annotate src/fftw-3.3.8/dft/simd/common/n2bv_6.c @ 82:d0c2a83c1364

Add FFTW 3.3.8 source, and a Linux build

/*
 * Copyright (c) 2003, 2007-14 Matteo Frigo
 * Copyright (c) 2003, 2007-14 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 Thu May 24 08:05:11 EDT 2018 */

#include "dft/codelet-dft.h"

#if defined(ARCH_PREFERS_FMA) || defined(ISA_EXTENSION_PREFERS_FMA)

/* Generated by: ../../../genfft/gen_notw_c.native -fma -simd -compact -variables 4 -pipeline-latency 8 -sign 1 -n 6 -name n2bv_6 -with-ostride 2 -include dft/simd/n2b.h -store-multiple 2 */

/*
 * This function contains 18 FP additions, 8 FP multiplications,
 * (or, 12 additions, 2 multiplications, 6 fused multiply/add),
 * 25 stack variables, 2 constants, and 15 memory accesses
 */
#include "dft/simd/n2b.h"

static void n2bv_6(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs)
{
     DVK(KP866025403, +0.866025403784438646763723170752936183471402627);
     DVK(KP500000000, +0.500000000000000000000000000000000000000000000);
     {
	  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(12, is), MAKE_VOLATILE_STRIDE(12, os)) {
	       V T3, Td, T6, Te, T9, Tf, Ta, Tg, T1, T2, Tj, Tk;
	       T1 = LD(&(xi[0]), ivs, &(xi[0]));
	       T2 = LD(&(xi[WS(is, 3)]), ivs, &(xi[WS(is, 1)]));
	       T3 = VSUB(T1, T2);
	       Td = VADD(T1, T2);
	       {
		    V T4, T5, T7, T8;
		    T4 = LD(&(xi[WS(is, 2)]), ivs, &(xi[0]));
		    T5 = LD(&(xi[WS(is, 5)]), ivs, &(xi[WS(is, 1)]));
		    T6 = VSUB(T4, T5);
		    Te = VADD(T4, T5);
		    T7 = LD(&(xi[WS(is, 4)]), ivs, &(xi[0]));
		    T8 = LD(&(xi[WS(is, 1)]), ivs, &(xi[WS(is, 1)]));
		    T9 = VSUB(T7, T8);
		    Tf = VADD(T7, T8);
	       }
	       Ta = VADD(T6, T9);
	       Tg = VADD(Te, Tf);
	       Tj = VADD(T3, Ta);
	       STM2(&(xo[6]), Tj, ovs, &(xo[2]));
	       Tk = VADD(Td, Tg);
	       STM2(&(xo[0]), Tk, ovs, &(xo[0]));
	       {
		    V Tm, Tb, Tc, Tl;
		    Tb = VFNMS(LDK(KP500000000), Ta, T3);
		    Tc = VMUL(LDK(KP866025403), VSUB(T6, T9));
		    Tl = VFMAI(Tc, Tb);
		    STM2(&(xo[2]), Tl, ovs, &(xo[2]));
		    STN2(&(xo[0]), Tk, Tl, ovs);
		    Tm = VFNMSI(Tc, Tb);
		    STM2(&(xo[10]), Tm, ovs, &(xo[2]));
		    {
			 V Th, Ti, Tn, To;
			 Th = VFNMS(LDK(KP500000000), Tg, Td);
			 Ti = VMUL(LDK(KP866025403), VSUB(Te, Tf));
			 Tn = VFNMSI(Ti, Th);
			 STM2(&(xo[4]), Tn, ovs, &(xo[0]));
			 STN2(&(xo[4]), Tn, Tj, ovs);
			 To = VFMAI(Ti, Th);
			 STM2(&(xo[8]), To, ovs, &(xo[0]));
			 STN2(&(xo[8]), To, Tm, ovs);
		    }
	       }
	  }
     }
     VLEAVE();
}

static const kdft_desc desc = { 6, XSIMD_STRING("n2bv_6"), {12, 2, 6, 0}, &GENUS, 0, 2, 0, 0 };

void XSIMD(codelet_n2bv_6) (planner *p) {
     X(kdft_register) (p, n2bv_6, &desc);
}

#else

/* Generated by: ../../../genfft/gen_notw_c.native -simd -compact -variables 4 -pipeline-latency 8 -sign 1 -n 6 -name n2bv_6 -with-ostride 2 -include dft/simd/n2b.h -store-multiple 2 */

/*
 * This function contains 18 FP additions, 4 FP multiplications,
 * (or, 16 additions, 2 multiplications, 2 fused multiply/add),
 * 25 stack variables, 2 constants, and 15 memory accesses
 */
#include "dft/simd/n2b.h"

static void n2bv_6(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs)
{
     DVK(KP500000000, +0.500000000000000000000000000000000000000000000);
     DVK(KP866025403, +0.866025403784438646763723170752936183471402627);
     {
	  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(12, is), MAKE_VOLATILE_STRIDE(12, os)) {
	       V Ta, Td, T3, Te, T6, Tf, Tb, Tg, T8, T9, Tj, Tk;
	       T8 = LD(&(xi[0]), ivs, &(xi[0]));
	       T9 = LD(&(xi[WS(is, 3)]), ivs, &(xi[WS(is, 1)]));
	       Ta = VSUB(T8, T9);
	       Td = VADD(T8, T9);
	       {
		    V T1, T2, T4, T5;
		    T1 = LD(&(xi[WS(is, 2)]), ivs, &(xi[0]));
		    T2 = LD(&(xi[WS(is, 5)]), ivs, &(xi[WS(is, 1)]));
		    T3 = VSUB(T1, T2);
		    Te = VADD(T1, T2);
		    T4 = LD(&(xi[WS(is, 4)]), ivs, &(xi[0]));
		    T5 = LD(&(xi[WS(is, 1)]), ivs, &(xi[WS(is, 1)]));
		    T6 = VSUB(T4, T5);
		    Tf = VADD(T4, T5);
	       }
	       Tb = VADD(T3, T6);
	       Tg = VADD(Te, Tf);
	       Tj = VADD(Ta, Tb);
	       STM2(&(xo[6]), Tj, ovs, &(xo[2]));
	       Tk = VADD(Td, Tg);
	       STM2(&(xo[0]), Tk, ovs, &(xo[0]));
	       {
		    V Tm, T7, Tc, Tl;
		    T7 = VBYI(VMUL(LDK(KP866025403), VSUB(T3, T6)));
		    Tc = VFNMS(LDK(KP500000000), Tb, Ta);
		    Tl = VADD(T7, Tc);
		    STM2(&(xo[2]), Tl, ovs, &(xo[2]));
		    STN2(&(xo[0]), Tk, Tl, ovs);
		    Tm = VSUB(Tc, T7);
		    STM2(&(xo[10]), Tm, ovs, &(xo[2]));
		    {
			 V Th, Ti, Tn, To;
			 Th = VFNMS(LDK(KP500000000), Tg, Td);
			 Ti = VBYI(VMUL(LDK(KP866025403), VSUB(Te, Tf)));
			 Tn = VSUB(Th, Ti);
			 STM2(&(xo[4]), Tn, ovs, &(xo[0]));
			 STN2(&(xo[4]), Tn, Tj, ovs);
			 To = VADD(Ti, Th);
			 STM2(&(xo[8]), To, ovs, &(xo[0]));
			 STN2(&(xo[8]), To, Tm, ovs);
		    }
	       }
	  }
     }
     VLEAVE();
}

static const kdft_desc desc = { 6, XSIMD_STRING("n2bv_6"), {16, 2, 2, 0}, &GENUS, 0, 2, 0, 0 };

void XSIMD(codelet_n2bv_6) (planner *p) {
     X(kdft_register) (p, n2bv_6, &desc);
}

#endif