Chris@42: /*
Chris@42:  * Copyright (c) 2003, 2007-14 Matteo Frigo
Chris@42:  * Copyright (c) 2003, 2007-14 Massachusetts Institute of Technology
Chris@42:  *
Chris@42:  * This program is free software; you can redistribute it and/or modify
Chris@42:  * it under the terms of the GNU General Public License as published by
Chris@42:  * the Free Software Foundation; either version 2 of the License, or
Chris@42:  * (at your option) any later version.
Chris@42:  *
Chris@42:  * This program is distributed in the hope that it will be useful,
Chris@42:  * but WITHOUT ANY WARRANTY; without even the implied warranty of
Chris@42:  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
Chris@42:  * GNU General Public License for more details.
Chris@42:  *
Chris@42:  * You should have received a copy of the GNU General Public License
Chris@42:  * along with this program; if not, write to the Free Software
Chris@42:  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02110-1301  USA
Chris@42:  *
Chris@42:  */
Chris@42: 
Chris@42: /* This file was automatically generated --- DO NOT EDIT */
Chris@42: /* Generated on Sat Jul 30 16:52:41 EDT 2016 */
Chris@42: 
Chris@42: #include "codelet-rdft.h"
Chris@42: 
Chris@42: #ifdef HAVE_FMA
Chris@42: 
Chris@42: /* Generated by: ../../../genfft/gen_hc2cdft_c.native -fma -reorder-insns -schedule-for-pipeline -simd -compact -variables 4 -pipeline-latency 8 -trivial-stores -variables 32 -no-generate-bytw -n 20 -dit -name hc2cfdftv_20 -include hc2cfv.h */
Chris@42: 
Chris@42: /*
Chris@42:  * This function contains 143 FP additions, 128 FP multiplications,
Chris@42:  * (or, 77 additions, 62 multiplications, 66 fused multiply/add),
Chris@42:  * 130 stack variables, 5 constants, and 40 memory accesses
Chris@42:  */
Chris@42: #include "hc2cfv.h"
Chris@42: 
Chris@42: static void hc2cfdftv_20(R *Rp, R *Ip, R *Rm, R *Im, const R *W, stride rs, INT mb, INT me, INT ms)
Chris@42: {
Chris@42:      DVK(KP559016994, +0.559016994374947424102293417182819058860154590);
Chris@42:      DVK(KP500000000, +0.500000000000000000000000000000000000000000000);
Chris@42:      DVK(KP951056516, +0.951056516295153572116439333379382143405698634);
Chris@42:      DVK(KP250000000, +0.250000000000000000000000000000000000000000000);
Chris@42:      DVK(KP618033988, +0.618033988749894848204586834365638117720309180);
Chris@42:      {
Chris@42: 	  INT m;
Chris@42: 	  for (m = mb, W = W + ((mb - 1) * ((TWVL / VL) * 38)); m < me; m = m + VL, Rp = Rp + (VL * ms), Ip = Ip + (VL * ms), Rm = Rm - (VL * ms), Im = Im - (VL * ms), W = W + (TWVL * 38), MAKE_VOLATILE_STRIDE(80, rs)) {
Chris@42: 	       V T2g, T2f, T2w, T2k, T2A, T2u, T2e, T2o, T1O, T2b, T2i, T1R, T1X, T1k, TN;
Chris@42: 	       V T1w, T1G, T1t, Ti, T2c, T12, T1x, T2j, T1U, T1y, T1d, T24, T2v, T2h, T2x;
Chris@42: 	       V T2B, T2p, T2l, T2z, T2y, T2D, T2C, T2r, T2q, T2n, T2m;
Chris@42: 	       {
Chris@42: 		    V T3, T7, TC, T1Y, Tc, Tg, Tn, T1P, T1Z, Tw, T1S, TS, TY, TZ, T1Q;
Chris@42: 		    V TL, T17, T21, TW, T19, TX, T1a, T8, T20, Th, Tx, T1u, T1v, TM, T10;
Chris@42: 		    V T1b, T22, T11, T1T, T1c, T23;
Chris@42: 		    {
Chris@42: 			 V Ta, Tb, Tz, Te, TB, Tf, Tl, T9, Td, Tk, T1, T2, Ty, T5, T6;
Chris@42: 			 V TA, T4, Tj, Tt, Tu, Ts, TQ, Tr, TP, Tp, Tq, Tm, To, TO, TG;
Chris@42: 			 V T14, TK, T16, TE, TF, Tv, TD, T13, TR, TI, TJ, TH, T15, TU, TV;
Chris@42: 			 V TT, T18;
Chris@42: 			 T1 = LD(&(Rp[0]), ms, &(Rp[0]));
Chris@42: 			 T2 = LD(&(Rm[0]), -ms, &(Rm[0]));
Chris@42: 			 Ty = LDW(&(W[0]));
Chris@42: 			 T5 = LD(&(Rp[WS(rs, 5)]), ms, &(Rp[WS(rs, 1)]));
Chris@42: 			 T6 = LD(&(Rm[WS(rs, 5)]), -ms, &(Rm[WS(rs, 1)]));
Chris@42: 			 TA = LDW(&(W[TWVL * 20]));
Chris@42: 			 T4 = LDW(&(W[TWVL * 18]));
Chris@42: 			 Ta = LD(&(Rp[WS(rs, 2)]), ms, &(Rp[0]));
Chris@42: 			 Tb = LD(&(Rm[WS(rs, 2)]), -ms, &(Rm[0]));
Chris@42: 			 T3 = VFMACONJ(T2, T1);
Chris@42: 			 Tz = VZMULIJ(Ty, VFNMSCONJ(T2, T1));
Chris@42: 			 Tj = LDW(&(W[TWVL * 6]));
Chris@42: 			 Te = LD(&(Rp[WS(rs, 7)]), ms, &(Rp[WS(rs, 1)]));
Chris@42: 			 TB = VZMULIJ(TA, VFNMSCONJ(T6, T5));
Chris@42: 			 T7 = VZMULJ(T4, VFMACONJ(T6, T5));
Chris@42: 			 Tf = LD(&(Rm[WS(rs, 7)]), -ms, &(Rm[WS(rs, 1)]));
Chris@42: 			 Tl = LDW(&(W[TWVL * 26]));
Chris@42: 			 T9 = LDW(&(W[TWVL * 8]));
Chris@42: 			 Td = LDW(&(W[TWVL * 28]));
Chris@42: 			 Tk = VZMULJ(Tj, VFMACONJ(Tb, Ta));
Chris@42: 			 Tp = LD(&(Rp[WS(rs, 4)]), ms, &(Rp[0]));
Chris@42: 			 TC = VADD(Tz, TB);
Chris@42: 			 T1Y = VSUB(TB, Tz);
Chris@42: 			 Tq = LD(&(Rm[WS(rs, 4)]), -ms, &(Rm[0]));
Chris@42: 			 Tm = VZMULJ(Tl, VFMACONJ(Tf, Te));
Chris@42: 			 Tc = VZMULIJ(T9, VFNMSCONJ(Tb, Ta));
Chris@42: 			 Tg = VZMULIJ(Td, VFNMSCONJ(Tf, Te));
Chris@42: 			 To = LDW(&(W[TWVL * 16]));
Chris@42: 			 TO = LDW(&(W[TWVL * 14]));
Chris@42: 			 Tt = LD(&(Rp[WS(rs, 9)]), ms, &(Rp[WS(rs, 1)]));
Chris@42: 			 Tu = LD(&(Rm[WS(rs, 9)]), -ms, &(Rm[WS(rs, 1)]));
Chris@42: 			 Ts = LDW(&(W[TWVL * 36]));
Chris@42: 			 Tn = VADD(Tk, Tm);
Chris@42: 			 T1P = VSUB(Tk, Tm);
Chris@42: 			 TQ = LDW(&(W[TWVL * 34]));
Chris@42: 			 Tr = VZMULIJ(To, VFNMSCONJ(Tq, Tp));
Chris@42: 			 TP = VZMULJ(TO, VFMACONJ(Tq, Tp));
Chris@42: 			 TE = LD(&(Rp[WS(rs, 8)]), ms, &(Rp[0]));
Chris@42: 			 TF = LD(&(Rm[WS(rs, 8)]), -ms, &(Rm[0]));
Chris@42: 			 Tv = VZMULIJ(Ts, VFNMSCONJ(Tu, Tt));
Chris@42: 			 TD = LDW(&(W[TWVL * 30]));
Chris@42: 			 T13 = LDW(&(W[TWVL * 32]));
Chris@42: 			 TR = VZMULJ(TQ, VFMACONJ(Tu, Tt));
Chris@42: 			 TI = LD(&(Rp[WS(rs, 3)]), ms, &(Rp[WS(rs, 1)]));
Chris@42: 			 TJ = LD(&(Rm[WS(rs, 3)]), -ms, &(Rm[WS(rs, 1)]));
Chris@42: 			 TH = LDW(&(W[TWVL * 10]));
Chris@42: 			 T15 = LDW(&(W[TWVL * 12]));
Chris@42: 			 T1Z = VSUB(Tv, Tr);
Chris@42: 			 Tw = VADD(Tr, Tv);
Chris@42: 			 TG = VZMULJ(TD, VFMACONJ(TF, TE));
Chris@42: 			 T14 = VZMULIJ(T13, VFNMSCONJ(TF, TE));
Chris@42: 			 T1S = VSUB(TP, TR);
Chris@42: 			 TS = VADD(TP, TR);
Chris@42: 			 TK = VZMULJ(TH, VFMACONJ(TJ, TI));
Chris@42: 			 T16 = VZMULIJ(T15, VFNMSCONJ(TJ, TI));
Chris@42: 			 TU = LD(&(Rp[WS(rs, 6)]), ms, &(Rp[0]));
Chris@42: 			 TV = LD(&(Rm[WS(rs, 6)]), -ms, &(Rm[0]));
Chris@42: 			 TT = LDW(&(W[TWVL * 24]));
Chris@42: 			 T18 = LDW(&(W[TWVL * 22]));
Chris@42: 			 TY = LD(&(Rp[WS(rs, 1)]), ms, &(Rp[WS(rs, 1)]));
Chris@42: 			 TZ = LD(&(Rm[WS(rs, 1)]), -ms, &(Rm[WS(rs, 1)]));
Chris@42: 			 T1Q = VSUB(TK, TG);
Chris@42: 			 TL = VADD(TG, TK);
Chris@42: 			 T17 = VADD(T14, T16);
Chris@42: 			 T21 = VSUB(T16, T14);
Chris@42: 			 TW = VZMULIJ(TT, VFNMSCONJ(TV, TU));
Chris@42: 			 T19 = VZMULJ(T18, VFMACONJ(TV, TU));
Chris@42: 			 TX = LDW(&(W[TWVL * 4]));
Chris@42: 			 T1a = LDW(&(W[TWVL * 2]));
Chris@42: 		    }
Chris@42: 		    T1O = VSUB(T3, T7);
Chris@42: 		    T8 = VADD(T3, T7);
Chris@42: 		    T20 = VADD(T1Y, T1Z);
Chris@42: 		    T2b = VSUB(T1Y, T1Z);
Chris@42: 		    T2i = VADD(T1P, T1Q);
Chris@42: 		    T1R = VSUB(T1P, T1Q);
Chris@42: 		    Th = VADD(Tc, Tg);
Chris@42: 		    T1X = VSUB(Tg, Tc);
Chris@42: 		    Tx = VSUB(Tn, Tw);
Chris@42: 		    T1u = VADD(Tn, Tw);
Chris@42: 		    T1v = VADD(TC, TL);
Chris@42: 		    TM = VSUB(TC, TL);
Chris@42: 		    T10 = VZMULIJ(TX, VFNMSCONJ(TZ, TY));
Chris@42: 		    T1b = VZMULJ(T1a, VFMACONJ(TZ, TY));
Chris@42: 		    T1k = VADD(Tx, TM);
Chris@42: 		    TN = VSUB(Tx, TM);
Chris@42: 		    T22 = VSUB(T10, TW);
Chris@42: 		    T11 = VADD(TW, T10);
Chris@42: 		    T1T = VSUB(T1b, T19);
Chris@42: 		    T1c = VADD(T19, T1b);
Chris@42: 		    T1w = VADD(T1u, T1v);
Chris@42: 		    T1G = VSUB(T1u, T1v);
Chris@42: 		    T1t = VADD(T8, Th);
Chris@42: 		    Ti = VSUB(T8, Th);
Chris@42: 		    T23 = VADD(T21, T22);
Chris@42: 		    T2c = VSUB(T21, T22);
Chris@42: 		    T12 = VSUB(TS, T11);
Chris@42: 		    T1x = VADD(TS, T11);
Chris@42: 		    T2j = VADD(T1S, T1T);
Chris@42: 		    T1U = VSUB(T1S, T1T);
Chris@42: 		    T1y = VADD(T17, T1c);
Chris@42: 		    T1d = VSUB(T17, T1c);
Chris@42: 		    T2g = VSUB(T23, T20);
Chris@42: 		    T24 = VADD(T20, T23);
Chris@42: 	       }
Chris@42: 	       {
Chris@42: 		    V T2d, T2t, T29, T25, T1m, T1q, T1i, T1H, T1L, T1D, T1A, T28, T1W, T1h, T1g;
Chris@42: 		    V T1e, T1l, T1z, T1F, T1V, T1f, T1C, T1B, T26, T27, T2a, T2s, T1j, T1p, T1K;
Chris@42: 		    V T1E, T1n, T1o, T1s, T1r, T1I, T1J, T1N, T1M;
Chris@42: 		    T2d = VFMA(LDK(KP618033988), T2c, T2b);
Chris@42: 		    T2t = VFNMS(LDK(KP618033988), T2b, T2c);
Chris@42: 		    T1e = VSUB(T12, T1d);
Chris@42: 		    T1l = VADD(T12, T1d);
Chris@42: 		    T1z = VADD(T1x, T1y);
Chris@42: 		    T1F = VSUB(T1x, T1y);
Chris@42: 		    T1V = VADD(T1R, T1U);
Chris@42: 		    T29 = VSUB(T1R, T1U);
Chris@42: 		    T2f = VFNMS(LDK(KP250000000), T24, T1X);
Chris@42: 		    T25 = VADD(T1X, T24);
Chris@42: 		    T1m = VMUL(LDK(KP951056516), VFMA(LDK(KP618033988), T1l, T1k));
Chris@42: 		    T1q = VMUL(LDK(KP951056516), VFNMS(LDK(KP618033988), T1k, T1l));
Chris@42: 		    T1i = VSUB(TN, T1e);
Chris@42: 		    T1f = VADD(TN, T1e);
Chris@42: 		    T1H = VMUL(LDK(KP951056516), VFNMS(LDK(KP618033988), T1G, T1F));
Chris@42: 		    T1L = VMUL(LDK(KP951056516), VFMA(LDK(KP618033988), T1F, T1G));
Chris@42: 		    T1D = VSUB(T1w, T1z);
Chris@42: 		    T1A = VADD(T1w, T1z);
Chris@42: 		    T28 = VFNMS(LDK(KP250000000), T1V, T1O);
Chris@42: 		    T1W = VADD(T1O, T1V);
Chris@42: 		    T1h = VFNMS(LDK(KP250000000), T1f, Ti);
Chris@42: 		    T1g = VMUL(LDK(KP500000000), VADD(Ti, T1f));
Chris@42: 		    T2w = VFNMS(LDK(KP618033988), T2i, T2j);
Chris@42: 		    T2k = VFMA(LDK(KP618033988), T2j, T2i);
Chris@42: 		    T1C = VFNMS(LDK(KP250000000), T1A, T1t);
Chris@42: 		    T1B = VCONJ(VMUL(LDK(KP500000000), VADD(T1t, T1A)));
Chris@42: 		    T26 = VMUL(LDK(KP500000000), VFNMSI(T25, T1W));
Chris@42: 		    T27 = VCONJ(VMUL(LDK(KP500000000), VFMAI(T25, T1W)));
Chris@42: 		    T2a = VFMA(LDK(KP559016994), T29, T28);
Chris@42: 		    T2s = VFNMS(LDK(KP559016994), T29, T28);
Chris@42: 		    ST(&(Rp[0]), T1g, ms, &(Rp[0]));
Chris@42: 		    T1j = VFMA(LDK(KP559016994), T1i, T1h);
Chris@42: 		    T1p = VFNMS(LDK(KP559016994), T1i, T1h);
Chris@42: 		    ST(&(Rm[WS(rs, 9)]), T1B, -ms, &(Rm[WS(rs, 1)]));
Chris@42: 		    T1K = VFMA(LDK(KP559016994), T1D, T1C);
Chris@42: 		    T1E = VFNMS(LDK(KP559016994), T1D, T1C);
Chris@42: 		    ST(&(Rm[WS(rs, 4)]), T27, -ms, &(Rm[0]));
Chris@42: 		    ST(&(Rp[WS(rs, 5)]), T26, ms, &(Rp[WS(rs, 1)]));
Chris@42: 		    T2A = VFMA(LDK(KP951056516), T2t, T2s);
Chris@42: 		    T2u = VFNMS(LDK(KP951056516), T2t, T2s);
Chris@42: 		    T2e = VFNMS(LDK(KP951056516), T2d, T2a);
Chris@42: 		    T2o = VFMA(LDK(KP951056516), T2d, T2a);
Chris@42: 		    T1n = VCONJ(VMUL(LDK(KP500000000), VFNMSI(T1m, T1j)));
Chris@42: 		    T1o = VMUL(LDK(KP500000000), VFMAI(T1m, T1j));
Chris@42: 		    T1s = VCONJ(VMUL(LDK(KP500000000), VFMAI(T1q, T1p)));
Chris@42: 		    T1r = VMUL(LDK(KP500000000), VFNMSI(T1q, T1p));
Chris@42: 		    T1I = VCONJ(VMUL(LDK(KP500000000), VFNMSI(T1H, T1E)));
Chris@42: 		    T1J = VMUL(LDK(KP500000000), VFMAI(T1H, T1E));
Chris@42: 		    T1N = VCONJ(VMUL(LDK(KP500000000), VFMAI(T1L, T1K)));
Chris@42: 		    T1M = VMUL(LDK(KP500000000), VFNMSI(T1L, T1K));
Chris@42: 		    ST(&(Rp[WS(rs, 4)]), T1o, ms, &(Rp[0]));
Chris@42: 		    ST(&(Rm[WS(rs, 3)]), T1n, -ms, &(Rm[WS(rs, 1)]));
Chris@42: 		    ST(&(Rp[WS(rs, 8)]), T1r, ms, &(Rp[0]));
Chris@42: 		    ST(&(Rm[WS(rs, 7)]), T1s, -ms, &(Rm[WS(rs, 1)]));
Chris@42: 		    ST(&(Rp[WS(rs, 2)]), T1J, ms, &(Rp[0]));
Chris@42: 		    ST(&(Rm[WS(rs, 1)]), T1I, -ms, &(Rm[WS(rs, 1)]));
Chris@42: 		    ST(&(Rp[WS(rs, 6)]), T1M, ms, &(Rp[0]));
Chris@42: 		    ST(&(Rm[WS(rs, 5)]), T1N, -ms, &(Rm[WS(rs, 1)]));
Chris@42: 	       }
Chris@42: 	       T2v = VFMA(LDK(KP559016994), T2g, T2f);
Chris@42: 	       T2h = VFNMS(LDK(KP559016994), T2g, T2f);
Chris@42: 	       T2x = VFNMS(LDK(KP951056516), T2w, T2v);
Chris@42: 	       T2B = VFMA(LDK(KP951056516), T2w, T2v);
Chris@42: 	       T2p = VFMA(LDK(KP951056516), T2k, T2h);
Chris@42: 	       T2l = VFNMS(LDK(KP951056516), T2k, T2h);
Chris@42: 	       T2z = VMUL(LDK(KP500000000), VFMAI(T2x, T2u));
Chris@42: 	       T2y = VCONJ(VMUL(LDK(KP500000000), VFNMSI(T2x, T2u)));
Chris@42: 	       T2D = VMUL(LDK(KP500000000), VFMAI(T2B, T2A));
Chris@42: 	       T2C = VCONJ(VMUL(LDK(KP500000000), VFNMSI(T2B, T2A)));
Chris@42: 	       T2r = VCONJ(VMUL(LDK(KP500000000), VFMAI(T2p, T2o)));
Chris@42: 	       T2q = VMUL(LDK(KP500000000), VFNMSI(T2p, T2o));
Chris@42: 	       T2n = VCONJ(VMUL(LDK(KP500000000), VFMAI(T2l, T2e)));
Chris@42: 	       T2m = VMUL(LDK(KP500000000), VFNMSI(T2l, T2e));
Chris@42: 	       ST(&(Rp[WS(rs, 3)]), T2z, ms, &(Rp[WS(rs, 1)]));
Chris@42: 	       ST(&(Rm[WS(rs, 2)]), T2y, -ms, &(Rm[0]));
Chris@42: 	       ST(&(Rp[WS(rs, 7)]), T2D, ms, &(Rp[WS(rs, 1)]));
Chris@42: 	       ST(&(Rm[WS(rs, 6)]), T2C, -ms, &(Rm[0]));
Chris@42: 	       ST(&(Rm[0]), T2r, -ms, &(Rm[0]));
Chris@42: 	       ST(&(Rp[WS(rs, 1)]), T2q, ms, &(Rp[WS(rs, 1)]));
Chris@42: 	       ST(&(Rm[WS(rs, 8)]), T2n, -ms, &(Rm[0]));
Chris@42: 	       ST(&(Rp[WS(rs, 9)]), T2m, ms, &(Rp[WS(rs, 1)]));
Chris@42: 	  }
Chris@42:      }
Chris@42:      VLEAVE();
Chris@42: }
Chris@42: 
Chris@42: static const tw_instr twinstr[] = {
Chris@42:      VTW(1, 1),
Chris@42:      VTW(1, 2),
Chris@42:      VTW(1, 3),
Chris@42:      VTW(1, 4),
Chris@42:      VTW(1, 5),
Chris@42:      VTW(1, 6),
Chris@42:      VTW(1, 7),
Chris@42:      VTW(1, 8),
Chris@42:      VTW(1, 9),
Chris@42:      VTW(1, 10),
Chris@42:      VTW(1, 11),
Chris@42:      VTW(1, 12),
Chris@42:      VTW(1, 13),
Chris@42:      VTW(1, 14),
Chris@42:      VTW(1, 15),
Chris@42:      VTW(1, 16),
Chris@42:      VTW(1, 17),
Chris@42:      VTW(1, 18),
Chris@42:      VTW(1, 19),
Chris@42:      {TW_NEXT, VL, 0}
Chris@42: };
Chris@42: 
Chris@42: static const hc2c_desc desc = { 20, XSIMD_STRING("hc2cfdftv_20"), twinstr, &GENUS, {77, 62, 66, 0} };
Chris@42: 
Chris@42: void XSIMD(codelet_hc2cfdftv_20) (planner *p) {
Chris@42:      X(khc2c_register) (p, hc2cfdftv_20, &desc, HC2C_VIA_DFT);
Chris@42: }
Chris@42: #else				/* HAVE_FMA */
Chris@42: 
Chris@42: /* Generated by: ../../../genfft/gen_hc2cdft_c.native -simd -compact -variables 4 -pipeline-latency 8 -trivial-stores -variables 32 -no-generate-bytw -n 20 -dit -name hc2cfdftv_20 -include hc2cfv.h */
Chris@42: 
Chris@42: /*
Chris@42:  * This function contains 143 FP additions, 77 FP multiplications,
Chris@42:  * (or, 131 additions, 65 multiplications, 12 fused multiply/add),
Chris@42:  * 141 stack variables, 9 constants, and 40 memory accesses
Chris@42:  */
Chris@42: #include "hc2cfv.h"
Chris@42: 
Chris@42: static void hc2cfdftv_20(R *Rp, R *Ip, R *Rm, R *Im, const R *W, stride rs, INT mb, INT me, INT ms)
Chris@42: {
Chris@42:      DVK(KP293892626, +0.293892626146236564584352977319536384298826219);
Chris@42:      DVK(KP475528258, +0.475528258147576786058219666689691071702849317);
Chris@42:      DVK(KP559016994, +0.559016994374947424102293417182819058860154590);
Chris@42:      DVK(KP250000000, +0.250000000000000000000000000000000000000000000);
Chris@42:      DVK(KP125000000, +0.125000000000000000000000000000000000000000000);
Chris@42:      DVK(KP279508497, +0.279508497187473712051146708591409529430077295);
Chris@42:      DVK(KP587785252, +0.587785252292473129168705954639072768597652438);
Chris@42:      DVK(KP951056516, +0.951056516295153572116439333379382143405698634);
Chris@42:      DVK(KP500000000, +0.500000000000000000000000000000000000000000000);
Chris@42:      {
Chris@42: 	  INT m;
Chris@42: 	  for (m = mb, W = W + ((mb - 1) * ((TWVL / VL) * 38)); m < me; m = m + VL, Rp = Rp + (VL * ms), Ip = Ip + (VL * ms), Rm = Rm - (VL * ms), Im = Im - (VL * ms), W = W + (TWVL * 38), MAKE_VOLATILE_STRIDE(80, rs)) {
Chris@42: 	       V TW, T1x, T2i, T2A, T1r, T1s, T1a, T1y, T1l, Tn, TK, TL, T1p, T1o, T27;
Chris@42: 	       V T2t, T2a, T2u, T2e, T2C, T20, T2w, T23, T2x, T2d, T2B, T1W, T1X, T1U, T1V;
Chris@42: 	       V T2z, T2K, T2G, T2N, T2J, T2v, T2y, T2F, T2D, T2E, T2M, T2H, T2I, T2L;
Chris@42: 	       {
Chris@42: 		    V T1u, T5, Tg, T1c, TV, T13, Ta, T1w, TQ, T11, TI, T1j, Tx, T18, Tl;
Chris@42: 		    V T1e, TD, T1h, Ts, T16, T2g, T2h, T14, T19, T1f, T1k, Tb, Tm, Ty, TJ;
Chris@42: 		    V T25, T26, T28, T29, T1Y, T1Z, T21, T22;
Chris@42: 		    {
Chris@42: 			 V T4, T3, T2, T1, Tf, Te, Td, Tc, T1b, TU, TT, TS, TR, T12, T9;
Chris@42: 			 V T8, T7, T6, T1v, TP, TO, TN, TM, T10, TH, TG, TF, TE, T1i, Tw;
Chris@42: 			 V Tv, Tu, Tt, T17, Tk, Tj, Ti, Th, T1d, TC, TB, TA, Tz, T1g, Tr;
Chris@42: 			 V Tq, Tp, To, T15;
Chris@42: 			 T4 = LD(&(Rp[0]), ms, &(Rp[0]));
Chris@42: 			 T2 = LD(&(Rm[0]), -ms, &(Rm[0]));
Chris@42: 			 T3 = VCONJ(T2);
Chris@42: 			 T1u = VADD(T4, T3);
Chris@42: 			 T1 = LDW(&(W[0]));
Chris@42: 			 T5 = VZMULIJ(T1, VSUB(T3, T4));
Chris@42: 			 Tf = LD(&(Rp[WS(rs, 4)]), ms, &(Rp[0]));
Chris@42: 			 Td = LD(&(Rm[WS(rs, 4)]), -ms, &(Rm[0]));
Chris@42: 			 Te = VCONJ(Td);
Chris@42: 			 Tc = LDW(&(W[TWVL * 16]));
Chris@42: 			 Tg = VZMULIJ(Tc, VSUB(Te, Tf));
Chris@42: 			 T1b = LDW(&(W[TWVL * 14]));
Chris@42: 			 T1c = VZMULJ(T1b, VADD(Te, Tf));
Chris@42: 			 TU = LD(&(Rp[WS(rs, 7)]), ms, &(Rp[WS(rs, 1)]));
Chris@42: 			 TS = LD(&(Rm[WS(rs, 7)]), -ms, &(Rm[WS(rs, 1)]));
Chris@42: 			 TT = VCONJ(TS);
Chris@42: 			 TR = LDW(&(W[TWVL * 28]));
Chris@42: 			 TV = VZMULIJ(TR, VSUB(TT, TU));
Chris@42: 			 T12 = LDW(&(W[TWVL * 26]));
Chris@42: 			 T13 = VZMULJ(T12, VADD(TT, TU));
Chris@42: 			 T9 = LD(&(Rp[WS(rs, 5)]), ms, &(Rp[WS(rs, 1)]));
Chris@42: 			 T7 = LD(&(Rm[WS(rs, 5)]), -ms, &(Rm[WS(rs, 1)]));
Chris@42: 			 T8 = VCONJ(T7);
Chris@42: 			 T6 = LDW(&(W[TWVL * 20]));
Chris@42: 			 Ta = VZMULIJ(T6, VSUB(T8, T9));
Chris@42: 			 T1v = LDW(&(W[TWVL * 18]));
Chris@42: 			 T1w = VZMULJ(T1v, VADD(T9, T8));
Chris@42: 			 TP = LD(&(Rp[WS(rs, 2)]), ms, &(Rp[0]));
Chris@42: 			 TN = LD(&(Rm[WS(rs, 2)]), -ms, &(Rm[0]));
Chris@42: 			 TO = VCONJ(TN);
Chris@42: 			 TM = LDW(&(W[TWVL * 8]));
Chris@42: 			 TQ = VZMULIJ(TM, VSUB(TO, TP));
Chris@42: 			 T10 = LDW(&(W[TWVL * 6]));
Chris@42: 			 T11 = VZMULJ(T10, VADD(TO, TP));
Chris@42: 			 TH = LD(&(Rp[WS(rs, 1)]), ms, &(Rp[WS(rs, 1)]));
Chris@42: 			 TF = LD(&(Rm[WS(rs, 1)]), -ms, &(Rm[WS(rs, 1)]));
Chris@42: 			 TG = VCONJ(TF);
Chris@42: 			 TE = LDW(&(W[TWVL * 4]));
Chris@42: 			 TI = VZMULIJ(TE, VSUB(TG, TH));
Chris@42: 			 T1i = LDW(&(W[TWVL * 2]));
Chris@42: 			 T1j = VZMULJ(T1i, VADD(TG, TH));
Chris@42: 			 Tw = LD(&(Rp[WS(rs, 3)]), ms, &(Rp[WS(rs, 1)]));
Chris@42: 			 Tu = LD(&(Rm[WS(rs, 3)]), -ms, &(Rm[WS(rs, 1)]));
Chris@42: 			 Tv = VCONJ(Tu);
Chris@42: 			 Tt = LDW(&(W[TWVL * 12]));
Chris@42: 			 Tx = VZMULIJ(Tt, VSUB(Tv, Tw));
Chris@42: 			 T17 = LDW(&(W[TWVL * 10]));
Chris@42: 			 T18 = VZMULJ(T17, VADD(Tw, Tv));
Chris@42: 			 Tk = LD(&(Rp[WS(rs, 9)]), ms, &(Rp[WS(rs, 1)]));
Chris@42: 			 Ti = LD(&(Rm[WS(rs, 9)]), -ms, &(Rm[WS(rs, 1)]));
Chris@42: 			 Tj = VCONJ(Ti);
Chris@42: 			 Th = LDW(&(W[TWVL * 36]));
Chris@42: 			 Tl = VZMULIJ(Th, VSUB(Tj, Tk));
Chris@42: 			 T1d = LDW(&(W[TWVL * 34]));
Chris@42: 			 T1e = VZMULJ(T1d, VADD(Tj, Tk));
Chris@42: 			 TC = LD(&(Rp[WS(rs, 6)]), ms, &(Rp[0]));
Chris@42: 			 TA = LD(&(Rm[WS(rs, 6)]), -ms, &(Rm[0]));
Chris@42: 			 TB = VCONJ(TA);
Chris@42: 			 Tz = LDW(&(W[TWVL * 24]));
Chris@42: 			 TD = VZMULIJ(Tz, VSUB(TB, TC));
Chris@42: 			 T1g = LDW(&(W[TWVL * 22]));
Chris@42: 			 T1h = VZMULJ(T1g, VADD(TB, TC));
Chris@42: 			 Tr = LD(&(Rp[WS(rs, 8)]), ms, &(Rp[0]));
Chris@42: 			 Tp = LD(&(Rm[WS(rs, 8)]), -ms, &(Rm[0]));
Chris@42: 			 Tq = VCONJ(Tp);
Chris@42: 			 To = LDW(&(W[TWVL * 32]));
Chris@42: 			 Ts = VZMULIJ(To, VSUB(Tq, Tr));
Chris@42: 			 T15 = LDW(&(W[TWVL * 30]));
Chris@42: 			 T16 = VZMULJ(T15, VADD(Tr, Tq));
Chris@42: 		    }
Chris@42: 		    TW = VSUB(TQ, TV);
Chris@42: 		    T1x = VSUB(T1u, T1w);
Chris@42: 		    T2g = VADD(T1u, T1w);
Chris@42: 		    T2h = VADD(TQ, TV);
Chris@42: 		    T2i = VADD(T2g, T2h);
Chris@42: 		    T2A = VSUB(T2g, T2h);
Chris@42: 		    T14 = VSUB(T11, T13);
Chris@42: 		    T19 = VSUB(T16, T18);
Chris@42: 		    T1r = VADD(T14, T19);
Chris@42: 		    T1f = VSUB(T1c, T1e);
Chris@42: 		    T1k = VSUB(T1h, T1j);
Chris@42: 		    T1s = VADD(T1f, T1k);
Chris@42: 		    T1a = VSUB(T14, T19);
Chris@42: 		    T1y = VADD(T1r, T1s);
Chris@42: 		    T1l = VSUB(T1f, T1k);
Chris@42: 		    Tb = VSUB(T5, Ta);
Chris@42: 		    Tm = VSUB(Tg, Tl);
Chris@42: 		    Tn = VADD(Tb, Tm);
Chris@42: 		    Ty = VSUB(Ts, Tx);
Chris@42: 		    TJ = VSUB(TD, TI);
Chris@42: 		    TK = VADD(Ty, TJ);
Chris@42: 		    TL = VADD(Tn, TK);
Chris@42: 		    T1p = VSUB(Ty, TJ);
Chris@42: 		    T1o = VSUB(Tb, Tm);
Chris@42: 		    T25 = VADD(T1c, T1e);
Chris@42: 		    T26 = VADD(TD, TI);
Chris@42: 		    T27 = VADD(T25, T26);
Chris@42: 		    T2t = VSUB(T25, T26);
Chris@42: 		    T28 = VADD(Ts, Tx);
Chris@42: 		    T29 = VADD(T1h, T1j);
Chris@42: 		    T2a = VADD(T28, T29);
Chris@42: 		    T2u = VSUB(T29, T28);
Chris@42: 		    T2e = VADD(T27, T2a);
Chris@42: 		    T2C = VADD(T2t, T2u);
Chris@42: 		    T1Y = VADD(T11, T13);
Chris@42: 		    T1Z = VADD(Tg, Tl);
Chris@42: 		    T20 = VADD(T1Y, T1Z);
Chris@42: 		    T2w = VSUB(T1Y, T1Z);
Chris@42: 		    T21 = VADD(T5, Ta);
Chris@42: 		    T22 = VADD(T16, T18);
Chris@42: 		    T23 = VADD(T21, T22);
Chris@42: 		    T2x = VSUB(T22, T21);
Chris@42: 		    T2d = VADD(T20, T23);
Chris@42: 		    T2B = VADD(T2w, T2x);
Chris@42: 	       }
Chris@42: 	       T1U = VADD(T1x, T1y);
Chris@42: 	       T1V = VBYI(VADD(TW, TL));
Chris@42: 	       T1W = VMUL(LDK(KP500000000), VSUB(T1U, T1V));
Chris@42: 	       T1X = VCONJ(VMUL(LDK(KP500000000), VADD(T1V, T1U)));
Chris@42: 	       ST(&(Rp[WS(rs, 5)]), T1W, ms, &(Rp[WS(rs, 1)]));
Chris@42: 	       ST(&(Rm[WS(rs, 4)]), T1X, -ms, &(Rm[0]));
Chris@42: 	       T2v = VSUB(T2t, T2u);
Chris@42: 	       T2y = VSUB(T2w, T2x);
Chris@42: 	       T2z = VMUL(LDK(KP500000000), VBYI(VFNMS(LDK(KP587785252), T2y, VMUL(LDK(KP951056516), T2v))));
Chris@42: 	       T2K = VMUL(LDK(KP500000000), VBYI(VFMA(LDK(KP951056516), T2y, VMUL(LDK(KP587785252), T2v))));
Chris@42: 	       T2F = VMUL(LDK(KP279508497), VSUB(T2B, T2C));
Chris@42: 	       T2D = VADD(T2B, T2C);
Chris@42: 	       T2E = VFNMS(LDK(KP125000000), T2D, VMUL(LDK(KP500000000), T2A));
Chris@42: 	       T2G = VSUB(T2E, T2F);
Chris@42: 	       T2N = VCONJ(VMUL(LDK(KP500000000), VADD(T2A, T2D)));
Chris@42: 	       T2J = VADD(T2F, T2E);
Chris@42: 	       ST(&(Rm[WS(rs, 9)]), T2N, -ms, &(Rm[WS(rs, 1)]));
Chris@42: 	       T2M = VCONJ(VADD(T2K, T2J));
Chris@42: 	       ST(&(Rm[WS(rs, 5)]), T2M, -ms, &(Rm[WS(rs, 1)]));
Chris@42: 	       T2H = VADD(T2z, T2G);
Chris@42: 	       ST(&(Rp[WS(rs, 2)]), T2H, ms, &(Rp[0]));
Chris@42: 	       T2I = VCONJ(VSUB(T2G, T2z));
Chris@42: 	       ST(&(Rm[WS(rs, 1)]), T2I, -ms, &(Rm[WS(rs, 1)]));
Chris@42: 	       T2L = VSUB(T2J, T2K);
Chris@42: 	       ST(&(Rp[WS(rs, 6)]), T2L, ms, &(Rp[0]));
Chris@42: 	       {
Chris@42: 		    V T2c, T2p, T2l, T2s, T2o, T24, T2b, T2f, T2j, T2k, T2r, T2m, T2n, T2q, T1n;
Chris@42: 		    V T1Q, T1E, T1K, T1B, T1R, T1F, T1N, T1m, T1J, TZ, T1I, TX, TY, T1q, T1M;
Chris@42: 		    V T1A, T1L, T1t, T1z, T1C, T1S, T1T, T1D, T1G, T1O, T1P, T1H;
Chris@42: 		    T24 = VSUB(T20, T23);
Chris@42: 		    T2b = VSUB(T27, T2a);
Chris@42: 		    T2c = VMUL(LDK(KP500000000), VBYI(VFMA(LDK(KP951056516), T24, VMUL(LDK(KP587785252), T2b))));
Chris@42: 		    T2p = VMUL(LDK(KP500000000), VBYI(VFNMS(LDK(KP587785252), T24, VMUL(LDK(KP951056516), T2b))));
Chris@42: 		    T2f = VMUL(LDK(KP279508497), VSUB(T2d, T2e));
Chris@42: 		    T2j = VADD(T2d, T2e);
Chris@42: 		    T2k = VFNMS(LDK(KP125000000), T2j, VMUL(LDK(KP500000000), T2i));
Chris@42: 		    T2l = VADD(T2f, T2k);
Chris@42: 		    T2s = VMUL(LDK(KP500000000), VADD(T2i, T2j));
Chris@42: 		    T2o = VSUB(T2k, T2f);
Chris@42: 		    ST(&(Rp[0]), T2s, ms, &(Rp[0]));
Chris@42: 		    T2r = VCONJ(VADD(T2p, T2o));
Chris@42: 		    ST(&(Rm[WS(rs, 7)]), T2r, -ms, &(Rm[WS(rs, 1)]));
Chris@42: 		    T2m = VADD(T2c, T2l);
Chris@42: 		    ST(&(Rp[WS(rs, 4)]), T2m, ms, &(Rp[0]));
Chris@42: 		    T2n = VCONJ(VSUB(T2l, T2c));
Chris@42: 		    ST(&(Rm[WS(rs, 3)]), T2n, -ms, &(Rm[WS(rs, 1)]));
Chris@42: 		    T2q = VSUB(T2o, T2p);
Chris@42: 		    ST(&(Rp[WS(rs, 8)]), T2q, ms, &(Rp[0]));
Chris@42: 		    T1m = VFMA(LDK(KP951056516), T1a, VMUL(LDK(KP587785252), T1l));
Chris@42: 		    T1J = VFNMS(LDK(KP587785252), T1a, VMUL(LDK(KP951056516), T1l));
Chris@42: 		    TX = VFMS(LDK(KP250000000), TL, TW);
Chris@42: 		    TY = VMUL(LDK(KP559016994), VSUB(TK, Tn));
Chris@42: 		    TZ = VADD(TX, TY);
Chris@42: 		    T1I = VSUB(TY, TX);
Chris@42: 		    T1n = VMUL(LDK(KP500000000), VBYI(VSUB(TZ, T1m)));
Chris@42: 		    T1Q = VMUL(LDK(KP500000000), VBYI(VADD(T1I, T1J)));
Chris@42: 		    T1E = VMUL(LDK(KP500000000), VBYI(VADD(TZ, T1m)));
Chris@42: 		    T1K = VMUL(LDK(KP500000000), VBYI(VSUB(T1I, T1J)));
Chris@42: 		    T1q = VFMA(LDK(KP475528258), T1o, VMUL(LDK(KP293892626), T1p));
Chris@42: 		    T1M = VFNMS(LDK(KP293892626), T1o, VMUL(LDK(KP475528258), T1p));
Chris@42: 		    T1t = VMUL(LDK(KP279508497), VSUB(T1r, T1s));
Chris@42: 		    T1z = VFNMS(LDK(KP125000000), T1y, VMUL(LDK(KP500000000), T1x));
Chris@42: 		    T1A = VADD(T1t, T1z);
Chris@42: 		    T1L = VSUB(T1z, T1t);
Chris@42: 		    T1B = VADD(T1q, T1A);
Chris@42: 		    T1R = VADD(T1M, T1L);
Chris@42: 		    T1F = VSUB(T1A, T1q);
Chris@42: 		    T1N = VSUB(T1L, T1M);
Chris@42: 		    T1C = VADD(T1n, T1B);
Chris@42: 		    ST(&(Rp[WS(rs, 1)]), T1C, ms, &(Rp[WS(rs, 1)]));
Chris@42: 		    T1S = VADD(T1Q, T1R);
Chris@42: 		    ST(&(Rp[WS(rs, 7)]), T1S, ms, &(Rp[WS(rs, 1)]));
Chris@42: 		    T1T = VCONJ(VSUB(T1R, T1Q));
Chris@42: 		    ST(&(Rm[WS(rs, 6)]), T1T, -ms, &(Rm[0]));
Chris@42: 		    T1D = VCONJ(VSUB(T1B, T1n));
Chris@42: 		    ST(&(Rm[0]), T1D, -ms, &(Rm[0]));
Chris@42: 		    T1G = VADD(T1E, T1F);
Chris@42: 		    ST(&(Rp[WS(rs, 9)]), T1G, ms, &(Rp[WS(rs, 1)]));
Chris@42: 		    T1O = VADD(T1K, T1N);
Chris@42: 		    ST(&(Rp[WS(rs, 3)]), T1O, ms, &(Rp[WS(rs, 1)]));
Chris@42: 		    T1P = VCONJ(VSUB(T1N, T1K));
Chris@42: 		    ST(&(Rm[WS(rs, 2)]), T1P, -ms, &(Rm[0]));
Chris@42: 		    T1H = VCONJ(VSUB(T1F, T1E));
Chris@42: 		    ST(&(Rm[WS(rs, 8)]), T1H, -ms, &(Rm[0]));
Chris@42: 	       }
Chris@42: 	  }
Chris@42:      }
Chris@42:      VLEAVE();
Chris@42: }
Chris@42: 
Chris@42: static const tw_instr twinstr[] = {
Chris@42:      VTW(1, 1),
Chris@42:      VTW(1, 2),
Chris@42:      VTW(1, 3),
Chris@42:      VTW(1, 4),
Chris@42:      VTW(1, 5),
Chris@42:      VTW(1, 6),
Chris@42:      VTW(1, 7),
Chris@42:      VTW(1, 8),
Chris@42:      VTW(1, 9),
Chris@42:      VTW(1, 10),
Chris@42:      VTW(1, 11),
Chris@42:      VTW(1, 12),
Chris@42:      VTW(1, 13),
Chris@42:      VTW(1, 14),
Chris@42:      VTW(1, 15),
Chris@42:      VTW(1, 16),
Chris@42:      VTW(1, 17),
Chris@42:      VTW(1, 18),
Chris@42:      VTW(1, 19),
Chris@42:      {TW_NEXT, VL, 0}
Chris@42: };
Chris@42: 
Chris@42: static const hc2c_desc desc = { 20, XSIMD_STRING("hc2cfdftv_20"), twinstr, &GENUS, {131, 65, 12, 0} };
Chris@42: 
Chris@42: void XSIMD(codelet_hc2cfdftv_20) (planner *p) {
Chris@42:      X(khc2c_register) (p, hc2cfdftv_20, &desc, HC2C_VIA_DFT);
Chris@42: }
Chris@42: #endif				/* HAVE_FMA */