static void t1bv_15(R *ri, R *ii, const R *W, stride rs, INT mb, INT me, INT ms) { 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 m; R *x; x = ii; for (m = mb, W = W + (mb * ((TWVL / VL) * 28)); m < me; m = m + VL, x = x + (VL * ms), W = W + (TWVL * 28), MAKE_VOLATILE_STRIDE(15, rs)) { V Tq, Ty, Th, TV, TK, Ts, T1f, T7, Tu, TA, TC, Tj, Tk, T1g, Tf; { V T1, T4, T2, T9, Te; T1 = LD(&(x[0]), ms, &(x[0])); T4 = LD(&(x[WS(rs, 10)]), ms, &(x[0])); T2 = LD(&(x[WS(rs, 5)]), ms, &(x[WS(rs, 1)])); { V T8, Tp, Tx, Tg; T8 = LD(&(x[WS(rs, 3)]), ms, &(x[WS(rs, 1)])); Tp = LD(&(x[WS(rs, 6)]), ms, &(x[0])); Tx = LD(&(x[WS(rs, 9)]), ms, &(x[WS(rs, 1)])); Tg = LD(&(x[WS(rs, 12)]), ms, &(x[0])); { V Tb, Td, Tr, T6, Tt, Tz, TB, Ti; { V T5, T3, Ta, Tc; Ta = LD(&(x[WS(rs, 8)]), ms, &(x[0])); Tc = LD(&(x[WS(rs, 13)]), ms, &(x[WS(rs, 1)])); T5 = BYTW(&(W[TWVL * 18]), T4); T3 = BYTW(&(W[TWVL * 8]), T2); T9 = BYTW(&(W[TWVL * 4]), T8); Tq = BYTW(&(W[TWVL * 10]), Tp); Ty = BYTW(&(W[TWVL * 16]), Tx); Th = BYTW(&(W[TWVL * 22]), Tg); Tb = BYTW(&(W[TWVL * 14]), Ta); Td = BYTW(&(W[TWVL * 24]), Tc); Tr = LD(&(x[WS(rs, 11)]), ms, &(x[WS(rs, 1)])); TV = VSUB(T3, T5); T6 = VADD(T3, T5); Tt = LD(&(x[WS(rs, 1)]), ms, &(x[WS(rs, 1)])); } Tz = LD(&(x[WS(rs, 14)]), ms, &(x[0])); TB = LD(&(x[WS(rs, 4)]), ms, &(x[0])); Ti = LD(&(x[WS(rs, 2)]), ms, &(x[0])); Te = VADD(Tb, Td); TK = VSUB(Tb, Td); Ts = BYTW(&(W[TWVL * 20]), Tr); T1f = VADD(T1, T6); T7 = VFNMS(LDK(KP500000000), T6, T1); Tu = BYTW(&(W[0]), Tt); TA = BYTW(&(W[TWVL * 26]), Tz); TC = BYTW(&(W[TWVL * 6]), TB); Tj = BYTW(&(W[TWVL * 2]), Ti); Tk = LD(&(x[WS(rs, 7)]), ms, &(x[WS(rs, 1)])); } } T1g = VADD(T9, Te); Tf = VFNMS(LDK(KP500000000), Te, T9); } { V Tv, TN, TD, TO, Tl; Tv = VADD(Ts, Tu); TN = VSUB(Ts, Tu); TD = VADD(TA, TC); TO = VSUB(TA, TC); Tl = BYTW(&(W[TWVL * 12]), Tk); { V Tw, T1j, TX, TP, TE, T1k, TL, Tm; Tw = VFNMS(LDK(KP500000000), Tv, Tq); T1j = VADD(Tq, Tv); TX = VADD(TN, TO); TP = VSUB(TN, TO); TE = VFNMS(LDK(KP500000000), TD, Ty); T1k = VADD(Ty, TD); TL = VSUB(Tj, Tl); Tm = VADD(Tj, Tl); { V TT, TF, T1q, T1l, TW, TM, T1h, Tn; TT = VSUB(Tw, TE); TF = VADD(Tw, TE); T1q = VSUB(T1j, T1k); T1l = VADD(T1j, T1k); TW = VADD(TK, TL); TM = VSUB(TK, TL); T1h = VADD(Th, Tm); Tn = VFNMS(LDK(KP500000000), Tm, Th); { V T10, TY, T16, TQ, T1r, T1i, TS, To, TZ, T1e; T10 = VSUB(TW, TX); TY = VADD(TW, TX); T16 = VFNMS(LDK(KP618033988), TM, TP); TQ = VFMA(LDK(KP618033988), TP, TM); T1r = VSUB(T1g, T1h); T1i = VADD(T1g, T1h); TS = VSUB(Tf, Tn); To = VADD(Tf, Tn); TZ = VFNMS(LDK(KP250000000), TY, TV); T1e = VMUL(LDK(KP866025403), VADD(TV, TY)); { V T1u, T1s, T1o, T18, TU, TG, TI, T19, T11, T1n, T1m; T1u = VMUL(LDK(KP951056516), VFMA(LDK(KP618033988), T1q, T1r)); T1s = VMUL(LDK(KP951056516), VFNMS(LDK(KP618033988), T1r, T1q)); T1m = VADD(T1i, T1l); T1o = VSUB(T1i, T1l); T18 = VFNMS(LDK(KP618033988), TS, TT); TU = VFMA(LDK(KP618033988), TT, TS); TG = VADD(To, TF); TI = VSUB(To, TF); T19 = VFNMS(LDK(KP559016994), T10, TZ); T11 = VFMA(LDK(KP559016994), T10, TZ); ST(&(x[0]), VADD(T1f, T1m), ms, &(x[0])); T1n = VFNMS(LDK(KP250000000), T1m, T1f); { V T1a, T1c, T14, T12, T1p, T1t, T15, TJ, T1d, TH; T1d = VADD(T7, TG); TH = VFNMS(LDK(KP250000000), TG, T7); T1a = VMUL(LDK(KP951056516), VFMA(LDK(KP910592997), T19, T18)); T1c = VMUL(LDK(KP951056516), VFNMS(LDK(KP910592997), T19, T18)); T14 = VMUL(LDK(KP951056516), VFNMS(LDK(KP910592997), T11, TU)); T12 = VMUL(LDK(KP951056516), VFMA(LDK(KP910592997), T11, TU)); T1p = VFNMS(LDK(KP559016994), T1o, T1n); T1t = VFMA(LDK(KP559016994), T1o, T1n); ST(&(x[WS(rs, 10)]), VFMAI(T1e, T1d), ms, &(x[0])); ST(&(x[WS(rs, 5)]), VFNMSI(T1e, T1d), ms, &(x[WS(rs, 1)])); T15 = VFNMS(LDK(KP559016994), TI, TH); TJ = VFMA(LDK(KP559016994), TI, TH); { V T17, T1b, T13, TR; ST(&(x[WS(rs, 12)]), VFNMSI(T1s, T1p), ms, &(x[0])); ST(&(x[WS(rs, 3)]), VFMAI(T1s, T1p), ms, &(x[WS(rs, 1)])); ST(&(x[WS(rs, 9)]), VFNMSI(T1u, T1t), ms, &(x[WS(rs, 1)])); ST(&(x[WS(rs, 6)]), VFMAI(T1u, T1t), ms, &(x[0])); T17 = VFNMS(LDK(KP823639103), T16, T15); T1b = VFMA(LDK(KP823639103), T16, T15); T13 = VFMA(LDK(KP823639103), TQ, TJ); TR = VFNMS(LDK(KP823639103), TQ, TJ); ST(&(x[WS(rs, 13)]), VFMAI(T1a, T17), ms, &(x[WS(rs, 1)])); ST(&(x[WS(rs, 2)]), VFNMSI(T1a, T17), ms, &(x[0])); ST(&(x[WS(rs, 8)]), VFMAI(T1c, T1b), ms, &(x[0])); ST(&(x[WS(rs, 7)]), VFNMSI(T1c, T1b), ms, &(x[WS(rs, 1)])); ST(&(x[WS(rs, 11)]), VFMAI(T14, T13), ms, &(x[WS(rs, 1)])); ST(&(x[WS(rs, 4)]), VFNMSI(T14, T13), ms, &(x[0])); ST(&(x[WS(rs, 14)]), VFNMSI(T12, TR), ms, &(x[0])); ST(&(x[WS(rs, 1)]), VFMAI(T12, TR), ms, &(x[WS(rs, 1)])); } } } } } } } } } VLEAVE(); }
static void t1buv_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(KP150383733, +0.150383733180435296639271897612501926072238258); DVK(KP342020143, +0.342020143325668733044099614682259580763083368); DVK(KP813797681, +0.813797681349373692844693217248393223289101568); DVK(KP984807753, +0.984807753012208059366743024589523013670643252); DVK(KP500000000, +0.500000000000000000000000000000000000000000000); DVK(KP866025403, +0.866025403784438646763723170752936183471402627); { INT m; R *x; x = ii; 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(9, rs)) { V T1, T6, Tu, Tg, Tf, TD, Tq, Tp, TE; T1 = LD(&(x[0]), ms, &(x[0])); { V T3, T5, T2, T4; T2 = LD(&(x[WS(rs, 3)]), ms, &(x[WS(rs, 1)])); T3 = BYTW(&(W[TWVL * 4]), T2); T4 = LD(&(x[WS(rs, 6)]), ms, &(x[0])); T5 = BYTW(&(W[TWVL * 10]), T4); T6 = VADD(T3, T5); Tu = VMUL(LDK(KP866025403), VSUB(T3, T5)); } { V T9, Td, Tb, T8, Tc, Ta, Te; T8 = LD(&(x[WS(rs, 1)]), ms, &(x[WS(rs, 1)])); T9 = BYTW(&(W[0]), T8); Tc = LD(&(x[WS(rs, 7)]), ms, &(x[WS(rs, 1)])); Td = BYTW(&(W[TWVL * 12]), Tc); Ta = LD(&(x[WS(rs, 4)]), ms, &(x[0])); Tb = BYTW(&(W[TWVL * 6]), Ta); Tg = VSUB(Tb, Td); Te = VADD(Tb, Td); Tf = VFNMS(LDK(KP500000000), Te, T9); TD = VADD(T9, Te); } { V Tj, Tn, Tl, Ti, Tm, Tk, To; Ti = LD(&(x[WS(rs, 2)]), ms, &(x[0])); Tj = BYTW(&(W[TWVL * 2]), Ti); Tm = LD(&(x[WS(rs, 8)]), ms, &(x[0])); Tn = BYTW(&(W[TWVL * 14]), Tm); Tk = LD(&(x[WS(rs, 5)]), ms, &(x[WS(rs, 1)])); Tl = BYTW(&(W[TWVL * 8]), Tk); Tq = VSUB(Tl, Tn); To = VADD(Tl, Tn); Tp = VFNMS(LDK(KP500000000), To, Tj); TE = VADD(Tj, To); } { V TF, TG, TH, TI; TF = VBYI(VMUL(LDK(KP866025403), VSUB(TD, TE))); TG = VADD(T1, T6); TH = VADD(TD, TE); TI = VFNMS(LDK(KP500000000), TH, TG); ST(&(x[WS(rs, 3)]), VADD(TF, TI), ms, &(x[WS(rs, 1)])); ST(&(x[0]), VADD(TG, TH), ms, &(x[0])); ST(&(x[WS(rs, 6)]), VSUB(TI, TF), ms, &(x[0])); } { V TC, Tv, Tw, Tx, Th, Tr, Ts, T7, TB; TC = VBYI(VSUB(VFMA(LDK(KP984807753), Tf, VFMA(LDK(KP813797681), Tq, VFNMS(LDK(KP150383733), Tg, VMUL(LDK(KP342020143), Tp)))), Tu)); Tv = VFMA(LDK(KP663413948), Tg, VMUL(LDK(KP642787609), Tf)); Tw = VFMA(LDK(KP150383733), Tq, VMUL(LDK(KP984807753), Tp)); Tx = VADD(Tv, Tw); Th = VFNMS(LDK(KP556670399), Tg, VMUL(LDK(KP766044443), Tf)); Tr = VFNMS(LDK(KP852868531), Tq, VMUL(LDK(KP173648177), Tp)); Ts = VADD(Th, Tr); T7 = VFNMS(LDK(KP500000000), T6, T1); TB = VFMA(LDK(KP852868531), Tg, VFMA(LDK(KP173648177), Tf, VFMA(LDK(KP296198132), Tq, VFNMS(LDK(KP939692620), Tp, T7)))); ST(&(x[WS(rs, 7)]), VSUB(TB, TC), ms, &(x[WS(rs, 1)])); ST(&(x[WS(rs, 2)]), VADD(TB, TC), ms, &(x[0])); { V Tt, Ty, Tz, TA; Tt = VADD(T7, Ts); Ty = VBYI(VADD(Tu, Tx)); ST(&(x[WS(rs, 8)]), VSUB(Tt, Ty), ms, &(x[0])); ST(&(x[WS(rs, 1)]), VADD(Tt, Ty), ms, &(x[WS(rs, 1)])); Tz = VBYI(VADD(Tu, VFNMS(LDK(KP500000000), Tx, VMUL(LDK(KP866025403), VSUB(Th, Tr))))); TA = VFMA(LDK(KP866025403), VSUB(Tw, Tv), VFNMS(LDK(KP500000000), Ts, T7)); ST(&(x[WS(rs, 4)]), VADD(Tz, TA), ms, &(x[0])); ST(&(x[WS(rs, 5)]), VSUB(TA, Tz), ms, &(x[WS(rs, 1)])); } } } } VLEAVE(); }
static void hc2cfdftv_10(R *Rp, R *Ip, R *Rm, R *Im, const R *W, stride rs, INT mb, INT me, INT ms) { DVK(KP559016994, +0.559016994374947424102293417182819058860154590); DVK(KP500000000, +0.500000000000000000000000000000000000000000000); DVK(KP250000000, +0.250000000000000000000000000000000000000000000); DVK(KP618033988, +0.618033988749894848204586834365638117720309180); DVK(KP951056516, +0.951056516295153572116439333379382143405698634); { INT m; for (m = mb, W = W + ((mb - 1) * ((TWVL / VL) * 18)); m < me; m = m + VL, Rp = Rp + (VL * ms), Ip = Ip + (VL * ms), Rm = Rm - (VL * ms), Im = Im - (VL * ms), W = W + (TWVL * 18), MAKE_VOLATILE_STRIDE(40, rs)) { V T5, T6, Tw, Tr, Tc, Tj, Tl, Tm, Tk, Ts, Tg, Ty, T3, T4, T1; V T2, Tv, Tq, Ta, Tb, T9, Ti, Te, Tf, Td, Tx, Tn, Tt, Th, TQ; V TT, Tz, T7, TR, To, Tu, TU; T1 = LD(&(Rp[0]), ms, &(Rp[0])); T2 = LD(&(Rm[0]), -ms, &(Rm[0])); Tv = LDW(&(W[0])); T5 = LD(&(Rp[WS(rs, 2)]), ms, &(Rp[0])); T6 = LD(&(Rm[WS(rs, 2)]), -ms, &(Rm[0])); Tq = LDW(&(W[TWVL * 6])); Ta = LD(&(Rp[WS(rs, 1)]), ms, &(Rp[WS(rs, 1)])); Tb = LD(&(Rm[WS(rs, 1)]), -ms, &(Rm[WS(rs, 1)])); T9 = LDW(&(W[TWVL * 2])); Ti = LDW(&(W[TWVL * 4])); Tw = VZMULIJ(Tv, VFNMSCONJ(T2, T1)); Te = LD(&(Rp[WS(rs, 3)]), ms, &(Rp[WS(rs, 1)])); Tf = LD(&(Rm[WS(rs, 3)]), -ms, &(Rm[WS(rs, 1)])); Tr = VZMULJ(Tq, VFMACONJ(T6, T5)); Td = LDW(&(W[TWVL * 12])); Tx = LDW(&(W[TWVL * 10])); Tc = VZMULJ(T9, VFMACONJ(Tb, Ta)); Tj = VZMULIJ(Ti, VFNMSCONJ(Tb, Ta)); Tl = LD(&(Rp[WS(rs, 4)]), ms, &(Rp[0])); Tm = LD(&(Rm[WS(rs, 4)]), -ms, &(Rm[0])); Tk = LDW(&(W[TWVL * 14])); Ts = LDW(&(W[TWVL * 16])); Tg = VZMULIJ(Td, VFNMSCONJ(Tf, Te)); Ty = VZMULJ(Tx, VFMACONJ(Tf, Te)); T3 = VFMACONJ(T2, T1); T4 = LDW(&(W[TWVL * 8])); Tn = VZMULJ(Tk, VFMACONJ(Tm, Tl)); Tt = VZMULIJ(Ts, VFNMSCONJ(Tm, Tl)); Th = VSUB(Tc, Tg); TQ = VADD(Tc, Tg); TT = VADD(Tw, Ty); Tz = VSUB(Tw, Ty); T7 = VZMULIJ(T4, VFNMSCONJ(T6, T5)); TR = VADD(Tj, Tn); To = VSUB(Tj, Tn); Tu = VSUB(Tr, Tt); TU = VADD(Tr, Tt); { V TP, T8, TS, T11, Tp, TH, TA, TG, TV, T12, TE, TB, TM, TI, TZ; V TW, T17, T13, TD, TC, TY, TX, TL, TF, T10, T16, TN, TO, TK, TJ; V T18, T19, T15, T14; TP = VADD(T3, T7); T8 = VSUB(T3, T7); TS = VADD(TQ, TR); T11 = VSUB(TQ, TR); Tp = VSUB(Th, To); TH = VADD(Th, To); TA = VSUB(Tu, Tz); TG = VADD(Tz, Tu); TV = VADD(TT, TU); T12 = VSUB(TU, TT); TE = VSUB(Tp, TA); TB = VADD(Tp, TA); TM = VMUL(LDK(KP951056516), VFMA(LDK(KP618033988), TG, TH)); TI = VMUL(LDK(KP951056516), VFNMS(LDK(KP618033988), TH, TG)); TZ = VSUB(TS, TV); TW = VADD(TS, TV); T17 = VMUL(LDK(KP951056516), VFNMS(LDK(KP618033988), T11, T12)); T13 = VMUL(LDK(KP951056516), VFMA(LDK(KP618033988), T12, T11)); TD = VFNMS(LDK(KP250000000), TB, T8); TC = VMUL(LDK(KP500000000), VADD(T8, TB)); TY = VFNMS(LDK(KP250000000), TW, TP); TX = VCONJ(VMUL(LDK(KP500000000), VADD(TP, TW))); TL = VFMA(LDK(KP559016994), TE, TD); TF = VFNMS(LDK(KP559016994), TE, TD); ST(&(Rp[0]), TC, ms, &(Rp[0])); T10 = VFMA(LDK(KP559016994), TZ, TY); T16 = VFNMS(LDK(KP559016994), TZ, TY); ST(&(Rm[WS(rs, 4)]), TX, -ms, &(Rm[0])); TN = VCONJ(VMUL(LDK(KP500000000), VFNMSI(TM, TL))); TO = VMUL(LDK(KP500000000), VFMAI(TM, TL)); TK = VMUL(LDK(KP500000000), VFMAI(TI, TF)); TJ = VCONJ(VMUL(LDK(KP500000000), VFNMSI(TI, TF))); T18 = VMUL(LDK(KP500000000), VFNMSI(T17, T16)); T19 = VCONJ(VMUL(LDK(KP500000000), VFMAI(T17, T16))); T15 = VCONJ(VMUL(LDK(KP500000000), VFMAI(T13, T10))); T14 = VMUL(LDK(KP500000000), VFNMSI(T13, T10)); ST(&(Rm[WS(rs, 3)]), TN, -ms, &(Rm[WS(rs, 1)])); ST(&(Rp[WS(rs, 4)]), TO, ms, &(Rp[0])); ST(&(Rp[WS(rs, 2)]), TK, ms, &(Rp[0])); ST(&(Rm[WS(rs, 1)]), TJ, -ms, &(Rm[WS(rs, 1)])); ST(&(Rp[WS(rs, 3)]), T18, ms, &(Rp[WS(rs, 1)])); ST(&(Rm[WS(rs, 2)]), T19, -ms, &(Rm[0])); ST(&(Rm[0]), T15, -ms, &(Rm[0])); ST(&(Rp[WS(rs, 1)]), T14, ms, &(Rp[WS(rs, 1)])); } } } VLEAVE(); }
static void t2bv_8(R *ri, R *ii, const R *W, stride rs, INT mb, INT me, INT ms) { DVK(KP707106781, +0.707106781186547524400844362104849039284835938); { INT m; R *x; x = ii; for (m = mb, W = W + (mb * ((TWVL / VL) * 14)); m < me; m = m + VL, x = x + (VL * ms), W = W + (TWVL * 14), MAKE_VOLATILE_STRIDE(rs)) { V T1, T2, Th, Tj, T5, T7, Ta, Tc; T1 = LD(&(x[0]), ms, &(x[0])); T2 = LD(&(x[WS(rs, 4)]), ms, &(x[0])); Th = LD(&(x[WS(rs, 2)]), ms, &(x[0])); Tj = LD(&(x[WS(rs, 6)]), ms, &(x[0])); T5 = LD(&(x[WS(rs, 1)]), ms, &(x[WS(rs, 1)])); T7 = LD(&(x[WS(rs, 5)]), ms, &(x[WS(rs, 1)])); Ta = LD(&(x[WS(rs, 7)]), ms, &(x[WS(rs, 1)])); Tc = LD(&(x[WS(rs, 3)]), ms, &(x[WS(rs, 1)])); { V T3, Ti, Tk, T6, T8, Tb, Td; T3 = BYTW(&(W[TWVL * 6]), T2); Ti = BYTW(&(W[TWVL * 2]), Th); Tk = BYTW(&(W[TWVL * 10]), Tj); T6 = BYTW(&(W[0]), T5); T8 = BYTW(&(W[TWVL * 8]), T7); Tb = BYTW(&(W[TWVL * 12]), Ta); Td = BYTW(&(W[TWVL * 4]), Tc); { V Tq, T4, Tr, Tl, Tt, T9, Tu, Te, Tw, Ts; Tq = VADD(T1, T3); T4 = VSUB(T1, T3); Tr = VADD(Ti, Tk); Tl = VSUB(Ti, Tk); Tt = VADD(T6, T8); T9 = VSUB(T6, T8); Tu = VADD(Tb, Td); Te = VSUB(Tb, Td); Tw = VADD(Tq, Tr); Ts = VSUB(Tq, Tr); { V Tx, Tv, Tm, Tf; Tx = VADD(Tt, Tu); Tv = VSUB(Tt, Tu); Tm = VSUB(T9, Te); Tf = VADD(T9, Te); { V Tp, Tn, To, Tg; ST(&(x[0]), VADD(Tw, Tx), ms, &(x[0])); ST(&(x[WS(rs, 4)]), VSUB(Tw, Tx), ms, &(x[0])); ST(&(x[WS(rs, 2)]), VFMAI(Tv, Ts), ms, &(x[0])); ST(&(x[WS(rs, 6)]), VFNMSI(Tv, Ts), ms, &(x[0])); Tp = VFMA(LDK(KP707106781), Tm, Tl); Tn = VFNMS(LDK(KP707106781), Tm, Tl); To = VFMA(LDK(KP707106781), Tf, T4); Tg = VFNMS(LDK(KP707106781), Tf, T4); ST(&(x[WS(rs, 1)]), VFMAI(Tp, To), ms, &(x[WS(rs, 1)])); ST(&(x[WS(rs, 7)]), VFNMSI(Tp, To), ms, &(x[WS(rs, 1)])); ST(&(x[WS(rs, 5)]), VFMAI(Tn, Tg), ms, &(x[WS(rs, 1)])); ST(&(x[WS(rs, 3)]), VFNMSI(Tn, Tg), ms, &(x[WS(rs, 1)])); } } } } } } VLEAVE(); }
static void n2bv_16(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs) { DVK(KP382683432, +0.382683432365089771728459984030398866761344562); DVK(KP923879532, +0.923879532511286756128183189396788286822416626); DVK(KP707106781, +0.707106781186547524400844362104849039284835938); { 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(32, is), MAKE_VOLATILE_STRIDE(32, os)) { V Tp, T13, Tu, TY, Tm, T14, Tv, TU, T7, T16, Tx, TN, Te, T17, Ty; V TQ; { V Tn, To, TX, Ts, Tt, TW; Tn = LD(&(xi[WS(is, 4)]), ivs, &(xi[0])); To = LD(&(xi[WS(is, 12)]), ivs, &(xi[0])); TX = VADD(Tn, To); Ts = LD(&(xi[0]), ivs, &(xi[0])); Tt = LD(&(xi[WS(is, 8)]), ivs, &(xi[0])); TW = VADD(Ts, Tt); Tp = VSUB(Tn, To); T13 = VADD(TW, TX); Tu = VSUB(Ts, Tt); TY = VSUB(TW, TX); } { V Ti, TS, Tl, TT; { V Tg, Th, Tj, Tk; Tg = LD(&(xi[WS(is, 2)]), ivs, &(xi[0])); Th = LD(&(xi[WS(is, 10)]), ivs, &(xi[0])); Ti = VSUB(Tg, Th); TS = VADD(Tg, Th); Tj = LD(&(xi[WS(is, 14)]), ivs, &(xi[0])); Tk = LD(&(xi[WS(is, 6)]), ivs, &(xi[0])); Tl = VSUB(Tj, Tk); TT = VADD(Tj, Tk); } Tm = VMUL(LDK(KP707106781), VSUB(Ti, Tl)); T14 = VADD(TS, TT); Tv = VMUL(LDK(KP707106781), VADD(Ti, Tl)); TU = VSUB(TS, TT); } { V T3, TL, T6, TM; { V T1, T2, T4, T5; T1 = LD(&(xi[WS(is, 1)]), ivs, &(xi[WS(is, 1)])); T2 = LD(&(xi[WS(is, 9)]), ivs, &(xi[WS(is, 1)])); T3 = VSUB(T1, T2); TL = VADD(T1, T2); T4 = LD(&(xi[WS(is, 5)]), ivs, &(xi[WS(is, 1)])); T5 = LD(&(xi[WS(is, 13)]), ivs, &(xi[WS(is, 1)])); T6 = VSUB(T4, T5); TM = VADD(T4, T5); } T7 = VFNMS(LDK(KP382683432), T6, VMUL(LDK(KP923879532), T3)); T16 = VADD(TL, TM); Tx = VFMA(LDK(KP382683432), T3, VMUL(LDK(KP923879532), T6)); TN = VSUB(TL, TM); } { V Ta, TO, Td, TP; { V T8, T9, Tb, Tc; T8 = LD(&(xi[WS(is, 15)]), ivs, &(xi[WS(is, 1)])); T9 = LD(&(xi[WS(is, 7)]), ivs, &(xi[WS(is, 1)])); Ta = VSUB(T8, T9); TO = VADD(T8, T9); Tb = LD(&(xi[WS(is, 3)]), ivs, &(xi[WS(is, 1)])); Tc = LD(&(xi[WS(is, 11)]), ivs, &(xi[WS(is, 1)])); Td = VSUB(Tb, Tc); TP = VADD(Tb, Tc); } Te = VFMA(LDK(KP923879532), Ta, VMUL(LDK(KP382683432), Td)); T17 = VADD(TO, TP); Ty = VFNMS(LDK(KP382683432), Ta, VMUL(LDK(KP923879532), Td)); TQ = VSUB(TO, TP); } { V T1b, T1c, T1d, T1e; { V T15, T18, T19, T1a; T15 = VSUB(T13, T14); T18 = VBYI(VSUB(T16, T17)); T1b = VSUB(T15, T18); STM2(&(xo[24]), T1b, ovs, &(xo[0])); T1c = VADD(T15, T18); STM2(&(xo[8]), T1c, ovs, &(xo[0])); T19 = VADD(T13, T14); T1a = VADD(T16, T17); T1d = VSUB(T19, T1a); STM2(&(xo[16]), T1d, ovs, &(xo[0])); T1e = VADD(T19, T1a); STM2(&(xo[0]), T1e, ovs, &(xo[0])); } { V T1f, T1g, T1h, T1i; { V TV, T11, T10, T12, TR, TZ; TR = VMUL(LDK(KP707106781), VSUB(TN, TQ)); TV = VBYI(VSUB(TR, TU)); T11 = VBYI(VADD(TU, TR)); TZ = VMUL(LDK(KP707106781), VADD(TN, TQ)); T10 = VSUB(TY, TZ); T12 = VADD(TY, TZ); T1f = VADD(TV, T10); STM2(&(xo[12]), T1f, ovs, &(xo[0])); T1g = VSUB(T12, T11); STM2(&(xo[28]), T1g, ovs, &(xo[0])); T1h = VSUB(T10, TV); STM2(&(xo[20]), T1h, ovs, &(xo[0])); T1i = VADD(T11, T12); STM2(&(xo[4]), T1i, ovs, &(xo[0])); } { V Tr, TB, TA, TC; { V Tf, Tq, Tw, Tz; Tf = VSUB(T7, Te); Tq = VSUB(Tm, Tp); Tr = VBYI(VSUB(Tf, Tq)); TB = VBYI(VADD(Tq, Tf)); Tw = VSUB(Tu, Tv); Tz = VSUB(Tx, Ty); TA = VSUB(Tw, Tz); TC = VADD(Tw, Tz); } { V T1j, T1k, T1l, T1m; T1j = VADD(Tr, TA); STM2(&(xo[10]), T1j, ovs, &(xo[2])); STN2(&(xo[8]), T1c, T1j, ovs); T1k = VSUB(TC, TB); STM2(&(xo[26]), T1k, ovs, &(xo[2])); STN2(&(xo[24]), T1b, T1k, ovs); T1l = VSUB(TA, Tr); STM2(&(xo[22]), T1l, ovs, &(xo[2])); STN2(&(xo[20]), T1h, T1l, ovs); T1m = VADD(TB, TC); STM2(&(xo[6]), T1m, ovs, &(xo[2])); STN2(&(xo[4]), T1i, T1m, ovs); } } { V TF, TJ, TI, TK; { V TD, TE, TG, TH; TD = VADD(Tu, Tv); TE = VADD(T7, Te); TF = VADD(TD, TE); TJ = VSUB(TD, TE); TG = VADD(Tp, Tm); TH = VADD(Tx, Ty); TI = VBYI(VADD(TG, TH)); TK = VBYI(VSUB(TH, TG)); } { V T1n, T1o, T1p, T1q; T1n = VSUB(TF, TI); STM2(&(xo[30]), T1n, ovs, &(xo[2])); STN2(&(xo[28]), T1g, T1n, ovs); T1o = VADD(TJ, TK); STM2(&(xo[14]), T1o, ovs, &(xo[2])); STN2(&(xo[12]), T1f, T1o, ovs); T1p = VADD(TF, TI); STM2(&(xo[2]), T1p, ovs, &(xo[2])); STN2(&(xo[0]), T1e, T1p, ovs); T1q = VSUB(TJ, TK); STM2(&(xo[18]), T1q, ovs, &(xo[2])); STN2(&(xo[16]), T1d, T1q, ovs); } } } } } } VLEAVE(); }
static void t2bv_20(R *ri, R *ii, const R *W, stride rs, INT mb, INT me, INT ms) { DVK(KP559016994, +0.559016994374947424102293417182819058860154590); DVK(KP618033988, +0.618033988749894848204586834365638117720309180); DVK(KP951056516, +0.951056516295153572116439333379382143405698634); DVK(KP250000000, +0.250000000000000000000000000000000000000000000); { INT m; R *x; x = ii; for (m = mb, W = W + (mb * ((TWVL / VL) * 38)); m < me; m = m + VL, x = x + (VL * ms), W = W + (TWVL * 38), MAKE_VOLATILE_STRIDE(20, rs)) { V T4, TX, T1m, T1K, TF, T14, T15, TQ, Tf, Tq, Tr, T1O, T1P, T1Q, T1w; V T1z, T1A, TY, TZ, T10, T1L, T1M, T1N, T1p, T1s, T1t, T1i, T1j; { V T1, TW, T3, TU, TV, T2, TT, T1k, T1l; T1 = LD(&(x[0]), ms, &(x[0])); TV = LD(&(x[WS(rs, 15)]), ms, &(x[WS(rs, 1)])); TW = BYTW(&(W[TWVL * 28]), TV); T2 = LD(&(x[WS(rs, 10)]), ms, &(x[0])); T3 = BYTW(&(W[TWVL * 18]), T2); TT = LD(&(x[WS(rs, 5)]), ms, &(x[WS(rs, 1)])); TU = BYTW(&(W[TWVL * 8]), TT); T4 = VSUB(T1, T3); TX = VSUB(TU, TW); T1k = VADD(T1, T3); T1l = VADD(TU, TW); T1m = VSUB(T1k, T1l); T1K = VADD(T1k, T1l); } { V T9, T1n, TK, T1v, TP, T1y, Te, T1q, Tk, T1u, Tz, T1o, TE, T1r, Tp; V T1x; { V T6, T8, T5, T7; T5 = LD(&(x[WS(rs, 4)]), ms, &(x[0])); T6 = BYTW(&(W[TWVL * 6]), T5); T7 = LD(&(x[WS(rs, 14)]), ms, &(x[0])); T8 = BYTW(&(W[TWVL * 26]), T7); T9 = VSUB(T6, T8); T1n = VADD(T6, T8); } { V TH, TJ, TG, TI; TG = LD(&(x[WS(rs, 13)]), ms, &(x[WS(rs, 1)])); TH = BYTW(&(W[TWVL * 24]), TG); TI = LD(&(x[WS(rs, 3)]), ms, &(x[WS(rs, 1)])); TJ = BYTW(&(W[TWVL * 4]), TI); TK = VSUB(TH, TJ); T1v = VADD(TH, TJ); } { V TM, TO, TL, TN; TL = LD(&(x[WS(rs, 17)]), ms, &(x[WS(rs, 1)])); TM = BYTW(&(W[TWVL * 32]), TL); TN = LD(&(x[WS(rs, 7)]), ms, &(x[WS(rs, 1)])); TO = BYTW(&(W[TWVL * 12]), TN); TP = VSUB(TM, TO); T1y = VADD(TM, TO); } { V Tb, Td, Ta, Tc; Ta = LD(&(x[WS(rs, 16)]), ms, &(x[0])); Tb = BYTW(&(W[TWVL * 30]), Ta); Tc = LD(&(x[WS(rs, 6)]), ms, &(x[0])); Td = BYTW(&(W[TWVL * 10]), Tc); Te = VSUB(Tb, Td); T1q = VADD(Tb, Td); } { V Th, Tj, Tg, Ti; Tg = LD(&(x[WS(rs, 8)]), ms, &(x[0])); Th = BYTW(&(W[TWVL * 14]), Tg); Ti = LD(&(x[WS(rs, 18)]), ms, &(x[0])); Tj = BYTW(&(W[TWVL * 34]), Ti); Tk = VSUB(Th, Tj); T1u = VADD(Th, Tj); } { V Tw, Ty, Tv, Tx; Tv = LD(&(x[WS(rs, 9)]), ms, &(x[WS(rs, 1)])); Tw = BYTW(&(W[TWVL * 16]), Tv); Tx = LD(&(x[WS(rs, 19)]), ms, &(x[WS(rs, 1)])); Ty = BYTW(&(W[TWVL * 36]), Tx); Tz = VSUB(Tw, Ty); T1o = VADD(Tw, Ty); } { V TB, TD, TA, TC; TA = LD(&(x[WS(rs, 1)]), ms, &(x[WS(rs, 1)])); TB = BYTW(&(W[0]), TA); TC = LD(&(x[WS(rs, 11)]), ms, &(x[WS(rs, 1)])); TD = BYTW(&(W[TWVL * 20]), TC); TE = VSUB(TB, TD); T1r = VADD(TB, TD); } { V Tm, To, Tl, Tn; Tl = LD(&(x[WS(rs, 12)]), ms, &(x[0])); Tm = BYTW(&(W[TWVL * 22]), Tl); Tn = LD(&(x[WS(rs, 2)]), ms, &(x[0])); To = BYTW(&(W[TWVL * 2]), Tn); Tp = VSUB(Tm, To); T1x = VADD(Tm, To); } TF = VSUB(Tz, TE); T14 = VSUB(T9, Te); T15 = VSUB(Tk, Tp); TQ = VSUB(TK, TP); Tf = VADD(T9, Te); Tq = VADD(Tk, Tp); Tr = VADD(Tf, Tq); T1O = VADD(T1u, T1v); T1P = VADD(T1x, T1y); T1Q = VADD(T1O, T1P); T1w = VSUB(T1u, T1v); T1z = VSUB(T1x, T1y); T1A = VADD(T1w, T1z); TY = VADD(Tz, TE); TZ = VADD(TK, TP); T10 = VADD(TY, TZ); T1L = VADD(T1n, T1o); T1M = VADD(T1q, T1r); T1N = VADD(T1L, T1M); T1p = VSUB(T1n, T1o); T1s = VSUB(T1q, T1r); T1t = VADD(T1p, T1s); } T1i = VADD(T4, Tr); T1j = VADD(TX, T10); ST(&(x[WS(rs, 15)]), VFNMSI(T1j, T1i), ms, &(x[WS(rs, 1)])); ST(&(x[WS(rs, 5)]), VFMAI(T1j, T1i), ms, &(x[WS(rs, 1)])); { V T1T, T1R, T1S, T1X, T1Z, T1V, T1W, T1Y, T1U; T1T = VSUB(T1N, T1Q); T1R = VADD(T1N, T1Q); T1S = VFNMS(LDK(KP250000000), T1R, T1K); T1V = VSUB(T1L, T1M); T1W = VSUB(T1O, T1P); T1X = VMUL(LDK(KP951056516), VFMA(LDK(KP618033988), T1W, T1V)); T1Z = VMUL(LDK(KP951056516), VFNMS(LDK(KP618033988), T1V, T1W)); ST(&(x[0]), VADD(T1K, T1R), ms, &(x[0])); T1Y = VFNMS(LDK(KP559016994), T1T, T1S); ST(&(x[WS(rs, 8)]), VFMAI(T1Z, T1Y), ms, &(x[0])); ST(&(x[WS(rs, 12)]), VFNMSI(T1Z, T1Y), ms, &(x[0])); T1U = VFMA(LDK(KP559016994), T1T, T1S); ST(&(x[WS(rs, 4)]), VFNMSI(T1X, T1U), ms, &(x[0])); ST(&(x[WS(rs, 16)]), VFMAI(T1X, T1U), ms, &(x[0])); } { V T1D, T1B, T1C, T1H, T1J, T1F, T1G, T1I, T1E; T1D = VSUB(T1t, T1A); T1B = VADD(T1t, T1A); T1C = VFNMS(LDK(KP250000000), T1B, T1m); T1F = VSUB(T1w, T1z); T1G = VSUB(T1p, T1s); T1H = VMUL(LDK(KP951056516), VFNMS(LDK(KP618033988), T1G, T1F)); T1J = VMUL(LDK(KP951056516), VFMA(LDK(KP618033988), T1F, T1G)); ST(&(x[WS(rs, 10)]), VADD(T1m, T1B), ms, &(x[0])); T1I = VFMA(LDK(KP559016994), T1D, T1C); ST(&(x[WS(rs, 6)]), VFMAI(T1J, T1I), ms, &(x[0])); ST(&(x[WS(rs, 14)]), VFNMSI(T1J, T1I), ms, &(x[0])); T1E = VFNMS(LDK(KP559016994), T1D, T1C); ST(&(x[WS(rs, 2)]), VFNMSI(T1H, T1E), ms, &(x[0])); ST(&(x[WS(rs, 18)]), VFMAI(T1H, T1E), ms, &(x[0])); } { V TR, T16, T1e, T1b, T13, T1d, Tu, T1a; TR = VFMA(LDK(KP618033988), TQ, TF); T16 = VFMA(LDK(KP618033988), T15, T14); T1e = VFNMS(LDK(KP618033988), T14, T15); T1b = VFNMS(LDK(KP618033988), TF, TQ); { V T11, T12, Ts, Tt; T11 = VFNMS(LDK(KP250000000), T10, TX); T12 = VSUB(TY, TZ); T13 = VFMA(LDK(KP559016994), T12, T11); T1d = VFNMS(LDK(KP559016994), T12, T11); Ts = VFNMS(LDK(KP250000000), Tr, T4); Tt = VSUB(Tf, Tq); Tu = VFMA(LDK(KP559016994), Tt, Ts); T1a = VFNMS(LDK(KP559016994), Tt, Ts); } { V TS, T17, T1g, T1h; TS = VFNMS(LDK(KP951056516), TR, Tu); T17 = VFMA(LDK(KP951056516), T16, T13); ST(&(x[WS(rs, 19)]), VFNMSI(T17, TS), ms, &(x[WS(rs, 1)])); ST(&(x[WS(rs, 1)]), VFMAI(T17, TS), ms, &(x[WS(rs, 1)])); T1g = VFNMS(LDK(KP951056516), T1b, T1a); T1h = VFMA(LDK(KP951056516), T1e, T1d); ST(&(x[WS(rs, 7)]), VFNMSI(T1h, T1g), ms, &(x[WS(rs, 1)])); ST(&(x[WS(rs, 13)]), VFMAI(T1h, T1g), ms, &(x[WS(rs, 1)])); } { V T18, T19, T1c, T1f; T18 = VFMA(LDK(KP951056516), TR, Tu); T19 = VFNMS(LDK(KP951056516), T16, T13); ST(&(x[WS(rs, 11)]), VFNMSI(T19, T18), ms, &(x[WS(rs, 1)])); ST(&(x[WS(rs, 9)]), VFMAI(T19, T18), ms, &(x[WS(rs, 1)])); T1c = VFMA(LDK(KP951056516), T1b, T1a); T1f = VFNMS(LDK(KP951056516), T1e, T1d); ST(&(x[WS(rs, 3)]), VFNMSI(T1f, T1c), ms, &(x[WS(rs, 1)])); ST(&(x[WS(rs, 17)]), VFMAI(T1f, T1c), ms, &(x[WS(rs, 1)])); } } } } VLEAVE(); }
static void n1bv_20(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs) { DVK(KP559016994, +0.559016994374947424102293417182819058860154590); DVK(KP618033988, +0.618033988749894848204586834365638117720309180); DVK(KP951056516, +0.951056516295153572116439333379382143405698634); DVK(KP250000000, +0.250000000000000000000000000000000000000000000); INT i; const R *xi; R *xo; xi = ii; xo = io; for (i = v; i > 0; i = i - VL, xi = xi + (VL * ivs), xo = xo + (VL * ovs), MAKE_VOLATILE_STRIDE(is), MAKE_VOLATILE_STRIDE(os)) { V TS, TA, TN, TV, TK, TU, TR, Tl; { V T3, TE, T1r, T13, Ta, TL, Tz, TG, Ts, TF, Th, TM, T1u, T1C, T1n; V T1a, T1m, T1h, T1x, T1D, Tk, Ti; { V T1, T2, TC, TD; T1 = LD(&(xi[0]), ivs, &(xi[0])); T2 = LD(&(xi[WS(is, 10)]), ivs, &(xi[0])); TC = LD(&(xi[WS(is, 5)]), ivs, &(xi[WS(is, 1)])); TD = LD(&(xi[WS(is, 15)]), ivs, &(xi[WS(is, 1)])); { V T14, T6, T1c, Tv, Tm, T1f, Ty, T17, T9, Tn, Tp, T1b, Td, Tq, Te; V Tf, T15, To; { V Tw, Tx, T7, T8, Tb, Tc; { V T4, T5, Tt, Tu, T11, T12; T4 = LD(&(xi[WS(is, 4)]), ivs, &(xi[0])); T5 = LD(&(xi[WS(is, 14)]), ivs, &(xi[0])); Tt = LD(&(xi[WS(is, 13)]), ivs, &(xi[WS(is, 1)])); Tu = LD(&(xi[WS(is, 3)]), ivs, &(xi[WS(is, 1)])); Tw = LD(&(xi[WS(is, 17)]), ivs, &(xi[WS(is, 1)])); T3 = VSUB(T1, T2); T11 = VADD(T1, T2); TE = VSUB(TC, TD); T12 = VADD(TC, TD); T14 = VADD(T4, T5); T6 = VSUB(T4, T5); T1c = VADD(Tt, Tu); Tv = VSUB(Tt, Tu); Tx = LD(&(xi[WS(is, 7)]), ivs, &(xi[WS(is, 1)])); T7 = LD(&(xi[WS(is, 16)]), ivs, &(xi[0])); T8 = LD(&(xi[WS(is, 6)]), ivs, &(xi[0])); T1r = VADD(T11, T12); T13 = VSUB(T11, T12); } Tb = LD(&(xi[WS(is, 8)]), ivs, &(xi[0])); Tc = LD(&(xi[WS(is, 18)]), ivs, &(xi[0])); Tm = LD(&(xi[WS(is, 9)]), ivs, &(xi[WS(is, 1)])); T1f = VADD(Tw, Tx); Ty = VSUB(Tw, Tx); T17 = VADD(T7, T8); T9 = VSUB(T7, T8); Tn = LD(&(xi[WS(is, 19)]), ivs, &(xi[WS(is, 1)])); Tp = LD(&(xi[WS(is, 1)]), ivs, &(xi[WS(is, 1)])); T1b = VADD(Tb, Tc); Td = VSUB(Tb, Tc); Tq = LD(&(xi[WS(is, 11)]), ivs, &(xi[WS(is, 1)])); Te = LD(&(xi[WS(is, 12)]), ivs, &(xi[0])); Tf = LD(&(xi[WS(is, 2)]), ivs, &(xi[0])); } Ta = VADD(T6, T9); TL = VSUB(T6, T9); T15 = VADD(Tm, Tn); To = VSUB(Tm, Tn); Tz = VSUB(Tv, Ty); TG = VADD(Tv, Ty); { V T1d, T1v, T18, Tr, T1e, Tg, T16, T1s; T1d = VSUB(T1b, T1c); T1v = VADD(T1b, T1c); T18 = VADD(Tp, Tq); Tr = VSUB(Tp, Tq); T1e = VADD(Te, Tf); Tg = VSUB(Te, Tf); T16 = VSUB(T14, T15); T1s = VADD(T14, T15); { V T1t, T19, T1w, T1g; T1t = VADD(T17, T18); T19 = VSUB(T17, T18); Ts = VSUB(To, Tr); TF = VADD(To, Tr); T1w = VADD(T1e, T1f); T1g = VSUB(T1e, T1f); Th = VADD(Td, Tg); TM = VSUB(Td, Tg); T1u = VADD(T1s, T1t); T1C = VSUB(T1s, T1t); T1n = VSUB(T16, T19); T1a = VADD(T16, T19); T1m = VSUB(T1d, T1g); T1h = VADD(T1d, T1g); T1x = VADD(T1v, T1w); T1D = VSUB(T1v, T1w); } } } } Tk = VSUB(Ta, Th); Ti = VADD(Ta, Th); { V TJ, T1k, T1A, TZ, Tj, T1E, T1G, TI, T10, T1j, T1z, T1i, T1y, TH; TJ = VSUB(TF, TG); TH = VADD(TF, TG); T1i = VADD(T1a, T1h); T1k = VSUB(T1a, T1h); T1y = VADD(T1u, T1x); T1A = VSUB(T1u, T1x); TZ = VADD(T3, Ti); Tj = VFNMS(LDK(KP250000000), Ti, T3); T1E = VMUL(LDK(KP951056516), VFMA(LDK(KP618033988), T1D, T1C)); T1G = VMUL(LDK(KP951056516), VFNMS(LDK(KP618033988), T1C, T1D)); TI = VFNMS(LDK(KP250000000), TH, TE); T10 = VADD(TE, TH); T1j = VFNMS(LDK(KP250000000), T1i, T13); ST(&(xo[0]), VADD(T1r, T1y), ovs, &(xo[0])); T1z = VFNMS(LDK(KP250000000), T1y, T1r); ST(&(xo[WS(os, 10)]), VADD(T13, T1i), ovs, &(xo[0])); { V T1p, T1l, T1o, T1q, T1F, T1B; TS = VFNMS(LDK(KP618033988), Ts, Tz); TA = VFMA(LDK(KP618033988), Tz, Ts); TN = VFMA(LDK(KP618033988), TM, TL); TV = VFNMS(LDK(KP618033988), TL, TM); ST(&(xo[WS(os, 5)]), VFMAI(T10, TZ), ovs, &(xo[WS(os, 1)])); ST(&(xo[WS(os, 15)]), VFNMSI(T10, TZ), ovs, &(xo[WS(os, 1)])); T1p = VFMA(LDK(KP559016994), T1k, T1j); T1l = VFNMS(LDK(KP559016994), T1k, T1j); T1o = VMUL(LDK(KP951056516), VFNMS(LDK(KP618033988), T1n, T1m)); T1q = VMUL(LDK(KP951056516), VFMA(LDK(KP618033988), T1m, T1n)); T1F = VFNMS(LDK(KP559016994), T1A, T1z); T1B = VFMA(LDK(KP559016994), T1A, T1z); ST(&(xo[WS(os, 14)]), VFNMSI(T1q, T1p), ovs, &(xo[0])); ST(&(xo[WS(os, 6)]), VFMAI(T1q, T1p), ovs, &(xo[0])); ST(&(xo[WS(os, 18)]), VFMAI(T1o, T1l), ovs, &(xo[0])); ST(&(xo[WS(os, 2)]), VFNMSI(T1o, T1l), ovs, &(xo[0])); ST(&(xo[WS(os, 16)]), VFMAI(T1E, T1B), ovs, &(xo[0])); ST(&(xo[WS(os, 4)]), VFNMSI(T1E, T1B), ovs, &(xo[0])); ST(&(xo[WS(os, 12)]), VFNMSI(T1G, T1F), ovs, &(xo[0])); ST(&(xo[WS(os, 8)]), VFMAI(T1G, T1F), ovs, &(xo[0])); TK = VFMA(LDK(KP559016994), TJ, TI); TU = VFNMS(LDK(KP559016994), TJ, TI); TR = VFNMS(LDK(KP559016994), Tk, Tj); Tl = VFMA(LDK(KP559016994), Tk, Tj); } } } { V TY, TW, TO, TQ, TB, TP, TX, TT; TY = VFMA(LDK(KP951056516), TV, TU); TW = VFNMS(LDK(KP951056516), TV, TU); TO = VFMA(LDK(KP951056516), TN, TK); TQ = VFNMS(LDK(KP951056516), TN, TK); TB = VFNMS(LDK(KP951056516), TA, Tl); TP = VFMA(LDK(KP951056516), TA, Tl); TX = VFNMS(LDK(KP951056516), TS, TR); TT = VFMA(LDK(KP951056516), TS, TR); ST(&(xo[WS(os, 9)]), VFMAI(TQ, TP), ovs, &(xo[WS(os, 1)])); ST(&(xo[WS(os, 11)]), VFNMSI(TQ, TP), ovs, &(xo[WS(os, 1)])); ST(&(xo[WS(os, 1)]), VFMAI(TO, TB), ovs, &(xo[WS(os, 1)])); ST(&(xo[WS(os, 19)]), VFNMSI(TO, TB), ovs, &(xo[WS(os, 1)])); ST(&(xo[WS(os, 17)]), VFMAI(TW, TT), ovs, &(xo[WS(os, 1)])); ST(&(xo[WS(os, 3)]), VFNMSI(TW, TT), ovs, &(xo[WS(os, 1)])); ST(&(xo[WS(os, 13)]), VFMAI(TY, TX), ovs, &(xo[WS(os, 1)])); ST(&(xo[WS(os, 7)]), VFNMSI(TY, TX), ovs, &(xo[WS(os, 1)])); } } }
static void n2bv_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 = 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 Tl, Ty, T7, Te, Tw, Tt, Tz, TA, TB, Tg, Th, Tm, Tj, Tk; Tj = LD(&(xi[0]), ivs, &(xi[0])); Tk = LD(&(xi[WS(is, 5)]), ivs, &(xi[WS(is, 1)])); Tl = VSUB(Tj, Tk); Ty = VADD(Tj, Tk); { V T3, Tr, Td, Tv, T6, Ts, Ta, Tu; { 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); Tr = 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); Tv = 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); Ts = 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); Tu = VADD(T8, T9); } T7 = VSUB(T3, T6); Te = VSUB(Ta, Td); Tw = VSUB(Tu, Tv); Tt = VSUB(Tr, Ts); Tz = VADD(Tr, Ts); TA = VADD(Tu, Tv); TB = VADD(Tz, TA); Tg = VADD(T3, T6); Th = VADD(Ta, Td); Tm = VADD(Tg, Th); } { V TH, TI, TK, TL, TM; TH = VADD(Tl, Tm); STM2(&(xo[10]), TH, ovs, &(xo[2])); TI = VADD(Ty, TB); STM2(&(xo[0]), TI, ovs, &(xo[0])); { V Tf, Tq, To, Tp, Ti, Tn, TJ; Tf = VBYI(VFMA(LDK(KP951056516), T7, VMUL(LDK(KP587785252), Te))); Tq = VBYI(VFNMS(LDK(KP951056516), Te, VMUL(LDK(KP587785252), T7))); Ti = VMUL(LDK(KP559016994), VSUB(Tg, Th)); Tn = VFNMS(LDK(KP250000000), Tm, Tl); To = VADD(Ti, Tn); Tp = VSUB(Tn, Ti); TJ = VADD(Tf, To); STM2(&(xo[2]), TJ, ovs, &(xo[2])); STN2(&(xo[0]), TI, TJ, ovs); TK = VADD(Tq, Tp); STM2(&(xo[14]), TK, ovs, &(xo[2])); TL = VSUB(To, Tf); STM2(&(xo[18]), TL, ovs, &(xo[2])); TM = VSUB(Tp, Tq); STM2(&(xo[6]), TM, ovs, &(xo[2])); } { V Tx, TG, TE, TF, TC, TD; Tx = VBYI(VFNMS(LDK(KP951056516), Tw, VMUL(LDK(KP587785252), Tt))); TG = VBYI(VFMA(LDK(KP951056516), Tt, VMUL(LDK(KP587785252), Tw))); TC = VFNMS(LDK(KP250000000), TB, Ty); TD = VMUL(LDK(KP559016994), VSUB(Tz, TA)); TE = VSUB(TC, TD); TF = VADD(TD, TC); { V TN, TO, TP, TQ; TN = VADD(Tx, TE); STM2(&(xo[4]), TN, ovs, &(xo[0])); STN2(&(xo[4]), TN, TM, ovs); TO = VADD(TG, TF); STM2(&(xo[12]), TO, ovs, &(xo[0])); STN2(&(xo[12]), TO, TK, ovs); TP = VSUB(TE, Tx); STM2(&(xo[16]), TP, ovs, &(xo[0])); STN2(&(xo[16]), TP, TL, ovs); TQ = VSUB(TF, TG); STM2(&(xo[8]), TQ, ovs, &(xo[0])); STN2(&(xo[8]), TQ, TH, ovs); } } } } }
static void n1bv_7(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs) { DVK(KP900968867, +0.900968867902419126236102319507445051165919162); DVK(KP692021471, +0.692021471630095869627814897002069140197260599); DVK(KP801937735, +0.801937735804838252472204639014890102331838324); DVK(KP974927912, +0.974927912181823607018131682993931217232785801); DVK(KP356895867, +0.356895867892209443894399510021300583399127187); DVK(KP554958132, +0.554958132087371191422194871006410481067288862); { INT i; const R *xi; R *xo; xi = ii; xo = io; for (i = v; i > 0; i = i - VL, xi = xi + (VL * ivs), xo = xo + (VL * ovs), MAKE_VOLATILE_STRIDE(is), MAKE_VOLATILE_STRIDE(os)) { V T1, T2, T3, T8, T9, T5, T6; T1 = LD(&(xi[0]), ivs, &(xi[0])); T2 = LD(&(xi[WS(is, 1)]), ivs, &(xi[WS(is, 1)])); T3 = LD(&(xi[WS(is, 6)]), ivs, &(xi[0])); T8 = LD(&(xi[WS(is, 3)]), ivs, &(xi[WS(is, 1)])); T9 = LD(&(xi[WS(is, 4)]), ivs, &(xi[0])); T5 = LD(&(xi[WS(is, 2)]), ivs, &(xi[0])); T6 = LD(&(xi[WS(is, 5)]), ivs, &(xi[WS(is, 1)])); { V Tg, T4, Te, Ta, Tf, T7; Tg = VSUB(T2, T3); T4 = VADD(T2, T3); Te = VSUB(T8, T9); Ta = VADD(T8, T9); Tf = VSUB(T5, T6); T7 = VADD(T5, T6); { V Tr, Tj, Tm, Th, To, Tb; Tr = VFMA(LDK(KP554958132), Te, Tg); Tj = VFNMS(LDK(KP356895867), T4, Ta); Tm = VFMA(LDK(KP554958132), Tf, Te); Th = VFNMS(LDK(KP554958132), Tg, Tf); ST(&(xo[0]), VADD(T1, VADD(T4, VADD(T7, Ta))), ovs, &(xo[0])); To = VFNMS(LDK(KP356895867), T7, T4); Tb = VFNMS(LDK(KP356895867), Ta, T7); { V Ts, Tk, Tn, Ti; Ts = VMUL(LDK(KP974927912), VFMA(LDK(KP801937735), Tr, Tf)); Tk = VFNMS(LDK(KP692021471), Tj, T7); Tn = VMUL(LDK(KP974927912), VFNMS(LDK(KP801937735), Tm, Tg)); Ti = VMUL(LDK(KP974927912), VFNMS(LDK(KP801937735), Th, Te)); { V Tp, Tc, Tl, Tq, Td; Tp = VFNMS(LDK(KP692021471), To, Ta); Tc = VFNMS(LDK(KP692021471), Tb, T4); Tl = VFNMS(LDK(KP900968867), Tk, T1); Tq = VFNMS(LDK(KP900968867), Tp, T1); Td = VFNMS(LDK(KP900968867), Tc, T1); ST(&(xo[WS(os, 5)]), VFNMSI(Tn, Tl), ovs, &(xo[WS(os, 1)])); ST(&(xo[WS(os, 2)]), VFMAI(Tn, Tl), ovs, &(xo[0])); ST(&(xo[WS(os, 6)]), VFNMSI(Ts, Tq), ovs, &(xo[0])); ST(&(xo[WS(os, 1)]), VFMAI(Ts, Tq), ovs, &(xo[WS(os, 1)])); ST(&(xo[WS(os, 4)]), VFNMSI(Ti, Td), ovs, &(xo[0])); ST(&(xo[WS(os, 3)]), VFMAI(Ti, Td), ovs, &(xo[WS(os, 1)])); } } } } } } VLEAVE(); }
static void n1bv_20(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs) { DVK(KP587785252, +0.587785252292473129168705954639072768597652438); DVK(KP951056516, +0.951056516295153572116439333379382143405698634); DVK(KP250000000, +0.250000000000000000000000000000000000000000000); DVK(KP559016994, +0.559016994374947424102293417182819058860154590); INT i; const R *xi; R *xo; xi = ii; xo = io; for (i = v; i > 0; i = i - VL, xi = xi + (VL * ivs), xo = xo + (VL * ovs), MAKE_VOLATILE_STRIDE(is), MAKE_VOLATILE_STRIDE(os)) { V T3, T1y, TH, T1i, Ts, TL, TM, Tz, T13, T16, T1j, T1u, T1v, T1w, T1r; V T1s, T1t, T1a, T1d, T1k, Ti, Tk, TE, TI, TZ, T10; { V T1, T2, T1g, TF, TG, T1h; T1 = LD(&(xi[0]), ivs, &(xi[0])); T2 = LD(&(xi[WS(is, 10)]), ivs, &(xi[0])); T1g = VADD(T1, T2); TF = LD(&(xi[WS(is, 5)]), ivs, &(xi[WS(is, 1)])); TG = LD(&(xi[WS(is, 15)]), ivs, &(xi[WS(is, 1)])); T1h = VADD(TF, TG); T3 = VSUB(T1, T2); T1y = VADD(T1g, T1h); TH = VSUB(TF, TG); T1i = VSUB(T1g, T1h); } { V T6, T11, Tv, T19, Ty, T1c, T9, T14, Td, T18, To, T12, Tr, T15, Tg; V T1b; { V T4, T5, Tt, Tu; T4 = LD(&(xi[WS(is, 4)]), ivs, &(xi[0])); T5 = LD(&(xi[WS(is, 14)]), ivs, &(xi[0])); T6 = VSUB(T4, T5); T11 = VADD(T4, T5); Tt = LD(&(xi[WS(is, 13)]), ivs, &(xi[WS(is, 1)])); Tu = LD(&(xi[WS(is, 3)]), ivs, &(xi[WS(is, 1)])); Tv = VSUB(Tt, Tu); T19 = VADD(Tt, Tu); } { V Tw, Tx, T7, T8; Tw = LD(&(xi[WS(is, 17)]), ivs, &(xi[WS(is, 1)])); Tx = LD(&(xi[WS(is, 7)]), ivs, &(xi[WS(is, 1)])); Ty = VSUB(Tw, Tx); T1c = VADD(Tw, Tx); T7 = LD(&(xi[WS(is, 16)]), ivs, &(xi[0])); T8 = LD(&(xi[WS(is, 6)]), ivs, &(xi[0])); T9 = VSUB(T7, T8); T14 = VADD(T7, T8); } { V Tb, Tc, Tm, Tn; Tb = LD(&(xi[WS(is, 8)]), ivs, &(xi[0])); Tc = LD(&(xi[WS(is, 18)]), ivs, &(xi[0])); Td = VSUB(Tb, Tc); T18 = VADD(Tb, Tc); Tm = LD(&(xi[WS(is, 9)]), ivs, &(xi[WS(is, 1)])); Tn = LD(&(xi[WS(is, 19)]), ivs, &(xi[WS(is, 1)])); To = VSUB(Tm, Tn); T12 = VADD(Tm, Tn); } { V Tp, Tq, Te, Tf; Tp = LD(&(xi[WS(is, 1)]), ivs, &(xi[WS(is, 1)])); Tq = LD(&(xi[WS(is, 11)]), ivs, &(xi[WS(is, 1)])); Tr = VSUB(Tp, Tq); T15 = VADD(Tp, Tq); Te = LD(&(xi[WS(is, 12)]), ivs, &(xi[0])); Tf = LD(&(xi[WS(is, 2)]), ivs, &(xi[0])); Tg = VSUB(Te, Tf); T1b = VADD(Te, Tf); } Ts = VSUB(To, Tr); TL = VSUB(T6, T9); TM = VSUB(Td, Tg); Tz = VSUB(Tv, Ty); T13 = VSUB(T11, T12); T16 = VSUB(T14, T15); T1j = VADD(T13, T16); T1u = VADD(T18, T19); T1v = VADD(T1b, T1c); T1w = VADD(T1u, T1v); T1r = VADD(T11, T12); T1s = VADD(T14, T15); T1t = VADD(T1r, T1s); T1a = VSUB(T18, T19); T1d = VSUB(T1b, T1c); T1k = VADD(T1a, T1d); { V Ta, Th, TC, TD; Ta = VADD(T6, T9); Th = VADD(Td, Tg); Ti = VADD(Ta, Th); Tk = VMUL(LDK(KP559016994), VSUB(Ta, Th)); TC = VADD(To, Tr); TD = VADD(Tv, Ty); TE = VMUL(LDK(KP559016994), VSUB(TC, TD)); TI = VADD(TC, TD); } } TZ = VADD(T3, Ti); T10 = VBYI(VADD(TH, TI)); ST(&(xo[WS(os, 15)]), VSUB(TZ, T10), ovs, &(xo[WS(os, 1)])); ST(&(xo[WS(os, 5)]), VADD(TZ, T10), ovs, &(xo[WS(os, 1)])); { V T1x, T1z, T1A, T1E, T1G, T1C, T1D, T1F, T1B; T1x = VMUL(LDK(KP559016994), VSUB(T1t, T1w)); T1z = VADD(T1t, T1w); T1A = VFNMS(LDK(KP250000000), T1z, T1y); T1C = VSUB(T1r, T1s); T1D = VSUB(T1u, T1v); T1E = VBYI(VFMA(LDK(KP951056516), T1C, VMUL(LDK(KP587785252), T1D))); T1G = VBYI(VFNMS(LDK(KP951056516), T1D, VMUL(LDK(KP587785252), T1C))); ST(&(xo[0]), VADD(T1y, T1z), ovs, &(xo[0])); T1F = VSUB(T1A, T1x); ST(&(xo[WS(os, 8)]), VSUB(T1F, T1G), ovs, &(xo[0])); ST(&(xo[WS(os, 12)]), VADD(T1G, T1F), ovs, &(xo[0])); T1B = VADD(T1x, T1A); ST(&(xo[WS(os, 4)]), VSUB(T1B, T1E), ovs, &(xo[0])); ST(&(xo[WS(os, 16)]), VADD(T1E, T1B), ovs, &(xo[0])); } { V T1n, T1l, T1m, T1f, T1p, T17, T1e, T1q, T1o; T1n = VMUL(LDK(KP559016994), VSUB(T1j, T1k)); T1l = VADD(T1j, T1k); T1m = VFNMS(LDK(KP250000000), T1l, T1i); T17 = VSUB(T13, T16); T1e = VSUB(T1a, T1d); T1f = VBYI(VFNMS(LDK(KP951056516), T1e, VMUL(LDK(KP587785252), T17))); T1p = VBYI(VFMA(LDK(KP951056516), T17, VMUL(LDK(KP587785252), T1e))); ST(&(xo[WS(os, 10)]), VADD(T1i, T1l), ovs, &(xo[0])); T1q = VADD(T1n, T1m); ST(&(xo[WS(os, 6)]), VADD(T1p, T1q), ovs, &(xo[0])); ST(&(xo[WS(os, 14)]), VSUB(T1q, T1p), ovs, &(xo[0])); T1o = VSUB(T1m, T1n); ST(&(xo[WS(os, 2)]), VADD(T1f, T1o), ovs, &(xo[0])); ST(&(xo[WS(os, 18)]), VSUB(T1o, T1f), ovs, &(xo[0])); } { V TA, TN, TU, TS, TK, TV, Tl, TR, TJ, Tj; TA = VFNMS(LDK(KP951056516), Tz, VMUL(LDK(KP587785252), Ts)); TN = VFNMS(LDK(KP951056516), TM, VMUL(LDK(KP587785252), TL)); TU = VFMA(LDK(KP951056516), TL, VMUL(LDK(KP587785252), TM)); TS = VFMA(LDK(KP951056516), Ts, VMUL(LDK(KP587785252), Tz)); TJ = VFNMS(LDK(KP250000000), TI, TH); TK = VSUB(TE, TJ); TV = VADD(TE, TJ); Tj = VFNMS(LDK(KP250000000), Ti, T3); Tl = VSUB(Tj, Tk); TR = VADD(Tk, Tj); { V TB, TO, TX, TY; TB = VSUB(Tl, TA); TO = VBYI(VSUB(TK, TN)); ST(&(xo[WS(os, 17)]), VSUB(TB, TO), ovs, &(xo[WS(os, 1)])); ST(&(xo[WS(os, 3)]), VADD(TB, TO), ovs, &(xo[WS(os, 1)])); TX = VADD(TR, TS); TY = VBYI(VSUB(TV, TU)); ST(&(xo[WS(os, 11)]), VSUB(TX, TY), ovs, &(xo[WS(os, 1)])); ST(&(xo[WS(os, 9)]), VADD(TX, TY), ovs, &(xo[WS(os, 1)])); } { V TP, TQ, TT, TW; TP = VADD(Tl, TA); TQ = VBYI(VADD(TN, TK)); ST(&(xo[WS(os, 13)]), VSUB(TP, TQ), ovs, &(xo[WS(os, 1)])); ST(&(xo[WS(os, 7)]), VADD(TP, TQ), ovs, &(xo[WS(os, 1)])); TT = VSUB(TR, TS); TW = VBYI(VADD(TU, TV)); ST(&(xo[WS(os, 19)]), VSUB(TT, TW), ovs, &(xo[WS(os, 1)])); ST(&(xo[WS(os, 1)]), VADD(TT, TW), ovs, &(xo[WS(os, 1)])); } } } }
static void hc2cfdftv_12(R *Rp, R *Ip, R *Rm, R *Im, const R *W, stride rs, INT mb, INT me, INT ms) { DVK(KP866025403, +0.866025403784438646763723170752936183471402627); DVK(KP500000000, +0.500000000000000000000000000000000000000000000); INT m; for (m = mb, W = W + ((mb - 1) * ((TWVL / VL) * 22)); m < me; m = m + VL, Rp = Rp + (VL * ms), Ip = Ip + (VL * ms), Rm = Rm - (VL * ms), Im = Im - (VL * ms), W = W + (TWVL * 22), MAKE_VOLATILE_STRIDE(rs)) { V T3, T7, TH, TE, Th, TC, Tq, T11, TU, Tx, Tb, Tz, Tu, Tw, Tp; V Tl, T9, Ta, T8, Ty, Tn, To, Tm, TG, T1, T2, Tt, T5, T6, T4; V Tv, Tj, Tk, Ti, TD, Tf, Tg, Te, TB, TT, TF, TR, Tr; T1 = LD(&(Rp[0]), ms, &(Rp[0])); T2 = LD(&(Rm[0]), -ms, &(Rm[0])); Tt = LDW(&(W[0])); T5 = LD(&(Rp[WS(rs, 2)]), ms, &(Rp[0])); T6 = LD(&(Rm[WS(rs, 2)]), -ms, &(Rm[0])); T4 = LDW(&(W[TWVL * 6])); Tv = LDW(&(W[TWVL * 8])); Tn = LD(&(Rp[WS(rs, 1)]), ms, &(Rp[WS(rs, 1)])); To = LD(&(Rm[WS(rs, 1)]), -ms, &(Rm[WS(rs, 1)])); T3 = VFMACONJ(T2, T1); Tu = VZMULIJ(Tt, VFNMSCONJ(T2, T1)); Tm = LDW(&(W[TWVL * 2])); TG = LDW(&(W[TWVL * 4])); T7 = VZMULJ(T4, VFMACONJ(T6, T5)); Tw = VZMULIJ(Tv, VFNMSCONJ(T6, T5)); Tj = LD(&(Rp[WS(rs, 5)]), ms, &(Rp[WS(rs, 1)])); Tk = LD(&(Rm[WS(rs, 5)]), -ms, &(Rm[WS(rs, 1)])); Ti = LDW(&(W[TWVL * 18])); TD = LDW(&(W[TWVL * 20])); Tp = VZMULJ(Tm, VFMACONJ(To, Tn)); TH = VZMULIJ(TG, VFNMSCONJ(To, Tn)); Tf = LD(&(Rp[WS(rs, 3)]), ms, &(Rp[WS(rs, 1)])); Tg = LD(&(Rm[WS(rs, 3)]), -ms, &(Rm[WS(rs, 1)])); Te = LDW(&(W[TWVL * 10])); TB = LDW(&(W[TWVL * 12])); Tl = VZMULJ(Ti, VFMACONJ(Tk, Tj)); TE = VZMULIJ(TD, VFNMSCONJ(Tk, Tj)); T9 = LD(&(Rp[WS(rs, 4)]), ms, &(Rp[0])); Ta = LD(&(Rm[WS(rs, 4)]), -ms, &(Rm[0])); T8 = LDW(&(W[TWVL * 14])); Ty = LDW(&(W[TWVL * 16])); Th = VZMULJ(Te, VFMACONJ(Tg, Tf)); TC = VZMULIJ(TB, VFNMSCONJ(Tg, Tf)); Tq = VADD(Tl, Tp); T11 = VSUB(Tp, Tl); TU = VSUB(Tu, Tw); Tx = VADD(Tu, Tw); Tb = VZMULJ(T8, VFMACONJ(Ta, T9)); Tz = VZMULIJ(Ty, VFNMSCONJ(Ta, T9)); TT = VSUB(TC, TE); TF = VADD(TC, TE); TR = VFNMS(LDK(KP500000000), Tq, Th); Tr = VADD(Th, Tq); { V TX, TA, T1d, TV, TY, TI, T1e, T12, TQ, Td, T10, Tc, T1a, TN, TJ; V T1j, T1f, T1b, TS, TM, Ts, T17, T13, TZ, T1i, T1c, T16, TW, TP, TO; V TL, TK, T1k, T1l, T1h, T1g, T18, T19, T15, T14; T10 = VSUB(Tb, T7); Tc = VADD(T7, Tb); TX = VFNMS(LDK(KP500000000), Tx, Tz); TA = VADD(Tx, Tz); T1d = VADD(TU, TT); TV = VSUB(TT, TU); TY = VFNMS(LDK(KP500000000), TF, TH); TI = VADD(TF, TH); T1e = VADD(T10, T11); T12 = VSUB(T10, T11); TQ = VFNMS(LDK(KP500000000), Tc, T3); Td = VADD(T3, Tc); T1a = VADD(TX, TY); TZ = VSUB(TX, TY); TN = VADD(TA, TI); TJ = VSUB(TA, TI); T1j = VMUL(LDK(KP866025403), VADD(T1d, T1e)); T1f = VMUL(LDK(KP866025403), VSUB(T1d, T1e)); T1b = VADD(TQ, TR); TS = VSUB(TQ, TR); TM = VADD(Td, Tr); Ts = VSUB(Td, Tr); T17 = VFMA(LDK(KP866025403), T12, TZ); T13 = VFNMS(LDK(KP866025403), T12, TZ); T1i = VSUB(T1b, T1a); T1c = VADD(T1a, T1b); T16 = VFNMS(LDK(KP866025403), TV, TS); TW = VFMA(LDK(KP866025403), TV, TS); TP = VCONJ(VMUL(LDK(KP500000000), VADD(TN, TM))); TO = VMUL(LDK(KP500000000), VSUB(TM, TN)); TL = VCONJ(VMUL(LDK(KP500000000), VFNMSI(TJ, Ts))); TK = VMUL(LDK(KP500000000), VFMAI(TJ, Ts)); T1k = VCONJ(VMUL(LDK(KP500000000), VFNMSI(T1j, T1i))); T1l = VMUL(LDK(KP500000000), VFMAI(T1j, T1i)); T1h = VMUL(LDK(KP500000000), VFMAI(T1f, T1c)); T1g = VCONJ(VMUL(LDK(KP500000000), VFNMSI(T1f, T1c))); T18 = VMUL(LDK(KP500000000), VFNMSI(T17, T16)); T19 = VCONJ(VMUL(LDK(KP500000000), VFMAI(T17, T16))); T15 = VCONJ(VMUL(LDK(KP500000000), VFMAI(T13, TW))); T14 = VMUL(LDK(KP500000000), VFNMSI(T13, TW)); ST(&(Rm[WS(rs, 5)]), TP, -ms, &(Rm[WS(rs, 1)])); ST(&(Rp[0]), TO, ms, &(Rp[0])); ST(&(Rm[WS(rs, 2)]), TL, -ms, &(Rm[0])); ST(&(Rp[WS(rs, 3)]), TK, ms, &(Rp[WS(rs, 1)])); ST(&(Rm[WS(rs, 3)]), T1k, -ms, &(Rm[WS(rs, 1)])); ST(&(Rp[WS(rs, 4)]), T1l, ms, &(Rp[0])); ST(&(Rp[WS(rs, 2)]), T1h, ms, &(Rp[0])); ST(&(Rm[WS(rs, 1)]), T1g, -ms, &(Rm[WS(rs, 1)])); ST(&(Rp[WS(rs, 5)]), T18, ms, &(Rp[WS(rs, 1)])); ST(&(Rm[WS(rs, 4)]), T19, -ms, &(Rm[0])); ST(&(Rm[0]), T15, -ms, &(Rm[0])); ST(&(Rp[WS(rs, 1)]), T14, ms, &(Rp[WS(rs, 1)])); } } }
static const R *t2fv_8(R *ri, R *ii, const R *W, stride ios, INT m, INT dist) { DVK(KP707106781, +0.707106781186547524400844362104849039284835938); INT i; R *x; x = ri; for (i = m; i > 0; i = i - VL, x = x + (VL * dist), W = W + (TWVL * 14), MAKE_VOLATILE_STRIDE(ios)) { V T1, T2, Th, Tj, T5, T7, Ta, Tc; T1 = LD(&(x[0]), dist, &(x[0])); T2 = LD(&(x[WS(ios, 4)]), dist, &(x[0])); Th = LD(&(x[WS(ios, 2)]), dist, &(x[0])); Tj = LD(&(x[WS(ios, 6)]), dist, &(x[0])); T5 = LD(&(x[WS(ios, 1)]), dist, &(x[WS(ios, 1)])); T7 = LD(&(x[WS(ios, 5)]), dist, &(x[WS(ios, 1)])); Ta = LD(&(x[WS(ios, 7)]), dist, &(x[WS(ios, 1)])); Tc = LD(&(x[WS(ios, 3)]), dist, &(x[WS(ios, 1)])); { V T3, Ti, Tk, T6, T8, Tb, Td; T3 = BYTWJ(&(W[TWVL * 6]), T2); Ti = BYTWJ(&(W[TWVL * 2]), Th); Tk = BYTWJ(&(W[TWVL * 10]), Tj); T6 = BYTWJ(&(W[0]), T5); T8 = BYTWJ(&(W[TWVL * 8]), T7); Tb = BYTWJ(&(W[TWVL * 12]), Ta); Td = BYTWJ(&(W[TWVL * 4]), Tc); { V Tq, T4, Tr, Tl, Tt, T9, Tu, Te, Tw, Ts; Tq = VADD(T1, T3); T4 = VSUB(T1, T3); Tr = VADD(Ti, Tk); Tl = VSUB(Ti, Tk); Tt = VADD(T6, T8); T9 = VSUB(T6, T8); Tu = VADD(Tb, Td); Te = VSUB(Tb, Td); Tw = VSUB(Tq, Tr); Ts = VADD(Tq, Tr); { V Tx, Tv, Tm, Tf; Tx = VSUB(Tu, Tt); Tv = VADD(Tt, Tu); Tm = VSUB(Te, T9); Tf = VADD(T9, Te); { V Tp, Tn, To, Tg; ST(&(x[WS(ios, 2)]), VFMAI(Tx, Tw), dist, &(x[0])); ST(&(x[WS(ios, 6)]), VFNMSI(Tx, Tw), dist, &(x[0])); ST(&(x[0]), VADD(Ts, Tv), dist, &(x[0])); ST(&(x[WS(ios, 4)]), VSUB(Ts, Tv), dist, &(x[0])); Tp = VFMA(LDK(KP707106781), Tm, Tl); Tn = VFNMS(LDK(KP707106781), Tm, Tl); To = VFNMS(LDK(KP707106781), Tf, T4); Tg = VFMA(LDK(KP707106781), Tf, T4); ST(&(x[WS(ios, 5)]), VFNMSI(Tp, To), dist, &(x[WS(ios, 1)])); ST(&(x[WS(ios, 3)]), VFMAI(Tp, To), dist, &(x[WS(ios, 1)])); ST(&(x[WS(ios, 7)]), VFMAI(Tn, Tg), dist, &(x[WS(ios, 1)])); ST(&(x[WS(ios, 1)]), VFNMSI(Tn, Tg), dist, &(x[WS(ios, 1)])); } } } } } return W; }
static void t1buv_9(R *ri, R *ii, const R *W, stride rs, INT mb, INT me, INT ms) { DVK(KP939692620, +0.939692620785908384054109277324731469936208134); DVK(KP907603734, +0.907603734547952313649323976213898122064543220); DVK(KP666666666, +0.666666666666666666666666666666666666666666667); DVK(KP852868531, +0.852868531952443209628250963940074071936020296); DVK(KP879385241, +0.879385241571816768108218554649462939872416269); DVK(KP984807753, +0.984807753012208059366743024589523013670643252); DVK(KP826351822, +0.826351822333069651148283373230685203999624323); DVK(KP347296355, +0.347296355333860697703433253538629592000751354); DVK(KP898197570, +0.898197570222573798468955502359086394667167570); DVK(KP673648177, +0.673648177666930348851716626769314796000375677); DVK(KP420276625, +0.420276625461206169731530603237061658838781920); DVK(KP866025403, +0.866025403784438646763723170752936183471402627); DVK(KP586256827, +0.586256827714544512072145703099641959914944179); DVK(KP968908795, +0.968908795874236621082202410917456709164223497); DVK(KP726681596, +0.726681596905677465811651808188092531873167623); DVK(KP439692620, +0.439692620785908384054109277324731469936208134); DVK(KP203604859, +0.203604859554852403062088995281827210665664861); DVK(KP152703644, +0.152703644666139302296566746461370407999248646); DVK(KP500000000, +0.500000000000000000000000000000000000000000000); { INT m; R *x; x = ii; 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(9, rs)) { V T1, T3, T5, T9, Tn, Tb, Td, Th, Tj, Tx, T6; T1 = LD(&(x[0]), ms, &(x[0])); { V T2, T4, T8, Tm; 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, 2)]), ms, &(x[0])); Tm = LD(&(x[WS(rs, 1)]), ms, &(x[WS(rs, 1)])); { V Ta, Tc, Tg, Ti; Ta = LD(&(x[WS(rs, 5)]), ms, &(x[WS(rs, 1)])); Tc = LD(&(x[WS(rs, 8)]), ms, &(x[0])); Tg = LD(&(x[WS(rs, 4)]), ms, &(x[0])); Ti = LD(&(x[WS(rs, 7)]), ms, &(x[WS(rs, 1)])); T3 = BYTW(&(W[TWVL * 4]), T2); T5 = BYTW(&(W[TWVL * 10]), T4); T9 = BYTW(&(W[TWVL * 2]), T8); Tn = BYTW(&(W[0]), Tm); Tb = BYTW(&(W[TWVL * 8]), Ta); Td = BYTW(&(W[TWVL * 14]), Tc); Th = BYTW(&(W[TWVL * 6]), Tg); Tj = BYTW(&(W[TWVL * 12]), Ti); } } Tx = VSUB(T3, T5); T6 = VADD(T3, T5); { V Tl, Te, Tk, To, T7, TN; Tl = VSUB(Td, Tb); Te = VADD(Tb, Td); Tk = VSUB(Th, Tj); To = VADD(Th, Tj); T7 = VFNMS(LDK(KP500000000), T6, T1); TN = VADD(T1, T6); { V Tf, TP, Tp, TO; Tf = VFNMS(LDK(KP500000000), Te, T9); TP = VADD(T9, Te); Tp = VFNMS(LDK(KP500000000), To, Tn); TO = VADD(Tn, To); { V Tz, TC, Tu, TD, TA, Tq, TQ, TS; Tz = VFNMS(LDK(KP152703644), Tl, Tf); TC = VFMA(LDK(KP203604859), Tf, Tl); Tu = VFNMS(LDK(KP439692620), Tk, Tf); TD = VFNMS(LDK(KP726681596), Tk, Tp); TA = VFMA(LDK(KP968908795), Tp, Tk); Tq = VFNMS(LDK(KP586256827), Tp, Tl); TQ = VADD(TO, TP); TS = VMUL(LDK(KP866025403), VSUB(TO, TP)); { V TI, TB, TH, TE, Tr, TR, Tw, Tv; Tv = VFNMS(LDK(KP420276625), Tu, Tl); TI = VFMA(LDK(KP673648177), TA, Tz); TB = VFNMS(LDK(KP673648177), TA, Tz); TH = VFNMS(LDK(KP898197570), TD, TC); TE = VFMA(LDK(KP898197570), TD, TC); Tr = VFNMS(LDK(KP347296355), Tq, Tk); ST(&(x[0]), VADD(TQ, TN), ms, &(x[0])); TR = VFNMS(LDK(KP500000000), TQ, TN); Tw = VFNMS(LDK(KP826351822), Tv, Tp); { V TM, TL, TF, TJ, Ts, Ty, TG, TK, Tt; TM = VMUL(LDK(KP984807753), VFMA(LDK(KP879385241), Tx, TI)); TL = VFMA(LDK(KP852868531), TE, T7); TF = VFNMS(LDK(KP500000000), TE, TB); TJ = VFMA(LDK(KP666666666), TI, TH); Ts = VFNMS(LDK(KP907603734), Tr, Tf); ST(&(x[WS(rs, 6)]), VFNMSI(TS, TR), ms, &(x[0])); ST(&(x[WS(rs, 3)]), VFMAI(TS, TR), ms, &(x[WS(rs, 1)])); Ty = VMUL(LDK(KP984807753), VFNMS(LDK(KP879385241), Tx, Tw)); ST(&(x[WS(rs, 8)]), VFNMSI(TM, TL), ms, &(x[0])); ST(&(x[WS(rs, 1)]), VFMAI(TM, TL), ms, &(x[WS(rs, 1)])); TG = VFMA(LDK(KP852868531), TF, T7); TK = VMUL(LDK(KP866025403), VFNMS(LDK(KP852868531), TJ, Tx)); Tt = VFNMS(LDK(KP939692620), Ts, T7); ST(&(x[WS(rs, 5)]), VFNMSI(TK, TG), ms, &(x[WS(rs, 1)])); ST(&(x[WS(rs, 4)]), VFMAI(TK, TG), ms, &(x[0])); ST(&(x[WS(rs, 2)]), VFMAI(Ty, Tt), ms, &(x[0])); ST(&(x[WS(rs, 7)]), VFNMSI(Ty, Tt), ms, &(x[WS(rs, 1)])); } } } } } } } VLEAVE(); }
static void n2fv_20(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs) { DVK(KP559016994, +0.559016994374947424102293417182819058860154590); DVK(KP618033988, +0.618033988749894848204586834365638117720309180); DVK(KP951056516, +0.951056516295153572116439333379382143405698634); DVK(KP250000000, +0.250000000000000000000000000000000000000000000); INT i; const R *xi; R *xo; xi = ri; xo = ro; for (i = v; i > 0; i = i - VL, xi = xi + (VL * ivs), xo = xo + (VL * ovs), MAKE_VOLATILE_STRIDE(is), MAKE_VOLATILE_STRIDE(os)) { V T1H, T1I, TU, TI, TP, TX, T1M, T1N, T1O, T1P, T1R, T1S, TM, TW, TT; V TF; { V T3, Tm, T1r, T13, Ta, TN, TH, TA, TG, Tt, Th, TO, T1u, T1C, T1n; V T1a, T1m, T1h, T1x, T1D, TE, Ti; { V T1, T2, Tk, Tl; T1 = LD(&(xi[0]), ivs, &(xi[0])); T2 = LD(&(xi[WS(is, 10)]), ivs, &(xi[0])); Tk = LD(&(xi[WS(is, 5)]), ivs, &(xi[WS(is, 1)])); Tl = LD(&(xi[WS(is, 15)]), ivs, &(xi[WS(is, 1)])); { V T14, T6, T1c, Tw, Tn, T1f, Tz, T17, T9, To, Tq, T1b, Td, Tr, Te; V Tf, T15, Tp; { V Tx, Ty, T7, T8, Tb, Tc; { V T4, T5, Tu, Tv, T11, T12; T4 = LD(&(xi[WS(is, 4)]), ivs, &(xi[0])); T5 = LD(&(xi[WS(is, 14)]), ivs, &(xi[0])); Tu = LD(&(xi[WS(is, 13)]), ivs, &(xi[WS(is, 1)])); Tv = LD(&(xi[WS(is, 3)]), ivs, &(xi[WS(is, 1)])); Tx = LD(&(xi[WS(is, 17)]), ivs, &(xi[WS(is, 1)])); T3 = VSUB(T1, T2); T11 = VADD(T1, T2); Tm = VSUB(Tk, Tl); T12 = VADD(Tk, Tl); T14 = VADD(T4, T5); T6 = VSUB(T4, T5); T1c = VADD(Tu, Tv); Tw = VSUB(Tu, Tv); Ty = LD(&(xi[WS(is, 7)]), ivs, &(xi[WS(is, 1)])); T7 = LD(&(xi[WS(is, 16)]), ivs, &(xi[0])); T8 = LD(&(xi[WS(is, 6)]), ivs, &(xi[0])); T1r = VADD(T11, T12); T13 = VSUB(T11, T12); } Tb = LD(&(xi[WS(is, 8)]), ivs, &(xi[0])); Tc = LD(&(xi[WS(is, 18)]), ivs, &(xi[0])); Tn = LD(&(xi[WS(is, 9)]), ivs, &(xi[WS(is, 1)])); T1f = VADD(Tx, Ty); Tz = VSUB(Tx, Ty); T17 = VADD(T7, T8); T9 = VSUB(T7, T8); To = LD(&(xi[WS(is, 19)]), ivs, &(xi[WS(is, 1)])); Tq = LD(&(xi[WS(is, 1)]), ivs, &(xi[WS(is, 1)])); T1b = VADD(Tb, Tc); Td = VSUB(Tb, Tc); Tr = LD(&(xi[WS(is, 11)]), ivs, &(xi[WS(is, 1)])); Te = LD(&(xi[WS(is, 12)]), ivs, &(xi[0])); Tf = LD(&(xi[WS(is, 2)]), ivs, &(xi[0])); } Ta = VADD(T6, T9); TN = VSUB(T6, T9); T15 = VADD(Tn, To); Tp = VSUB(Tn, To); TH = VSUB(Tz, Tw); TA = VADD(Tw, Tz); { V T1d, T1v, T18, Ts, T1e, Tg, T16, T1s; T1d = VSUB(T1b, T1c); T1v = VADD(T1b, T1c); T18 = VADD(Tq, Tr); Ts = VSUB(Tq, Tr); T1e = VADD(Te, Tf); Tg = VSUB(Te, Tf); T16 = VSUB(T14, T15); T1s = VADD(T14, T15); { V T1t, T19, T1w, T1g; T1t = VADD(T17, T18); T19 = VSUB(T17, T18); TG = VSUB(Ts, Tp); Tt = VADD(Tp, Ts); T1w = VADD(T1e, T1f); T1g = VSUB(T1e, T1f); Th = VADD(Td, Tg); TO = VSUB(Td, Tg); T1u = VADD(T1s, T1t); T1C = VSUB(T1s, T1t); T1n = VSUB(T16, T19); T1a = VADD(T16, T19); T1m = VSUB(T1d, T1g); T1h = VADD(T1d, T1g); T1x = VADD(T1v, T1w); T1D = VSUB(T1v, T1w); } } } } TE = VSUB(Ta, Th); Ti = VADD(Ta, Th); { V TL, T1k, T1A, Tj, TD, T1E, T1G, TK, TC, T1j, T1z, T1i, T1y, TB; TL = VSUB(TA, Tt); TB = VADD(Tt, TA); T1i = VADD(T1a, T1h); T1k = VSUB(T1a, T1h); T1y = VADD(T1u, T1x); T1A = VSUB(T1u, T1x); Tj = VADD(T3, Ti); TD = VFNMS(LDK(KP250000000), Ti, T3); T1E = VMUL(LDK(KP951056516), VFMA(LDK(KP618033988), T1D, T1C)); T1G = VMUL(LDK(KP951056516), VFNMS(LDK(KP618033988), T1C, T1D)); TK = VFNMS(LDK(KP250000000), TB, Tm); TC = VADD(Tm, TB); T1j = VFNMS(LDK(KP250000000), T1i, T13); T1H = VADD(T1r, T1y); STM2(&(xo[0]), T1H, ovs, &(xo[0])); T1z = VFNMS(LDK(KP250000000), T1y, T1r); T1I = VADD(T13, T1i); STM2(&(xo[20]), T1I, ovs, &(xo[0])); { V T1J, T1K, T1p, T1l, T1o, T1q, T1F, T1B, T1L, T1Q; TU = VFNMS(LDK(KP618033988), TG, TH); TI = VFMA(LDK(KP618033988), TH, TG); TP = VFMA(LDK(KP618033988), TO, TN); TX = VFNMS(LDK(KP618033988), TN, TO); T1J = VFMAI(TC, Tj); STM2(&(xo[30]), T1J, ovs, &(xo[2])); T1K = VFNMSI(TC, Tj); STM2(&(xo[10]), T1K, ovs, &(xo[2])); T1p = VFMA(LDK(KP559016994), T1k, T1j); T1l = VFNMS(LDK(KP559016994), T1k, T1j); T1o = VMUL(LDK(KP951056516), VFNMS(LDK(KP618033988), T1n, T1m)); T1q = VMUL(LDK(KP951056516), VFMA(LDK(KP618033988), T1m, T1n)); T1F = VFNMS(LDK(KP559016994), T1A, T1z); T1B = VFMA(LDK(KP559016994), T1A, T1z); T1L = VFMAI(T1q, T1p); STM2(&(xo[28]), T1L, ovs, &(xo[0])); STN2(&(xo[28]), T1L, T1J, ovs); T1M = VFNMSI(T1q, T1p); STM2(&(xo[12]), T1M, ovs, &(xo[0])); T1N = VFNMSI(T1o, T1l); STM2(&(xo[36]), T1N, ovs, &(xo[0])); T1O = VFMAI(T1o, T1l); STM2(&(xo[4]), T1O, ovs, &(xo[0])); T1P = VFNMSI(T1E, T1B); STM2(&(xo[32]), T1P, ovs, &(xo[0])); T1Q = VFMAI(T1E, T1B); STM2(&(xo[8]), T1Q, ovs, &(xo[0])); STN2(&(xo[8]), T1Q, T1K, ovs); T1R = VFMAI(T1G, T1F); STM2(&(xo[24]), T1R, ovs, &(xo[0])); T1S = VFNMSI(T1G, T1F); STM2(&(xo[16]), T1S, ovs, &(xo[0])); TM = VFNMS(LDK(KP559016994), TL, TK); TW = VFMA(LDK(KP559016994), TL, TK); TT = VFNMS(LDK(KP559016994), TE, TD); TF = VFMA(LDK(KP559016994), TE, TD); } } } { V T10, TY, TQ, TS, TJ, TR, TZ, TV; T10 = VFMA(LDK(KP951056516), TX, TW); TY = VFNMS(LDK(KP951056516), TX, TW); TQ = VFMA(LDK(KP951056516), TP, TM); TS = VFNMS(LDK(KP951056516), TP, TM); TJ = VFMA(LDK(KP951056516), TI, TF); TR = VFNMS(LDK(KP951056516), TI, TF); TZ = VFMA(LDK(KP951056516), TU, TT); TV = VFNMS(LDK(KP951056516), TU, TT); { V T1T, T1U, T1V, T1W; T1T = VFMAI(TS, TR); STM2(&(xo[22]), T1T, ovs, &(xo[2])); STN2(&(xo[20]), T1I, T1T, ovs); T1U = VFNMSI(TS, TR); STM2(&(xo[18]), T1U, ovs, &(xo[2])); STN2(&(xo[16]), T1S, T1U, ovs); T1V = VFMAI(TQ, TJ); STM2(&(xo[38]), T1V, ovs, &(xo[2])); STN2(&(xo[36]), T1N, T1V, ovs); T1W = VFNMSI(TQ, TJ); STM2(&(xo[2]), T1W, ovs, &(xo[2])); STN2(&(xo[0]), T1H, T1W, ovs); { V T1X, T1Y, T1Z, T20; T1X = VFMAI(TY, TV); STM2(&(xo[6]), T1X, ovs, &(xo[2])); STN2(&(xo[4]), T1O, T1X, ovs); T1Y = VFNMSI(TY, TV); STM2(&(xo[34]), T1Y, ovs, &(xo[2])); STN2(&(xo[32]), T1P, T1Y, ovs); T1Z = VFMAI(T10, TZ); STM2(&(xo[14]), T1Z, ovs, &(xo[2])); STN2(&(xo[12]), T1M, T1Z, ovs); T20 = VFNMSI(T10, TZ); STM2(&(xo[26]), T20, ovs, &(xo[2])); STN2(&(xo[24]), T1R, T20, ovs); } } } } }
static void m2bv_32_0(const R *xi, R *xo, stride is, int ivs, int ovs) { 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); { V T2f, T2k, T2N, T2M, T19, T1B, Tb, T1p, TT, T1v, TY, T1w, T2E, T2F, T2G; V T24, T2o, TC, T1s, TH, T1t, T2B, T2C, T2D, T1X, T2n, T2I, T2J, Tq, T1A; V T14, T1q, T2c, T2l; { V T3, T2i, T18, T2j, T6, T2d, T9, T2e, T15, Ta; { V T1, T2, T16, T17; T1 = LD(&(xi[0]), ivs, &(xi[0])); T2 = LD(&(xi[WS(is, 16)]), ivs, &(xi[0])); T3 = VSUB(T1, T2); T2i = VADD(T1, T2); T16 = LD(&(xi[WS(is, 8)]), ivs, &(xi[0])); T17 = LD(&(xi[WS(is, 24)]), ivs, &(xi[0])); T18 = VSUB(T16, T17); T2j = VADD(T16, T17); } { V T4, T5, T7, T8; T4 = LD(&(xi[WS(is, 4)]), ivs, &(xi[0])); T5 = LD(&(xi[WS(is, 20)]), ivs, &(xi[0])); T6 = VSUB(T4, T5); T2d = VADD(T4, T5); T7 = LD(&(xi[WS(is, 28)]), ivs, &(xi[0])); T8 = LD(&(xi[WS(is, 12)]), ivs, &(xi[0])); T9 = VSUB(T7, T8); T2e = VADD(T7, T8); } T2f = VSUB(T2d, T2e); T2k = VSUB(T2i, T2j); T2N = VADD(T2d, T2e); T2M = VADD(T2i, T2j); T15 = VMUL(LDK(KP707106781), VSUB(T6, T9)); T19 = VSUB(T15, T18); T1B = VADD(T18, T15); Ta = VMUL(LDK(KP707106781), VADD(T6, T9)); Tb = VSUB(T3, Ta); T1p = VADD(T3, Ta); } { V TL, T21, TW, T1Y, TO, T22, TS, T1Z; { V TJ, TK, TU, TV; TJ = LD(&(xi[WS(is, 3)]), ivs, &(xi[WS(is, 1)])); TK = LD(&(xi[WS(is, 19)]), ivs, &(xi[WS(is, 1)])); TL = VSUB(TJ, TK); T21 = VADD(TJ, TK); TU = LD(&(xi[WS(is, 31)]), ivs, &(xi[WS(is, 1)])); TV = LD(&(xi[WS(is, 15)]), ivs, &(xi[WS(is, 1)])); TW = VSUB(TU, TV); T1Y = VADD(TU, TV); } { V TM, TN, TQ, TR; TM = LD(&(xi[WS(is, 27)]), ivs, &(xi[WS(is, 1)])); TN = LD(&(xi[WS(is, 11)]), ivs, &(xi[WS(is, 1)])); TO = VSUB(TM, TN); T22 = VADD(TM, TN); TQ = LD(&(xi[WS(is, 7)]), ivs, &(xi[WS(is, 1)])); TR = LD(&(xi[WS(is, 23)]), ivs, &(xi[WS(is, 1)])); TS = VSUB(TQ, TR); T1Z = VADD(TQ, TR); } { V TP, TX, T20, T23; TP = VMUL(LDK(KP707106781), VSUB(TL, TO)); TT = VSUB(TP, TS); T1v = VADD(TS, TP); TX = VMUL(LDK(KP707106781), VADD(TL, TO)); TY = VSUB(TW, TX); T1w = VADD(TW, TX); T2E = VADD(T1Y, T1Z); T2F = VADD(T21, T22); T2G = VSUB(T2E, T2F); T20 = VSUB(T1Y, T1Z); T23 = VSUB(T21, T22); T24 = VFMA(LDK(KP923879532), T20, VMUL(LDK(KP382683432), T23)); T2o = VFNMS(LDK(KP382683432), T20, VMUL(LDK(KP923879532), T23)); } } { V Tu, T1U, TF, T1R, Tx, T1V, TB, T1S; { V Ts, Tt, TD, TE; Ts = LD(&(xi[WS(is, 5)]), ivs, &(xi[WS(is, 1)])); Tt = LD(&(xi[WS(is, 21)]), ivs, &(xi[WS(is, 1)])); Tu = VSUB(Ts, Tt); T1U = VADD(Ts, Tt); TD = LD(&(xi[WS(is, 1)]), ivs, &(xi[WS(is, 1)])); TE = LD(&(xi[WS(is, 17)]), ivs, &(xi[WS(is, 1)])); TF = VSUB(TD, TE); T1R = VADD(TD, TE); } { V Tv, Tw, Tz, TA; Tv = LD(&(xi[WS(is, 29)]), ivs, &(xi[WS(is, 1)])); Tw = LD(&(xi[WS(is, 13)]), ivs, &(xi[WS(is, 1)])); Tx = VSUB(Tv, Tw); T1V = VADD(Tv, Tw); Tz = LD(&(xi[WS(is, 9)]), ivs, &(xi[WS(is, 1)])); TA = LD(&(xi[WS(is, 25)]), ivs, &(xi[WS(is, 1)])); TB = VSUB(Tz, TA); T1S = VADD(Tz, TA); } { V Ty, TG, T1T, T1W; Ty = VMUL(LDK(KP707106781), VSUB(Tu, Tx)); TC = VSUB(Ty, TB); T1s = VADD(TB, Ty); TG = VMUL(LDK(KP707106781), VADD(Tu, Tx)); TH = VSUB(TF, TG); T1t = VADD(TF, TG); T2B = VADD(T1R, T1S); T2C = VADD(T1U, T1V); T2D = VSUB(T2B, T2C); T1T = VSUB(T1R, T1S); T1W = VSUB(T1U, T1V); T1X = VFNMS(LDK(KP382683432), T1W, VMUL(LDK(KP923879532), T1T)); T2n = VFMA(LDK(KP382683432), T1T, VMUL(LDK(KP923879532), T1W)); } } { V Te, T26, To, T29, Th, T27, Tl, T2a, Ti, Tp; { V Tc, Td, Tm, Tn; Tc = LD(&(xi[WS(is, 2)]), ivs, &(xi[0])); Td = LD(&(xi[WS(is, 18)]), ivs, &(xi[0])); Te = VSUB(Tc, Td); T26 = VADD(Tc, Td); Tm = LD(&(xi[WS(is, 30)]), ivs, &(xi[0])); Tn = LD(&(xi[WS(is, 14)]), ivs, &(xi[0])); To = VSUB(Tm, Tn); T29 = VADD(Tm, Tn); } { V Tf, Tg, Tj, Tk; Tf = LD(&(xi[WS(is, 10)]), ivs, &(xi[0])); Tg = LD(&(xi[WS(is, 26)]), ivs, &(xi[0])); Th = VSUB(Tf, Tg); T27 = VADD(Tf, Tg); Tj = LD(&(xi[WS(is, 6)]), ivs, &(xi[0])); Tk = LD(&(xi[WS(is, 22)]), ivs, &(xi[0])); Tl = VSUB(Tj, Tk); T2a = VADD(Tj, Tk); } T2I = VADD(T26, T27); T2J = VADD(T29, T2a); Ti = VFMA(LDK(KP382683432), Te, VMUL(LDK(KP923879532), Th)); Tp = VFNMS(LDK(KP382683432), To, VMUL(LDK(KP923879532), Tl)); Tq = VSUB(Ti, Tp); T1A = VADD(Ti, Tp); { V T12, T13, T28, T2b; T12 = VFNMS(LDK(KP382683432), Th, VMUL(LDK(KP923879532), Te)); T13 = VFMA(LDK(KP923879532), To, VMUL(LDK(KP382683432), Tl)); T14 = VSUB(T12, T13); T1q = VADD(T12, T13); T28 = VSUB(T26, T27); T2b = VSUB(T29, T2a); T2c = VMUL(LDK(KP707106781), VSUB(T28, T2b)); T2l = VMUL(LDK(KP707106781), VADD(T28, T2b)); } } { V T2L, T2R, T2Q, T2S; { V T2H, T2K, T2O, T2P; T2H = VMUL(LDK(KP707106781), VSUB(T2D, T2G)); T2K = VSUB(T2I, T2J); T2L = VBYI(VSUB(T2H, T2K)); T2R = VBYI(VADD(T2K, T2H)); T2O = VSUB(T2M, T2N); T2P = VMUL(LDK(KP707106781), VADD(T2D, T2G)); T2Q = VSUB(T2O, T2P); T2S = VADD(T2O, T2P); } ST(&(xo[24]), VADD(T2L, T2Q), ovs, &(xo[0])); ST(&(xo[56]), VSUB(T2S, T2R), ovs, &(xo[0])); ST(&(xo[40]), VSUB(T2Q, T2L), ovs, &(xo[0])); ST(&(xo[8]), VADD(T2R, T2S), ovs, &(xo[0])); } { V T2h, T2r, T2q, T2s; { V T25, T2g, T2m, T2p; T25 = VSUB(T1X, T24); T2g = VSUB(T2c, T2f); T2h = VBYI(VSUB(T25, T2g)); T2r = VBYI(VADD(T2g, T25)); T2m = VSUB(T2k, T2l); T2p = VSUB(T2n, T2o); T2q = VSUB(T2m, T2p); T2s = VADD(T2m, T2p); } ST(&(xo[20]), VADD(T2h, T2q), ovs, &(xo[0])); ST(&(xo[52]), VSUB(T2s, T2r), ovs, &(xo[0])); ST(&(xo[44]), VSUB(T2q, T2h), ovs, &(xo[0])); ST(&(xo[12]), VADD(T2r, T2s), ovs, &(xo[0])); } { V T2V, T2Z, T2Y, T30; { V T2T, T2U, T2W, T2X; T2T = VADD(T2M, T2N); T2U = VADD(T2I, T2J); T2V = VSUB(T2T, T2U); T2Z = VADD(T2T, T2U); T2W = VADD(T2B, T2C); T2X = VADD(T2E, T2F); T2Y = VBYI(VSUB(T2W, T2X)); T30 = VADD(T2W, T2X); } ST(&(xo[48]), VSUB(T2V, T2Y), ovs, &(xo[0])); ST(&(xo[0]), VADD(T2Z, T30), ovs, &(xo[0])); ST(&(xo[16]), VADD(T2V, T2Y), ovs, &(xo[0])); ST(&(xo[32]), VSUB(T2Z, T30), ovs, &(xo[0])); } { V T2v, T2z, T2y, T2A; { V T2t, T2u, T2w, T2x; T2t = VADD(T2k, T2l); T2u = VADD(T1X, T24); T2v = VADD(T2t, T2u); T2z = VSUB(T2t, T2u); T2w = VADD(T2f, T2c); T2x = VADD(T2n, T2o); T2y = VBYI(VADD(T2w, T2x)); T2A = VBYI(VSUB(T2x, T2w)); } ST(&(xo[60]), VSUB(T2v, T2y), ovs, &(xo[0])); ST(&(xo[28]), VADD(T2z, T2A), ovs, &(xo[0])); ST(&(xo[4]), VADD(T2v, T2y), ovs, &(xo[0])); ST(&(xo[36]), VSUB(T2z, T2A), ovs, &(xo[0])); } { V T1r, T1C, T1M, T1K, T1F, T1N, T1y, T1J; T1r = VSUB(T1p, T1q); T1C = VSUB(T1A, T1B); T1M = VADD(T1p, T1q); T1K = VADD(T1B, T1A); { V T1D, T1E, T1u, T1x; T1D = VFNMS(LDK(KP195090322), T1s, VMUL(LDK(KP980785280), T1t)); T1E = VFMA(LDK(KP195090322), T1v, VMUL(LDK(KP980785280), T1w)); T1F = VSUB(T1D, T1E); T1N = VADD(T1D, T1E); T1u = VFMA(LDK(KP980785280), T1s, VMUL(LDK(KP195090322), T1t)); T1x = VFNMS(LDK(KP195090322), T1w, VMUL(LDK(KP980785280), T1v)); T1y = VSUB(T1u, T1x); T1J = VADD(T1u, T1x); } { V T1z, T1G, T1P, T1Q; T1z = VADD(T1r, T1y); T1G = VBYI(VADD(T1C, T1F)); ST(&(xo[50]), VSUB(T1z, T1G), ovs, &(xo[2])); ST(&(xo[14]), VADD(T1z, T1G), ovs, &(xo[2])); T1P = VBYI(VADD(T1K, T1J)); T1Q = VADD(T1M, T1N); ST(&(xo[2]), VADD(T1P, T1Q), ovs, &(xo[2])); ST(&(xo[62]), VSUB(T1Q, T1P), ovs, &(xo[2])); } { V T1H, T1I, T1L, T1O; T1H = VSUB(T1r, T1y); T1I = VBYI(VSUB(T1F, T1C)); ST(&(xo[46]), VSUB(T1H, T1I), ovs, &(xo[2])); ST(&(xo[18]), VADD(T1H, T1I), ovs, &(xo[2])); T1L = VBYI(VSUB(T1J, T1K)); T1O = VSUB(T1M, T1N); ST(&(xo[30]), VADD(T1L, T1O), ovs, &(xo[2])); ST(&(xo[34]), VSUB(T1O, T1L), ovs, &(xo[2])); } } { V Tr, T1a, T1k, T1i, T1d, T1l, T10, T1h; Tr = VSUB(Tb, Tq); T1a = VSUB(T14, T19); T1k = VADD(Tb, Tq); T1i = VADD(T19, T14); { V T1b, T1c, TI, TZ; T1b = VFNMS(LDK(KP555570233), TC, VMUL(LDK(KP831469612), TH)); T1c = VFMA(LDK(KP555570233), TT, VMUL(LDK(KP831469612), TY)); T1d = VSUB(T1b, T1c); T1l = VADD(T1b, T1c); TI = VFMA(LDK(KP831469612), TC, VMUL(LDK(KP555570233), TH)); TZ = VFNMS(LDK(KP555570233), TY, VMUL(LDK(KP831469612), TT)); T10 = VSUB(TI, TZ); T1h = VADD(TI, TZ); } { V T11, T1e, T1n, T1o; T11 = VADD(Tr, T10); T1e = VBYI(VADD(T1a, T1d)); ST(&(xo[54]), VSUB(T11, T1e), ovs, &(xo[2])); ST(&(xo[10]), VADD(T11, T1e), ovs, &(xo[2])); T1n = VBYI(VADD(T1i, T1h)); T1o = VADD(T1k, T1l); ST(&(xo[6]), VADD(T1n, T1o), ovs, &(xo[2])); ST(&(xo[58]), VSUB(T1o, T1n), ovs, &(xo[2])); } { V T1f, T1g, T1j, T1m; T1f = VSUB(Tr, T10); T1g = VBYI(VSUB(T1d, T1a)); ST(&(xo[42]), VSUB(T1f, T1g), ovs, &(xo[2])); ST(&(xo[22]), VADD(T1f, T1g), ovs, &(xo[2])); T1j = VBYI(VSUB(T1h, T1i)); T1m = VSUB(T1k, T1l); ST(&(xo[26]), VADD(T1j, T1m), ovs, &(xo[2])); ST(&(xo[38]), VSUB(T1m, T1j), ovs, &(xo[2])); } } } }
static void t1bv_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 = ii; for (m = mb, W = W + (mb * ((TWVL / VL) * 12)); m < me; m = m + VL, x = x + (VL * ms), W = W + (TWVL * 12), MAKE_VOLATILE_STRIDE(rs)) { V T1, T2, T4, Te, Tc, T9, T7; T1 = LD(&(x[0]), ms, &(x[0])); T2 = LD(&(x[WS(rs, 1)]), ms, &(x[WS(rs, 1)])); T4 = LD(&(x[WS(rs, 6)]), ms, &(x[0])); Te = LD(&(x[WS(rs, 4)]), ms, &(x[0])); Tc = LD(&(x[WS(rs, 3)]), ms, &(x[WS(rs, 1)])); T9 = LD(&(x[WS(rs, 5)]), ms, &(x[WS(rs, 1)])); T7 = LD(&(x[WS(rs, 2)]), ms, &(x[0])); { V T3, T5, Tf, Td, Ta, T8; T3 = BYTW(&(W[0]), T2); T5 = BYTW(&(W[TWVL * 10]), T4); Tf = BYTW(&(W[TWVL * 6]), Te); Td = BYTW(&(W[TWVL * 4]), Tc); Ta = BYTW(&(W[TWVL * 8]), T9); T8 = BYTW(&(W[TWVL * 2]), T7); { V T6, Tm, Tg, Tk, Tb, Tl; T6 = VADD(T3, T5); Tm = VSUB(T3, T5); Tg = VADD(Td, Tf); Tk = VSUB(Td, Tf); Tb = VADD(T8, Ta); Tl = VSUB(T8, Ta); { V Tp, Tx, Tu, Th, Ts, Tn, Tq, Ty; Tp = VFNMS(LDK(KP356895867), T6, Tg); Tx = VFMA(LDK(KP554958132), Tk, Tm); ST(&(x[0]), VADD(T1, VADD(T6, VADD(Tb, Tg))), ms, &(x[0])); Tu = VFNMS(LDK(KP356895867), Tb, T6); Th = VFNMS(LDK(KP356895867), Tg, Tb); Ts = VFMA(LDK(KP554958132), Tl, Tk); Tn = VFNMS(LDK(KP554958132), Tm, Tl); Tq = VFNMS(LDK(KP692021471), Tp, Tb); Ty = VMUL(LDK(KP974927912), VFMA(LDK(KP801937735), Tx, Tl)); { V Tv, Ti, Tt, To, Tr, Tw, Tj; Tv = VFNMS(LDK(KP692021471), Tu, Tg); Ti = VFNMS(LDK(KP692021471), Th, T6); Tt = VMUL(LDK(KP974927912), VFNMS(LDK(KP801937735), Ts, Tm)); To = VMUL(LDK(KP974927912), VFNMS(LDK(KP801937735), Tn, Tk)); Tr = VFNMS(LDK(KP900968867), Tq, T1); Tw = VFNMS(LDK(KP900968867), Tv, T1); Tj = VFNMS(LDK(KP900968867), Ti, T1); ST(&(x[WS(rs, 5)]), VFNMSI(Tt, Tr), ms, &(x[WS(rs, 1)])); ST(&(x[WS(rs, 2)]), VFMAI(Tt, Tr), ms, &(x[0])); ST(&(x[WS(rs, 6)]), VFNMSI(Ty, Tw), ms, &(x[0])); ST(&(x[WS(rs, 1)]), VFMAI(Ty, Tw), ms, &(x[WS(rs, 1)])); ST(&(x[WS(rs, 4)]), VFNMSI(To, Tj), ms, &(x[0])); ST(&(x[WS(rs, 3)]), VFMAI(To, Tj), ms, &(x[WS(rs, 1)])); } } } } } }
static void t1sv_16(R *ri, R *ii, const R *W, stride rs, INT mb, INT me, INT ms) { DVK(KP923879532, +0.923879532511286756128183189396788286822416626); DVK(KP414213562, +0.414213562373095048801688724209698078569671875); DVK(KP707106781, +0.707106781186547524400844362104849039284835938); { INT m; for (m = mb, W = W + (mb * 30); m < me; m = m + (2 * VL), ri = ri + ((2 * VL) * ms), ii = ii + ((2 * VL) * ms), W = W + ((2 * VL) * 30), MAKE_VOLATILE_STRIDE(32, rs)) { V T2S, T2O, T2B, T2j, T2A, T24, T3J, T3L, T2Q, T2I, T2R, T2L, T2C, T2y, T3D; V T3F; { V T3o, T3z, T1I, T8, T35, T2o, T1s, T2r, T36, T2w, T1F, T2p, T1N, T3k, Tl; V T3A, T2V, T1T, Tz, T1U, T30, T29, T11, T2c, TH, TK, TJ, T31, T2h, T1e; V T2a, T1Z, TI, T1Y, TF; { V Ta, Td, Tg, Tj, T2t, T1y, Tf, T1J, Tb, Tc, T2v, T1E, Ti; { V T1, T3n, T3, T6, T5, T1h, T1k, T1n, T1q, T1m, T3l, T4, T1j, T1p, T2k; V T1i, T2, T1g; T1 = LD(&(ri[0]), ms, &(ri[0])); T3n = LD(&(ii[0]), ms, &(ii[0])); T3 = LD(&(ri[WS(rs, 8)]), ms, &(ri[0])); T6 = LD(&(ii[WS(rs, 8)]), ms, &(ii[0])); T2 = LDW(&(W[TWVL * 14])); T5 = LDW(&(W[TWVL * 15])); T1h = LD(&(ri[WS(rs, 15)]), ms, &(ri[WS(rs, 1)])); T1k = LD(&(ii[WS(rs, 15)]), ms, &(ii[WS(rs, 1)])); T1g = LDW(&(W[TWVL * 28])); T1n = LD(&(ri[WS(rs, 7)]), ms, &(ri[WS(rs, 1)])); T1q = LD(&(ii[WS(rs, 7)]), ms, &(ii[WS(rs, 1)])); T1m = LDW(&(W[TWVL * 12])); T3l = VMUL(T2, T6); T4 = VMUL(T2, T3); T1j = LDW(&(W[TWVL * 29])); T1p = LDW(&(W[TWVL * 13])); T2k = VMUL(T1g, T1k); T1i = VMUL(T1g, T1h); { V T1u, T1x, T1A, T2s, T1v, T1D, T1z, T1w, T1C, T2u, T1B, T9; { V T2l, T1l, T1t, T2n, T1r; { V T2m, T1o, T3m, T7; T1u = LD(&(ri[WS(rs, 3)]), ms, &(ri[WS(rs, 1)])); T2m = VMUL(T1m, T1q); T1o = VMUL(T1m, T1n); T3m = VFNMS(T5, T3, T3l); T7 = VFMA(T5, T6, T4); T1x = LD(&(ii[WS(rs, 3)]), ms, &(ii[WS(rs, 1)])); T2l = VFNMS(T1j, T1h, T2k); T1l = VFMA(T1j, T1k, T1i); T1t = LDW(&(W[TWVL * 4])); T2n = VFNMS(T1p, T1n, T2m); T1r = VFMA(T1p, T1q, T1o); T3o = VADD(T3m, T3n); T3z = VSUB(T3n, T3m); T1I = VSUB(T1, T7); T8 = VADD(T1, T7); } T1A = LD(&(ri[WS(rs, 11)]), ms, &(ri[WS(rs, 1)])); T2s = VMUL(T1t, T1x); T1v = VMUL(T1t, T1u); T35 = VADD(T2l, T2n); T2o = VSUB(T2l, T2n); T1s = VADD(T1l, T1r); T2r = VSUB(T1l, T1r); T1D = LD(&(ii[WS(rs, 11)]), ms, &(ii[WS(rs, 1)])); T1z = LDW(&(W[TWVL * 20])); } T1w = LDW(&(W[TWVL * 5])); T1C = LDW(&(W[TWVL * 21])); Ta = LD(&(ri[WS(rs, 4)]), ms, &(ri[0])); Td = LD(&(ii[WS(rs, 4)]), ms, &(ii[0])); T9 = LDW(&(W[TWVL * 6])); Tg = LD(&(ri[WS(rs, 12)]), ms, &(ri[0])); Tj = LD(&(ii[WS(rs, 12)]), ms, &(ii[0])); T2u = VMUL(T1z, T1D); T1B = VMUL(T1z, T1A); T2t = VFNMS(T1w, T1u, T2s); T1y = VFMA(T1w, T1x, T1v); Tf = LDW(&(W[TWVL * 22])); T1J = VMUL(T9, Td); Tb = VMUL(T9, Ta); Tc = LDW(&(W[TWVL * 7])); T2v = VFNMS(T1C, T1A, T2u); T1E = VFMA(T1C, T1D, T1B); Ti = LDW(&(W[TWVL * 23])); } } { V TW, TZ, TY, T27, TX, T26, TU; { V To, Tr, Tu, Tx, Tq, Tw, T1P, Tp, T1R, Tv; { V T1K, Te, T1M, Tk, Tn, Tt, T1L, Th; To = LD(&(ri[WS(rs, 2)]), ms, &(ri[0])); T1L = VMUL(Tf, Tj); Th = VMUL(Tf, Tg); Tr = LD(&(ii[WS(rs, 2)]), ms, &(ii[0])); T1K = VFNMS(Tc, Ta, T1J); Te = VFMA(Tc, Td, Tb); T36 = VADD(T2t, T2v); T2w = VSUB(T2t, T2v); T1F = VADD(T1y, T1E); T2p = VSUB(T1y, T1E); T1M = VFNMS(Ti, Tg, T1L); Tk = VFMA(Ti, Tj, Th); Tn = LDW(&(W[TWVL * 2])); Tu = LD(&(ri[WS(rs, 10)]), ms, &(ri[0])); Tx = LD(&(ii[WS(rs, 10)]), ms, &(ii[0])); Tt = LDW(&(W[TWVL * 18])); Tq = LDW(&(W[TWVL * 3])); Tw = LDW(&(W[TWVL * 19])); T1N = VSUB(T1K, T1M); T3k = VADD(T1K, T1M); Tl = VADD(Te, Tk); T3A = VSUB(Te, Tk); T1P = VMUL(Tn, Tr); Tp = VMUL(Tn, To); T1R = VMUL(Tt, Tx); Tv = VMUL(Tt, Tu); } { V TQ, TT, T1Q, Ts, T1S, Ty, TV, T25, TR, TP, TS; TQ = LD(&(ri[WS(rs, 1)]), ms, &(ri[WS(rs, 1)])); TT = LD(&(ii[WS(rs, 1)]), ms, &(ii[WS(rs, 1)])); TP = LDW(&(W[0])); TW = LD(&(ri[WS(rs, 9)]), ms, &(ri[WS(rs, 1)])); T1Q = VFNMS(Tq, To, T1P); Ts = VFMA(Tq, Tr, Tp); T1S = VFNMS(Tw, Tu, T1R); Ty = VFMA(Tw, Tx, Tv); TZ = LD(&(ii[WS(rs, 9)]), ms, &(ii[WS(rs, 1)])); TV = LDW(&(W[TWVL * 16])); T25 = VMUL(TP, TT); TR = VMUL(TP, TQ); TS = LDW(&(W[TWVL * 1])); TY = LDW(&(W[TWVL * 17])); T2V = VADD(T1Q, T1S); T1T = VSUB(T1Q, T1S); Tz = VADD(Ts, Ty); T1U = VSUB(Ts, Ty); T27 = VMUL(TV, TZ); TX = VMUL(TV, TW); T26 = VFNMS(TS, TQ, T25); TU = VFMA(TS, TT, TR); } } { V T19, T1c, T1b, T2f, T1a, T2e, T17; { V T13, T16, T12, T28, T10, T18, T15, T2d, T14; T13 = LD(&(ri[WS(rs, 5)]), ms, &(ri[WS(rs, 1)])); T16 = LD(&(ii[WS(rs, 5)]), ms, &(ii[WS(rs, 1)])); T12 = LDW(&(W[TWVL * 8])); T19 = LD(&(ri[WS(rs, 13)]), ms, &(ri[WS(rs, 1)])); T28 = VFNMS(TY, TW, T27); T10 = VFMA(TY, TZ, TX); T1c = LD(&(ii[WS(rs, 13)]), ms, &(ii[WS(rs, 1)])); T18 = LDW(&(W[TWVL * 24])); T15 = LDW(&(W[TWVL * 9])); T1b = LDW(&(W[TWVL * 25])); T2d = VMUL(T12, T16); T14 = VMUL(T12, T13); T30 = VADD(T26, T28); T29 = VSUB(T26, T28); T11 = VADD(TU, T10); T2c = VSUB(TU, T10); T2f = VMUL(T18, T1c); T1a = VMUL(T18, T19); T2e = VFNMS(T15, T13, T2d); T17 = VFMA(T15, T16, T14); } { V TB, TE, TA, T2g, T1d, TG, TD, T1X, TC; TB = LD(&(ri[WS(rs, 14)]), ms, &(ri[0])); TE = LD(&(ii[WS(rs, 14)]), ms, &(ii[0])); TA = LDW(&(W[TWVL * 26])); TH = LD(&(ri[WS(rs, 6)]), ms, &(ri[0])); T2g = VFNMS(T1b, T19, T2f); T1d = VFMA(T1b, T1c, T1a); TK = LD(&(ii[WS(rs, 6)]), ms, &(ii[0])); TG = LDW(&(W[TWVL * 10])); TD = LDW(&(W[TWVL * 27])); TJ = LDW(&(W[TWVL * 11])); T1X = VMUL(TA, TE); TC = VMUL(TA, TB); T31 = VADD(T2e, T2g); T2h = VSUB(T2e, T2g); T1e = VADD(T17, T1d); T2a = VSUB(T17, T1d); T1Z = VMUL(TG, TK); TI = VMUL(TG, TH); T1Y = VFNMS(TD, TB, T1X); TF = VFMA(TD, TE, TC); } } } } { V T2U, Tm, T3p, T3u, T34, T1G, T1f, T2Z, T20, TL, T32, T3f, T3g, T37; T2U = VSUB(T8, Tl); Tm = VADD(T8, Tl); T3p = VADD(T3k, T3o); T3u = VSUB(T3o, T3k); T34 = VSUB(T1s, T1F); T1G = VADD(T1s, T1F); T1f = VADD(T11, T1e); T2Z = VSUB(T11, T1e); T20 = VFNMS(TJ, TH, T1Z); TL = VFMA(TJ, TK, TI); T32 = VSUB(T30, T31); T3f = VADD(T30, T31); T3g = VADD(T35, T36); T37 = VSUB(T35, T36); { V T3r, T1H, T21, T1W, T3i, T3h, T3j, T2X, TN, T3t, T2W, TM; T3r = VSUB(T1G, T1f); T1H = VADD(T1f, T1G); T21 = VSUB(T1Y, T20); T2W = VADD(T1Y, T20); T1W = VSUB(TF, TL); TM = VADD(TF, TL); T3i = VADD(T3f, T3g); T3h = VSUB(T3f, T3g); T3j = VADD(T2V, T2W); T2X = VSUB(T2V, T2W); TN = VADD(Tz, TM); T3t = VSUB(TM, Tz); { V T2E, T1O, T3B, T3H, T2x, T2q, T2K, T2J, T3C, T23, T3I, T2H; { V T2F, T1V, T22, T2G; T2E = VADD(T1I, T1N); T1O = VSUB(T1I, T1N); { V T3b, T33, T3c, T38; T3b = VSUB(T32, T2Z); T33 = VADD(T2Z, T32); T3c = VADD(T34, T37); T38 = VSUB(T34, T37); { V T3a, T2Y, T3s, T3q; T3a = VSUB(T2U, T2X); T2Y = VADD(T2U, T2X); T3s = VSUB(T3p, T3j); T3q = VADD(T3j, T3p); { V T3x, T3v, T3e, TO; T3x = VSUB(T3u, T3t); T3v = VADD(T3t, T3u); T3e = VSUB(Tm, TN); TO = VADD(Tm, TN); { V T3d, T3w, T3y, T39; T3d = VSUB(T3b, T3c); T3w = VADD(T3b, T3c); T3y = VSUB(T38, T33); T39 = VADD(T33, T38); ST(&(ii[WS(rs, 4)]), VADD(T3r, T3s), ms, &(ii[0])); ST(&(ii[WS(rs, 12)]), VSUB(T3s, T3r), ms, &(ii[0])); ST(&(ii[0]), VADD(T3i, T3q), ms, &(ii[0])); ST(&(ii[WS(rs, 8)]), VSUB(T3q, T3i), ms, &(ii[0])); ST(&(ri[WS(rs, 4)]), VADD(T3e, T3h), ms, &(ri[0])); ST(&(ri[WS(rs, 12)]), VSUB(T3e, T3h), ms, &(ri[0])); ST(&(ri[0]), VADD(TO, T1H), ms, &(ri[0])); ST(&(ri[WS(rs, 8)]), VSUB(TO, T1H), ms, &(ri[0])); ST(&(ri[WS(rs, 6)]), VFMA(LDK(KP707106781), T3d, T3a), ms, &(ri[0])); ST(&(ri[WS(rs, 14)]), VFNMS(LDK(KP707106781), T3d, T3a), ms, &(ri[0])); ST(&(ii[WS(rs, 10)]), VFNMS(LDK(KP707106781), T3w, T3v), ms, &(ii[0])); ST(&(ii[WS(rs, 2)]), VFMA(LDK(KP707106781), T3w, T3v), ms, &(ii[0])); ST(&(ii[WS(rs, 14)]), VFNMS(LDK(KP707106781), T3y, T3x), ms, &(ii[0])); ST(&(ii[WS(rs, 6)]), VFMA(LDK(KP707106781), T3y, T3x), ms, &(ii[0])); ST(&(ri[WS(rs, 2)]), VFMA(LDK(KP707106781), T39, T2Y), ms, &(ri[0])); ST(&(ri[WS(rs, 10)]), VFNMS(LDK(KP707106781), T39, T2Y), ms, &(ri[0])); T3B = VSUB(T3z, T3A); T3H = VADD(T3A, T3z); } } } } T2F = VADD(T1U, T1T); T1V = VSUB(T1T, T1U); T22 = VADD(T1W, T21); T2G = VSUB(T1W, T21); { V T2M, T2N, T2b, T2i; T2x = VSUB(T2r, T2w); T2M = VADD(T2r, T2w); T2N = VSUB(T2o, T2p); T2q = VADD(T2o, T2p); T2K = VSUB(T29, T2a); T2b = VADD(T29, T2a); T2i = VSUB(T2c, T2h); T2J = VADD(T2c, T2h); T3C = VADD(T1V, T22); T23 = VSUB(T1V, T22); T2S = VFMA(LDK(KP414213562), T2M, T2N); T2O = VFNMS(LDK(KP414213562), T2N, T2M); T3I = VSUB(T2G, T2F); T2H = VADD(T2F, T2G); T2B = VFNMS(LDK(KP414213562), T2b, T2i); T2j = VFMA(LDK(KP414213562), T2i, T2b); } } T2A = VFNMS(LDK(KP707106781), T23, T1O); T24 = VFMA(LDK(KP707106781), T23, T1O); T3J = VFMA(LDK(KP707106781), T3I, T3H); T3L = VFNMS(LDK(KP707106781), T3I, T3H); T2Q = VFNMS(LDK(KP707106781), T2H, T2E); T2I = VFMA(LDK(KP707106781), T2H, T2E); T2R = VFNMS(LDK(KP414213562), T2J, T2K); T2L = VFMA(LDK(KP414213562), T2K, T2J); T2C = VFMA(LDK(KP414213562), T2q, T2x); T2y = VFNMS(LDK(KP414213562), T2x, T2q); T3D = VFMA(LDK(KP707106781), T3C, T3B); T3F = VFNMS(LDK(KP707106781), T3C, T3B); } } } } { V T3E, T2T, T2P, T3G; T3E = VADD(T2R, T2S); T2T = VSUB(T2R, T2S); T2P = VADD(T2L, T2O); T3G = VSUB(T2O, T2L); { V T3K, T2D, T2z, T3M; T3K = VSUB(T2C, T2B); T2D = VADD(T2B, T2C); T2z = VSUB(T2j, T2y); T3M = VADD(T2j, T2y); ST(&(ri[WS(rs, 5)]), VFMA(LDK(KP923879532), T2T, T2Q), ms, &(ri[WS(rs, 1)])); ST(&(ri[WS(rs, 13)]), VFNMS(LDK(KP923879532), T2T, T2Q), ms, &(ri[WS(rs, 1)])); ST(&(ii[WS(rs, 9)]), VFNMS(LDK(KP923879532), T3E, T3D), ms, &(ii[WS(rs, 1)])); ST(&(ii[WS(rs, 1)]), VFMA(LDK(KP923879532), T3E, T3D), ms, &(ii[WS(rs, 1)])); ST(&(ii[WS(rs, 13)]), VFNMS(LDK(KP923879532), T3G, T3F), ms, &(ii[WS(rs, 1)])); ST(&(ii[WS(rs, 5)]), VFMA(LDK(KP923879532), T3G, T3F), ms, &(ii[WS(rs, 1)])); ST(&(ri[WS(rs, 1)]), VFMA(LDK(KP923879532), T2P, T2I), ms, &(ri[WS(rs, 1)])); ST(&(ri[WS(rs, 9)]), VFNMS(LDK(KP923879532), T2P, T2I), ms, &(ri[WS(rs, 1)])); ST(&(ri[WS(rs, 15)]), VFMA(LDK(KP923879532), T2D, T2A), ms, &(ri[WS(rs, 1)])); ST(&(ri[WS(rs, 7)]), VFNMS(LDK(KP923879532), T2D, T2A), ms, &(ri[WS(rs, 1)])); ST(&(ii[WS(rs, 11)]), VFNMS(LDK(KP923879532), T3K, T3J), ms, &(ii[WS(rs, 1)])); ST(&(ii[WS(rs, 3)]), VFMA(LDK(KP923879532), T3K, T3J), ms, &(ii[WS(rs, 1)])); ST(&(ii[WS(rs, 15)]), VFMA(LDK(KP923879532), T3M, T3L), ms, &(ii[WS(rs, 1)])); ST(&(ii[WS(rs, 7)]), VFNMS(LDK(KP923879532), T3M, T3L), ms, &(ii[WS(rs, 1)])); ST(&(ri[WS(rs, 3)]), VFMA(LDK(KP923879532), T2z, T24), ms, &(ri[WS(rs, 1)])); ST(&(ri[WS(rs, 11)]), VFNMS(LDK(KP923879532), T2z, T24), ms, &(ri[WS(rs, 1)])); } } } } VLEAVE(); }
static void n2bv_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 = 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, 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; 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)); { V Tz, TH, Tj, TI; Tz = VFNMS(LDK(KP250000000), Ty, Tr); TH = VADD(Tr, Ty); STM2(&(xo[0]), TH, ovs, &(xo[0])); Tj = VFNMS(LDK(KP250000000), Ti, T3); TI = VADD(T3, Ti); STM2(&(xo[10]), TI, ovs, &(xo[2])); { 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); { V TJ, TK, TL, TM; TJ = VFNMSI(TG, TF); STM2(&(xo[8]), TJ, ovs, &(xo[0])); STN2(&(xo[8]), TJ, TI, ovs); TK = VFMAI(TG, TF); STM2(&(xo[12]), TK, ovs, &(xo[0])); TL = VFMAI(TE, TB); STM2(&(xo[16]), TL, ovs, &(xo[0])); TM = VFNMSI(TE, TB); STM2(&(xo[4]), TM, ovs, &(xo[0])); { V TN, TO, TP, TQ; TN = VFMAI(Tq, Tp); STM2(&(xo[6]), TN, ovs, &(xo[2])); STN2(&(xo[4]), TM, TN, ovs); TO = VFNMSI(Tq, Tp); STM2(&(xo[14]), TO, ovs, &(xo[2])); STN2(&(xo[12]), TK, TO, ovs); TP = VFNMSI(To, Tl); STM2(&(xo[18]), TP, ovs, &(xo[2])); STN2(&(xo[16]), TL, TP, ovs); TQ = VFMAI(To, Tl); STM2(&(xo[2]), TQ, ovs, &(xo[2])); STN2(&(xo[0]), TH, TQ, ovs); } } } } } } } } }
static void t1sv_16(R *ri, R *ii, const R *W, stride rs, INT mb, INT me, INT ms) { DVK(KP382683432, +0.382683432365089771728459984030398866761344562); DVK(KP923879532, +0.923879532511286756128183189396788286822416626); DVK(KP707106781, +0.707106781186547524400844362104849039284835938); { INT m; for (m = mb, W = W + (mb * 30); m < me; m = m + (2 * VL), ri = ri + ((2 * VL) * ms), ii = ii + ((2 * VL) * ms), W = W + ((2 * VL) * 30), MAKE_VOLATILE_STRIDE(32, rs)) { V T7, T37, T1t, T2U, Ti, T38, T1w, T2R, Tu, T2s, T1C, T2c, TF, T2t, T1H; V T2d, T1f, T1q, T2B, T2C, T2D, T2E, T1Z, T2j, T24, T2k, TS, T13, T2w, T2x; V T2y, T2z, T1O, T2g, T1T, T2h; { V T1, T2T, T6, T2S; T1 = LD(&(ri[0]), ms, &(ri[0])); T2T = LD(&(ii[0]), ms, &(ii[0])); { V T3, T5, T2, T4; T3 = LD(&(ri[WS(rs, 8)]), ms, &(ri[0])); T5 = LD(&(ii[WS(rs, 8)]), ms, &(ii[0])); T2 = LDW(&(W[TWVL * 14])); T4 = LDW(&(W[TWVL * 15])); T6 = VFMA(T2, T3, VMUL(T4, T5)); T2S = VFNMS(T4, T3, VMUL(T2, T5)); } T7 = VADD(T1, T6); T37 = VSUB(T2T, T2S); T1t = VSUB(T1, T6); T2U = VADD(T2S, T2T); } { V Tc, T1u, Th, T1v; { V T9, Tb, T8, Ta; T9 = LD(&(ri[WS(rs, 4)]), ms, &(ri[0])); Tb = LD(&(ii[WS(rs, 4)]), ms, &(ii[0])); T8 = LDW(&(W[TWVL * 6])); Ta = LDW(&(W[TWVL * 7])); Tc = VFMA(T8, T9, VMUL(Ta, Tb)); T1u = VFNMS(Ta, T9, VMUL(T8, Tb)); } { V Te, Tg, Td, Tf; Te = LD(&(ri[WS(rs, 12)]), ms, &(ri[0])); Tg = LD(&(ii[WS(rs, 12)]), ms, &(ii[0])); Td = LDW(&(W[TWVL * 22])); Tf = LDW(&(W[TWVL * 23])); Th = VFMA(Td, Te, VMUL(Tf, Tg)); T1v = VFNMS(Tf, Te, VMUL(Td, Tg)); } Ti = VADD(Tc, Th); T38 = VSUB(Tc, Th); T1w = VSUB(T1u, T1v); T2R = VADD(T1u, T1v); } { V To, T1y, Tt, T1z, T1A, T1B; { V Tl, Tn, Tk, Tm; Tl = LD(&(ri[WS(rs, 2)]), ms, &(ri[0])); Tn = LD(&(ii[WS(rs, 2)]), ms, &(ii[0])); Tk = LDW(&(W[TWVL * 2])); Tm = LDW(&(W[TWVL * 3])); To = VFMA(Tk, Tl, VMUL(Tm, Tn)); T1y = VFNMS(Tm, Tl, VMUL(Tk, Tn)); } { V Tq, Ts, Tp, Tr; Tq = LD(&(ri[WS(rs, 10)]), ms, &(ri[0])); Ts = LD(&(ii[WS(rs, 10)]), ms, &(ii[0])); Tp = LDW(&(W[TWVL * 18])); Tr = LDW(&(W[TWVL * 19])); Tt = VFMA(Tp, Tq, VMUL(Tr, Ts)); T1z = VFNMS(Tr, Tq, VMUL(Tp, Ts)); } Tu = VADD(To, Tt); T2s = VADD(T1y, T1z); T1A = VSUB(T1y, T1z); T1B = VSUB(To, Tt); T1C = VSUB(T1A, T1B); T2c = VADD(T1B, T1A); } { V Tz, T1E, TE, T1F, T1D, T1G; { V Tw, Ty, Tv, Tx; Tw = LD(&(ri[WS(rs, 14)]), ms, &(ri[0])); Ty = LD(&(ii[WS(rs, 14)]), ms, &(ii[0])); Tv = LDW(&(W[TWVL * 26])); Tx = LDW(&(W[TWVL * 27])); Tz = VFMA(Tv, Tw, VMUL(Tx, Ty)); T1E = VFNMS(Tx, Tw, VMUL(Tv, Ty)); } { V TB, TD, TA, TC; TB = LD(&(ri[WS(rs, 6)]), ms, &(ri[0])); TD = LD(&(ii[WS(rs, 6)]), ms, &(ii[0])); TA = LDW(&(W[TWVL * 10])); TC = LDW(&(W[TWVL * 11])); TE = VFMA(TA, TB, VMUL(TC, TD)); T1F = VFNMS(TC, TB, VMUL(TA, TD)); } TF = VADD(Tz, TE); T2t = VADD(T1E, T1F); T1D = VSUB(Tz, TE); T1G = VSUB(T1E, T1F); T1H = VADD(T1D, T1G); T2d = VSUB(T1D, T1G); } { V T19, T20, T1p, T1X, T1e, T21, T1k, T1W; { V T16, T18, T15, T17; T16 = LD(&(ri[WS(rs, 15)]), ms, &(ri[WS(rs, 1)])); T18 = LD(&(ii[WS(rs, 15)]), ms, &(ii[WS(rs, 1)])); T15 = LDW(&(W[TWVL * 28])); T17 = LDW(&(W[TWVL * 29])); T19 = VFMA(T15, T16, VMUL(T17, T18)); T20 = VFNMS(T17, T16, VMUL(T15, T18)); } { V T1m, T1o, T1l, T1n; T1m = LD(&(ri[WS(rs, 11)]), ms, &(ri[WS(rs, 1)])); T1o = LD(&(ii[WS(rs, 11)]), ms, &(ii[WS(rs, 1)])); T1l = LDW(&(W[TWVL * 20])); T1n = LDW(&(W[TWVL * 21])); T1p = VFMA(T1l, T1m, VMUL(T1n, T1o)); T1X = VFNMS(T1n, T1m, VMUL(T1l, T1o)); } { V T1b, T1d, T1a, T1c; T1b = LD(&(ri[WS(rs, 7)]), ms, &(ri[WS(rs, 1)])); T1d = LD(&(ii[WS(rs, 7)]), ms, &(ii[WS(rs, 1)])); T1a = LDW(&(W[TWVL * 12])); T1c = LDW(&(W[TWVL * 13])); T1e = VFMA(T1a, T1b, VMUL(T1c, T1d)); T21 = VFNMS(T1c, T1b, VMUL(T1a, T1d)); } { V T1h, T1j, T1g, T1i; T1h = LD(&(ri[WS(rs, 3)]), ms, &(ri[WS(rs, 1)])); T1j = LD(&(ii[WS(rs, 3)]), ms, &(ii[WS(rs, 1)])); T1g = LDW(&(W[TWVL * 4])); T1i = LDW(&(W[TWVL * 5])); T1k = VFMA(T1g, T1h, VMUL(T1i, T1j)); T1W = VFNMS(T1i, T1h, VMUL(T1g, T1j)); } T1f = VADD(T19, T1e); T1q = VADD(T1k, T1p); T2B = VSUB(T1f, T1q); T2C = VADD(T20, T21); T2D = VADD(T1W, T1X); T2E = VSUB(T2C, T2D); { V T1V, T1Y, T22, T23; T1V = VSUB(T19, T1e); T1Y = VSUB(T1W, T1X); T1Z = VSUB(T1V, T1Y); T2j = VADD(T1V, T1Y); T22 = VSUB(T20, T21); T23 = VSUB(T1k, T1p); T24 = VADD(T22, T23); T2k = VSUB(T22, T23); } } { V TM, T1K, T12, T1R, TR, T1L, TX, T1Q; { V TJ, TL, TI, TK; TJ = LD(&(ri[WS(rs, 1)]), ms, &(ri[WS(rs, 1)])); TL = LD(&(ii[WS(rs, 1)]), ms, &(ii[WS(rs, 1)])); TI = LDW(&(W[0])); TK = LDW(&(W[TWVL * 1])); TM = VFMA(TI, TJ, VMUL(TK, TL)); T1K = VFNMS(TK, TJ, VMUL(TI, TL)); } { V TZ, T11, TY, T10; TZ = LD(&(ri[WS(rs, 13)]), ms, &(ri[WS(rs, 1)])); T11 = LD(&(ii[WS(rs, 13)]), ms, &(ii[WS(rs, 1)])); TY = LDW(&(W[TWVL * 24])); T10 = LDW(&(W[TWVL * 25])); T12 = VFMA(TY, TZ, VMUL(T10, T11)); T1R = VFNMS(T10, TZ, VMUL(TY, T11)); } { V TO, TQ, TN, TP; TO = LD(&(ri[WS(rs, 9)]), ms, &(ri[WS(rs, 1)])); TQ = LD(&(ii[WS(rs, 9)]), ms, &(ii[WS(rs, 1)])); TN = LDW(&(W[TWVL * 16])); TP = LDW(&(W[TWVL * 17])); TR = VFMA(TN, TO, VMUL(TP, TQ)); T1L = VFNMS(TP, TO, VMUL(TN, TQ)); } { V TU, TW, TT, TV; TU = LD(&(ri[WS(rs, 5)]), ms, &(ri[WS(rs, 1)])); TW = LD(&(ii[WS(rs, 5)]), ms, &(ii[WS(rs, 1)])); TT = LDW(&(W[TWVL * 8])); TV = LDW(&(W[TWVL * 9])); TX = VFMA(TT, TU, VMUL(TV, TW)); T1Q = VFNMS(TV, TU, VMUL(TT, TW)); } TS = VADD(TM, TR); T13 = VADD(TX, T12); T2w = VSUB(TS, T13); T2x = VADD(T1K, T1L); T2y = VADD(T1Q, T1R); T2z = VSUB(T2x, T2y); { V T1M, T1N, T1P, T1S; T1M = VSUB(T1K, T1L); T1N = VSUB(TX, T12); T1O = VADD(T1M, T1N); T2g = VSUB(T1M, T1N); T1P = VSUB(TM, TR); T1S = VSUB(T1Q, T1R); T1T = VSUB(T1P, T1S); T2h = VADD(T1P, T1S); } } { V T1J, T27, T3g, T3i, T26, T3h, T2a, T3d; { V T1x, T1I, T3e, T3f; T1x = VSUB(T1t, T1w); T1I = VMUL(LDK(KP707106781), VSUB(T1C, T1H)); T1J = VADD(T1x, T1I); T27 = VSUB(T1x, T1I); T3e = VMUL(LDK(KP707106781), VSUB(T2d, T2c)); T3f = VADD(T38, T37); T3g = VADD(T3e, T3f); T3i = VSUB(T3f, T3e); } { V T1U, T25, T28, T29; T1U = VFMA(LDK(KP923879532), T1O, VMUL(LDK(KP382683432), T1T)); T25 = VFNMS(LDK(KP923879532), T24, VMUL(LDK(KP382683432), T1Z)); T26 = VADD(T1U, T25); T3h = VSUB(T25, T1U); T28 = VFNMS(LDK(KP923879532), T1T, VMUL(LDK(KP382683432), T1O)); T29 = VFMA(LDK(KP382683432), T24, VMUL(LDK(KP923879532), T1Z)); T2a = VSUB(T28, T29); T3d = VADD(T28, T29); } ST(&(ri[WS(rs, 11)]), VSUB(T1J, T26), ms, &(ri[WS(rs, 1)])); ST(&(ii[WS(rs, 11)]), VSUB(T3g, T3d), ms, &(ii[WS(rs, 1)])); ST(&(ri[WS(rs, 3)]), VADD(T1J, T26), ms, &(ri[WS(rs, 1)])); ST(&(ii[WS(rs, 3)]), VADD(T3d, T3g), ms, &(ii[WS(rs, 1)])); ST(&(ri[WS(rs, 15)]), VSUB(T27, T2a), ms, &(ri[WS(rs, 1)])); ST(&(ii[WS(rs, 15)]), VSUB(T3i, T3h), ms, &(ii[WS(rs, 1)])); ST(&(ri[WS(rs, 7)]), VADD(T27, T2a), ms, &(ri[WS(rs, 1)])); ST(&(ii[WS(rs, 7)]), VADD(T3h, T3i), ms, &(ii[WS(rs, 1)])); } { V T2v, T2H, T32, T34, T2G, T33, T2K, T2Z; { V T2r, T2u, T30, T31; T2r = VSUB(T7, Ti); T2u = VSUB(T2s, T2t); T2v = VADD(T2r, T2u); T2H = VSUB(T2r, T2u); T30 = VSUB(TF, Tu); T31 = VSUB(T2U, T2R); T32 = VADD(T30, T31); T34 = VSUB(T31, T30); } { V T2A, T2F, T2I, T2J; T2A = VADD(T2w, T2z); T2F = VSUB(T2B, T2E); T2G = VMUL(LDK(KP707106781), VADD(T2A, T2F)); T33 = VMUL(LDK(KP707106781), VSUB(T2F, T2A)); T2I = VSUB(T2z, T2w); T2J = VADD(T2B, T2E); T2K = VMUL(LDK(KP707106781), VSUB(T2I, T2J)); T2Z = VMUL(LDK(KP707106781), VADD(T2I, T2J)); } ST(&(ri[WS(rs, 10)]), VSUB(T2v, T2G), ms, &(ri[0])); ST(&(ii[WS(rs, 10)]), VSUB(T32, T2Z), ms, &(ii[0])); ST(&(ri[WS(rs, 2)]), VADD(T2v, T2G), ms, &(ri[0])); ST(&(ii[WS(rs, 2)]), VADD(T2Z, T32), ms, &(ii[0])); ST(&(ri[WS(rs, 14)]), VSUB(T2H, T2K), ms, &(ri[0])); ST(&(ii[WS(rs, 14)]), VSUB(T34, T33), ms, &(ii[0])); ST(&(ri[WS(rs, 6)]), VADD(T2H, T2K), ms, &(ri[0])); ST(&(ii[WS(rs, 6)]), VADD(T33, T34), ms, &(ii[0])); } { V T2f, T2n, T3a, T3c, T2m, T3b, T2q, T35; { V T2b, T2e, T36, T39; T2b = VADD(T1t, T1w); T2e = VMUL(LDK(KP707106781), VADD(T2c, T2d)); T2f = VADD(T2b, T2e); T2n = VSUB(T2b, T2e); T36 = VMUL(LDK(KP707106781), VADD(T1C, T1H)); T39 = VSUB(T37, T38); T3a = VADD(T36, T39); T3c = VSUB(T39, T36); } { V T2i, T2l, T2o, T2p; T2i = VFMA(LDK(KP382683432), T2g, VMUL(LDK(KP923879532), T2h)); T2l = VFNMS(LDK(KP382683432), T2k, VMUL(LDK(KP923879532), T2j)); T2m = VADD(T2i, T2l); T3b = VSUB(T2l, T2i); T2o = VFNMS(LDK(KP382683432), T2h, VMUL(LDK(KP923879532), T2g)); T2p = VFMA(LDK(KP923879532), T2k, VMUL(LDK(KP382683432), T2j)); T2q = VSUB(T2o, T2p); T35 = VADD(T2o, T2p); } ST(&(ri[WS(rs, 9)]), VSUB(T2f, T2m), ms, &(ri[WS(rs, 1)])); ST(&(ii[WS(rs, 9)]), VSUB(T3a, T35), ms, &(ii[WS(rs, 1)])); ST(&(ri[WS(rs, 1)]), VADD(T2f, T2m), ms, &(ri[WS(rs, 1)])); ST(&(ii[WS(rs, 1)]), VADD(T35, T3a), ms, &(ii[WS(rs, 1)])); ST(&(ri[WS(rs, 13)]), VSUB(T2n, T2q), ms, &(ri[WS(rs, 1)])); ST(&(ii[WS(rs, 13)]), VSUB(T3c, T3b), ms, &(ii[WS(rs, 1)])); ST(&(ri[WS(rs, 5)]), VADD(T2n, T2q), ms, &(ri[WS(rs, 1)])); ST(&(ii[WS(rs, 5)]), VADD(T3b, T3c), ms, &(ii[WS(rs, 1)])); } { V TH, T2L, T2W, T2Y, T1s, T2X, T2O, T2P; { V Tj, TG, T2Q, T2V; Tj = VADD(T7, Ti); TG = VADD(Tu, TF); TH = VADD(Tj, TG); T2L = VSUB(Tj, TG); T2Q = VADD(T2s, T2t); T2V = VADD(T2R, T2U); T2W = VADD(T2Q, T2V); T2Y = VSUB(T2V, T2Q); } { V T14, T1r, T2M, T2N; T14 = VADD(TS, T13); T1r = VADD(T1f, T1q); T1s = VADD(T14, T1r); T2X = VSUB(T1r, T14); T2M = VADD(T2x, T2y); T2N = VADD(T2C, T2D); T2O = VSUB(T2M, T2N); T2P = VADD(T2M, T2N); } ST(&(ri[WS(rs, 8)]), VSUB(TH, T1s), ms, &(ri[0])); ST(&(ii[WS(rs, 8)]), VSUB(T2W, T2P), ms, &(ii[0])); ST(&(ri[0]), VADD(TH, T1s), ms, &(ri[0])); ST(&(ii[0]), VADD(T2P, T2W), ms, &(ii[0])); ST(&(ri[WS(rs, 12)]), VSUB(T2L, T2O), ms, &(ri[0])); ST(&(ii[WS(rs, 12)]), VSUB(T2Y, T2X), ms, &(ii[0])); ST(&(ri[WS(rs, 4)]), VADD(T2L, T2O), ms, &(ri[0])); ST(&(ii[WS(rs, 4)]), VADD(T2X, T2Y), ms, &(ii[0])); } } } VLEAVE(); }
static void n2bv_16(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs) { DVK(KP923879532, +0.923879532511286756128183189396788286822416626); DVK(KP414213562, +0.414213562373095048801688724209698078569671875); DVK(KP707106781, +0.707106781186547524400844362104849039284835938); { 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(32, is), MAKE_VOLATILE_STRIDE(32, os)) { V T7, Tu, TF, TB, T13, TL, TO, TX, TC, Te, TP, Th, TQ, Tk, TW; V T16; { V TH, TU, Tz, Tf, TK, TV, TA, TM, Ta, TN, Td, Tg, Ti, Tj; { V T1, T2, T4, T5, To, Tp, Tr, Ts; T1 = LD(&(xi[0]), ivs, &(xi[0])); T2 = LD(&(xi[WS(is, 8)]), ivs, &(xi[0])); T4 = LD(&(xi[WS(is, 4)]), ivs, &(xi[0])); T5 = LD(&(xi[WS(is, 12)]), ivs, &(xi[0])); To = LD(&(xi[WS(is, 2)]), ivs, &(xi[0])); Tp = LD(&(xi[WS(is, 10)]), ivs, &(xi[0])); Tr = LD(&(xi[WS(is, 14)]), ivs, &(xi[0])); Ts = LD(&(xi[WS(is, 6)]), ivs, &(xi[0])); { V T8, TI, Tq, TJ, Tt, T9, Tb, Tc, T3, T6; T8 = LD(&(xi[WS(is, 1)]), ivs, &(xi[WS(is, 1)])); TH = VSUB(T1, T2); T3 = VADD(T1, T2); TU = VSUB(T4, T5); T6 = VADD(T4, T5); TI = VSUB(To, Tp); Tq = VADD(To, Tp); TJ = VSUB(Tr, Ts); Tt = VADD(Tr, Ts); T9 = LD(&(xi[WS(is, 9)]), ivs, &(xi[WS(is, 1)])); Tb = LD(&(xi[WS(is, 5)]), ivs, &(xi[WS(is, 1)])); Tc = LD(&(xi[WS(is, 13)]), ivs, &(xi[WS(is, 1)])); T7 = VSUB(T3, T6); Tz = VADD(T3, T6); Tf = LD(&(xi[WS(is, 15)]), ivs, &(xi[WS(is, 1)])); TK = VADD(TI, TJ); TV = VSUB(TI, TJ); TA = VADD(Tq, Tt); Tu = VSUB(Tq, Tt); TM = VSUB(T8, T9); Ta = VADD(T8, T9); TN = VSUB(Tb, Tc); Td = VADD(Tb, Tc); Tg = LD(&(xi[WS(is, 7)]), ivs, &(xi[WS(is, 1)])); Ti = LD(&(xi[WS(is, 3)]), ivs, &(xi[WS(is, 1)])); Tj = LD(&(xi[WS(is, 11)]), ivs, &(xi[WS(is, 1)])); } } TF = VADD(Tz, TA); TB = VSUB(Tz, TA); T13 = VFNMS(LDK(KP707106781), TK, TH); TL = VFMA(LDK(KP707106781), TK, TH); TO = VFNMS(LDK(KP414213562), TN, TM); TX = VFMA(LDK(KP414213562), TM, TN); TC = VADD(Ta, Td); Te = VSUB(Ta, Td); TP = VSUB(Tf, Tg); Th = VADD(Tf, Tg); TQ = VSUB(Tj, Ti); Tk = VADD(Ti, Tj); TW = VFMA(LDK(KP707106781), TV, TU); T16 = VFNMS(LDK(KP707106781), TV, TU); } { V TY, TR, Tl, TD; TY = VFMA(LDK(KP414213562), TP, TQ); TR = VFNMS(LDK(KP414213562), TQ, TP); Tl = VSUB(Th, Tk); TD = VADD(Th, Tk); { V TS, T17, TZ, T14; TS = VADD(TO, TR); T17 = VSUB(TO, TR); TZ = VSUB(TX, TY); T14 = VADD(TX, TY); { V TE, TG, Tm, Tv; TE = VSUB(TC, TD); TG = VADD(TC, TD); Tm = VADD(Te, Tl); Tv = VSUB(Te, Tl); { V T18, T1a, TT, T11; T18 = VFMA(LDK(KP923879532), T17, T16); T1a = VFNMS(LDK(KP923879532), T17, T16); TT = VFNMS(LDK(KP923879532), TS, TL); T11 = VFMA(LDK(KP923879532), TS, TL); { V T15, T19, T10, T12; T15 = VFNMS(LDK(KP923879532), T14, T13); T19 = VFMA(LDK(KP923879532), T14, T13); T10 = VFNMS(LDK(KP923879532), TZ, TW); T12 = VFMA(LDK(KP923879532), TZ, TW); { V T1b, T1c, T1d, T1e; T1b = VADD(TF, TG); STM2(&(xo[0]), T1b, ovs, &(xo[0])); T1c = VSUB(TF, TG); STM2(&(xo[16]), T1c, ovs, &(xo[0])); T1d = VFMAI(TE, TB); STM2(&(xo[8]), T1d, ovs, &(xo[0])); T1e = VFNMSI(TE, TB); STM2(&(xo[24]), T1e, ovs, &(xo[0])); { V Tw, Ty, Tn, Tx; Tw = VFNMS(LDK(KP707106781), Tv, Tu); Ty = VFMA(LDK(KP707106781), Tv, Tu); Tn = VFNMS(LDK(KP707106781), Tm, T7); Tx = VFMA(LDK(KP707106781), Tm, T7); { V T1f, T1g, T1h, T1i; T1f = VFNMSI(T1a, T19); STM2(&(xo[6]), T1f, ovs, &(xo[2])); T1g = VFMAI(T1a, T19); STM2(&(xo[26]), T1g, ovs, &(xo[2])); STN2(&(xo[24]), T1e, T1g, ovs); T1h = VFNMSI(T18, T15); STM2(&(xo[22]), T1h, ovs, &(xo[2])); T1i = VFMAI(T18, T15); STM2(&(xo[10]), T1i, ovs, &(xo[2])); STN2(&(xo[8]), T1d, T1i, ovs); { V T1j, T1k, T1l, T1m; T1j = VFNMSI(T12, T11); STM2(&(xo[30]), T1j, ovs, &(xo[2])); T1k = VFMAI(T12, T11); STM2(&(xo[2]), T1k, ovs, &(xo[2])); STN2(&(xo[0]), T1b, T1k, ovs); T1l = VFMAI(T10, TT); STM2(&(xo[18]), T1l, ovs, &(xo[2])); STN2(&(xo[16]), T1c, T1l, ovs); T1m = VFNMSI(T10, TT); STM2(&(xo[14]), T1m, ovs, &(xo[2])); { V T1n, T1o, T1p, T1q; T1n = VFMAI(Ty, Tx); STM2(&(xo[4]), T1n, ovs, &(xo[0])); STN2(&(xo[4]), T1n, T1f, ovs); T1o = VFNMSI(Ty, Tx); STM2(&(xo[28]), T1o, ovs, &(xo[0])); STN2(&(xo[28]), T1o, T1j, ovs); T1p = VFMAI(Tw, Tn); STM2(&(xo[20]), T1p, ovs, &(xo[0])); STN2(&(xo[20]), T1p, T1h, ovs); T1q = VFNMSI(Tw, Tn); STM2(&(xo[12]), T1q, ovs, &(xo[0])); STN2(&(xo[12]), T1q, T1m, ovs); } } } } } } } } } } } } VLEAVE(); }
static void t1fv_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 = ri; 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(rs)) { V Tr, TH, Tg, Tl, Tm, TA, TB, TJ, T5, Ta, Tb, TD, TE, TI, To; V Tq, Tp; To = LD(&(x[0]), ms, &(x[0])); Tp = LD(&(x[WS(rs, 5)]), ms, &(x[WS(rs, 1)])); Tq = BYTWJ(&(W[TWVL * 8]), Tp); Tr = VSUB(To, Tq); TH = VADD(To, Tq); { V Td, Tk, Tf, Ti; { V Tc, Tj, Te, Th; Tc = LD(&(x[WS(rs, 4)]), ms, &(x[0])); Td = BYTWJ(&(W[TWVL * 6]), Tc); Tj = LD(&(x[WS(rs, 1)]), ms, &(x[WS(rs, 1)])); Tk = BYTWJ(&(W[0]), Tj); Te = LD(&(x[WS(rs, 9)]), ms, &(x[WS(rs, 1)])); Tf = BYTWJ(&(W[TWVL * 16]), Te); Th = LD(&(x[WS(rs, 6)]), ms, &(x[0])); Ti = BYTWJ(&(W[TWVL * 10]), Th); } Tg = VSUB(Td, Tf); Tl = VSUB(Ti, Tk); Tm = VADD(Tg, Tl); TA = VADD(Td, Tf); TB = VADD(Ti, Tk); TJ = VADD(TA, TB); } { V T2, T9, T4, T7; { V T1, T8, T3, T6; T1 = LD(&(x[WS(rs, 2)]), ms, &(x[0])); T2 = BYTWJ(&(W[TWVL * 2]), T1); T8 = LD(&(x[WS(rs, 3)]), ms, &(x[WS(rs, 1)])); T9 = BYTWJ(&(W[TWVL * 4]), T8); T3 = LD(&(x[WS(rs, 7)]), ms, &(x[WS(rs, 1)])); T4 = BYTWJ(&(W[TWVL * 12]), T3); T6 = LD(&(x[WS(rs, 8)]), ms, &(x[0])); T7 = BYTWJ(&(W[TWVL * 14]), T6); } T5 = VSUB(T2, T4); Ta = VSUB(T7, T9); Tb = VADD(T5, Ta); TD = VADD(T2, T4); TE = VADD(T7, T9); TI = VADD(TD, TE); } { V Tn, Ts, Tt, Tx, Tz, Tv, Tw, Ty, Tu; Tn = VMUL(LDK(KP559016994), VSUB(Tb, Tm)); Ts = VADD(Tb, Tm); Tt = VFNMS(LDK(KP250000000), Ts, Tr); Tv = VSUB(T5, Ta); Tw = VSUB(Tg, Tl); Tx = VBYI(VFMA(LDK(KP951056516), Tv, VMUL(LDK(KP587785252), Tw))); Tz = VBYI(VFNMS(LDK(KP587785252), Tv, VMUL(LDK(KP951056516), Tw))); ST(&(x[WS(rs, 5)]), VADD(Tr, Ts), ms, &(x[WS(rs, 1)])); Ty = VSUB(Tt, Tn); 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)])); Tu = VADD(Tn, Tt); ST(&(x[WS(rs, 1)]), VSUB(Tu, Tx), ms, &(x[WS(rs, 1)])); ST(&(x[WS(rs, 9)]), VADD(Tx, Tu), ms, &(x[WS(rs, 1)])); } { V TM, TK, TL, TG, TO, TC, TF, TP, 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(KP587785252), TF, VMUL(LDK(KP951056516), TC))); TO = VBYI(VFMA(LDK(KP951056516), TF, VMUL(LDK(KP587785252), TC))); ST(&(x[0]), VADD(TH, TK), ms, &(x[0])); TP = VADD(TM, TL); ST(&(x[WS(rs, 4)]), VADD(TO, TP), ms, &(x[0])); ST(&(x[WS(rs, 6)]), VSUB(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])); } } }
static void hc2cfdftv_10(R *Rp, R *Ip, R *Rm, R *Im, const R *W, stride rs, INT mb, INT me, INT ms) { DVK(KP125000000, +0.125000000000000000000000000000000000000000000); DVK(KP279508497, +0.279508497187473712051146708591409529430077295); DVK(KP587785252, +0.587785252292473129168705954639072768597652438); DVK(KP951056516, +0.951056516295153572116439333379382143405698634); DVK(KP500000000, +0.500000000000000000000000000000000000000000000); { INT m; for (m = mb, W = W + ((mb - 1) * ((TWVL / VL) * 18)); m < me; m = m + VL, Rp = Rp + (VL * ms), Ip = Ip + (VL * ms), Rm = Rm - (VL * ms), Im = Im - (VL * ms), W = W + (TWVL * 18), MAKE_VOLATILE_STRIDE(40, rs)) { V Tl, Tt, Tu, TY, TZ, T10, Tz, TE, TF, TV, TW, TX, Ta, TU, TN; V TR, TH, TQ, TK, TL, TM, TI, TG, TJ, TT, TO, TP, TS, T18, T1c; V T12, T1b, T15, T16, T17, T14, T11, T13, T1e, T19, T1a, T1d; { V T1, T3, Ty, T8, T7, TB, Tf, Ts, Tk, Tw, Tq, TD, T2, Tx, T6; V TA, Tc, Te, Td, Tb, Tr, Tj, Ti, Th, Tg, Tv, Tn, Tp, To, Tm; V TC, T4, T9, T5; T1 = LD(&(Rp[0]), ms, &(Rp[0])); T2 = LD(&(Rm[0]), -ms, &(Rm[0])); T3 = VCONJ(T2); Tx = LDW(&(W[0])); Ty = VZMULIJ(Tx, VSUB(T3, T1)); T8 = LD(&(Rp[WS(rs, 2)]), ms, &(Rp[0])); T6 = LD(&(Rm[WS(rs, 2)]), -ms, &(Rm[0])); T7 = VCONJ(T6); TA = LDW(&(W[TWVL * 6])); TB = VZMULJ(TA, VADD(T7, T8)); Tc = LD(&(Rp[WS(rs, 1)]), ms, &(Rp[WS(rs, 1)])); Td = LD(&(Rm[WS(rs, 1)]), -ms, &(Rm[WS(rs, 1)])); Te = VCONJ(Td); Tb = LDW(&(W[TWVL * 2])); Tf = VZMULJ(Tb, VADD(Tc, Te)); Tr = LDW(&(W[TWVL * 4])); Ts = VZMULIJ(Tr, VSUB(Te, Tc)); Tj = LD(&(Rp[WS(rs, 3)]), ms, &(Rp[WS(rs, 1)])); Th = LD(&(Rm[WS(rs, 3)]), -ms, &(Rm[WS(rs, 1)])); Ti = VCONJ(Th); Tg = LDW(&(W[TWVL * 12])); Tk = VZMULIJ(Tg, VSUB(Ti, Tj)); Tv = LDW(&(W[TWVL * 10])); Tw = VZMULJ(Tv, VADD(Ti, Tj)); Tn = LD(&(Rp[WS(rs, 4)]), ms, &(Rp[0])); To = LD(&(Rm[WS(rs, 4)]), -ms, &(Rm[0])); Tp = VCONJ(To); Tm = LDW(&(W[TWVL * 14])); Tq = VZMULJ(Tm, VADD(Tn, Tp)); TC = LDW(&(W[TWVL * 16])); TD = VZMULIJ(TC, VSUB(Tp, Tn)); Tl = VSUB(Tf, Tk); Tt = VSUB(Tq, Ts); Tu = VADD(Tl, Tt); TY = VADD(Ty, Tw); TZ = VADD(TB, TD); T10 = VADD(TY, TZ); Tz = VSUB(Tw, Ty); TE = VSUB(TB, TD); TF = VADD(Tz, TE); TV = VADD(Tf, Tk); TW = VADD(Ts, Tq); TX = VADD(TV, TW); T4 = VADD(T1, T3); T5 = LDW(&(W[TWVL * 8])); T9 = VZMULIJ(T5, VSUB(T7, T8)); Ta = VSUB(T4, T9); TU = VADD(T4, T9); } TL = VSUB(Tl, Tt); TM = VSUB(TE, Tz); TN = VMUL(LDK(KP500000000), VBYI(VFMA(LDK(KP951056516), TL, VMUL(LDK(KP587785252), TM)))); TR = VMUL(LDK(KP500000000), VBYI(VFNMS(LDK(KP587785252), TL, VMUL(LDK(KP951056516), TM)))); TI = VMUL(LDK(KP279508497), VSUB(Tu, TF)); TG = VADD(Tu, TF); TJ = VFNMS(LDK(KP125000000), TG, VMUL(LDK(KP500000000), Ta)); TH = VCONJ(VMUL(LDK(KP500000000), VADD(Ta, TG))); TQ = VSUB(TJ, TI); TK = VADD(TI, TJ); ST(&(Rm[WS(rs, 4)]), TH, -ms, &(Rm[0])); TT = VCONJ(VADD(TQ, TR)); ST(&(Rm[WS(rs, 2)]), TT, -ms, &(Rm[0])); TO = VSUB(TK, TN); ST(&(Rp[WS(rs, 1)]), TO, ms, &(Rp[WS(rs, 1)])); TP = VCONJ(VADD(TK, TN)); ST(&(Rm[0]), TP, -ms, &(Rm[0])); TS = VSUB(TQ, TR); ST(&(Rp[WS(rs, 3)]), TS, ms, &(Rp[WS(rs, 1)])); T16 = VSUB(TZ, TY); T17 = VSUB(TV, TW); T18 = VMUL(LDK(KP500000000), VBYI(VFNMS(LDK(KP587785252), T17, VMUL(LDK(KP951056516), T16)))); T1c = VMUL(LDK(KP500000000), VBYI(VFMA(LDK(KP951056516), T17, VMUL(LDK(KP587785252), T16)))); T14 = VMUL(LDK(KP279508497), VSUB(TX, T10)); T11 = VADD(TX, T10); T13 = VFNMS(LDK(KP125000000), T11, VMUL(LDK(KP500000000), TU)); T12 = VMUL(LDK(KP500000000), VADD(TU, T11)); T1b = VADD(T14, T13); T15 = VSUB(T13, T14); ST(&(Rp[0]), T12, ms, &(Rp[0])); T1e = VADD(T1b, T1c); ST(&(Rp[WS(rs, 4)]), T1e, ms, &(Rp[0])); T19 = VCONJ(VSUB(T15, T18)); ST(&(Rm[WS(rs, 1)]), T19, -ms, &(Rm[WS(rs, 1)])); T1a = VADD(T15, T18); ST(&(Rp[WS(rs, 2)]), T1a, ms, &(Rp[0])); T1d = VCONJ(VSUB(T1b, T1c)); ST(&(Rm[WS(rs, 3)]), T1d, -ms, &(Rm[WS(rs, 1)])); } } VLEAVE(); }
static void t1fv_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 = ri; 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(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 = BYTWJ(&(W[TWVL * 8]), T2); Th = BYTWJ(&(W[TWVL * 6]), Tg); To = BYTWJ(&(W[0]), Tn); Tj = BYTWJ(&(W[TWVL * 16]), Ti); Tm = BYTWJ(&(W[TWVL * 10]), Tl); T6 = BYTWJ(&(W[TWVL * 2]), T5); Td = BYTWJ(&(W[TWVL * 4]), Tc); T8 = BYTWJ(&(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 = BYTWJ(&(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)]), VFNMSI(TN, TK), ms, &(x[0])); ST(&(x[WS(rs, 2)]), VFMAI(TN, TK), ms, &(x[0])); ST(&(x[WS(rs, 6)]), VFNMSI(TP, TO), ms, &(x[0])); ST(&(x[WS(rs, 4)]), VFMAI(TP, TO), ms, &(x[0])); ST(&(x[WS(rs, 9)]), VFMAI(Tx, Tu), ms, &(x[WS(rs, 1)])); ST(&(x[WS(rs, 1)]), VFNMSI(Tx, Tu), ms, &(x[WS(rs, 1)])); ST(&(x[WS(rs, 7)]), VFMAI(Tz, Ty), ms, &(x[WS(rs, 1)])); ST(&(x[WS(rs, 3)]), VFNMSI(Tz, Ty), ms, &(x[WS(rs, 1)])); } } } } } }
static void t3fv_20(R *ri, R *ii, const R *W, stride rs, INT mb, INT me, INT ms) { DVK(KP559016994, +0.559016994374947424102293417182819058860154590); DVK(KP618033988, +0.618033988749894848204586834365638117720309180); DVK(KP951056516, +0.951056516295153572116439333379382143405698634); 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 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, T1F, T23, T1i, T1p, T1q, T1j, Tp, TE, TF, T27, T28, T29, T1P; V T1S, T1T, TY, T1c, T1d, T24, T25, T26, T1I, T1L, T1M, TG, T1e; { V T1, TL, T6, TI, TK, T5, TH, T1D, T1E; 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); T1D = VADD(T1, T6); T1E = VADD(TI, TL); T1F = VSUB(T1D, T1E); T23 = VADD(T1D, T1E); } { V Th, T1G, T14, T1O, T1b, T1R, To, T1J, Tw, T1N, TR, T1H, TX, T1K, TD; V T1Q; { 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); T1G = 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); T1O = 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); T1R = 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); T1J = 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); T1N = 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); T1H = 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); T1K = 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); T1Q = 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); T27 = VADD(T1N, T1O); T28 = VADD(T1Q, T1R); T29 = VADD(T27, T28); T1P = VSUB(T1N, T1O); T1S = VSUB(T1Q, T1R); T1T = VADD(T1P, T1S); TY = VADD(TR, TX); T1c = VADD(T14, T1b); T1d = VADD(TY, T1c); T24 = VADD(T1G, T1H); T25 = VADD(T1J, T1K); T26 = VADD(T24, T25); T1I = VSUB(T1G, T1H); T1L = VSUB(T1J, T1K); T1M = VADD(T1I, T1L); } TG = VADD(T7, TF); T1e = VADD(TM, T1d); ST(&(x[WS(rs, 5)]), VFNMSI(T1e, TG), ms, &(x[WS(rs, 1)])); ST(&(x[WS(rs, 15)]), VFMAI(T1e, TG), ms, &(x[WS(rs, 1)])); { V T2c, T2a, T2b, T2g, T2i, T2e, T2f, T2h, T2d; T2c = VSUB(T26, T29); T2a = VADD(T26, T29); T2b = VFNMS(LDK(KP250000000), T2a, T23); T2e = VSUB(T24, T25); T2f = VSUB(T27, T28); T2g = VMUL(LDK(KP951056516), VFMA(LDK(KP618033988), T2f, T2e)); T2i = VMUL(LDK(KP951056516), VFNMS(LDK(KP618033988), T2e, T2f)); ST(&(x[0]), VADD(T23, T2a), ms, &(x[0])); T2h = VFNMS(LDK(KP559016994), T2c, T2b); ST(&(x[WS(rs, 8)]), VFNMSI(T2i, T2h), ms, &(x[0])); ST(&(x[WS(rs, 12)]), VFMAI(T2i, T2h), ms, &(x[0])); T2d = VFMA(LDK(KP559016994), T2c, T2b); ST(&(x[WS(rs, 4)]), VFMAI(T2g, T2d), ms, &(x[0])); ST(&(x[WS(rs, 16)]), VFNMSI(T2g, T2d), ms, &(x[0])); } { V T1W, T1U, T1V, T20, T22, T1Y, T1Z, T21, T1X; T1W = VSUB(T1M, T1T); T1U = VADD(T1M, T1T); T1V = VFNMS(LDK(KP250000000), T1U, T1F); T1Y = VSUB(T1P, T1S); T1Z = VSUB(T1I, T1L); T20 = VMUL(LDK(KP951056516), VFNMS(LDK(KP618033988), T1Z, T1Y)); T22 = VMUL(LDK(KP951056516), VFMA(LDK(KP618033988), T1Y, T1Z)); ST(&(x[WS(rs, 10)]), VADD(T1F, T1U), ms, &(x[0])); T21 = VFMA(LDK(KP559016994), T1W, T1V); ST(&(x[WS(rs, 6)]), VFNMSI(T22, T21), ms, &(x[0])); ST(&(x[WS(rs, 14)]), VFMAI(T22, T21), ms, &(x[0])); T1X = VFNMS(LDK(KP559016994), T1W, T1V); ST(&(x[WS(rs, 2)]), VFMAI(T20, T1X), ms, &(x[0])); ST(&(x[WS(rs, 18)]), VFNMSI(T20, T1X), ms, &(x[0])); } { V T1k, T1r, T1z, T1w, T1o, T1y, T1h, T1v; T1k = VFMA(LDK(KP618033988), T1j, T1i); T1r = VFMA(LDK(KP618033988), T1q, T1p); T1z = VFNMS(LDK(KP618033988), T1p, T1q); T1w = VFNMS(LDK(KP618033988), T1i, T1j); { V T1m, T1n, T1f, T1g; T1m = VFNMS(LDK(KP250000000), T1d, TM); T1n = VSUB(T1c, TY); T1o = VFNMS(LDK(KP559016994), T1n, T1m); T1y = VFMA(LDK(KP559016994), T1n, T1m); T1f = VFNMS(LDK(KP250000000), TF, T7); T1g = VSUB(Tp, TE); T1h = VFMA(LDK(KP559016994), T1g, T1f); T1v = VFNMS(LDK(KP559016994), T1g, T1f); } { V T1l, T1s, T1B, T1C; T1l = VFMA(LDK(KP951056516), T1k, T1h); T1s = VFMA(LDK(KP951056516), T1r, T1o); ST(&(x[WS(rs, 1)]), VFNMSI(T1s, T1l), ms, &(x[WS(rs, 1)])); ST(&(x[WS(rs, 19)]), VFMAI(T1s, T1l), ms, &(x[WS(rs, 1)])); T1B = VFMA(LDK(KP951056516), T1w, T1v); T1C = VFMA(LDK(KP951056516), T1z, T1y); ST(&(x[WS(rs, 13)]), VFNMSI(T1C, T1B), ms, &(x[WS(rs, 1)])); ST(&(x[WS(rs, 7)]), VFMAI(T1C, T1B), ms, &(x[WS(rs, 1)])); } { V T1t, T1u, T1x, T1A; T1t = VFNMS(LDK(KP951056516), T1k, T1h); T1u = VFNMS(LDK(KP951056516), T1r, T1o); ST(&(x[WS(rs, 9)]), VFNMSI(T1u, T1t), ms, &(x[WS(rs, 1)])); ST(&(x[WS(rs, 11)]), VFMAI(T1u, T1t), ms, &(x[WS(rs, 1)])); T1x = VFNMS(LDK(KP951056516), T1w, T1v); T1A = VFNMS(LDK(KP951056516), T1z, T1y); ST(&(x[WS(rs, 17)]), VFNMSI(T1A, T1x), ms, &(x[WS(rs, 1)])); ST(&(x[WS(rs, 3)]), VFMAI(T1A, T1x), ms, &(x[WS(rs, 1)])); } } } } } VLEAVE(); }
static void n2bv_8(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs) { DVK(KP707106781, +0.707106781186547524400844362104849039284835938); INT i; const R *xi; R *xo; xi = 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, Tc, Td, T4, T5, T7, T8; T1 = LD(&(xi[0]), ivs, &(xi[0])); T2 = LD(&(xi[WS(is, 4)]), ivs, &(xi[0])); Tc = LD(&(xi[WS(is, 2)]), ivs, &(xi[0])); Td = LD(&(xi[WS(is, 6)]), ivs, &(xi[0])); T4 = LD(&(xi[WS(is, 1)]), ivs, &(xi[WS(is, 1)])); T5 = LD(&(xi[WS(is, 5)]), ivs, &(xi[WS(is, 1)])); T7 = LD(&(xi[WS(is, 7)]), ivs, &(xi[WS(is, 1)])); T8 = LD(&(xi[WS(is, 3)]), ivs, &(xi[WS(is, 1)])); { V T3, Tj, Te, Tk, T6, Tm, T9, Tn, Tp, Tl; T3 = VSUB(T1, T2); Tj = VADD(T1, T2); Te = VSUB(Tc, Td); Tk = VADD(Tc, Td); T6 = VSUB(T4, T5); Tm = VADD(T4, T5); T9 = VSUB(T7, T8); Tn = VADD(T7, T8); Tp = VADD(Tj, Tk); Tl = VSUB(Tj, Tk); { V Tq, To, Ta, Tf; Tq = VADD(Tm, Tn); To = VSUB(Tm, Tn); Ta = VADD(T6, T9); Tf = VSUB(T6, T9); { V Tr, Ts, Tt, Tu, Tg, Ti, Tb, Th; Tr = VFMAI(To, Tl); STM2(&(xo[4]), Tr, ovs, &(xo[0])); Ts = VFNMSI(To, Tl); STM2(&(xo[12]), Ts, ovs, &(xo[0])); Tt = VADD(Tp, Tq); STM2(&(xo[0]), Tt, ovs, &(xo[0])); Tu = VSUB(Tp, Tq); STM2(&(xo[8]), Tu, ovs, &(xo[0])); Tg = VFNMS(LDK(KP707106781), Tf, Te); Ti = VFMA(LDK(KP707106781), Tf, Te); Tb = VFNMS(LDK(KP707106781), Ta, T3); Th = VFMA(LDK(KP707106781), Ta, T3); { V Tv, Tw, Tx, Ty; Tv = VFNMSI(Ti, Th); STM2(&(xo[14]), Tv, ovs, &(xo[2])); STN2(&(xo[12]), Ts, Tv, ovs); Tw = VFMAI(Ti, Th); STM2(&(xo[2]), Tw, ovs, &(xo[2])); STN2(&(xo[0]), Tt, Tw, ovs); Tx = VFMAI(Tg, Tb); STM2(&(xo[10]), Tx, ovs, &(xo[2])); STN2(&(xo[8]), Tu, Tx, ovs); Ty = VFNMSI(Tg, Tb); STM2(&(xo[6]), Ty, ovs, &(xo[2])); STN2(&(xo[4]), Tr, Ty, ovs); } } } } } }
static void t1bv_15(R *ri, R *ii, const R *W, stride rs, INT mb, INT me, INT ms) { DVK(KP216506350, +0.216506350946109661690930792688234045867850657); DVK(KP484122918, +0.484122918275927110647408174972799951354115213); DVK(KP866025403, +0.866025403784438646763723170752936183471402627); DVK(KP509036960, +0.509036960455127183450980863393907648510733164); DVK(KP823639103, +0.823639103546331925877420039278190003029660514); DVK(KP951056516, +0.951056516295153572116439333379382143405698634); DVK(KP587785252, +0.587785252292473129168705954639072768597652438); DVK(KP250000000, +0.250000000000000000000000000000000000000000000); DVK(KP559016994, +0.559016994374947424102293417182819058860154590); DVK(KP500000000, +0.500000000000000000000000000000000000000000000); { INT m; R *x; x = ii; for (m = mb, W = W + (mb * ((TWVL / VL) * 28)); m < me; m = m + VL, x = x + (VL * ms), W = W + (TWVL * 28), MAKE_VOLATILE_STRIDE(15, rs)) { V Ts, TV, T1f, TZ, T10, Tb, Tm, Tt, T1j, T1k, T1l, TI, TM, TR, Tz; V TD, TQ, T1g, T1h, T1i; { V TT, Tr, Tp, Tq, To, TU; TT = LD(&(x[0]), ms, &(x[0])); Tq = LD(&(x[WS(rs, 10)]), ms, &(x[0])); Tr = BYTW(&(W[TWVL * 18]), Tq); To = LD(&(x[WS(rs, 5)]), ms, &(x[WS(rs, 1)])); Tp = BYTW(&(W[TWVL * 8]), To); Ts = VSUB(Tp, Tr); TU = VADD(Tp, Tr); TV = VFNMS(LDK(KP500000000), TU, TT); T1f = VADD(TT, TU); } { V Tx, TG, TK, TB, T5, Ty, Tg, TH, Tl, TL, Ta, TC; { V Tw, TF, TJ, TA; Tw = LD(&(x[WS(rs, 3)]), ms, &(x[WS(rs, 1)])); Tx = BYTW(&(W[TWVL * 4]), Tw); TF = LD(&(x[WS(rs, 6)]), ms, &(x[0])); TG = BYTW(&(W[TWVL * 10]), TF); TJ = LD(&(x[WS(rs, 9)]), ms, &(x[WS(rs, 1)])); TK = BYTW(&(W[TWVL * 16]), TJ); TA = LD(&(x[WS(rs, 12)]), ms, &(x[0])); TB = BYTW(&(W[TWVL * 22]), TA); } { V T2, T4, T1, T3; T1 = LD(&(x[WS(rs, 8)]), ms, &(x[0])); T2 = BYTW(&(W[TWVL * 14]), T1); T3 = LD(&(x[WS(rs, 13)]), ms, &(x[WS(rs, 1)])); T4 = BYTW(&(W[TWVL * 24]), T3); T5 = VSUB(T2, T4); Ty = VADD(T2, T4); } { V Td, Tf, Tc, Te; Tc = LD(&(x[WS(rs, 11)]), ms, &(x[WS(rs, 1)])); Td = BYTW(&(W[TWVL * 20]), Tc); Te = LD(&(x[WS(rs, 1)]), ms, &(x[WS(rs, 1)])); Tf = BYTW(&(W[0]), Te); Tg = VSUB(Td, Tf); TH = VADD(Td, Tf); } { V Ti, Tk, Th, Tj; Th = LD(&(x[WS(rs, 14)]), ms, &(x[0])); Ti = BYTW(&(W[TWVL * 26]), Th); Tj = LD(&(x[WS(rs, 4)]), ms, &(x[0])); Tk = BYTW(&(W[TWVL * 6]), Tj); Tl = VSUB(Ti, Tk); TL = VADD(Ti, Tk); } { V T7, T9, T6, T8; T6 = LD(&(x[WS(rs, 2)]), ms, &(x[0])); T7 = BYTW(&(W[TWVL * 2]), T6); T8 = LD(&(x[WS(rs, 7)]), ms, &(x[WS(rs, 1)])); T9 = BYTW(&(W[TWVL * 12]), T8); Ta = VSUB(T7, T9); TC = VADD(T7, T9); } TZ = VSUB(T5, Ta); T10 = VSUB(Tg, Tl); Tb = VADD(T5, Ta); Tm = VADD(Tg, Tl); Tt = VADD(Tb, Tm); T1j = VADD(TG, TH); T1k = VADD(TK, TL); T1l = VADD(T1j, T1k); TI = VFNMS(LDK(KP500000000), TH, TG); TM = VFNMS(LDK(KP500000000), TL, TK); TR = VADD(TI, TM); Tz = VFNMS(LDK(KP500000000), Ty, Tx); TD = VFNMS(LDK(KP500000000), TC, TB); TQ = VADD(Tz, TD); T1g = VADD(Tx, Ty); T1h = VADD(TB, TC); T1i = VADD(T1g, T1h); } { V T1o, T1m, T1n, T1s, T1t, T1q, T1r, T1u, T1p; T1o = VMUL(LDK(KP559016994), VSUB(T1i, T1l)); T1m = VADD(T1i, T1l); T1n = VFNMS(LDK(KP250000000), T1m, T1f); T1q = VSUB(T1g, T1h); T1r = VSUB(T1j, T1k); T1s = VBYI(VFNMS(LDK(KP951056516), T1r, VMUL(LDK(KP587785252), T1q))); T1t = VBYI(VFMA(LDK(KP951056516), T1q, VMUL(LDK(KP587785252), T1r))); ST(&(x[0]), VADD(T1f, T1m), ms, &(x[0])); T1u = VADD(T1o, T1n); ST(&(x[WS(rs, 6)]), VADD(T1t, T1u), ms, &(x[0])); ST(&(x[WS(rs, 9)]), VSUB(T1u, T1t), ms, &(x[WS(rs, 1)])); T1p = VSUB(T1n, T1o); ST(&(x[WS(rs, 3)]), VSUB(T1p, T1s), ms, &(x[WS(rs, 1)])); ST(&(x[WS(rs, 12)]), VADD(T1s, T1p), ms, &(x[0])); } { V T11, T18, T1e, TO, T16, Tv, T15, TY, T1d, T19, TE, TN; T11 = VFMA(LDK(KP823639103), TZ, VMUL(LDK(KP509036960), T10)); T18 = VFNMS(LDK(KP823639103), T10, VMUL(LDK(KP509036960), TZ)); T1e = VBYI(VMUL(LDK(KP866025403), VADD(Ts, Tt))); TE = VSUB(Tz, TD); TN = VSUB(TI, TM); TO = VFMA(LDK(KP951056516), TE, VMUL(LDK(KP587785252), TN)); T16 = VFNMS(LDK(KP951056516), TN, VMUL(LDK(KP587785252), TE)); { V Tn, Tu, TS, TW, TX; Tn = VMUL(LDK(KP484122918), VSUB(Tb, Tm)); Tu = VFNMS(LDK(KP216506350), Tt, VMUL(LDK(KP866025403), Ts)); Tv = VADD(Tn, Tu); T15 = VSUB(Tn, Tu); TS = VMUL(LDK(KP559016994), VSUB(TQ, TR)); TW = VADD(TQ, TR); TX = VFNMS(LDK(KP250000000), TW, TV); TY = VADD(TS, TX); T1d = VADD(TV, TW); T19 = VSUB(TX, TS); } { V TP, T12, T1b, T1c; ST(&(x[WS(rs, 5)]), VSUB(T1d, T1e), ms, &(x[WS(rs, 1)])); ST(&(x[WS(rs, 10)]), VADD(T1e, T1d), ms, &(x[0])); TP = VBYI(VADD(Tv, TO)); T12 = VSUB(TY, T11); ST(&(x[WS(rs, 1)]), VADD(TP, T12), ms, &(x[WS(rs, 1)])); ST(&(x[WS(rs, 14)]), VSUB(T12, TP), ms, &(x[0])); T1b = VBYI(VSUB(T16, T15)); T1c = VSUB(T19, T18); ST(&(x[WS(rs, 7)]), VADD(T1b, T1c), ms, &(x[WS(rs, 1)])); ST(&(x[WS(rs, 8)]), VSUB(T1c, T1b), ms, &(x[0])); { V T17, T1a, T13, T14; T17 = VBYI(VADD(T15, T16)); T1a = VADD(T18, T19); ST(&(x[WS(rs, 2)]), VADD(T17, T1a), ms, &(x[0])); ST(&(x[WS(rs, 13)]), VSUB(T1a, T17), ms, &(x[WS(rs, 1)])); T13 = VBYI(VSUB(Tv, TO)); T14 = VADD(T11, TY); ST(&(x[WS(rs, 4)]), VADD(T13, T14), ms, &(x[0])); ST(&(x[WS(rs, 11)]), VSUB(T14, T13), ms, &(x[WS(rs, 1)])); } } } } } VLEAVE(); }
static void n2fv_20(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs) { DVK(KP587785252, +0.587785252292473129168705954639072768597652438); DVK(KP951056516, +0.951056516295153572116439333379382143405698634); DVK(KP250000000, +0.250000000000000000000000000000000000000000000); DVK(KP559016994, +0.559016994374947424102293417182819058860154590); INT i; const R *xi; R *xo; xi = ri; xo = ro; for (i = v; i > 0; i = i - VL, xi = xi + (VL * ivs), xo = xo + (VL * ovs), MAKE_VOLATILE_STRIDE(is), MAKE_VOLATILE_STRIDE(os)) { V T3, T1B, Tm, T1i, TG, TN, TO, TH, T13, T16, T1k, T1u, T1v, T1z, T1r; V T1s, T1y, T1a, T1d, T1j, Ti, TD, TB, TL; { V T1, T2, T1g, Tk, Tl, T1h; T1 = LD(&(xi[0]), ivs, &(xi[0])); T2 = LD(&(xi[WS(is, 10)]), ivs, &(xi[0])); T1g = VADD(T1, T2); Tk = LD(&(xi[WS(is, 5)]), ivs, &(xi[WS(is, 1)])); Tl = LD(&(xi[WS(is, 15)]), ivs, &(xi[WS(is, 1)])); T1h = VADD(Tk, Tl); T3 = VSUB(T1, T2); T1B = VADD(T1g, T1h); Tm = VSUB(Tk, Tl); T1i = VSUB(T1g, T1h); } { V T6, T18, Tw, T12, Tz, T15, T9, T1b, Td, T11, Tp, T19, Ts, T1c, Tg; V T14; { V T4, T5, Tu, Tv; T4 = LD(&(xi[WS(is, 4)]), ivs, &(xi[0])); T5 = LD(&(xi[WS(is, 14)]), ivs, &(xi[0])); T6 = VSUB(T4, T5); T18 = VADD(T4, T5); Tu = LD(&(xi[WS(is, 13)]), ivs, &(xi[WS(is, 1)])); Tv = LD(&(xi[WS(is, 3)]), ivs, &(xi[WS(is, 1)])); Tw = VSUB(Tu, Tv); T12 = VADD(Tu, Tv); } { V Tx, Ty, T7, T8; Tx = LD(&(xi[WS(is, 17)]), ivs, &(xi[WS(is, 1)])); Ty = LD(&(xi[WS(is, 7)]), ivs, &(xi[WS(is, 1)])); Tz = VSUB(Tx, Ty); T15 = VADD(Tx, Ty); T7 = LD(&(xi[WS(is, 16)]), ivs, &(xi[0])); T8 = LD(&(xi[WS(is, 6)]), ivs, &(xi[0])); T9 = VSUB(T7, T8); T1b = VADD(T7, T8); } { V Tb, Tc, Tn, To; Tb = LD(&(xi[WS(is, 8)]), ivs, &(xi[0])); Tc = LD(&(xi[WS(is, 18)]), ivs, &(xi[0])); Td = VSUB(Tb, Tc); T11 = VADD(Tb, Tc); Tn = LD(&(xi[WS(is, 9)]), ivs, &(xi[WS(is, 1)])); To = LD(&(xi[WS(is, 19)]), ivs, &(xi[WS(is, 1)])); Tp = VSUB(Tn, To); T19 = VADD(Tn, To); } { V Tq, Tr, Te, Tf; Tq = LD(&(xi[WS(is, 1)]), ivs, &(xi[WS(is, 1)])); Tr = LD(&(xi[WS(is, 11)]), ivs, &(xi[WS(is, 1)])); Ts = VSUB(Tq, Tr); T1c = VADD(Tq, Tr); Te = LD(&(xi[WS(is, 12)]), ivs, &(xi[0])); Tf = LD(&(xi[WS(is, 2)]), ivs, &(xi[0])); Tg = VSUB(Te, Tf); T14 = VADD(Te, Tf); } TG = VSUB(Ts, Tp); TN = VSUB(T6, T9); TO = VSUB(Td, Tg); TH = VSUB(Tz, Tw); T13 = VSUB(T11, T12); T16 = VSUB(T14, T15); T1k = VADD(T13, T16); T1u = VADD(T11, T12); T1v = VADD(T14, T15); T1z = VADD(T1u, T1v); T1r = VADD(T18, T19); T1s = VADD(T1b, T1c); T1y = VADD(T1r, T1s); T1a = VSUB(T18, T19); T1d = VSUB(T1b, T1c); T1j = VADD(T1a, T1d); { V Ta, Th, Tt, TA; Ta = VADD(T6, T9); Th = VADD(Td, Tg); Ti = VADD(Ta, Th); TD = VMUL(LDK(KP559016994), VSUB(Ta, Th)); Tt = VADD(Tp, Ts); TA = VADD(Tw, Tz); TB = VADD(Tt, TA); TL = VMUL(LDK(KP559016994), VSUB(TA, Tt)); } } { V T1I, T1J, T1K, T1L, T1N, T1H, Tj, TC; Tj = VADD(T3, Ti); TC = VBYI(VADD(Tm, TB)); T1H = VSUB(Tj, TC); STM2(&(xo[10]), T1H, ovs, &(xo[2])); T1I = VADD(Tj, TC); STM2(&(xo[30]), T1I, ovs, &(xo[2])); { V T1A, T1C, T1D, T1x, T1G, T1t, T1w, T1F, T1E, T1M; T1A = VMUL(LDK(KP559016994), VSUB(T1y, T1z)); T1C = VADD(T1y, T1z); T1D = VFNMS(LDK(KP250000000), T1C, T1B); T1t = VSUB(T1r, T1s); T1w = VSUB(T1u, T1v); T1x = VBYI(VFMA(LDK(KP951056516), T1t, VMUL(LDK(KP587785252), T1w))); T1G = VBYI(VFNMS(LDK(KP587785252), T1t, VMUL(LDK(KP951056516), T1w))); T1J = VADD(T1B, T1C); STM2(&(xo[0]), T1J, ovs, &(xo[0])); T1F = VSUB(T1D, T1A); T1K = VSUB(T1F, T1G); STM2(&(xo[16]), T1K, ovs, &(xo[0])); T1L = VADD(T1G, T1F); STM2(&(xo[24]), T1L, ovs, &(xo[0])); T1E = VADD(T1A, T1D); T1M = VADD(T1x, T1E); STM2(&(xo[8]), T1M, ovs, &(xo[0])); STN2(&(xo[8]), T1M, T1H, ovs); T1N = VSUB(T1E, T1x); STM2(&(xo[32]), T1N, ovs, &(xo[0])); } { V T1O, T1P, T1R, T1S; { V T1n, T1l, T1m, T1f, T1q, T17, T1e, T1p, T1Q, T1o; T1n = VMUL(LDK(KP559016994), VSUB(T1j, T1k)); T1l = VADD(T1j, T1k); T1m = VFNMS(LDK(KP250000000), T1l, T1i); T17 = VSUB(T13, T16); T1e = VSUB(T1a, T1d); T1f = VBYI(VFNMS(LDK(KP587785252), T1e, VMUL(LDK(KP951056516), T17))); T1q = VBYI(VFMA(LDK(KP951056516), T1e, VMUL(LDK(KP587785252), T17))); T1O = VADD(T1i, T1l); STM2(&(xo[20]), T1O, ovs, &(xo[0])); T1p = VADD(T1n, T1m); T1P = VSUB(T1p, T1q); STM2(&(xo[12]), T1P, ovs, &(xo[0])); T1Q = VADD(T1q, T1p); STM2(&(xo[28]), T1Q, ovs, &(xo[0])); STN2(&(xo[28]), T1Q, T1I, ovs); T1o = VSUB(T1m, T1n); T1R = VADD(T1f, T1o); STM2(&(xo[4]), T1R, ovs, &(xo[0])); T1S = VSUB(T1o, T1f); STM2(&(xo[36]), T1S, ovs, &(xo[0])); } { V TI, TP, TX, TU, TM, TW, TF, TT, TK, TE; TI = VFMA(LDK(KP951056516), TG, VMUL(LDK(KP587785252), TH)); TP = VFMA(LDK(KP951056516), TN, VMUL(LDK(KP587785252), TO)); TX = VFNMS(LDK(KP587785252), TN, VMUL(LDK(KP951056516), TO)); TU = VFNMS(LDK(KP587785252), TG, VMUL(LDK(KP951056516), TH)); TK = VFMS(LDK(KP250000000), TB, Tm); TM = VADD(TK, TL); TW = VSUB(TL, TK); TE = VFNMS(LDK(KP250000000), Ti, T3); TF = VADD(TD, TE); TT = VSUB(TE, TD); { V TJ, TQ, T1T, T1U; TJ = VADD(TF, TI); TQ = VBYI(VSUB(TM, TP)); T1T = VSUB(TJ, TQ); STM2(&(xo[38]), T1T, ovs, &(xo[2])); STN2(&(xo[36]), T1S, T1T, ovs); T1U = VADD(TJ, TQ); STM2(&(xo[2]), T1U, ovs, &(xo[2])); STN2(&(xo[0]), T1J, T1U, ovs); } { V TZ, T10, T1V, T1W; TZ = VADD(TT, TU); T10 = VBYI(VADD(TX, TW)); T1V = VSUB(TZ, T10); STM2(&(xo[26]), T1V, ovs, &(xo[2])); STN2(&(xo[24]), T1L, T1V, ovs); T1W = VADD(TZ, T10); STM2(&(xo[14]), T1W, ovs, &(xo[2])); STN2(&(xo[12]), T1P, T1W, ovs); } { V TR, TS, T1X, T1Y; TR = VSUB(TF, TI); TS = VBYI(VADD(TP, TM)); T1X = VSUB(TR, TS); STM2(&(xo[22]), T1X, ovs, &(xo[2])); STN2(&(xo[20]), T1O, T1X, ovs); T1Y = VADD(TR, TS); STM2(&(xo[18]), T1Y, ovs, &(xo[2])); STN2(&(xo[16]), T1K, T1Y, ovs); } { V TV, TY, T1Z, T20; TV = VSUB(TT, TU); TY = VBYI(VSUB(TW, TX)); T1Z = VSUB(TV, TY); STM2(&(xo[34]), T1Z, ovs, &(xo[2])); STN2(&(xo[32]), T1N, T1Z, ovs); T20 = VADD(TV, TY); STM2(&(xo[6]), T20, ovs, &(xo[2])); STN2(&(xo[4]), T1R, T20, ovs); } } } } } }
static void n1fv_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(KP587785252, +0.587785252292473129168705954639072768597652438); DVK(KP951056516, +0.951056516295153572116439333379382143405698634); 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 = ri; xo = ro; for (i = v; i > 0; i = i - VL, xi = xi + (VL * ivs), xo = xo + (VL * ovs), MAKE_VOLATILE_STRIDE(is), MAKE_VOLATILE_STRIDE(os)) { V T5, T10, TB, TO, TU, TV, TR, Ta, Tf, Tg, Tl, Tq, Tr, TE, TH; V TI, TZ, T11, T1f, T1g; { V T1, T2, T3, T4; 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])); T4 = VADD(T2, T3); T5 = VADD(T1, T4); T10 = VSUB(T3, T2); TB = VFNMS(LDK(KP500000000), T4, T1); } { V T6, T9, TC, TP, Tm, Tp, TG, TN, Tb, Te, TD, TQ, Th, Tk, TF; V TM, TX, TY; { V T7, T8, 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)])); T9 = VADD(T7, T8); TC = VFNMS(LDK(KP500000000), T9, T6); TP = VSUB(T8, T7); 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])); Tp = VADD(Tn, To); TG = VFNMS(LDK(KP500000000), Tp, Tm); TN = VSUB(To, Tn); } { V Tc, Td, Ti, Tj; Tb = LD(&(xi[WS(is, 12)]), ivs, &(xi[0])); Tc = LD(&(xi[WS(is, 2)]), ivs, &(xi[0])); Td = LD(&(xi[WS(is, 7)]), ivs, &(xi[WS(is, 1)])); Te = VADD(Tc, Td); TD = VFNMS(LDK(KP500000000), Te, Tb); TQ = VSUB(Td, Tc); Th = LD(&(xi[WS(is, 6)]), ivs, &(xi[0])); Ti = LD(&(xi[WS(is, 11)]), ivs, &(xi[WS(is, 1)])); Tj = LD(&(xi[WS(is, 1)]), ivs, &(xi[WS(is, 1)])); Tk = VADD(Ti, Tj); TF = VFNMS(LDK(KP500000000), Tk, Th); TM = VSUB(Tj, Ti); } TO = VSUB(TM, TN); TU = VSUB(TF, TG); TV = VSUB(TC, TD); TR = VSUB(TP, TQ); Ta = VADD(T6, T9); Tf = VADD(Tb, Te); Tg = VADD(Ta, Tf); Tl = VADD(Th, Tk); Tq = VADD(Tm, Tp); Tr = VADD(Tl, Tq); TE = VADD(TC, TD); TH = VADD(TF, TG); TI = VADD(TE, TH); TX = VADD(TP, TQ); TY = VADD(TM, TN); TZ = VMUL(LDK(KP484122918), VSUB(TX, TY)); T11 = VADD(TX, TY); } T1f = VADD(TB, TI); T1g = VBYI(VMUL(LDK(KP866025403), VADD(T10, T11))); ST(&(xo[WS(os, 5)]), VSUB(T1f, T1g), ovs, &(xo[WS(os, 1)])); ST(&(xo[WS(os, 10)]), VADD(T1f, T1g), ovs, &(xo[0])); { V Tu, Ts, Tt, Ty, TA, Tw, Tx, Tz, Tv; Tu = VMUL(LDK(KP559016994), VSUB(Tg, Tr)); Ts = VADD(Tg, Tr); Tt = VFNMS(LDK(KP250000000), Ts, T5); Tw = VSUB(Tl, Tq); Tx = VSUB(Ta, Tf); Ty = VBYI(VFNMS(LDK(KP587785252), Tx, VMUL(LDK(KP951056516), Tw))); TA = VBYI(VFMA(LDK(KP951056516), Tx, VMUL(LDK(KP587785252), Tw))); ST(&(xo[0]), VADD(T5, Ts), ovs, &(xo[0])); Tz = VADD(Tu, Tt); ST(&(xo[WS(os, 6)]), VSUB(Tz, TA), ovs, &(xo[0])); ST(&(xo[WS(os, 9)]), VADD(TA, Tz), ovs, &(xo[WS(os, 1)])); Tv = VSUB(Tt, Tu); ST(&(xo[WS(os, 3)]), VSUB(Tv, Ty), ovs, &(xo[WS(os, 1)])); ST(&(xo[WS(os, 12)]), VADD(Ty, Tv), ovs, &(xo[0])); } { V TS, TW, T1b, T18, T13, T1a, TL, T17, T12, TJ, TK; TS = VFNMS(LDK(KP509036960), TR, VMUL(LDK(KP823639103), TO)); TW = VFNMS(LDK(KP587785252), TV, VMUL(LDK(KP951056516), TU)); T1b = VFMA(LDK(KP951056516), TV, VMUL(LDK(KP587785252), TU)); T18 = VFMA(LDK(KP823639103), TR, VMUL(LDK(KP509036960), TO)); T12 = VFNMS(LDK(KP216506350), T11, VMUL(LDK(KP866025403), T10)); T13 = VSUB(TZ, T12); T1a = VADD(TZ, T12); TJ = VFNMS(LDK(KP250000000), TI, TB); TK = VMUL(LDK(KP559016994), VSUB(TE, TH)); TL = VSUB(TJ, TK); T17 = VADD(TK, TJ); { V TT, T14, T1d, T1e; TT = VSUB(TL, TS); T14 = VBYI(VSUB(TW, T13)); ST(&(xo[WS(os, 8)]), VSUB(TT, T14), ovs, &(xo[0])); ST(&(xo[WS(os, 7)]), VADD(TT, T14), ovs, &(xo[WS(os, 1)])); T1d = VSUB(T17, T18); T1e = VBYI(VADD(T1b, T1a)); ST(&(xo[WS(os, 11)]), VSUB(T1d, T1e), ovs, &(xo[WS(os, 1)])); ST(&(xo[WS(os, 4)]), VADD(T1d, T1e), ovs, &(xo[0])); } { V T15, T16, T19, T1c; T15 = VADD(TL, TS); T16 = VBYI(VADD(TW, T13)); ST(&(xo[WS(os, 13)]), VSUB(T15, T16), ovs, &(xo[WS(os, 1)])); ST(&(xo[WS(os, 2)]), VADD(T15, T16), ovs, &(xo[0])); T19 = VADD(T17, T18); T1c = VBYI(VSUB(T1a, T1b)); ST(&(xo[WS(os, 14)]), VSUB(T19, T1c), ovs, &(xo[0])); ST(&(xo[WS(os, 1)]), VADD(T19, T1c), ovs, &(xo[WS(os, 1)])); } } } }
static void hc2cbdftv_8(R *Rp, R *Ip, R *Rm, R *Im, const R *W, stride rs, INT mb, INT me, INT ms) { DVK(KP707106781, +0.707106781186547524400844362104849039284835938); { INT m; for (m = mb, W = W + ((mb - 1) * ((TWVL / VL) * 14)); m < me; m = m + VL, Rp = Rp + (VL * ms), Ip = Ip + (VL * ms), Rm = Rm - (VL * ms), Im = Im - (VL * ms), W = W + (TWVL * 14), MAKE_VOLATILE_STRIDE(32, rs)) { V TJ, T4, Tf, TB, TD, TE, Tm, T1, Tj, TF, Tp, Tb, Tg, Tt, Tx; V T2, T3, Td, Te, T5, T6, T8, T9, Tn, T7, To, Ta, Tk, Tl, TG; V TL, Tq, Tc, Tu, Th, Tv, Ty, Tw, TC, Ti, TK, TA, Tz, TI, TH; V Ts, Tr, TN, TM; T2 = LD(&(Rp[0]), ms, &(Rp[0])); T3 = LD(&(Rm[WS(rs, 3)]), -ms, &(Rm[WS(rs, 1)])); Td = LD(&(Rp[WS(rs, 2)]), ms, &(Rp[0])); Te = LD(&(Rm[WS(rs, 1)]), -ms, &(Rm[WS(rs, 1)])); T5 = LD(&(Rp[WS(rs, 1)]), ms, &(Rp[WS(rs, 1)])); T6 = LD(&(Rm[WS(rs, 2)]), -ms, &(Rm[0])); T8 = LD(&(Rp[WS(rs, 3)]), ms, &(Rp[WS(rs, 1)])); T9 = LD(&(Rm[0]), -ms, &(Rm[0])); TJ = LDW(&(W[0])); Tk = VFMACONJ(T3, T2); T4 = VFNMSCONJ(T3, T2); Tl = VFMACONJ(Te, Td); Tf = VFNMSCONJ(Te, Td); Tn = VFMACONJ(T6, T5); T7 = VFNMSCONJ(T6, T5); To = VFMACONJ(T9, T8); Ta = VFMSCONJ(T9, T8); TB = LDW(&(W[TWVL * 8])); TD = LDW(&(W[TWVL * 6])); TE = VADD(Tk, Tl); Tm = VSUB(Tk, Tl); T1 = LDW(&(W[TWVL * 12])); Tj = LDW(&(W[TWVL * 10])); TF = VADD(Tn, To); Tp = VSUB(Tn, To); Tb = VADD(T7, Ta); Tg = VSUB(T7, Ta); Tt = LDW(&(W[TWVL * 4])); Tx = LDW(&(W[TWVL * 2])); TG = VZMUL(TD, VSUB(TE, TF)); TL = VADD(TE, TF); Tq = VZMUL(Tj, VFNMSI(Tp, Tm)); Tc = VFMA(LDK(KP707106781), Tb, T4); Tu = VFNMS(LDK(KP707106781), Tb, T4); Th = VFMA(LDK(KP707106781), Tg, Tf); Tv = VFNMS(LDK(KP707106781), Tg, Tf); Ty = VZMUL(Tx, VFMAI(Tp, Tm)); Tw = VZMULI(Tt, VFNMSI(Tv, Tu)); TC = VZMULI(TB, VFMAI(Tv, Tu)); Ti = VZMULI(T1, VFNMSI(Th, Tc)); TK = VZMULI(TJ, VFMAI(Th, Tc)); TA = VCONJ(VSUB(Ty, Tw)); Tz = VADD(Tw, Ty); TI = VCONJ(VSUB(TG, TC)); TH = VADD(TC, TG); Ts = VCONJ(VSUB(Tq, Ti)); Tr = VADD(Ti, Tq); TN = VCONJ(VSUB(TL, TK)); TM = VADD(TK, TL); ST(&(Rm[WS(rs, 1)]), TA, -ms, &(Rm[WS(rs, 1)])); ST(&(Rp[WS(rs, 1)]), Tz, ms, &(Rp[WS(rs, 1)])); ST(&(Rm[WS(rs, 2)]), TI, -ms, &(Rm[0])); ST(&(Rp[WS(rs, 2)]), TH, ms, &(Rp[0])); ST(&(Rm[WS(rs, 3)]), Ts, -ms, &(Rm[WS(rs, 1)])); ST(&(Rp[WS(rs, 3)]), Tr, ms, &(Rp[WS(rs, 1)])); ST(&(Rm[0]), TN, -ms, &(Rm[0])); ST(&(Rp[0]), TM, ms, &(Rp[0])); } } VLEAVE(); }