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 08:04:10 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.native -fma -compact -variables 4 -pipeline-latency 4 -n 12 -name n1_12 -include dft/scalar/n.h */
Chris@82: 
Chris@82: /*
Chris@82:  * This function contains 96 FP additions, 24 FP multiplications,
Chris@82:  * (or, 72 additions, 0 multiplications, 24 fused multiply/add),
Chris@82:  * 43 stack variables, 2 constants, and 48 memory accesses
Chris@82:  */
Chris@82: #include "dft/scalar/n.h"
Chris@82: 
Chris@82: static void n1_12(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs)
Chris@82: {
Chris@82:      DK(KP866025403, +0.866025403784438646763723170752936183471402627);
Chris@82:      DK(KP500000000, +0.500000000000000000000000000000000000000000000);
Chris@82:      {
Chris@82: 	  INT i;
Chris@82: 	  for (i = v; i > 0; i = i - 1, ri = ri + ivs, ii = ii + ivs, ro = ro + ovs, io = io + ovs, MAKE_VOLATILE_STRIDE(48, is), MAKE_VOLATILE_STRIDE(48, os)) {
Chris@82: 	       E T5, TR, TA, Ts, TS, Tz, Ta, TU, TD, Tx, TV, TC, Tg, T1d, TG;
Chris@82: 	       E TJ, T1u, T1c, Tl, T1i, TL, TO, T1v, T1h;
Chris@82: 	       {
Chris@82: 		    E T1, T2, T3, T4;
Chris@82: 		    T1 = ri[0];
Chris@82: 		    T2 = ri[WS(is, 4)];
Chris@82: 		    T3 = ri[WS(is, 8)];
Chris@82: 		    T4 = T2 + T3;
Chris@82: 		    T5 = T1 + T4;
Chris@82: 		    TR = FNMS(KP500000000, T4, T1);
Chris@82: 		    TA = T3 - T2;
Chris@82: 	       }
Chris@82: 	       {
Chris@82: 		    E To, Tp, Tq, Tr;
Chris@82: 		    To = ii[0];
Chris@82: 		    Tp = ii[WS(is, 4)];
Chris@82: 		    Tq = ii[WS(is, 8)];
Chris@82: 		    Tr = Tp + Tq;
Chris@82: 		    Ts = To + Tr;
Chris@82: 		    TS = Tp - Tq;
Chris@82: 		    Tz = FNMS(KP500000000, Tr, To);
Chris@82: 	       }
Chris@82: 	       {
Chris@82: 		    E T6, T7, T8, T9;
Chris@82: 		    T6 = ri[WS(is, 6)];
Chris@82: 		    T7 = ri[WS(is, 10)];
Chris@82: 		    T8 = ri[WS(is, 2)];
Chris@82: 		    T9 = T7 + T8;
Chris@82: 		    Ta = T6 + T9;
Chris@82: 		    TU = FNMS(KP500000000, T9, T6);
Chris@82: 		    TD = T8 - T7;
Chris@82: 	       }
Chris@82: 	       {
Chris@82: 		    E Tt, Tu, Tv, Tw;
Chris@82: 		    Tt = ii[WS(is, 6)];
Chris@82: 		    Tu = ii[WS(is, 10)];
Chris@82: 		    Tv = ii[WS(is, 2)];
Chris@82: 		    Tw = Tu + Tv;
Chris@82: 		    Tx = Tt + Tw;
Chris@82: 		    TV = Tu - Tv;
Chris@82: 		    TC = FNMS(KP500000000, Tw, Tt);
Chris@82: 	       }
Chris@82: 	       {
Chris@82: 		    E Tc, Td, Te, Tf;
Chris@82: 		    Tc = ri[WS(is, 3)];
Chris@82: 		    Td = ri[WS(is, 7)];
Chris@82: 		    Te = ri[WS(is, 11)];
Chris@82: 		    Tf = Td + Te;
Chris@82: 		    Tg = Tc + Tf;
Chris@82: 		    T1d = Te - Td;
Chris@82: 		    TG = FNMS(KP500000000, Tf, Tc);
Chris@82: 	       }
Chris@82: 	       {
Chris@82: 		    E T1a, TH, TI, T1b;
Chris@82: 		    T1a = ii[WS(is, 3)];
Chris@82: 		    TH = ii[WS(is, 7)];
Chris@82: 		    TI = ii[WS(is, 11)];
Chris@82: 		    T1b = TH + TI;
Chris@82: 		    TJ = TH - TI;
Chris@82: 		    T1u = T1a + T1b;
Chris@82: 		    T1c = FNMS(KP500000000, T1b, T1a);
Chris@82: 	       }
Chris@82: 	       {
Chris@82: 		    E Th, Ti, Tj, Tk;
Chris@82: 		    Th = ri[WS(is, 9)];
Chris@82: 		    Ti = ri[WS(is, 1)];
Chris@82: 		    Tj = ri[WS(is, 5)];
Chris@82: 		    Tk = Ti + Tj;
Chris@82: 		    Tl = Th + Tk;
Chris@82: 		    T1i = Tj - Ti;
Chris@82: 		    TL = FNMS(KP500000000, Tk, Th);
Chris@82: 	       }
Chris@82: 	       {
Chris@82: 		    E T1f, TM, TN, T1g;
Chris@82: 		    T1f = ii[WS(is, 9)];
Chris@82: 		    TM = ii[WS(is, 1)];
Chris@82: 		    TN = ii[WS(is, 5)];
Chris@82: 		    T1g = TM + TN;
Chris@82: 		    TO = TM - TN;
Chris@82: 		    T1v = T1f + T1g;
Chris@82: 		    T1h = FNMS(KP500000000, T1g, T1f);
Chris@82: 	       }
Chris@82: 	       {
Chris@82: 		    E Tb, Tm, T1t, T1w;
Chris@82: 		    Tb = T5 + Ta;
Chris@82: 		    Tm = Tg + Tl;
Chris@82: 		    ro[WS(os, 6)] = Tb - Tm;
Chris@82: 		    ro[0] = Tb + Tm;
Chris@82: 		    {
Chris@82: 			 E T1x, T1y, Tn, Ty;
Chris@82: 			 T1x = Ts + Tx;
Chris@82: 			 T1y = T1u + T1v;
Chris@82: 			 io[WS(os, 6)] = T1x - T1y;
Chris@82: 			 io[0] = T1x + T1y;
Chris@82: 			 Tn = Tg - Tl;
Chris@82: 			 Ty = Ts - Tx;
Chris@82: 			 io[WS(os, 3)] = Tn + Ty;
Chris@82: 			 io[WS(os, 9)] = Ty - Tn;
Chris@82: 		    }
Chris@82: 		    T1t = T5 - Ta;
Chris@82: 		    T1w = T1u - T1v;
Chris@82: 		    ro[WS(os, 3)] = T1t - T1w;
Chris@82: 		    ro[WS(os, 9)] = T1t + T1w;
Chris@82: 		    {
Chris@82: 			 E T11, T1l, T1k, T1m, T14, T18, T17, T19;
Chris@82: 			 {
Chris@82: 			      E TZ, T10, T1e, T1j;
Chris@82: 			      TZ = FMA(KP866025403, TA, Tz);
Chris@82: 			      T10 = FMA(KP866025403, TD, TC);
Chris@82: 			      T11 = TZ - T10;
Chris@82: 			      T1l = TZ + T10;
Chris@82: 			      T1e = FMA(KP866025403, T1d, T1c);
Chris@82: 			      T1j = FMA(KP866025403, T1i, T1h);
Chris@82: 			      T1k = T1e - T1j;
Chris@82: 			      T1m = T1e + T1j;
Chris@82: 			 }
Chris@82: 			 {
Chris@82: 			      E T12, T13, T15, T16;
Chris@82: 			      T12 = FMA(KP866025403, TJ, TG);
Chris@82: 			      T13 = FMA(KP866025403, TO, TL);
Chris@82: 			      T14 = T12 - T13;
Chris@82: 			      T18 = T12 + T13;
Chris@82: 			      T15 = FMA(KP866025403, TS, TR);
Chris@82: 			      T16 = FMA(KP866025403, TV, TU);
Chris@82: 			      T17 = T15 + T16;
Chris@82: 			      T19 = T15 - T16;
Chris@82: 			 }
Chris@82: 			 io[WS(os, 1)] = T11 - T14;
Chris@82: 			 ro[WS(os, 1)] = T19 + T1k;
Chris@82: 			 io[WS(os, 7)] = T11 + T14;
Chris@82: 			 ro[WS(os, 7)] = T19 - T1k;
Chris@82: 			 ro[WS(os, 10)] = T17 - T18;
Chris@82: 			 io[WS(os, 10)] = T1l - T1m;
Chris@82: 			 ro[WS(os, 4)] = T17 + T18;
Chris@82: 			 io[WS(os, 4)] = T1l + T1m;
Chris@82: 		    }
Chris@82: 		    {
Chris@82: 			 E TF, T1r, T1q, T1s, TQ, TY, TX, T1n;
Chris@82: 			 {
Chris@82: 			      E TB, TE, T1o, T1p;
Chris@82: 			      TB = FNMS(KP866025403, TA, Tz);
Chris@82: 			      TE = FNMS(KP866025403, TD, TC);
Chris@82: 			      TF = TB - TE;
Chris@82: 			      T1r = TB + TE;
Chris@82: 			      T1o = FNMS(KP866025403, T1d, T1c);
Chris@82: 			      T1p = FNMS(KP866025403, T1i, T1h);
Chris@82: 			      T1q = T1o - T1p;
Chris@82: 			      T1s = T1o + T1p;
Chris@82: 			 }
Chris@82: 			 {
Chris@82: 			      E TK, TP, TT, TW;
Chris@82: 			      TK = FNMS(KP866025403, TJ, TG);
Chris@82: 			      TP = FNMS(KP866025403, TO, TL);
Chris@82: 			      TQ = TK - TP;
Chris@82: 			      TY = TK + TP;
Chris@82: 			      TT = FNMS(KP866025403, TS, TR);
Chris@82: 			      TW = FNMS(KP866025403, TV, TU);
Chris@82: 			      TX = TT + TW;
Chris@82: 			      T1n = TT - TW;
Chris@82: 			 }
Chris@82: 			 io[WS(os, 5)] = TF - TQ;
Chris@82: 			 ro[WS(os, 5)] = T1n + T1q;
Chris@82: 			 io[WS(os, 11)] = TF + TQ;
Chris@82: 			 ro[WS(os, 11)] = T1n - T1q;
Chris@82: 			 ro[WS(os, 2)] = TX - TY;
Chris@82: 			 io[WS(os, 2)] = T1r - T1s;
Chris@82: 			 ro[WS(os, 8)] = TX + TY;
Chris@82: 			 io[WS(os, 8)] = T1r + T1s;
Chris@82: 		    }
Chris@82: 	       }
Chris@82: 	  }
Chris@82:      }
Chris@82: }
Chris@82: 
Chris@82: static const kdft_desc desc = { 12, "n1_12", {72, 0, 24, 0}, &GENUS, 0, 0, 0, 0 };
Chris@82: 
Chris@82: void X(codelet_n1_12) (planner *p) {
Chris@82:      X(kdft_register) (p, n1_12, &desc);
Chris@82: }
Chris@82: 
Chris@82: #else
Chris@82: 
Chris@82: /* Generated by: ../../../genfft/gen_notw.native -compact -variables 4 -pipeline-latency 4 -n 12 -name n1_12 -include dft/scalar/n.h */
Chris@82: 
Chris@82: /*
Chris@82:  * This function contains 96 FP additions, 16 FP multiplications,
Chris@82:  * (or, 88 additions, 8 multiplications, 8 fused multiply/add),
Chris@82:  * 43 stack variables, 2 constants, and 48 memory accesses
Chris@82:  */
Chris@82: #include "dft/scalar/n.h"
Chris@82: 
Chris@82: static void n1_12(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs)
Chris@82: {
Chris@82:      DK(KP866025403, +0.866025403784438646763723170752936183471402627);
Chris@82:      DK(KP500000000, +0.500000000000000000000000000000000000000000000);
Chris@82:      {
Chris@82: 	  INT i;
Chris@82: 	  for (i = v; i > 0; i = i - 1, ri = ri + ivs, ii = ii + ivs, ro = ro + ovs, io = io + ovs, MAKE_VOLATILE_STRIDE(48, is), MAKE_VOLATILE_STRIDE(48, os)) {
Chris@82: 	       E T5, TR, TA, Ts, TS, Tz, Ta, TU, TD, Tx, TV, TC, Tg, T1a, TG;
Chris@82: 	       E TJ, T1u, T1d, Tl, T1f, TL, TO, T1v, T1i;
Chris@82: 	       {
Chris@82: 		    E T1, T2, T3, T4;
Chris@82: 		    T1 = ri[0];
Chris@82: 		    T2 = ri[WS(is, 4)];
Chris@82: 		    T3 = ri[WS(is, 8)];
Chris@82: 		    T4 = T2 + T3;
Chris@82: 		    T5 = T1 + T4;
Chris@82: 		    TR = FNMS(KP500000000, T4, T1);
Chris@82: 		    TA = KP866025403 * (T3 - T2);
Chris@82: 	       }
Chris@82: 	       {
Chris@82: 		    E To, Tp, Tq, Tr;
Chris@82: 		    To = ii[0];
Chris@82: 		    Tp = ii[WS(is, 4)];
Chris@82: 		    Tq = ii[WS(is, 8)];
Chris@82: 		    Tr = Tp + Tq;
Chris@82: 		    Ts = To + Tr;
Chris@82: 		    TS = KP866025403 * (Tp - Tq);
Chris@82: 		    Tz = FNMS(KP500000000, Tr, To);
Chris@82: 	       }
Chris@82: 	       {
Chris@82: 		    E T6, T7, T8, T9;
Chris@82: 		    T6 = ri[WS(is, 6)];
Chris@82: 		    T7 = ri[WS(is, 10)];
Chris@82: 		    T8 = ri[WS(is, 2)];
Chris@82: 		    T9 = T7 + T8;
Chris@82: 		    Ta = T6 + T9;
Chris@82: 		    TU = FNMS(KP500000000, T9, T6);
Chris@82: 		    TD = KP866025403 * (T8 - T7);
Chris@82: 	       }
Chris@82: 	       {
Chris@82: 		    E Tt, Tu, Tv, Tw;
Chris@82: 		    Tt = ii[WS(is, 6)];
Chris@82: 		    Tu = ii[WS(is, 10)];
Chris@82: 		    Tv = ii[WS(is, 2)];
Chris@82: 		    Tw = Tu + Tv;
Chris@82: 		    Tx = Tt + Tw;
Chris@82: 		    TV = KP866025403 * (Tu - Tv);
Chris@82: 		    TC = FNMS(KP500000000, Tw, Tt);
Chris@82: 	       }
Chris@82: 	       {
Chris@82: 		    E Tc, Td, Te, Tf;
Chris@82: 		    Tc = ri[WS(is, 3)];
Chris@82: 		    Td = ri[WS(is, 7)];
Chris@82: 		    Te = ri[WS(is, 11)];
Chris@82: 		    Tf = Td + Te;
Chris@82: 		    Tg = Tc + Tf;
Chris@82: 		    T1a = KP866025403 * (Te - Td);
Chris@82: 		    TG = FNMS(KP500000000, Tf, Tc);
Chris@82: 	       }
Chris@82: 	       {
Chris@82: 		    E T1b, TH, TI, T1c;
Chris@82: 		    T1b = ii[WS(is, 3)];
Chris@82: 		    TH = ii[WS(is, 7)];
Chris@82: 		    TI = ii[WS(is, 11)];
Chris@82: 		    T1c = TH + TI;
Chris@82: 		    TJ = KP866025403 * (TH - TI);
Chris@82: 		    T1u = T1b + T1c;
Chris@82: 		    T1d = FNMS(KP500000000, T1c, T1b);
Chris@82: 	       }
Chris@82: 	       {
Chris@82: 		    E Th, Ti, Tj, Tk;
Chris@82: 		    Th = ri[WS(is, 9)];
Chris@82: 		    Ti = ri[WS(is, 1)];
Chris@82: 		    Tj = ri[WS(is, 5)];
Chris@82: 		    Tk = Ti + Tj;
Chris@82: 		    Tl = Th + Tk;
Chris@82: 		    T1f = KP866025403 * (Tj - Ti);
Chris@82: 		    TL = FNMS(KP500000000, Tk, Th);
Chris@82: 	       }
Chris@82: 	       {
Chris@82: 		    E T1g, TM, TN, T1h;
Chris@82: 		    T1g = ii[WS(is, 9)];
Chris@82: 		    TM = ii[WS(is, 1)];
Chris@82: 		    TN = ii[WS(is, 5)];
Chris@82: 		    T1h = TM + TN;
Chris@82: 		    TO = KP866025403 * (TM - TN);
Chris@82: 		    T1v = T1g + T1h;
Chris@82: 		    T1i = FNMS(KP500000000, T1h, T1g);
Chris@82: 	       }
Chris@82: 	       {
Chris@82: 		    E Tb, Tm, T1t, T1w;
Chris@82: 		    Tb = T5 + Ta;
Chris@82: 		    Tm = Tg + Tl;
Chris@82: 		    ro[WS(os, 6)] = Tb - Tm;
Chris@82: 		    ro[0] = Tb + Tm;
Chris@82: 		    {
Chris@82: 			 E T1x, T1y, Tn, Ty;
Chris@82: 			 T1x = Ts + Tx;
Chris@82: 			 T1y = T1u + T1v;
Chris@82: 			 io[WS(os, 6)] = T1x - T1y;
Chris@82: 			 io[0] = T1x + T1y;
Chris@82: 			 Tn = Tg - Tl;
Chris@82: 			 Ty = Ts - Tx;
Chris@82: 			 io[WS(os, 3)] = Tn + Ty;
Chris@82: 			 io[WS(os, 9)] = Ty - Tn;
Chris@82: 		    }
Chris@82: 		    T1t = T5 - Ta;
Chris@82: 		    T1w = T1u - T1v;
Chris@82: 		    ro[WS(os, 3)] = T1t - T1w;
Chris@82: 		    ro[WS(os, 9)] = T1t + T1w;
Chris@82: 		    {
Chris@82: 			 E T11, T1l, T1k, T1m, T14, T18, T17, T19;
Chris@82: 			 {
Chris@82: 			      E TZ, T10, T1e, T1j;
Chris@82: 			      TZ = TA + Tz;
Chris@82: 			      T10 = TD + TC;
Chris@82: 			      T11 = TZ - T10;
Chris@82: 			      T1l = TZ + T10;
Chris@82: 			      T1e = T1a + T1d;
Chris@82: 			      T1j = T1f + T1i;
Chris@82: 			      T1k = T1e - T1j;
Chris@82: 			      T1m = T1e + T1j;
Chris@82: 			 }
Chris@82: 			 {
Chris@82: 			      E T12, T13, T15, T16;
Chris@82: 			      T12 = TG + TJ;
Chris@82: 			      T13 = TL + TO;
Chris@82: 			      T14 = T12 - T13;
Chris@82: 			      T18 = T12 + T13;
Chris@82: 			      T15 = TR + TS;
Chris@82: 			      T16 = TU + TV;
Chris@82: 			      T17 = T15 + T16;
Chris@82: 			      T19 = T15 - T16;
Chris@82: 			 }
Chris@82: 			 io[WS(os, 1)] = T11 - T14;
Chris@82: 			 ro[WS(os, 1)] = T19 + T1k;
Chris@82: 			 io[WS(os, 7)] = T11 + T14;
Chris@82: 			 ro[WS(os, 7)] = T19 - T1k;
Chris@82: 			 ro[WS(os, 10)] = T17 - T18;
Chris@82: 			 io[WS(os, 10)] = T1l - T1m;
Chris@82: 			 ro[WS(os, 4)] = T17 + T18;
Chris@82: 			 io[WS(os, 4)] = T1l + T1m;
Chris@82: 		    }
Chris@82: 		    {
Chris@82: 			 E TF, T1r, T1q, T1s, TQ, TY, TX, T1n;
Chris@82: 			 {
Chris@82: 			      E TB, TE, T1o, T1p;
Chris@82: 			      TB = Tz - TA;
Chris@82: 			      TE = TC - TD;
Chris@82: 			      TF = TB - TE;
Chris@82: 			      T1r = TB + TE;
Chris@82: 			      T1o = T1d - T1a;
Chris@82: 			      T1p = T1i - T1f;
Chris@82: 			      T1q = T1o - T1p;
Chris@82: 			      T1s = T1o + T1p;
Chris@82: 			 }
Chris@82: 			 {
Chris@82: 			      E TK, TP, TT, TW;
Chris@82: 			      TK = TG - TJ;
Chris@82: 			      TP = TL - TO;
Chris@82: 			      TQ = TK - TP;
Chris@82: 			      TY = TK + TP;
Chris@82: 			      TT = TR - TS;
Chris@82: 			      TW = TU - TV;
Chris@82: 			      TX = TT + TW;
Chris@82: 			      T1n = TT - TW;
Chris@82: 			 }
Chris@82: 			 io[WS(os, 5)] = TF - TQ;
Chris@82: 			 ro[WS(os, 5)] = T1n + T1q;
Chris@82: 			 io[WS(os, 11)] = TF + TQ;
Chris@82: 			 ro[WS(os, 11)] = T1n - T1q;
Chris@82: 			 ro[WS(os, 2)] = TX - TY;
Chris@82: 			 io[WS(os, 2)] = T1r - T1s;
Chris@82: 			 ro[WS(os, 8)] = TX + TY;
Chris@82: 			 io[WS(os, 8)] = T1r + T1s;
Chris@82: 		    }
Chris@82: 	       }
Chris@82: 	  }
Chris@82:      }
Chris@82: }
Chris@82: 
Chris@82: static const kdft_desc desc = { 12, "n1_12", {88, 8, 8, 0}, &GENUS, 0, 0, 0, 0 };
Chris@82: 
Chris@82: void X(codelet_n1_12) (planner *p) {
Chris@82:      X(kdft_register) (p, n1_12, &desc);
Chris@82: }
Chris@82: 
Chris@82: #endif