static void hc2cbdftv_4(R *Rp, R *Ip, R *Rm, R *Im, const R *W, stride rs, INT mb, INT me, INT ms)
{
     {
	  INT m;
	  for (m = mb, W = W + ((mb - 1) * ((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 T2, T3, T5, T6, Tf, T1, T9, Ta, T4, Tb, T7, Tc, Th, T8, Tg;
	       V Te, Td, Ti, Tj;
	       T2 = LD(&(Rp[0]), ms, &(Rp[0]));
	       T3 = LD(&(Rm[WS(rs, 1)]), -ms, &(Rm[WS(rs, 1)]));
	       T5 = LD(&(Rp[WS(rs, 1)]), ms, &(Rp[WS(rs, 1)]));
	       T6 = LD(&(Rm[0]), -ms, &(Rm[0]));
	       Tf = LDW(&(W[0]));
	       T1 = LDW(&(W[TWVL * 4]));
	       T9 = LDW(&(W[TWVL * 2]));
	       Ta = VFMACONJ(T3, T2);
	       T4 = VFNMSCONJ(T3, T2);
	       Tb = VFMACONJ(T6, T5);
	       T7 = VFNMSCONJ(T6, T5);
	       Tc = VZMUL(T9, VSUB(Ta, Tb));
	       Th = VADD(Ta, Tb);
	       T8 = VZMULI(T1, VFNMSI(T7, T4));
	       Tg = VZMULI(Tf, VFMAI(T7, T4));
	       Te = VCONJ(VSUB(Tc, T8));
	       Td = VADD(T8, Tc);
	       Ti = VADD(Tg, Th);
	       Tj = VCONJ(VSUB(Th, Tg));
	       ST(&(Rm[WS(rs, 1)]), Te, -ms, &(Rm[WS(rs, 1)]));
	       ST(&(Rp[WS(rs, 1)]), Td, ms, &(Rp[WS(rs, 1)]));
	       ST(&(Rp[0]), Ti, ms, &(Rp[0]));
	       ST(&(Rm[0]), Tj, -ms, &(Rm[0]));
	  }
     }
     VLEAVE();
}
Пример #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(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();
}
Пример #3
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]));
     }
}
Пример #4
0
static void hc2cbdftv_4(R *Rp, R *Ip, R *Rm, R *Im, const R *W, stride rs, INT mb, INT me, INT ms)
{
     INT m;
     for (m = mb, W = W + ((mb - 1) * ((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 T5, Tc, T9, Td, T2, T4, T3, T6, T8, T7, Tj, Ti, Th, Tk, Tl;
	  V Ta, Te, T1, Tb, Tf, Tg;
	  T2 = LD(&(Rp[0]), ms, &(Rp[0]));
	  T3 = LD(&(Rm[WS(rs, 1)]), -ms, &(Rm[WS(rs, 1)]));
	  T4 = VCONJ(T3);
	  T5 = VSUB(T2, T4);
	  Tc = VADD(T2, T4);
	  T6 = LD(&(Rp[WS(rs, 1)]), ms, &(Rp[WS(rs, 1)]));
	  T7 = LD(&(Rm[0]), -ms, &(Rm[0]));
	  T8 = VCONJ(T7);
	  T9 = VBYI(VSUB(T6, T8));
	  Td = VADD(T6, T8);
	  Tj = VADD(Tc, Td);
	  Th = LDW(&(W[0]));
	  Ti = VZMULI(Th, VADD(T5, T9));
	  Tk = VADD(Ti, Tj);
	  ST(&(Rp[0]), Tk, ms, &(Rp[0]));
	  Tl = VCONJ(VSUB(Tj, Ti));
	  ST(&(Rm[0]), Tl, -ms, &(Rm[0]));
	  T1 = LDW(&(W[TWVL * 4]));
	  Ta = VZMULI(T1, VSUB(T5, T9));
	  Tb = LDW(&(W[TWVL * 2]));
	  Te = VZMUL(Tb, VSUB(Tc, Td));
	  Tf = VADD(Ta, Te);
	  ST(&(Rp[WS(rs, 1)]), Tf, ms, &(Rp[WS(rs, 1)]));
	  Tg = VCONJ(VSUB(Te, Ta));
	  ST(&(Rm[WS(rs, 1)]), Tg, -ms, &(Rm[WS(rs, 1)]));
     }
}
Пример #5
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();
}
Пример #6
0
static void hc2cbdftv_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(rs)) {
	  V Tv, Tn, Tr, Te, T4, Tg, Ta, Tf, T7, T1, Td, T2, T3, T8, T9;
	  V T5, T6, Th, Tj, Tb, Tp, Tx, Ti, Tc, To, Tk, Ts, Tq, Tw, Tm;
	  V Tl, Tu, Tt, Tz, Ty;
	  T2 = LD(&(Rp[0]), ms, &(Rp[0]));
	  T3 = LD(&(Rm[WS(rs, 2)]), -ms, &(Rm[0]));
	  T8 = LD(&(Rp[WS(rs, 1)]), ms, &(Rp[WS(rs, 1)]));
	  T9 = LD(&(Rm[WS(rs, 1)]), -ms, &(Rm[WS(rs, 1)]));
	  T5 = LD(&(Rp[WS(rs, 2)]), ms, &(Rp[0]));
	  T6 = LD(&(Rm[0]), -ms, &(Rm[0]));
	  Tv = LDW(&(W[0]));
	  Tn = LDW(&(W[TWVL * 8]));
	  Tr = LDW(&(W[TWVL * 6]));
	  Te = VFMACONJ(T3, T2);
	  T4 = VFNMSCONJ(T3, T2);
	  Tg = VFMACONJ(T9, T8);
	  Ta = VFMSCONJ(T9, T8);
	  Tf = VFMACONJ(T6, T5);
	  T7 = VFNMSCONJ(T6, T5);
	  T1 = LDW(&(W[TWVL * 4]));
	  Td = LDW(&(W[TWVL * 2]));
	  Th = VADD(Tf, Tg);
	  Tj = VMUL(LDK(KP866025403), VSUB(Tf, Tg));
	  Tb = VADD(T7, Ta);
	  Tp = VMUL(LDK(KP866025403), VSUB(T7, Ta));
	  Tx = VADD(Te, Th);
	  Ti = VFNMS(LDK(KP500000000), Th, Te);
	  Tc = VZMULI(T1, VADD(T4, Tb));
	  To = VFNMS(LDK(KP500000000), Tb, T4);
	  Tk = VZMUL(Td, VFNMSI(Tj, Ti));
	  Ts = VZMUL(Tr, VFMAI(Tj, Ti));
	  Tq = VZMULI(Tn, VFNMSI(Tp, To));
	  Tw = VZMULI(Tv, VFMAI(Tp, To));
	  Tm = VCONJ(VSUB(Tk, Tc));
	  Tl = VADD(Tc, Tk);
	  Tu = VCONJ(VSUB(Ts, Tq));
	  Tt = VADD(Tq, Ts);
	  Tz = VCONJ(VSUB(Tx, Tw));
	  Ty = VADD(Tw, Tx);
	  ST(&(Rm[WS(rs, 1)]), Tm, -ms, &(Rm[WS(rs, 1)]));
	  ST(&(Rp[WS(rs, 1)]), Tl, ms, &(Rp[WS(rs, 1)]));
	  ST(&(Rm[WS(rs, 2)]), Tu, -ms, &(Rm[0]));
	  ST(&(Rp[WS(rs, 2)]), Tt, ms, &(Rp[0]));
	  ST(&(Rm[0]), Tz, -ms, &(Rm[0]));
	  ST(&(Rp[0]), Ty, ms, &(Rp[0]));
     }
}
Пример #7
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 T4, T6, T1, T3, T2, T5, T7, T8;
	  T1 = LD(&(Rp[0]), ms, &(Rp[0]));
	  T2 = LD(&(Rm[0]), -ms, &(Rm[0]));
	  T3 = VCONJ(T2);
	  T4 = VADD(T1, T3);
	  T5 = LDW(&(W[0]));
	  T6 = VZMULIJ(T5, VSUB(T3, T1));
	  T7 = VCONJ(VMUL(LDK(KP500000000), VSUB(T4, T6)));
	  ST(&(Rm[0]), T7, -ms, &(Rm[0]));
	  T8 = VMUL(LDK(KP500000000), VADD(T4, T6));
	  ST(&(Rp[0]), T8, ms, &(Rp[0]));
     }
}
Пример #8
0
static void hc2cbdftv_2(R *Rp, R *Ip, R *Rm, R *Im, const R *W, stride rs, INT mb, INT me, INT ms)
{
     {
	  INT m;
	  for (m = mb, W = W + ((mb - 1) * ((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(8, rs)) {
	       V T6, T5, T2, T4, T3, T1, T7, T8;
	       T2 = LD(&(Rp[0]), ms, &(Rp[0]));
	       T3 = LD(&(Rm[0]), -ms, &(Rm[0]));
	       T4 = VCONJ(T3);
	       T6 = VADD(T2, T4);
	       T1 = LDW(&(W[0]));
	       T5 = VZMULI(T1, VSUB(T2, T4));
	       T7 = VADD(T5, T6);
	       ST(&(Rp[0]), T7, ms, &(Rp[0]));
	       T8 = VCONJ(VSUB(T6, T5));
	       ST(&(Rm[0]), T8, -ms, &(Rm[0]));
	  }
     }
     VLEAVE();
}
Пример #9
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();
}
Пример #10
0
static void hc2cbdftv_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(rs)) {
	  V T5, Th, Te, Ts, Tk, Tm, T2, T4, T3, T6, Tc, T8, Tb, T7, Ta;
	  V T9, Td, Ti, Tj, TA, Tf, Tn, Tv, Tt, Tz, T1, Tl, Tg, Tu, Tr;
	  V Tq, Ty, To, Tp, TC, TB, Tx, Tw;
	  T2 = LD(&(Rp[0]), ms, &(Rp[0]));
	  T3 = LD(&(Rm[WS(rs, 2)]), -ms, &(Rm[0]));
	  T4 = VCONJ(T3);
	  T5 = VSUB(T2, T4);
	  Th = VADD(T2, T4);
	  T6 = LD(&(Rp[WS(rs, 2)]), ms, &(Rp[0]));
	  Tc = LD(&(Rp[WS(rs, 1)]), ms, &(Rp[WS(rs, 1)]));
	  T7 = LD(&(Rm[0]), -ms, &(Rm[0]));
	  T8 = VCONJ(T7);
	  Ta = LD(&(Rm[WS(rs, 1)]), -ms, &(Rm[WS(rs, 1)]));
	  Tb = VCONJ(Ta);
	  T9 = VSUB(T6, T8);
	  Td = VSUB(Tb, Tc);
	  Te = VADD(T9, Td);
	  Ts = VBYI(VMUL(LDK(KP866025403), VSUB(T9, Td)));
	  Ti = VADD(T6, T8);
	  Tj = VADD(Tb, Tc);
	  Tk = VADD(Ti, Tj);
	  Tm = VBYI(VMUL(LDK(KP866025403), VSUB(Ti, Tj)));
	  TA = VADD(Th, Tk);
	  T1 = LDW(&(W[TWVL * 4]));
	  Tf = VZMULI(T1, VADD(T5, Te));
	  Tl = VFNMS(LDK(KP500000000), Tk, Th);
	  Tg = LDW(&(W[TWVL * 2]));
	  Tn = VZMUL(Tg, VSUB(Tl, Tm));
	  Tu = LDW(&(W[TWVL * 6]));
	  Tv = VZMUL(Tu, VADD(Tm, Tl));
	  Tr = VFNMS(LDK(KP500000000), Te, T5);
	  Tq = LDW(&(W[TWVL * 8]));
	  Tt = VZMULI(Tq, VSUB(Tr, Ts));
	  Ty = LDW(&(W[0]));
	  Tz = VZMULI(Ty, VADD(Ts, Tr));
	  To = VADD(Tf, Tn);
	  ST(&(Rp[WS(rs, 1)]), To, ms, &(Rp[WS(rs, 1)]));
	  Tp = VCONJ(VSUB(Tn, Tf));
	  ST(&(Rm[WS(rs, 1)]), Tp, -ms, &(Rm[WS(rs, 1)]));
	  TC = VCONJ(VSUB(TA, Tz));
	  ST(&(Rm[0]), TC, -ms, &(Rm[0]));
	  TB = VADD(Tz, TA);
	  ST(&(Rp[0]), TB, ms, &(Rp[0]));
	  Tx = VCONJ(VSUB(Tv, Tt));
	  ST(&(Rm[WS(rs, 2)]), Tx, -ms, &(Rm[0]));
	  Tw = VADD(Tt, Tv);
	  ST(&(Rp[WS(rs, 2)]), Tw, ms, &(Rp[0]));
     }
}
Пример #11
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();
}
static void hc2cbdftv_8(R *Rp, R *Ip, R *Rm, R *Im, const R *W, stride rs, INT mb, INT me, INT ms)
{
     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(rs)) {
	  V T5, Tj, Tq, TI, Te, Tk, Tt, TJ, T2, Tg, T4, Ti, T3, Th, To;
	  V Tp, T6, Tc, T8, Tb, T7, Ta, T9, Td, Tr, Ts, TP, Tu, Tm, TO;
	  V Tn, Tf, Tl, T1, TN, Tv, TR, Tw, TQ, TC, TK, TA, TG, TB, TH;
	  V Ty, Tz, Tx, TF, TD, TM, TE, TL;
	  T2 = LD(&(Rp[0]), ms, &(Rp[0]));
	  Tg = LD(&(Rp[WS(rs, 2)]), ms, &(Rp[0]));
	  T3 = LD(&(Rm[WS(rs, 3)]), -ms, &(Rm[WS(rs, 1)]));
	  T4 = VCONJ(T3);
	  Th = LD(&(Rm[WS(rs, 1)]), -ms, &(Rm[WS(rs, 1)]));
	  Ti = VCONJ(Th);
	  T5 = VSUB(T2, T4);
	  Tj = VSUB(Tg, Ti);
	  To = VADD(T2, T4);
	  Tp = VADD(Tg, Ti);
	  Tq = VSUB(To, Tp);
	  TI = VADD(To, Tp);
	  T6 = LD(&(Rp[WS(rs, 1)]), ms, &(Rp[WS(rs, 1)]));
	  Tc = LD(&(Rp[WS(rs, 3)]), ms, &(Rp[WS(rs, 1)]));
	  T7 = LD(&(Rm[WS(rs, 2)]), -ms, &(Rm[0]));
	  T8 = VCONJ(T7);
	  Ta = LD(&(Rm[0]), -ms, &(Rm[0]));
	  Tb = VCONJ(Ta);
	  T9 = VSUB(T6, T8);
	  Td = VSUB(Tb, Tc);
	  Te = VMUL(LDK(KP707106781), VADD(T9, Td));
	  Tk = VMUL(LDK(KP707106781), VSUB(T9, Td));
	  Tr = VADD(T6, T8);
	  Ts = VADD(Tb, Tc);
	  Tt = VBYI(VSUB(Tr, Ts));
	  TJ = VADD(Tr, Ts);
	  TP = VADD(TI, TJ);
	  Tn = LDW(&(W[TWVL * 10]));
	  Tu = VZMUL(Tn, VSUB(Tq, Tt));
	  Tf = VADD(T5, Te);
	  Tl = VBYI(VADD(Tj, Tk));
	  T1 = LDW(&(W[TWVL * 12]));
	  Tm = VZMULI(T1, VSUB(Tf, Tl));
	  TN = LDW(&(W[0]));
	  TO = VZMULI(TN, VADD(Tl, Tf));
	  Tv = VADD(Tm, Tu);
	  ST(&(Rp[WS(rs, 3)]), Tv, ms, &(Rp[WS(rs, 1)]));
	  TR = VCONJ(VSUB(TP, TO));
	  ST(&(Rm[0]), TR, -ms, &(Rm[0]));
	  Tw = VCONJ(VSUB(Tu, Tm));
	  ST(&(Rm[WS(rs, 3)]), Tw, -ms, &(Rm[WS(rs, 1)]));
	  TQ = VADD(TO, TP);
	  ST(&(Rp[0]), TQ, ms, &(Rp[0]));
	  TB = LDW(&(W[TWVL * 2]));
	  TC = VZMUL(TB, VADD(Tq, Tt));
	  TH = LDW(&(W[TWVL * 6]));
	  TK = VZMUL(TH, VSUB(TI, TJ));
	  Ty = VBYI(VSUB(Tk, Tj));
	  Tz = VSUB(T5, Te);
	  Tx = LDW(&(W[TWVL * 4]));
	  TA = VZMULI(Tx, VADD(Ty, Tz));
	  TF = LDW(&(W[TWVL * 8]));
	  TG = VZMULI(TF, VSUB(Tz, Ty));
	  TD = VADD(TA, TC);
	  ST(&(Rp[WS(rs, 1)]), TD, ms, &(Rp[WS(rs, 1)]));
	  TM = VCONJ(VSUB(TK, TG));
	  ST(&(Rm[WS(rs, 2)]), TM, -ms, &(Rm[0]));
	  TE = VCONJ(VSUB(TC, TA));
	  ST(&(Rm[WS(rs, 1)]), TE, -ms, &(Rm[WS(rs, 1)]));
	  TL = VADD(TG, TK);
	  ST(&(Rp[WS(rs, 2)]), TL, ms, &(Rp[0]));
     }
}
Пример #13
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();
}
Пример #14
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();
}
static void hc2cbdftv_8(R *Rp, R *Ip, R *Rm, R *Im, const R *W, stride rs, INT mb, INT me, INT ms)
{
     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(rs)) {
	  V TJ, T4, Tf, TB, TD, TE, Tm, T1, Tj, TF, Tp, Tb, Tg, Tt, Tx;
	  V T2, T3, Td, Te, T5, T6, T8, T9, Tn, T7, To, Ta, Tk, Tl, TG;
	  V TL, Tq, Tc, Tu, Th, Tv, Ty, Tw, TC, Ti, TK, TA, Tz, TI, TH;
	  V Ts, Tr, TN, TM;
	  T2 = LD(&(Rp[0]), ms, &(Rp[0]));
	  T3 = LD(&(Rm[WS(rs, 3)]), -ms, &(Rm[WS(rs, 1)]));
	  Td = LD(&(Rp[WS(rs, 2)]), ms, &(Rp[0]));
	  Te = LD(&(Rm[WS(rs, 1)]), -ms, &(Rm[WS(rs, 1)]));
	  T5 = LD(&(Rp[WS(rs, 1)]), ms, &(Rp[WS(rs, 1)]));
	  T6 = LD(&(Rm[WS(rs, 2)]), -ms, &(Rm[0]));
	  T8 = LD(&(Rp[WS(rs, 3)]), ms, &(Rp[WS(rs, 1)]));
	  T9 = LD(&(Rm[0]), -ms, &(Rm[0]));
	  TJ = LDW(&(W[0]));
	  Tk = VFMACONJ(T3, T2);
	  T4 = VFNMSCONJ(T3, T2);
	  Tl = VFMACONJ(Te, Td);
	  Tf = VFNMSCONJ(Te, Td);
	  Tn = VFMACONJ(T6, T5);
	  T7 = VFNMSCONJ(T6, T5);
	  To = VFMACONJ(T9, T8);
	  Ta = VFMSCONJ(T9, T8);
	  TB = LDW(&(W[TWVL * 8]));
	  TD = LDW(&(W[TWVL * 6]));
	  TE = VADD(Tk, Tl);
	  Tm = VSUB(Tk, Tl);
	  T1 = LDW(&(W[TWVL * 12]));
	  Tj = LDW(&(W[TWVL * 10]));
	  TF = VADD(Tn, To);
	  Tp = VSUB(Tn, To);
	  Tb = VADD(T7, Ta);
	  Tg = VSUB(T7, Ta);
	  Tt = LDW(&(W[TWVL * 4]));
	  Tx = LDW(&(W[TWVL * 2]));
	  TG = VZMUL(TD, VSUB(TE, TF));
	  TL = VADD(TE, TF);
	  Tq = VZMUL(Tj, VFNMSI(Tp, Tm));
	  Tc = VFMA(LDK(KP707106781), Tb, T4);
	  Tu = VFNMS(LDK(KP707106781), Tb, T4);
	  Th = VFMA(LDK(KP707106781), Tg, Tf);
	  Tv = VFNMS(LDK(KP707106781), Tg, Tf);
	  Ty = VZMUL(Tx, VFMAI(Tp, Tm));
	  Tw = VZMULI(Tt, VFNMSI(Tv, Tu));
	  TC = VZMULI(TB, VFMAI(Tv, Tu));
	  Ti = VZMULI(T1, VFNMSI(Th, Tc));
	  TK = VZMULI(TJ, VFMAI(Th, Tc));
	  TA = VCONJ(VSUB(Ty, Tw));
	  Tz = VADD(Tw, Ty);
	  TI = VCONJ(VSUB(TG, TC));
	  TH = VADD(TC, TG);
	  Ts = VCONJ(VSUB(Tq, Ti));
	  Tr = VADD(Ti, Tq);
	  TN = VCONJ(VSUB(TL, TK));
	  TM = VADD(TK, TL);
	  ST(&(Rm[WS(rs, 1)]), TA, -ms, &(Rm[WS(rs, 1)]));
	  ST(&(Rp[WS(rs, 1)]), Tz, ms, &(Rp[WS(rs, 1)]));
	  ST(&(Rm[WS(rs, 2)]), TI, -ms, &(Rm[0]));
	  ST(&(Rp[WS(rs, 2)]), TH, ms, &(Rp[0]));
	  ST(&(Rm[WS(rs, 3)]), Ts, -ms, &(Rm[WS(rs, 1)]));
	  ST(&(Rp[WS(rs, 3)]), Tr, ms, &(Rp[WS(rs, 1)]));
	  ST(&(Rm[0]), TN, -ms, &(Rm[0]));
	  ST(&(Rp[0]), TM, ms, &(Rp[0]));
     }
}
Пример #16
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)]));
	  }
     }
}
Пример #17
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)]));
	  }
     }
}
Пример #18
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();
}
static void hc2cbdftv_10(R *Rp, R *Ip, R *Rm, R *Im, const R *W, stride rs, INT mb, INT me, INT ms)
{
     DVK(KP250000000, +0.250000000000000000000000000000000000000000000);
     DVK(KP951056516, +0.951056516295153572116439333379382143405698634);
     DVK(KP587785252, +0.587785252292473129168705954639072768597652438);
     DVK(KP559016994, +0.559016994374947424102293417182819058860154590);
     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(rs)) {
	  V T5, TE, Ts, Tt, TC, Tz, TH, TJ, To, Tq, T2, T4, T3, T9, Tx;
	  V Tm, TB, Td, Ty, Ti, TA, T6, T8, T7, Tl, Tk, Tj, Tc, Tb, Ta;
	  V Tf, Th, Tg, TF, TG, Te, Tn;
	  T2 = LD(&(Rp[0]), ms, &(Rp[0]));
	  T3 = LD(&(Rm[WS(rs, 4)]), -ms, &(Rm[0]));
	  T4 = VCONJ(T3);
	  T5 = VSUB(T2, T4);
	  TE = VADD(T2, T4);
	  T6 = LD(&(Rp[WS(rs, 2)]), ms, &(Rp[0]));
	  T7 = LD(&(Rm[WS(rs, 2)]), -ms, &(Rm[0]));
	  T8 = VCONJ(T7);
	  T9 = VSUB(T6, T8);
	  Tx = VADD(T6, T8);
	  Tl = LD(&(Rp[WS(rs, 1)]), ms, &(Rp[WS(rs, 1)]));
	  Tj = LD(&(Rm[WS(rs, 3)]), -ms, &(Rm[WS(rs, 1)]));
	  Tk = VCONJ(Tj);
	  Tm = VSUB(Tk, Tl);
	  TB = VADD(Tk, Tl);
	  Tc = LD(&(Rp[WS(rs, 3)]), ms, &(Rp[WS(rs, 1)]));
	  Ta = LD(&(Rm[WS(rs, 1)]), -ms, &(Rm[WS(rs, 1)]));
	  Tb = VCONJ(Ta);
	  Td = VSUB(Tb, Tc);
	  Ty = VADD(Tb, Tc);
	  Tf = LD(&(Rp[WS(rs, 4)]), ms, &(Rp[0]));
	  Tg = LD(&(Rm[0]), -ms, &(Rm[0]));
	  Th = VCONJ(Tg);
	  Ti = VSUB(Tf, Th);
	  TA = VADD(Tf, Th);
	  Ts = VSUB(T9, Td);
	  Tt = VSUB(Ti, Tm);
	  TC = VSUB(TA, TB);
	  Tz = VSUB(Tx, Ty);
	  TF = VADD(Tx, Ty);
	  TG = VADD(TA, TB);
	  TH = VADD(TF, TG);
	  TJ = VMUL(LDK(KP559016994), VSUB(TF, TG));
	  Te = VADD(T9, Td);
	  Tn = VADD(Ti, Tm);
	  To = VADD(Te, Tn);
	  Tq = VMUL(LDK(KP559016994), VSUB(Te, Tn));
	  {
	       V T1c, TX, Tv, T1b, TR, T15, TL, T17, TT, T11, TW, Tu, TQ, Tr, TP;
	       V Tp, T1, T1a, TO, T14, TD, T10, TK, TZ, TI, Tw, T16, TS, TY, TM;
	       V TU, T1e, TN, T1d, T19, T13, TV, T18, T12;
	       T1c = VADD(TE, TH);
	       TW = LDW(&(W[TWVL * 8]));
	       TX = VZMULI(TW, VADD(T5, To));
	       Tu = VBYI(VFNMS(LDK(KP951056516), Tt, VMUL(LDK(KP587785252), Ts)));
	       TQ = VBYI(VFMA(LDK(KP951056516), Ts, VMUL(LDK(KP587785252), Tt)));
	       Tp = VFNMS(LDK(KP250000000), To, T5);
	       Tr = VSUB(Tp, Tq);
	       TP = VADD(Tq, Tp);
	       T1 = LDW(&(W[TWVL * 4]));
	       Tv = VZMULI(T1, VSUB(Tr, Tu));
	       T1a = LDW(&(W[0]));
	       T1b = VZMULI(T1a, VADD(TQ, TP));
	       TO = LDW(&(W[TWVL * 16]));
	       TR = VZMULI(TO, VSUB(TP, TQ));
	       T14 = LDW(&(W[TWVL * 12]));
	       T15 = VZMULI(T14, VADD(Tu, Tr));
	       TD = VBYI(VFNMS(LDK(KP951056516), TC, VMUL(LDK(KP587785252), Tz)));
	       T10 = VBYI(VFMA(LDK(KP951056516), Tz, VMUL(LDK(KP587785252), TC)));
	       TI = VFNMS(LDK(KP250000000), TH, TE);
	       TK = VSUB(TI, TJ);
	       TZ = VADD(TJ, TI);
	       Tw = LDW(&(W[TWVL * 2]));
	       TL = VZMUL(Tw, VADD(TD, TK));
	       T16 = LDW(&(W[TWVL * 10]));
	       T17 = VZMUL(T16, VADD(T10, TZ));
	       TS = LDW(&(W[TWVL * 14]));
	       TT = VZMUL(TS, VSUB(TK, TD));
	       TY = LDW(&(W[TWVL * 6]));
	       T11 = VZMUL(TY, VSUB(TZ, T10));
	       TM = VADD(Tv, TL);
	       ST(&(Rp[WS(rs, 1)]), TM, ms, &(Rp[WS(rs, 1)]));
	       TU = VADD(TR, TT);
	       ST(&(Rp[WS(rs, 4)]), TU, ms, &(Rp[0]));
	       T1e = VCONJ(VSUB(T1c, T1b));
	       ST(&(Rm[0]), T1e, -ms, &(Rm[0]));
	       TN = VCONJ(VSUB(TL, Tv));
	       ST(&(Rm[WS(rs, 1)]), TN, -ms, &(Rm[WS(rs, 1)]));
	       T1d = VADD(T1b, T1c);
	       ST(&(Rp[0]), T1d, ms, &(Rp[0]));
	       T19 = VCONJ(VSUB(T17, T15));
	       ST(&(Rm[WS(rs, 3)]), T19, -ms, &(Rm[WS(rs, 1)]));
	       T13 = VCONJ(VSUB(T11, TX));
	       ST(&(Rm[WS(rs, 2)]), T13, -ms, &(Rm[0]));
	       TV = VCONJ(VSUB(TT, TR));
	       ST(&(Rm[WS(rs, 4)]), TV, -ms, &(Rm[0]));
	       T18 = VADD(T15, T17);
	       ST(&(Rp[WS(rs, 3)]), T18, ms, &(Rp[WS(rs, 1)]));
	       T12 = VADD(TX, T11);
	       ST(&(Rp[WS(rs, 2)]), T12, ms, &(Rp[0]));
	  }
     }
}
Пример #20
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(KP382683432, +0.382683432365089771728459984030398866761344562);
     DVK(KP923879532, +0.923879532511286756128183189396788286822416626);
     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 Tf, T16, TZ, T1C, TI, T1a, TV, T1D, T1F, T1G, Ty, T19, TC, T17, TS;
	       V T10;
	       {
		    V T2, TD, T4, TF, Tc, Tb, Td, T6, T8, T9, T3, TE, Ta, T7, T5;
		    V Te, TX, TY, TG, TH, TT, TU, Tj, TM, Tw, TQ, Tn, TN, Ts, TP;
		    V Tg, Ti, Th, Tt, Tv, Tu, Tk, Tm, Tl, Tr, Tq, Tp, To, Tx, TA;
		    V TB, TO, TR;
		    T2 = LD(&(Rp[0]), ms, &(Rp[0]));
		    TD = LD(&(Rp[WS(rs, 4)]), ms, &(Rp[0]));
		    T3 = LD(&(Rm[WS(rs, 7)]), -ms, &(Rm[WS(rs, 1)]));
		    T4 = VCONJ(T3);
		    TE = LD(&(Rm[WS(rs, 3)]), -ms, &(Rm[WS(rs, 1)]));
		    TF = VCONJ(TE);
		    Tc = LD(&(Rp[WS(rs, 6)]), ms, &(Rp[0]));
		    Ta = LD(&(Rm[WS(rs, 1)]), -ms, &(Rm[WS(rs, 1)]));
		    Tb = VCONJ(Ta);
		    Td = VSUB(Tb, Tc);
		    T6 = LD(&(Rp[WS(rs, 2)]), ms, &(Rp[0]));
		    T7 = LD(&(Rm[WS(rs, 5)]), -ms, &(Rm[WS(rs, 1)]));
		    T8 = VCONJ(T7);
		    T9 = VSUB(T6, T8);
		    T5 = VSUB(T2, T4);
		    Te = VMUL(LDK(KP707106781), VADD(T9, Td));
		    Tf = VADD(T5, Te);
		    T16 = VSUB(T5, Te);
		    TX = VADD(T2, T4);
		    TY = VADD(TD, TF);
		    TZ = VSUB(TX, TY);
		    T1C = VADD(TX, TY);
		    TG = VSUB(TD, TF);
		    TH = VMUL(LDK(KP707106781), VSUB(T9, Td));
		    TI = VADD(TG, TH);
		    T1a = VSUB(TH, TG);
		    TT = VADD(T6, T8);
		    TU = VADD(Tb, Tc);
		    TV = VSUB(TT, TU);
		    T1D = VADD(TT, TU);
		    Tg = LD(&(Rp[WS(rs, 1)]), ms, &(Rp[WS(rs, 1)]));
		    Th = LD(&(Rm[WS(rs, 6)]), -ms, &(Rm[0]));
		    Ti = VCONJ(Th);
		    Tj = VSUB(Tg, Ti);
		    TM = VADD(Tg, Ti);
		    Tt = LD(&(Rp[WS(rs, 3)]), ms, &(Rp[WS(rs, 1)]));
		    Tu = LD(&(Rm[WS(rs, 4)]), -ms, &(Rm[0]));
		    Tv = VCONJ(Tu);
		    Tw = VSUB(Tt, Tv);
		    TQ = VADD(Tt, Tv);
		    Tk = LD(&(Rp[WS(rs, 5)]), ms, &(Rp[WS(rs, 1)]));
		    Tl = LD(&(Rm[WS(rs, 2)]), -ms, &(Rm[0]));
		    Tm = VCONJ(Tl);
		    Tn = VSUB(Tk, Tm);
		    TN = VADD(Tk, Tm);
		    Tr = LD(&(Rp[WS(rs, 7)]), ms, &(Rp[WS(rs, 1)]));
		    Tp = LD(&(Rm[0]), -ms, &(Rm[0]));
		    Tq = VCONJ(Tp);
		    Ts = VSUB(Tq, Tr);
		    TP = VADD(Tq, Tr);
		    T1F = VADD(TM, TN);
		    T1G = VADD(TP, TQ);
		    To = VFNMS(LDK(KP382683432), Tn, VMUL(LDK(KP923879532), Tj));
		    Tx = VFMA(LDK(KP923879532), Ts, VMUL(LDK(KP382683432), Tw));
		    Ty = VADD(To, Tx);
		    T19 = VSUB(To, Tx);
		    TA = VFMA(LDK(KP382683432), Tj, VMUL(LDK(KP923879532), Tn));
		    TB = VFNMS(LDK(KP382683432), Ts, VMUL(LDK(KP923879532), Tw));
		    TC = VADD(TA, TB);
		    T17 = VSUB(TA, TB);
		    TO = VSUB(TM, TN);
		    TR = VSUB(TP, TQ);
		    TS = VMUL(LDK(KP707106781), VSUB(TO, TR));
		    T10 = VMUL(LDK(KP707106781), VADD(TO, TR));
	       }
	       {
		    V T21, T1W, T1u, T20, T1I, T1O, TK, T1S, T12, T1e, T1k, T1A, T1o, T1w, T1c;
		    V T1M, T1U, T1V, T1T, T1s, T1t, T1r, T1Z, T1E, T1H, T1B, T1N, Tz, TJ, T1;
		    V T1R, TW, T11, TL, T1d, T1i, T1j, T1h, T1z, T1m, T1n, T1l, T1v, T18, T1b;
		    V T15, T1L, T13, T1g, T1X, T23, T14, T1f, T1Y, T22, T1p, T1y, T1J, T1Q, T1q;
		    V T1x, T1K, T1P;
		    T1U = VADD(T1C, T1D);
		    T1V = VADD(T1F, T1G);
		    T21 = VADD(T1U, T1V);
		    T1T = LDW(&(W[TWVL * 14]));
		    T1W = VZMUL(T1T, VSUB(T1U, T1V));
		    T1s = VADD(Tf, Ty);
		    T1t = VBYI(VADD(TI, TC));
		    T1r = LDW(&(W[TWVL * 28]));
		    T1u = VZMULI(T1r, VSUB(T1s, T1t));
		    T1Z = LDW(&(W[0]));
		    T20 = VZMULI(T1Z, VADD(T1s, T1t));
		    T1E = VSUB(T1C, T1D);
		    T1H = VBYI(VSUB(T1F, T1G));
		    T1B = LDW(&(W[TWVL * 22]));
		    T1I = VZMUL(T1B, VSUB(T1E, T1H));
		    T1N = LDW(&(W[TWVL * 6]));
		    T1O = VZMUL(T1N, VADD(T1E, T1H));
		    Tz = VSUB(Tf, Ty);
		    TJ = VBYI(VSUB(TC, TI));
		    T1 = LDW(&(W[TWVL * 12]));
		    TK = VZMULI(T1, VADD(Tz, TJ));
		    T1R = LDW(&(W[TWVL * 16]));
		    T1S = VZMULI(T1R, VSUB(Tz, TJ));
		    TW = VBYI(VSUB(TS, TV));
		    T11 = VSUB(TZ, T10);
		    TL = LDW(&(W[TWVL * 10]));
		    T12 = VZMUL(TL, VADD(TW, T11));
		    T1d = LDW(&(W[TWVL * 18]));
		    T1e = VZMUL(T1d, VSUB(T11, TW));
		    T1i = VBYI(VADD(T1a, T19));
		    T1j = VADD(T16, T17);
		    T1h = LDW(&(W[TWVL * 4]));
		    T1k = VZMULI(T1h, VADD(T1i, T1j));
		    T1z = LDW(&(W[TWVL * 24]));
		    T1A = VZMULI(T1z, VSUB(T1j, T1i));
		    T1m = VBYI(VADD(TV, TS));
		    T1n = VADD(TZ, T10);
		    T1l = LDW(&(W[TWVL * 2]));
		    T1o = VZMUL(T1l, VADD(T1m, T1n));
		    T1v = LDW(&(W[TWVL * 26]));
		    T1w = VZMUL(T1v, VSUB(T1n, T1m));
		    T18 = VSUB(T16, T17);
		    T1b = VBYI(VSUB(T19, T1a));
		    T15 = LDW(&(W[TWVL * 20]));
		    T1c = VZMULI(T15, VSUB(T18, T1b));
		    T1L = LDW(&(W[TWVL * 8]));
		    T1M = VZMULI(T1L, VADD(T1b, T18));
		    T13 = VADD(TK, T12);
		    ST(&(Rp[WS(rs, 3)]), T13, ms, &(Rp[WS(rs, 1)]));
		    T1g = VCONJ(VSUB(T1e, T1c));
		    ST(&(Rm[WS(rs, 5)]), T1g, -ms, &(Rm[WS(rs, 1)]));
		    T1X = VADD(T1S, T1W);
		    ST(&(Rp[WS(rs, 4)]), T1X, ms, &(Rp[0]));
		    T23 = VCONJ(VSUB(T21, T20));
		    ST(&(Rm[0]), T23, -ms, &(Rm[0]));
		    T14 = VCONJ(VSUB(T12, TK));
		    ST(&(Rm[WS(rs, 3)]), T14, -ms, &(Rm[WS(rs, 1)]));
		    T1f = VADD(T1c, T1e);
		    ST(&(Rp[WS(rs, 5)]), T1f, ms, &(Rp[WS(rs, 1)]));
		    T1Y = VCONJ(VSUB(T1W, T1S));
		    ST(&(Rm[WS(rs, 4)]), T1Y, -ms, &(Rm[0]));
		    T22 = VADD(T20, T21);
		    ST(&(Rp[0]), T22, ms, &(Rp[0]));
		    T1p = VADD(T1k, T1o);
		    ST(&(Rp[WS(rs, 1)]), T1p, ms, &(Rp[WS(rs, 1)]));
		    T1y = VCONJ(VSUB(T1w, T1u));
		    ST(&(Rm[WS(rs, 7)]), T1y, -ms, &(Rm[WS(rs, 1)]));
		    T1J = VADD(T1A, T1I);
		    ST(&(Rp[WS(rs, 6)]), T1J, ms, &(Rp[0]));
		    T1Q = VCONJ(VSUB(T1O, T1M));
		    ST(&(Rm[WS(rs, 2)]), T1Q, -ms, &(Rm[0]));
		    T1q = VCONJ(VSUB(T1o, T1k));
		    ST(&(Rm[WS(rs, 1)]), T1q, -ms, &(Rm[WS(rs, 1)]));
		    T1x = VADD(T1u, T1w);
		    ST(&(Rp[WS(rs, 7)]), T1x, ms, &(Rp[WS(rs, 1)]));
		    T1K = VCONJ(VSUB(T1I, T1A));
		    ST(&(Rm[WS(rs, 6)]), T1K, -ms, &(Rm[0]));
		    T1P = VADD(T1M, T1O);
		    ST(&(Rp[WS(rs, 2)]), T1P, ms, &(Rp[0]));
	       }
	  }
     }
     VLEAVE();
}
static void hc2cbdftv_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(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(rs)) {
	  V Ts, T4, TR, T1, TZ, TD, Ty, Tn, Ti, TT, T11, TJ, T15, Tr, TN;
	  V TE, Tv, To, Tb, T8, Tw, Te, Tx, Th, Tt, T7, T9, T2, T3, Tc;
	  V Td, Tf, Tg, T5, T6, Tu, Ta;
	  T2 = LD(&(Rp[0]), ms, &(Rp[0]));
	  T3 = LD(&(Rm[WS(rs, 4)]), -ms, &(Rm[0]));
	  Tc = LD(&(Rp[WS(rs, 4)]), ms, &(Rp[0]));
	  Td = LD(&(Rm[0]), -ms, &(Rm[0]));
	  Tf = LD(&(Rp[WS(rs, 1)]), ms, &(Rp[WS(rs, 1)]));
	  Tg = LD(&(Rm[WS(rs, 3)]), -ms, &(Rm[WS(rs, 1)]));
	  T5 = LD(&(Rp[WS(rs, 2)]), ms, &(Rp[0]));
	  T6 = LD(&(Rm[WS(rs, 2)]), -ms, &(Rm[0]));
	  T8 = LD(&(Rp[WS(rs, 3)]), ms, &(Rp[WS(rs, 1)]));
	  Ts = VFMACONJ(T3, T2);
	  T4 = VFNMSCONJ(T3, T2);
	  Tw = VFMACONJ(Td, Tc);
	  Te = VFNMSCONJ(Td, Tc);
	  Tx = VFMACONJ(Tg, Tf);
	  Th = VFMSCONJ(Tg, Tf);
	  Tt = VFMACONJ(T6, T5);
	  T7 = VFNMSCONJ(T6, T5);
	  T9 = LD(&(Rm[WS(rs, 1)]), -ms, &(Rm[WS(rs, 1)]));
	  TR = LDW(&(W[TWVL * 8]));
	  T1 = LDW(&(W[TWVL * 4]));
	  TZ = LDW(&(W[TWVL * 12]));
	  TD = VSUB(Tw, Tx);
	  Ty = VADD(Tw, Tx);
	  Tn = VSUB(Te, Th);
	  Ti = VADD(Te, Th);
	  Tu = VFMACONJ(T9, T8);
	  Ta = VFMSCONJ(T9, T8);
	  TT = LDW(&(W[TWVL * 6]));
	  T11 = LDW(&(W[TWVL * 10]));
	  TJ = LDW(&(W[TWVL * 16]));
	  T15 = LDW(&(W[0]));
	  Tr = LDW(&(W[TWVL * 2]));
	  TN = LDW(&(W[TWVL * 14]));
	  TE = VSUB(Tt, Tu);
	  Tv = VADD(Tt, Tu);
	  To = VSUB(T7, Ta);
	  Tb = VADD(T7, Ta);
	  {
	       V TV, TF, Tz, TB, TL, Tp, Tj, Tl, T17, TA, TS, Tk, TC, TU, TK;
	       V Tm, TO, TG, T12, TW, T16, TM, T10, Tq, TX, TY, T18, T19, TQ, TP;
	       V T13, T14, TI, TH;
	       TV = VMUL(LDK(KP951056516), VFMA(LDK(KP618033988), TD, TE));
	       TF = VMUL(LDK(KP951056516), VFNMS(LDK(KP618033988), TE, TD));
	       Tz = VADD(Tv, Ty);
	       TB = VSUB(Tv, Ty);
	       TL = VMUL(LDK(KP951056516), VFMA(LDK(KP618033988), Tn, To));
	       Tp = VMUL(LDK(KP951056516), VFNMS(LDK(KP618033988), To, Tn));
	       Tj = VADD(Tb, Ti);
	       Tl = VSUB(Tb, Ti);
	       T17 = VADD(Ts, Tz);
	       TA = VFNMS(LDK(KP250000000), Tz, Ts);
	       TS = VZMULI(TR, VADD(T4, Tj));
	       Tk = VFNMS(LDK(KP250000000), Tj, T4);
	       TC = VFNMS(LDK(KP559016994), TB, TA);
	       TU = VFMA(LDK(KP559016994), TB, TA);
	       TK = VFMA(LDK(KP559016994), Tl, Tk);
	       Tm = VFNMS(LDK(KP559016994), Tl, Tk);
	       TO = VZMUL(TN, VFMAI(TF, TC));
	       TG = VZMUL(Tr, VFNMSI(TF, TC));
	       T12 = VZMUL(T11, VFMAI(TV, TU));
	       TW = VZMUL(TT, VFNMSI(TV, TU));
	       T16 = VZMULI(T15, VFMAI(TL, TK));
	       TM = VZMULI(TJ, VFNMSI(TL, TK));
	       T10 = VZMULI(TZ, VFNMSI(Tp, Tm));
	       Tq = VZMULI(T1, VFMAI(Tp, Tm));
	       TX = VADD(TS, TW);
	       TY = VCONJ(VSUB(TW, TS));
	       T18 = VADD(T16, T17);
	       T19 = VCONJ(VSUB(T17, T16));
	       TQ = VCONJ(VSUB(TO, TM));
	       TP = VADD(TM, TO);
	       T13 = VADD(T10, T12);
	       T14 = VCONJ(VSUB(T12, T10));
	       TI = VCONJ(VSUB(TG, Tq));
	       TH = VADD(Tq, TG);
	       ST(&(Rp[WS(rs, 2)]), TX, ms, &(Rp[0]));
	       ST(&(Rm[WS(rs, 2)]), TY, -ms, &(Rm[0]));
	       ST(&(Rp[0]), T18, ms, &(Rp[0]));
	       ST(&(Rm[0]), T19, -ms, &(Rm[0]));
	       ST(&(Rm[WS(rs, 4)]), TQ, -ms, &(Rm[0]));
	       ST(&(Rp[WS(rs, 4)]), TP, ms, &(Rp[0]));
	       ST(&(Rp[WS(rs, 3)]), T13, ms, &(Rp[WS(rs, 1)]));
	       ST(&(Rm[WS(rs, 3)]), T14, -ms, &(Rm[WS(rs, 1)]));
	       ST(&(Rm[WS(rs, 1)]), TI, -ms, &(Rm[WS(rs, 1)]));
	       ST(&(Rp[WS(rs, 1)]), TH, ms, &(Rp[WS(rs, 1)]));
	  }
     }
}
Пример #22
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();
}