		    Tl = VFMACONJ(Tk, Tj);
		    TJ = VFNMSCONJ(Tk, Tj);
		    Tc = LD(&(Rp[WS(rs, 5)]), ms, &(Rp[WS(rs, 1)]));
		    Td = LD(&(Rm[WS(rs, 2)]), -ms, &(Rm[0]));
		    Te = VFMACONJ(Td, Tc);
		    TG = VFNMSCONJ(Td, Tc);
		    Tg = LD(&(Rp[WS(rs, 7)]), ms, &(Rp[WS(rs, 1)]));
		    Th = LD(&(Rm[0]), -ms, &(Rm[0]));
		    Ti = VFMACONJ(Th, Tg);
		    TI = VFMSCONJ(Th, Tg);
		    Tf = VSUB(Tb, Te);
		    Tm = VSUB(Ti, Tl);
		    Tn = VADD(Tf, Tm);
		    Tw = VSUB(Tf, Tm);
		    T11 = VADD(Tb, Te);
		    T12 = VADD(Ti, Tl);
		    T13 = VADD(T11, T12);
		    T1q = VSUB(T11, T12);
		    TH = VFNMS(LDK(KP414213562), TG, TF);
		    TK = VFMA(LDK(KP414213562), TJ, TI);
		    TL = VADD(TH, TK);
		    T1x = VSUB(TH, TK);
		    TQ = VFMA(LDK(KP414213562), TF, TG);
		    TR = VFNMS(LDK(KP414213562), TI, TJ);
		    TS = VADD(TQ, TR);
		    T1u = VSUB(TQ, TR);
	       }
	       {
		    V T1j, T1R, T1c, T1J, T1g, T1l, T1N, T1T, T1Q, T1a, T1b, T19, T1I, T1e, T1f;
		    V T1d, T1k, T1L, T1M, T1K, T1S, T1h, T1U, T1V, T1i, T1m, T1O, T1P, T1n, T14;
		    V T1r, Ty, T1D, TU, T16, T1z, T1F, TX, T1o, To, Tx, T1, T1C, TM, TT;
		    V Tz, T15, T1v, T1y, T1s, T1E, TV, T1G, T1H, TW, T17, T1A, T1B, T18;
		    T1j = VADD(T10, T13);
		    T1Q = LDW(&(W[TWVL * 22]));
		    T1R = VZMUL(T1Q, VFNMSI(T1q, T1p));
		    T1a = VFMA(LDK(KP707106781), Tn, T8);
		    T1b = VFMA(LDK(KP707106781), Tw, Tv);
		    T19 = LDW(&(W[TWVL * 26]));
		    T1c = VZMUL(T19, VFNMSI(T1b, T1a));
		    T1I = LDW(&(W[TWVL * 2]));
		    T1J = VZMUL(T1I, VFMAI(T1b, T1a));
		    T1e = VFMA(LDK(KP923879532), TL, TE);
		    T1f = VFMA(LDK(KP923879532), TS, TP);
		    T1d = LDW(&(W[TWVL * 28]));
		    T1g = VZMULI(T1d, VFNMSI(T1f, T1e));
		    T1k = LDW(&(W[0]));
		    T1l = VZMULI(T1k, VFMAI(T1f, T1e));
		    T1L = VFMA(LDK(KP923879532), T1u, T1t);
		    T1M = VFNMS(LDK(KP923879532), T1x, T1w);
		    T1K = LDW(&(W[TWVL * 4]));
		    T1N = VZMULI(T1K, VFNMSI(T1M, T1L));
		    T1S = LDW(&(W[TWVL * 24]));
		    T1T = VZMULI(T1S, VFMAI(T1M, T1L));
		    T1h = VCONJ(VSUB(T1c, T1g));
		    ST(&(Rm[WS(rs, 7)]), T1h, -ms, &(Rm[WS(rs, 1)]));
		    T1U = VCONJ(VSUB(T1R, T1T));
		    ST(&(Rm[WS(rs, 6)]), T1U, -ms, &(Rm[0]));
		    T1V = VADD(T1R, T1T);
		    ST(&(Rp[WS(rs, 6)]), T1V, ms, &(Rp[0]));
		    T1i = VADD(T1c, T1g);
		    ST(&(Rp[WS(rs, 7)]), T1i, ms, &(Rp[WS(rs, 1)]));
		    T1m = VCONJ(VSUB(T1j, T1l));
		    ST(&(Rm[0]), T1m, -ms, &(Rm[0]));
		    T1O = VCONJ(VSUB(T1J, T1N));
		    ST(&(Rm[WS(rs, 1)]), T1O, -ms, &(Rm[WS(rs, 1)]));
		    T1P = VADD(T1J, T1N);
		    ST(&(Rp[WS(rs, 1)]), T1P, ms, &(Rp[WS(rs, 1)]));
		    T1n = VADD(T1j, T1l);
		    ST(&(Rp[0]), T1n, ms, &(Rp[0]));
		    TX = LDW(&(W[TWVL * 14]));
		    T14 = VZMUL(TX, VSUB(T10, T13));
		    T1o = LDW(&(W[TWVL * 6]));
		    T1r = VZMUL(T1o, VFMAI(T1q, T1p));
		    To = VFNMS(LDK(KP707106781), Tn, T8);
		    Tx = VFNMS(LDK(KP707106781), Tw, Tv);
		    T1 = LDW(&(W[TWVL * 10]));
		    Ty = VZMUL(T1, VFNMSI(Tx, To));
		    T1C = LDW(&(W[TWVL * 18]));
		    T1D = VZMUL(T1C, VFMAI(Tx, To));
		    TM = VFNMS(LDK(KP923879532), TL, TE);
		    TT = VFNMS(LDK(KP923879532), TS, TP);
		    Tz = LDW(&(W[TWVL * 12]));
		    TU = VZMULI(Tz, VFNMSI(TT, TM));
		    T15 = LDW(&(W[TWVL * 16]));
		    T16 = VZMULI(T15, VFMAI(TT, TM));
		    T1v = VFNMS(LDK(KP923879532), T1u, T1t);
		    T1y = VFMA(LDK(KP923879532), T1x, T1w);
		    T1s = LDW(&(W[TWVL * 8]));
		    T1z = VZMULI(T1s, VFMAI(T1y, T1v));
		    T1E = LDW(&(W[TWVL * 20]));
		    T1F = VZMULI(T1E, VFNMSI(T1y, T1v));
		    TV = VCONJ(VSUB(Ty, TU));
		    ST(&(Rm[WS(rs, 3)]), TV, -ms, &(Rm[WS(rs, 1)]));
		    T1G = VCONJ(VSUB(T1D, T1F));
		    ST(&(Rm[WS(rs, 5)]), T1G, -ms, &(Rm[WS(rs, 1)]));
		    T1H = VADD(T1D, T1F);
		    ST(&(Rp[WS(rs, 5)]), T1H, ms, &(Rp[WS(rs, 1)]));
		    TW = VADD(Ty, TU);
		    ST(&(Rp[WS(rs, 3)]), TW, ms, &(Rp[WS(rs, 1)]));
		    T17 = VCONJ(VSUB(T14, T16));
		    ST(&(Rm[WS(rs, 4)]), T17, -ms, &(Rm[0]));
		    T1A = VCONJ(VSUB(T1r, T1z));
		    ST(&(Rm[WS(rs, 2)]), T1A, -ms, &(Rm[0]));
		    T1B = VADD(T1r, T1z);
		    ST(&(Rp[WS(rs, 2)]), T1B, ms, &(Rp[0]));
		    T18 = VADD(T14, T16);
		    ST(&(Rp[WS(rs, 4)]), T18, ms, &(Rp[0]));
	       }
	  }
     }
     VLEAVE();
}