Chris@82: /*
Chris@82:  * Copyright (c) 2003, 2007-14 Matteo Frigo
Chris@82:  * Copyright (c) 2003, 2007-14 Massachusetts Institute of Technology
Chris@82:  *
Chris@82:  * This program is free software; you can redistribute it and/or modify
Chris@82:  * it under the terms of the GNU General Public License as published by
Chris@82:  * the Free Software Foundation; either version 2 of the License, or
Chris@82:  * (at your option) any later version.
Chris@82:  *
Chris@82:  * This program is distributed in the hope that it will be useful,
Chris@82:  * but WITHOUT ANY WARRANTY; without even the implied warranty of
Chris@82:  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
Chris@82:  * GNU General Public License for more details.
Chris@82:  *
Chris@82:  * You should have received a copy of the GNU General Public License
Chris@82:  * along with this program; if not, write to the Free Software
Chris@82:  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02110-1301  USA
Chris@82:  *
Chris@82:  */
Chris@82: 
Chris@82: /* This file was automatically generated --- DO NOT EDIT */
Chris@82: /* Generated on Thu May 24 08:05:59 EDT 2018 */
Chris@82: 
Chris@82: #include "dft/codelet-dft.h"
Chris@82: 
Chris@82: #if defined(ARCH_PREFERS_FMA) || defined(ISA_EXTENSION_PREFERS_FMA)
Chris@82: 
Chris@82: /* Generated by: ../../../genfft/gen_twiddle_c.native -fma -simd -compact -variables 4 -pipeline-latency 8 -n 15 -name t1bv_15 -include dft/simd/t1b.h -sign 1 */
Chris@82: 
Chris@82: /*
Chris@82:  * This function contains 92 FP additions, 77 FP multiplications,
Chris@82:  * (or, 50 additions, 35 multiplications, 42 fused multiply/add),
Chris@82:  * 50 stack variables, 8 constants, and 30 memory accesses
Chris@82:  */
Chris@82: #include "dft/simd/t1b.h"
Chris@82: 
Chris@82: static void t1bv_15(R *ri, R *ii, const R *W, stride rs, INT mb, INT me, INT ms)
Chris@82: {
Chris@82:      DVK(KP910592997, +0.910592997310029334643087372129977886038870291);
Chris@82:      DVK(KP823639103, +0.823639103546331925877420039278190003029660514);
Chris@82:      DVK(KP866025403, +0.866025403784438646763723170752936183471402627);
Chris@82:      DVK(KP559016994, +0.559016994374947424102293417182819058860154590);
Chris@82:      DVK(KP618033988, +0.618033988749894848204586834365638117720309180);
Chris@82:      DVK(KP951056516, +0.951056516295153572116439333379382143405698634);
Chris@82:      DVK(KP250000000, +0.250000000000000000000000000000000000000000000);
Chris@82:      DVK(KP500000000, +0.500000000000000000000000000000000000000000000);
Chris@82:      {
Chris@82: 	  INT m;
Chris@82: 	  R *x;
Chris@82: 	  x = ii;
Chris@82: 	  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)) {
Chris@82: 	       V TV, T7, T1f, TM, TP, Tf, Tn, To, T1j, T1k, T1l, TW, TX, TY, Tw;
Chris@82: 	       V TE, TF, T1g, T1h, T1i;
Chris@82: 	       {
Chris@82: 		    V T1, T5, T3, T4, T2, T6;
Chris@82: 		    T1 = LD(&(x[0]), ms, &(x[0]));
Chris@82: 		    T4 = LD(&(x[WS(rs, 10)]), ms, &(x[0]));
Chris@82: 		    T5 = BYTW(&(W[TWVL * 18]), T4);
Chris@82: 		    T2 = LD(&(x[WS(rs, 5)]), ms, &(x[WS(rs, 1)]));
Chris@82: 		    T3 = BYTW(&(W[TWVL * 8]), T2);
Chris@82: 		    TV = VSUB(T3, T5);
Chris@82: 		    T6 = VADD(T3, T5);
Chris@82: 		    T7 = VFNMS(LDK(KP500000000), T6, T1);
Chris@82: 		    T1f = VADD(T1, T6);
Chris@82: 	       }
Chris@82: 	       {
Chris@82: 		    V T9, Tq, Ty, Th, Te, TK, Tv, TN, TD, TO, Tm, TL;
Chris@82: 		    {
Chris@82: 			 V T8, Tp, Tx, Tg;
Chris@82: 			 T8 = LD(&(x[WS(rs, 3)]), ms, &(x[WS(rs, 1)]));
Chris@82: 			 T9 = BYTW(&(W[TWVL * 4]), T8);
Chris@82: 			 Tp = LD(&(x[WS(rs, 6)]), ms, &(x[0]));
Chris@82: 			 Tq = BYTW(&(W[TWVL * 10]), Tp);
Chris@82: 			 Tx = LD(&(x[WS(rs, 9)]), ms, &(x[WS(rs, 1)]));
Chris@82: 			 Ty = BYTW(&(W[TWVL * 16]), Tx);
Chris@82: 			 Tg = LD(&(x[WS(rs, 12)]), ms, &(x[0]));
Chris@82: 			 Th = BYTW(&(W[TWVL * 22]), Tg);
Chris@82: 		    }
Chris@82: 		    {
Chris@82: 			 V Tb, Td, Ta, Tc;
Chris@82: 			 Ta = LD(&(x[WS(rs, 8)]), ms, &(x[0]));
Chris@82: 			 Tb = BYTW(&(W[TWVL * 14]), Ta);
Chris@82: 			 Tc = LD(&(x[WS(rs, 13)]), ms, &(x[WS(rs, 1)]));
Chris@82: 			 Td = BYTW(&(W[TWVL * 24]), Tc);
Chris@82: 			 Te = VADD(Tb, Td);
Chris@82: 			 TK = VSUB(Tb, Td);
Chris@82: 		    }
Chris@82: 		    {
Chris@82: 			 V Ts, Tu, Tr, Tt;
Chris@82: 			 Tr = LD(&(x[WS(rs, 11)]), ms, &(x[WS(rs, 1)]));
Chris@82: 			 Ts = BYTW(&(W[TWVL * 20]), Tr);
Chris@82: 			 Tt = LD(&(x[WS(rs, 1)]), ms, &(x[WS(rs, 1)]));
Chris@82: 			 Tu = BYTW(&(W[0]), Tt);
Chris@82: 			 Tv = VADD(Ts, Tu);
Chris@82: 			 TN = VSUB(Ts, Tu);
Chris@82: 		    }
Chris@82: 		    {
Chris@82: 			 V TA, TC, Tz, TB;
Chris@82: 			 Tz = LD(&(x[WS(rs, 14)]), ms, &(x[0]));
Chris@82: 			 TA = BYTW(&(W[TWVL * 26]), Tz);
Chris@82: 			 TB = LD(&(x[WS(rs, 4)]), ms, &(x[0]));
Chris@82: 			 TC = BYTW(&(W[TWVL * 6]), TB);
Chris@82: 			 TD = VADD(TA, TC);
Chris@82: 			 TO = VSUB(TA, TC);
Chris@82: 		    }
Chris@82: 		    {
Chris@82: 			 V Tj, Tl, Ti, Tk;
Chris@82: 			 Ti = LD(&(x[WS(rs, 2)]), ms, &(x[0]));
Chris@82: 			 Tj = BYTW(&(W[TWVL * 2]), Ti);
Chris@82: 			 Tk = LD(&(x[WS(rs, 7)]), ms, &(x[WS(rs, 1)]));
Chris@82: 			 Tl = BYTW(&(W[TWVL * 12]), Tk);
Chris@82: 			 Tm = VADD(Tj, Tl);
Chris@82: 			 TL = VSUB(Tj, Tl);
Chris@82: 		    }
Chris@82: 		    TM = VSUB(TK, TL);
Chris@82: 		    TP = VSUB(TN, TO);
Chris@82: 		    Tf = VFNMS(LDK(KP500000000), Te, T9);
Chris@82: 		    Tn = VFNMS(LDK(KP500000000), Tm, Th);
Chris@82: 		    To = VADD(Tf, Tn);
Chris@82: 		    T1j = VADD(Tq, Tv);
Chris@82: 		    T1k = VADD(Ty, TD);
Chris@82: 		    T1l = VADD(T1j, T1k);
Chris@82: 		    TW = VADD(TK, TL);
Chris@82: 		    TX = VADD(TN, TO);
Chris@82: 		    TY = VADD(TW, TX);
Chris@82: 		    Tw = VFNMS(LDK(KP500000000), Tv, Tq);
Chris@82: 		    TE = VFNMS(LDK(KP500000000), TD, Ty);
Chris@82: 		    TF = VADD(Tw, TE);
Chris@82: 		    T1g = VADD(T9, Te);
Chris@82: 		    T1h = VADD(Th, Tm);
Chris@82: 		    T1i = VADD(T1g, T1h);
Chris@82: 	       }
Chris@82: 	       {
Chris@82: 		    V T1o, T1m, T1n, T1s, T1u, T1q, T1r, T1t, T1p;
Chris@82: 		    T1o = VSUB(T1i, T1l);
Chris@82: 		    T1m = VADD(T1i, T1l);
Chris@82: 		    T1n = VFNMS(LDK(KP250000000), T1m, T1f);
Chris@82: 		    T1q = VSUB(T1j, T1k);
Chris@82: 		    T1r = VSUB(T1g, T1h);
Chris@82: 		    T1s = VMUL(LDK(KP951056516), VFNMS(LDK(KP618033988), T1r, T1q));
Chris@82: 		    T1u = VMUL(LDK(KP951056516), VFMA(LDK(KP618033988), T1q, T1r));
Chris@82: 		    ST(&(x[0]), VADD(T1f, T1m), ms, &(x[0]));
Chris@82: 		    T1t = VFMA(LDK(KP559016994), T1o, T1n);
Chris@82: 		    ST(&(x[WS(rs, 6)]), VFMAI(T1u, T1t), ms, &(x[0]));
Chris@82: 		    ST(&(x[WS(rs, 9)]), VFNMSI(T1u, T1t), ms, &(x[WS(rs, 1)]));
Chris@82: 		    T1p = VFNMS(LDK(KP559016994), T1o, T1n);
Chris@82: 		    ST(&(x[WS(rs, 3)]), VFMAI(T1s, T1p), ms, &(x[WS(rs, 1)]));
Chris@82: 		    ST(&(x[WS(rs, 12)]), VFNMSI(T1s, T1p), ms, &(x[0]));
Chris@82: 	       }
Chris@82: 	       {
Chris@82: 		    V TQ, T16, T1e, T11, T19, TU, T18, TJ, T1d, T15, TZ, T10;
Chris@82: 		    TQ = VFMA(LDK(KP618033988), TP, TM);
Chris@82: 		    T16 = VFNMS(LDK(KP618033988), TM, TP);
Chris@82: 		    T1e = VMUL(LDK(KP866025403), VADD(TV, TY));
Chris@82: 		    TZ = VFNMS(LDK(KP250000000), TY, TV);
Chris@82: 		    T10 = VSUB(TW, TX);
Chris@82: 		    T11 = VFMA(LDK(KP559016994), T10, TZ);
Chris@82: 		    T19 = VFNMS(LDK(KP559016994), T10, TZ);
Chris@82: 		    {
Chris@82: 			 V TS, TT, TI, TG, TH;
Chris@82: 			 TS = VSUB(Tf, Tn);
Chris@82: 			 TT = VSUB(Tw, TE);
Chris@82: 			 TU = VFMA(LDK(KP618033988), TT, TS);
Chris@82: 			 T18 = VFNMS(LDK(KP618033988), TS, TT);
Chris@82: 			 TI = VSUB(To, TF);
Chris@82: 			 TG = VADD(To, TF);
Chris@82: 			 TH = VFNMS(LDK(KP250000000), TG, T7);
Chris@82: 			 TJ = VFMA(LDK(KP559016994), TI, TH);
Chris@82: 			 T1d = VADD(T7, TG);
Chris@82: 			 T15 = VFNMS(LDK(KP559016994), TI, TH);
Chris@82: 		    }
Chris@82: 		    {
Chris@82: 			 V TR, T12, T1b, T1c;
Chris@82: 			 ST(&(x[WS(rs, 5)]), VFNMSI(T1e, T1d), ms, &(x[WS(rs, 1)]));
Chris@82: 			 ST(&(x[WS(rs, 10)]), VFMAI(T1e, T1d), ms, &(x[0]));
Chris@82: 			 TR = VFNMS(LDK(KP823639103), TQ, TJ);
Chris@82: 			 T12 = VMUL(LDK(KP951056516), VFMA(LDK(KP910592997), T11, TU));
Chris@82: 			 ST(&(x[WS(rs, 1)]), VFMAI(T12, TR), ms, &(x[WS(rs, 1)]));
Chris@82: 			 ST(&(x[WS(rs, 14)]), VFNMSI(T12, TR), ms, &(x[0]));
Chris@82: 			 T1b = VFMA(LDK(KP823639103), T16, T15);
Chris@82: 			 T1c = VMUL(LDK(KP951056516), VFNMS(LDK(KP910592997), T19, T18));
Chris@82: 			 ST(&(x[WS(rs, 7)]), VFNMSI(T1c, T1b), ms, &(x[WS(rs, 1)]));
Chris@82: 			 ST(&(x[WS(rs, 8)]), VFMAI(T1c, T1b), ms, &(x[0]));
Chris@82: 			 {
Chris@82: 			      V T17, T1a, T13, T14;
Chris@82: 			      T17 = VFNMS(LDK(KP823639103), T16, T15);
Chris@82: 			      T1a = VMUL(LDK(KP951056516), VFMA(LDK(KP910592997), T19, T18));
Chris@82: 			      ST(&(x[WS(rs, 2)]), VFNMSI(T1a, T17), ms, &(x[0]));
Chris@82: 			      ST(&(x[WS(rs, 13)]), VFMAI(T1a, T17), ms, &(x[WS(rs, 1)]));
Chris@82: 			      T13 = VFMA(LDK(KP823639103), TQ, TJ);
Chris@82: 			      T14 = VMUL(LDK(KP951056516), VFNMS(LDK(KP910592997), T11, TU));
Chris@82: 			      ST(&(x[WS(rs, 4)]), VFNMSI(T14, T13), ms, &(x[0]));
Chris@82: 			      ST(&(x[WS(rs, 11)]), VFMAI(T14, T13), ms, &(x[WS(rs, 1)]));
Chris@82: 			 }
Chris@82: 		    }
Chris@82: 	       }
Chris@82: 	  }
Chris@82:      }
Chris@82:      VLEAVE();
Chris@82: }
Chris@82: 
Chris@82: static const tw_instr twinstr[] = {
Chris@82:      VTW(0, 1),
Chris@82:      VTW(0, 2),
Chris@82:      VTW(0, 3),
Chris@82:      VTW(0, 4),
Chris@82:      VTW(0, 5),
Chris@82:      VTW(0, 6),
Chris@82:      VTW(0, 7),
Chris@82:      VTW(0, 8),
Chris@82:      VTW(0, 9),
Chris@82:      VTW(0, 10),
Chris@82:      VTW(0, 11),
Chris@82:      VTW(0, 12),
Chris@82:      VTW(0, 13),
Chris@82:      VTW(0, 14),
Chris@82:      {TW_NEXT, VL, 0}
Chris@82: };
Chris@82: 
Chris@82: static const ct_desc desc = { 15, XSIMD_STRING("t1bv_15"), twinstr, &GENUS, {50, 35, 42, 0}, 0, 0, 0 };
Chris@82: 
Chris@82: void XSIMD(codelet_t1bv_15) (planner *p) {
Chris@82:      X(kdft_dit_register) (p, t1bv_15, &desc);
Chris@82: }
Chris@82: #else
Chris@82: 
Chris@82: /* Generated by: ../../../genfft/gen_twiddle_c.native -simd -compact -variables 4 -pipeline-latency 8 -n 15 -name t1bv_15 -include dft/simd/t1b.h -sign 1 */
Chris@82: 
Chris@82: /*
Chris@82:  * This function contains 92 FP additions, 53 FP multiplications,
Chris@82:  * (or, 78 additions, 39 multiplications, 14 fused multiply/add),
Chris@82:  * 52 stack variables, 10 constants, and 30 memory accesses
Chris@82:  */
Chris@82: #include "dft/simd/t1b.h"
Chris@82: 
Chris@82: static void t1bv_15(R *ri, R *ii, const R *W, stride rs, INT mb, INT me, INT ms)
Chris@82: {
Chris@82:      DVK(KP216506350, +0.216506350946109661690930792688234045867850657);
Chris@82:      DVK(KP484122918, +0.484122918275927110647408174972799951354115213);
Chris@82:      DVK(KP866025403, +0.866025403784438646763723170752936183471402627);
Chris@82:      DVK(KP509036960, +0.509036960455127183450980863393907648510733164);
Chris@82:      DVK(KP823639103, +0.823639103546331925877420039278190003029660514);
Chris@82:      DVK(KP951056516, +0.951056516295153572116439333379382143405698634);
Chris@82:      DVK(KP587785252, +0.587785252292473129168705954639072768597652438);
Chris@82:      DVK(KP250000000, +0.250000000000000000000000000000000000000000000);
Chris@82:      DVK(KP559016994, +0.559016994374947424102293417182819058860154590);
Chris@82:      DVK(KP500000000, +0.500000000000000000000000000000000000000000000);
Chris@82:      {
Chris@82: 	  INT m;
Chris@82: 	  R *x;
Chris@82: 	  x = ii;
Chris@82: 	  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)) {
Chris@82: 	       V Ts, TV, T1f, TZ, T10, Tb, Tm, Tt, T1j, T1k, T1l, TI, TM, TR, Tz;
Chris@82: 	       V TD, TQ, T1g, T1h, T1i;
Chris@82: 	       {
Chris@82: 		    V TT, Tr, Tp, Tq, To, TU;
Chris@82: 		    TT = LD(&(x[0]), ms, &(x[0]));
Chris@82: 		    Tq = LD(&(x[WS(rs, 10)]), ms, &(x[0]));
Chris@82: 		    Tr = BYTW(&(W[TWVL * 18]), Tq);
Chris@82: 		    To = LD(&(x[WS(rs, 5)]), ms, &(x[WS(rs, 1)]));
Chris@82: 		    Tp = BYTW(&(W[TWVL * 8]), To);
Chris@82: 		    Ts = VSUB(Tp, Tr);
Chris@82: 		    TU = VADD(Tp, Tr);
Chris@82: 		    TV = VFNMS(LDK(KP500000000), TU, TT);
Chris@82: 		    T1f = VADD(TT, TU);
Chris@82: 	       }
Chris@82: 	       {
Chris@82: 		    V Tx, TG, TK, TB, T5, Ty, Tg, TH, Tl, TL, Ta, TC;
Chris@82: 		    {
Chris@82: 			 V Tw, TF, TJ, TA;
Chris@82: 			 Tw = LD(&(x[WS(rs, 3)]), ms, &(x[WS(rs, 1)]));
Chris@82: 			 Tx = BYTW(&(W[TWVL * 4]), Tw);
Chris@82: 			 TF = LD(&(x[WS(rs, 6)]), ms, &(x[0]));
Chris@82: 			 TG = BYTW(&(W[TWVL * 10]), TF);
Chris@82: 			 TJ = LD(&(x[WS(rs, 9)]), ms, &(x[WS(rs, 1)]));
Chris@82: 			 TK = BYTW(&(W[TWVL * 16]), TJ);
Chris@82: 			 TA = LD(&(x[WS(rs, 12)]), ms, &(x[0]));
Chris@82: 			 TB = BYTW(&(W[TWVL * 22]), TA);
Chris@82: 		    }
Chris@82: 		    {
Chris@82: 			 V T2, T4, T1, T3;
Chris@82: 			 T1 = LD(&(x[WS(rs, 8)]), ms, &(x[0]));
Chris@82: 			 T2 = BYTW(&(W[TWVL * 14]), T1);
Chris@82: 			 T3 = LD(&(x[WS(rs, 13)]), ms, &(x[WS(rs, 1)]));
Chris@82: 			 T4 = BYTW(&(W[TWVL * 24]), T3);
Chris@82: 			 T5 = VSUB(T2, T4);
Chris@82: 			 Ty = VADD(T2, T4);
Chris@82: 		    }
Chris@82: 		    {
Chris@82: 			 V Td, Tf, Tc, Te;
Chris@82: 			 Tc = LD(&(x[WS(rs, 11)]), ms, &(x[WS(rs, 1)]));
Chris@82: 			 Td = BYTW(&(W[TWVL * 20]), Tc);
Chris@82: 			 Te = LD(&(x[WS(rs, 1)]), ms, &(x[WS(rs, 1)]));
Chris@82: 			 Tf = BYTW(&(W[0]), Te);
Chris@82: 			 Tg = VSUB(Td, Tf);
Chris@82: 			 TH = VADD(Td, Tf);
Chris@82: 		    }
Chris@82: 		    {
Chris@82: 			 V Ti, Tk, Th, Tj;
Chris@82: 			 Th = LD(&(x[WS(rs, 14)]), ms, &(x[0]));
Chris@82: 			 Ti = BYTW(&(W[TWVL * 26]), Th);
Chris@82: 			 Tj = LD(&(x[WS(rs, 4)]), ms, &(x[0]));
Chris@82: 			 Tk = BYTW(&(W[TWVL * 6]), Tj);
Chris@82: 			 Tl = VSUB(Ti, Tk);
Chris@82: 			 TL = VADD(Ti, Tk);
Chris@82: 		    }
Chris@82: 		    {
Chris@82: 			 V T7, T9, T6, T8;
Chris@82: 			 T6 = LD(&(x[WS(rs, 2)]), ms, &(x[0]));
Chris@82: 			 T7 = BYTW(&(W[TWVL * 2]), T6);
Chris@82: 			 T8 = LD(&(x[WS(rs, 7)]), ms, &(x[WS(rs, 1)]));
Chris@82: 			 T9 = BYTW(&(W[TWVL * 12]), T8);
Chris@82: 			 Ta = VSUB(T7, T9);
Chris@82: 			 TC = VADD(T7, T9);
Chris@82: 		    }
Chris@82: 		    TZ = VSUB(T5, Ta);
Chris@82: 		    T10 = VSUB(Tg, Tl);
Chris@82: 		    Tb = VADD(T5, Ta);
Chris@82: 		    Tm = VADD(Tg, Tl);
Chris@82: 		    Tt = VADD(Tb, Tm);
Chris@82: 		    T1j = VADD(TG, TH);
Chris@82: 		    T1k = VADD(TK, TL);
Chris@82: 		    T1l = VADD(T1j, T1k);
Chris@82: 		    TI = VFNMS(LDK(KP500000000), TH, TG);
Chris@82: 		    TM = VFNMS(LDK(KP500000000), TL, TK);
Chris@82: 		    TR = VADD(TI, TM);
Chris@82: 		    Tz = VFNMS(LDK(KP500000000), Ty, Tx);
Chris@82: 		    TD = VFNMS(LDK(KP500000000), TC, TB);
Chris@82: 		    TQ = VADD(Tz, TD);
Chris@82: 		    T1g = VADD(Tx, Ty);
Chris@82: 		    T1h = VADD(TB, TC);
Chris@82: 		    T1i = VADD(T1g, T1h);
Chris@82: 	       }
Chris@82: 	       {
Chris@82: 		    V T1o, T1m, T1n, T1s, T1t, T1q, T1r, T1u, T1p;
Chris@82: 		    T1o = VMUL(LDK(KP559016994), VSUB(T1i, T1l));
Chris@82: 		    T1m = VADD(T1i, T1l);
Chris@82: 		    T1n = VFNMS(LDK(KP250000000), T1m, T1f);
Chris@82: 		    T1q = VSUB(T1g, T1h);
Chris@82: 		    T1r = VSUB(T1j, T1k);
Chris@82: 		    T1s = VBYI(VFNMS(LDK(KP951056516), T1r, VMUL(LDK(KP587785252), T1q)));
Chris@82: 		    T1t = VBYI(VFMA(LDK(KP951056516), T1q, VMUL(LDK(KP587785252), T1r)));
Chris@82: 		    ST(&(x[0]), VADD(T1f, T1m), ms, &(x[0]));
Chris@82: 		    T1u = VADD(T1o, T1n);
Chris@82: 		    ST(&(x[WS(rs, 6)]), VADD(T1t, T1u), ms, &(x[0]));
Chris@82: 		    ST(&(x[WS(rs, 9)]), VSUB(T1u, T1t), ms, &(x[WS(rs, 1)]));
Chris@82: 		    T1p = VSUB(T1n, T1o);
Chris@82: 		    ST(&(x[WS(rs, 3)]), VSUB(T1p, T1s), ms, &(x[WS(rs, 1)]));
Chris@82: 		    ST(&(x[WS(rs, 12)]), VADD(T1s, T1p), ms, &(x[0]));
Chris@82: 	       }
Chris@82: 	       {
Chris@82: 		    V T11, T18, T1e, TO, T16, Tv, T15, TY, T1d, T19, TE, TN;
Chris@82: 		    T11 = VFMA(LDK(KP823639103), TZ, VMUL(LDK(KP509036960), T10));
Chris@82: 		    T18 = VFNMS(LDK(KP823639103), T10, VMUL(LDK(KP509036960), TZ));
Chris@82: 		    T1e = VBYI(VMUL(LDK(KP866025403), VADD(Ts, Tt)));
Chris@82: 		    TE = VSUB(Tz, TD);
Chris@82: 		    TN = VSUB(TI, TM);
Chris@82: 		    TO = VFMA(LDK(KP951056516), TE, VMUL(LDK(KP587785252), TN));
Chris@82: 		    T16 = VFNMS(LDK(KP951056516), TN, VMUL(LDK(KP587785252), TE));
Chris@82: 		    {
Chris@82: 			 V Tn, Tu, TS, TW, TX;
Chris@82: 			 Tn = VMUL(LDK(KP484122918), VSUB(Tb, Tm));
Chris@82: 			 Tu = VFNMS(LDK(KP216506350), Tt, VMUL(LDK(KP866025403), Ts));
Chris@82: 			 Tv = VADD(Tn, Tu);
Chris@82: 			 T15 = VSUB(Tn, Tu);
Chris@82: 			 TS = VMUL(LDK(KP559016994), VSUB(TQ, TR));
Chris@82: 			 TW = VADD(TQ, TR);
Chris@82: 			 TX = VFNMS(LDK(KP250000000), TW, TV);
Chris@82: 			 TY = VADD(TS, TX);
Chris@82: 			 T1d = VADD(TV, TW);
Chris@82: 			 T19 = VSUB(TX, TS);
Chris@82: 		    }
Chris@82: 		    {
Chris@82: 			 V TP, T12, T1b, T1c;
Chris@82: 			 ST(&(x[WS(rs, 5)]), VSUB(T1d, T1e), ms, &(x[WS(rs, 1)]));
Chris@82: 			 ST(&(x[WS(rs, 10)]), VADD(T1e, T1d), ms, &(x[0]));
Chris@82: 			 TP = VBYI(VADD(Tv, TO));
Chris@82: 			 T12 = VSUB(TY, T11);
Chris@82: 			 ST(&(x[WS(rs, 1)]), VADD(TP, T12), ms, &(x[WS(rs, 1)]));
Chris@82: 			 ST(&(x[WS(rs, 14)]), VSUB(T12, TP), ms, &(x[0]));
Chris@82: 			 T1b = VBYI(VSUB(T16, T15));
Chris@82: 			 T1c = VSUB(T19, T18);
Chris@82: 			 ST(&(x[WS(rs, 7)]), VADD(T1b, T1c), ms, &(x[WS(rs, 1)]));
Chris@82: 			 ST(&(x[WS(rs, 8)]), VSUB(T1c, T1b), ms, &(x[0]));
Chris@82: 			 {
Chris@82: 			      V T17, T1a, T13, T14;
Chris@82: 			      T17 = VBYI(VADD(T15, T16));
Chris@82: 			      T1a = VADD(T18, T19);
Chris@82: 			      ST(&(x[WS(rs, 2)]), VADD(T17, T1a), ms, &(x[0]));
Chris@82: 			      ST(&(x[WS(rs, 13)]), VSUB(T1a, T17), ms, &(x[WS(rs, 1)]));
Chris@82: 			      T13 = VBYI(VSUB(Tv, TO));
Chris@82: 			      T14 = VADD(T11, TY);
Chris@82: 			      ST(&(x[WS(rs, 4)]), VADD(T13, T14), ms, &(x[0]));
Chris@82: 			      ST(&(x[WS(rs, 11)]), VSUB(T14, T13), ms, &(x[WS(rs, 1)]));
Chris@82: 			 }
Chris@82: 		    }
Chris@82: 	       }
Chris@82: 	  }
Chris@82:      }
Chris@82:      VLEAVE();
Chris@82: }
Chris@82: 
Chris@82: static const tw_instr twinstr[] = {
Chris@82:      VTW(0, 1),
Chris@82:      VTW(0, 2),
Chris@82:      VTW(0, 3),
Chris@82:      VTW(0, 4),
Chris@82:      VTW(0, 5),
Chris@82:      VTW(0, 6),
Chris@82:      VTW(0, 7),
Chris@82:      VTW(0, 8),
Chris@82:      VTW(0, 9),
Chris@82:      VTW(0, 10),
Chris@82:      VTW(0, 11),
Chris@82:      VTW(0, 12),
Chris@82:      VTW(0, 13),
Chris@82:      VTW(0, 14),
Chris@82:      {TW_NEXT, VL, 0}
Chris@82: };
Chris@82: 
Chris@82: static const ct_desc desc = { 15, XSIMD_STRING("t1bv_15"), twinstr, &GENUS, {78, 39, 14, 0}, 0, 0, 0 };
Chris@82: 
Chris@82: void XSIMD(codelet_t1bv_15) (planner *p) {
Chris@82:      X(kdft_dit_register) (p, t1bv_15, &desc);
Chris@82: }
Chris@82: #endif