static void n1bv_5(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs) { DVK(KP559016994, +0.559016994374947424102293417182819058860154590); DVK(KP250000000, +0.250000000000000000000000000000000000000000000); DVK(KP618033988, +0.618033988749894848204586834365638117720309180); DVK(KP951056516, +0.951056516295153572116439333379382143405698634); { INT i; const R *xi; R *xo; xi = ii; xo = io; for (i = v; i > 0; i = i - VL, xi = xi + (VL * ivs), xo = xo + (VL * ovs), MAKE_VOLATILE_STRIDE(is), MAKE_VOLATILE_STRIDE(os)) { V T1, T2, T3, T5, T6; T1 = LD(&(xi[0]), ivs, &(xi[0])); T2 = LD(&(xi[WS(is, 1)]), ivs, &(xi[WS(is, 1)])); T3 = LD(&(xi[WS(is, 4)]), ivs, &(xi[0])); T5 = LD(&(xi[WS(is, 2)]), ivs, &(xi[0])); T6 = LD(&(xi[WS(is, 3)]), ivs, &(xi[WS(is, 1)])); { V Tc, T4, Td, T7; Tc = VSUB(T2, T3); T4 = VADD(T2, T3); Td = VSUB(T5, T6); T7 = VADD(T5, T6); { V Tg, Te, Ta, T8, T9, Tf, Tb; Tg = VMUL(LDK(KP951056516), VFNMS(LDK(KP618033988), Tc, Td)); Te = VMUL(LDK(KP951056516), VFMA(LDK(KP618033988), Td, Tc)); Ta = VSUB(T4, T7); T8 = VADD(T4, T7); T9 = VFNMS(LDK(KP250000000), T8, T1); ST(&(xo[0]), VADD(T1, T8), ovs, &(xo[0])); Tf = VFNMS(LDK(KP559016994), Ta, T9); Tb = VFMA(LDK(KP559016994), Ta, T9); ST(&(xo[WS(os, 2)]), VFNMSI(Tg, Tf), ovs, &(xo[0])); ST(&(xo[WS(os, 3)]), VFMAI(Tg, Tf), ovs, &(xo[WS(os, 1)])); ST(&(xo[WS(os, 4)]), VFNMSI(Te, Tb), ovs, &(xo[0])); ST(&(xo[WS(os, 1)]), VFMAI(Te, Tb), ovs, &(xo[WS(os, 1)])); } } } } VLEAVE(); }
static void n1bv_5(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs) { DVK(KP250000000, +0.250000000000000000000000000000000000000000000); DVK(KP587785252, +0.587785252292473129168705954639072768597652438); DVK(KP951056516, +0.951056516295153572116439333379382143405698634); DVK(KP559016994, +0.559016994374947424102293417182819058860154590); { INT i; const R *xi; R *xo; xi = ii; xo = io; for (i = v; i > 0; i = i - VL, xi = xi + (VL * ivs), xo = xo + (VL * ovs), MAKE_VOLATILE_STRIDE(is), MAKE_VOLATILE_STRIDE(os)) { V Tb, T3, Tc, T6, Ta; Tb = LD(&(xi[0]), ivs, &(xi[0])); { V T1, T2, T8, T4, T5, T9; T1 = LD(&(xi[WS(is, 1)]), ivs, &(xi[WS(is, 1)])); T2 = LD(&(xi[WS(is, 4)]), ivs, &(xi[0])); T8 = VADD(T1, T2); T4 = LD(&(xi[WS(is, 2)]), ivs, &(xi[0])); T5 = LD(&(xi[WS(is, 3)]), ivs, &(xi[WS(is, 1)])); T9 = VADD(T4, T5); T3 = VSUB(T1, T2); Tc = VADD(T8, T9); T6 = VSUB(T4, T5); Ta = VMUL(LDK(KP559016994), VSUB(T8, T9)); } ST(&(xo[0]), VADD(Tb, Tc), ovs, &(xo[0])); { V T7, Tf, Te, Tg, Td; T7 = VBYI(VFMA(LDK(KP951056516), T3, VMUL(LDK(KP587785252), T6))); Tf = VBYI(VFNMS(LDK(KP951056516), T6, VMUL(LDK(KP587785252), T3))); Td = VFNMS(LDK(KP250000000), Tc, Tb); Te = VADD(Ta, Td); Tg = VSUB(Td, Ta); ST(&(xo[WS(os, 1)]), VADD(T7, Te), ovs, &(xo[WS(os, 1)])); ST(&(xo[WS(os, 3)]), VSUB(Tg, Tf), ovs, &(xo[WS(os, 1)])); ST(&(xo[WS(os, 4)]), VSUB(Te, T7), ovs, &(xo[0])); ST(&(xo[WS(os, 2)]), VADD(Tf, Tg), ovs, &(xo[0])); } } } VLEAVE(); }
static void n1bv_2(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs) { { INT i; const R *xi; R *xo; xi = ii; xo = io; for (i = v; i > 0; i = i - VL, xi = xi + (VL * ivs), xo = xo + (VL * ovs), MAKE_VOLATILE_STRIDE(is), MAKE_VOLATILE_STRIDE(os)) { V T1, T2; T1 = LD(&(xi[0]), ivs, &(xi[0])); T2 = LD(&(xi[WS(is, 1)]), ivs, &(xi[WS(is, 1)])); ST(&(xo[WS(os, 1)]), VSUB(T1, T2), ovs, &(xo[WS(os, 1)])); ST(&(xo[0]), VADD(T1, T2), ovs, &(xo[0])); } } VLEAVE(); }
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 T2, T3, T1, T5, T4, T7, T6; T2 = LD(&(Rp[0]), ms, &(Rp[0])); T3 = LD(&(Rm[0]), -ms, &(Rm[0])); T1 = LDW(&(W[0])); T5 = VFMACONJ(T3, T2); T4 = VZMULI(T1, VFNMSCONJ(T3, T2)); T7 = VCONJ(VSUB(T5, T4)); T6 = VADD(T4, T5); ST(&(Rm[0]), T7, -ms, &(Rm[0])); ST(&(Rp[0]), T6, ms, &(Rp[0])); } } VLEAVE(); }
static void n2fv_2(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs) { { INT i; const R *xi; R *xo; xi = ri; xo = ro; for (i = v; i > 0; i = i - VL, xi = xi + (VL * ivs), xo = xo + (VL * ovs), MAKE_VOLATILE_STRIDE(4, is), MAKE_VOLATILE_STRIDE(4, os)) { V T1, T2, T3, T4; T1 = LD(&(xi[0]), ivs, &(xi[0])); T2 = LD(&(xi[WS(is, 1)]), ivs, &(xi[WS(is, 1)])); T3 = VSUB(T1, T2); STM2(&(xo[2]), T3, ovs, &(xo[2])); T4 = VADD(T1, T2); STM2(&(xo[0]), T4, ovs, &(xo[0])); STN2(&(xo[0]), T4, T3, ovs); } } VLEAVE(); }
static void q1fv_2(R *ri, R *ii, const R *W, stride rs, stride vs, INT mb, INT me, INT ms) { { INT m; R *x; x = ri; for (m = mb, W = W + (mb * ((TWVL / VL) * 2)); m < me; m = m + VL, x = x + (VL * ms), W = W + (TWVL * 2), MAKE_VOLATILE_STRIDE(4, rs), MAKE_VOLATILE_STRIDE(4, vs)) { V T1, T2, T3, T4, T5, T6; T1 = LD(&(x[0]), ms, &(x[0])); T2 = LD(&(x[WS(rs, 1)]), ms, &(x[WS(rs, 1)])); T3 = BYTWJ(&(W[0]), VSUB(T1, T2)); T4 = LD(&(x[WS(vs, 1)]), ms, &(x[WS(vs, 1)])); T5 = LD(&(x[WS(vs, 1) + WS(rs, 1)]), ms, &(x[WS(vs, 1) + WS(rs, 1)])); T6 = BYTWJ(&(W[0]), VSUB(T4, T5)); ST(&(x[WS(vs, 1)]), T3, ms, &(x[WS(vs, 1)])); ST(&(x[WS(vs, 1) + WS(rs, 1)]), T6, ms, &(x[WS(vs, 1) + WS(rs, 1)])); ST(&(x[0]), VADD(T1, T2), ms, &(x[0])); ST(&(x[WS(rs, 1)]), VADD(T4, T5), ms, &(x[WS(rs, 1)])); } } VLEAVE(); }
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 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])); } } VLEAVE(); }
static void t3fv_4(R *ri, R *ii, const R *W, stride rs, INT mb, INT me, INT ms) { { INT m; R *x; x = ri; for (m = mb, W = W + (mb * ((TWVL / VL) * 4)); m < me; m = m + VL, x = x + (VL * ms), W = W + (TWVL * 4), MAKE_VOLATILE_STRIDE(4, rs)) { V T2, T3, T4; T2 = LDW(&(W[0])); T3 = LDW(&(W[TWVL * 2])); T4 = VZMULJ(T2, T3); { V T1, Tb, T6, T9, Ta, T5, T8; T1 = LD(&(x[0]), ms, &(x[0])); Ta = LD(&(x[WS(rs, 3)]), ms, &(x[WS(rs, 1)])); Tb = VZMULJ(T3, Ta); T5 = LD(&(x[WS(rs, 2)]), ms, &(x[0])); T6 = VZMULJ(T4, T5); T8 = LD(&(x[WS(rs, 1)]), ms, &(x[WS(rs, 1)])); T9 = VZMULJ(T2, T8); { V T7, Tc, Td, Te; T7 = VSUB(T1, T6); Tc = VBYI(VSUB(T9, Tb)); ST(&(x[WS(rs, 1)]), VSUB(T7, Tc), ms, &(x[WS(rs, 1)])); ST(&(x[WS(rs, 3)]), VADD(T7, Tc), ms, &(x[WS(rs, 1)])); Td = VADD(T1, T6); Te = VADD(T9, Tb); ST(&(x[WS(rs, 2)]), VSUB(Td, Te), ms, &(x[0])); ST(&(x[0]), VADD(Td, Te), ms, &(x[0])); } } } } VLEAVE(); }
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)])); } } VLEAVE(); }
static void t3bv_10(R *ri, R *ii, const R *W, stride rs, INT mb, INT me, INT ms) { DVK(KP559016994, +0.559016994374947424102293417182819058860154590); DVK(KP250000000, +0.250000000000000000000000000000000000000000000); DVK(KP618033988, +0.618033988749894848204586834365638117720309180); DVK(KP951056516, +0.951056516295153572116439333379382143405698634); { INT m; R *x; x = ii; for (m = mb, W = W + (mb * ((TWVL / VL) * 6)); m < me; m = m + VL, x = x + (VL * ms), W = W + (TWVL * 6), MAKE_VOLATILE_STRIDE(10, rs)) { V T1, T7, Th, Tx, Tr, Td, Tp, T6, Tv, Tc, Te, Ti, Tl, T2, T3; V T5; T2 = LDW(&(W[0])); T3 = LDW(&(W[TWVL * 2])); T5 = LDW(&(W[TWVL * 4])); T1 = LD(&(x[0]), ms, &(x[0])); T7 = LD(&(x[WS(rs, 5)]), ms, &(x[WS(rs, 1)])); { V To, Tw, Tq, Tu, Ta, T4, Tt, Tk, Tb; To = LD(&(x[WS(rs, 4)]), ms, &(x[0])); Tw = LD(&(x[WS(rs, 1)]), ms, &(x[WS(rs, 1)])); Tq = LD(&(x[WS(rs, 9)]), ms, &(x[WS(rs, 1)])); Tu = LD(&(x[WS(rs, 6)]), ms, &(x[0])); Ta = VZMULJ(T2, T3); T4 = VZMUL(T2, T3); Th = VZMULJ(T2, T5); Tt = VZMULJ(T3, T5); Tb = LD(&(x[WS(rs, 2)]), ms, &(x[0])); Tx = VZMUL(T2, Tw); Tr = VZMUL(T5, Tq); Tk = LD(&(x[WS(rs, 3)]), ms, &(x[WS(rs, 1)])); Td = VZMULJ(Ta, T5); Tp = VZMUL(T4, To); T6 = VZMULJ(T4, T5); Tv = VZMUL(Tt, Tu); Tc = VZMUL(Ta, Tb); Te = LD(&(x[WS(rs, 7)]), ms, &(x[WS(rs, 1)])); Ti = LD(&(x[WS(rs, 8)]), ms, &(x[0])); Tl = VZMUL(T3, Tk); } { V TN, Ts, T8, Ty, TO, Tf, Tj; TN = VADD(Tp, Tr); Ts = VSUB(Tp, Tr); T8 = VZMUL(T6, T7); Ty = VSUB(Tv, Tx); TO = VADD(Tv, Tx); Tf = VZMUL(Td, Te); Tj = VZMUL(Th, Ti); { V T9, TJ, TP, TU, Tz, TF, Tg, TK, Tm, TL; T9 = VSUB(T1, T8); TJ = VADD(T1, T8); TP = VADD(TN, TO); TU = VSUB(TN, TO); Tz = VADD(Ts, Ty); TF = VSUB(Ts, Ty); Tg = VSUB(Tc, Tf); TK = VADD(Tc, Tf); Tm = VSUB(Tj, Tl); TL = VADD(Tj, Tl); { V TM, TV, Tn, TE; TM = VADD(TK, TL); TV = VSUB(TK, TL); Tn = VADD(Tg, Tm); TE = VSUB(Tg, Tm); { V TW, TY, TS, TQ, TG, TI, TC, TA, TR, TB; TW = VMUL(LDK(KP951056516), VFNMS(LDK(KP618033988), TV, TU)); TY = VMUL(LDK(KP951056516), VFMA(LDK(KP618033988), TU, TV)); TS = VSUB(TM, TP); TQ = VADD(TM, TP); TG = VMUL(LDK(KP951056516), VFMA(LDK(KP618033988), TF, TE)); TI = VMUL(LDK(KP951056516), VFNMS(LDK(KP618033988), TE, TF)); TC = VSUB(Tn, Tz); TA = VADD(Tn, Tz); ST(&(x[0]), VADD(TJ, TQ), ms, &(x[0])); TR = VFNMS(LDK(KP250000000), TQ, TJ); ST(&(x[WS(rs, 5)]), VADD(T9, TA), ms, &(x[WS(rs, 1)])); TB = VFNMS(LDK(KP250000000), TA, T9); { V TX, TT, TH, TD; TX = VFMA(LDK(KP559016994), TS, TR); TT = VFNMS(LDK(KP559016994), TS, TR); TH = VFNMS(LDK(KP559016994), TC, TB); TD = VFMA(LDK(KP559016994), TC, TB); ST(&(x[WS(rs, 8)]), VFMAI(TW, TT), ms, &(x[0])); ST(&(x[WS(rs, 2)]), VFNMSI(TW, TT), ms, &(x[0])); ST(&(x[WS(rs, 6)]), VFMAI(TY, TX), ms, &(x[0])); ST(&(x[WS(rs, 4)]), VFNMSI(TY, TX), ms, &(x[0])); ST(&(x[WS(rs, 9)]), VFNMSI(TG, TD), ms, &(x[WS(rs, 1)])); ST(&(x[WS(rs, 1)]), VFMAI(TG, TD), ms, &(x[WS(rs, 1)])); ST(&(x[WS(rs, 7)]), VFNMSI(TI, TH), ms, &(x[WS(rs, 1)])); ST(&(x[WS(rs, 3)]), VFMAI(TI, TH), ms, &(x[WS(rs, 1)])); } } } } } } } VLEAVE(); }
static void t3bv_10(R *ri, R *ii, const R *W, stride rs, INT mb, INT me, INT ms) { DVK(KP587785252, +0.587785252292473129168705954639072768597652438); DVK(KP951056516, +0.951056516295153572116439333379382143405698634); DVK(KP250000000, +0.250000000000000000000000000000000000000000000); DVK(KP559016994, +0.559016994374947424102293417182819058860154590); { INT m; R *x; x = ii; for (m = mb, W = W + (mb * ((TWVL / VL) * 6)); m < me; m = m + VL, x = x + (VL * ms), W = W + (TWVL * 6), MAKE_VOLATILE_STRIDE(10, rs)) { V T1, T2, T3, Ti, T6, T7, TA, Tb, To; T1 = LDW(&(W[0])); T2 = LDW(&(W[TWVL * 2])); T3 = VZMULJ(T1, T2); Ti = VZMUL(T1, T2); T6 = LDW(&(W[TWVL * 4])); T7 = VZMULJ(T3, T6); TA = VZMULJ(Ti, T6); Tb = VZMULJ(T1, T6); To = VZMULJ(T2, T6); { V TD, TQ, Tn, Tt, Tx, TM, TN, TS, Ta, Tg, Tw, TJ, TK, TR, Tz; V TC, TB; Tz = LD(&(x[0]), ms, &(x[0])); TB = LD(&(x[WS(rs, 5)]), ms, &(x[WS(rs, 1)])); TC = VZMUL(TA, TB); TD = VSUB(Tz, TC); TQ = VADD(Tz, TC); { V Tk, Ts, Tm, Tq; { V Tj, Tr, Tl, Tp; Tj = LD(&(x[WS(rs, 4)]), ms, &(x[0])); Tk = VZMUL(Ti, Tj); Tr = LD(&(x[WS(rs, 1)]), ms, &(x[WS(rs, 1)])); Ts = VZMUL(T1, Tr); Tl = LD(&(x[WS(rs, 9)]), ms, &(x[WS(rs, 1)])); Tm = VZMUL(T6, Tl); Tp = LD(&(x[WS(rs, 6)]), ms, &(x[0])); Tq = VZMUL(To, Tp); } Tn = VSUB(Tk, Tm); Tt = VSUB(Tq, Ts); Tx = VADD(Tn, Tt); TM = VADD(Tk, Tm); TN = VADD(Tq, Ts); TS = VADD(TM, TN); } { V T5, Tf, T9, Td; { V T4, Te, T8, Tc; T4 = LD(&(x[WS(rs, 2)]), ms, &(x[0])); T5 = VZMUL(T3, T4); Te = LD(&(x[WS(rs, 3)]), ms, &(x[WS(rs, 1)])); Tf = VZMUL(T2, Te); T8 = LD(&(x[WS(rs, 7)]), ms, &(x[WS(rs, 1)])); T9 = VZMUL(T7, T8); Tc = LD(&(x[WS(rs, 8)]), ms, &(x[0])); Td = VZMUL(Tb, Tc); } Ta = VSUB(T5, T9); Tg = VSUB(Td, Tf); Tw = VADD(Ta, Tg); TJ = VADD(T5, T9); TK = VADD(Td, Tf); TR = VADD(TJ, TK); } { V Ty, TE, TF, Tv, TI, Th, Tu, TH, TG; Ty = VMUL(LDK(KP559016994), VSUB(Tw, Tx)); TE = VADD(Tw, Tx); TF = VFNMS(LDK(KP250000000), TE, TD); Th = VSUB(Ta, Tg); Tu = VSUB(Tn, Tt); Tv = VBYI(VFMA(LDK(KP951056516), Th, VMUL(LDK(KP587785252), Tu))); TI = VBYI(VFNMS(LDK(KP951056516), Tu, VMUL(LDK(KP587785252), Th))); ST(&(x[WS(rs, 5)]), VADD(TD, TE), ms, &(x[WS(rs, 1)])); TH = VSUB(TF, Ty); ST(&(x[WS(rs, 3)]), VSUB(TH, TI), ms, &(x[WS(rs, 1)])); ST(&(x[WS(rs, 7)]), VADD(TI, TH), ms, &(x[WS(rs, 1)])); TG = VADD(Ty, TF); ST(&(x[WS(rs, 1)]), VADD(Tv, TG), ms, &(x[WS(rs, 1)])); ST(&(x[WS(rs, 9)]), VSUB(TG, Tv), ms, &(x[WS(rs, 1)])); } { V TV, TT, TU, TP, TY, TL, TO, TX, TW; TV = VMUL(LDK(KP559016994), VSUB(TR, TS)); TT = VADD(TR, TS); TU = VFNMS(LDK(KP250000000), TT, TQ); TL = VSUB(TJ, TK); TO = VSUB(TM, TN); TP = VBYI(VFNMS(LDK(KP951056516), TO, VMUL(LDK(KP587785252), TL))); TY = VBYI(VFMA(LDK(KP951056516), TL, VMUL(LDK(KP587785252), TO))); ST(&(x[0]), VADD(TQ, TT), ms, &(x[0])); TX = VADD(TV, TU); ST(&(x[WS(rs, 4)]), VSUB(TX, TY), ms, &(x[0])); ST(&(x[WS(rs, 6)]), VADD(TY, TX), ms, &(x[0])); TW = VSUB(TU, TV); ST(&(x[WS(rs, 2)]), VADD(TP, TW), ms, &(x[0])); ST(&(x[WS(rs, 8)]), VSUB(TW, TP), ms, &(x[0])); } } } } VLEAVE(); }
static void t2bv_10(R *ri, R *ii, const R *W, stride rs, INT mb, INT me, INT ms) { DVK(KP559016994, +0.559016994374947424102293417182819058860154590); DVK(KP250000000, +0.250000000000000000000000000000000000000000000); DVK(KP618033988, +0.618033988749894848204586834365638117720309180); DVK(KP951056516, +0.951056516295153572116439333379382143405698634); { INT m; R *x; x = ii; for (m = mb, W = W + (mb * ((TWVL / VL) * 18)); m < me; m = m + VL, x = x + (VL * ms), W = W + (TWVL * 18), MAKE_VOLATILE_STRIDE(10, rs)) { V Td, TA, T4, Ta, Tk, TE, Tp, TF, TB, T9, T1, T2, Tb; T1 = LD(&(x[0]), ms, &(x[0])); T2 = LD(&(x[WS(rs, 5)]), ms, &(x[WS(rs, 1)])); { V Tg, Tn, Ti, Tl; Tg = LD(&(x[WS(rs, 4)]), ms, &(x[0])); Tn = LD(&(x[WS(rs, 1)]), ms, &(x[WS(rs, 1)])); Ti = LD(&(x[WS(rs, 9)]), ms, &(x[WS(rs, 1)])); Tl = LD(&(x[WS(rs, 6)]), ms, &(x[0])); { V T6, T8, T5, Tc; T5 = LD(&(x[WS(rs, 2)]), ms, &(x[0])); Tc = LD(&(x[WS(rs, 3)]), ms, &(x[WS(rs, 1)])); { V T3, Th, To, Tj, Tm, T7; T7 = LD(&(x[WS(rs, 7)]), ms, &(x[WS(rs, 1)])); T3 = BYTW(&(W[TWVL * 8]), T2); Th = BYTW(&(W[TWVL * 6]), Tg); To = BYTW(&(W[0]), Tn); Tj = BYTW(&(W[TWVL * 16]), Ti); Tm = BYTW(&(W[TWVL * 10]), Tl); T6 = BYTW(&(W[TWVL * 2]), T5); Td = BYTW(&(W[TWVL * 4]), Tc); T8 = BYTW(&(W[TWVL * 12]), T7); TA = VADD(T1, T3); T4 = VSUB(T1, T3); Ta = LD(&(x[WS(rs, 8)]), ms, &(x[0])); Tk = VSUB(Th, Tj); TE = VADD(Th, Tj); Tp = VSUB(Tm, To); TF = VADD(Tm, To); } TB = VADD(T6, T8); T9 = VSUB(T6, T8); } } Tb = BYTW(&(W[TWVL * 14]), Ta); { V TL, TG, Tw, Tq, TC, Te; TL = VSUB(TE, TF); TG = VADD(TE, TF); Tw = VSUB(Tk, Tp); Tq = VADD(Tk, Tp); TC = VADD(Tb, Td); Te = VSUB(Tb, Td); { V TM, TD, Tv, Tf; TM = VSUB(TB, TC); TD = VADD(TB, TC); Tv = VSUB(T9, Te); Tf = VADD(T9, Te); { V TP, TN, TH, TJ, Tz, Tx, Tr, Tt, TI, Ts; TP = VMUL(LDK(KP951056516), VFMA(LDK(KP618033988), TL, TM)); TN = VMUL(LDK(KP951056516), VFNMS(LDK(KP618033988), TM, TL)); TH = VADD(TD, TG); TJ = VSUB(TD, TG); Tz = VMUL(LDK(KP951056516), VFNMS(LDK(KP618033988), Tv, Tw)); Tx = VMUL(LDK(KP951056516), VFMA(LDK(KP618033988), Tw, Tv)); Tr = VADD(Tf, Tq); Tt = VSUB(Tf, Tq); ST(&(x[0]), VADD(TA, TH), ms, &(x[0])); TI = VFNMS(LDK(KP250000000), TH, TA); ST(&(x[WS(rs, 5)]), VADD(T4, Tr), ms, &(x[WS(rs, 1)])); Ts = VFNMS(LDK(KP250000000), Tr, T4); { V TK, TO, Tu, Ty; TK = VFNMS(LDK(KP559016994), TJ, TI); TO = VFMA(LDK(KP559016994), TJ, TI); Tu = VFMA(LDK(KP559016994), Tt, Ts); Ty = VFNMS(LDK(KP559016994), Tt, Ts); ST(&(x[WS(rs, 8)]), VFMAI(TN, TK), ms, &(x[0])); ST(&(x[WS(rs, 2)]), VFNMSI(TN, TK), ms, &(x[0])); ST(&(x[WS(rs, 6)]), VFMAI(TP, TO), ms, &(x[0])); ST(&(x[WS(rs, 4)]), VFNMSI(TP, TO), ms, &(x[0])); ST(&(x[WS(rs, 9)]), VFNMSI(Tx, Tu), ms, &(x[WS(rs, 1)])); ST(&(x[WS(rs, 1)]), VFMAI(Tx, Tu), ms, &(x[WS(rs, 1)])); ST(&(x[WS(rs, 7)]), VFNMSI(Tz, Ty), ms, &(x[WS(rs, 1)])); ST(&(x[WS(rs, 3)]), VFMAI(Tz, Ty), ms, &(x[WS(rs, 1)])); } } } } } } VLEAVE(); }
static void t2bv_10(R *ri, R *ii, const R *W, stride rs, INT mb, INT me, INT ms) { DVK(KP587785252, +0.587785252292473129168705954639072768597652438); DVK(KP951056516, +0.951056516295153572116439333379382143405698634); DVK(KP250000000, +0.250000000000000000000000000000000000000000000); DVK(KP559016994, +0.559016994374947424102293417182819058860154590); { INT m; R *x; x = ii; for (m = mb, W = W + (mb * ((TWVL / VL) * 18)); m < me; m = m + VL, x = x + (VL * ms), W = W + (TWVL * 18), MAKE_VOLATILE_STRIDE(10, rs)) { V Tu, TH, Tg, Tl, Tp, TD, TE, TJ, T5, Ta, To, TA, TB, TI, Tr; V Tt, Ts; Tr = LD(&(x[0]), ms, &(x[0])); Ts = LD(&(x[WS(rs, 5)]), ms, &(x[WS(rs, 1)])); Tt = BYTW(&(W[TWVL * 8]), Ts); Tu = VSUB(Tr, Tt); TH = VADD(Tr, Tt); { V Td, Tk, Tf, Ti; { V Tc, Tj, Te, Th; Tc = LD(&(x[WS(rs, 4)]), ms, &(x[0])); Td = BYTW(&(W[TWVL * 6]), Tc); Tj = LD(&(x[WS(rs, 1)]), ms, &(x[WS(rs, 1)])); Tk = BYTW(&(W[0]), Tj); Te = LD(&(x[WS(rs, 9)]), ms, &(x[WS(rs, 1)])); Tf = BYTW(&(W[TWVL * 16]), Te); Th = LD(&(x[WS(rs, 6)]), ms, &(x[0])); Ti = BYTW(&(W[TWVL * 10]), Th); } Tg = VSUB(Td, Tf); Tl = VSUB(Ti, Tk); Tp = VADD(Tg, Tl); TD = VADD(Td, Tf); TE = VADD(Ti, Tk); TJ = VADD(TD, TE); } { V T2, T9, T4, T7; { V T1, T8, T3, T6; T1 = LD(&(x[WS(rs, 2)]), ms, &(x[0])); T2 = BYTW(&(W[TWVL * 2]), T1); T8 = LD(&(x[WS(rs, 3)]), ms, &(x[WS(rs, 1)])); T9 = BYTW(&(W[TWVL * 4]), T8); T3 = LD(&(x[WS(rs, 7)]), ms, &(x[WS(rs, 1)])); T4 = BYTW(&(W[TWVL * 12]), T3); T6 = LD(&(x[WS(rs, 8)]), ms, &(x[0])); T7 = BYTW(&(W[TWVL * 14]), T6); } T5 = VSUB(T2, T4); Ta = VSUB(T7, T9); To = VADD(T5, Ta); TA = VADD(T2, T4); TB = VADD(T7, T9); TI = VADD(TA, TB); } { V Tq, Tv, Tw, Tn, Tz, Tb, Tm, Ty, Tx; Tq = VMUL(LDK(KP559016994), VSUB(To, Tp)); Tv = VADD(To, Tp); Tw = VFNMS(LDK(KP250000000), Tv, Tu); Tb = VSUB(T5, Ta); Tm = VSUB(Tg, Tl); Tn = VBYI(VFMA(LDK(KP951056516), Tb, VMUL(LDK(KP587785252), Tm))); Tz = VBYI(VFNMS(LDK(KP951056516), Tm, VMUL(LDK(KP587785252), Tb))); ST(&(x[WS(rs, 5)]), VADD(Tu, Tv), ms, &(x[WS(rs, 1)])); Ty = VSUB(Tw, Tq); ST(&(x[WS(rs, 3)]), VSUB(Ty, Tz), ms, &(x[WS(rs, 1)])); ST(&(x[WS(rs, 7)]), VADD(Tz, Ty), ms, &(x[WS(rs, 1)])); Tx = VADD(Tq, Tw); ST(&(x[WS(rs, 1)]), VADD(Tn, Tx), ms, &(x[WS(rs, 1)])); ST(&(x[WS(rs, 9)]), VSUB(Tx, Tn), ms, &(x[WS(rs, 1)])); } { V TM, TK, TL, TG, TP, TC, TF, TO, TN; TM = VMUL(LDK(KP559016994), VSUB(TI, TJ)); TK = VADD(TI, TJ); TL = VFNMS(LDK(KP250000000), TK, TH); TC = VSUB(TA, TB); TF = VSUB(TD, TE); TG = VBYI(VFNMS(LDK(KP951056516), TF, VMUL(LDK(KP587785252), TC))); TP = VBYI(VFMA(LDK(KP951056516), TC, VMUL(LDK(KP587785252), TF))); ST(&(x[0]), VADD(TH, TK), ms, &(x[0])); TO = VADD(TM, TL); ST(&(x[WS(rs, 4)]), VSUB(TO, TP), ms, &(x[0])); ST(&(x[WS(rs, 6)]), VADD(TP, TO), ms, &(x[0])); TN = VSUB(TL, TM); ST(&(x[WS(rs, 2)]), VADD(TG, TN), ms, &(x[0])); ST(&(x[WS(rs, 8)]), VSUB(TN, TG), ms, &(x[0])); } } } VLEAVE(); }
static void t1fuv_9(R *ri, R *ii, const R *W, stride rs, INT mb, INT me, INT ms) { DVK(KP939692620, +0.939692620785908384054109277324731469936208134); DVK(KP826351822, +0.826351822333069651148283373230685203999624323); DVK(KP879385241, +0.879385241571816768108218554649462939872416269); DVK(KP984807753, +0.984807753012208059366743024589523013670643252); DVK(KP666666666, +0.666666666666666666666666666666666666666666667); DVK(KP852868531, +0.852868531952443209628250963940074071936020296); DVK(KP907603734, +0.907603734547952313649323976213898122064543220); DVK(KP420276625, +0.420276625461206169731530603237061658838781920); DVK(KP673648177, +0.673648177666930348851716626769314796000375677); DVK(KP898197570, +0.898197570222573798468955502359086394667167570); DVK(KP347296355, +0.347296355333860697703433253538629592000751354); DVK(KP866025403, +0.866025403784438646763723170752936183471402627); DVK(KP439692620, +0.439692620785908384054109277324731469936208134); DVK(KP203604859, +0.203604859554852403062088995281827210665664861); DVK(KP152703644, +0.152703644666139302296566746461370407999248646); DVK(KP586256827, +0.586256827714544512072145703099641959914944179); DVK(KP968908795, +0.968908795874236621082202410917456709164223497); DVK(KP726681596, +0.726681596905677465811651808188092531873167623); DVK(KP500000000, +0.500000000000000000000000000000000000000000000); { INT m; R *x; x = ri; for (m = mb, W = W + (mb * ((TWVL / VL) * 16)); m < me; m = m + VL, x = x + (VL * ms), W = W + (TWVL * 16), MAKE_VOLATILE_STRIDE(rs)) { V T1, T3, T5, T9, Th, Tb, Td, Tj, Tl, TD, T6; T1 = LD(&(x[0]), ms, &(x[0])); { V T2, T4, T8, Tg; T2 = LD(&(x[WS(rs, 3)]), ms, &(x[WS(rs, 1)])); T4 = LD(&(x[WS(rs, 6)]), ms, &(x[0])); T8 = LD(&(x[WS(rs, 1)]), ms, &(x[WS(rs, 1)])); Tg = LD(&(x[WS(rs, 2)]), ms, &(x[0])); { V Ta, Tc, Ti, Tk; Ta = LD(&(x[WS(rs, 4)]), ms, &(x[0])); Tc = LD(&(x[WS(rs, 7)]), ms, &(x[WS(rs, 1)])); Ti = LD(&(x[WS(rs, 5)]), ms, &(x[WS(rs, 1)])); Tk = LD(&(x[WS(rs, 8)]), ms, &(x[0])); T3 = BYTWJ(&(W[TWVL * 4]), T2); T5 = BYTWJ(&(W[TWVL * 10]), T4); T9 = BYTWJ(&(W[0]), T8); Th = BYTWJ(&(W[TWVL * 2]), Tg); Tb = BYTWJ(&(W[TWVL * 6]), Ta); Td = BYTWJ(&(W[TWVL * 12]), Tc); Tj = BYTWJ(&(W[TWVL * 8]), Ti); Tl = BYTWJ(&(W[TWVL * 14]), Tk); } } TD = VSUB(T5, T3); T6 = VADD(T3, T5); { V Tt, Te, Tu, Tm, Tr, T7; Tt = VSUB(Tb, Td); Te = VADD(Tb, Td); Tu = VSUB(Tl, Tj); Tm = VADD(Tj, Tl); Tr = VFNMS(LDK(KP500000000), T6, T1); T7 = VADD(T1, T6); { V Tv, Tf, Ts, Tn; Tv = VFNMS(LDK(KP500000000), Te, T9); Tf = VADD(T9, Te); Ts = VFNMS(LDK(KP500000000), Tm, Th); Tn = VADD(Th, Tm); { V TG, TK, Tw, TJ, TF, TA, To, Tq; TG = VFNMS(LDK(KP726681596), Tt, Tv); TK = VFMA(LDK(KP968908795), Tv, Tt); Tw = VFNMS(LDK(KP586256827), Tv, Tu); TJ = VFNMS(LDK(KP152703644), Tu, Ts); TF = VFMA(LDK(KP203604859), Ts, Tu); TA = VFNMS(LDK(KP439692620), Tt, Ts); To = VADD(Tf, Tn); Tq = VMUL(LDK(KP866025403), VSUB(Tn, Tf)); { V TQ, TH, TL, TN, TB, Tp, Ty, TI, Tx; Tx = VFNMS(LDK(KP347296355), Tw, Tt); TQ = VFNMS(LDK(KP898197570), TG, TF); TH = VFMA(LDK(KP898197570), TG, TF); TL = VFMA(LDK(KP673648177), TK, TJ); TN = VFNMS(LDK(KP673648177), TK, TJ); TB = VFNMS(LDK(KP420276625), TA, Tu); ST(&(x[0]), VADD(T7, To), ms, &(x[0])); Tp = VFNMS(LDK(KP500000000), To, T7); Ty = VFNMS(LDK(KP907603734), Tx, Ts); TI = VFMA(LDK(KP852868531), TH, Tr); { V TO, TR, TM, TC, Tz, TP, TS, TE; TO = VFNMS(LDK(KP500000000), TH, TN); TR = VFMA(LDK(KP666666666), TL, TQ); TM = VMUL(LDK(KP984807753), VFNMS(LDK(KP879385241), TD, TL)); TC = VFNMS(LDK(KP826351822), TB, Tv); ST(&(x[WS(rs, 6)]), VFNMSI(Tq, Tp), ms, &(x[0])); ST(&(x[WS(rs, 3)]), VFMAI(Tq, Tp), ms, &(x[WS(rs, 1)])); Tz = VFNMS(LDK(KP939692620), Ty, Tr); TP = VFMA(LDK(KP852868531), TO, Tr); TS = VMUL(LDK(KP866025403), VFMA(LDK(KP852868531), TR, TD)); ST(&(x[WS(rs, 8)]), VFMAI(TM, TI), ms, &(x[0])); ST(&(x[WS(rs, 1)]), VFNMSI(TM, TI), ms, &(x[WS(rs, 1)])); TE = VMUL(LDK(KP984807753), VFMA(LDK(KP879385241), TD, TC)); ST(&(x[WS(rs, 4)]), VFMAI(TS, TP), ms, &(x[0])); ST(&(x[WS(rs, 5)]), VFNMSI(TS, TP), ms, &(x[WS(rs, 1)])); ST(&(x[WS(rs, 7)]), VFMAI(TE, Tz), ms, &(x[WS(rs, 1)])); ST(&(x[WS(rs, 2)]), VFNMSI(TE, Tz), ms, &(x[0])); } } } } } } } VLEAVE(); }
static void t1fuv_9(R *ri, R *ii, const R *W, stride rs, INT mb, INT me, INT ms) { DVK(KP939692620, +0.939692620785908384054109277324731469936208134); DVK(KP296198132, +0.296198132726023843175338011893050938967728390); DVK(KP852868531, +0.852868531952443209628250963940074071936020296); DVK(KP173648177, +0.173648177666930348851716626769314796000375677); DVK(KP556670399, +0.556670399226419366452912952047023132968291906); DVK(KP766044443, +0.766044443118978035202392650555416673935832457); DVK(KP642787609, +0.642787609686539326322643409907263432907559884); DVK(KP663413948, +0.663413948168938396205421319635891297216863310); DVK(KP984807753, +0.984807753012208059366743024589523013670643252); DVK(KP150383733, +0.150383733180435296639271897612501926072238258); DVK(KP342020143, +0.342020143325668733044099614682259580763083368); DVK(KP813797681, +0.813797681349373692844693217248393223289101568); DVK(KP500000000, +0.500000000000000000000000000000000000000000000); DVK(KP866025403, +0.866025403784438646763723170752936183471402627); { INT m; R *x; x = ri; for (m = mb, W = W + (mb * ((TWVL / VL) * 16)); m < me; m = m + VL, x = x + (VL * ms), W = W + (TWVL * 16), MAKE_VOLATILE_STRIDE(rs)) { V T1, T6, TA, Tt, Tf, Ts, Tw, Tn, Tv; T1 = LD(&(x[0]), ms, &(x[0])); { V T3, T5, T2, T4; T2 = LD(&(x[WS(rs, 3)]), ms, &(x[WS(rs, 1)])); T3 = BYTWJ(&(W[TWVL * 4]), T2); T4 = LD(&(x[WS(rs, 6)]), ms, &(x[0])); T5 = BYTWJ(&(W[TWVL * 10]), T4); T6 = VADD(T3, T5); TA = VMUL(LDK(KP866025403), VSUB(T5, T3)); } { V T9, Td, Tb, T8, Tc, Ta, Te; T8 = LD(&(x[WS(rs, 1)]), ms, &(x[WS(rs, 1)])); T9 = BYTWJ(&(W[0]), T8); Tc = LD(&(x[WS(rs, 7)]), ms, &(x[WS(rs, 1)])); Td = BYTWJ(&(W[TWVL * 12]), Tc); Ta = LD(&(x[WS(rs, 4)]), ms, &(x[0])); Tb = BYTWJ(&(W[TWVL * 6]), Ta); Tt = VSUB(Td, Tb); Te = VADD(Tb, Td); Tf = VADD(T9, Te); Ts = VFNMS(LDK(KP500000000), Te, T9); } { V Th, Tl, Tj, Tg, Tk, Ti, Tm; Tg = LD(&(x[WS(rs, 2)]), ms, &(x[0])); Th = BYTWJ(&(W[TWVL * 2]), Tg); Tk = LD(&(x[WS(rs, 8)]), ms, &(x[0])); Tl = BYTWJ(&(W[TWVL * 14]), Tk); Ti = LD(&(x[WS(rs, 5)]), ms, &(x[WS(rs, 1)])); Tj = BYTWJ(&(W[TWVL * 8]), Ti); Tw = VSUB(Tl, Tj); Tm = VADD(Tj, Tl); Tn = VADD(Th, Tm); Tv = VFNMS(LDK(KP500000000), Tm, Th); } { V Tq, T7, To, Tp; Tq = VBYI(VMUL(LDK(KP866025403), VSUB(Tn, Tf))); T7 = VADD(T1, T6); To = VADD(Tf, Tn); Tp = VFNMS(LDK(KP500000000), To, T7); ST(&(x[0]), VADD(T7, To), ms, &(x[0])); ST(&(x[WS(rs, 3)]), VADD(Tp, Tq), ms, &(x[WS(rs, 1)])); ST(&(x[WS(rs, 6)]), VSUB(Tp, Tq), ms, &(x[0])); } { V TI, TB, TC, TD, Tu, Tx, Ty, Tr, TH; TI = VBYI(VSUB(VFNMS(LDK(KP342020143), Tv, VFNMS(LDK(KP150383733), Tt, VFNMS(LDK(KP984807753), Ts, VMUL(LDK(KP813797681), Tw)))), TA)); TB = VFNMS(LDK(KP642787609), Ts, VMUL(LDK(KP663413948), Tt)); TC = VFNMS(LDK(KP984807753), Tv, VMUL(LDK(KP150383733), Tw)); TD = VADD(TB, TC); Tu = VFMA(LDK(KP766044443), Ts, VMUL(LDK(KP556670399), Tt)); Tx = VFMA(LDK(KP173648177), Tv, VMUL(LDK(KP852868531), Tw)); Ty = VADD(Tu, Tx); Tr = VFNMS(LDK(KP500000000), T6, T1); TH = VFMA(LDK(KP173648177), Ts, VFNMS(LDK(KP296198132), Tw, VFNMS(LDK(KP939692620), Tv, VFNMS(LDK(KP852868531), Tt, Tr)))); ST(&(x[WS(rs, 7)]), VSUB(TH, TI), ms, &(x[WS(rs, 1)])); ST(&(x[WS(rs, 2)]), VADD(TH, TI), ms, &(x[0])); { V Tz, TE, TF, TG; Tz = VADD(Tr, Ty); TE = VBYI(VADD(TA, TD)); ST(&(x[WS(rs, 8)]), VSUB(Tz, TE), ms, &(x[0])); ST(&(x[WS(rs, 1)]), VADD(TE, Tz), ms, &(x[WS(rs, 1)])); TF = VFMA(LDK(KP866025403), VSUB(TB, TC), VFNMS(LDK(KP500000000), Ty, Tr)); TG = VBYI(VADD(TA, VFNMS(LDK(KP500000000), TD, VMUL(LDK(KP866025403), VSUB(Tx, Tu))))); ST(&(x[WS(rs, 5)]), VSUB(TF, TG), ms, &(x[WS(rs, 1)])); ST(&(x[WS(rs, 4)]), VADD(TF, TG), ms, &(x[0])); } } } } VLEAVE(); }
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 t1fuv_7(R *ri, R *ii, const R *W, stride rs, INT mb, INT me, INT ms) { DVK(KP900968867, +0.900968867902419126236102319507445051165919162); DVK(KP801937735, +0.801937735804838252472204639014890102331838324); DVK(KP974927912, +0.974927912181823607018131682993931217232785801); DVK(KP692021471, +0.692021471630095869627814897002069140197260599); DVK(KP554958132, +0.554958132087371191422194871006410481067288862); DVK(KP356895867, +0.356895867892209443894399510021300583399127187); { INT m; R *x; x = ri; for (m = mb, W = W + (mb * ((TWVL / VL) * 12)); m < me; m = m + VL, x = x + (VL * ms), W = W + (TWVL * 12), MAKE_VOLATILE_STRIDE(7, rs)) { V T1, T2, T4, Te, Tc, T9, T7; T1 = LD(&(x[0]), ms, &(x[0])); T2 = LD(&(x[WS(rs, 1)]), ms, &(x[WS(rs, 1)])); T4 = LD(&(x[WS(rs, 6)]), ms, &(x[0])); Te = LD(&(x[WS(rs, 4)]), ms, &(x[0])); Tc = LD(&(x[WS(rs, 3)]), ms, &(x[WS(rs, 1)])); T9 = LD(&(x[WS(rs, 5)]), ms, &(x[WS(rs, 1)])); T7 = LD(&(x[WS(rs, 2)]), ms, &(x[0])); { V T3, T5, Tf, Td, Ta, T8; T3 = BYTWJ(&(W[0]), T2); T5 = BYTWJ(&(W[TWVL * 10]), T4); Tf = BYTWJ(&(W[TWVL * 6]), Te); Td = BYTWJ(&(W[TWVL * 4]), Tc); Ta = BYTWJ(&(W[TWVL * 8]), T9); T8 = BYTWJ(&(W[TWVL * 2]), T7); { V T6, Tk, Tg, Tl, Tb, Tm; T6 = VADD(T3, T5); Tk = VSUB(T5, T3); Tg = VADD(Td, Tf); Tl = VSUB(Tf, Td); Tb = VADD(T8, Ta); Tm = VSUB(Ta, T8); { V Th, Ts, Tp, Tu, Tn, Tx, Ti, Tt; Th = VFNMS(LDK(KP356895867), T6, Tg); Ts = VFMA(LDK(KP554958132), Tl, Tk); ST(&(x[0]), VADD(T1, VADD(T6, VADD(Tb, Tg))), ms, &(x[0])); Tp = VFNMS(LDK(KP356895867), Tb, T6); Tu = VFNMS(LDK(KP356895867), Tg, Tb); Tn = VFMA(LDK(KP554958132), Tm, Tl); Tx = VFNMS(LDK(KP554958132), Tk, Tm); Ti = VFNMS(LDK(KP692021471), Th, Tb); Tt = VMUL(LDK(KP974927912), VFMA(LDK(KP801937735), Ts, Tm)); { V Tq, Tv, To, Ty, Tj, Tr, Tw; Tq = VFNMS(LDK(KP692021471), Tp, Tg); Tv = VFNMS(LDK(KP692021471), Tu, T6); To = VMUL(LDK(KP974927912), VFNMS(LDK(KP801937735), Tn, Tk)); Ty = VMUL(LDK(KP974927912), VFNMS(LDK(KP801937735), Tx, Tl)); Tj = VFNMS(LDK(KP900968867), Ti, T1); Tr = VFNMS(LDK(KP900968867), Tq, T1); Tw = VFNMS(LDK(KP900968867), Tv, T1); ST(&(x[WS(rs, 2)]), VFMAI(To, Tj), ms, &(x[0])); ST(&(x[WS(rs, 5)]), VFNMSI(To, Tj), ms, &(x[WS(rs, 1)])); ST(&(x[WS(rs, 1)]), VFMAI(Tt, Tr), ms, &(x[WS(rs, 1)])); ST(&(x[WS(rs, 6)]), VFNMSI(Tt, Tr), ms, &(x[0])); ST(&(x[WS(rs, 3)]), VFMAI(Ty, Tw), ms, &(x[WS(rs, 1)])); ST(&(x[WS(rs, 4)]), VFNMSI(Ty, Tw), ms, &(x[0])); } } } } } } VLEAVE(); }
static void t1fv_12(R *ri, R *ii, const R *W, stride rs, INT mb, INT me, INT ms) { DVK(KP866025403, +0.866025403784438646763723170752936183471402627); DVK(KP500000000, +0.500000000000000000000000000000000000000000000); { INT m; R *x; x = ri; for (m = mb, W = W + (mb * ((TWVL / VL) * 22)); m < me; m = m + VL, x = x + (VL * ms), W = W + (TWVL * 22), MAKE_VOLATILE_STRIDE(12, rs)) { V T1, TC, T6, T7, Ty, Tq, Tz, TA, T9, TD, Te, Tf, Tu, Tl, Tv; V Tw; { V T5, T3, T4, T2; T1 = LD(&(x[0]), ms, &(x[0])); T4 = LD(&(x[WS(rs, 8)]), ms, &(x[0])); T5 = BYTWJ(&(W[TWVL * 14]), T4); T2 = LD(&(x[WS(rs, 4)]), ms, &(x[0])); T3 = BYTWJ(&(W[TWVL * 6]), T2); TC = VSUB(T5, T3); T6 = VADD(T3, T5); T7 = VFNMS(LDK(KP500000000), T6, T1); } { V Tn, Tp, Tm, Tx, To; Tm = LD(&(x[WS(rs, 1)]), ms, &(x[WS(rs, 1)])); Tn = BYTWJ(&(W[0]), Tm); Tx = LD(&(x[WS(rs, 9)]), ms, &(x[WS(rs, 1)])); Ty = BYTWJ(&(W[TWVL * 16]), Tx); To = LD(&(x[WS(rs, 5)]), ms, &(x[WS(rs, 1)])); Tp = BYTWJ(&(W[TWVL * 8]), To); Tq = VSUB(Tn, Tp); Tz = VADD(Tn, Tp); TA = VFNMS(LDK(KP500000000), Tz, Ty); } { V Td, Tb, T8, Tc, Ta; T8 = LD(&(x[WS(rs, 6)]), ms, &(x[0])); T9 = BYTWJ(&(W[TWVL * 10]), T8); Tc = LD(&(x[WS(rs, 2)]), ms, &(x[0])); Td = BYTWJ(&(W[TWVL * 2]), Tc); Ta = LD(&(x[WS(rs, 10)]), ms, &(x[0])); Tb = BYTWJ(&(W[TWVL * 18]), Ta); TD = VSUB(Td, Tb); Te = VADD(Tb, Td); Tf = VFNMS(LDK(KP500000000), Te, T9); } { V Ti, Tk, Th, Tt, Tj; Th = LD(&(x[WS(rs, 11)]), ms, &(x[WS(rs, 1)])); Ti = BYTWJ(&(W[TWVL * 20]), Th); Tt = LD(&(x[WS(rs, 3)]), ms, &(x[WS(rs, 1)])); Tu = BYTWJ(&(W[TWVL * 4]), Tt); Tj = LD(&(x[WS(rs, 7)]), ms, &(x[WS(rs, 1)])); Tk = BYTWJ(&(W[TWVL * 12]), Tj); Tl = VSUB(Ti, Tk); Tv = VADD(Tk, Ti); Tw = VFNMS(LDK(KP500000000), Tv, Tu); } { V Ts, TG, TF, TH; { V Tg, Tr, TB, TE; Tg = VSUB(T7, Tf); Tr = VADD(Tl, Tq); Ts = VFMA(LDK(KP866025403), Tr, Tg); TG = VFNMS(LDK(KP866025403), Tr, Tg); TB = VSUB(Tw, TA); TE = VSUB(TC, TD); TF = VFNMS(LDK(KP866025403), TE, TB); TH = VFMA(LDK(KP866025403), TE, TB); } ST(&(x[WS(rs, 1)]), VFNMSI(TF, Ts), ms, &(x[WS(rs, 1)])); ST(&(x[WS(rs, 7)]), VFMAI(TH, TG), ms, &(x[WS(rs, 1)])); ST(&(x[WS(rs, 11)]), VFMAI(TF, Ts), ms, &(x[WS(rs, 1)])); ST(&(x[WS(rs, 5)]), VFNMSI(TH, TG), ms, &(x[WS(rs, 1)])); } { V TS, TW, TV, TX; { V TQ, TR, TT, TU; TQ = VADD(T1, T6); TR = VADD(T9, Te); TS = VSUB(TQ, TR); TW = VADD(TQ, TR); TT = VADD(Tu, Tv); TU = VADD(Ty, Tz); TV = VSUB(TT, TU); TX = VADD(TT, TU); } ST(&(x[WS(rs, 9)]), VFNMSI(TV, TS), ms, &(x[WS(rs, 1)])); ST(&(x[0]), VADD(TW, TX), ms, &(x[0])); ST(&(x[WS(rs, 3)]), VFMAI(TV, TS), ms, &(x[WS(rs, 1)])); ST(&(x[WS(rs, 6)]), VSUB(TW, TX), ms, &(x[0])); } { V TK, TO, TN, TP; { V TI, TJ, TL, TM; TI = VADD(T7, Tf); TJ = VADD(Tw, TA); TK = VSUB(TI, TJ); TO = VADD(TI, TJ); TL = VSUB(Tl, Tq); TM = VADD(TC, TD); TN = VMUL(LDK(KP866025403), VSUB(TL, TM)); TP = VMUL(LDK(KP866025403), VADD(TM, TL)); } ST(&(x[WS(rs, 2)]), VFMAI(TN, TK), ms, &(x[0])); ST(&(x[WS(rs, 8)]), VFNMSI(TP, TO), ms, &(x[0])); ST(&(x[WS(rs, 10)]), VFNMSI(TN, TK), ms, &(x[0])); ST(&(x[WS(rs, 4)]), VFMAI(TP, TO), ms, &(x[0])); } } } VLEAVE(); }
static void n1bv_15(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs) { DVK(KP216506350, +0.216506350946109661690930792688234045867850657); DVK(KP509036960, +0.509036960455127183450980863393907648510733164); DVK(KP823639103, +0.823639103546331925877420039278190003029660514); DVK(KP951056516, +0.951056516295153572116439333379382143405698634); DVK(KP587785252, +0.587785252292473129168705954639072768597652438); DVK(KP250000000, +0.250000000000000000000000000000000000000000000); DVK(KP559016994, +0.559016994374947424102293417182819058860154590); DVK(KP866025403, +0.866025403784438646763723170752936183471402627); DVK(KP484122918, +0.484122918275927110647408174972799951354115213); DVK(KP500000000, +0.500000000000000000000000000000000000000000000); { INT i; const R *xi; R *xo; xi = ii; xo = io; for (i = v; i > 0; i = i - VL, xi = xi + (VL * ivs), xo = xo + (VL * ovs), MAKE_VOLATILE_STRIDE(is), MAKE_VOLATILE_STRIDE(os)) { V Ti, T11, TH, Ts, TL, TM, Tz, TC, TD, TI, T12, T13, T14, T15, T16; V T17, Tf, Tj, TZ, T10; { V TF, Tg, Th, TG; TF = LD(&(xi[0]), ivs, &(xi[0])); Tg = LD(&(xi[WS(is, 5)]), ivs, &(xi[WS(is, 1)])); Th = LD(&(xi[WS(is, 10)]), ivs, &(xi[0])); TG = VADD(Tg, Th); Ti = VSUB(Tg, Th); T11 = VADD(TF, TG); TH = VFNMS(LDK(KP500000000), TG, TF); } { V Tm, Tn, T3, To, Tw, Tx, Td, Ty, Tp, Tq, T6, Tr, Tt, Tu, Ta; V Tv, T7, Te; { V T1, T2, Tb, Tc; Tm = LD(&(xi[WS(is, 3)]), ivs, &(xi[WS(is, 1)])); T1 = LD(&(xi[WS(is, 8)]), ivs, &(xi[0])); T2 = LD(&(xi[WS(is, 13)]), ivs, &(xi[WS(is, 1)])); Tn = VADD(T1, T2); T3 = VSUB(T1, T2); To = VFNMS(LDK(KP500000000), Tn, Tm); Tw = LD(&(xi[WS(is, 9)]), ivs, &(xi[WS(is, 1)])); Tb = LD(&(xi[WS(is, 14)]), ivs, &(xi[0])); Tc = LD(&(xi[WS(is, 4)]), ivs, &(xi[0])); Tx = VADD(Tb, Tc); Td = VSUB(Tb, Tc); Ty = VFNMS(LDK(KP500000000), Tx, Tw); } { V T4, T5, T8, T9; Tp = LD(&(xi[WS(is, 12)]), ivs, &(xi[0])); T4 = LD(&(xi[WS(is, 2)]), ivs, &(xi[0])); T5 = LD(&(xi[WS(is, 7)]), ivs, &(xi[WS(is, 1)])); Tq = VADD(T4, T5); T6 = VSUB(T4, T5); Tr = VFNMS(LDK(KP500000000), Tq, Tp); Tt = LD(&(xi[WS(is, 6)]), ivs, &(xi[0])); T8 = LD(&(xi[WS(is, 11)]), ivs, &(xi[WS(is, 1)])); T9 = LD(&(xi[WS(is, 1)]), ivs, &(xi[WS(is, 1)])); Tu = VADD(T8, T9); Ta = VSUB(T8, T9); Tv = VFNMS(LDK(KP500000000), Tu, Tt); } Ts = VSUB(To, Tr); TL = VSUB(T3, T6); TM = VSUB(Ta, Td); Tz = VSUB(Tv, Ty); TC = VADD(To, Tr); TD = VADD(Tv, Ty); TI = VADD(TC, TD); T12 = VADD(Tm, Tn); T13 = VADD(Tp, Tq); T14 = VADD(T12, T13); T15 = VADD(Tt, Tu); T16 = VADD(Tw, Tx); T17 = VADD(T15, T16); T7 = VADD(T3, T6); Te = VADD(Ta, Td); Tf = VMUL(LDK(KP484122918), VSUB(T7, Te)); Tj = VADD(T7, Te); } TZ = VADD(TH, TI); T10 = VBYI(VMUL(LDK(KP866025403), VADD(Ti, Tj))); ST(&(xo[WS(os, 5)]), VSUB(TZ, T10), ovs, &(xo[WS(os, 1)])); ST(&(xo[WS(os, 10)]), VADD(T10, TZ), ovs, &(xo[0])); { V T1a, T18, T19, T1e, T1f, T1c, T1d, T1g, T1b; T1a = VMUL(LDK(KP559016994), VSUB(T14, T17)); T18 = VADD(T14, T17); T19 = VFNMS(LDK(KP250000000), T18, T11); T1c = VSUB(T12, T13); T1d = VSUB(T15, T16); T1e = VBYI(VFNMS(LDK(KP951056516), T1d, VMUL(LDK(KP587785252), T1c))); T1f = VBYI(VFMA(LDK(KP951056516), T1c, VMUL(LDK(KP587785252), T1d))); ST(&(xo[0]), VADD(T11, T18), ovs, &(xo[0])); T1g = VADD(T1a, T19); ST(&(xo[WS(os, 6)]), VADD(T1f, T1g), ovs, &(xo[0])); ST(&(xo[WS(os, 9)]), VSUB(T1g, T1f), ovs, &(xo[WS(os, 1)])); T1b = VSUB(T19, T1a); ST(&(xo[WS(os, 3)]), VSUB(T1b, T1e), ovs, &(xo[WS(os, 1)])); ST(&(xo[WS(os, 12)]), VADD(T1e, T1b), ovs, &(xo[0])); } { V TA, TN, TU, TS, Tl, TR, TK, TV, Tk, TE, TJ; TA = VFMA(LDK(KP951056516), Ts, VMUL(LDK(KP587785252), Tz)); TN = VFMA(LDK(KP823639103), TL, VMUL(LDK(KP509036960), TM)); TU = VFNMS(LDK(KP823639103), TM, VMUL(LDK(KP509036960), TL)); TS = VFNMS(LDK(KP951056516), Tz, VMUL(LDK(KP587785252), Ts)); Tk = VFNMS(LDK(KP216506350), Tj, VMUL(LDK(KP866025403), Ti)); Tl = VADD(Tf, Tk); TR = VSUB(Tf, Tk); TE = VMUL(LDK(KP559016994), VSUB(TC, TD)); TJ = VFNMS(LDK(KP250000000), TI, TH); TK = VADD(TE, TJ); TV = VSUB(TJ, TE); { V TB, TO, TX, TY; TB = VBYI(VADD(Tl, TA)); TO = VSUB(TK, TN); ST(&(xo[WS(os, 1)]), VADD(TB, TO), ovs, &(xo[WS(os, 1)])); ST(&(xo[WS(os, 14)]), VSUB(TO, TB), ovs, &(xo[0])); TX = VBYI(VSUB(TS, TR)); TY = VSUB(TV, TU); ST(&(xo[WS(os, 7)]), VADD(TX, TY), ovs, &(xo[WS(os, 1)])); ST(&(xo[WS(os, 8)]), VSUB(TY, TX), ovs, &(xo[0])); } { V TP, TQ, TT, TW; TP = VBYI(VSUB(Tl, TA)); TQ = VADD(TN, TK); ST(&(xo[WS(os, 4)]), VADD(TP, TQ), ovs, &(xo[0])); ST(&(xo[WS(os, 11)]), VSUB(TQ, TP), ovs, &(xo[WS(os, 1)])); TT = VBYI(VADD(TR, TS)); TW = VADD(TU, TV); ST(&(xo[WS(os, 2)]), VADD(TT, TW), ovs, &(xo[0])); ST(&(xo[WS(os, 13)]), VSUB(TW, TT), ovs, &(xo[WS(os, 1)])); } } } } VLEAVE(); }
static void n1bv_12(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs) { DVK(KP866025403, +0.866025403784438646763723170752936183471402627); DVK(KP500000000, +0.500000000000000000000000000000000000000000000); { INT i; const R *xi; R *xo; xi = ii; xo = io; for (i = v; i > 0; i = i - VL, xi = xi + (VL * ivs), xo = xo + (VL * ovs), MAKE_VOLATILE_STRIDE(24, is), MAKE_VOLATILE_STRIDE(24, os)) { V T1, T6, Tc, Th, Td, Te, Ti, Tz, T4, TA, T9, Tj, Tf, Tw; { V T2, T3, T7, T8; T1 = LD(&(xi[0]), ivs, &(xi[0])); T6 = LD(&(xi[WS(is, 6)]), ivs, &(xi[0])); T2 = LD(&(xi[WS(is, 4)]), ivs, &(xi[0])); T3 = LD(&(xi[WS(is, 8)]), ivs, &(xi[0])); T7 = LD(&(xi[WS(is, 10)]), ivs, &(xi[0])); T8 = LD(&(xi[WS(is, 2)]), ivs, &(xi[0])); Tc = LD(&(xi[WS(is, 3)]), ivs, &(xi[WS(is, 1)])); Th = LD(&(xi[WS(is, 9)]), ivs, &(xi[WS(is, 1)])); Td = LD(&(xi[WS(is, 7)]), ivs, &(xi[WS(is, 1)])); Te = LD(&(xi[WS(is, 11)]), ivs, &(xi[WS(is, 1)])); Ti = LD(&(xi[WS(is, 1)]), ivs, &(xi[WS(is, 1)])); Tz = VSUB(T2, T3); T4 = VADD(T2, T3); TA = VSUB(T7, T8); T9 = VADD(T7, T8); Tj = LD(&(xi[WS(is, 5)]), ivs, &(xi[WS(is, 1)])); } Tf = VADD(Td, Te); Tw = VSUB(Td, Te); { V T5, Tp, TJ, TB, Ta, Tq, Tk, Tx, Tg, Ts; T5 = VADD(T1, T4); Tp = VFNMS(LDK(KP500000000), T4, T1); TJ = VSUB(Tz, TA); TB = VADD(Tz, TA); Ta = VADD(T6, T9); Tq = VFNMS(LDK(KP500000000), T9, T6); Tk = VADD(Ti, Tj); Tx = VSUB(Tj, Ti); Tg = VADD(Tc, Tf); Ts = VFNMS(LDK(KP500000000), Tf, Tc); { V Tr, TF, Tb, Tn, TG, Ty, Tl, Tt; Tr = VADD(Tp, Tq); TF = VSUB(Tp, Tq); Tb = VSUB(T5, Ta); Tn = VADD(T5, Ta); TG = VADD(Tw, Tx); Ty = VSUB(Tw, Tx); Tl = VADD(Th, Tk); Tt = VFNMS(LDK(KP500000000), Tk, Th); { V TC, TE, TH, TL, Tu, TI, Tm, To; TC = VMUL(LDK(KP866025403), VSUB(Ty, TB)); TE = VMUL(LDK(KP866025403), VADD(TB, Ty)); TH = VFNMS(LDK(KP866025403), TG, TF); TL = VFMA(LDK(KP866025403), TG, TF); Tu = VADD(Ts, Tt); TI = VSUB(Ts, Tt); Tm = VSUB(Tg, Tl); To = VADD(Tg, Tl); { V TK, TM, Tv, TD; TK = VFMA(LDK(KP866025403), TJ, TI); TM = VFNMS(LDK(KP866025403), TJ, TI); Tv = VSUB(Tr, Tu); TD = VADD(Tr, Tu); ST(&(xo[0]), VADD(Tn, To), ovs, &(xo[0])); ST(&(xo[WS(os, 6)]), VSUB(Tn, To), ovs, &(xo[0])); ST(&(xo[WS(os, 9)]), VFMAI(Tm, Tb), ovs, &(xo[WS(os, 1)])); ST(&(xo[WS(os, 3)]), VFNMSI(Tm, Tb), ovs, &(xo[WS(os, 1)])); ST(&(xo[WS(os, 5)]), VFMAI(TM, TL), ovs, &(xo[WS(os, 1)])); ST(&(xo[WS(os, 7)]), VFNMSI(TM, TL), ovs, &(xo[WS(os, 1)])); ST(&(xo[WS(os, 11)]), VFNMSI(TK, TH), ovs, &(xo[WS(os, 1)])); ST(&(xo[WS(os, 1)]), VFMAI(TK, TH), ovs, &(xo[WS(os, 1)])); ST(&(xo[WS(os, 8)]), VFNMSI(TE, TD), ovs, &(xo[0])); ST(&(xo[WS(os, 4)]), VFMAI(TE, TD), ovs, &(xo[0])); ST(&(xo[WS(os, 2)]), VFMAI(TC, Tv), ovs, &(xo[0])); ST(&(xo[WS(os, 10)]), VFNMSI(TC, Tv), ovs, &(xo[0])); } } } } } } VLEAVE(); }
static void n1bv_12(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs) { DVK(KP866025403, +0.866025403784438646763723170752936183471402627); DVK(KP500000000, +0.500000000000000000000000000000000000000000000); { INT i; const R *xi; R *xo; xi = ii; xo = io; for (i = v; i > 0; i = i - VL, xi = xi + (VL * ivs), xo = xo + (VL * ovs), MAKE_VOLATILE_STRIDE(24, is), MAKE_VOLATILE_STRIDE(24, os)) { V T5, Ta, TG, TF, Ty, Tm, Ti, Tp, TJ, TI, Tx, Ts; { V T1, T6, T4, Tk, T9, Tl; T1 = LD(&(xi[0]), ivs, &(xi[0])); T6 = LD(&(xi[WS(is, 6)]), ivs, &(xi[0])); { V T2, T3, T7, T8; T2 = LD(&(xi[WS(is, 4)]), ivs, &(xi[0])); T3 = LD(&(xi[WS(is, 8)]), ivs, &(xi[0])); T4 = VADD(T2, T3); Tk = VSUB(T2, T3); T7 = LD(&(xi[WS(is, 10)]), ivs, &(xi[0])); T8 = LD(&(xi[WS(is, 2)]), ivs, &(xi[0])); T9 = VADD(T7, T8); Tl = VSUB(T7, T8); } T5 = VFNMS(LDK(KP500000000), T4, T1); Ta = VFNMS(LDK(KP500000000), T9, T6); TG = VADD(T6, T9); TF = VADD(T1, T4); Ty = VADD(Tk, Tl); Tm = VMUL(LDK(KP866025403), VSUB(Tk, Tl)); } { V Tn, Tq, Te, To, Th, Tr; Tn = LD(&(xi[WS(is, 3)]), ivs, &(xi[WS(is, 1)])); Tq = LD(&(xi[WS(is, 9)]), ivs, &(xi[WS(is, 1)])); { V Tc, Td, Tf, Tg; Tc = LD(&(xi[WS(is, 7)]), ivs, &(xi[WS(is, 1)])); Td = LD(&(xi[WS(is, 11)]), ivs, &(xi[WS(is, 1)])); Te = VSUB(Tc, Td); To = VADD(Tc, Td); Tf = LD(&(xi[WS(is, 1)]), ivs, &(xi[WS(is, 1)])); Tg = LD(&(xi[WS(is, 5)]), ivs, &(xi[WS(is, 1)])); Th = VSUB(Tf, Tg); Tr = VADD(Tf, Tg); } Ti = VMUL(LDK(KP866025403), VSUB(Te, Th)); Tp = VFNMS(LDK(KP500000000), To, Tn); TJ = VADD(Tq, Tr); TI = VADD(Tn, To); Tx = VADD(Te, Th); Ts = VFNMS(LDK(KP500000000), Tr, Tq); } { V TH, TK, TL, TM; TH = VSUB(TF, TG); TK = VBYI(VSUB(TI, TJ)); ST(&(xo[WS(os, 3)]), VSUB(TH, TK), ovs, &(xo[WS(os, 1)])); ST(&(xo[WS(os, 9)]), VADD(TH, TK), ovs, &(xo[WS(os, 1)])); TL = VADD(TF, TG); TM = VADD(TI, TJ); ST(&(xo[WS(os, 6)]), VSUB(TL, TM), ovs, &(xo[0])); ST(&(xo[0]), VADD(TL, TM), ovs, &(xo[0])); } { V Tj, Tv, Tu, Tw, Tb, Tt; Tb = VSUB(T5, Ta); Tj = VSUB(Tb, Ti); Tv = VADD(Tb, Ti); Tt = VSUB(Tp, Ts); Tu = VBYI(VADD(Tm, Tt)); Tw = VBYI(VSUB(Tt, Tm)); ST(&(xo[WS(os, 11)]), VSUB(Tj, Tu), ovs, &(xo[WS(os, 1)])); ST(&(xo[WS(os, 5)]), VADD(Tv, Tw), ovs, &(xo[WS(os, 1)])); ST(&(xo[WS(os, 1)]), VADD(Tj, Tu), ovs, &(xo[WS(os, 1)])); ST(&(xo[WS(os, 7)]), VSUB(Tv, Tw), ovs, &(xo[WS(os, 1)])); } { V Tz, TD, TC, TE, TA, TB; Tz = VBYI(VMUL(LDK(KP866025403), VSUB(Tx, Ty))); TD = VBYI(VMUL(LDK(KP866025403), VADD(Ty, Tx))); TA = VADD(T5, Ta); TB = VADD(Tp, Ts); TC = VSUB(TA, TB); TE = VADD(TA, TB); ST(&(xo[WS(os, 2)]), VADD(Tz, TC), ovs, &(xo[0])); ST(&(xo[WS(os, 8)]), VSUB(TE, TD), ovs, &(xo[0])); ST(&(xo[WS(os, 10)]), VSUB(TC, Tz), ovs, &(xo[0])); ST(&(xo[WS(os, 4)]), VADD(TD, TE), ovs, &(xo[0])); } } } VLEAVE(); }
static void t2fv_32(R *ri, R *ii, const R *W, stride rs, INT mb, INT me, INT ms) { DVK(KP555570233, +0.555570233019602224742830813948532874374937191); DVK(KP831469612, +0.831469612302545237078788377617905756738560812); DVK(KP195090322, +0.195090322016128267848284868477022240927691618); DVK(KP980785280, +0.980785280403230449126182236134239036973933731); DVK(KP382683432, +0.382683432365089771728459984030398866761344562); DVK(KP923879532, +0.923879532511286756128183189396788286822416626); DVK(KP707106781, +0.707106781186547524400844362104849039284835938); { INT m; R *x; x = ri; for (m = mb, W = W + (mb * ((TWVL / VL) * 62)); m < me; m = m + VL, x = x + (VL * ms), W = W + (TWVL * 62), MAKE_VOLATILE_STRIDE(rs)) { V T4, T1A, T2o, T32, Tf, T1v, T2r, T3f, TC, T1C, T2L, T34, Tr, T1D, T2O; V T33, T1k, T20, T2F, T3b, T1r, T21, T2C, T3a, TV, T1X, T2y, T38, T12, T1Y; V T2v, T37; { V T1, T1z, T3, T1x, T1y, T2, T1w, T2m, T2n; T1 = LD(&(x[0]), ms, &(x[0])); T1y = LD(&(x[WS(rs, 24)]), ms, &(x[0])); T1z = BYTWJ(&(W[TWVL * 46]), T1y); T2 = LD(&(x[WS(rs, 16)]), ms, &(x[0])); T3 = BYTWJ(&(W[TWVL * 30]), T2); T1w = LD(&(x[WS(rs, 8)]), ms, &(x[0])); T1x = BYTWJ(&(W[TWVL * 14]), T1w); T4 = VSUB(T1, T3); T1A = VSUB(T1x, T1z); T2m = VADD(T1, T3); T2n = VADD(T1x, T1z); T2o = VADD(T2m, T2n); T32 = VSUB(T2m, T2n); } { V T6, Td, T8, Tb; { V T5, Tc, T7, Ta; T5 = LD(&(x[WS(rs, 4)]), ms, &(x[0])); T6 = BYTWJ(&(W[TWVL * 6]), T5); Tc = LD(&(x[WS(rs, 12)]), ms, &(x[0])); Td = BYTWJ(&(W[TWVL * 22]), Tc); T7 = LD(&(x[WS(rs, 20)]), ms, &(x[0])); T8 = BYTWJ(&(W[TWVL * 38]), T7); Ta = LD(&(x[WS(rs, 28)]), ms, &(x[0])); Tb = BYTWJ(&(W[TWVL * 54]), Ta); } { V T9, Te, T2p, T2q; T9 = VSUB(T6, T8); Te = VSUB(Tb, Td); Tf = VMUL(LDK(KP707106781), VADD(T9, Te)); T1v = VMUL(LDK(KP707106781), VSUB(Te, T9)); T2p = VADD(T6, T8); T2q = VADD(Tb, Td); T2r = VADD(T2p, T2q); T3f = VSUB(T2q, T2p); } } { V Tt, TA, Tv, Ty; { V Ts, Tz, Tu, Tx; Ts = LD(&(x[WS(rs, 30)]), ms, &(x[0])); Tt = BYTWJ(&(W[TWVL * 58]), Ts); Tz = LD(&(x[WS(rs, 22)]), ms, &(x[0])); TA = BYTWJ(&(W[TWVL * 42]), Tz); Tu = LD(&(x[WS(rs, 14)]), ms, &(x[0])); Tv = BYTWJ(&(W[TWVL * 26]), Tu); Tx = LD(&(x[WS(rs, 6)]), ms, &(x[0])); Ty = BYTWJ(&(W[TWVL * 10]), Tx); } { V Tw, TB, T2J, T2K; Tw = VSUB(Tt, Tv); TB = VSUB(Ty, TA); TC = VFMA(LDK(KP923879532), Tw, VMUL(LDK(KP382683432), TB)); T1C = VFNMS(LDK(KP923879532), TB, VMUL(LDK(KP382683432), Tw)); T2J = VADD(Tt, Tv); T2K = VADD(Ty, TA); T2L = VADD(T2J, T2K); T34 = VSUB(T2J, T2K); } } { V Ti, Tp, Tk, Tn; { V Th, To, Tj, Tm; Th = LD(&(x[WS(rs, 2)]), ms, &(x[0])); Ti = BYTWJ(&(W[TWVL * 2]), Th); To = LD(&(x[WS(rs, 26)]), ms, &(x[0])); Tp = BYTWJ(&(W[TWVL * 50]), To); Tj = LD(&(x[WS(rs, 18)]), ms, &(x[0])); Tk = BYTWJ(&(W[TWVL * 34]), Tj); Tm = LD(&(x[WS(rs, 10)]), ms, &(x[0])); Tn = BYTWJ(&(W[TWVL * 18]), Tm); } { V Tl, Tq, T2M, T2N; Tl = VSUB(Ti, Tk); Tq = VSUB(Tn, Tp); Tr = VFNMS(LDK(KP382683432), Tq, VMUL(LDK(KP923879532), Tl)); T1D = VFMA(LDK(KP382683432), Tl, VMUL(LDK(KP923879532), Tq)); T2M = VADD(Ti, Tk); T2N = VADD(Tn, Tp); T2O = VADD(T2M, T2N); T33 = VSUB(T2M, T2N); } } { V T15, T17, T1p, T1n, T1f, T1h, T1i, T1a, T1c, T1d; { V T14, T16, T1o, T1m; T14 = LD(&(x[WS(rs, 31)]), ms, &(x[WS(rs, 1)])); T15 = BYTWJ(&(W[TWVL * 60]), T14); T16 = LD(&(x[WS(rs, 15)]), ms, &(x[WS(rs, 1)])); T17 = BYTWJ(&(W[TWVL * 28]), T16); T1o = LD(&(x[WS(rs, 23)]), ms, &(x[WS(rs, 1)])); T1p = BYTWJ(&(W[TWVL * 44]), T1o); T1m = LD(&(x[WS(rs, 7)]), ms, &(x[WS(rs, 1)])); T1n = BYTWJ(&(W[TWVL * 12]), T1m); { V T1e, T1g, T19, T1b; T1e = LD(&(x[WS(rs, 27)]), ms, &(x[WS(rs, 1)])); T1f = BYTWJ(&(W[TWVL * 52]), T1e); T1g = LD(&(x[WS(rs, 11)]), ms, &(x[WS(rs, 1)])); T1h = BYTWJ(&(W[TWVL * 20]), T1g); T1i = VSUB(T1f, T1h); T19 = LD(&(x[WS(rs, 3)]), ms, &(x[WS(rs, 1)])); T1a = BYTWJ(&(W[TWVL * 4]), T19); T1b = LD(&(x[WS(rs, 19)]), ms, &(x[WS(rs, 1)])); T1c = BYTWJ(&(W[TWVL * 36]), T1b); T1d = VSUB(T1a, T1c); } } { V T18, T1j, T2D, T2E; T18 = VSUB(T15, T17); T1j = VMUL(LDK(KP707106781), VADD(T1d, T1i)); T1k = VADD(T18, T1j); T20 = VSUB(T18, T1j); T2D = VADD(T1a, T1c); T2E = VADD(T1f, T1h); T2F = VADD(T2D, T2E); T3b = VSUB(T2E, T2D); } { V T1l, T1q, T2A, T2B; T1l = VMUL(LDK(KP707106781), VSUB(T1i, T1d)); T1q = VSUB(T1n, T1p); T1r = VSUB(T1l, T1q); T21 = VADD(T1q, T1l); T2A = VADD(T15, T17); T2B = VADD(T1n, T1p); T2C = VADD(T2A, T2B); T3a = VSUB(T2A, T2B); } } { V TG, TI, T10, TY, TQ, TS, TT, TL, TN, TO; { V TF, TH, TZ, TX; TF = LD(&(x[WS(rs, 1)]), ms, &(x[WS(rs, 1)])); TG = BYTWJ(&(W[0]), TF); TH = LD(&(x[WS(rs, 17)]), ms, &(x[WS(rs, 1)])); TI = BYTWJ(&(W[TWVL * 32]), TH); TZ = LD(&(x[WS(rs, 25)]), ms, &(x[WS(rs, 1)])); T10 = BYTWJ(&(W[TWVL * 48]), TZ); TX = LD(&(x[WS(rs, 9)]), ms, &(x[WS(rs, 1)])); TY = BYTWJ(&(W[TWVL * 16]), TX); { V TP, TR, TK, TM; TP = LD(&(x[WS(rs, 29)]), ms, &(x[WS(rs, 1)])); TQ = BYTWJ(&(W[TWVL * 56]), TP); TR = LD(&(x[WS(rs, 13)]), ms, &(x[WS(rs, 1)])); TS = BYTWJ(&(W[TWVL * 24]), TR); TT = VSUB(TQ, TS); TK = LD(&(x[WS(rs, 5)]), ms, &(x[WS(rs, 1)])); TL = BYTWJ(&(W[TWVL * 8]), TK); TM = LD(&(x[WS(rs, 21)]), ms, &(x[WS(rs, 1)])); TN = BYTWJ(&(W[TWVL * 40]), TM); TO = VSUB(TL, TN); } } { V TJ, TU, T2w, T2x; TJ = VSUB(TG, TI); TU = VMUL(LDK(KP707106781), VADD(TO, TT)); TV = VADD(TJ, TU); T1X = VSUB(TJ, TU); T2w = VADD(TL, TN); T2x = VADD(TQ, TS); T2y = VADD(T2w, T2x); T38 = VSUB(T2x, T2w); } { V TW, T11, T2t, T2u; TW = VMUL(LDK(KP707106781), VSUB(TT, TO)); T11 = VSUB(TY, T10); T12 = VSUB(TW, T11); T1Y = VADD(T11, TW); T2t = VADD(TG, TI); T2u = VADD(TY, T10); T2v = VADD(T2t, T2u); T37 = VSUB(T2t, T2u); } } { V T2W, T30, T2Z, T31; { V T2U, T2V, T2X, T2Y; T2U = VADD(T2o, T2r); T2V = VADD(T2O, T2L); T2W = VADD(T2U, T2V); T30 = VSUB(T2U, T2V); T2X = VADD(T2v, T2y); T2Y = VADD(T2C, T2F); T2Z = VADD(T2X, T2Y); T31 = VBYI(VSUB(T2Y, T2X)); } ST(&(x[WS(rs, 16)]), VSUB(T2W, T2Z), ms, &(x[0])); ST(&(x[WS(rs, 8)]), VADD(T30, T31), ms, &(x[0])); ST(&(x[0]), VADD(T2W, T2Z), ms, &(x[0])); ST(&(x[WS(rs, 24)]), VSUB(T30, T31), ms, &(x[0])); } { V T2s, T2P, T2H, T2Q, T2z, T2G; T2s = VSUB(T2o, T2r); T2P = VSUB(T2L, T2O); T2z = VSUB(T2v, T2y); T2G = VSUB(T2C, T2F); T2H = VMUL(LDK(KP707106781), VADD(T2z, T2G)); T2Q = VMUL(LDK(KP707106781), VSUB(T2G, T2z)); { V T2I, T2R, T2S, T2T; T2I = VADD(T2s, T2H); T2R = VBYI(VADD(T2P, T2Q)); ST(&(x[WS(rs, 28)]), VSUB(T2I, T2R), ms, &(x[0])); ST(&(x[WS(rs, 4)]), VADD(T2I, T2R), ms, &(x[0])); T2S = VSUB(T2s, T2H); T2T = VBYI(VSUB(T2Q, T2P)); ST(&(x[WS(rs, 20)]), VSUB(T2S, T2T), ms, &(x[0])); ST(&(x[WS(rs, 12)]), VADD(T2S, T2T), ms, &(x[0])); } } { V T36, T3r, T3h, T3p, T3d, T3o, T3k, T3s, T35, T3g; T35 = VMUL(LDK(KP707106781), VADD(T33, T34)); T36 = VADD(T32, T35); T3r = VSUB(T32, T35); T3g = VMUL(LDK(KP707106781), VSUB(T34, T33)); T3h = VADD(T3f, T3g); T3p = VSUB(T3g, T3f); { V T39, T3c, T3i, T3j; T39 = VFMA(LDK(KP923879532), T37, VMUL(LDK(KP382683432), T38)); T3c = VFNMS(LDK(KP382683432), T3b, VMUL(LDK(KP923879532), T3a)); T3d = VADD(T39, T3c); T3o = VSUB(T3c, T39); T3i = VFNMS(LDK(KP382683432), T37, VMUL(LDK(KP923879532), T38)); T3j = VFMA(LDK(KP382683432), T3a, VMUL(LDK(KP923879532), T3b)); T3k = VADD(T3i, T3j); T3s = VSUB(T3j, T3i); } { V T3e, T3l, T3u, T3v; T3e = VADD(T36, T3d); T3l = VBYI(VADD(T3h, T3k)); ST(&(x[WS(rs, 30)]), VSUB(T3e, T3l), ms, &(x[0])); ST(&(x[WS(rs, 2)]), VADD(T3e, T3l), ms, &(x[0])); T3u = VBYI(VADD(T3p, T3o)); T3v = VADD(T3r, T3s); ST(&(x[WS(rs, 6)]), VADD(T3u, T3v), ms, &(x[0])); ST(&(x[WS(rs, 26)]), VSUB(T3v, T3u), ms, &(x[0])); } { V T3m, T3n, T3q, T3t; T3m = VSUB(T36, T3d); T3n = VBYI(VSUB(T3k, T3h)); ST(&(x[WS(rs, 18)]), VSUB(T3m, T3n), ms, &(x[0])); ST(&(x[WS(rs, 14)]), VADD(T3m, T3n), ms, &(x[0])); T3q = VBYI(VSUB(T3o, T3p)); T3t = VSUB(T3r, T3s); ST(&(x[WS(rs, 10)]), VADD(T3q, T3t), ms, &(x[0])); ST(&(x[WS(rs, 22)]), VSUB(T3t, T3q), ms, &(x[0])); } } { V TE, T1P, T1I, T1Q, T1t, T1M, T1F, T1N; { V Tg, TD, T1G, T1H; Tg = VADD(T4, Tf); TD = VADD(Tr, TC); TE = VADD(Tg, TD); T1P = VSUB(Tg, TD); T1G = VFNMS(LDK(KP195090322), TV, VMUL(LDK(KP980785280), T12)); T1H = VFMA(LDK(KP195090322), T1k, VMUL(LDK(KP980785280), T1r)); T1I = VADD(T1G, T1H); T1Q = VSUB(T1H, T1G); } { V T13, T1s, T1B, T1E; T13 = VFMA(LDK(KP980785280), TV, VMUL(LDK(KP195090322), T12)); T1s = VFNMS(LDK(KP195090322), T1r, VMUL(LDK(KP980785280), T1k)); T1t = VADD(T13, T1s); T1M = VSUB(T1s, T13); T1B = VSUB(T1v, T1A); T1E = VSUB(T1C, T1D); T1F = VADD(T1B, T1E); T1N = VSUB(T1E, T1B); } { V T1u, T1J, T1S, T1T; T1u = VADD(TE, T1t); T1J = VBYI(VADD(T1F, T1I)); ST(&(x[WS(rs, 31)]), VSUB(T1u, T1J), ms, &(x[WS(rs, 1)])); ST(&(x[WS(rs, 1)]), VADD(T1u, T1J), ms, &(x[WS(rs, 1)])); T1S = VBYI(VADD(T1N, T1M)); T1T = VADD(T1P, T1Q); ST(&(x[WS(rs, 7)]), VADD(T1S, T1T), ms, &(x[WS(rs, 1)])); ST(&(x[WS(rs, 25)]), VSUB(T1T, T1S), ms, &(x[WS(rs, 1)])); } { V T1K, T1L, T1O, T1R; T1K = VSUB(TE, T1t); T1L = VBYI(VSUB(T1I, T1F)); ST(&(x[WS(rs, 17)]), VSUB(T1K, T1L), ms, &(x[WS(rs, 1)])); ST(&(x[WS(rs, 15)]), VADD(T1K, T1L), ms, &(x[WS(rs, 1)])); T1O = VBYI(VSUB(T1M, T1N)); T1R = VSUB(T1P, T1Q); ST(&(x[WS(rs, 9)]), VADD(T1O, T1R), ms, &(x[WS(rs, 1)])); ST(&(x[WS(rs, 23)]), VSUB(T1R, T1O), ms, &(x[WS(rs, 1)])); } } { V T1W, T2h, T2a, T2i, T23, T2e, T27, T2f; { V T1U, T1V, T28, T29; T1U = VSUB(T4, Tf); T1V = VADD(T1D, T1C); T1W = VADD(T1U, T1V); T2h = VSUB(T1U, T1V); T28 = VFNMS(LDK(KP555570233), T1X, VMUL(LDK(KP831469612), T1Y)); T29 = VFMA(LDK(KP555570233), T20, VMUL(LDK(KP831469612), T21)); T2a = VADD(T28, T29); T2i = VSUB(T29, T28); } { V T1Z, T22, T25, T26; T1Z = VFMA(LDK(KP831469612), T1X, VMUL(LDK(KP555570233), T1Y)); T22 = VFNMS(LDK(KP555570233), T21, VMUL(LDK(KP831469612), T20)); T23 = VADD(T1Z, T22); T2e = VSUB(T22, T1Z); T25 = VADD(T1A, T1v); T26 = VSUB(TC, Tr); T27 = VADD(T25, T26); T2f = VSUB(T26, T25); } { V T24, T2b, T2k, T2l; T24 = VADD(T1W, T23); T2b = VBYI(VADD(T27, T2a)); ST(&(x[WS(rs, 29)]), VSUB(T24, T2b), ms, &(x[WS(rs, 1)])); ST(&(x[WS(rs, 3)]), VADD(T24, T2b), ms, &(x[WS(rs, 1)])); T2k = VBYI(VADD(T2f, T2e)); T2l = VADD(T2h, T2i); ST(&(x[WS(rs, 5)]), VADD(T2k, T2l), ms, &(x[WS(rs, 1)])); ST(&(x[WS(rs, 27)]), VSUB(T2l, T2k), ms, &(x[WS(rs, 1)])); } { V T2c, T2d, T2g, T2j; T2c = VSUB(T1W, T23); T2d = VBYI(VSUB(T2a, T27)); ST(&(x[WS(rs, 19)]), VSUB(T2c, T2d), ms, &(x[WS(rs, 1)])); ST(&(x[WS(rs, 13)]), VADD(T2c, T2d), ms, &(x[WS(rs, 1)])); T2g = VBYI(VSUB(T2e, T2f)); T2j = VSUB(T2h, T2i); ST(&(x[WS(rs, 11)]), VADD(T2g, T2j), ms, &(x[WS(rs, 1)])); ST(&(x[WS(rs, 21)]), VSUB(T2j, T2g), ms, &(x[WS(rs, 1)])); } } } } VLEAVE(); }
static void t2fv_32(R *ri, R *ii, const R *W, stride rs, INT mb, INT me, INT ms) { DVK(KP831469612, +0.831469612302545237078788377617905756738560812); DVK(KP980785280, +0.980785280403230449126182236134239036973933731); DVK(KP668178637, +0.668178637919298919997757686523080761552472251); DVK(KP198912367, +0.198912367379658006911597622644676228597850501); DVK(KP923879532, +0.923879532511286756128183189396788286822416626); DVK(KP707106781, +0.707106781186547524400844362104849039284835938); DVK(KP414213562, +0.414213562373095048801688724209698078569671875); { INT m; R *x; x = ri; for (m = mb, W = W + (mb * ((TWVL / VL) * 62)); m < me; m = m + VL, x = x + (VL * ms), W = W + (TWVL * 62), MAKE_VOLATILE_STRIDE(rs)) { V T26, T25, T1Z, T22, T1W, T2a, T2k, T2g; { V T4, T1z, T2o, T32, T2r, T3f, Tf, T1A, T34, T2L, T1D, TC, T33, T2O, T1C; V Tr, T2C, T3a, T2F, T3b, T1r, T21, T1k, T20, TQ, TM, TS, TL, T2t, TJ; V T10, T2u; { V Tt, T9, T2p, Te, T2q, TA, Tu, Tx; { V T1, T1x, T2, T1v; T1 = LD(&(x[0]), ms, &(x[0])); T1x = LD(&(x[WS(rs, 24)]), ms, &(x[0])); T2 = LD(&(x[WS(rs, 16)]), ms, &(x[0])); T1v = LD(&(x[WS(rs, 8)]), ms, &(x[0])); { V T5, Tc, T7, Ta, T2m, T2n; T5 = LD(&(x[WS(rs, 4)]), ms, &(x[0])); Tc = LD(&(x[WS(rs, 12)]), ms, &(x[0])); T7 = LD(&(x[WS(rs, 20)]), ms, &(x[0])); Ta = LD(&(x[WS(rs, 28)]), ms, &(x[0])); { V T1y, T3, T1w, T6, Td, T8, Tb, Ts, Tz; Ts = LD(&(x[WS(rs, 30)]), ms, &(x[0])); T1y = BYTWJ(&(W[TWVL * 46]), T1x); T3 = BYTWJ(&(W[TWVL * 30]), T2); T1w = BYTWJ(&(W[TWVL * 14]), T1v); T6 = BYTWJ(&(W[TWVL * 6]), T5); Td = BYTWJ(&(W[TWVL * 22]), Tc); T8 = BYTWJ(&(W[TWVL * 38]), T7); Tb = BYTWJ(&(W[TWVL * 54]), Ta); Tt = BYTWJ(&(W[TWVL * 58]), Ts); Tz = LD(&(x[WS(rs, 6)]), ms, &(x[0])); T4 = VSUB(T1, T3); T2m = VADD(T1, T3); T1z = VSUB(T1w, T1y); T2n = VADD(T1w, T1y); T9 = VSUB(T6, T8); T2p = VADD(T6, T8); Te = VSUB(Tb, Td); T2q = VADD(Tb, Td); TA = BYTWJ(&(W[TWVL * 10]), Tz); } Tu = LD(&(x[WS(rs, 14)]), ms, &(x[0])); T2o = VADD(T2m, T2n); T32 = VSUB(T2m, T2n); Tx = LD(&(x[WS(rs, 22)]), ms, &(x[0])); } } { V Tv, To, Ty, Ti, Tj, Tm, Th; Th = LD(&(x[WS(rs, 2)]), ms, &(x[0])); T2r = VADD(T2p, T2q); T3f = VSUB(T2q, T2p); Tf = VADD(T9, Te); T1A = VSUB(Te, T9); Tv = BYTWJ(&(W[TWVL * 26]), Tu); To = LD(&(x[WS(rs, 26)]), ms, &(x[0])); Ty = BYTWJ(&(W[TWVL * 42]), Tx); Ti = BYTWJ(&(W[TWVL * 2]), Th); Tj = LD(&(x[WS(rs, 18)]), ms, &(x[0])); Tm = LD(&(x[WS(rs, 10)]), ms, &(x[0])); { V T1f, T1h, T1a, T1c, T18, T2A, T2B, T1p; { V T15, T17, T1o, T1m; { V Tw, T2J, Tp, T2K, TB, Tk, Tn, T1n, T14, T16; T14 = LD(&(x[WS(rs, 31)]), ms, &(x[WS(rs, 1)])); T16 = LD(&(x[WS(rs, 15)]), ms, &(x[WS(rs, 1)])); Tw = VSUB(Tt, Tv); T2J = VADD(Tt, Tv); Tp = BYTWJ(&(W[TWVL * 50]), To); T2K = VADD(TA, Ty); TB = VSUB(Ty, TA); Tk = BYTWJ(&(W[TWVL * 34]), Tj); Tn = BYTWJ(&(W[TWVL * 18]), Tm); T15 = BYTWJ(&(W[TWVL * 60]), T14); T17 = BYTWJ(&(W[TWVL * 28]), T16); T1n = LD(&(x[WS(rs, 7)]), ms, &(x[WS(rs, 1)])); { V T2M, Tl, T2N, Tq, T1l; T1l = LD(&(x[WS(rs, 23)]), ms, &(x[WS(rs, 1)])); T34 = VSUB(T2J, T2K); T2L = VADD(T2J, T2K); T1D = VFMA(LDK(KP414213562), Tw, TB); TC = VFNMS(LDK(KP414213562), TB, Tw); T2M = VADD(Ti, Tk); Tl = VSUB(Ti, Tk); T2N = VADD(Tn, Tp); Tq = VSUB(Tn, Tp); T1o = BYTWJ(&(W[TWVL * 12]), T1n); T1m = BYTWJ(&(W[TWVL * 44]), T1l); { V T1e, T1g, T19, T1b; T1e = LD(&(x[WS(rs, 27)]), ms, &(x[WS(rs, 1)])); T1g = LD(&(x[WS(rs, 11)]), ms, &(x[WS(rs, 1)])); T19 = LD(&(x[WS(rs, 3)]), ms, &(x[WS(rs, 1)])); T1b = LD(&(x[WS(rs, 19)]), ms, &(x[WS(rs, 1)])); T33 = VSUB(T2M, T2N); T2O = VADD(T2M, T2N); T1C = VFMA(LDK(KP414213562), Tl, Tq); Tr = VFNMS(LDK(KP414213562), Tq, Tl); T1f = BYTWJ(&(W[TWVL * 52]), T1e); T1h = BYTWJ(&(W[TWVL * 20]), T1g); T1a = BYTWJ(&(W[TWVL * 4]), T19); T1c = BYTWJ(&(W[TWVL * 36]), T1b); } } } T18 = VSUB(T15, T17); T2A = VADD(T15, T17); T2B = VADD(T1o, T1m); T1p = VSUB(T1m, T1o); } { V TG, TI, TZ, TX; { V T1i, T2E, T1d, T2D, TH, TY, TF; TF = LD(&(x[WS(rs, 1)]), ms, &(x[WS(rs, 1)])); T1i = VSUB(T1f, T1h); T2E = VADD(T1f, T1h); T1d = VSUB(T1a, T1c); T2D = VADD(T1a, T1c); TH = LD(&(x[WS(rs, 17)]), ms, &(x[WS(rs, 1)])); TY = LD(&(x[WS(rs, 25)]), ms, &(x[WS(rs, 1)])); T2C = VADD(T2A, T2B); T3a = VSUB(T2A, T2B); TG = BYTWJ(&(W[0]), TF); { V TW, T1j, T1q, TP, TR, TK; TW = LD(&(x[WS(rs, 9)]), ms, &(x[WS(rs, 1)])); T2F = VADD(T2D, T2E); T3b = VSUB(T2E, T2D); T1j = VADD(T1d, T1i); T1q = VSUB(T1i, T1d); TI = BYTWJ(&(W[TWVL * 32]), TH); TZ = BYTWJ(&(W[TWVL * 48]), TY); TP = LD(&(x[WS(rs, 29)]), ms, &(x[WS(rs, 1)])); TX = BYTWJ(&(W[TWVL * 16]), TW); TR = LD(&(x[WS(rs, 13)]), ms, &(x[WS(rs, 1)])); TK = LD(&(x[WS(rs, 5)]), ms, &(x[WS(rs, 1)])); T1r = VFMA(LDK(KP707106781), T1q, T1p); T21 = VFNMS(LDK(KP707106781), T1q, T1p); T1k = VFMA(LDK(KP707106781), T1j, T18); T20 = VFNMS(LDK(KP707106781), T1j, T18); TQ = BYTWJ(&(W[TWVL * 56]), TP); TM = LD(&(x[WS(rs, 21)]), ms, &(x[WS(rs, 1)])); TS = BYTWJ(&(W[TWVL * 24]), TR); TL = BYTWJ(&(W[TWVL * 8]), TK); } } T2t = VADD(TG, TI); TJ = VSUB(TG, TI); T10 = VSUB(TX, TZ); T2u = VADD(TX, TZ); } } } } { V T2s, TT, T2x, T2P, T2Y, T2G, T37, T2v, T2w, TO, T2W, T30, T2U, TN, T2V; T2s = VSUB(T2o, T2r); T2U = VADD(T2o, T2r); TN = BYTWJ(&(W[TWVL * 40]), TM); TT = VSUB(TQ, TS); T2x = VADD(TQ, TS); T2P = VSUB(T2L, T2O); T2V = VADD(T2O, T2L); T2Y = VADD(T2C, T2F); T2G = VSUB(T2C, T2F); T37 = VSUB(T2t, T2u); T2v = VADD(T2t, T2u); T2w = VADD(TL, TN); TO = VSUB(TL, TN); T2W = VADD(T2U, T2V); T30 = VSUB(T2U, T2V); { V T3i, T3o, T36, T3r, T3h, T3j, T12, T1Y, TV, T1X, T3s, T3d, T2Q, T2H, T31; V T2Z; { V T35, T3g, T38, T2y, T11, TU; T35 = VADD(T33, T34); T3g = VSUB(T34, T33); T38 = VSUB(T2w, T2x); T2y = VADD(T2w, T2x); T11 = VSUB(TO, TT); TU = VADD(TO, TT); { V T3c, T39, T2X, T2z; T3c = VFNMS(LDK(KP414213562), T3b, T3a); T3i = VFMA(LDK(KP414213562), T3a, T3b); T3o = VFNMS(LDK(KP707106781), T35, T32); T36 = VFMA(LDK(KP707106781), T35, T32); T3r = VFNMS(LDK(KP707106781), T3g, T3f); T3h = VFMA(LDK(KP707106781), T3g, T3f); T39 = VFNMS(LDK(KP414213562), T38, T37); T3j = VFMA(LDK(KP414213562), T37, T38); T2X = VADD(T2v, T2y); T2z = VSUB(T2v, T2y); T12 = VFMA(LDK(KP707106781), T11, T10); T1Y = VFNMS(LDK(KP707106781), T11, T10); TV = VFMA(LDK(KP707106781), TU, TJ); T1X = VFNMS(LDK(KP707106781), TU, TJ); T3s = VSUB(T3c, T39); T3d = VADD(T39, T3c); T2Q = VSUB(T2G, T2z); T2H = VADD(T2z, T2G); T31 = VSUB(T2Y, T2X); T2Z = VADD(T2X, T2Y); } } { V Tg, T1U, TD, T1G, T13, T1s, T1H, T1B, T1V, T1E, T3k, T3p, T2e, T2f; Tg = VFMA(LDK(KP707106781), Tf, T4); T1U = VFNMS(LDK(KP707106781), Tf, T4); T3k = VSUB(T3i, T3j); T3p = VADD(T3j, T3i); { V T3v, T3t, T3e, T3m; T3v = VFNMS(LDK(KP923879532), T3s, T3r); T3t = VFMA(LDK(KP923879532), T3s, T3r); T3e = VFNMS(LDK(KP923879532), T3d, T36); T3m = VFMA(LDK(KP923879532), T3d, T36); { V T2R, T2T, T2I, T2S; T2R = VFNMS(LDK(KP707106781), T2Q, T2P); T2T = VFMA(LDK(KP707106781), T2Q, T2P); T2I = VFNMS(LDK(KP707106781), T2H, T2s); T2S = VFMA(LDK(KP707106781), T2H, T2s); ST(&(x[WS(rs, 24)]), VFNMSI(T31, T30), ms, &(x[0])); ST(&(x[WS(rs, 8)]), VFMAI(T31, T30), ms, &(x[0])); ST(&(x[0]), VADD(T2W, T2Z), ms, &(x[0])); ST(&(x[WS(rs, 16)]), VSUB(T2W, T2Z), ms, &(x[0])); { V T3u, T3q, T3l, T3n; T3u = VFMA(LDK(KP923879532), T3p, T3o); T3q = VFNMS(LDK(KP923879532), T3p, T3o); T3l = VFNMS(LDK(KP923879532), T3k, T3h); T3n = VFMA(LDK(KP923879532), T3k, T3h); ST(&(x[WS(rs, 4)]), VFMAI(T2T, T2S), ms, &(x[0])); ST(&(x[WS(rs, 28)]), VFNMSI(T2T, T2S), ms, &(x[0])); ST(&(x[WS(rs, 20)]), VFMAI(T2R, T2I), ms, &(x[0])); ST(&(x[WS(rs, 12)]), VFNMSI(T2R, T2I), ms, &(x[0])); ST(&(x[WS(rs, 22)]), VFNMSI(T3t, T3q), ms, &(x[0])); ST(&(x[WS(rs, 10)]), VFMAI(T3t, T3q), ms, &(x[0])); ST(&(x[WS(rs, 26)]), VFMAI(T3v, T3u), ms, &(x[0])); ST(&(x[WS(rs, 6)]), VFNMSI(T3v, T3u), ms, &(x[0])); ST(&(x[WS(rs, 2)]), VFMAI(T3n, T3m), ms, &(x[0])); ST(&(x[WS(rs, 30)]), VFNMSI(T3n, T3m), ms, &(x[0])); ST(&(x[WS(rs, 18)]), VFMAI(T3l, T3e), ms, &(x[0])); ST(&(x[WS(rs, 14)]), VFNMSI(T3l, T3e), ms, &(x[0])); T26 = VSUB(TC, Tr); TD = VADD(Tr, TC); } } } T1G = VFMA(LDK(KP198912367), TV, T12); T13 = VFNMS(LDK(KP198912367), T12, TV); T1s = VFNMS(LDK(KP198912367), T1r, T1k); T1H = VFMA(LDK(KP198912367), T1k, T1r); T1B = VFNMS(LDK(KP707106781), T1A, T1z); T25 = VFMA(LDK(KP707106781), T1A, T1z); T1V = VADD(T1C, T1D); T1E = VSUB(T1C, T1D); { V T1S, T1O, T1K, T1u, T1R, T1T, T1L, T1J; { V TE, T1M, T1I, T1N, T1t, T1Q, T1F, T1P, T28, T29; TE = VFMA(LDK(KP923879532), TD, Tg); T1M = VFNMS(LDK(KP923879532), TD, Tg); T1I = VSUB(T1G, T1H); T1N = VADD(T1G, T1H); T1t = VADD(T13, T1s); T1Q = VSUB(T1s, T13); T1F = VFMA(LDK(KP923879532), T1E, T1B); T1P = VFNMS(LDK(KP923879532), T1E, T1B); T28 = VFNMS(LDK(KP668178637), T1X, T1Y); T1Z = VFMA(LDK(KP668178637), T1Y, T1X); T1S = VFMA(LDK(KP980785280), T1N, T1M); T1O = VFNMS(LDK(KP980785280), T1N, T1M); T22 = VFMA(LDK(KP668178637), T21, T20); T29 = VFNMS(LDK(KP668178637), T20, T21); T1K = VFMA(LDK(KP980785280), T1t, TE); T1u = VFNMS(LDK(KP980785280), T1t, TE); T1R = VFNMS(LDK(KP980785280), T1Q, T1P); T1T = VFMA(LDK(KP980785280), T1Q, T1P); T1L = VFMA(LDK(KP980785280), T1I, T1F); T1J = VFNMS(LDK(KP980785280), T1I, T1F); T2e = VFNMS(LDK(KP923879532), T1V, T1U); T1W = VFMA(LDK(KP923879532), T1V, T1U); T2a = VSUB(T28, T29); T2f = VADD(T28, T29); } ST(&(x[WS(rs, 23)]), VFMAI(T1R, T1O), ms, &(x[WS(rs, 1)])); ST(&(x[WS(rs, 9)]), VFNMSI(T1R, T1O), ms, &(x[WS(rs, 1)])); ST(&(x[WS(rs, 25)]), VFNMSI(T1T, T1S), ms, &(x[WS(rs, 1)])); ST(&(x[WS(rs, 7)]), VFMAI(T1T, T1S), ms, &(x[WS(rs, 1)])); ST(&(x[WS(rs, 31)]), VFMAI(T1L, T1K), ms, &(x[WS(rs, 1)])); ST(&(x[WS(rs, 1)]), VFNMSI(T1L, T1K), ms, &(x[WS(rs, 1)])); ST(&(x[WS(rs, 15)]), VFMAI(T1J, T1u), ms, &(x[WS(rs, 1)])); ST(&(x[WS(rs, 17)]), VFNMSI(T1J, T1u), ms, &(x[WS(rs, 1)])); } T2k = VFNMS(LDK(KP831469612), T2f, T2e); T2g = VFMA(LDK(KP831469612), T2f, T2e); } } } } { V T2i, T23, T2h, T27; T2i = VSUB(T22, T1Z); T23 = VADD(T1Z, T22); T2h = VFNMS(LDK(KP923879532), T26, T25); T27 = VFMA(LDK(KP923879532), T26, T25); { V T2c, T24, T2j, T2l, T2d, T2b; T2c = VFMA(LDK(KP831469612), T23, T1W); T24 = VFNMS(LDK(KP831469612), T23, T1W); T2j = VFMA(LDK(KP831469612), T2i, T2h); T2l = VFNMS(LDK(KP831469612), T2i, T2h); T2d = VFMA(LDK(KP831469612), T2a, T27); T2b = VFNMS(LDK(KP831469612), T2a, T27); ST(&(x[WS(rs, 21)]), VFNMSI(T2j, T2g), ms, &(x[WS(rs, 1)])); ST(&(x[WS(rs, 11)]), VFMAI(T2j, T2g), ms, &(x[WS(rs, 1)])); ST(&(x[WS(rs, 27)]), VFMAI(T2l, T2k), ms, &(x[WS(rs, 1)])); ST(&(x[WS(rs, 5)]), VFNMSI(T2l, T2k), ms, &(x[WS(rs, 1)])); ST(&(x[WS(rs, 3)]), VFMAI(T2d, T2c), ms, &(x[WS(rs, 1)])); ST(&(x[WS(rs, 29)]), VFNMSI(T2d, T2c), ms, &(x[WS(rs, 1)])); ST(&(x[WS(rs, 19)]), VFMAI(T2b, T24), ms, &(x[WS(rs, 1)])); ST(&(x[WS(rs, 13)]), VFNMSI(T2b, T24), ms, &(x[WS(rs, 1)])); } } } } VLEAVE(); }
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(); }
static void t1fuv_7(R *ri, R *ii, const R *W, stride rs, INT mb, INT me, INT ms) { DVK(KP900968867, +0.900968867902419126236102319507445051165919162); DVK(KP222520933, +0.222520933956314404288902564496794759466355569); DVK(KP623489801, +0.623489801858733530525004884004239810632274731); DVK(KP781831482, +0.781831482468029808708444526674057750232334519); DVK(KP974927912, +0.974927912181823607018131682993931217232785801); DVK(KP433883739, +0.433883739117558120475768332848358754609990728); { INT m; R *x; x = ri; for (m = mb, W = W + (mb * ((TWVL / VL) * 12)); m < me; m = m + VL, x = x + (VL * ms), W = W + (TWVL * 12), MAKE_VOLATILE_STRIDE(7, rs)) { V T1, Tg, Tj, T6, Ti, Tb, Tk, Tp, To; T1 = LD(&(x[0]), ms, &(x[0])); { V Td, Tf, Tc, Te; Tc = LD(&(x[WS(rs, 3)]), ms, &(x[WS(rs, 1)])); Td = BYTWJ(&(W[TWVL * 4]), Tc); Te = LD(&(x[WS(rs, 4)]), ms, &(x[0])); Tf = BYTWJ(&(W[TWVL * 6]), Te); Tg = VADD(Td, Tf); Tj = VSUB(Tf, Td); } { V T3, T5, T2, T4; T2 = LD(&(x[WS(rs, 1)]), ms, &(x[WS(rs, 1)])); T3 = BYTWJ(&(W[0]), T2); T4 = LD(&(x[WS(rs, 6)]), ms, &(x[0])); T5 = BYTWJ(&(W[TWVL * 10]), T4); T6 = VADD(T3, T5); Ti = VSUB(T5, T3); } { V T8, Ta, T7, T9; T7 = LD(&(x[WS(rs, 2)]), ms, &(x[0])); T8 = BYTWJ(&(W[TWVL * 2]), T7); T9 = LD(&(x[WS(rs, 5)]), ms, &(x[WS(rs, 1)])); Ta = BYTWJ(&(W[TWVL * 8]), T9); Tb = VADD(T8, Ta); Tk = VSUB(Ta, T8); } ST(&(x[0]), VADD(T1, VADD(T6, VADD(Tb, Tg))), ms, &(x[0])); Tp = VBYI(VFMA(LDK(KP433883739), Ti, VFNMS(LDK(KP781831482), Tk, VMUL(LDK(KP974927912), Tj)))); To = VFMA(LDK(KP623489801), Tb, VFNMS(LDK(KP222520933), Tg, VFNMS(LDK(KP900968867), T6, T1))); ST(&(x[WS(rs, 4)]), VSUB(To, Tp), ms, &(x[0])); ST(&(x[WS(rs, 3)]), VADD(To, Tp), ms, &(x[WS(rs, 1)])); { V Tl, Th, Tn, Tm; Tl = VBYI(VFNMS(LDK(KP781831482), Tj, VFNMS(LDK(KP433883739), Tk, VMUL(LDK(KP974927912), Ti)))); Th = VFMA(LDK(KP623489801), Tg, VFNMS(LDK(KP900968867), Tb, VFNMS(LDK(KP222520933), T6, T1))); ST(&(x[WS(rs, 5)]), VSUB(Th, Tl), ms, &(x[WS(rs, 1)])); ST(&(x[WS(rs, 2)]), VADD(Th, Tl), ms, &(x[0])); Tn = VBYI(VFMA(LDK(KP781831482), Ti, VFMA(LDK(KP974927912), Tk, VMUL(LDK(KP433883739), Tj)))); Tm = VFMA(LDK(KP623489801), T6, VFNMS(LDK(KP900968867), Tg, VFNMS(LDK(KP222520933), Tb, T1))); ST(&(x[WS(rs, 6)]), VSUB(Tm, Tn), ms, &(x[0])); ST(&(x[WS(rs, 1)]), VADD(Tm, Tn), ms, &(x[WS(rs, 1)])); } } } VLEAVE(); }
static void n1fv_10(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs) { DVK(KP250000000, +0.250000000000000000000000000000000000000000000); DVK(KP559016994, +0.559016994374947424102293417182819058860154590); DVK(KP587785252, +0.587785252292473129168705954639072768597652438); DVK(KP951056516, +0.951056516295153572116439333379382143405698634); { INT i; const R *xi; R *xo; xi = ri; xo = ro; for (i = v; i > 0; i = i - VL, xi = xi + (VL * ivs), xo = xo + (VL * ovs), MAKE_VOLATILE_STRIDE(20, is), MAKE_VOLATILE_STRIDE(20, os)) { V Ti, Ty, Tm, Tn, Tw, Tt, Tz, TA, TB, T7, Te, Tj, Tg, Th; Tg = LD(&(xi[0]), ivs, &(xi[0])); Th = LD(&(xi[WS(is, 5)]), ivs, &(xi[WS(is, 1)])); Ti = VSUB(Tg, Th); Ty = VADD(Tg, Th); { V T3, Tu, Td, Ts, T6, Tv, Ta, Tr; { V T1, T2, Tb, Tc; T1 = LD(&(xi[WS(is, 2)]), ivs, &(xi[0])); T2 = LD(&(xi[WS(is, 7)]), ivs, &(xi[WS(is, 1)])); T3 = VSUB(T1, T2); Tu = VADD(T1, T2); Tb = LD(&(xi[WS(is, 6)]), ivs, &(xi[0])); Tc = LD(&(xi[WS(is, 1)]), ivs, &(xi[WS(is, 1)])); Td = VSUB(Tb, Tc); Ts = VADD(Tb, Tc); } { V T4, T5, T8, T9; T4 = LD(&(xi[WS(is, 8)]), ivs, &(xi[0])); T5 = LD(&(xi[WS(is, 3)]), ivs, &(xi[WS(is, 1)])); T6 = VSUB(T4, T5); Tv = VADD(T4, T5); T8 = LD(&(xi[WS(is, 4)]), ivs, &(xi[0])); T9 = LD(&(xi[WS(is, 9)]), ivs, &(xi[WS(is, 1)])); Ta = VSUB(T8, T9); Tr = VADD(T8, T9); } Tm = VSUB(T3, T6); Tn = VSUB(Ta, Td); Tw = VSUB(Tu, Tv); Tt = VSUB(Tr, Ts); Tz = VADD(Tu, Tv); TA = VADD(Tr, Ts); TB = VADD(Tz, TA); T7 = VADD(T3, T6); Te = VADD(Ta, Td); Tj = VADD(T7, Te); } ST(&(xo[WS(os, 5)]), VADD(Ti, Tj), ovs, &(xo[WS(os, 1)])); ST(&(xo[0]), VADD(Ty, TB), ovs, &(xo[0])); { V To, Tq, Tl, Tp, Tf, Tk; To = VBYI(VFMA(LDK(KP951056516), Tm, VMUL(LDK(KP587785252), Tn))); Tq = VBYI(VFNMS(LDK(KP587785252), Tm, VMUL(LDK(KP951056516), Tn))); Tf = VMUL(LDK(KP559016994), VSUB(T7, Te)); Tk = VFNMS(LDK(KP250000000), Tj, Ti); Tl = VADD(Tf, Tk); Tp = VSUB(Tk, Tf); ST(&(xo[WS(os, 1)]), VSUB(Tl, To), ovs, &(xo[WS(os, 1)])); ST(&(xo[WS(os, 7)]), VADD(Tq, Tp), ovs, &(xo[WS(os, 1)])); ST(&(xo[WS(os, 9)]), VADD(To, Tl), ovs, &(xo[WS(os, 1)])); ST(&(xo[WS(os, 3)]), VSUB(Tp, Tq), ovs, &(xo[WS(os, 1)])); } { V Tx, TF, TE, TG, TC, TD; Tx = VBYI(VFNMS(LDK(KP587785252), Tw, VMUL(LDK(KP951056516), Tt))); TF = VBYI(VFMA(LDK(KP951056516), Tw, VMUL(LDK(KP587785252), Tt))); TC = VFNMS(LDK(KP250000000), TB, Ty); TD = VMUL(LDK(KP559016994), VSUB(Tz, TA)); TE = VSUB(TC, TD); TG = VADD(TD, TC); ST(&(xo[WS(os, 2)]), VADD(Tx, TE), ovs, &(xo[0])); ST(&(xo[WS(os, 6)]), VSUB(TG, TF), ovs, &(xo[0])); ST(&(xo[WS(os, 8)]), VSUB(TE, Tx), ovs, &(xo[0])); ST(&(xo[WS(os, 4)]), VADD(TF, TG), ovs, &(xo[0])); } } } VLEAVE(); }
static void t3fv_20(R *ri, R *ii, const R *W, stride rs, INT mb, INT me, INT ms) { DVK(KP587785252, +0.587785252292473129168705954639072768597652438); DVK(KP951056516, +0.951056516295153572116439333379382143405698634); DVK(KP250000000, +0.250000000000000000000000000000000000000000000); DVK(KP559016994, +0.559016994374947424102293417182819058860154590); { INT m; R *x; x = ri; for (m = mb, W = W + (mb * ((TWVL / VL) * 8)); m < me; m = m + VL, x = x + (VL * ms), W = W + (TWVL * 8), MAKE_VOLATILE_STRIDE(20, rs)) { V T2, T8, T9, TA, T3, Tc, T4, TZ, T18, Tl, Tq, Tx, TU, Td, Te; V T15, Ti, Tt, TJ; T2 = LDW(&(W[0])); T8 = LDW(&(W[TWVL * 2])); T9 = VZMUL(T2, T8); TA = VZMULJ(T2, T8); T3 = LDW(&(W[TWVL * 4])); Tc = VZMULJ(T9, T3); T4 = VZMUL(T2, T3); TZ = VZMUL(T9, T3); T18 = VZMULJ(TA, T3); Tl = VZMULJ(T8, T3); Tq = VZMULJ(T2, T3); Tx = VZMUL(T8, T3); TU = VZMUL(TA, T3); Td = LDW(&(W[TWVL * 6])); Te = VZMULJ(Tc, Td); T15 = VZMULJ(TA, Td); Ti = VZMULJ(T8, Td); Tt = VZMULJ(T2, Td); TJ = VZMULJ(T9, Td); { V T7, TM, T1U, T2d, T1i, T1p, T1q, T1j, Tp, TE, TF, T26, T27, T2b, T1M; V T1P, T1V, TY, T1c, T1d, T23, T24, T2a, T1F, T1I, T1W, TG, T1e; { V T1, TL, T6, TI, TK, T5, TH, T1S, T1T; T1 = LD(&(x[0]), ms, &(x[0])); TK = LD(&(x[WS(rs, 15)]), ms, &(x[WS(rs, 1)])); TL = VZMULJ(TJ, TK); T5 = LD(&(x[WS(rs, 10)]), ms, &(x[0])); T6 = VZMULJ(T4, T5); TH = LD(&(x[WS(rs, 5)]), ms, &(x[WS(rs, 1)])); TI = VZMULJ(Tc, TH); T7 = VSUB(T1, T6); TM = VSUB(TI, TL); T1S = VADD(T1, T6); T1T = VADD(TI, TL); T1U = VSUB(T1S, T1T); T2d = VADD(T1S, T1T); } { V Th, T1K, T14, T1E, T1b, T1H, To, T1N, Tw, T1D, TR, T1L, TX, T1O, TD; V T1G; { V Tb, Tg, Ta, Tf; Ta = LD(&(x[WS(rs, 4)]), ms, &(x[0])); Tb = VZMULJ(T9, Ta); Tf = LD(&(x[WS(rs, 14)]), ms, &(x[0])); Tg = VZMULJ(Te, Tf); Th = VSUB(Tb, Tg); T1K = VADD(Tb, Tg); } { V T11, T13, T10, T12; T10 = LD(&(x[WS(rs, 13)]), ms, &(x[WS(rs, 1)])); T11 = VZMULJ(TZ, T10); T12 = LD(&(x[WS(rs, 3)]), ms, &(x[WS(rs, 1)])); T13 = VZMULJ(T8, T12); T14 = VSUB(T11, T13); T1E = VADD(T11, T13); } { V T17, T1a, T16, T19; T16 = LD(&(x[WS(rs, 17)]), ms, &(x[WS(rs, 1)])); T17 = VZMULJ(T15, T16); T19 = LD(&(x[WS(rs, 7)]), ms, &(x[WS(rs, 1)])); T1a = VZMULJ(T18, T19); T1b = VSUB(T17, T1a); T1H = VADD(T17, T1a); } { V Tk, Tn, Tj, Tm; Tj = LD(&(x[WS(rs, 16)]), ms, &(x[0])); Tk = VZMULJ(Ti, Tj); Tm = LD(&(x[WS(rs, 6)]), ms, &(x[0])); Tn = VZMULJ(Tl, Tm); To = VSUB(Tk, Tn); T1N = VADD(Tk, Tn); } { V Ts, Tv, Tr, Tu; Tr = LD(&(x[WS(rs, 8)]), ms, &(x[0])); Ts = VZMULJ(Tq, Tr); Tu = LD(&(x[WS(rs, 18)]), ms, &(x[0])); Tv = VZMULJ(Tt, Tu); Tw = VSUB(Ts, Tv); T1D = VADD(Ts, Tv); } { V TO, TQ, TN, TP; TN = LD(&(x[WS(rs, 9)]), ms, &(x[WS(rs, 1)])); TO = VZMULJ(T3, TN); TP = LD(&(x[WS(rs, 19)]), ms, &(x[WS(rs, 1)])); TQ = VZMULJ(Td, TP); TR = VSUB(TO, TQ); T1L = VADD(TO, TQ); } { V TT, TW, TS, TV; TS = LD(&(x[WS(rs, 1)]), ms, &(x[WS(rs, 1)])); TT = VZMULJ(T2, TS); TV = LD(&(x[WS(rs, 11)]), ms, &(x[WS(rs, 1)])); TW = VZMULJ(TU, TV); TX = VSUB(TT, TW); T1O = VADD(TT, TW); } { V Tz, TC, Ty, TB; Ty = LD(&(x[WS(rs, 12)]), ms, &(x[0])); Tz = VZMULJ(Tx, Ty); TB = LD(&(x[WS(rs, 2)]), ms, &(x[0])); TC = VZMULJ(TA, TB); TD = VSUB(Tz, TC); T1G = VADD(Tz, TC); } T1i = VSUB(TX, TR); T1p = VSUB(Th, To); T1q = VSUB(Tw, TD); T1j = VSUB(T1b, T14); Tp = VADD(Th, To); TE = VADD(Tw, TD); TF = VADD(Tp, TE); T26 = VADD(T1D, T1E); T27 = VADD(T1G, T1H); T2b = VADD(T26, T27); T1M = VSUB(T1K, T1L); T1P = VSUB(T1N, T1O); T1V = VADD(T1M, T1P); TY = VADD(TR, TX); T1c = VADD(T14, T1b); T1d = VADD(TY, T1c); T23 = VADD(T1K, T1L); T24 = VADD(T1N, T1O); T2a = VADD(T23, T24); T1F = VSUB(T1D, T1E); T1I = VSUB(T1G, T1H); T1W = VADD(T1F, T1I); } TG = VADD(T7, TF); T1e = VBYI(VADD(TM, T1d)); ST(&(x[WS(rs, 5)]), VSUB(TG, T1e), ms, &(x[WS(rs, 1)])); ST(&(x[WS(rs, 15)]), VADD(TG, T1e), ms, &(x[WS(rs, 1)])); { V T2c, T2e, T2f, T29, T2i, T25, T28, T2h, T2g; T2c = VMUL(LDK(KP559016994), VSUB(T2a, T2b)); T2e = VADD(T2a, T2b); T2f = VFNMS(LDK(KP250000000), T2e, T2d); T25 = VSUB(T23, T24); T28 = VSUB(T26, T27); T29 = VBYI(VFMA(LDK(KP951056516), T25, VMUL(LDK(KP587785252), T28))); T2i = VBYI(VFNMS(LDK(KP587785252), T25, VMUL(LDK(KP951056516), T28))); ST(&(x[0]), VADD(T2d, T2e), ms, &(x[0])); T2h = VSUB(T2f, T2c); ST(&(x[WS(rs, 8)]), VSUB(T2h, T2i), ms, &(x[0])); ST(&(x[WS(rs, 12)]), VADD(T2i, T2h), ms, &(x[0])); T2g = VADD(T2c, T2f); ST(&(x[WS(rs, 4)]), VADD(T29, T2g), ms, &(x[0])); ST(&(x[WS(rs, 16)]), VSUB(T2g, T29), ms, &(x[0])); } { V T1Z, T1X, T1Y, T1R, T22, T1J, T1Q, T21, T20; T1Z = VMUL(LDK(KP559016994), VSUB(T1V, T1W)); T1X = VADD(T1V, T1W); T1Y = VFNMS(LDK(KP250000000), T1X, T1U); T1J = VSUB(T1F, T1I); T1Q = VSUB(T1M, T1P); T1R = VBYI(VFNMS(LDK(KP587785252), T1Q, VMUL(LDK(KP951056516), T1J))); T22 = VBYI(VFMA(LDK(KP951056516), T1Q, VMUL(LDK(KP587785252), T1J))); ST(&(x[WS(rs, 10)]), VADD(T1U, T1X), ms, &(x[0])); T21 = VADD(T1Z, T1Y); ST(&(x[WS(rs, 6)]), VSUB(T21, T22), ms, &(x[0])); ST(&(x[WS(rs, 14)]), VADD(T22, T21), ms, &(x[0])); T20 = VSUB(T1Y, T1Z); ST(&(x[WS(rs, 2)]), VADD(T1R, T20), ms, &(x[0])); ST(&(x[WS(rs, 18)]), VSUB(T20, T1R), ms, &(x[0])); } { V T1k, T1r, T1z, T1w, T1o, T1y, T1h, T1v; T1k = VFMA(LDK(KP951056516), T1i, VMUL(LDK(KP587785252), T1j)); T1r = VFMA(LDK(KP951056516), T1p, VMUL(LDK(KP587785252), T1q)); T1z = VFNMS(LDK(KP587785252), T1p, VMUL(LDK(KP951056516), T1q)); T1w = VFNMS(LDK(KP587785252), T1i, VMUL(LDK(KP951056516), T1j)); { V T1m, T1n, T1f, T1g; T1m = VFMS(LDK(KP250000000), T1d, TM); T1n = VMUL(LDK(KP559016994), VSUB(T1c, TY)); T1o = VADD(T1m, T1n); T1y = VSUB(T1n, T1m); T1f = VMUL(LDK(KP559016994), VSUB(Tp, TE)); T1g = VFNMS(LDK(KP250000000), TF, T7); T1h = VADD(T1f, T1g); T1v = VSUB(T1g, T1f); } { V T1l, T1s, T1B, T1C; T1l = VADD(T1h, T1k); T1s = VBYI(VSUB(T1o, T1r)); ST(&(x[WS(rs, 19)]), VSUB(T1l, T1s), ms, &(x[WS(rs, 1)])); ST(&(x[WS(rs, 1)]), VADD(T1l, T1s), ms, &(x[WS(rs, 1)])); T1B = VADD(T1v, T1w); T1C = VBYI(VADD(T1z, T1y)); ST(&(x[WS(rs, 13)]), VSUB(T1B, T1C), ms, &(x[WS(rs, 1)])); ST(&(x[WS(rs, 7)]), VADD(T1B, T1C), ms, &(x[WS(rs, 1)])); } { V T1t, T1u, T1x, T1A; T1t = VSUB(T1h, T1k); T1u = VBYI(VADD(T1r, T1o)); ST(&(x[WS(rs, 11)]), VSUB(T1t, T1u), ms, &(x[WS(rs, 1)])); ST(&(x[WS(rs, 9)]), VADD(T1t, T1u), ms, &(x[WS(rs, 1)])); T1x = VSUB(T1v, T1w); T1A = VBYI(VSUB(T1y, T1z)); ST(&(x[WS(rs, 17)]), VSUB(T1x, T1A), ms, &(x[WS(rs, 1)])); ST(&(x[WS(rs, 3)]), VADD(T1x, T1A), ms, &(x[WS(rs, 1)])); } } } } } VLEAVE(); }
static void n1fv_10(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs) { DVK(KP559016994, +0.559016994374947424102293417182819058860154590); DVK(KP250000000, +0.250000000000000000000000000000000000000000000); DVK(KP618033988, +0.618033988749894848204586834365638117720309180); DVK(KP951056516, +0.951056516295153572116439333379382143405698634); { INT i; const R *xi; R *xo; xi = ri; xo = ro; for (i = v; i > 0; i = i - VL, xi = xi + (VL * ivs), xo = xo + (VL * ovs), MAKE_VOLATILE_STRIDE(20, is), MAKE_VOLATILE_STRIDE(20, os)) { V Tb, Tr, T3, Ts, T6, Tw, Tg, Tt, T9, Tc, T1, T2; T1 = LD(&(xi[0]), ivs, &(xi[0])); T2 = LD(&(xi[WS(is, 5)]), ivs, &(xi[WS(is, 1)])); { V T4, T5, Te, Tf, T7, T8; T4 = LD(&(xi[WS(is, 2)]), ivs, &(xi[0])); T5 = LD(&(xi[WS(is, 7)]), ivs, &(xi[WS(is, 1)])); Te = LD(&(xi[WS(is, 6)]), ivs, &(xi[0])); Tf = LD(&(xi[WS(is, 1)]), ivs, &(xi[WS(is, 1)])); T7 = LD(&(xi[WS(is, 8)]), ivs, &(xi[0])); T8 = LD(&(xi[WS(is, 3)]), ivs, &(xi[WS(is, 1)])); Tb = LD(&(xi[WS(is, 4)]), ivs, &(xi[0])); Tr = VADD(T1, T2); T3 = VSUB(T1, T2); Ts = VADD(T4, T5); T6 = VSUB(T4, T5); Tw = VADD(Te, Tf); Tg = VSUB(Te, Tf); Tt = VADD(T7, T8); T9 = VSUB(T7, T8); Tc = LD(&(xi[WS(is, 9)]), ivs, &(xi[WS(is, 1)])); } { V TD, Tu, Tm, Ta, Td, Tv; TD = VSUB(Ts, Tt); Tu = VADD(Ts, Tt); Tm = VSUB(T6, T9); Ta = VADD(T6, T9); Td = VSUB(Tb, Tc); Tv = VADD(Tb, Tc); { V TC, Tx, Tn, Th; TC = VSUB(Tv, Tw); Tx = VADD(Tv, Tw); Tn = VSUB(Td, Tg); Th = VADD(Td, Tg); { V Ty, TA, TE, TG, Ti, Tk, To, Tq, Tz, Tj; Ty = VADD(Tu, Tx); TA = VSUB(Tu, Tx); TE = VMUL(LDK(KP951056516), VFNMS(LDK(KP618033988), TD, TC)); TG = VMUL(LDK(KP951056516), VFMA(LDK(KP618033988), TC, TD)); Ti = VADD(Ta, Th); Tk = VSUB(Ta, Th); To = VMUL(LDK(KP951056516), VFMA(LDK(KP618033988), Tn, Tm)); Tq = VMUL(LDK(KP951056516), VFNMS(LDK(KP618033988), Tm, Tn)); Tz = VFNMS(LDK(KP250000000), Ty, Tr); ST(&(xo[0]), VADD(Tr, Ty), ovs, &(xo[0])); Tj = VFNMS(LDK(KP250000000), Ti, T3); ST(&(xo[WS(os, 5)]), VADD(T3, Ti), ovs, &(xo[WS(os, 1)])); { V TB, TF, Tl, Tp; TB = VFNMS(LDK(KP559016994), TA, Tz); TF = VFMA(LDK(KP559016994), TA, Tz); Tl = VFMA(LDK(KP559016994), Tk, Tj); Tp = VFNMS(LDK(KP559016994), Tk, Tj); ST(&(xo[WS(os, 4)]), VFMAI(TG, TF), ovs, &(xo[0])); ST(&(xo[WS(os, 6)]), VFNMSI(TG, TF), ovs, &(xo[0])); ST(&(xo[WS(os, 8)]), VFNMSI(TE, TB), ovs, &(xo[0])); ST(&(xo[WS(os, 2)]), VFMAI(TE, TB), ovs, &(xo[0])); ST(&(xo[WS(os, 3)]), VFNMSI(Tq, Tp), ovs, &(xo[WS(os, 1)])); ST(&(xo[WS(os, 7)]), VFMAI(Tq, Tp), ovs, &(xo[WS(os, 1)])); ST(&(xo[WS(os, 9)]), VFMAI(To, Tl), ovs, &(xo[WS(os, 1)])); ST(&(xo[WS(os, 1)]), VFNMSI(To, Tl), ovs, &(xo[WS(os, 1)])); } } } } } } VLEAVE(); }
static void n1bv_15(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs) { DVK(KP823639103, +0.823639103546331925877420039278190003029660514); DVK(KP910592997, +0.910592997310029334643087372129977886038870291); DVK(KP559016994, +0.559016994374947424102293417182819058860154590); DVK(KP951056516, +0.951056516295153572116439333379382143405698634); DVK(KP866025403, +0.866025403784438646763723170752936183471402627); DVK(KP250000000, +0.250000000000000000000000000000000000000000000); DVK(KP618033988, +0.618033988749894848204586834365638117720309180); DVK(KP500000000, +0.500000000000000000000000000000000000000000000); { INT i; const R *xi; R *xo; xi = ii; xo = io; for (i = v; i > 0; i = i - VL, xi = xi + (VL * ivs), xo = xo + (VL * ovs), MAKE_VOLATILE_STRIDE(is), MAKE_VOLATILE_STRIDE(os)) { V Tb, TH, Tw, TA, Th, T11, T5, Ti, T12, Ta, Tx, Te, Tq, T16, Tj; V T1, T2, T3; T1 = LD(&(xi[0]), ivs, &(xi[0])); T2 = LD(&(xi[WS(is, 5)]), ivs, &(xi[WS(is, 1)])); T3 = LD(&(xi[WS(is, 10)]), ivs, &(xi[0])); { V T6, T7, T8, Tm, Tn, To; T6 = LD(&(xi[WS(is, 3)]), ivs, &(xi[WS(is, 1)])); T7 = LD(&(xi[WS(is, 8)]), ivs, &(xi[0])); T8 = LD(&(xi[WS(is, 13)]), ivs, &(xi[WS(is, 1)])); Tm = LD(&(xi[WS(is, 9)]), ivs, &(xi[WS(is, 1)])); Tn = LD(&(xi[WS(is, 14)]), ivs, &(xi[0])); To = LD(&(xi[WS(is, 4)]), ivs, &(xi[0])); { V T4, Tc, T9, Td, Tp; Tb = LD(&(xi[WS(is, 12)]), ivs, &(xi[0])); T4 = VADD(T2, T3); TH = VSUB(T2, T3); Tc = LD(&(xi[WS(is, 2)]), ivs, &(xi[0])); Tw = VSUB(T7, T8); T9 = VADD(T7, T8); Td = LD(&(xi[WS(is, 7)]), ivs, &(xi[WS(is, 1)])); Tp = VADD(Tn, To); TA = VSUB(Tn, To); Th = LD(&(xi[WS(is, 6)]), ivs, &(xi[0])); T11 = VADD(T1, T4); T5 = VFNMS(LDK(KP500000000), T4, T1); Ti = LD(&(xi[WS(is, 11)]), ivs, &(xi[WS(is, 1)])); T12 = VADD(T6, T9); Ta = VFNMS(LDK(KP500000000), T9, T6); Tx = VSUB(Tc, Td); Te = VADD(Tc, Td); Tq = VFNMS(LDK(KP500000000), Tp, Tm); T16 = VADD(Tm, Tp); Tj = LD(&(xi[WS(is, 1)]), ivs, &(xi[WS(is, 1)])); } } { V TI, Ty, T13, Tf, Tz, Tk; TI = VADD(Tw, Tx); Ty = VSUB(Tw, Tx); T13 = VADD(Tb, Te); Tf = VFNMS(LDK(KP500000000), Te, Tb); Tz = VSUB(Ti, Tj); Tk = VADD(Ti, Tj); { V T1d, T14, Tg, TE, TJ, TB, T15, Tl; T1d = VSUB(T12, T13); T14 = VADD(T12, T13); Tg = VADD(Ta, Tf); TE = VSUB(Ta, Tf); TJ = VADD(Tz, TA); TB = VSUB(Tz, TA); T15 = VADD(Th, Tk); Tl = VFNMS(LDK(KP500000000), Tk, Th); { V TM, TK, TS, TC, T1c, T17, Tr, TF, TL, T10; TM = VSUB(TI, TJ); TK = VADD(TI, TJ); TS = VFNMS(LDK(KP618033988), Ty, TB); TC = VFMA(LDK(KP618033988), TB, Ty); T1c = VSUB(T15, T16); T17 = VADD(T15, T16); Tr = VADD(Tl, Tq); TF = VSUB(Tl, Tq); TL = VFNMS(LDK(KP250000000), TK, TH); T10 = VMUL(LDK(KP866025403), VADD(TH, TK)); { V T1g, T1e, T1a, Tu, Ts, TU, TG, TV, TN, T19, T18, Tt, TZ; T1g = VMUL(LDK(KP951056516), VFMA(LDK(KP618033988), T1c, T1d)); T1e = VMUL(LDK(KP951056516), VFNMS(LDK(KP618033988), T1d, T1c)); T18 = VADD(T14, T17); T1a = VSUB(T14, T17); Tu = VSUB(Tg, Tr); Ts = VADD(Tg, Tr); TU = VFNMS(LDK(KP618033988), TE, TF); TG = VFMA(LDK(KP618033988), TF, TE); TV = VFNMS(LDK(KP559016994), TM, TL); TN = VFMA(LDK(KP559016994), TM, TL); ST(&(xo[0]), VADD(T11, T18), ovs, &(xo[0])); T19 = VFNMS(LDK(KP250000000), T18, T11); Tt = VFNMS(LDK(KP250000000), Ts, T5); TZ = VADD(T5, Ts); { V TW, TY, TQ, TO, T1b, T1f, TR, Tv; TW = VMUL(LDK(KP951056516), VFMA(LDK(KP910592997), TV, TU)); TY = VMUL(LDK(KP951056516), VFNMS(LDK(KP910592997), TV, TU)); TQ = VMUL(LDK(KP951056516), VFNMS(LDK(KP910592997), TN, TG)); TO = VMUL(LDK(KP951056516), VFMA(LDK(KP910592997), TN, TG)); T1b = VFNMS(LDK(KP559016994), T1a, T19); T1f = VFMA(LDK(KP559016994), T1a, T19); TR = VFNMS(LDK(KP559016994), Tu, Tt); Tv = VFMA(LDK(KP559016994), Tu, Tt); ST(&(xo[WS(os, 10)]), VFMAI(T10, TZ), ovs, &(xo[0])); ST(&(xo[WS(os, 5)]), VFNMSI(T10, TZ), ovs, &(xo[WS(os, 1)])); { V TT, TX, TP, TD; ST(&(xo[WS(os, 12)]), VFNMSI(T1e, T1b), ovs, &(xo[0])); ST(&(xo[WS(os, 3)]), VFMAI(T1e, T1b), ovs, &(xo[WS(os, 1)])); ST(&(xo[WS(os, 9)]), VFNMSI(T1g, T1f), ovs, &(xo[WS(os, 1)])); ST(&(xo[WS(os, 6)]), VFMAI(T1g, T1f), ovs, &(xo[0])); TT = VFNMS(LDK(KP823639103), TS, TR); TX = VFMA(LDK(KP823639103), TS, TR); TP = VFMA(LDK(KP823639103), TC, Tv); TD = VFNMS(LDK(KP823639103), TC, Tv); ST(&(xo[WS(os, 13)]), VFMAI(TW, TT), ovs, &(xo[WS(os, 1)])); ST(&(xo[WS(os, 2)]), VFNMSI(TW, TT), ovs, &(xo[0])); ST(&(xo[WS(os, 8)]), VFMAI(TY, TX), ovs, &(xo[0])); ST(&(xo[WS(os, 7)]), VFNMSI(TY, TX), ovs, &(xo[WS(os, 1)])); ST(&(xo[WS(os, 11)]), VFMAI(TQ, TP), ovs, &(xo[WS(os, 1)])); ST(&(xo[WS(os, 4)]), VFNMSI(TQ, TP), ovs, &(xo[0])); ST(&(xo[WS(os, 14)]), VFNMSI(TO, TD), ovs, &(xo[0])); ST(&(xo[WS(os, 1)]), VFMAI(TO, TD), ovs, &(xo[WS(os, 1)])); } } } } } } } } VLEAVE(); }
static void t3fv_20(R *ri, R *ii, const R *W, stride rs, INT mb, INT me, INT ms) { DVK(KP559016994, +0.559016994374947424102293417182819058860154590); DVK(KP951056516, +0.951056516295153572116439333379382143405698634); DVK(KP618033988, +0.618033988749894848204586834365638117720309180); DVK(KP250000000, +0.250000000000000000000000000000000000000000000); { INT m; R *x; x = ri; for (m = mb, W = W + (mb * ((TWVL / VL) * 8)); m < me; m = m + VL, x = x + (VL * ms), W = W + (TWVL * 8), MAKE_VOLATILE_STRIDE(20, rs)) { V T1k, T1w, T1r, T1z, T1o, T1y, T1v, T1h; { V T2, T8, T3, Td; T2 = LDW(&(W[0])); T8 = LDW(&(W[TWVL * 2])); T3 = LDW(&(W[TWVL * 4])); Td = LDW(&(W[TWVL * 6])); { V T7, TM, T1F, T23, T1p, Tp, T1j, T27, T1P, T1I, T1i, T1L, T28, T1S, T1q; V TE, T1n, T1d, T26, T2e; { V T1, TK, T5, TH; T1 = LD(&(x[0]), ms, &(x[0])); TK = LD(&(x[WS(rs, 15)]), ms, &(x[WS(rs, 1)])); T5 = LD(&(x[WS(rs, 10)]), ms, &(x[0])); TH = LD(&(x[WS(rs, 5)]), ms, &(x[WS(rs, 1)])); { V TA, Tx, TU, T1O, T14, Th, T1G, T1R, T1b, T1J, To, Ts, TV, Tv, TO; V TQ, TT, Ty, TB; { V Tq, Tt, T17, T1a, Tk, Tn; { V Tl, Ti, T15, T18, TZ, Tc, T6, Tb, Tf, T10, T12, TL; { V TJ, Ta, T9, T4; Ta = LD(&(x[WS(rs, 4)]), ms, &(x[0])); TA = VZMULJ(T2, T8); T9 = VZMUL(T2, T8); Tx = VZMUL(T8, T3); Tl = VZMULJ(T8, T3); T4 = VZMUL(T2, T3); Tq = VZMULJ(T2, T3); Tt = VZMULJ(T2, Td); Ti = VZMULJ(T8, Td); T15 = VZMULJ(TA, Td); T18 = VZMULJ(TA, T3); TU = VZMUL(TA, T3); TJ = VZMULJ(T9, Td); TZ = VZMUL(T9, T3); Tc = VZMULJ(T9, T3); T6 = VZMULJ(T4, T5); Tb = VZMULJ(T9, Ta); Tf = LD(&(x[WS(rs, 14)]), ms, &(x[0])); T10 = LD(&(x[WS(rs, 13)]), ms, &(x[WS(rs, 1)])); T12 = LD(&(x[WS(rs, 3)]), ms, &(x[WS(rs, 1)])); TL = VZMULJ(TJ, TK); } { V T1D, T11, T13, T19, T1E, Tg, T16, TI, Te, Tj, Tm; T16 = LD(&(x[WS(rs, 17)]), ms, &(x[WS(rs, 1)])); TI = VZMULJ(Tc, TH); Te = VZMULJ(Tc, Td); T7 = VSUB(T1, T6); T1D = VADD(T1, T6); T11 = VZMULJ(TZ, T10); T13 = VZMULJ(T8, T12); T19 = LD(&(x[WS(rs, 7)]), ms, &(x[WS(rs, 1)])); T17 = VZMULJ(T15, T16); TM = VSUB(TI, TL); T1E = VADD(TI, TL); Tg = VZMULJ(Te, Tf); Tj = LD(&(x[WS(rs, 16)]), ms, &(x[0])); Tm = LD(&(x[WS(rs, 6)]), ms, &(x[0])); T1O = VADD(T11, T13); T14 = VSUB(T11, T13); T1a = VZMULJ(T18, T19); T1F = VSUB(T1D, T1E); T23 = VADD(T1D, T1E); Th = VSUB(Tb, Tg); T1G = VADD(Tb, Tg); Tk = VZMULJ(Ti, Tj); Tn = VZMULJ(Tl, Tm); } } { V Tr, Tu, TN, TP, TS; Tr = LD(&(x[WS(rs, 8)]), ms, &(x[0])); T1R = VADD(T17, T1a); T1b = VSUB(T17, T1a); Tu = LD(&(x[WS(rs, 18)]), ms, &(x[0])); TN = LD(&(x[WS(rs, 9)]), ms, &(x[WS(rs, 1)])); TP = LD(&(x[WS(rs, 19)]), ms, &(x[WS(rs, 1)])); TS = LD(&(x[WS(rs, 1)]), ms, &(x[WS(rs, 1)])); T1J = VADD(Tk, Tn); To = VSUB(Tk, Tn); Ts = VZMULJ(Tq, Tr); TV = LD(&(x[WS(rs, 11)]), ms, &(x[WS(rs, 1)])); Tv = VZMULJ(Tt, Tu); TO = VZMULJ(T3, TN); TQ = VZMULJ(Td, TP); TT = VZMULJ(T2, TS); Ty = LD(&(x[WS(rs, 12)]), ms, &(x[0])); TB = LD(&(x[WS(rs, 2)]), ms, &(x[0])); } } { V T1N, Tw, T1H, TR, Tz, TC, T1c, TX, T1K, TW; T1p = VSUB(Th, To); Tp = VADD(Th, To); TW = VZMULJ(TU, TV); T1N = VADD(Ts, Tv); Tw = VSUB(Ts, Tv); T1H = VADD(TO, TQ); TR = VSUB(TO, TQ); Tz = VZMULJ(Tx, Ty); TC = VZMULJ(TA, TB); T1j = VSUB(T1b, T14); T1c = VADD(T14, T1b); TX = VSUB(TT, TW); T1K = VADD(TT, TW); T27 = VADD(T1N, T1O); T1P = VSUB(T1N, T1O); { V TD, T1Q, T24, TY, T25; TD = VSUB(Tz, TC); T1Q = VADD(Tz, TC); T1I = VSUB(T1G, T1H); T24 = VADD(T1G, T1H); TY = VADD(TR, TX); T1i = VSUB(TX, TR); T25 = VADD(T1J, T1K); T1L = VSUB(T1J, T1K); T28 = VADD(T1Q, T1R); T1S = VSUB(T1Q, T1R); T1q = VSUB(Tw, TD); TE = VADD(Tw, TD); T1n = VSUB(T1c, TY); T1d = VADD(TY, T1c); T26 = VADD(T24, T25); T2e = VSUB(T24, T25); } } } } { V T1M, T1Z, T1Y, T1T, T29, T2f, T1g, TF, T1m, T1e; T1M = VADD(T1I, T1L); T1Z = VSUB(T1I, T1L); T1Y = VSUB(T1P, T1S); T1T = VADD(T1P, T1S); T29 = VADD(T27, T28); T2f = VSUB(T27, T28); T1g = VSUB(Tp, TE); TF = VADD(Tp, TE); T1m = VFNMS(LDK(KP250000000), T1d, TM); T1e = VADD(TM, T1d); { V T1W, T2c, T1f, T2i, T2g, T22, T20, T1V, T2b, T1U, T2a, TG; T1k = VFMA(LDK(KP618033988), T1j, T1i); T1w = VFNMS(LDK(KP618033988), T1i, T1j); T1W = VSUB(T1M, T1T); T1U = VADD(T1M, T1T); T2c = VSUB(T26, T29); T2a = VADD(T26, T29); T1f = VFNMS(LDK(KP250000000), TF, T7); TG = VADD(T7, TF); T2i = VMUL(LDK(KP951056516), VFNMS(LDK(KP618033988), T2e, T2f)); T2g = VMUL(LDK(KP951056516), VFMA(LDK(KP618033988), T2f, T2e)); T22 = VMUL(LDK(KP951056516), VFMA(LDK(KP618033988), T1Y, T1Z)); T20 = VMUL(LDK(KP951056516), VFNMS(LDK(KP618033988), T1Z, T1Y)); ST(&(x[WS(rs, 10)]), VADD(T1F, T1U), ms, &(x[0])); T1V = VFNMS(LDK(KP250000000), T1U, T1F); ST(&(x[0]), VADD(T23, T2a), ms, &(x[0])); T2b = VFNMS(LDK(KP250000000), T2a, T23); ST(&(x[WS(rs, 15)]), VFMAI(T1e, TG), ms, &(x[WS(rs, 1)])); ST(&(x[WS(rs, 5)]), VFNMSI(T1e, TG), ms, &(x[WS(rs, 1)])); T1r = VFMA(LDK(KP618033988), T1q, T1p); T1z = VFNMS(LDK(KP618033988), T1p, T1q); { V T21, T1X, T2h, T2d; T21 = VFMA(LDK(KP559016994), T1W, T1V); T1X = VFNMS(LDK(KP559016994), T1W, T1V); T2h = VFNMS(LDK(KP559016994), T2c, T2b); T2d = VFMA(LDK(KP559016994), T2c, T2b); ST(&(x[WS(rs, 18)]), VFNMSI(T20, T1X), ms, &(x[0])); ST(&(x[WS(rs, 2)]), VFMAI(T20, T1X), ms, &(x[0])); ST(&(x[WS(rs, 14)]), VFMAI(T22, T21), ms, &(x[0])); ST(&(x[WS(rs, 6)]), VFNMSI(T22, T21), ms, &(x[0])); ST(&(x[WS(rs, 16)]), VFNMSI(T2g, T2d), ms, &(x[0])); ST(&(x[WS(rs, 4)]), VFMAI(T2g, T2d), ms, &(x[0])); ST(&(x[WS(rs, 12)]), VFMAI(T2i, T2h), ms, &(x[0])); ST(&(x[WS(rs, 8)]), VFNMSI(T2i, T2h), ms, &(x[0])); T1o = VFNMS(LDK(KP559016994), T1n, T1m); T1y = VFMA(LDK(KP559016994), T1n, T1m); T1v = VFNMS(LDK(KP559016994), T1g, T1f); T1h = VFMA(LDK(KP559016994), T1g, T1f); } } } } } { V T1C, T1A, T1s, T1u, T1l, T1t, T1B, T1x; T1C = VFMA(LDK(KP951056516), T1z, T1y); T1A = VFNMS(LDK(KP951056516), T1z, T1y); T1s = VFMA(LDK(KP951056516), T1r, T1o); T1u = VFNMS(LDK(KP951056516), T1r, T1o); T1l = VFMA(LDK(KP951056516), T1k, T1h); T1t = VFNMS(LDK(KP951056516), T1k, T1h); T1B = VFMA(LDK(KP951056516), T1w, T1v); T1x = VFNMS(LDK(KP951056516), T1w, T1v); ST(&(x[WS(rs, 11)]), VFMAI(T1u, T1t), ms, &(x[WS(rs, 1)])); ST(&(x[WS(rs, 9)]), VFNMSI(T1u, T1t), ms, &(x[WS(rs, 1)])); ST(&(x[WS(rs, 19)]), VFMAI(T1s, T1l), ms, &(x[WS(rs, 1)])); ST(&(x[WS(rs, 1)]), VFNMSI(T1s, T1l), ms, &(x[WS(rs, 1)])); ST(&(x[WS(rs, 3)]), VFMAI(T1A, T1x), ms, &(x[WS(rs, 1)])); ST(&(x[WS(rs, 17)]), VFNMSI(T1A, T1x), ms, &(x[WS(rs, 1)])); ST(&(x[WS(rs, 7)]), VFMAI(T1C, T1B), ms, &(x[WS(rs, 1)])); ST(&(x[WS(rs, 13)]), VFNMSI(T1C, T1B), ms, &(x[WS(rs, 1)])); } } } VLEAVE(); }