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_4XN
48 #ifdef GMX_NBNXN_HALF_WIDTH_SIMD
49 #define GMX_USE_HALF_WIDTH_SIMD_HERE
52 #include "gromacs/simd/macros.h"
53 #include "gromacs/simd/vector_operations.h"
54 #if !(GMX_SIMD_REAL_WIDTH == 2 || GMX_SIMD_REAL_WIDTH == 4 || GMX_SIMD_REAL_WIDTH == 8)
55 #error "unsupported SIMD width"
58 #define GMX_SIMD_J_UNROLL_SIZE 1
59 #include "nbnxn_kernel_simd_4xn.h"
60 #include "../nbnxn_kernel_common.h"
61 #include "gmx_omp_nthreads.h"
62 #include "types/force_flags.h"
63 #include "gmx_fatal.h"
65 /*! \brief Kinds of electrostatic treatments in SIMD Verlet kernels
68 coultRF, coultTAB, coultTAB_TWIN, coultEWALD, coultEWALD_TWIN, coultNR
71 /* Declare and define the kernel function pointer lookup tables.
72 * The minor index of the array goes over both the LJ combination rules,
73 * which is only supported by plain cut-off, and the LJ switch functions.
75 static p_nbk_func_noener p_nbk_noener[coultNR][ljcrNR+2] =
78 nbnxn_kernel_ElecRF_VdwLJCombGeom_F_4xn,
79 nbnxn_kernel_ElecRF_VdwLJCombLB_F_4xn,
80 nbnxn_kernel_ElecRF_VdwLJ_F_4xn,
81 nbnxn_kernel_ElecRF_VdwLJFSw_F_4xn,
82 nbnxn_kernel_ElecRF_VdwLJPSw_F_4xn,
85 nbnxn_kernel_ElecQSTab_VdwLJCombGeom_F_4xn,
86 nbnxn_kernel_ElecQSTab_VdwLJCombLB_F_4xn,
87 nbnxn_kernel_ElecQSTab_VdwLJ_F_4xn,
88 nbnxn_kernel_ElecQSTab_VdwLJFSw_F_4xn,
89 nbnxn_kernel_ElecQSTab_VdwLJPSw_F_4xn,
92 nbnxn_kernel_ElecQSTabTwinCut_VdwLJCombGeom_F_4xn,
93 nbnxn_kernel_ElecQSTabTwinCut_VdwLJCombLB_F_4xn,
94 nbnxn_kernel_ElecQSTabTwinCut_VdwLJ_F_4xn,
95 nbnxn_kernel_ElecQSTabTwinCut_VdwLJFSw_F_4xn,
96 nbnxn_kernel_ElecQSTabTwinCut_VdwLJPSw_F_4xn,
99 nbnxn_kernel_ElecEw_VdwLJCombGeom_F_4xn,
100 nbnxn_kernel_ElecEw_VdwLJCombLB_F_4xn,
101 nbnxn_kernel_ElecEw_VdwLJ_F_4xn,
102 nbnxn_kernel_ElecEw_VdwLJFSw_F_4xn,
103 nbnxn_kernel_ElecEw_VdwLJPSw_F_4xn,
106 nbnxn_kernel_ElecEwTwinCut_VdwLJCombGeom_F_4xn,
107 nbnxn_kernel_ElecEwTwinCut_VdwLJCombLB_F_4xn,
108 nbnxn_kernel_ElecEwTwinCut_VdwLJ_F_4xn,
109 nbnxn_kernel_ElecEwTwinCut_VdwLJFSw_F_4xn,
110 nbnxn_kernel_ElecEwTwinCut_VdwLJPSw_F_4xn,
114 static p_nbk_func_ener p_nbk_ener[coultNR][ljcrNR+2] =
117 nbnxn_kernel_ElecRF_VdwLJCombGeom_VF_4xn,
118 nbnxn_kernel_ElecRF_VdwLJCombLB_VF_4xn,
119 nbnxn_kernel_ElecRF_VdwLJ_VF_4xn,
120 nbnxn_kernel_ElecRF_VdwLJFSw_VF_4xn,
121 nbnxn_kernel_ElecRF_VdwLJPSw_VF_4xn,
124 nbnxn_kernel_ElecQSTab_VdwLJCombGeom_VF_4xn,
125 nbnxn_kernel_ElecQSTab_VdwLJCombLB_VF_4xn,
126 nbnxn_kernel_ElecQSTab_VdwLJ_VF_4xn,
127 nbnxn_kernel_ElecQSTab_VdwLJFSw_VF_4xn,
128 nbnxn_kernel_ElecQSTab_VdwLJPSw_VF_4xn,
131 nbnxn_kernel_ElecQSTabTwinCut_VdwLJCombGeom_VF_4xn,
132 nbnxn_kernel_ElecQSTabTwinCut_VdwLJCombLB_VF_4xn,
133 nbnxn_kernel_ElecQSTabTwinCut_VdwLJ_VF_4xn,
134 nbnxn_kernel_ElecQSTabTwinCut_VdwLJFSw_VF_4xn,
135 nbnxn_kernel_ElecQSTabTwinCut_VdwLJPSw_VF_4xn,
138 nbnxn_kernel_ElecEw_VdwLJCombGeom_VF_4xn,
139 nbnxn_kernel_ElecEw_VdwLJCombLB_VF_4xn,
140 nbnxn_kernel_ElecEw_VdwLJ_VF_4xn,
141 nbnxn_kernel_ElecEw_VdwLJFSw_VF_4xn,
142 nbnxn_kernel_ElecEw_VdwLJPSw_VF_4xn,
145 nbnxn_kernel_ElecEwTwinCut_VdwLJCombGeom_VF_4xn,
146 nbnxn_kernel_ElecEwTwinCut_VdwLJCombLB_VF_4xn,
147 nbnxn_kernel_ElecEwTwinCut_VdwLJ_VF_4xn,
148 nbnxn_kernel_ElecEwTwinCut_VdwLJFSw_VF_4xn,
149 nbnxn_kernel_ElecEwTwinCut_VdwLJPSw_VF_4xn,
153 static p_nbk_func_ener p_nbk_energrp[coultNR][ljcrNR+2] =
156 nbnxn_kernel_ElecRF_VdwLJCombGeom_VgrpF_4xn,
157 nbnxn_kernel_ElecRF_VdwLJCombLB_VgrpF_4xn,
158 nbnxn_kernel_ElecRF_VdwLJ_VgrpF_4xn,
159 nbnxn_kernel_ElecRF_VdwLJFSw_VgrpF_4xn,
160 nbnxn_kernel_ElecRF_VdwLJPSw_VgrpF_4xn,
163 nbnxn_kernel_ElecQSTab_VdwLJCombGeom_VgrpF_4xn,
164 nbnxn_kernel_ElecQSTab_VdwLJCombLB_VgrpF_4xn,
165 nbnxn_kernel_ElecQSTab_VdwLJ_VgrpF_4xn,
166 nbnxn_kernel_ElecQSTab_VdwLJFSw_VgrpF_4xn,
167 nbnxn_kernel_ElecQSTab_VdwLJPSw_VgrpF_4xn,
170 nbnxn_kernel_ElecQSTabTwinCut_VdwLJCombGeom_VgrpF_4xn,
171 nbnxn_kernel_ElecQSTabTwinCut_VdwLJCombLB_VgrpF_4xn,
172 nbnxn_kernel_ElecQSTabTwinCut_VdwLJ_VgrpF_4xn,
173 nbnxn_kernel_ElecQSTabTwinCut_VdwLJFSw_VgrpF_4xn,
174 nbnxn_kernel_ElecQSTabTwinCut_VdwLJPSw_VgrpF_4xn,
177 nbnxn_kernel_ElecEw_VdwLJCombGeom_VgrpF_4xn,
178 nbnxn_kernel_ElecEw_VdwLJCombLB_VgrpF_4xn,
179 nbnxn_kernel_ElecEw_VdwLJ_VgrpF_4xn,
180 nbnxn_kernel_ElecEw_VdwLJFSw_VgrpF_4xn,
181 nbnxn_kernel_ElecEw_VdwLJPSw_VgrpF_4xn,
184 nbnxn_kernel_ElecEwTwinCut_VdwLJCombGeom_VgrpF_4xn,
185 nbnxn_kernel_ElecEwTwinCut_VdwLJCombLB_VgrpF_4xn,
186 nbnxn_kernel_ElecEwTwinCut_VdwLJ_VgrpF_4xn,
187 nbnxn_kernel_ElecEwTwinCut_VdwLJFSw_VgrpF_4xn,
188 nbnxn_kernel_ElecEwTwinCut_VdwLJPSw_VgrpF_4xn,
194 reduce_group_energies(int ng, int ng_2log,
195 const real *VSvdw, const real *VSc,
196 real *Vvdw, real *Vc)
198 const int unrollj = GMX_SIMD_REAL_WIDTH/GMX_SIMD_J_UNROLL_SIZE;
199 const int unrollj_half = unrollj/2;
200 int ng_p2, i, j, j0, j1, c, s;
202 ng_p2 = (1<<ng_2log);
204 /* The size of the x86 SIMD energy group buffer array is:
205 * ng*ng*ng_p2*unrollj_half*simd_width
207 for (i = 0; i < ng; i++)
209 for (j = 0; j < ng; j++)
215 for (j1 = 0; j1 < ng; j1++)
217 for (j0 = 0; j0 < ng; j0++)
219 c = ((i*ng + j1)*ng_p2 + j0)*unrollj_half*unrollj;
220 for (s = 0; s < unrollj_half; s++)
222 Vvdw[i*ng+j0] += VSvdw[c+0];
223 Vvdw[i*ng+j1] += VSvdw[c+1];
224 Vc [i*ng+j0] += VSc [c+0];
225 Vc [i*ng+j1] += VSc [c+1];
233 #else /* GMX_NBNXN_SIMD_4XN */
235 #include "gmx_fatal.h"
237 #endif /* GMX_NBNXN_SIMD_4XN */
240 nbnxn_kernel_simd_4xn(nbnxn_pairlist_set_t gmx_unused *nbl_list,
241 const nbnxn_atomdata_t gmx_unused *nbat,
242 const interaction_const_t gmx_unused *ic,
243 int gmx_unused ewald_excl,
244 rvec gmx_unused *shift_vec,
245 int gmx_unused force_flags,
246 int gmx_unused clearF,
247 real gmx_unused *fshift,
249 real gmx_unused *Vvdw)
250 #ifdef GMX_NBNXN_SIMD_4XN
253 nbnxn_pairlist_t **nbl;
254 int coult, ljtreatment = 0;
257 nnbl = nbl_list->nnbl;
260 if (EEL_RF(ic->eeltype) || ic->eeltype == eelCUT)
266 if (ewald_excl == ewaldexclTable)
268 if (ic->rcoulomb == ic->rvdw)
274 coult = coultTAB_TWIN;
279 if (ic->rcoulomb == ic->rvdw)
285 coult = coultEWALD_TWIN;
290 switch (ic->vdw_modifier)
293 case eintmodPOTSHIFT:
294 ljtreatment = nbat->comb_rule;
296 /* Switch functions follow after cut-off combination rule kernels */
297 case eintmodFORCESWITCH:
298 ljtreatment = ljcrNR;
300 case eintmodPOTSWITCH:
301 ljtreatment = ljcrNR + 1;
304 gmx_incons("Unsupported VdW interaction modifier");
307 #pragma omp parallel for schedule(static) num_threads(gmx_omp_nthreads_get(emntNonbonded))
308 for (nb = 0; nb < nnbl; nb++)
310 nbnxn_atomdata_output_t *out;
313 out = &nbat->out[nb];
315 if (clearF == enbvClearFYes)
317 clear_f(nbat, nb, out->f);
320 if ((force_flags & GMX_FORCE_VIRIAL) && nnbl == 1)
326 fshift_p = out->fshift;
328 if (clearF == enbvClearFYes)
330 clear_fshift(fshift_p);
334 if (!(force_flags & GMX_FORCE_ENERGY))
336 /* Don't calculate energies */
337 p_nbk_noener[coult][ljtreatment](nbl[nb], nbat,
343 else if (out->nV == 1)
345 /* No energy groups */
349 p_nbk_ener[coult][ljtreatment](nbl[nb], nbat,
359 /* Calculate energy group contributions */
362 for (i = 0; i < out->nVS; i++)
366 for (i = 0; i < out->nVS; i++)
371 p_nbk_energrp[coult][ljtreatment](nbl[nb], nbat,
379 reduce_group_energies(nbat->nenergrp, nbat->neg_2log,
380 out->VSvdw, out->VSc,
385 if (force_flags & GMX_FORCE_ENERGY)
387 reduce_energies_over_lists(nbat, nnbl, Vvdw, Vc);
392 gmx_incons("nbnxn_kernel_simd_4xn called when such kernels "
393 " are not enabled.");
396 #undef GMX_SIMD_J_UNROLL_SIZE