view src/fftw-3.3.3/dft/simd/common/n2fv_20.c @ 23:619f715526df sv_v2.1

Update Vamp plugin SDK to 2.5

/*
 * Copyright (c) 2003, 2007-11 Matteo Frigo
 * Copyright (c) 2003, 2007-11 Massachusetts Institute of Technology
 *
 * This program is free software; you can redistribute it and/or modify
 * it under the terms of the GNU General Public License as published by
 * the Free Software Foundation; either version 2 of the License, or
 * (at your option) any later version.
 *
 * This program is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 * GNU General Public License for more details.
 *
 * You should have received a copy of the GNU General Public License
 * along with this program; if not, write to the Free Software
 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02110-1301  USA
 *
 */

/* This file was automatically generated --- DO NOT EDIT */
/* Generated on Sun Nov 25 07:37:28 EST 2012 */

#include "codelet-dft.h"

#ifdef HAVE_FMA

/* Generated by: ../../../genfft/gen_notw_c.native -fma -reorder-insns -schedule-for-pipeline -simd -compact -variables 4 -pipeline-latency 8 -n 20 -name n2fv_20 -with-ostride 2 -include n2f.h -store-multiple 2 */

/*
 * This function contains 104 FP additions, 50 FP multiplications,
 * (or, 58 additions, 4 multiplications, 46 fused multiply/add),
 * 79 stack variables, 4 constants, and 50 memory accesses
 */
#include "n2f.h"

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(40, is), MAKE_VOLATILE_STRIDE(40, 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);
			 }
		    }
	       }
	  }
     }
     VLEAVE();
}

static const kdft_desc desc = { 20, XSIMD_STRING("n2fv_20"), {58, 4, 46, 0}, &GENUS, 0, 2, 0, 0 };

void XSIMD(codelet_n2fv_20) (planner *p) {
     X(kdft_register) (p, n2fv_20, &desc);
}

#else				/* HAVE_FMA */

/* Generated by: ../../../genfft/gen_notw_c.native -simd -compact -variables 4 -pipeline-latency 8 -n 20 -name n2fv_20 -with-ostride 2 -include n2f.h -store-multiple 2 */

/*
 * This function contains 104 FP additions, 24 FP multiplications,
 * (or, 92 additions, 12 multiplications, 12 fused multiply/add),
 * 57 stack variables, 4 constants, and 50 memory accesses
 */
#include "n2f.h"

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(40, is), MAKE_VOLATILE_STRIDE(40, 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);
			      }
			 }
		    }
	       }
	  }
     }
     VLEAVE();
}

static const kdft_desc desc = { 20, XSIMD_STRING("n2fv_20"), {92, 12, 12, 0}, &GENUS, 0, 2, 0, 0 };

void XSIMD(codelet_n2fv_20) (planner *p) {
     X(kdft_register) (p, n2fv_20, &desc);
}

#endif				/* HAVE_FMA */