2 * This file is part of the GROMACS molecular simulation package.
4 * Copyright (c) 2012,2013,2014, 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.
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.
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.
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.
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.
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.
41 #include "gromacs/mdlib/nbnxn_simd.h"
45 {1}#include "gromacs/simd/vector_operations.h"
48 #define GMX_SIMD_J_UNROLL_SIZE {3}
50 #include "../nbnxn_kernel_common.h"
51 #include "gmx_omp_nthreads.h"
52 #include "types/force_flags.h"
53 #include "gmx_fatal.h"
55 /*! \brief Kinds of electrostatic treatments in SIMD Verlet kernels
58 coulktRF, coulktTAB, coulktTAB_TWIN, coulktEWALD, coulktEWALD_TWIN, coulktNR
61 /*! \brief Kinds of Van der Waals treatments in SIMD Verlet kernels
64 vdwktLJCUT_COMBGEOM, vdwktLJCUT_COMBLB, vdwktLJCUT_COMBNONE, vdwktLJFORCESWITCH, vdwktLJPOTSWITCH, vdwktLJEWALDCOMBGEOM, vdwktNR
67 /* Declare and define the kernel function pointer lookup tables.
68 * The minor index of the array goes over both the LJ combination rules,
69 * which is only supported by plain cut-off, and the LJ switch/PME functions.
71 static p_nbk_func_noener p_nbk_noener[coulktNR][vdwktNR] =
73 static p_nbk_func_ener p_nbk_ener[coulktNR][vdwktNR] =
75 static p_nbk_func_ener p_nbk_energrp[coulktNR][vdwktNR] =
79 reduce_group_energies(int ng, int ng_2log,
80 const real *VSvdw, const real *VSc,
83 const int unrollj = GMX_SIMD_REAL_WIDTH/GMX_SIMD_J_UNROLL_SIZE;
84 const int unrollj_half = unrollj/2;
85 int ng_p2, i, j, j0, j1, c, s;
89 /* The size of the x86 SIMD energy group buffer array is:
90 * ng*ng*ng_p2*unrollj_half*simd_width
92 for (i = 0; i < ng; i++)
94 for (j = 0; j < ng; j++)
100 for (j1 = 0; j1 < ng; j1++)
102 for (j0 = 0; j0 < ng; j0++)
104 c = ((i*ng + j1)*ng_p2 + j0)*unrollj_half*unrollj;
105 for (s = 0; s < unrollj_half; s++)
107 Vvdw[i*ng+j0] += VSvdw[c+0];
108 Vvdw[i*ng+j1] += VSvdw[c+1];
109 Vc [i*ng+j0] += VSc [c+0];
110 Vc [i*ng+j1] += VSc [c+1];
120 #include "gmx_fatal.h"
125 {5}(nbnxn_pairlist_set_t gmx_unused *nbl_list,
126 {6}const nbnxn_atomdata_t gmx_unused *nbat,
127 {6}const interaction_const_t gmx_unused *ic,
128 {6}int gmx_unused ewald_excl,
129 {6}rvec gmx_unused *shift_vec,
130 {6}int gmx_unused force_flags,
131 {6}int gmx_unused clearF,
132 {6}real gmx_unused *fshift,
133 {6}real gmx_unused *Vc,
134 {6}real gmx_unused *Vvdw)
138 nbnxn_pairlist_t **nbl;
139 int coulkt, vdwkt = 0;
141 int nthreads gmx_unused;
143 nnbl = nbl_list->nnbl;
146 if (EEL_RF(ic->eeltype) || ic->eeltype == eelCUT)
152 if (ewald_excl == ewaldexclTable)
154 if (ic->rcoulomb == ic->rvdw)
160 coulkt = coulktTAB_TWIN;
165 if (ic->rcoulomb == ic->rvdw)
167 coulkt = coulktEWALD;
171 coulkt = coulktEWALD_TWIN;
176 if (ic->vdwtype == evdwCUT)
178 switch (ic->vdw_modifier)
181 case eintmodPOTSHIFT:
182 switch (nbat->comb_rule)
184 case ljcrGEOM: vdwkt = vdwktLJCUT_COMBGEOM; break;
185 case ljcrLB: vdwkt = vdwktLJCUT_COMBLB; break;
186 case ljcrNONE: vdwkt = vdwktLJCUT_COMBNONE; break;
187 default: gmx_incons("Unknown combination rule");
190 case eintmodFORCESWITCH:
191 vdwkt = vdwktLJFORCESWITCH;
193 case eintmodPOTSWITCH:
194 vdwkt = vdwktLJPOTSWITCH;
197 gmx_incons("Unsupported VdW interaction modifier");
200 else if (ic->vdwtype == evdwPME)
202 if (ic->ljpme_comb_rule == eljpmeLB)
204 gmx_incons("The nbnxn SIMD kernels don't suport LJ-PME with LB");
206 vdwkt = vdwktLJEWALDCOMBGEOM;
210 gmx_incons("Unsupported VdW interaction type");
213 nthreads = gmx_omp_nthreads_get(emntNonbonded);
214 #pragma omp parallel for schedule(static) num_threads(nthreads)
215 for (nb = 0; nb < nnbl; nb++)
217 nbnxn_atomdata_output_t *out;
220 out = &nbat->out[nb];
222 if (clearF == enbvClearFYes)
224 clear_f(nbat, nb, out->f);
227 if ((force_flags & GMX_FORCE_VIRIAL) && nnbl == 1)
233 fshift_p = out->fshift;
235 if (clearF == enbvClearFYes)
237 clear_fshift(fshift_p);
241 if (!(force_flags & GMX_FORCE_ENERGY))
243 /* Don't calculate energies */
244 p_nbk_noener[coulkt][vdwkt](nbl[nb], nbat,
250 else if (out->nV == 1)
252 /* No energy groups */
256 p_nbk_ener[coulkt][vdwkt](nbl[nb], nbat,
266 /* Calculate energy group contributions */
269 for (i = 0; i < out->nVS; i++)
273 for (i = 0; i < out->nVS; i++)
278 p_nbk_energrp[coulkt][vdwkt](nbl[nb], nbat,
286 reduce_group_energies(nbat->nenergrp, nbat->neg_2log,
287 out->VSvdw, out->VSc,
292 if (force_flags & GMX_FORCE_ENERGY)
294 reduce_energies_over_lists(nbat, nnbl, Vvdw, Vc);
299 gmx_incons("{5} called when such kernels "
300 " are not enabled.");
303 #undef GMX_SIMD_J_UNROLL_SIZE