annotate src/fftw-3.3.3/dft/simd/common/n1bv_16.c @ 138:eb184393b244

Rebuild with DW2 exception handling to match Qt

/*
 * 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 07:37:04 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 -sign 1 -n 16 -name n1bv_16 -include n1b.h */

/*
 * This function contains 72 FP additions, 34 FP multiplications,
 * (or, 38 additions, 0 multiplications, 34 fused multiply/add),
 * 54 stack variables, 3 constants, and 32 memory accesses
 */
#include "n1b.h"

static void n1bv_16(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs)
{
     DVK(KP923879532, +0.923879532511286756128183189396788286822416626);
     DVK(KP414213562, +0.414213562373095048801688724209698078569671875);
     DVK(KP707106781, +0.707106781186547524400844362104849039284835938);
     {
	  INT i;
	  const R *xi;
	  R *xo;
	  xi = ii;
	  xo = io;
	  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)) {
	       V T7, Tu, TF, TB, T13, TL, TO, TX, TC, Te, TP, Th, TQ, Tk, TW;
	       V T16;
	       {
		    V TH, TU, Tz, Tf, TK, TV, TA, TM, Ta, TN, Td, Tg, Ti, Tj;
		    {
			 V T1, T2, T4, T5, To, Tp, Tr, Ts;
			 T1 = LD(&(xi[0]), ivs, &(xi[0]));
			 T2 = LD(&(xi[WS(is, 8)]), ivs, &(xi[0]));
			 T4 = LD(&(xi[WS(is, 4)]), ivs, &(xi[0]));
			 T5 = LD(&(xi[WS(is, 12)]), ivs, &(xi[0]));
			 To = LD(&(xi[WS(is, 2)]), ivs, &(xi[0]));
			 Tp = LD(&(xi[WS(is, 10)]), ivs, &(xi[0]));
			 Tr = LD(&(xi[WS(is, 14)]), ivs, &(xi[0]));
			 Ts = LD(&(xi[WS(is, 6)]), ivs, &(xi[0]));
			 {
			      V T8, TI, Tq, TJ, Tt, T9, Tb, Tc, T3, T6;
			      T8 = LD(&(xi[WS(is, 1)]), ivs, &(xi[WS(is, 1)]));
			      TH = VSUB(T1, T2);
			      T3 = VADD(T1, T2);
			      TU = VSUB(T4, T5);
			      T6 = VADD(T4, T5);
			      TI = VSUB(To, Tp);
			      Tq = VADD(To, Tp);
			      TJ = VSUB(Tr, Ts);
			      Tt = VADD(Tr, Ts);
			      T9 = LD(&(xi[WS(is, 9)]), ivs, &(xi[WS(is, 1)]));
			      Tb = LD(&(xi[WS(is, 5)]), ivs, &(xi[WS(is, 1)]));
			      Tc = LD(&(xi[WS(is, 13)]), ivs, &(xi[WS(is, 1)]));
			      T7 = VSUB(T3, T6);
			      Tz = VADD(T3, T6);
			      Tf = LD(&(xi[WS(is, 15)]), ivs, &(xi[WS(is, 1)]));
			      TK = VADD(TI, TJ);
			      TV = VSUB(TI, TJ);
			      TA = VADD(Tq, Tt);
			      Tu = VSUB(Tq, Tt);
			      TM = VSUB(T8, T9);
			      Ta = VADD(T8, T9);
			      TN = VSUB(Tb, Tc);
			      Td = VADD(Tb, Tc);
			      Tg = LD(&(xi[WS(is, 7)]), ivs, &(xi[WS(is, 1)]));
			      Ti = LD(&(xi[WS(is, 3)]), ivs, &(xi[WS(is, 1)]));
			      Tj = LD(&(xi[WS(is, 11)]), ivs, &(xi[WS(is, 1)]));
			 }
		    }
		    TF = VADD(Tz, TA);
		    TB = VSUB(Tz, TA);
		    T13 = VFNMS(LDK(KP707106781), TK, TH);
		    TL = VFMA(LDK(KP707106781), TK, TH);
		    TO = VFNMS(LDK(KP414213562), TN, TM);
		    TX = VFMA(LDK(KP414213562), TM, TN);
		    TC = VADD(Ta, Td);
		    Te = VSUB(Ta, Td);
		    TP = VSUB(Tf, Tg);
		    Th = VADD(Tf, Tg);
		    TQ = VSUB(Tj, Ti);
		    Tk = VADD(Ti, Tj);
		    TW = VFMA(LDK(KP707106781), TV, TU);
		    T16 = VFNMS(LDK(KP707106781), TV, TU);
	       }
	       {
		    V TY, TR, Tl, TD;
		    TY = VFMA(LDK(KP414213562), TP, TQ);
		    TR = VFNMS(LDK(KP414213562), TQ, TP);
		    Tl = VSUB(Th, Tk);
		    TD = VADD(Th, Tk);
		    {
			 V TS, T17, TZ, T14;
			 TS = VADD(TO, TR);
			 T17 = VSUB(TO, TR);
			 TZ = VSUB(TX, TY);
			 T14 = VADD(TX, TY);
			 {
			      V TE, TG, Tm, Tv;
			      TE = VSUB(TC, TD);
			      TG = VADD(TC, TD);
			      Tm = VADD(Te, Tl);
			      Tv = VSUB(Te, Tl);
			      {
				   V T18, T1a, TT, T11;
				   T18 = VFMA(LDK(KP923879532), T17, T16);
				   T1a = VFNMS(LDK(KP923879532), T17, T16);
				   TT = VFNMS(LDK(KP923879532), TS, TL);
				   T11 = VFMA(LDK(KP923879532), TS, TL);
				   {
					V T15, T19, T10, T12;
					T15 = VFNMS(LDK(KP923879532), T14, T13);
					T19 = VFMA(LDK(KP923879532), T14, T13);
					T10 = VFNMS(LDK(KP923879532), TZ, TW);
					T12 = VFMA(LDK(KP923879532), TZ, TW);
					ST(&(xo[0]), VADD(TF, TG), ovs, &(xo[0]));
					ST(&(xo[WS(os, 8)]), VSUB(TF, TG), ovs, &(xo[0]));
					ST(&(xo[WS(os, 4)]), VFMAI(TE, TB), ovs, &(xo[0]));
					ST(&(xo[WS(os, 12)]), VFNMSI(TE, TB), ovs, &(xo[0]));
					{
					     V Tw, Ty, Tn, Tx;
					     Tw = VFNMS(LDK(KP707106781), Tv, Tu);
					     Ty = VFMA(LDK(KP707106781), Tv, Tu);
					     Tn = VFNMS(LDK(KP707106781), Tm, T7);
					     Tx = VFMA(LDK(KP707106781), Tm, T7);
					     ST(&(xo[WS(os, 3)]), VFNMSI(T1a, T19), ovs, &(xo[WS(os, 1)]));
					     ST(&(xo[WS(os, 13)]), VFMAI(T1a, T19), ovs, &(xo[WS(os, 1)]));
					     ST(&(xo[WS(os, 11)]), VFNMSI(T18, T15), ovs, &(xo[WS(os, 1)]));
					     ST(&(xo[WS(os, 5)]), VFMAI(T18, T15), ovs, &(xo[WS(os, 1)]));
					     ST(&(xo[WS(os, 15)]), VFNMSI(T12, T11), ovs, &(xo[WS(os, 1)]));
					     ST(&(xo[WS(os, 1)]), VFMAI(T12, T11), ovs, &(xo[WS(os, 1)]));
					     ST(&(xo[WS(os, 9)]), VFMAI(T10, TT), ovs, &(xo[WS(os, 1)]));
					     ST(&(xo[WS(os, 7)]), VFNMSI(T10, TT), ovs, &(xo[WS(os, 1)]));
					     ST(&(xo[WS(os, 2)]), VFMAI(Ty, Tx), ovs, &(xo[0]));
					     ST(&(xo[WS(os, 14)]), VFNMSI(Ty, Tx), ovs, &(xo[0]));
					     ST(&(xo[WS(os, 10)]), VFMAI(Tw, Tn), ovs, &(xo[0]));
					     ST(&(xo[WS(os, 6)]), VFNMSI(Tw, Tn), ovs, &(xo[0]));
					}
				   }
			      }
			 }
		    }
	       }
	  }
     }
     VLEAVE();
}

static const kdft_desc desc = { 16, XSIMD_STRING("n1bv_16"), {38, 0, 34, 0}, &GENUS, 0, 0, 0, 0 };

void XSIMD(codelet_n1bv_16) (planner *p) {
     X(kdft_register) (p, n1bv_16, &desc);
}

#else				/* HAVE_FMA */

/* Generated by: ../../../genfft/gen_notw_c.native -simd -compact -variables 4 -pipeline-latency 8 -sign 1 -n 16 -name n1bv_16 -include n1b.h */

/*
 * This function contains 72 FP additions, 12 FP multiplications,
 * (or, 68 additions, 8 multiplications, 4 fused multiply/add),
 * 30 stack variables, 3 constants, and 32 memory accesses
 */
#include "n1b.h"

static void n1bv_16(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs)
{
     DVK(KP382683432, +0.382683432365089771728459984030398866761344562);
     DVK(KP923879532, +0.923879532511286756128183189396788286822416626);
     DVK(KP707106781, +0.707106781186547524400844362104849039284835938);
     {
	  INT i;
	  const R *xi;
	  R *xo;
	  xi = ii;
	  xo = io;
	  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)) {
	       V Tp, T13, Tu, TY, Tm, T14, Tv, TU, T7, T16, Tx, TN, Te, T17, Ty;
	       V TQ;
	       {
		    V Tn, To, TX, Ts, Tt, TW;
		    Tn = LD(&(xi[WS(is, 4)]), ivs, &(xi[0]));
		    To = LD(&(xi[WS(is, 12)]), ivs, &(xi[0]));
		    TX = VADD(Tn, To);
		    Ts = LD(&(xi[0]), ivs, &(xi[0]));
		    Tt = LD(&(xi[WS(is, 8)]), ivs, &(xi[0]));
		    TW = VADD(Ts, Tt);
		    Tp = VSUB(Tn, To);
		    T13 = VADD(TW, TX);
		    Tu = VSUB(Ts, Tt);
		    TY = VSUB(TW, TX);
	       }
	       {
		    V Ti, TS, Tl, TT;
		    {
			 V Tg, Th, Tj, Tk;
			 Tg = LD(&(xi[WS(is, 2)]), ivs, &(xi[0]));
			 Th = LD(&(xi[WS(is, 10)]), ivs, &(xi[0]));
			 Ti = VSUB(Tg, Th);
			 TS = VADD(Tg, Th);
			 Tj = LD(&(xi[WS(is, 14)]), ivs, &(xi[0]));
			 Tk = LD(&(xi[WS(is, 6)]), ivs, &(xi[0]));
			 Tl = VSUB(Tj, Tk);
			 TT = VADD(Tj, Tk);
		    }
		    Tm = VMUL(LDK(KP707106781), VSUB(Ti, Tl));
		    T14 = VADD(TS, TT);
		    Tv = VMUL(LDK(KP707106781), VADD(Ti, Tl));
		    TU = VSUB(TS, TT);
	       }
	       {
		    V T3, TL, T6, TM;
		    {
			 V T1, T2, T4, T5;
			 T1 = LD(&(xi[WS(is, 1)]), ivs, &(xi[WS(is, 1)]));
			 T2 = LD(&(xi[WS(is, 9)]), ivs, &(xi[WS(is, 1)]));
			 T3 = VSUB(T1, T2);
			 TL = VADD(T1, T2);
			 T4 = LD(&(xi[WS(is, 5)]), ivs, &(xi[WS(is, 1)]));
			 T5 = LD(&(xi[WS(is, 13)]), ivs, &(xi[WS(is, 1)]));
			 T6 = VSUB(T4, T5);
			 TM = VADD(T4, T5);
		    }
		    T7 = VFNMS(LDK(KP382683432), T6, VMUL(LDK(KP923879532), T3));
		    T16 = VADD(TL, TM);
		    Tx = VFMA(LDK(KP382683432), T3, VMUL(LDK(KP923879532), T6));
		    TN = VSUB(TL, TM);
	       }
	       {
		    V Ta, TO, Td, TP;
		    {
			 V T8, T9, Tb, Tc;
			 T8 = LD(&(xi[WS(is, 15)]), ivs, &(xi[WS(is, 1)]));
			 T9 = LD(&(xi[WS(is, 7)]), ivs, &(xi[WS(is, 1)]));
			 Ta = VSUB(T8, T9);
			 TO = VADD(T8, T9);
			 Tb = LD(&(xi[WS(is, 3)]), ivs, &(xi[WS(is, 1)]));
			 Tc = LD(&(xi[WS(is, 11)]), ivs, &(xi[WS(is, 1)]));
			 Td = VSUB(Tb, Tc);
			 TP = VADD(Tb, Tc);
		    }
		    Te = VFMA(LDK(KP923879532), Ta, VMUL(LDK(KP382683432), Td));
		    T17 = VADD(TO, TP);
		    Ty = VFNMS(LDK(KP382683432), Ta, VMUL(LDK(KP923879532), Td));
		    TQ = VSUB(TO, TP);
	       }
	       {
		    V T15, T18, T19, T1a;
		    T15 = VSUB(T13, T14);
		    T18 = VBYI(VSUB(T16, T17));
		    ST(&(xo[WS(os, 12)]), VSUB(T15, T18), ovs, &(xo[0]));
		    ST(&(xo[WS(os, 4)]), VADD(T15, T18), ovs, &(xo[0]));
		    T19 = VADD(T13, T14);
		    T1a = VADD(T16, T17);
		    ST(&(xo[WS(os, 8)]), VSUB(T19, T1a), ovs, &(xo[0]));
		    ST(&(xo[0]), VADD(T19, T1a), ovs, &(xo[0]));
	       }
	       {
		    V TV, T11, T10, T12, TR, TZ;
		    TR = VMUL(LDK(KP707106781), VSUB(TN, TQ));
		    TV = VBYI(VSUB(TR, TU));
		    T11 = VBYI(VADD(TU, TR));
		    TZ = VMUL(LDK(KP707106781), VADD(TN, TQ));
		    T10 = VSUB(TY, TZ);
		    T12 = VADD(TY, TZ);
		    ST(&(xo[WS(os, 6)]), VADD(TV, T10), ovs, &(xo[0]));
		    ST(&(xo[WS(os, 14)]), VSUB(T12, T11), ovs, &(xo[0]));
		    ST(&(xo[WS(os, 10)]), VSUB(T10, TV), ovs, &(xo[0]));
		    ST(&(xo[WS(os, 2)]), VADD(T11, T12), ovs, &(xo[0]));
	       }
	       {
		    V Tr, TB, TA, TC;
		    {
			 V Tf, Tq, Tw, Tz;
			 Tf = VSUB(T7, Te);
			 Tq = VSUB(Tm, Tp);
			 Tr = VBYI(VSUB(Tf, Tq));
			 TB = VBYI(VADD(Tq, Tf));
			 Tw = VSUB(Tu, Tv);
			 Tz = VSUB(Tx, Ty);
			 TA = VSUB(Tw, Tz);
			 TC = VADD(Tw, Tz);
		    }
		    ST(&(xo[WS(os, 5)]), VADD(Tr, TA), ovs, &(xo[WS(os, 1)]));
		    ST(&(xo[WS(os, 13)]), VSUB(TC, TB), ovs, &(xo[WS(os, 1)]));
		    ST(&(xo[WS(os, 11)]), VSUB(TA, Tr), ovs, &(xo[WS(os, 1)]));
		    ST(&(xo[WS(os, 3)]), VADD(TB, TC), ovs, &(xo[WS(os, 1)]));
	       }
	       {
		    V TF, TJ, TI, TK;
		    {
			 V TD, TE, TG, TH;
			 TD = VADD(Tu, Tv);
			 TE = VADD(T7, Te);
			 TF = VADD(TD, TE);
			 TJ = VSUB(TD, TE);
			 TG = VADD(Tp, Tm);
			 TH = VADD(Tx, Ty);
			 TI = VBYI(VADD(TG, TH));
			 TK = VBYI(VSUB(TH, TG));
		    }
		    ST(&(xo[WS(os, 15)]), VSUB(TF, TI), ovs, &(xo[WS(os, 1)]));
		    ST(&(xo[WS(os, 7)]), VADD(TJ, TK), ovs, &(xo[WS(os, 1)]));
		    ST(&(xo[WS(os, 1)]), VADD(TF, TI), ovs, &(xo[WS(os, 1)]));
		    ST(&(xo[WS(os, 9)]), VSUB(TJ, TK), ovs, &(xo[WS(os, 1)]));
	       }
	  }
     }
     VLEAVE();
}

static const kdft_desc desc = { 16, XSIMD_STRING("n1bv_16"), {68, 8, 4, 0}, &GENUS, 0, 0, 0, 0 };

void XSIMD(codelet_n1bv_16) (planner *p) {
     X(kdft_register) (p, n1bv_16, &desc);
}

#endif				/* HAVE_FMA */