src/gromacs/mdlib/nbnxn_kernels/nbnxn_kernel_file_generator/nbnxn_kernel_simd_template.c.pre

   1 /*
   2  * This file is part of the GROMACS molecular simulation package.
   3  *
   4  * Copyright (c) 2012,2013, by the GROMACS development team, led by
   5  * Mark Abraham, David van der Spoel, Berk Hess, and Erik Lindahl,
   6  * and including many others, as listed in the AUTHORS file in the
   7  * top-level source directory and at http://www.gromacs.org.
   8  *
   9  * GROMACS is free software; you can redistribute it and/or
  10  * modify it under the terms of the GNU Lesser General Public License
  11  * as published by the Free Software Foundation; either version 2.1
  12  * of the License, or (at your option) any later version.
  13  *
  14  * GROMACS is distributed in the hope that it will be useful,
  15  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  16  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  17  * Lesser General Public License for more details.
  18  *
  19  * You should have received a copy of the GNU Lesser General Public
  20  * License along with GROMACS; if not, see
  21  * http://www.gnu.org/licenses, or write to the Free Software Foundation,
  22  * Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02110-1301  USA.
  23  *
  24  * If you want to redistribute modifications to GROMACS, please
  25  * consider that scientific software is very special. Version
  26  * control is crucial - bugs must be traceable. We will be happy to
  27  * consider code for inclusion in the official distribution, but
  28  * derived work must not be called official GROMACS. Details are found
  29  * in the README & COPYING files - if they are missing, get the
  30  * official version at http://www.gromacs.org.
  31  *
  32  * To help us fund GROMACS development, we humbly ask that you cite
  33  * the research papers on the package. Check out http://www.gromacs.org.
  34  */
  35 #ifdef HAVE_CONFIG_H
  36 #include <config.h>
  37 #endif
  38
  39 #include "typedefs.h"
  40
  41 #ifdef {0}
  42
  43 {1}
  44 #include "gmx_simd_macros.h"
  45 #include "gmx_simd_vec.h"
  46 {2}
  47 #define GMX_SIMD_J_UNROLL_SIZE {3}
  48 #include "{4}"
  49 #include "../nbnxn_kernel_common.h"
  50 #include "gmx_omp_nthreads.h"
  51 #include "types/force_flags.h"
  52
  53 /*! \brief Kinds of electrostatic treatments in SIMD Verlet kernels
  54  */
  55 enum {{
  56     coultRF, coultTAB, coultTAB_TWIN, coultEWALD, coultEWALD_TWIN, coultNR
  57 }};
  58
  59 /* Declare and define the kernel function pointer lookup tables. */
  60 static p_nbk_func_ener p_nbk_ener[coultNR][ljcrNR] =
  61 {7}
  62 static p_nbk_func_ener p_nbk_energrp[coultNR][ljcrNR] =
  63 {8}
  64 static p_nbk_func_noener p_nbk_noener[coultNR][ljcrNR] =
  65 {9}
  66
  67 static void
  68 reduce_group_energies(int ng, int ng_2log,
  69                       const real *VSvdw, const real *VSc,
  70                       real *Vvdw, real *Vc)
  71 {{
  72     const int unrollj      = GMX_SIMD_WIDTH_HERE/GMX_SIMD_J_UNROLL_SIZE;
  73     const int unrollj_half = unrollj/2;
  74     int       ng_p2, i, j, j0, j1, c, s;
  75
  76     ng_p2 = (1<<ng_2log);
  77
  78     /* The size of the x86 SIMD energy group buffer array is:
  79      * ng*ng*ng_p2*unrollj_half*simd_width
  80      */
  81     for (i = 0; i < ng; i++)
  82     {{
  83         for (j = 0; j < ng; j++)
  84         {{
  85             Vvdw[i*ng+j] = 0;
  86             Vc[i*ng+j]   = 0;
  87         }}
  88
  89         for (j1 = 0; j1 < ng; j1++)
  90         {{
  91             for (j0 = 0; j0 < ng; j0++)
  92             {{
  93                 c = ((i*ng + j1)*ng_p2 + j0)*unrollj_half*unrollj;
  94                 for (s = 0; s < unrollj_half; s++)
  95                 {{
  96                     Vvdw[i*ng+j0] += VSvdw[c+0];
  97                     Vvdw[i*ng+j1] += VSvdw[c+1];
  98                     Vc  [i*ng+j0] += VSc  [c+0];
  99                     Vc  [i*ng+j1] += VSc  [c+1];
 100                     c             += unrollj + 2;
 101                 }}
 102             }}
 103         }}
 104     }}
 105 }}
 106
 107 #else /* {0} */
 108
 109 #include "gmx_fatal.h"
 110
 111 #endif /* {0} */
 112
 113 void
 114 {5}(nbnxn_pairlist_set_t      gmx_unused *nbl_list,
 115 {6}const nbnxn_atomdata_t    gmx_unused *nbat,
 116 {6}const interaction_const_t gmx_unused *ic,
 117 {6}int                       gmx_unused  ewald_excl,
 118 {6}rvec                      gmx_unused *shift_vec,
 119 {6}int                       gmx_unused  force_flags,
 120 {6}int                       gmx_unused  clearF,
 121 {6}real                      gmx_unused *fshift,
 122 {6}real                      gmx_unused *Vc,
 123 {6}real                      gmx_unused *Vvdw)
 124 #ifdef {0}
 125 {{
 126     int                nnbl;
 127     nbnxn_pairlist_t **nbl;
 128     int                coult;
 129     int                nb;
 130
 131     nnbl = nbl_list->nnbl;
 132     nbl  = nbl_list->nbl;
 133
 134     if (EEL_RF(ic->eeltype) || ic->eeltype == eelCUT)
 135     {{
 136         coult = coultRF;
 137     }}
 138     else
 139     {{
 140         if (ewald_excl == ewaldexclTable)
 141         {{
 142             if (ic->rcoulomb == ic->rvdw)
 143             {{
 144                 coult = coultTAB;
 145             }}
 146             else
 147             {{
 148                 coult = coultTAB_TWIN;
 149             }}
 150         }}
 151         else
 152         {{
 153             if (ic->rcoulomb == ic->rvdw)
 154             {{
 155                 coult = coultEWALD;
 156             }}
 157             else
 158             {{
 159                 coult = coultEWALD_TWIN;
 160             }}
 161         }}
 162     }}
 163
 164 #pragma omp parallel for schedule(static) num_threads(gmx_omp_nthreads_get(emntNonbonded))
 165     for (nb = 0; nb < nnbl; nb++)
 166     {{
 167         nbnxn_atomdata_output_t *out;
 168         real                    *fshift_p;
 169
 170         out = &nbat->out[nb];
 171
 172         if (clearF == enbvClearFYes)
 173         {{
 174             clear_f(nbat, nb, out->f);
 175         }}
 176
 177         if ((force_flags & GMX_FORCE_VIRIAL) && nnbl == 1)
 178         {{
 179             fshift_p = fshift;
 180         }}
 181         else
 182         {{
 183             fshift_p = out->fshift;
 184
 185             if (clearF == enbvClearFYes)
 186             {{
 187                 clear_fshift(fshift_p);
 188             }}
 189         }}
 190
 191         if (!(force_flags & GMX_FORCE_ENERGY))
 192         {{
 193             /* Don't calculate energies */
 194             p_nbk_noener[coult][nbat->comb_rule](nbl[nb], nbat,
 195                                                  ic,
 196                                                  shift_vec,
 197                                                  out->f,
 198                                                  fshift_p);
 199         }}
 200         else if (out->nV == 1)
 201         {{
 202             /* No energy groups */
 203             out->Vvdw[0] = 0;
 204             out->Vc[0]   = 0;
 205
 206             p_nbk_ener[coult][nbat->comb_rule](nbl[nb], nbat,
 207                                                ic,
 208                                                shift_vec,
 209                                                out->f,
 210                                                fshift_p,
 211                                                out->Vvdw,
 212                                                out->Vc);
 213         }}
 214         else
 215         {{
 216             /* Calculate energy group contributions */
 217             int i;
 218
 219             for (i = 0; i < out->nVS; i++)
 220             {{
 221                 out->VSvdw[i] = 0;
 222             }}
 223             for (i = 0; i < out->nVS; i++)
 224             {{
 225                 out->VSc[i] = 0;
 226             }}
 227
 228             p_nbk_energrp[coult][nbat->comb_rule](nbl[nb], nbat,
 229                                                   ic,
 230                                                   shift_vec,
 231                                                   out->f,
 232                                                   fshift_p,
 233                                                   out->VSvdw,
 234                                                   out->VSc);
 235
 236             reduce_group_energies(nbat->nenergrp, nbat->neg_2log,
 237                                   out->VSvdw, out->VSc,
 238                                   out->Vvdw, out->Vc);
 239         }}
 240     }}
 241
 242     if (force_flags & GMX_FORCE_ENERGY)
 243     {{
 244         reduce_energies_over_lists(nbat, nnbl, Vvdw, Vc);
 245     }}
 246 }}
 247 #else
 248 {{
 249     gmx_incons("{5} called when such kernels "
 250                " are not enabled.");
 251 }}
 252 #endif
 253 #undef GMX_SIMD_J_UNROLL_SIZE