예제 #1
0
static void hc2cfdftv_4(R *Rp, R *Ip, R *Rm, R *Im, const R *W, stride rs, INT mb, INT me, INT ms)
{
     DVK(KP500000000, +0.500000000000000000000000000000000000000000000);
     {
	  INT m;
	  for (m = mb, W = W + ((mb - 1) * ((TWVL / VL) * 6)); m < me; m = m + VL, Rp = Rp + (VL * ms), Ip = Ip + (VL * ms), Rm = Rm - (VL * ms), Im = Im - (VL * ms), W = W + (TWVL * 6), MAKE_VOLATILE_STRIDE(16, rs)) {
	       V T1, T2, Tb, T5, T6, T4, T9, T3, Tc, T7, Ta, Tg, T8, Td, Th;
	       V Tf, Te, Ti, Tj;
	       T1 = LD(&(Rp[0]), ms, &(Rp[0]));
	       T2 = LD(&(Rm[0]), -ms, &(Rm[0]));
	       Tb = LDW(&(W[0]));
	       T5 = LD(&(Rp[WS(rs, 1)]), ms, &(Rp[WS(rs, 1)]));
	       T6 = LD(&(Rm[WS(rs, 1)]), -ms, &(Rm[WS(rs, 1)]));
	       T4 = LDW(&(W[TWVL * 2]));
	       T9 = LDW(&(W[TWVL * 4]));
	       T3 = VFMACONJ(T2, T1);
	       Tc = VZMULIJ(Tb, VFNMSCONJ(T2, T1));
	       T7 = VZMULJ(T4, VFMACONJ(T6, T5));
	       Ta = VZMULIJ(T9, VFNMSCONJ(T6, T5));
	       Tg = VADD(T3, T7);
	       T8 = VSUB(T3, T7);
	       Td = VSUB(Ta, Tc);
	       Th = VADD(Tc, Ta);
	       Tf = VCONJ(VMUL(LDK(KP500000000), VFMAI(Td, T8)));
	       Te = VMUL(LDK(KP500000000), VFNMSI(Td, T8));
	       Ti = VMUL(LDK(KP500000000), VSUB(Tg, Th));
	       Tj = VCONJ(VMUL(LDK(KP500000000), VADD(Th, Tg)));
	       ST(&(Rm[0]), Tf, -ms, &(Rm[0]));
	       ST(&(Rp[WS(rs, 1)]), Te, ms, &(Rp[WS(rs, 1)]));
	       ST(&(Rp[0]), Ti, ms, &(Rp[0]));
	       ST(&(Rm[WS(rs, 1)]), Tj, -ms, &(Rm[WS(rs, 1)]));
	  }
     }
     VLEAVE();
}
예제 #2
0
static void hc2cfdftv_4(R *Rp, R *Ip, R *Rm, R *Im, const R *W, stride rs, INT mb, INT me, INT ms)
{
     DVK(KP500000000, +0.500000000000000000000000000000000000000000000);
     INT m;
     for (m = mb, W = W + ((mb - 1) * ((TWVL / VL) * 6)); m < me; m = m + VL, Rp = Rp + (VL * ms), Ip = Ip + (VL * ms), Rm = Rm - (VL * ms), Im = Im - (VL * ms), W = W + (TWVL * 6), MAKE_VOLATILE_STRIDE(rs)) {
	  V T4, Tc, T9, Te, T1, T3, T2, Tb, T6, T8, T7, T5, Td, Tg, Th;
	  V Ta, Tf, Tk, Tl, Ti, Tj;
	  T1 = LD(&(Rp[0]), ms, &(Rp[0]));
	  T2 = LD(&(Rm[0]), -ms, &(Rm[0]));
	  T3 = VCONJ(T2);
	  T4 = VADD(T1, T3);
	  Tb = LDW(&(W[0]));
	  Tc = VZMULIJ(Tb, VSUB(T3, T1));
	  T6 = LD(&(Rp[WS(rs, 1)]), ms, &(Rp[WS(rs, 1)]));
	  T7 = LD(&(Rm[WS(rs, 1)]), -ms, &(Rm[WS(rs, 1)]));
	  T8 = VCONJ(T7);
	  T5 = LDW(&(W[TWVL * 2]));
	  T9 = VZMULJ(T5, VADD(T6, T8));
	  Td = LDW(&(W[TWVL * 4]));
	  Te = VZMULIJ(Td, VSUB(T8, T6));
	  Ta = VSUB(T4, T9);
	  Tf = VBYI(VSUB(Tc, Te));
	  Tg = VMUL(LDK(KP500000000), VSUB(Ta, Tf));
	  Th = VCONJ(VMUL(LDK(KP500000000), VADD(Ta, Tf)));
	  ST(&(Rp[WS(rs, 1)]), Tg, ms, &(Rp[WS(rs, 1)]));
	  ST(&(Rm[0]), Th, -ms, &(Rm[0]));
	  Ti = VADD(T4, T9);
	  Tj = VADD(Tc, Te);
	  Tk = VCONJ(VMUL(LDK(KP500000000), VSUB(Ti, Tj)));
	  Tl = VMUL(LDK(KP500000000), VADD(Ti, Tj));
	  ST(&(Rm[WS(rs, 1)]), Tk, -ms, &(Rm[WS(rs, 1)]));
	  ST(&(Rp[0]), Tl, ms, &(Rp[0]));
     }
}
예제 #3
0
static void hc2cfdftv_6(R *Rp, R *Ip, R *Rm, R *Im, const R *W, stride rs, INT mb, INT me, INT ms)
{
     DVK(KP250000000, +0.250000000000000000000000000000000000000000000);
     DVK(KP866025403, +0.866025403784438646763723170752936183471402627);
     DVK(KP500000000, +0.500000000000000000000000000000000000000000000);
     {
	  INT m;
	  for (m = mb, W = W + ((mb - 1) * ((TWVL / VL) * 10)); m < me; m = m + VL, Rp = Rp + (VL * ms), Ip = Ip + (VL * ms), Rm = Rm - (VL * ms), Im = Im - (VL * ms), W = W + (TWVL * 10), MAKE_VOLATILE_STRIDE(24, rs)) {
	       V Ta, Tu, Tn, Tw, Ti, Tv, T1, T8, Tg, Tf, T7, T3, Te, T6, T2;
	       V T4, T9, T5, Tk, Tm, Tj, Tl, Tc, Th, Tb, Td, Tr, Tp, Tq, To;
	       V Tt, Ts, TA, Ty, Tz, Tx, TC, TB;
	       T1 = LD(&(Rp[0]), ms, &(Rp[0]));
	       T8 = LD(&(Rp[WS(rs, 1)]), ms, &(Rp[WS(rs, 1)]));
	       Tg = LD(&(Rp[WS(rs, 2)]), ms, &(Rp[0]));
	       Te = LD(&(Rm[WS(rs, 2)]), -ms, &(Rm[0]));
	       Tf = VCONJ(Te);
	       T6 = LD(&(Rm[WS(rs, 1)]), -ms, &(Rm[WS(rs, 1)]));
	       T7 = VCONJ(T6);
	       T2 = LD(&(Rm[0]), -ms, &(Rm[0]));
	       T3 = VCONJ(T2);
	       T4 = VADD(T1, T3);
	       T5 = LDW(&(W[TWVL * 4]));
	       T9 = VZMULIJ(T5, VSUB(T7, T8));
	       Ta = VADD(T4, T9);
	       Tu = VSUB(T4, T9);
	       Tj = LDW(&(W[0]));
	       Tk = VZMULIJ(Tj, VSUB(T3, T1));
	       Tl = LDW(&(W[TWVL * 6]));
	       Tm = VZMULJ(Tl, VADD(Tf, Tg));
	       Tn = VADD(Tk, Tm);
	       Tw = VSUB(Tm, Tk);
	       Tb = LDW(&(W[TWVL * 2]));
	       Tc = VZMULJ(Tb, VADD(T7, T8));
	       Td = LDW(&(W[TWVL * 8]));
	       Th = VZMULIJ(Td, VSUB(Tf, Tg));
	       Ti = VADD(Tc, Th);
	       Tv = VSUB(Tc, Th);
	       Tr = VMUL(LDK(KP500000000), VBYI(VMUL(LDK(KP866025403), VSUB(Tn, Ti))));
	       To = VADD(Ti, Tn);
	       Tp = VMUL(LDK(KP500000000), VADD(Ta, To));
	       Tq = VFNMS(LDK(KP250000000), To, VMUL(LDK(KP500000000), Ta));
	       ST(&(Rp[0]), Tp, ms, &(Rp[0]));
	       Tt = VCONJ(VADD(Tq, Tr));
	       ST(&(Rm[WS(rs, 1)]), Tt, -ms, &(Rm[WS(rs, 1)]));
	       Ts = VSUB(Tq, Tr);
	       ST(&(Rp[WS(rs, 2)]), Ts, ms, &(Rp[0]));
	       TA = VMUL(LDK(KP500000000), VBYI(VMUL(LDK(KP866025403), VSUB(Tw, Tv))));
	       Tx = VADD(Tv, Tw);
	       Ty = VCONJ(VMUL(LDK(KP500000000), VADD(Tu, Tx)));
	       Tz = VFNMS(LDK(KP250000000), Tx, VMUL(LDK(KP500000000), Tu));
	       ST(&(Rm[WS(rs, 2)]), Ty, -ms, &(Rm[0]));
	       TC = VADD(Tz, TA);
	       ST(&(Rp[WS(rs, 1)]), TC, ms, &(Rp[WS(rs, 1)]));
	       TB = VCONJ(VSUB(Tz, TA));
	       ST(&(Rm[0]), TB, -ms, &(Rm[0]));
	  }
     }
     VLEAVE();
}
예제 #4
0
static void hc2cfdftv_6(R *Rp, R *Ip, R *Rm, R *Im, const R *W, stride rs, INT mb, INT me, INT ms)
{
     DVK(KP500000000, +0.500000000000000000000000000000000000000000000);
     DVK(KP866025403, +0.866025403784438646763723170752936183471402627);
     {
	  INT m;
	  for (m = mb, W = W + ((mb - 1) * ((TWVL / VL) * 10)); m < me; m = m + VL, Rp = Rp + (VL * ms), Ip = Ip + (VL * ms), Rm = Rm - (VL * ms), Im = Im - (VL * ms), W = W + (TWVL * 10), MAKE_VOLATILE_STRIDE(24, rs)) {
	       V T5, T6, T3, Tj, T4, T9, Te, Th, T1, T2, Ti, Tc, Td, Tb, Tg;
	       V T7, Ta, Tt, Tk, Tr, T8, Ts, Tf, Tx, Tu, To, Tl, Tw, Tv, Tn;
	       V Tm, Tz, Ty, Tp, Tq;
	       T1 = LD(&(Rp[0]), ms, &(Rp[0]));
	       T2 = LD(&(Rm[0]), -ms, &(Rm[0]));
	       Ti = LDW(&(W[0]));
	       Tc = LD(&(Rp[WS(rs, 2)]), ms, &(Rp[0]));
	       Td = LD(&(Rm[WS(rs, 2)]), -ms, &(Rm[0]));
	       Tb = LDW(&(W[TWVL * 8]));
	       Tg = LDW(&(W[TWVL * 6]));
	       T5 = LD(&(Rp[WS(rs, 1)]), ms, &(Rp[WS(rs, 1)]));
	       T6 = LD(&(Rm[WS(rs, 1)]), -ms, &(Rm[WS(rs, 1)]));
	       T3 = VFMACONJ(T2, T1);
	       Tj = VZMULIJ(Ti, VFNMSCONJ(T2, T1));
	       T4 = LDW(&(W[TWVL * 4]));
	       T9 = LDW(&(W[TWVL * 2]));
	       Te = VZMULIJ(Tb, VFNMSCONJ(Td, Tc));
	       Th = VZMULJ(Tg, VFMACONJ(Td, Tc));
	       T7 = VZMULIJ(T4, VFNMSCONJ(T6, T5));
	       Ta = VZMULJ(T9, VFMACONJ(T6, T5));
	       Tt = VADD(Tj, Th);
	       Tk = VSUB(Th, Tj);
	       Tr = VADD(T3, T7);
	       T8 = VSUB(T3, T7);
	       Ts = VADD(Ta, Te);
	       Tf = VSUB(Ta, Te);
	       Tx = VMUL(LDK(KP866025403), VSUB(Tt, Ts));
	       Tu = VADD(Ts, Tt);
	       To = VMUL(LDK(KP866025403), VSUB(Tk, Tf));
	       Tl = VADD(Tf, Tk);
	       Tw = VFNMS(LDK(KP500000000), Tu, Tr);
	       Tv = VCONJ(VMUL(LDK(KP500000000), VADD(Tr, Tu)));
	       Tn = VFNMS(LDK(KP500000000), Tl, T8);
	       Tm = VMUL(LDK(KP500000000), VADD(T8, Tl));
	       Tz = VMUL(LDK(KP500000000), VFMAI(Tx, Tw));
	       Ty = VCONJ(VMUL(LDK(KP500000000), VFNMSI(Tx, Tw)));
	       ST(&(Rm[WS(rs, 2)]), Tv, -ms, &(Rm[0]));
	       Tp = VMUL(LDK(KP500000000), VFNMSI(To, Tn));
	       Tq = VCONJ(VMUL(LDK(KP500000000), VFMAI(To, Tn)));
	       ST(&(Rp[0]), Tm, ms, &(Rp[0]));
	       ST(&(Rp[WS(rs, 1)]), Tz, ms, &(Rp[WS(rs, 1)]));
	       ST(&(Rm[0]), Ty, -ms, &(Rm[0]));
	       ST(&(Rm[WS(rs, 1)]), Tq, -ms, &(Rm[WS(rs, 1)]));
	       ST(&(Rp[WS(rs, 2)]), Tp, ms, &(Rp[0]));
	  }
     }
     VLEAVE();
}
예제 #5
0
static void hc2cfdftv_2(R *Rp, R *Ip, R *Rm, R *Im, const R *W, stride rs, INT mb, INT me, INT ms)
{
     DVK(KP500000000, +0.500000000000000000000000000000000000000000000);
     INT m;
     for (m = mb, W = W + ((mb - 1) * ((TWVL / VL) * 2)); m < me; m = m + VL, Rp = Rp + (VL * ms), Ip = Ip + (VL * ms), Rm = Rm - (VL * ms), Im = Im - (VL * ms), W = W + (TWVL * 2), MAKE_VOLATILE_STRIDE(rs)) {
	  V T1, T2, T4, T3, T5, T7, T6;
	  T1 = LD(&(Rp[0]), ms, &(Rp[0]));
	  T2 = LD(&(Rm[0]), -ms, &(Rm[0]));
	  T4 = LDW(&(W[0]));
	  T3 = VFMACONJ(T2, T1);
	  T5 = VZMULIJ(T4, VFNMSCONJ(T2, T1));
	  T7 = VCONJ(VMUL(LDK(KP500000000), VADD(T3, T5)));
	  T6 = VMUL(LDK(KP500000000), VSUB(T3, T5));
	  ST(&(Rm[0]), T7, -ms, &(Rm[0]));
	  ST(&(Rp[0]), T6, ms, &(Rp[0]));
     }
}
예제 #6
0
static void hc2cfdftv_8(R *Rp, R *Ip, R *Rm, R *Im, const R *W, stride rs, INT mb, INT me, INT ms)
{
     DVK(KP500000000, +0.500000000000000000000000000000000000000000000);
     DVK(KP707106781, +0.707106781186547524400844362104849039284835938);
     {
	  INT m;
	  for (m = mb, W = W + ((mb - 1) * ((TWVL / VL) * 14)); m < me; m = m + VL, Rp = Rp + (VL * ms), Ip = Ip + (VL * ms), Rm = Rm - (VL * ms), Im = Im - (VL * ms), W = W + (TWVL * 14), MAKE_VOLATILE_STRIDE(32, rs)) {
	       V T3, Tc, Tl, Ts, Tf, Tg, Te, Tp, T7, Ta, T1, T2, Tb, Tj, Tk;
	       V Ti, Tr, T5, T6, T4, T9, Th, Tq, TC, T8, Td, TF, Tm, TG, TD;
	       V Tt, Tu, Tn, TH, TL, TE, TK, Tz, Tv, Ty, To, TJ, TI, TN, TM;
	       V TB, TA, Tx, Tw;
	       T1 = LD(&(Rp[0]), ms, &(Rp[0]));
	       T2 = LD(&(Rm[0]), -ms, &(Rm[0]));
	       Tb = LDW(&(W[0]));
	       Tj = LD(&(Rp[WS(rs, 3)]), ms, &(Rp[WS(rs, 1)]));
	       Tk = LD(&(Rm[WS(rs, 3)]), -ms, &(Rm[WS(rs, 1)]));
	       Ti = LDW(&(W[TWVL * 12]));
	       Tr = LDW(&(W[TWVL * 10]));
	       T5 = LD(&(Rp[WS(rs, 2)]), ms, &(Rp[0]));
	       T6 = LD(&(Rm[WS(rs, 2)]), -ms, &(Rm[0]));
	       T3 = VFMACONJ(T2, T1);
	       Tc = VZMULIJ(Tb, VFNMSCONJ(T2, T1));
	       T4 = LDW(&(W[TWVL * 6]));
	       T9 = LDW(&(W[TWVL * 8]));
	       Tl = VZMULIJ(Ti, VFNMSCONJ(Tk, Tj));
	       Ts = VZMULJ(Tr, VFMACONJ(Tk, Tj));
	       Tf = LD(&(Rp[WS(rs, 1)]), ms, &(Rp[WS(rs, 1)]));
	       Tg = LD(&(Rm[WS(rs, 1)]), -ms, &(Rm[WS(rs, 1)]));
	       Te = LDW(&(W[TWVL * 4]));
	       Tp = LDW(&(W[TWVL * 2]));
	       T7 = VZMULJ(T4, VFMACONJ(T6, T5));
	       Ta = VZMULIJ(T9, VFNMSCONJ(T6, T5));
	       Th = VZMULIJ(Te, VFNMSCONJ(Tg, Tf));
	       Tq = VZMULJ(Tp, VFMACONJ(Tg, Tf));
	       TC = VADD(T3, T7);
	       T8 = VSUB(T3, T7);
	       Td = VSUB(Ta, Tc);
	       TF = VADD(Tc, Ta);
	       Tm = VSUB(Th, Tl);
	       TG = VADD(Th, Tl);
	       TD = VADD(Tq, Ts);
	       Tt = VSUB(Tq, Ts);
	       Tu = VSUB(Tm, Td);
	       Tn = VADD(Td, Tm);
	       TH = VSUB(TF, TG);
	       TL = VADD(TF, TG);
	       TE = VSUB(TC, TD);
	       TK = VADD(TC, TD);
	       Tz = VFMA(LDK(KP707106781), Tu, Tt);
	       Tv = VFNMS(LDK(KP707106781), Tu, Tt);
	       Ty = VFNMS(LDK(KP707106781), Tn, T8);
	       To = VFMA(LDK(KP707106781), Tn, T8);
	       TJ = VCONJ(VMUL(LDK(KP500000000), VFNMSI(TH, TE)));
	       TI = VMUL(LDK(KP500000000), VFMAI(TH, TE));
	       TN = VCONJ(VMUL(LDK(KP500000000), VADD(TL, TK)));
	       TM = VMUL(LDK(KP500000000), VSUB(TK, TL));
	       TB = VMUL(LDK(KP500000000), VFMAI(Tz, Ty));
	       TA = VCONJ(VMUL(LDK(KP500000000), VFNMSI(Tz, Ty)));
	       Tx = VCONJ(VMUL(LDK(KP500000000), VFMAI(Tv, To)));
	       Tw = VMUL(LDK(KP500000000), VFNMSI(Tv, To));
	       ST(&(Rm[WS(rs, 1)]), TJ, -ms, &(Rm[WS(rs, 1)]));
	       ST(&(Rp[WS(rs, 2)]), TI, ms, &(Rp[0]));
	       ST(&(Rm[WS(rs, 3)]), TN, -ms, &(Rm[WS(rs, 1)]));
	       ST(&(Rp[0]), TM, ms, &(Rp[0]));
	       ST(&(Rp[WS(rs, 3)]), TB, ms, &(Rp[WS(rs, 1)]));
	       ST(&(Rm[WS(rs, 2)]), TA, -ms, &(Rm[0]));
	       ST(&(Rm[0]), Tx, -ms, &(Rm[0]));
	       ST(&(Rp[WS(rs, 1)]), Tw, ms, &(Rp[WS(rs, 1)]));
	  }
     }
     VLEAVE();
}
예제 #7
0
static void hc2cfdftv_8(R *Rp, R *Ip, R *Rm, R *Im, const R *W, stride rs, INT mb, INT me, INT ms)
{
     DVK(KP707106781, +0.707106781186547524400844362104849039284835938);
     DVK(KP353553390, +0.353553390593273762200422181052424519642417969);
     DVK(KP500000000, +0.500000000000000000000000000000000000000000000);
     {
	  INT m;
	  for (m = mb, W = W + ((mb - 1) * ((TWVL / VL) * 14)); m < me; m = m + VL, Rp = Rp + (VL * ms), Ip = Ip + (VL * ms), Rm = Rm - (VL * ms), Im = Im - (VL * ms), W = W + (TWVL * 14), MAKE_VOLATILE_STRIDE(32, rs)) {
	       V Ta, TE, Tr, TF, Tl, TK, Tw, TG, T1, T6, T3, T8, T2, T7, T4;
	       V T9, T5, To, Tq, Tn, Tp, Tc, Th, Te, Tj, Td, Ti, Tf, Tk, Tb;
	       V Tg, Tt, Tv, Ts, Tu, Ty, Tz, Tm, Tx, TC, TD, TA, TB, TI, TO;
	       V TL, TP, TH, TJ, TM, TR, TN, TQ;
	       T1 = LD(&(Rp[0]), ms, &(Rp[0]));
	       T6 = LD(&(Rp[WS(rs, 2)]), ms, &(Rp[0]));
	       T2 = LD(&(Rm[0]), -ms, &(Rm[0]));
	       T3 = VCONJ(T2);
	       T7 = LD(&(Rm[WS(rs, 2)]), -ms, &(Rm[0]));
	       T8 = VCONJ(T7);
	       T4 = VADD(T1, T3);
	       T5 = LDW(&(W[TWVL * 6]));
	       T9 = VZMULJ(T5, VADD(T6, T8));
	       Ta = VADD(T4, T9);
	       TE = VMUL(LDK(KP500000000), VSUB(T4, T9));
	       Tn = LDW(&(W[0]));
	       To = VZMULIJ(Tn, VSUB(T3, T1));
	       Tp = LDW(&(W[TWVL * 8]));
	       Tq = VZMULIJ(Tp, VSUB(T8, T6));
	       Tr = VADD(To, Tq);
	       TF = VSUB(To, Tq);
	       Tc = LD(&(Rp[WS(rs, 1)]), ms, &(Rp[WS(rs, 1)]));
	       Th = LD(&(Rp[WS(rs, 3)]), ms, &(Rp[WS(rs, 1)]));
	       Td = LD(&(Rm[WS(rs, 1)]), -ms, &(Rm[WS(rs, 1)]));
	       Te = VCONJ(Td);
	       Ti = LD(&(Rm[WS(rs, 3)]), -ms, &(Rm[WS(rs, 1)]));
	       Tj = VCONJ(Ti);
	       Tb = LDW(&(W[TWVL * 2]));
	       Tf = VZMULJ(Tb, VADD(Tc, Te));
	       Tg = LDW(&(W[TWVL * 10]));
	       Tk = VZMULJ(Tg, VADD(Th, Tj));
	       Tl = VADD(Tf, Tk);
	       TK = VSUB(Tf, Tk);
	       Ts = LDW(&(W[TWVL * 4]));
	       Tt = VZMULIJ(Ts, VSUB(Te, Tc));
	       Tu = LDW(&(W[TWVL * 12]));
	       Tv = VZMULIJ(Tu, VSUB(Tj, Th));
	       Tw = VADD(Tt, Tv);
	       TG = VSUB(Tv, Tt);
	       Tm = VADD(Ta, Tl);
	       Tx = VADD(Tr, Tw);
	       Ty = VCONJ(VMUL(LDK(KP500000000), VSUB(Tm, Tx)));
	       Tz = VMUL(LDK(KP500000000), VADD(Tm, Tx));
	       ST(&(Rm[WS(rs, 3)]), Ty, -ms, &(Rm[WS(rs, 1)]));
	       ST(&(Rp[0]), Tz, ms, &(Rp[0]));
	       TA = VSUB(Ta, Tl);
	       TB = VBYI(VSUB(Tw, Tr));
	       TC = VCONJ(VMUL(LDK(KP500000000), VSUB(TA, TB)));
	       TD = VMUL(LDK(KP500000000), VADD(TA, TB));
	       ST(&(Rm[WS(rs, 1)]), TC, -ms, &(Rm[WS(rs, 1)]));
	       ST(&(Rp[WS(rs, 2)]), TD, ms, &(Rp[0]));
	       TH = VMUL(LDK(KP353553390), VADD(TF, TG));
	       TI = VADD(TE, TH);
	       TO = VSUB(TE, TH);
	       TJ = VMUL(LDK(KP707106781), VSUB(TG, TF));
	       TL = VMUL(LDK(KP500000000), VBYI(VSUB(TJ, TK)));
	       TP = VMUL(LDK(KP500000000), VBYI(VADD(TK, TJ)));
	       TM = VCONJ(VSUB(TI, TL));
	       ST(&(Rm[0]), TM, -ms, &(Rm[0]));
	       TR = VADD(TO, TP);
	       ST(&(Rp[WS(rs, 3)]), TR, ms, &(Rp[WS(rs, 1)]));
	       TN = VADD(TI, TL);
	       ST(&(Rp[WS(rs, 1)]), TN, ms, &(Rp[WS(rs, 1)]));
	       TQ = VCONJ(VSUB(TO, TP));
	       ST(&(Rm[WS(rs, 2)]), TQ, -ms, &(Rm[0]));
	  }
     }
     VLEAVE();
}
예제 #8
0
static void hc2cfdftv_12(R *Rp, R *Ip, R *Rm, R *Im, const R *W, stride rs, INT mb, INT me, INT ms)
{
     DVK(KP866025403, +0.866025403784438646763723170752936183471402627);
     DVK(KP500000000, +0.500000000000000000000000000000000000000000000);
     INT m;
     for (m = mb, W = W + ((mb - 1) * ((TWVL / VL) * 22)); m < me; m = m + VL, Rp = Rp + (VL * ms), Ip = Ip + (VL * ms), Rm = Rm - (VL * ms), Im = Im - (VL * ms), W = W + (TWVL * 22), MAKE_VOLATILE_STRIDE(rs)) {
	  V T3, T7, TH, TE, Th, TC, Tq, T11, TU, Tx, Tb, Tz, Tu, Tw, Tp;
	  V Tl, T9, Ta, T8, Ty, Tn, To, Tm, TG, T1, T2, Tt, T5, T6, T4;
	  V Tv, Tj, Tk, Ti, TD, Tf, Tg, Te, TB, TT, TF, TR, Tr;
	  T1 = LD(&(Rp[0]), ms, &(Rp[0]));
	  T2 = LD(&(Rm[0]), -ms, &(Rm[0]));
	  Tt = LDW(&(W[0]));
	  T5 = LD(&(Rp[WS(rs, 2)]), ms, &(Rp[0]));
	  T6 = LD(&(Rm[WS(rs, 2)]), -ms, &(Rm[0]));
	  T4 = LDW(&(W[TWVL * 6]));
	  Tv = LDW(&(W[TWVL * 8]));
	  Tn = LD(&(Rp[WS(rs, 1)]), ms, &(Rp[WS(rs, 1)]));
	  To = LD(&(Rm[WS(rs, 1)]), -ms, &(Rm[WS(rs, 1)]));
	  T3 = VFMACONJ(T2, T1);
	  Tu = VZMULIJ(Tt, VFNMSCONJ(T2, T1));
	  Tm = LDW(&(W[TWVL * 2]));
	  TG = LDW(&(W[TWVL * 4]));
	  T7 = VZMULJ(T4, VFMACONJ(T6, T5));
	  Tw = VZMULIJ(Tv, VFNMSCONJ(T6, T5));
	  Tj = LD(&(Rp[WS(rs, 5)]), ms, &(Rp[WS(rs, 1)]));
	  Tk = LD(&(Rm[WS(rs, 5)]), -ms, &(Rm[WS(rs, 1)]));
	  Ti = LDW(&(W[TWVL * 18]));
	  TD = LDW(&(W[TWVL * 20]));
	  Tp = VZMULJ(Tm, VFMACONJ(To, Tn));
	  TH = VZMULIJ(TG, VFNMSCONJ(To, Tn));
	  Tf = LD(&(Rp[WS(rs, 3)]), ms, &(Rp[WS(rs, 1)]));
	  Tg = LD(&(Rm[WS(rs, 3)]), -ms, &(Rm[WS(rs, 1)]));
	  Te = LDW(&(W[TWVL * 10]));
	  TB = LDW(&(W[TWVL * 12]));
	  Tl = VZMULJ(Ti, VFMACONJ(Tk, Tj));
	  TE = VZMULIJ(TD, VFNMSCONJ(Tk, Tj));
	  T9 = LD(&(Rp[WS(rs, 4)]), ms, &(Rp[0]));
	  Ta = LD(&(Rm[WS(rs, 4)]), -ms, &(Rm[0]));
	  T8 = LDW(&(W[TWVL * 14]));
	  Ty = LDW(&(W[TWVL * 16]));
	  Th = VZMULJ(Te, VFMACONJ(Tg, Tf));
	  TC = VZMULIJ(TB, VFNMSCONJ(Tg, Tf));
	  Tq = VADD(Tl, Tp);
	  T11 = VSUB(Tp, Tl);
	  TU = VSUB(Tu, Tw);
	  Tx = VADD(Tu, Tw);
	  Tb = VZMULJ(T8, VFMACONJ(Ta, T9));
	  Tz = VZMULIJ(Ty, VFNMSCONJ(Ta, T9));
	  TT = VSUB(TC, TE);
	  TF = VADD(TC, TE);
	  TR = VFNMS(LDK(KP500000000), Tq, Th);
	  Tr = VADD(Th, Tq);
	  {
	       V TX, TA, T1d, TV, TY, TI, T1e, T12, TQ, Td, T10, Tc, T1a, TN, TJ;
	       V T1j, T1f, T1b, TS, TM, Ts, T17, T13, TZ, T1i, T1c, T16, TW, TP, TO;
	       V TL, TK, T1k, T1l, T1h, T1g, T18, T19, T15, T14;
	       T10 = VSUB(Tb, T7);
	       Tc = VADD(T7, Tb);
	       TX = VFNMS(LDK(KP500000000), Tx, Tz);
	       TA = VADD(Tx, Tz);
	       T1d = VADD(TU, TT);
	       TV = VSUB(TT, TU);
	       TY = VFNMS(LDK(KP500000000), TF, TH);
	       TI = VADD(TF, TH);
	       T1e = VADD(T10, T11);
	       T12 = VSUB(T10, T11);
	       TQ = VFNMS(LDK(KP500000000), Tc, T3);
	       Td = VADD(T3, Tc);
	       T1a = VADD(TX, TY);
	       TZ = VSUB(TX, TY);
	       TN = VADD(TA, TI);
	       TJ = VSUB(TA, TI);
	       T1j = VMUL(LDK(KP866025403), VADD(T1d, T1e));
	       T1f = VMUL(LDK(KP866025403), VSUB(T1d, T1e));
	       T1b = VADD(TQ, TR);
	       TS = VSUB(TQ, TR);
	       TM = VADD(Td, Tr);
	       Ts = VSUB(Td, Tr);
	       T17 = VFMA(LDK(KP866025403), T12, TZ);
	       T13 = VFNMS(LDK(KP866025403), T12, TZ);
	       T1i = VSUB(T1b, T1a);
	       T1c = VADD(T1a, T1b);
	       T16 = VFNMS(LDK(KP866025403), TV, TS);
	       TW = VFMA(LDK(KP866025403), TV, TS);
	       TP = VCONJ(VMUL(LDK(KP500000000), VADD(TN, TM)));
	       TO = VMUL(LDK(KP500000000), VSUB(TM, TN));
	       TL = VCONJ(VMUL(LDK(KP500000000), VFNMSI(TJ, Ts)));
	       TK = VMUL(LDK(KP500000000), VFMAI(TJ, Ts));
	       T1k = VCONJ(VMUL(LDK(KP500000000), VFNMSI(T1j, T1i)));
	       T1l = VMUL(LDK(KP500000000), VFMAI(T1j, T1i));
	       T1h = VMUL(LDK(KP500000000), VFMAI(T1f, T1c));
	       T1g = VCONJ(VMUL(LDK(KP500000000), VFNMSI(T1f, T1c)));
	       T18 = VMUL(LDK(KP500000000), VFNMSI(T17, T16));
	       T19 = VCONJ(VMUL(LDK(KP500000000), VFMAI(T17, T16)));
	       T15 = VCONJ(VMUL(LDK(KP500000000), VFMAI(T13, TW)));
	       T14 = VMUL(LDK(KP500000000), VFNMSI(T13, TW));
	       ST(&(Rm[WS(rs, 5)]), TP, -ms, &(Rm[WS(rs, 1)]));
	       ST(&(Rp[0]), TO, ms, &(Rp[0]));
	       ST(&(Rm[WS(rs, 2)]), TL, -ms, &(Rm[0]));
	       ST(&(Rp[WS(rs, 3)]), TK, ms, &(Rp[WS(rs, 1)]));
	       ST(&(Rm[WS(rs, 3)]), T1k, -ms, &(Rm[WS(rs, 1)]));
	       ST(&(Rp[WS(rs, 4)]), T1l, ms, &(Rp[0]));
	       ST(&(Rp[WS(rs, 2)]), T1h, ms, &(Rp[0]));
	       ST(&(Rm[WS(rs, 1)]), T1g, -ms, &(Rm[WS(rs, 1)]));
	       ST(&(Rp[WS(rs, 5)]), T18, ms, &(Rp[WS(rs, 1)]));
	       ST(&(Rm[WS(rs, 4)]), T19, -ms, &(Rm[0]));
	       ST(&(Rm[0]), T15, -ms, &(Rm[0]));
	       ST(&(Rp[WS(rs, 1)]), T14, ms, &(Rp[WS(rs, 1)]));
	  }
     }
}
예제 #9
0
static void hc2cfdftv_12(R *Rp, R *Ip, R *Rm, R *Im, const R *W, stride rs, INT mb, INT me, INT ms)
{
     DVK(KP433012701, +0.433012701892219323381861585376468091735701313);
     DVK(KP866025403, +0.866025403784438646763723170752936183471402627);
     DVK(KP250000000, +0.250000000000000000000000000000000000000000000);
     DVK(KP500000000, +0.500000000000000000000000000000000000000000000);
     INT m;
     for (m = mb, W = W + ((mb - 1) * ((TWVL / VL) * 22)); m < me; m = m + VL, Rp = Rp + (VL * ms), Ip = Ip + (VL * ms), Rm = Rm - (VL * ms), Im = Im - (VL * ms), W = W + (TWVL * 22), MAKE_VOLATILE_STRIDE(rs)) {
	  V TX, T13, T4, Tf, TZ, TD, TF, T17, TW, T14, Tw, Tl, T10, TL, TN;
	  V T16;
	  {
	       V T1, T3, TA, Tb, Td, Te, T9, TC, T2, Tz, Tc, Ta, T6, T8, T7;
	       V T5, TB, TE, Ti, Tk, TI, Ts, Tu, Tv, Tq, TK, Tj, TH, Tt, Tr;
	       V Tn, Tp, To, Tm, TJ, Th, TM;
	       T1 = LD(&(Rp[0]), ms, &(Rp[0]));
	       T2 = LD(&(Rm[0]), -ms, &(Rm[0]));
	       T3 = VCONJ(T2);
	       Tz = LDW(&(W[0]));
	       TA = VZMULIJ(Tz, VSUB(T3, T1));
	       Tb = LD(&(Rp[WS(rs, 4)]), ms, &(Rp[0]));
	       Tc = LD(&(Rm[WS(rs, 4)]), -ms, &(Rm[0]));
	       Td = VCONJ(Tc);
	       Ta = LDW(&(W[TWVL * 14]));
	       Te = VZMULJ(Ta, VADD(Tb, Td));
	       T6 = LD(&(Rp[WS(rs, 2)]), ms, &(Rp[0]));
	       T7 = LD(&(Rm[WS(rs, 2)]), -ms, &(Rm[0]));
	       T8 = VCONJ(T7);
	       T5 = LDW(&(W[TWVL * 6]));
	       T9 = VZMULJ(T5, VADD(T6, T8));
	       TB = LDW(&(W[TWVL * 8]));
	       TC = VZMULIJ(TB, VSUB(T8, T6));
	       TX = VSUB(TC, TA);
	       T13 = VSUB(Te, T9);
	       T4 = VADD(T1, T3);
	       Tf = VADD(T9, Te);
	       TZ = VFNMS(LDK(KP250000000), Tf, VMUL(LDK(KP500000000), T4));
	       TD = VADD(TA, TC);
	       TE = LDW(&(W[TWVL * 16]));
	       TF = VZMULIJ(TE, VSUB(Td, Tb));
	       T17 = VFNMS(LDK(KP500000000), TD, TF);
	       Ti = LD(&(Rp[WS(rs, 3)]), ms, &(Rp[WS(rs, 1)]));
	       Tj = LD(&(Rm[WS(rs, 3)]), -ms, &(Rm[WS(rs, 1)]));
	       Tk = VCONJ(Tj);
	       TH = LDW(&(W[TWVL * 12]));
	       TI = VZMULIJ(TH, VSUB(Tk, Ti));
	       Ts = LD(&(Rp[WS(rs, 1)]), ms, &(Rp[WS(rs, 1)]));
	       Tt = LD(&(Rm[WS(rs, 1)]), -ms, &(Rm[WS(rs, 1)]));
	       Tu = VCONJ(Tt);
	       Tr = LDW(&(W[TWVL * 2]));
	       Tv = VZMULJ(Tr, VADD(Ts, Tu));
	       Tn = LD(&(Rp[WS(rs, 5)]), ms, &(Rp[WS(rs, 1)]));
	       To = LD(&(Rm[WS(rs, 5)]), -ms, &(Rm[WS(rs, 1)]));
	       Tp = VCONJ(To);
	       Tm = LDW(&(W[TWVL * 18]));
	       Tq = VZMULJ(Tm, VADD(Tn, Tp));
	       TJ = LDW(&(W[TWVL * 20]));
	       TK = VZMULIJ(TJ, VSUB(Tp, Tn));
	       TW = VSUB(TK, TI);
	       T14 = VSUB(Tv, Tq);
	       Tw = VADD(Tq, Tv);
	       Th = LDW(&(W[TWVL * 10]));
	       Tl = VZMULJ(Th, VADD(Ti, Tk));
	       T10 = VFNMS(LDK(KP250000000), Tw, VMUL(LDK(KP500000000), Tl));
	       TL = VADD(TI, TK);
	       TM = LDW(&(W[TWVL * 4]));
	       TN = VZMULIJ(TM, VSUB(Tu, Ts));
	       T16 = VFNMS(LDK(KP500000000), TL, TN);
	  }
	  {
	       V Ty, TS, TP, TT, Tg, Tx, TG, TO, TQ, TV, TR, TU, T1i, T1o, T1l;
	       V T1p, T1g, T1h, T1j, T1k, T1m, T1r, T1n, T1q, T12, T1c, T19, T1d, TY, T11;
	       V T15, T18, T1a, T1f, T1b, T1e;
	       Tg = VADD(T4, Tf);
	       Tx = VADD(Tl, Tw);
	       Ty = VADD(Tg, Tx);
	       TS = VSUB(Tg, Tx);
	       TG = VADD(TD, TF);
	       TO = VADD(TL, TN);
	       TP = VADD(TG, TO);
	       TT = VBYI(VSUB(TO, TG));
	       TQ = VCONJ(VMUL(LDK(KP500000000), VSUB(Ty, TP)));
	       ST(&(Rm[WS(rs, 5)]), TQ, -ms, &(Rm[WS(rs, 1)]));
	       TV = VMUL(LDK(KP500000000), VADD(TS, TT));
	       ST(&(Rp[WS(rs, 3)]), TV, ms, &(Rp[WS(rs, 1)]));
	       TR = VMUL(LDK(KP500000000), VADD(Ty, TP));
	       ST(&(Rp[0]), TR, ms, &(Rp[0]));
	       TU = VCONJ(VMUL(LDK(KP500000000), VSUB(TS, TT)));
	       ST(&(Rm[WS(rs, 2)]), TU, -ms, &(Rm[0]));
	       T1g = VADD(TX, TW);
	       T1h = VADD(T13, T14);
	       T1i = VMUL(LDK(KP500000000), VBYI(VMUL(LDK(KP866025403), VSUB(T1g, T1h))));
	       T1o = VMUL(LDK(KP500000000), VBYI(VMUL(LDK(KP866025403), VADD(T1g, T1h))));
	       T1j = VADD(TZ, T10);
	       T1k = VMUL(LDK(KP500000000), VADD(T17, T16));
	       T1l = VSUB(T1j, T1k);
	       T1p = VADD(T1j, T1k);
	       T1m = VADD(T1i, T1l);
	       ST(&(Rp[WS(rs, 2)]), T1m, ms, &(Rp[0]));
	       T1r = VCONJ(VSUB(T1p, T1o));
	       ST(&(Rm[WS(rs, 3)]), T1r, -ms, &(Rm[WS(rs, 1)]));
	       T1n = VCONJ(VSUB(T1l, T1i));
	       ST(&(Rm[WS(rs, 1)]), T1n, -ms, &(Rm[WS(rs, 1)]));
	       T1q = VADD(T1o, T1p);
	       ST(&(Rp[WS(rs, 4)]), T1q, ms, &(Rp[0]));
	       TY = VMUL(LDK(KP433012701), VSUB(TW, TX));
	       T11 = VSUB(TZ, T10);
	       T12 = VADD(TY, T11);
	       T1c = VSUB(T11, TY);
	       T15 = VMUL(LDK(KP866025403), VSUB(T13, T14));
	       T18 = VSUB(T16, T17);
	       T19 = VMUL(LDK(KP500000000), VBYI(VSUB(T15, T18)));
	       T1d = VMUL(LDK(KP500000000), VBYI(VADD(T15, T18)));
	       T1a = VCONJ(VSUB(T12, T19));
	       ST(&(Rm[0]), T1a, -ms, &(Rm[0]));
	       T1f = VCONJ(VADD(T1c, T1d));
	       ST(&(Rm[WS(rs, 4)]), T1f, -ms, &(Rm[0]));
	       T1b = VADD(T12, T19);
	       ST(&(Rp[WS(rs, 1)]), T1b, ms, &(Rp[WS(rs, 1)]));
	       T1e = VSUB(T1c, T1d);
	       ST(&(Rp[WS(rs, 5)]), T1e, ms, &(Rp[WS(rs, 1)]));
	  }
     }
}
예제 #10
0
static void hc2cfdftv_10(R *Rp, R *Ip, R *Rm, R *Im, const R *W, stride rs, INT mb, INT me, INT ms)
{
     DVK(KP559016994, +0.559016994374947424102293417182819058860154590);
     DVK(KP500000000, +0.500000000000000000000000000000000000000000000);
     DVK(KP250000000, +0.250000000000000000000000000000000000000000000);
     DVK(KP618033988, +0.618033988749894848204586834365638117720309180);
     DVK(KP951056516, +0.951056516295153572116439333379382143405698634);
     {
	  INT m;
	  for (m = mb, W = W + ((mb - 1) * ((TWVL / VL) * 18)); m < me; m = m + VL, Rp = Rp + (VL * ms), Ip = Ip + (VL * ms), Rm = Rm - (VL * ms), Im = Im - (VL * ms), W = W + (TWVL * 18), MAKE_VOLATILE_STRIDE(40, rs)) {
	       V T5, T6, Tw, Tr, Tc, Tj, Tl, Tm, Tk, Ts, Tg, Ty, T3, T4, T1;
	       V T2, Tv, Tq, Ta, Tb, T9, Ti, Te, Tf, Td, Tx, Tn, Tt, Th, TQ;
	       V TT, Tz, T7, TR, To, Tu, TU;
	       T1 = LD(&(Rp[0]), ms, &(Rp[0]));
	       T2 = LD(&(Rm[0]), -ms, &(Rm[0]));
	       Tv = LDW(&(W[0]));
	       T5 = LD(&(Rp[WS(rs, 2)]), ms, &(Rp[0]));
	       T6 = LD(&(Rm[WS(rs, 2)]), -ms, &(Rm[0]));
	       Tq = LDW(&(W[TWVL * 6]));
	       Ta = LD(&(Rp[WS(rs, 1)]), ms, &(Rp[WS(rs, 1)]));
	       Tb = LD(&(Rm[WS(rs, 1)]), -ms, &(Rm[WS(rs, 1)]));
	       T9 = LDW(&(W[TWVL * 2]));
	       Ti = LDW(&(W[TWVL * 4]));
	       Tw = VZMULIJ(Tv, VFNMSCONJ(T2, T1));
	       Te = LD(&(Rp[WS(rs, 3)]), ms, &(Rp[WS(rs, 1)]));
	       Tf = LD(&(Rm[WS(rs, 3)]), -ms, &(Rm[WS(rs, 1)]));
	       Tr = VZMULJ(Tq, VFMACONJ(T6, T5));
	       Td = LDW(&(W[TWVL * 12]));
	       Tx = LDW(&(W[TWVL * 10]));
	       Tc = VZMULJ(T9, VFMACONJ(Tb, Ta));
	       Tj = VZMULIJ(Ti, VFNMSCONJ(Tb, Ta));
	       Tl = LD(&(Rp[WS(rs, 4)]), ms, &(Rp[0]));
	       Tm = LD(&(Rm[WS(rs, 4)]), -ms, &(Rm[0]));
	       Tk = LDW(&(W[TWVL * 14]));
	       Ts = LDW(&(W[TWVL * 16]));
	       Tg = VZMULIJ(Td, VFNMSCONJ(Tf, Te));
	       Ty = VZMULJ(Tx, VFMACONJ(Tf, Te));
	       T3 = VFMACONJ(T2, T1);
	       T4 = LDW(&(W[TWVL * 8]));
	       Tn = VZMULJ(Tk, VFMACONJ(Tm, Tl));
	       Tt = VZMULIJ(Ts, VFNMSCONJ(Tm, Tl));
	       Th = VSUB(Tc, Tg);
	       TQ = VADD(Tc, Tg);
	       TT = VADD(Tw, Ty);
	       Tz = VSUB(Tw, Ty);
	       T7 = VZMULIJ(T4, VFNMSCONJ(T6, T5));
	       TR = VADD(Tj, Tn);
	       To = VSUB(Tj, Tn);
	       Tu = VSUB(Tr, Tt);
	       TU = VADD(Tr, Tt);
	       {
		    V TP, T8, TS, T11, Tp, TH, TA, TG, TV, T12, TE, TB, TM, TI, TZ;
		    V TW, T17, T13, TD, TC, TY, TX, TL, TF, T10, T16, TN, TO, TK, TJ;
		    V T18, T19, T15, T14;
		    TP = VADD(T3, T7);
		    T8 = VSUB(T3, T7);
		    TS = VADD(TQ, TR);
		    T11 = VSUB(TQ, TR);
		    Tp = VSUB(Th, To);
		    TH = VADD(Th, To);
		    TA = VSUB(Tu, Tz);
		    TG = VADD(Tz, Tu);
		    TV = VADD(TT, TU);
		    T12 = VSUB(TU, TT);
		    TE = VSUB(Tp, TA);
		    TB = VADD(Tp, TA);
		    TM = VMUL(LDK(KP951056516), VFMA(LDK(KP618033988), TG, TH));
		    TI = VMUL(LDK(KP951056516), VFNMS(LDK(KP618033988), TH, TG));
		    TZ = VSUB(TS, TV);
		    TW = VADD(TS, TV);
		    T17 = VMUL(LDK(KP951056516), VFNMS(LDK(KP618033988), T11, T12));
		    T13 = VMUL(LDK(KP951056516), VFMA(LDK(KP618033988), T12, T11));
		    TD = VFNMS(LDK(KP250000000), TB, T8);
		    TC = VMUL(LDK(KP500000000), VADD(T8, TB));
		    TY = VFNMS(LDK(KP250000000), TW, TP);
		    TX = VCONJ(VMUL(LDK(KP500000000), VADD(TP, TW)));
		    TL = VFMA(LDK(KP559016994), TE, TD);
		    TF = VFNMS(LDK(KP559016994), TE, TD);
		    ST(&(Rp[0]), TC, ms, &(Rp[0]));
		    T10 = VFMA(LDK(KP559016994), TZ, TY);
		    T16 = VFNMS(LDK(KP559016994), TZ, TY);
		    ST(&(Rm[WS(rs, 4)]), TX, -ms, &(Rm[0]));
		    TN = VCONJ(VMUL(LDK(KP500000000), VFNMSI(TM, TL)));
		    TO = VMUL(LDK(KP500000000), VFMAI(TM, TL));
		    TK = VMUL(LDK(KP500000000), VFMAI(TI, TF));
		    TJ = VCONJ(VMUL(LDK(KP500000000), VFNMSI(TI, TF)));
		    T18 = VMUL(LDK(KP500000000), VFNMSI(T17, T16));
		    T19 = VCONJ(VMUL(LDK(KP500000000), VFMAI(T17, T16)));
		    T15 = VCONJ(VMUL(LDK(KP500000000), VFMAI(T13, T10)));
		    T14 = VMUL(LDK(KP500000000), VFNMSI(T13, T10));
		    ST(&(Rm[WS(rs, 3)]), TN, -ms, &(Rm[WS(rs, 1)]));
		    ST(&(Rp[WS(rs, 4)]), TO, ms, &(Rp[0]));
		    ST(&(Rp[WS(rs, 2)]), TK, ms, &(Rp[0]));
		    ST(&(Rm[WS(rs, 1)]), TJ, -ms, &(Rm[WS(rs, 1)]));
		    ST(&(Rp[WS(rs, 3)]), T18, ms, &(Rp[WS(rs, 1)]));
		    ST(&(Rm[WS(rs, 2)]), T19, -ms, &(Rm[0]));
		    ST(&(Rm[0]), T15, -ms, &(Rm[0]));
		    ST(&(Rp[WS(rs, 1)]), T14, ms, &(Rp[WS(rs, 1)]));
	       }
	  }
     }
     VLEAVE();
}
예제 #11
0
static void hc2cfdftv_10(R *Rp, R *Ip, R *Rm, R *Im, const R *W, stride rs, INT mb, INT me, INT ms)
{
     DVK(KP125000000, +0.125000000000000000000000000000000000000000000);
     DVK(KP279508497, +0.279508497187473712051146708591409529430077295);
     DVK(KP587785252, +0.587785252292473129168705954639072768597652438);
     DVK(KP951056516, +0.951056516295153572116439333379382143405698634);
     DVK(KP500000000, +0.500000000000000000000000000000000000000000000);
     {
	  INT m;
	  for (m = mb, W = W + ((mb - 1) * ((TWVL / VL) * 18)); m < me; m = m + VL, Rp = Rp + (VL * ms), Ip = Ip + (VL * ms), Rm = Rm - (VL * ms), Im = Im - (VL * ms), W = W + (TWVL * 18), MAKE_VOLATILE_STRIDE(40, rs)) {
	       V Tl, Tt, Tu, TY, TZ, T10, Tz, TE, TF, TV, TW, TX, Ta, TU, TN;
	       V TR, TH, TQ, TK, TL, TM, TI, TG, TJ, TT, TO, TP, TS, T18, T1c;
	       V T12, T1b, T15, T16, T17, T14, T11, T13, T1e, T19, T1a, T1d;
	       {
		    V T1, T3, Ty, T8, T7, TB, Tf, Ts, Tk, Tw, Tq, TD, T2, Tx, T6;
		    V TA, Tc, Te, Td, Tb, Tr, Tj, Ti, Th, Tg, Tv, Tn, Tp, To, Tm;
		    V TC, T4, T9, T5;
		    T1 = LD(&(Rp[0]), ms, &(Rp[0]));
		    T2 = LD(&(Rm[0]), -ms, &(Rm[0]));
		    T3 = VCONJ(T2);
		    Tx = LDW(&(W[0]));
		    Ty = VZMULIJ(Tx, VSUB(T3, T1));
		    T8 = LD(&(Rp[WS(rs, 2)]), ms, &(Rp[0]));
		    T6 = LD(&(Rm[WS(rs, 2)]), -ms, &(Rm[0]));
		    T7 = VCONJ(T6);
		    TA = LDW(&(W[TWVL * 6]));
		    TB = VZMULJ(TA, VADD(T7, T8));
		    Tc = LD(&(Rp[WS(rs, 1)]), ms, &(Rp[WS(rs, 1)]));
		    Td = LD(&(Rm[WS(rs, 1)]), -ms, &(Rm[WS(rs, 1)]));
		    Te = VCONJ(Td);
		    Tb = LDW(&(W[TWVL * 2]));
		    Tf = VZMULJ(Tb, VADD(Tc, Te));
		    Tr = LDW(&(W[TWVL * 4]));
		    Ts = VZMULIJ(Tr, VSUB(Te, Tc));
		    Tj = LD(&(Rp[WS(rs, 3)]), ms, &(Rp[WS(rs, 1)]));
		    Th = LD(&(Rm[WS(rs, 3)]), -ms, &(Rm[WS(rs, 1)]));
		    Ti = VCONJ(Th);
		    Tg = LDW(&(W[TWVL * 12]));
		    Tk = VZMULIJ(Tg, VSUB(Ti, Tj));
		    Tv = LDW(&(W[TWVL * 10]));
		    Tw = VZMULJ(Tv, VADD(Ti, Tj));
		    Tn = LD(&(Rp[WS(rs, 4)]), ms, &(Rp[0]));
		    To = LD(&(Rm[WS(rs, 4)]), -ms, &(Rm[0]));
		    Tp = VCONJ(To);
		    Tm = LDW(&(W[TWVL * 14]));
		    Tq = VZMULJ(Tm, VADD(Tn, Tp));
		    TC = LDW(&(W[TWVL * 16]));
		    TD = VZMULIJ(TC, VSUB(Tp, Tn));
		    Tl = VSUB(Tf, Tk);
		    Tt = VSUB(Tq, Ts);
		    Tu = VADD(Tl, Tt);
		    TY = VADD(Ty, Tw);
		    TZ = VADD(TB, TD);
		    T10 = VADD(TY, TZ);
		    Tz = VSUB(Tw, Ty);
		    TE = VSUB(TB, TD);
		    TF = VADD(Tz, TE);
		    TV = VADD(Tf, Tk);
		    TW = VADD(Ts, Tq);
		    TX = VADD(TV, TW);
		    T4 = VADD(T1, T3);
		    T5 = LDW(&(W[TWVL * 8]));
		    T9 = VZMULIJ(T5, VSUB(T7, T8));
		    Ta = VSUB(T4, T9);
		    TU = VADD(T4, T9);
	       }
	       TL = VSUB(Tl, Tt);
	       TM = VSUB(TE, Tz);
	       TN = VMUL(LDK(KP500000000), VBYI(VFMA(LDK(KP951056516), TL, VMUL(LDK(KP587785252), TM))));
	       TR = VMUL(LDK(KP500000000), VBYI(VFNMS(LDK(KP587785252), TL, VMUL(LDK(KP951056516), TM))));
	       TI = VMUL(LDK(KP279508497), VSUB(Tu, TF));
	       TG = VADD(Tu, TF);
	       TJ = VFNMS(LDK(KP125000000), TG, VMUL(LDK(KP500000000), Ta));
	       TH = VCONJ(VMUL(LDK(KP500000000), VADD(Ta, TG)));
	       TQ = VSUB(TJ, TI);
	       TK = VADD(TI, TJ);
	       ST(&(Rm[WS(rs, 4)]), TH, -ms, &(Rm[0]));
	       TT = VCONJ(VADD(TQ, TR));
	       ST(&(Rm[WS(rs, 2)]), TT, -ms, &(Rm[0]));
	       TO = VSUB(TK, TN);
	       ST(&(Rp[WS(rs, 1)]), TO, ms, &(Rp[WS(rs, 1)]));
	       TP = VCONJ(VADD(TK, TN));
	       ST(&(Rm[0]), TP, -ms, &(Rm[0]));
	       TS = VSUB(TQ, TR);
	       ST(&(Rp[WS(rs, 3)]), TS, ms, &(Rp[WS(rs, 1)]));
	       T16 = VSUB(TZ, TY);
	       T17 = VSUB(TV, TW);
	       T18 = VMUL(LDK(KP500000000), VBYI(VFNMS(LDK(KP587785252), T17, VMUL(LDK(KP951056516), T16))));
	       T1c = VMUL(LDK(KP500000000), VBYI(VFMA(LDK(KP951056516), T17, VMUL(LDK(KP587785252), T16))));
	       T14 = VMUL(LDK(KP279508497), VSUB(TX, T10));
	       T11 = VADD(TX, T10);
	       T13 = VFNMS(LDK(KP125000000), T11, VMUL(LDK(KP500000000), TU));
	       T12 = VMUL(LDK(KP500000000), VADD(TU, T11));
	       T1b = VADD(T14, T13);
	       T15 = VSUB(T13, T14);
	       ST(&(Rp[0]), T12, ms, &(Rp[0]));
	       T1e = VADD(T1b, T1c);
	       ST(&(Rp[WS(rs, 4)]), T1e, ms, &(Rp[0]));
	       T19 = VCONJ(VSUB(T15, T18));
	       ST(&(Rm[WS(rs, 1)]), T19, -ms, &(Rm[WS(rs, 1)]));
	       T1a = VADD(T15, T18);
	       ST(&(Rp[WS(rs, 2)]), T1a, ms, &(Rp[0]));
	       T1d = VCONJ(VSUB(T1b, T1c));
	       ST(&(Rm[WS(rs, 3)]), T1d, -ms, &(Rm[WS(rs, 1)]));
	  }
     }
     VLEAVE();
}