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 08:04:52 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_notw_c.native -fma -simd -compact -variables 4 -pipeline-latency 8 -n 16 -name n1fv_16 -include dft/simd/n1f.h */
Chris@82: 
Chris@82: /*
Chris@82:  * This function contains 72 FP additions, 34 FP multiplications,
Chris@82:  * (or, 38 additions, 0 multiplications, 34 fused multiply/add),
Chris@82:  * 30 stack variables, 3 constants, and 32 memory accesses
Chris@82:  */
Chris@82: #include "dft/simd/n1f.h"
Chris@82: 
Chris@82: static void n1fv_16(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs)
Chris@82: {
Chris@82:      DVK(KP923879532, +0.923879532511286756128183189396788286822416626);
Chris@82:      DVK(KP707106781, +0.707106781186547524400844362104849039284835938);
Chris@82:      DVK(KP414213562, +0.414213562373095048801688724209698078569671875);
Chris@82:      {
Chris@82: 	  INT i;
Chris@82: 	  const R *xi;
Chris@82: 	  R *xo;
Chris@82: 	  xi = ri;
Chris@82: 	  xo = ro;
Chris@82: 	  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)) {
Chris@82: 	       V T7, TU, Tz, TH, Tu, TV, TA, TK, Te, TX, TC, TO, Tl, TY, TD;
Chris@82: 	       V TR;
Chris@82: 	       {
Chris@82: 		    V T1, T2, T3, T4, T5, T6;
Chris@82: 		    T1 = LD(&(xi[0]), ivs, &(xi[0]));
Chris@82: 		    T2 = LD(&(xi[WS(is, 8)]), ivs, &(xi[0]));
Chris@82: 		    T3 = VADD(T1, T2);
Chris@82: 		    T4 = LD(&(xi[WS(is, 4)]), ivs, &(xi[0]));
Chris@82: 		    T5 = LD(&(xi[WS(is, 12)]), ivs, &(xi[0]));
Chris@82: 		    T6 = VADD(T4, T5);
Chris@82: 		    T7 = VSUB(T3, T6);
Chris@82: 		    TU = VSUB(T4, T5);
Chris@82: 		    Tz = VADD(T3, T6);
Chris@82: 		    TH = VSUB(T1, T2);
Chris@82: 	       }
Chris@82: 	       {
Chris@82: 		    V Tq, TJ, Tt, TI;
Chris@82: 		    {
Chris@82: 			 V To, Tp, Tr, Ts;
Chris@82: 			 To = LD(&(xi[WS(is, 14)]), ivs, &(xi[0]));
Chris@82: 			 Tp = LD(&(xi[WS(is, 6)]), ivs, &(xi[0]));
Chris@82: 			 Tq = VADD(To, Tp);
Chris@82: 			 TJ = VSUB(To, Tp);
Chris@82: 			 Tr = LD(&(xi[WS(is, 2)]), ivs, &(xi[0]));
Chris@82: 			 Ts = LD(&(xi[WS(is, 10)]), ivs, &(xi[0]));
Chris@82: 			 Tt = VADD(Tr, Ts);
Chris@82: 			 TI = VSUB(Tr, Ts);
Chris@82: 		    }
Chris@82: 		    Tu = VSUB(Tq, Tt);
Chris@82: 		    TV = VSUB(TJ, TI);
Chris@82: 		    TA = VADD(Tt, Tq);
Chris@82: 		    TK = VADD(TI, TJ);
Chris@82: 	       }
Chris@82: 	       {
Chris@82: 		    V Ta, TM, Td, TN;
Chris@82: 		    {
Chris@82: 			 V T8, T9, Tb, Tc;
Chris@82: 			 T8 = LD(&(xi[WS(is, 1)]), ivs, &(xi[WS(is, 1)]));
Chris@82: 			 T9 = LD(&(xi[WS(is, 9)]), ivs, &(xi[WS(is, 1)]));
Chris@82: 			 Ta = VADD(T8, T9);
Chris@82: 			 TM = VSUB(T8, T9);
Chris@82: 			 Tb = LD(&(xi[WS(is, 5)]), ivs, &(xi[WS(is, 1)]));
Chris@82: 			 Tc = LD(&(xi[WS(is, 13)]), ivs, &(xi[WS(is, 1)]));
Chris@82: 			 Td = VADD(Tb, Tc);
Chris@82: 			 TN = VSUB(Tb, Tc);
Chris@82: 		    }
Chris@82: 		    Te = VSUB(Ta, Td);
Chris@82: 		    TX = VFMA(LDK(KP414213562), TM, TN);
Chris@82: 		    TC = VADD(Ta, Td);
Chris@82: 		    TO = VFNMS(LDK(KP414213562), TN, TM);
Chris@82: 	       }
Chris@82: 	       {
Chris@82: 		    V Th, TP, Tk, TQ;
Chris@82: 		    {
Chris@82: 			 V Tf, Tg, Ti, Tj;
Chris@82: 			 Tf = LD(&(xi[WS(is, 15)]), ivs, &(xi[WS(is, 1)]));
Chris@82: 			 Tg = LD(&(xi[WS(is, 7)]), ivs, &(xi[WS(is, 1)]));
Chris@82: 			 Th = VADD(Tf, Tg);
Chris@82: 			 TP = VSUB(Tf, Tg);
Chris@82: 			 Ti = LD(&(xi[WS(is, 3)]), ivs, &(xi[WS(is, 1)]));
Chris@82: 			 Tj = LD(&(xi[WS(is, 11)]), ivs, &(xi[WS(is, 1)]));
Chris@82: 			 Tk = VADD(Ti, Tj);
Chris@82: 			 TQ = VSUB(Tj, Ti);
Chris@82: 		    }
Chris@82: 		    Tl = VSUB(Th, Tk);
Chris@82: 		    TY = VFMA(LDK(KP414213562), TP, TQ);
Chris@82: 		    TD = VADD(Th, Tk);
Chris@82: 		    TR = VFNMS(LDK(KP414213562), TQ, TP);
Chris@82: 	       }
Chris@82: 	       {
Chris@82: 		    V TB, TE, TF, TG;
Chris@82: 		    TB = VADD(Tz, TA);
Chris@82: 		    TE = VADD(TC, TD);
Chris@82: 		    ST(&(xo[WS(os, 8)]), VSUB(TB, TE), ovs, &(xo[0]));
Chris@82: 		    ST(&(xo[0]), VADD(TB, TE), ovs, &(xo[0]));
Chris@82: 		    TF = VSUB(Tz, TA);
Chris@82: 		    TG = VSUB(TD, TC);
Chris@82: 		    ST(&(xo[WS(os, 12)]), VFNMSI(TG, TF), ovs, &(xo[0]));
Chris@82: 		    ST(&(xo[WS(os, 4)]), VFMAI(TG, TF), ovs, &(xo[0]));
Chris@82: 	       }
Chris@82: 	       {
Chris@82: 		    V Tn, Tx, Tw, Ty, Tm, Tv;
Chris@82: 		    Tm = VADD(Te, Tl);
Chris@82: 		    Tn = VFNMS(LDK(KP707106781), Tm, T7);
Chris@82: 		    Tx = VFMA(LDK(KP707106781), Tm, T7);
Chris@82: 		    Tv = VSUB(Tl, Te);
Chris@82: 		    Tw = VFNMS(LDK(KP707106781), Tv, Tu);
Chris@82: 		    Ty = VFMA(LDK(KP707106781), Tv, Tu);
Chris@82: 		    ST(&(xo[WS(os, 6)]), VFNMSI(Tw, Tn), ovs, &(xo[0]));
Chris@82: 		    ST(&(xo[WS(os, 2)]), VFMAI(Ty, Tx), ovs, &(xo[0]));
Chris@82: 		    ST(&(xo[WS(os, 10)]), VFMAI(Tw, Tn), ovs, &(xo[0]));
Chris@82: 		    ST(&(xo[WS(os, 14)]), VFNMSI(Ty, Tx), ovs, &(xo[0]));
Chris@82: 	       }
Chris@82: 	       {
Chris@82: 		    V TT, T11, T10, T12;
Chris@82: 		    {
Chris@82: 			 V TL, TS, TW, TZ;
Chris@82: 			 TL = VFMA(LDK(KP707106781), TK, TH);
Chris@82: 			 TS = VADD(TO, TR);
Chris@82: 			 TT = VFNMS(LDK(KP923879532), TS, TL);
Chris@82: 			 T11 = VFMA(LDK(KP923879532), TS, TL);
Chris@82: 			 TW = VFNMS(LDK(KP707106781), TV, TU);
Chris@82: 			 TZ = VSUB(TX, TY);
Chris@82: 			 T10 = VFNMS(LDK(KP923879532), TZ, TW);
Chris@82: 			 T12 = VFMA(LDK(KP923879532), TZ, TW);
Chris@82: 		    }
Chris@82: 		    ST(&(xo[WS(os, 9)]), VFNMSI(T10, TT), ovs, &(xo[WS(os, 1)]));
Chris@82: 		    ST(&(xo[WS(os, 15)]), VFMAI(T12, T11), ovs, &(xo[WS(os, 1)]));
Chris@82: 		    ST(&(xo[WS(os, 7)]), VFMAI(T10, TT), ovs, &(xo[WS(os, 1)]));
Chris@82: 		    ST(&(xo[WS(os, 1)]), VFNMSI(T12, T11), ovs, &(xo[WS(os, 1)]));
Chris@82: 	       }
Chris@82: 	       {
Chris@82: 		    V T15, T19, T18, T1a;
Chris@82: 		    {
Chris@82: 			 V T13, T14, T16, T17;
Chris@82: 			 T13 = VFNMS(LDK(KP707106781), TK, TH);
Chris@82: 			 T14 = VADD(TX, TY);
Chris@82: 			 T15 = VFNMS(LDK(KP923879532), T14, T13);
Chris@82: 			 T19 = VFMA(LDK(KP923879532), T14, T13);
Chris@82: 			 T16 = VFMA(LDK(KP707106781), TV, TU);
Chris@82: 			 T17 = VSUB(TR, TO);
Chris@82: 			 T18 = VFNMS(LDK(KP923879532), T17, T16);
Chris@82: 			 T1a = VFMA(LDK(KP923879532), T17, T16);
Chris@82: 		    }
Chris@82: 		    ST(&(xo[WS(os, 5)]), VFNMSI(T18, T15), ovs, &(xo[WS(os, 1)]));
Chris@82: 		    ST(&(xo[WS(os, 13)]), VFNMSI(T1a, T19), ovs, &(xo[WS(os, 1)]));
Chris@82: 		    ST(&(xo[WS(os, 11)]), VFMAI(T18, T15), ovs, &(xo[WS(os, 1)]));
Chris@82: 		    ST(&(xo[WS(os, 3)]), VFMAI(T1a, T19), ovs, &(xo[WS(os, 1)]));
Chris@82: 	       }
Chris@82: 	  }
Chris@82:      }
Chris@82:      VLEAVE();
Chris@82: }
Chris@82: 
Chris@82: static const kdft_desc desc = { 16, XSIMD_STRING("n1fv_16"), {38, 0, 34, 0}, &GENUS, 0, 0, 0, 0 };
Chris@82: 
Chris@82: void XSIMD(codelet_n1fv_16) (planner *p) {
Chris@82:      X(kdft_register) (p, n1fv_16, &desc);
Chris@82: }
Chris@82: 
Chris@82: #else
Chris@82: 
Chris@82: /* Generated by: ../../../genfft/gen_notw_c.native -simd -compact -variables 4 -pipeline-latency 8 -n 16 -name n1fv_16 -include dft/simd/n1f.h */
Chris@82: 
Chris@82: /*
Chris@82:  * This function contains 72 FP additions, 12 FP multiplications,
Chris@82:  * (or, 68 additions, 8 multiplications, 4 fused multiply/add),
Chris@82:  * 30 stack variables, 3 constants, and 32 memory accesses
Chris@82:  */
Chris@82: #include "dft/simd/n1f.h"
Chris@82: 
Chris@82: static void n1fv_16(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs)
Chris@82: {
Chris@82:      DVK(KP923879532, +0.923879532511286756128183189396788286822416626);
Chris@82:      DVK(KP382683432, +0.382683432365089771728459984030398866761344562);
Chris@82:      DVK(KP707106781, +0.707106781186547524400844362104849039284835938);
Chris@82:      {
Chris@82: 	  INT i;
Chris@82: 	  const R *xi;
Chris@82: 	  R *xo;
Chris@82: 	  xi = ri;
Chris@82: 	  xo = ro;
Chris@82: 	  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)) {
Chris@82: 	       V Tp, T13, Tu, TN, Tm, T14, Tv, TY, T7, T17, Ty, TT, Te, T16, Tx;
Chris@82: 	       V TQ;
Chris@82: 	       {
Chris@82: 		    V Tn, To, TM, Ts, Tt, TL;
Chris@82: 		    Tn = LD(&(xi[WS(is, 4)]), ivs, &(xi[0]));
Chris@82: 		    To = LD(&(xi[WS(is, 12)]), ivs, &(xi[0]));
Chris@82: 		    TM = VADD(Tn, To);
Chris@82: 		    Ts = LD(&(xi[0]), ivs, &(xi[0]));
Chris@82: 		    Tt = LD(&(xi[WS(is, 8)]), ivs, &(xi[0]));
Chris@82: 		    TL = VADD(Ts, Tt);
Chris@82: 		    Tp = VSUB(Tn, To);
Chris@82: 		    T13 = VADD(TL, TM);
Chris@82: 		    Tu = VSUB(Ts, Tt);
Chris@82: 		    TN = VSUB(TL, TM);
Chris@82: 	       }
Chris@82: 	       {
Chris@82: 		    V Ti, TW, Tl, TX;
Chris@82: 		    {
Chris@82: 			 V Tg, Th, Tj, Tk;
Chris@82: 			 Tg = LD(&(xi[WS(is, 14)]), ivs, &(xi[0]));
Chris@82: 			 Th = LD(&(xi[WS(is, 6)]), ivs, &(xi[0]));
Chris@82: 			 Ti = VSUB(Tg, Th);
Chris@82: 			 TW = VADD(Tg, Th);
Chris@82: 			 Tj = LD(&(xi[WS(is, 2)]), ivs, &(xi[0]));
Chris@82: 			 Tk = LD(&(xi[WS(is, 10)]), ivs, &(xi[0]));
Chris@82: 			 Tl = VSUB(Tj, Tk);
Chris@82: 			 TX = VADD(Tj, Tk);
Chris@82: 		    }
Chris@82: 		    Tm = VMUL(LDK(KP707106781), VSUB(Ti, Tl));
Chris@82: 		    T14 = VADD(TX, TW);
Chris@82: 		    Tv = VMUL(LDK(KP707106781), VADD(Tl, Ti));
Chris@82: 		    TY = VSUB(TW, TX);
Chris@82: 	       }
Chris@82: 	       {
Chris@82: 		    V T3, TR, T6, TS;
Chris@82: 		    {
Chris@82: 			 V T1, T2, T4, T5;
Chris@82: 			 T1 = LD(&(xi[WS(is, 15)]), ivs, &(xi[WS(is, 1)]));
Chris@82: 			 T2 = LD(&(xi[WS(is, 7)]), ivs, &(xi[WS(is, 1)]));
Chris@82: 			 T3 = VSUB(T1, T2);
Chris@82: 			 TR = VADD(T1, T2);
Chris@82: 			 T4 = LD(&(xi[WS(is, 3)]), ivs, &(xi[WS(is, 1)]));
Chris@82: 			 T5 = LD(&(xi[WS(is, 11)]), ivs, &(xi[WS(is, 1)]));
Chris@82: 			 T6 = VSUB(T4, T5);
Chris@82: 			 TS = VADD(T4, T5);
Chris@82: 		    }
Chris@82: 		    T7 = VFNMS(LDK(KP923879532), T6, VMUL(LDK(KP382683432), T3));
Chris@82: 		    T17 = VADD(TR, TS);
Chris@82: 		    Ty = VFMA(LDK(KP923879532), T3, VMUL(LDK(KP382683432), T6));
Chris@82: 		    TT = VSUB(TR, TS);
Chris@82: 	       }
Chris@82: 	       {
Chris@82: 		    V Ta, TO, Td, TP;
Chris@82: 		    {
Chris@82: 			 V T8, T9, Tb, Tc;
Chris@82: 			 T8 = LD(&(xi[WS(is, 1)]), ivs, &(xi[WS(is, 1)]));
Chris@82: 			 T9 = LD(&(xi[WS(is, 9)]), ivs, &(xi[WS(is, 1)]));
Chris@82: 			 Ta = VSUB(T8, T9);
Chris@82: 			 TO = VADD(T8, T9);
Chris@82: 			 Tb = LD(&(xi[WS(is, 5)]), ivs, &(xi[WS(is, 1)]));
Chris@82: 			 Tc = LD(&(xi[WS(is, 13)]), ivs, &(xi[WS(is, 1)]));
Chris@82: 			 Td = VSUB(Tb, Tc);
Chris@82: 			 TP = VADD(Tb, Tc);
Chris@82: 		    }
Chris@82: 		    Te = VFMA(LDK(KP382683432), Ta, VMUL(LDK(KP923879532), Td));
Chris@82: 		    T16 = VADD(TO, TP);
Chris@82: 		    Tx = VFNMS(LDK(KP382683432), Td, VMUL(LDK(KP923879532), Ta));
Chris@82: 		    TQ = VSUB(TO, TP);
Chris@82: 	       }
Chris@82: 	       {
Chris@82: 		    V T15, T18, T19, T1a;
Chris@82: 		    T15 = VADD(T13, T14);
Chris@82: 		    T18 = VADD(T16, T17);
Chris@82: 		    ST(&(xo[WS(os, 8)]), VSUB(T15, T18), ovs, &(xo[0]));
Chris@82: 		    ST(&(xo[0]), VADD(T15, T18), ovs, &(xo[0]));
Chris@82: 		    T19 = VSUB(T13, T14);
Chris@82: 		    T1a = VBYI(VSUB(T17, T16));
Chris@82: 		    ST(&(xo[WS(os, 12)]), VSUB(T19, T1a), ovs, &(xo[0]));
Chris@82: 		    ST(&(xo[WS(os, 4)]), VADD(T19, T1a), ovs, &(xo[0]));
Chris@82: 	       }
Chris@82: 	       {
Chris@82: 		    V TV, T11, T10, T12, TU, TZ;
Chris@82: 		    TU = VMUL(LDK(KP707106781), VADD(TQ, TT));
Chris@82: 		    TV = VADD(TN, TU);
Chris@82: 		    T11 = VSUB(TN, TU);
Chris@82: 		    TZ = VMUL(LDK(KP707106781), VSUB(TT, TQ));
Chris@82: 		    T10 = VBYI(VADD(TY, TZ));
Chris@82: 		    T12 = VBYI(VSUB(TZ, TY));
Chris@82: 		    ST(&(xo[WS(os, 14)]), VSUB(TV, T10), ovs, &(xo[0]));
Chris@82: 		    ST(&(xo[WS(os, 6)]), VADD(T11, T12), ovs, &(xo[0]));
Chris@82: 		    ST(&(xo[WS(os, 2)]), VADD(TV, T10), ovs, &(xo[0]));
Chris@82: 		    ST(&(xo[WS(os, 10)]), VSUB(T11, T12), ovs, &(xo[0]));
Chris@82: 	       }
Chris@82: 	       {
Chris@82: 		    V Tr, TB, TA, TC;
Chris@82: 		    {
Chris@82: 			 V Tf, Tq, Tw, Tz;
Chris@82: 			 Tf = VSUB(T7, Te);
Chris@82: 			 Tq = VSUB(Tm, Tp);
Chris@82: 			 Tr = VBYI(VSUB(Tf, Tq));
Chris@82: 			 TB = VBYI(VADD(Tq, Tf));
Chris@82: 			 Tw = VADD(Tu, Tv);
Chris@82: 			 Tz = VADD(Tx, Ty);
Chris@82: 			 TA = VSUB(Tw, Tz);
Chris@82: 			 TC = VADD(Tw, Tz);
Chris@82: 		    }
Chris@82: 		    ST(&(xo[WS(os, 7)]), VADD(Tr, TA), ovs, &(xo[WS(os, 1)]));
Chris@82: 		    ST(&(xo[WS(os, 15)]), VSUB(TC, TB), ovs, &(xo[WS(os, 1)]));
Chris@82: 		    ST(&(xo[WS(os, 9)]), VSUB(TA, Tr), ovs, &(xo[WS(os, 1)]));
Chris@82: 		    ST(&(xo[WS(os, 1)]), VADD(TB, TC), ovs, &(xo[WS(os, 1)]));
Chris@82: 	       }
Chris@82: 	       {
Chris@82: 		    V TF, TJ, TI, TK;
Chris@82: 		    {
Chris@82: 			 V TD, TE, TG, TH;
Chris@82: 			 TD = VSUB(Tu, Tv);
Chris@82: 			 TE = VADD(Te, T7);
Chris@82: 			 TF = VADD(TD, TE);
Chris@82: 			 TJ = VSUB(TD, TE);
Chris@82: 			 TG = VADD(Tp, Tm);
Chris@82: 			 TH = VSUB(Ty, Tx);
Chris@82: 			 TI = VBYI(VADD(TG, TH));
Chris@82: 			 TK = VBYI(VSUB(TH, TG));
Chris@82: 		    }
Chris@82: 		    ST(&(xo[WS(os, 13)]), VSUB(TF, TI), ovs, &(xo[WS(os, 1)]));
Chris@82: 		    ST(&(xo[WS(os, 5)]), VADD(TJ, TK), ovs, &(xo[WS(os, 1)]));
Chris@82: 		    ST(&(xo[WS(os, 3)]), VADD(TF, TI), ovs, &(xo[WS(os, 1)]));
Chris@82: 		    ST(&(xo[WS(os, 11)]), VSUB(TJ, TK), ovs, &(xo[WS(os, 1)]));
Chris@82: 	       }
Chris@82: 	  }
Chris@82:      }
Chris@82:      VLEAVE();
Chris@82: }
Chris@82: 
Chris@82: static const kdft_desc desc = { 16, XSIMD_STRING("n1fv_16"), {68, 8, 4, 0}, &GENUS, 0, 0, 0, 0 };
Chris@82: 
Chris@82: void XSIMD(codelet_n1fv_16) (planner *p) {
Chris@82:      X(kdft_register) (p, n1fv_16, &desc);
Chris@82: }
Chris@82: 
Chris@82: #endif