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.
36 * Note: this file was generated by the Verlet kernel generator for
46 #ifdef GMX_NBNXN_SIMD_2XNN
48 /* Include the full-width SIMD macros */
50 #include "gromacs/simd/macros.h"
51 #include "gromacs/simd/vector_operations.h"
53 #if !(GMX_SIMD_REAL_WIDTH == 8 || GMX_SIMD_REAL_WIDTH == 16)
54 #error "unsupported SIMD width"
57 #define GMX_SIMD_J_UNROLL_SIZE 2
58 #include "nbnxn_kernel_simd_2xnn.h"
59 #include "../nbnxn_kernel_common.h"
60 #include "gmx_omp_nthreads.h"
61 #include "types/force_flags.h"
62 #include "gmx_fatal.h"
64 /*! \brief Kinds of electrostatic treatments in SIMD Verlet kernels
67 coultRF, coultTAB, coultTAB_TWIN, coultEWALD, coultEWALD_TWIN, coultNR
70 /* Declare and define the kernel function pointer lookup tables.
71 * The minor index of the array goes over both the LJ combination rules,
72 * which is only supported by plain cut-off, and the LJ switch functions.
74 static p_nbk_func_noener p_nbk_noener[coultNR][ljcrNR+2] =
77 nbnxn_kernel_ElecRF_VdwLJCombGeom_F_2xnn,
78 nbnxn_kernel_ElecRF_VdwLJCombLB_F_2xnn,
79 nbnxn_kernel_ElecRF_VdwLJ_F_2xnn,
80 nbnxn_kernel_ElecRF_VdwLJFSw_F_2xnn,
81 nbnxn_kernel_ElecRF_VdwLJPSw_F_2xnn,
84 nbnxn_kernel_ElecQSTab_VdwLJCombGeom_F_2xnn,
85 nbnxn_kernel_ElecQSTab_VdwLJCombLB_F_2xnn,
86 nbnxn_kernel_ElecQSTab_VdwLJ_F_2xnn,
87 nbnxn_kernel_ElecQSTab_VdwLJFSw_F_2xnn,
88 nbnxn_kernel_ElecQSTab_VdwLJPSw_F_2xnn,
91 nbnxn_kernel_ElecQSTabTwinCut_VdwLJCombGeom_F_2xnn,
92 nbnxn_kernel_ElecQSTabTwinCut_VdwLJCombLB_F_2xnn,
93 nbnxn_kernel_ElecQSTabTwinCut_VdwLJ_F_2xnn,
94 nbnxn_kernel_ElecQSTabTwinCut_VdwLJFSw_F_2xnn,
95 nbnxn_kernel_ElecQSTabTwinCut_VdwLJPSw_F_2xnn,
98 nbnxn_kernel_ElecEw_VdwLJCombGeom_F_2xnn,
99 nbnxn_kernel_ElecEw_VdwLJCombLB_F_2xnn,
100 nbnxn_kernel_ElecEw_VdwLJ_F_2xnn,
101 nbnxn_kernel_ElecEw_VdwLJFSw_F_2xnn,
102 nbnxn_kernel_ElecEw_VdwLJPSw_F_2xnn,
105 nbnxn_kernel_ElecEwTwinCut_VdwLJCombGeom_F_2xnn,
106 nbnxn_kernel_ElecEwTwinCut_VdwLJCombLB_F_2xnn,
107 nbnxn_kernel_ElecEwTwinCut_VdwLJ_F_2xnn,
108 nbnxn_kernel_ElecEwTwinCut_VdwLJFSw_F_2xnn,
109 nbnxn_kernel_ElecEwTwinCut_VdwLJPSw_F_2xnn,
113 static p_nbk_func_ener p_nbk_ener[coultNR][ljcrNR+2] =
116 nbnxn_kernel_ElecRF_VdwLJCombGeom_VF_2xnn,
117 nbnxn_kernel_ElecRF_VdwLJCombLB_VF_2xnn,
118 nbnxn_kernel_ElecRF_VdwLJ_VF_2xnn,
119 nbnxn_kernel_ElecRF_VdwLJFSw_VF_2xnn,
120 nbnxn_kernel_ElecRF_VdwLJPSw_VF_2xnn,
123 nbnxn_kernel_ElecQSTab_VdwLJCombGeom_VF_2xnn,
124 nbnxn_kernel_ElecQSTab_VdwLJCombLB_VF_2xnn,
125 nbnxn_kernel_ElecQSTab_VdwLJ_VF_2xnn,
126 nbnxn_kernel_ElecQSTab_VdwLJFSw_VF_2xnn,
127 nbnxn_kernel_ElecQSTab_VdwLJPSw_VF_2xnn,
130 nbnxn_kernel_ElecQSTabTwinCut_VdwLJCombGeom_VF_2xnn,
131 nbnxn_kernel_ElecQSTabTwinCut_VdwLJCombLB_VF_2xnn,
132 nbnxn_kernel_ElecQSTabTwinCut_VdwLJ_VF_2xnn,
133 nbnxn_kernel_ElecQSTabTwinCut_VdwLJFSw_VF_2xnn,
134 nbnxn_kernel_ElecQSTabTwinCut_VdwLJPSw_VF_2xnn,
137 nbnxn_kernel_ElecEw_VdwLJCombGeom_VF_2xnn,
138 nbnxn_kernel_ElecEw_VdwLJCombLB_VF_2xnn,
139 nbnxn_kernel_ElecEw_VdwLJ_VF_2xnn,
140 nbnxn_kernel_ElecEw_VdwLJFSw_VF_2xnn,
141 nbnxn_kernel_ElecEw_VdwLJPSw_VF_2xnn,
144 nbnxn_kernel_ElecEwTwinCut_VdwLJCombGeom_VF_2xnn,
145 nbnxn_kernel_ElecEwTwinCut_VdwLJCombLB_VF_2xnn,
146 nbnxn_kernel_ElecEwTwinCut_VdwLJ_VF_2xnn,
147 nbnxn_kernel_ElecEwTwinCut_VdwLJFSw_VF_2xnn,
148 nbnxn_kernel_ElecEwTwinCut_VdwLJPSw_VF_2xnn,
152 static p_nbk_func_ener p_nbk_energrp[coultNR][ljcrNR+2] =
155 nbnxn_kernel_ElecRF_VdwLJCombGeom_VgrpF_2xnn,
156 nbnxn_kernel_ElecRF_VdwLJCombLB_VgrpF_2xnn,
157 nbnxn_kernel_ElecRF_VdwLJ_VgrpF_2xnn,
158 nbnxn_kernel_ElecRF_VdwLJFSw_VgrpF_2xnn,
159 nbnxn_kernel_ElecRF_VdwLJPSw_VgrpF_2xnn,
162 nbnxn_kernel_ElecQSTab_VdwLJCombGeom_VgrpF_2xnn,
163 nbnxn_kernel_ElecQSTab_VdwLJCombLB_VgrpF_2xnn,
164 nbnxn_kernel_ElecQSTab_VdwLJ_VgrpF_2xnn,
165 nbnxn_kernel_ElecQSTab_VdwLJFSw_VgrpF_2xnn,
166 nbnxn_kernel_ElecQSTab_VdwLJPSw_VgrpF_2xnn,
169 nbnxn_kernel_ElecQSTabTwinCut_VdwLJCombGeom_VgrpF_2xnn,
170 nbnxn_kernel_ElecQSTabTwinCut_VdwLJCombLB_VgrpF_2xnn,
171 nbnxn_kernel_ElecQSTabTwinCut_VdwLJ_VgrpF_2xnn,
172 nbnxn_kernel_ElecQSTabTwinCut_VdwLJFSw_VgrpF_2xnn,
173 nbnxn_kernel_ElecQSTabTwinCut_VdwLJPSw_VgrpF_2xnn,
176 nbnxn_kernel_ElecEw_VdwLJCombGeom_VgrpF_2xnn,
177 nbnxn_kernel_ElecEw_VdwLJCombLB_VgrpF_2xnn,
178 nbnxn_kernel_ElecEw_VdwLJ_VgrpF_2xnn,
179 nbnxn_kernel_ElecEw_VdwLJFSw_VgrpF_2xnn,
180 nbnxn_kernel_ElecEw_VdwLJPSw_VgrpF_2xnn,
183 nbnxn_kernel_ElecEwTwinCut_VdwLJCombGeom_VgrpF_2xnn,
184 nbnxn_kernel_ElecEwTwinCut_VdwLJCombLB_VgrpF_2xnn,
185 nbnxn_kernel_ElecEwTwinCut_VdwLJ_VgrpF_2xnn,
186 nbnxn_kernel_ElecEwTwinCut_VdwLJFSw_VgrpF_2xnn,
187 nbnxn_kernel_ElecEwTwinCut_VdwLJPSw_VgrpF_2xnn,
193 reduce_group_energies(int ng, int ng_2log,
194 const real *VSvdw, const real *VSc,
195 real *Vvdw, real *Vc)
197 const int unrollj = GMX_SIMD_REAL_WIDTH/GMX_SIMD_J_UNROLL_SIZE;
198 const int unrollj_half = unrollj/2;
199 int ng_p2, i, j, j0, j1, c, s;
201 ng_p2 = (1<<ng_2log);
203 /* The size of the x86 SIMD energy group buffer array is:
204 * ng*ng*ng_p2*unrollj_half*simd_width
206 for (i = 0; i < ng; i++)
208 for (j = 0; j < ng; j++)
214 for (j1 = 0; j1 < ng; j1++)
216 for (j0 = 0; j0 < ng; j0++)
218 c = ((i*ng + j1)*ng_p2 + j0)*unrollj_half*unrollj;
219 for (s = 0; s < unrollj_half; s++)
221 Vvdw[i*ng+j0] += VSvdw[c+0];
222 Vvdw[i*ng+j1] += VSvdw[c+1];
223 Vc [i*ng+j0] += VSc [c+0];
224 Vc [i*ng+j1] += VSc [c+1];
232 #else /* GMX_NBNXN_SIMD_2XNN */
234 #include "gmx_fatal.h"
236 #endif /* GMX_NBNXN_SIMD_2XNN */
239 nbnxn_kernel_simd_2xnn(nbnxn_pairlist_set_t gmx_unused *nbl_list,
240 const nbnxn_atomdata_t gmx_unused *nbat,
241 const interaction_const_t gmx_unused *ic,
242 int gmx_unused ewald_excl,
243 rvec gmx_unused *shift_vec,
244 int gmx_unused force_flags,
245 int gmx_unused clearF,
246 real gmx_unused *fshift,
248 real gmx_unused *Vvdw)
249 #ifdef GMX_NBNXN_SIMD_2XNN
252 nbnxn_pairlist_t **nbl;
253 int coult, ljtreatment = 0;
256 nnbl = nbl_list->nnbl;
259 if (EEL_RF(ic->eeltype) || ic->eeltype == eelCUT)
265 if (ewald_excl == ewaldexclTable)
267 if (ic->rcoulomb == ic->rvdw)
273 coult = coultTAB_TWIN;
278 if (ic->rcoulomb == ic->rvdw)
284 coult = coultEWALD_TWIN;
289 switch (ic->vdw_modifier)
292 case eintmodPOTSHIFT:
293 ljtreatment = nbat->comb_rule;
295 /* Switch functions follow after cut-off combination rule kernels */
296 case eintmodFORCESWITCH:
297 ljtreatment = ljcrNR;
299 case eintmodPOTSWITCH:
300 ljtreatment = ljcrNR + 1;
303 gmx_incons("Unsupported VdW interaction modifier");
306 #pragma omp parallel for schedule(static) num_threads(gmx_omp_nthreads_get(emntNonbonded))
307 for (nb = 0; nb < nnbl; nb++)
309 nbnxn_atomdata_output_t *out;
312 out = &nbat->out[nb];
314 if (clearF == enbvClearFYes)
316 clear_f(nbat, nb, out->f);
319 if ((force_flags & GMX_FORCE_VIRIAL) && nnbl == 1)
325 fshift_p = out->fshift;
327 if (clearF == enbvClearFYes)
329 clear_fshift(fshift_p);
333 if (!(force_flags & GMX_FORCE_ENERGY))
335 /* Don't calculate energies */
336 p_nbk_noener[coult][ljtreatment](nbl[nb], nbat,
342 else if (out->nV == 1)
344 /* No energy groups */
348 p_nbk_ener[coult][ljtreatment](nbl[nb], nbat,
358 /* Calculate energy group contributions */
361 for (i = 0; i < out->nVS; i++)
365 for (i = 0; i < out->nVS; i++)
370 p_nbk_energrp[coult][ljtreatment](nbl[nb], nbat,
378 reduce_group_energies(nbat->nenergrp, nbat->neg_2log,
379 out->VSvdw, out->VSc,
384 if (force_flags & GMX_FORCE_ENERGY)
386 reduce_energies_over_lists(nbat, nnbl, Vvdw, Vc);
391 gmx_incons("nbnxn_kernel_simd_2xnn called when such kernels "
392 " are not enabled.");
395 #undef GMX_SIMD_J_UNROLL_SIZE