File: | gromacs/mdlib/sim_util.c |
Location: | line 842, column 5 |
Description: | Value stored to 'cg0' is never read |
1 | /* |
2 | * This file is part of the GROMACS molecular simulation package. |
3 | * |
4 | * Copyright (c) 1991-2000, University of Groningen, The Netherlands. |
5 | * Copyright (c) 2001-2004, The GROMACS development team. |
6 | * Copyright (c) 2013,2014, by the GROMACS development team, led by |
7 | * Mark Abraham, David van der Spoel, Berk Hess, and Erik Lindahl, |
8 | * and including many others, as listed in the AUTHORS file in the |
9 | * top-level source directory and at http://www.gromacs.org. |
10 | * |
11 | * GROMACS is free software; you can redistribute it and/or |
12 | * modify it under the terms of the GNU Lesser General Public License |
13 | * as published by the Free Software Foundation; either version 2.1 |
14 | * of the License, or (at your option) any later version. |
15 | * |
16 | * GROMACS is distributed in the hope that it will be useful, |
17 | * but WITHOUT ANY WARRANTY; without even the implied warranty of |
18 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU |
19 | * Lesser General Public License for more details. |
20 | * |
21 | * You should have received a copy of the GNU Lesser General Public |
22 | * License along with GROMACS; if not, see |
23 | * http://www.gnu.org/licenses, or write to the Free Software Foundation, |
24 | * Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA. |
25 | * |
26 | * If you want to redistribute modifications to GROMACS, please |
27 | * consider that scientific software is very special. Version |
28 | * control is crucial - bugs must be traceable. We will be happy to |
29 | * consider code for inclusion in the official distribution, but |
30 | * derived work must not be called official GROMACS. Details are found |
31 | * in the README & COPYING files - if they are missing, get the |
32 | * official version at http://www.gromacs.org. |
33 | * |
34 | * To help us fund GROMACS development, we humbly ask that you cite |
35 | * the research papers on the package. Check out http://www.gromacs.org. |
36 | */ |
37 | #ifdef HAVE_CONFIG_H1 |
38 | #include <config.h> |
39 | #endif |
40 | |
41 | #include <assert.h> |
42 | #include <math.h> |
43 | #include <stdio.h> |
44 | #include <string.h> |
45 | #ifdef HAVE_SYS_TIME_H |
46 | #include <sys/time.h> |
47 | #endif |
48 | |
49 | #include "typedefs.h" |
50 | #include "gromacs/utility/cstringutil.h" |
51 | #include "gromacs/utility/smalloc.h" |
52 | #include "names.h" |
53 | #include "txtdump.h" |
54 | #include "pbc.h" |
55 | #include "chargegroup.h" |
56 | #include "gromacs/math/vec.h" |
57 | #include "nrnb.h" |
58 | #include "mshift.h" |
59 | #include "mdrun.h" |
60 | #include "sim_util.h" |
61 | #include "update.h" |
62 | #include "physics.h" |
63 | #include "mdatoms.h" |
64 | #include "force.h" |
65 | #include "bondf.h" |
66 | #include "pme.h" |
67 | #include "disre.h" |
68 | #include "orires.h" |
69 | #include "network.h" |
70 | #include "calcmu.h" |
71 | #include "constr.h" |
72 | #include "copyrite.h" |
73 | #include "domdec.h" |
74 | #include "genborn.h" |
75 | #include "nbnxn_atomdata.h" |
76 | #include "nbnxn_search.h" |
77 | #include "nbnxn_kernels/nbnxn_kernel_ref.h" |
78 | #include "nbnxn_kernels/simd_4xn/nbnxn_kernel_simd_4xn.h" |
79 | #include "nbnxn_kernels/simd_2xnn/nbnxn_kernel_simd_2xnn.h" |
80 | #include "nbnxn_kernels/nbnxn_kernel_gpu_ref.h" |
81 | #include "nonbonded.h" |
82 | #include "../gmxlib/nonbonded/nb_kernel.h" |
83 | #include "../gmxlib/nonbonded/nb_free_energy.h" |
84 | |
85 | #include "gromacs/timing/wallcycle.h" |
86 | #include "gromacs/timing/walltime_accounting.h" |
87 | #include "gromacs/utility/gmxmpi.h" |
88 | #include "gromacs/essentialdynamics/edsam.h" |
89 | #include "gromacs/pulling/pull.h" |
90 | #include "gromacs/pulling/pull_rotation.h" |
91 | #include "gromacs/imd/imd.h" |
92 | #include "adress.h" |
93 | #include "qmmm.h" |
94 | |
95 | #include "gmx_omp_nthreads.h" |
96 | |
97 | #include "nbnxn_cuda_data_mgmt.h" |
98 | #include "nbnxn_cuda/nbnxn_cuda.h" |
99 | |
100 | void print_time(FILE *out, |
101 | gmx_walltime_accounting_t walltime_accounting, |
102 | gmx_int64_t step, |
103 | t_inputrec *ir, |
104 | t_commrec gmx_unused__attribute__ ((unused)) *cr) |
105 | { |
106 | time_t finish; |
107 | char timebuf[STRLEN4096]; |
108 | double dt, elapsed_seconds, time_per_step; |
109 | char buf[48]; |
110 | |
111 | #ifndef GMX_THREAD_MPI |
112 | if (!PAR(cr)((cr)->nnodes > 1)) |
113 | #endif |
114 | { |
115 | fprintf(out, "\r"); |
116 | } |
117 | fprintf(out, "step %s", gmx_step_str(step, buf)); |
118 | if ((step >= ir->nstlist)) |
119 | { |
120 | double seconds_since_epoch = gmx_gettime(); |
121 | elapsed_seconds = seconds_since_epoch - walltime_accounting_get_start_time_stamp(walltime_accounting); |
122 | time_per_step = elapsed_seconds/(step - ir->init_step + 1); |
123 | dt = (ir->nsteps + ir->init_step - step) * time_per_step; |
124 | |
125 | if (ir->nsteps >= 0) |
126 | { |
127 | if (dt >= 300) |
128 | { |
129 | finish = (time_t) (seconds_since_epoch + dt); |
130 | gmx_ctime_r(&finish, timebuf, STRLEN4096); |
131 | sprintf(buf, "%s", timebuf); |
132 | buf[strlen(buf)-1] = '\0'; |
133 | fprintf(out, ", will finish %s", buf); |
134 | } |
135 | else |
136 | { |
137 | fprintf(out, ", remaining wall clock time: %5d s ", (int)dt); |
138 | } |
139 | } |
140 | else |
141 | { |
142 | fprintf(out, " performance: %.1f ns/day ", |
143 | ir->delta_t/1000*24*60*60/time_per_step); |
144 | } |
145 | } |
146 | #ifndef GMX_THREAD_MPI |
147 | if (PAR(cr)((cr)->nnodes > 1)) |
148 | { |
149 | fprintf(out, "\n"); |
150 | } |
151 | #endif |
152 | |
153 | fflush(out); |
154 | } |
155 | |
156 | void print_date_and_time(FILE *fplog, int nodeid, const char *title, |
157 | double the_time) |
158 | { |
159 | char time_string[STRLEN4096]; |
160 | |
161 | if (!fplog) |
162 | { |
163 | return; |
164 | } |
165 | |
166 | { |
167 | int i; |
168 | char timebuf[STRLEN4096]; |
169 | time_t temp_time = (time_t) the_time; |
170 | |
171 | gmx_ctime_r(&temp_time, timebuf, STRLEN4096); |
172 | for (i = 0; timebuf[i] >= ' '; i++) |
173 | { |
174 | time_string[i] = timebuf[i]; |
175 | } |
176 | time_string[i] = '\0'; |
177 | } |
178 | |
179 | fprintf(fplog, "%s on node %d %s\n", title, nodeid, time_string); |
180 | } |
181 | |
182 | void print_start(FILE *fplog, t_commrec *cr, |
183 | gmx_walltime_accounting_t walltime_accounting, |
184 | const char *name) |
185 | { |
186 | char buf[STRLEN4096]; |
187 | |
188 | sprintf(buf, "Started %s", name); |
189 | print_date_and_time(fplog, cr->nodeid, buf, |
190 | walltime_accounting_get_start_time_stamp(walltime_accounting)); |
191 | } |
192 | |
193 | static void sum_forces(int start, int end, rvec f[], rvec flr[]) |
194 | { |
195 | int i; |
196 | |
197 | if (gmx_debug_at) |
198 | { |
199 | pr_rvecs(debug, 0, "fsr", f+start, end-start); |
200 | pr_rvecs(debug, 0, "flr", flr+start, end-start); |
201 | } |
202 | for (i = start; (i < end); i++) |
203 | { |
204 | rvec_inc(f[i], flr[i]); |
205 | } |
206 | } |
207 | |
208 | /* |
209 | * calc_f_el calculates forces due to an electric field. |
210 | * |
211 | * force is kJ mol^-1 nm^-1 = e * kJ mol^-1 nm^-1 / e |
212 | * |
213 | * Et[] contains the parameters for the time dependent |
214 | * part of the field (not yet used). |
215 | * Ex[] contains the parameters for |
216 | * the spatial dependent part of the field. You can have cool periodic |
217 | * fields in principle, but only a constant field is supported |
218 | * now. |
219 | * The function should return the energy due to the electric field |
220 | * (if any) but for now returns 0. |
221 | * |
222 | * WARNING: |
223 | * There can be problems with the virial. |
224 | * Since the field is not self-consistent this is unavoidable. |
225 | * For neutral molecules the virial is correct within this approximation. |
226 | * For neutral systems with many charged molecules the error is small. |
227 | * But for systems with a net charge or a few charged molecules |
228 | * the error can be significant when the field is high. |
229 | * Solution: implement a self-consitent electric field into PME. |
230 | */ |
231 | static void calc_f_el(FILE *fp, int start, int homenr, |
232 | real charge[], rvec f[], |
233 | t_cosines Ex[], t_cosines Et[], double t) |
234 | { |
235 | rvec Ext; |
236 | real t0; |
237 | int i, m; |
238 | |
239 | for (m = 0; (m < DIM3); m++) |
240 | { |
241 | if (Et[m].n > 0) |
242 | { |
243 | if (Et[m].n == 3) |
244 | { |
245 | t0 = Et[m].a[1]; |
246 | Ext[m] = cos(Et[m].a[0]*(t-t0))*exp(-sqr(t-t0)/(2.0*sqr(Et[m].a[2]))); |
247 | } |
248 | else |
249 | { |
250 | Ext[m] = cos(Et[m].a[0]*t); |
251 | } |
252 | } |
253 | else |
254 | { |
255 | Ext[m] = 1.0; |
256 | } |
257 | if (Ex[m].n > 0) |
258 | { |
259 | /* Convert the field strength from V/nm to MD-units */ |
260 | Ext[m] *= Ex[m].a[0]*FIELDFAC(((1.60217733e-19)*(6.0221367e23))/(1e3)); |
261 | for (i = start; (i < start+homenr); i++) |
262 | { |
263 | f[i][m] += charge[i]*Ext[m]; |
264 | } |
265 | } |
266 | else |
267 | { |
268 | Ext[m] = 0; |
269 | } |
270 | } |
271 | if (fp != NULL((void*)0)) |
272 | { |
273 | fprintf(fp, "%10g %10g %10g %10g #FIELD\n", t, |
274 | Ext[XX0]/FIELDFAC(((1.60217733e-19)*(6.0221367e23))/(1e3)), Ext[YY1]/FIELDFAC(((1.60217733e-19)*(6.0221367e23))/(1e3)), Ext[ZZ2]/FIELDFAC(((1.60217733e-19)*(6.0221367e23))/(1e3))); |
275 | } |
276 | } |
277 | |
278 | static void calc_virial(int start, int homenr, rvec x[], rvec f[], |
279 | tensor vir_part, t_graph *graph, matrix box, |
280 | t_nrnb *nrnb, const t_forcerec *fr, int ePBC) |
281 | { |
282 | int i, j; |
283 | tensor virtest; |
284 | |
285 | /* The short-range virial from surrounding boxes */ |
286 | clear_mat(vir_part); |
287 | calc_vir(SHIFTS((2*1 +1)*(2*1 +1)*(2*2 +1)), fr->shift_vec, fr->fshift, vir_part, ePBC == epbcSCREW, box); |
288 | inc_nrnb(nrnb, eNR_VIRIAL, SHIFTS)(nrnb)->n[eNR_VIRIAL] += ((2*1 +1)*(2*1 +1)*(2*2 +1)); |
289 | |
290 | /* Calculate partial virial, for local atoms only, based on short range. |
291 | * Total virial is computed in global_stat, called from do_md |
292 | */ |
293 | f_calc_vir(start, start+homenr, x, f, vir_part, graph, box); |
294 | inc_nrnb(nrnb, eNR_VIRIAL, homenr)(nrnb)->n[eNR_VIRIAL] += homenr; |
295 | |
296 | /* Add position restraint contribution */ |
297 | for (i = 0; i < DIM3; i++) |
298 | { |
299 | vir_part[i][i] += fr->vir_diag_posres[i]; |
300 | } |
301 | |
302 | /* Add wall contribution */ |
303 | for (i = 0; i < DIM3; i++) |
304 | { |
305 | vir_part[i][ZZ2] += fr->vir_wall_z[i]; |
306 | } |
307 | |
308 | if (debug) |
309 | { |
310 | pr_rvecs(debug, 0, "vir_part", vir_part, DIM3); |
311 | } |
312 | } |
313 | |
314 | static void posres_wrapper(FILE *fplog, |
315 | int flags, |
316 | gmx_bool bSepDVDL, |
317 | t_inputrec *ir, |
318 | t_nrnb *nrnb, |
319 | gmx_localtop_t *top, |
320 | matrix box, rvec x[], |
321 | gmx_enerdata_t *enerd, |
322 | real *lambda, |
323 | t_forcerec *fr) |
324 | { |
325 | t_pbc pbc; |
326 | real v, dvdl; |
327 | int i; |
328 | |
329 | /* Position restraints always require full pbc */ |
330 | set_pbc(&pbc, ir->ePBC, box); |
331 | dvdl = 0; |
332 | v = posres(top->idef.il[F_POSRES].nr, top->idef.il[F_POSRES].iatoms, |
333 | top->idef.iparams_posres, |
334 | (const rvec*)x, fr->f_novirsum, fr->vir_diag_posres, |
335 | ir->ePBC == epbcNONE ? NULL((void*)0) : &pbc, |
336 | lambda[efptRESTRAINT], &dvdl, |
337 | fr->rc_scaling, fr->ePBC, fr->posres_com, fr->posres_comB); |
338 | if (bSepDVDL) |
339 | { |
340 | gmx_print_sepdvdl(fplog, interaction_function[F_POSRES].longname, v, dvdl); |
341 | } |
342 | enerd->term[F_POSRES] += v; |
343 | /* If just the force constant changes, the FEP term is linear, |
344 | * but if k changes, it is not. |
345 | */ |
346 | enerd->dvdl_nonlin[efptRESTRAINT] += dvdl; |
347 | inc_nrnb(nrnb, eNR_POSRES, top->idef.il[F_POSRES].nr/2)(nrnb)->n[eNR_POSRES] += top->idef.il[F_POSRES].nr/2; |
348 | |
349 | if ((ir->fepvals->n_lambda > 0) && (flags & GMX_FORCE_DHDL(1<<10))) |
350 | { |
351 | for (i = 0; i < enerd->n_lambda; i++) |
352 | { |
353 | real dvdl_dum, lambda_dum; |
354 | |
355 | lambda_dum = (i == 0 ? lambda[efptRESTRAINT] : ir->fepvals->all_lambda[efptRESTRAINT][i-1]); |
356 | v = posres(top->idef.il[F_POSRES].nr, top->idef.il[F_POSRES].iatoms, |
357 | top->idef.iparams_posres, |
358 | (const rvec*)x, NULL((void*)0), NULL((void*)0), |
359 | ir->ePBC == epbcNONE ? NULL((void*)0) : &pbc, lambda_dum, &dvdl, |
360 | fr->rc_scaling, fr->ePBC, fr->posres_com, fr->posres_comB); |
361 | enerd->enerpart_lambda[i] += v; |
362 | } |
363 | } |
364 | } |
365 | |
366 | static void fbposres_wrapper(t_inputrec *ir, |
367 | t_nrnb *nrnb, |
368 | gmx_localtop_t *top, |
369 | matrix box, rvec x[], |
370 | gmx_enerdata_t *enerd, |
371 | t_forcerec *fr) |
372 | { |
373 | t_pbc pbc; |
374 | real v; |
375 | |
376 | /* Flat-bottomed position restraints always require full pbc */ |
377 | set_pbc(&pbc, ir->ePBC, box); |
378 | v = fbposres(top->idef.il[F_FBPOSRES].nr, top->idef.il[F_FBPOSRES].iatoms, |
379 | top->idef.iparams_fbposres, |
380 | (const rvec*)x, fr->f_novirsum, fr->vir_diag_posres, |
381 | ir->ePBC == epbcNONE ? NULL((void*)0) : &pbc, |
382 | fr->rc_scaling, fr->ePBC, fr->posres_com); |
383 | enerd->term[F_FBPOSRES] += v; |
384 | inc_nrnb(nrnb, eNR_FBPOSRES, top->idef.il[F_FBPOSRES].nr/2)(nrnb)->n[eNR_FBPOSRES] += top->idef.il[F_FBPOSRES].nr/ 2; |
385 | } |
386 | |
387 | static void pull_potential_wrapper(FILE *fplog, |
388 | gmx_bool bSepDVDL, |
389 | t_commrec *cr, |
390 | t_inputrec *ir, |
391 | matrix box, rvec x[], |
392 | rvec f[], |
393 | tensor vir_force, |
394 | t_mdatoms *mdatoms, |
395 | gmx_enerdata_t *enerd, |
396 | real *lambda, |
397 | double t) |
398 | { |
399 | t_pbc pbc; |
400 | real dvdl; |
401 | |
402 | /* Calculate the center of mass forces, this requires communication, |
403 | * which is why pull_potential is called close to other communication. |
404 | * The virial contribution is calculated directly, |
405 | * which is why we call pull_potential after calc_virial. |
406 | */ |
407 | set_pbc(&pbc, ir->ePBC, box); |
408 | dvdl = 0; |
409 | enerd->term[F_COM_PULL] += |
410 | pull_potential(ir->ePull, ir->pull, mdatoms, &pbc, |
411 | cr, t, lambda[efptRESTRAINT], x, f, vir_force, &dvdl); |
412 | if (bSepDVDL) |
413 | { |
414 | gmx_print_sepdvdl(fplog, "Com pull", enerd->term[F_COM_PULL], dvdl); |
415 | } |
416 | enerd->dvdl_lin[efptRESTRAINT] += dvdl; |
417 | } |
418 | |
419 | static void pme_receive_force_ener(FILE *fplog, |
420 | gmx_bool bSepDVDL, |
421 | t_commrec *cr, |
422 | gmx_wallcycle_t wcycle, |
423 | gmx_enerdata_t *enerd, |
424 | t_forcerec *fr) |
425 | { |
426 | real e_q, e_lj, v, dvdl_q, dvdl_lj; |
427 | float cycles_ppdpme, cycles_seppme; |
428 | |
429 | cycles_ppdpme = wallcycle_stop(wcycle, ewcPPDURINGPME); |
430 | dd_cycles_add(cr->dd, cycles_ppdpme, ddCyclPPduringPME); |
431 | |
432 | /* In case of node-splitting, the PP nodes receive the long-range |
433 | * forces, virial and energy from the PME nodes here. |
434 | */ |
435 | wallcycle_start(wcycle, ewcPP_PMEWAITRECVF); |
436 | dvdl_q = 0; |
437 | dvdl_lj = 0; |
438 | gmx_pme_receive_f(cr, fr->f_novirsum, fr->vir_el_recip, &e_q, |
439 | fr->vir_lj_recip, &e_lj, &dvdl_q, &dvdl_lj, |
440 | &cycles_seppme); |
441 | if (bSepDVDL) |
442 | { |
443 | gmx_print_sepdvdl(fplog, "Electrostatic PME mesh", e_q, dvdl_q); |
444 | gmx_print_sepdvdl(fplog, "Lennard-Jones PME mesh", e_lj, dvdl_lj); |
445 | } |
446 | enerd->term[F_COUL_RECIP] += e_q; |
447 | enerd->term[F_LJ_RECIP] += e_lj; |
448 | enerd->dvdl_lin[efptCOUL] += dvdl_q; |
449 | enerd->dvdl_lin[efptVDW] += dvdl_lj; |
450 | |
451 | if (wcycle) |
452 | { |
453 | dd_cycles_add(cr->dd, cycles_seppme, ddCyclPME); |
454 | } |
455 | wallcycle_stop(wcycle, ewcPP_PMEWAITRECVF); |
456 | } |
457 | |
458 | static void print_large_forces(FILE *fp, t_mdatoms *md, t_commrec *cr, |
459 | gmx_int64_t step, real pforce, rvec *x, rvec *f) |
460 | { |
461 | int i; |
462 | real pf2, fn2; |
463 | char buf[STEPSTRSIZE22]; |
464 | |
465 | pf2 = sqr(pforce); |
466 | for (i = 0; i < md->homenr; i++) |
467 | { |
468 | fn2 = norm2(f[i]); |
469 | /* We also catch NAN, if the compiler does not optimize this away. */ |
470 | if (fn2 >= pf2 || fn2 != fn2) |
471 | { |
472 | fprintf(fp, "step %s atom %6d x %8.3f %8.3f %8.3f force %12.5e\n", |
473 | gmx_step_str(step, buf), |
474 | ddglatnr(cr->dd, i), x[i][XX0], x[i][YY1], x[i][ZZ2], sqrt(fn2)); |
475 | } |
476 | } |
477 | } |
478 | |
479 | static void post_process_forces(t_commrec *cr, |
480 | gmx_int64_t step, |
481 | t_nrnb *nrnb, gmx_wallcycle_t wcycle, |
482 | gmx_localtop_t *top, |
483 | matrix box, rvec x[], |
484 | rvec f[], |
485 | tensor vir_force, |
486 | t_mdatoms *mdatoms, |
487 | t_graph *graph, |
488 | t_forcerec *fr, gmx_vsite_t *vsite, |
489 | int flags) |
490 | { |
491 | if (fr->bF_NoVirSum) |
492 | { |
493 | if (vsite) |
494 | { |
495 | /* Spread the mesh force on virtual sites to the other particles... |
496 | * This is parallellized. MPI communication is performed |
497 | * if the constructing atoms aren't local. |
498 | */ |
499 | wallcycle_start(wcycle, ewcVSITESPREAD); |
500 | spread_vsite_f(vsite, x, fr->f_novirsum, NULL((void*)0), |
501 | (flags & GMX_FORCE_VIRIAL(1<<8)), fr->vir_el_recip, |
502 | nrnb, |
503 | &top->idef, fr->ePBC, fr->bMolPBC, graph, box, cr); |
504 | wallcycle_stop(wcycle, ewcVSITESPREAD); |
505 | } |
506 | if (flags & GMX_FORCE_VIRIAL(1<<8)) |
507 | { |
508 | /* Now add the forces, this is local */ |
509 | if (fr->bDomDec) |
510 | { |
511 | sum_forces(0, fr->f_novirsum_n, f, fr->f_novirsum); |
512 | } |
513 | else |
514 | { |
515 | sum_forces(0, mdatoms->homenr, |
516 | f, fr->f_novirsum); |
517 | } |
518 | if (EEL_FULL(fr->eeltype)((((fr->eeltype) == eelPME || (fr->eeltype) == eelPMESWITCH || (fr->eeltype) == eelPMEUSER || (fr->eeltype) == eelPMEUSERSWITCH || (fr->eeltype) == eelP3M_AD) || (fr->eeltype) == eelEWALD ) || (fr->eeltype) == eelPOISSON)) |
519 | { |
520 | /* Add the mesh contribution to the virial */ |
521 | m_add(vir_force, fr->vir_el_recip, vir_force); |
522 | } |
523 | if (EVDW_PME(fr->vdwtype)((fr->vdwtype) == evdwPME)) |
524 | { |
525 | /* Add the mesh contribution to the virial */ |
526 | m_add(vir_force, fr->vir_lj_recip, vir_force); |
527 | } |
528 | if (debug) |
529 | { |
530 | pr_rvecs(debug, 0, "vir_force", vir_force, DIM3); |
531 | } |
532 | } |
533 | } |
534 | |
535 | if (fr->print_force >= 0) |
536 | { |
537 | print_large_forces(stderrstderr, mdatoms, cr, step, fr->print_force, x, f); |
538 | } |
539 | } |
540 | |
541 | static void do_nb_verlet(t_forcerec *fr, |
542 | interaction_const_t *ic, |
543 | gmx_enerdata_t *enerd, |
544 | int flags, int ilocality, |
545 | int clearF, |
546 | t_nrnb *nrnb, |
547 | gmx_wallcycle_t wcycle) |
548 | { |
549 | int nnbl, kernel_type, enr_nbnxn_kernel_ljc, enr_nbnxn_kernel_lj; |
550 | char *env; |
551 | nonbonded_verlet_group_t *nbvg; |
552 | gmx_bool bCUDA; |
553 | |
554 | if (!(flags & GMX_FORCE_NONBONDED(1<<6))) |
555 | { |
556 | /* skip non-bonded calculation */ |
557 | return; |
558 | } |
559 | |
560 | nbvg = &fr->nbv->grp[ilocality]; |
561 | |
562 | /* CUDA kernel launch overhead is already timed separately */ |
563 | if (fr->cutoff_scheme != ecutsVERLET) |
564 | { |
565 | gmx_incons("Invalid cut-off scheme passed!")_gmx_error("incons", "Invalid cut-off scheme passed!", "/home/alexxy/Develop/gromacs/src/gromacs/mdlib/sim_util.c" , 565); |
566 | } |
567 | |
568 | bCUDA = (nbvg->kernel_type == nbnxnk8x8x8_CUDA); |
569 | |
570 | if (!bCUDA) |
571 | { |
572 | wallcycle_sub_start(wcycle, ewcsNONBONDED); |
573 | } |
574 | switch (nbvg->kernel_type) |
575 | { |
576 | case nbnxnk4x4_PlainC: |
577 | nbnxn_kernel_ref(&nbvg->nbl_lists, |
578 | nbvg->nbat, ic, |
579 | fr->shift_vec, |
580 | flags, |
581 | clearF, |
582 | fr->fshift[0], |
583 | enerd->grpp.ener[egCOULSR], |
584 | fr->bBHAM ? |
585 | enerd->grpp.ener[egBHAMSR] : |
586 | enerd->grpp.ener[egLJSR]); |
587 | break; |
588 | |
589 | case nbnxnk4xN_SIMD_4xN: |
590 | nbnxn_kernel_simd_4xn(&nbvg->nbl_lists, |
591 | nbvg->nbat, ic, |
592 | nbvg->ewald_excl, |
593 | fr->shift_vec, |
594 | flags, |
595 | clearF, |
596 | fr->fshift[0], |
597 | enerd->grpp.ener[egCOULSR], |
598 | fr->bBHAM ? |
599 | enerd->grpp.ener[egBHAMSR] : |
600 | enerd->grpp.ener[egLJSR]); |
601 | break; |
602 | case nbnxnk4xN_SIMD_2xNN: |
603 | nbnxn_kernel_simd_2xnn(&nbvg->nbl_lists, |
604 | nbvg->nbat, ic, |
605 | nbvg->ewald_excl, |
606 | fr->shift_vec, |
607 | flags, |
608 | clearF, |
609 | fr->fshift[0], |
610 | enerd->grpp.ener[egCOULSR], |
611 | fr->bBHAM ? |
612 | enerd->grpp.ener[egBHAMSR] : |
613 | enerd->grpp.ener[egLJSR]); |
614 | break; |
615 | |
616 | case nbnxnk8x8x8_CUDA: |
617 | nbnxn_cuda_launch_kernel(fr->nbv->cu_nbv, nbvg->nbat, flags, ilocality); |
618 | break; |
619 | |
620 | case nbnxnk8x8x8_PlainC: |
621 | nbnxn_kernel_gpu_ref(nbvg->nbl_lists.nbl[0], |
622 | nbvg->nbat, ic, |
623 | fr->shift_vec, |
624 | flags, |
625 | clearF, |
626 | nbvg->nbat->out[0].f, |
627 | fr->fshift[0], |
628 | enerd->grpp.ener[egCOULSR], |
629 | fr->bBHAM ? |
630 | enerd->grpp.ener[egBHAMSR] : |
631 | enerd->grpp.ener[egLJSR]); |
632 | break; |
633 | |
634 | default: |
635 | gmx_incons("Invalid nonbonded kernel type passed!")_gmx_error("incons", "Invalid nonbonded kernel type passed!", "/home/alexxy/Develop/gromacs/src/gromacs/mdlib/sim_util.c", 635); |
636 | |
637 | } |
638 | if (!bCUDA) |
639 | { |
640 | wallcycle_sub_stop(wcycle, ewcsNONBONDED); |
641 | } |
642 | |
643 | if (EEL_RF(ic->eeltype)((ic->eeltype) == eelRF || (ic->eeltype) == eelGRF || ( ic->eeltype) == eelRF_NEC || (ic->eeltype) == eelRF_ZERO ) || ic->eeltype == eelCUT) |
644 | { |
645 | enr_nbnxn_kernel_ljc = eNR_NBNXN_LJ_RF; |
646 | } |
647 | else if ((!bCUDA && nbvg->ewald_excl == ewaldexclAnalytical) || |
648 | (bCUDA && nbnxn_cuda_is_kernel_ewald_analytical(fr->nbv->cu_nbv))) |
649 | { |
650 | enr_nbnxn_kernel_ljc = eNR_NBNXN_LJ_EWALD; |
651 | } |
652 | else |
653 | { |
654 | enr_nbnxn_kernel_ljc = eNR_NBNXN_LJ_TAB; |
655 | } |
656 | enr_nbnxn_kernel_lj = eNR_NBNXN_LJ; |
657 | if (flags & GMX_FORCE_ENERGY(1<<9)) |
658 | { |
659 | /* In eNR_??? the nbnxn F+E kernels are always the F kernel + 1 */ |
660 | enr_nbnxn_kernel_ljc += 1; |
661 | enr_nbnxn_kernel_lj += 1; |
662 | } |
663 | |
664 | inc_nrnb(nrnb, enr_nbnxn_kernel_ljc,(nrnb)->n[enr_nbnxn_kernel_ljc] += nbvg->nbl_lists.natpair_ljq |
665 | nbvg->nbl_lists.natpair_ljq)(nrnb)->n[enr_nbnxn_kernel_ljc] += nbvg->nbl_lists.natpair_ljq; |
666 | inc_nrnb(nrnb, enr_nbnxn_kernel_lj,(nrnb)->n[enr_nbnxn_kernel_lj] += nbvg->nbl_lists.natpair_lj |
667 | nbvg->nbl_lists.natpair_lj)(nrnb)->n[enr_nbnxn_kernel_lj] += nbvg->nbl_lists.natpair_lj; |
668 | /* The Coulomb-only kernels are offset -eNR_NBNXN_LJ_RF+eNR_NBNXN_RF */ |
669 | inc_nrnb(nrnb, enr_nbnxn_kernel_ljc-eNR_NBNXN_LJ_RF+eNR_NBNXN_RF,(nrnb)->n[enr_nbnxn_kernel_ljc-eNR_NBNXN_LJ_RF+eNR_NBNXN_RF ] += nbvg->nbl_lists.natpair_q |
670 | nbvg->nbl_lists.natpair_q)(nrnb)->n[enr_nbnxn_kernel_ljc-eNR_NBNXN_LJ_RF+eNR_NBNXN_RF ] += nbvg->nbl_lists.natpair_q; |
671 | |
672 | if (ic->vdw_modifier == eintmodFORCESWITCH) |
673 | { |
674 | /* We add up the switch cost separately */ |
675 | inc_nrnb(nrnb, eNR_NBNXN_ADD_LJ_FSW+((flags & GMX_FORCE_ENERGY) ? 1 : 0),(nrnb)->n[eNR_NBNXN_ADD_LJ_FSW+((flags & (1<<9)) ? 1 : 0)] += nbvg->nbl_lists.natpair_ljq + nbvg->nbl_lists .natpair_lj |
676 | nbvg->nbl_lists.natpair_ljq + nbvg->nbl_lists.natpair_lj)(nrnb)->n[eNR_NBNXN_ADD_LJ_FSW+((flags & (1<<9)) ? 1 : 0)] += nbvg->nbl_lists.natpair_ljq + nbvg->nbl_lists .natpair_lj; |
677 | } |
678 | if (ic->vdw_modifier == eintmodPOTSWITCH) |
679 | { |
680 | /* We add up the switch cost separately */ |
681 | inc_nrnb(nrnb, eNR_NBNXN_ADD_LJ_PSW+((flags & GMX_FORCE_ENERGY) ? 1 : 0),(nrnb)->n[eNR_NBNXN_ADD_LJ_PSW+((flags & (1<<9)) ? 1 : 0)] += nbvg->nbl_lists.natpair_ljq + nbvg->nbl_lists .natpair_lj |
682 | nbvg->nbl_lists.natpair_ljq + nbvg->nbl_lists.natpair_lj)(nrnb)->n[eNR_NBNXN_ADD_LJ_PSW+((flags & (1<<9)) ? 1 : 0)] += nbvg->nbl_lists.natpair_ljq + nbvg->nbl_lists .natpair_lj; |
683 | } |
684 | if (ic->vdwtype == evdwPME) |
685 | { |
686 | /* We add up the LJ Ewald cost separately */ |
687 | inc_nrnb(nrnb, eNR_NBNXN_ADD_LJ_EWALD+((flags & GMX_FORCE_ENERGY) ? 1 : 0),(nrnb)->n[eNR_NBNXN_ADD_LJ_EWALD+((flags & (1<<9 )) ? 1 : 0)] += nbvg->nbl_lists.natpair_ljq + nbvg->nbl_lists .natpair_lj |
688 | nbvg->nbl_lists.natpair_ljq + nbvg->nbl_lists.natpair_lj)(nrnb)->n[eNR_NBNXN_ADD_LJ_EWALD+((flags & (1<<9 )) ? 1 : 0)] += nbvg->nbl_lists.natpair_ljq + nbvg->nbl_lists .natpair_lj; |
689 | } |
690 | } |
691 | |
692 | static void do_nb_verlet_fep(nbnxn_pairlist_set_t *nbl_lists, |
693 | t_forcerec *fr, |
694 | rvec x[], |
695 | rvec f[], |
696 | t_mdatoms *mdatoms, |
697 | t_lambda *fepvals, |
698 | real *lambda, |
699 | gmx_enerdata_t *enerd, |
700 | int flags, |
701 | t_nrnb *nrnb, |
702 | gmx_wallcycle_t wcycle) |
703 | { |
704 | int donb_flags; |
705 | nb_kernel_data_t kernel_data; |
706 | real lam_i[efptNR]; |
707 | real dvdl_nb[efptNR]; |
708 | int th; |
709 | int i, j; |
710 | |
711 | donb_flags = 0; |
712 | /* Add short-range interactions */ |
713 | donb_flags |= GMX_NONBONDED_DO_SR(1<<5); |
714 | |
715 | /* Currently all group scheme kernels always calculate (shift-)forces */ |
716 | if (flags & GMX_FORCE_FORCES(1<<7)) |
717 | { |
718 | donb_flags |= GMX_NONBONDED_DO_FORCE(1<<1); |
719 | } |
720 | if (flags & GMX_FORCE_VIRIAL(1<<8)) |
721 | { |
722 | donb_flags |= GMX_NONBONDED_DO_SHIFTFORCE(1<<2); |
723 | } |
724 | if (flags & GMX_FORCE_ENERGY(1<<9)) |
725 | { |
726 | donb_flags |= GMX_NONBONDED_DO_POTENTIAL(1<<4); |
727 | } |
728 | if (flags & GMX_FORCE_DO_LR(1<<11)) |
729 | { |
730 | donb_flags |= GMX_NONBONDED_DO_LR(1<<0); |
731 | } |
732 | |
733 | kernel_data.flags = donb_flags; |
734 | kernel_data.lambda = lambda; |
735 | kernel_data.dvdl = dvdl_nb; |
736 | |
737 | kernel_data.energygrp_elec = enerd->grpp.ener[egCOULSR]; |
738 | kernel_data.energygrp_vdw = enerd->grpp.ener[egLJSR]; |
739 | |
740 | /* reset free energy components */ |
741 | for (i = 0; i < efptNR; i++) |
742 | { |
743 | dvdl_nb[i] = 0; |
744 | } |
745 | |
746 | assert(gmx_omp_nthreads_get(emntNonbonded) == nbl_lists->nnbl)((void) (0)); |
747 | |
748 | wallcycle_sub_start(wcycle, ewcsNONBONDED); |
749 | #pragma omp parallel for schedule(static) num_threads(nbl_lists->nnbl) |
750 | for (th = 0; th < nbl_lists->nnbl; th++) |
751 | { |
752 | gmx_nb_free_energy_kernel(nbl_lists->nbl_fep[th], |
753 | x, f, fr, mdatoms, &kernel_data, nrnb); |
754 | } |
755 | |
756 | if (fepvals->sc_alpha != 0) |
757 | { |
758 | enerd->dvdl_nonlin[efptVDW] += dvdl_nb[efptVDW]; |
759 | enerd->dvdl_nonlin[efptCOUL] += dvdl_nb[efptCOUL]; |
760 | } |
761 | else |
762 | { |
763 | enerd->dvdl_lin[efptVDW] += dvdl_nb[efptVDW]; |
764 | enerd->dvdl_lin[efptCOUL] += dvdl_nb[efptCOUL]; |
765 | } |
766 | |
767 | /* If we do foreign lambda and we have soft-core interactions |
768 | * we have to recalculate the (non-linear) energies contributions. |
769 | */ |
770 | if (fepvals->n_lambda > 0 && (flags & GMX_FORCE_DHDL(1<<10)) && fepvals->sc_alpha != 0) |
771 | { |
772 | kernel_data.flags = (donb_flags & ~(GMX_NONBONDED_DO_FORCE(1<<1) | GMX_NONBONDED_DO_SHIFTFORCE(1<<2))) | GMX_NONBONDED_DO_FOREIGNLAMBDA(1<<3); |
773 | kernel_data.lambda = lam_i; |
774 | kernel_data.energygrp_elec = enerd->foreign_grpp.ener[egCOULSR]; |
775 | kernel_data.energygrp_vdw = enerd->foreign_grpp.ener[egLJSR]; |
776 | /* Note that we add to kernel_data.dvdl, but ignore the result */ |
777 | |
778 | for (i = 0; i < enerd->n_lambda; i++) |
779 | { |
780 | for (j = 0; j < efptNR; j++) |
781 | { |
782 | lam_i[j] = (i == 0 ? lambda[j] : fepvals->all_lambda[j][i-1]); |
783 | } |
784 | reset_foreign_enerdata(enerd); |
785 | #pragma omp parallel for schedule(static) num_threads(nbl_lists->nnbl) |
786 | for (th = 0; th < nbl_lists->nnbl; th++) |
787 | { |
788 | gmx_nb_free_energy_kernel(nbl_lists->nbl_fep[th], |
789 | x, f, fr, mdatoms, &kernel_data, nrnb); |
790 | } |
791 | |
792 | sum_epot(&(enerd->foreign_grpp), enerd->foreign_term); |
793 | enerd->enerpart_lambda[i] += enerd->foreign_term[F_EPOT]; |
794 | } |
795 | } |
796 | |
797 | wallcycle_sub_stop(wcycle, ewcsNONBONDED); |
798 | } |
799 | |
800 | void do_force_cutsVERLET(FILE *fplog, t_commrec *cr, |
801 | t_inputrec *inputrec, |
802 | gmx_int64_t step, t_nrnb *nrnb, gmx_wallcycle_t wcycle, |
803 | gmx_localtop_t *top, |
804 | gmx_groups_t gmx_unused__attribute__ ((unused)) *groups, |
805 | matrix box, rvec x[], history_t *hist, |
806 | rvec f[], |
807 | tensor vir_force, |
808 | t_mdatoms *mdatoms, |
809 | gmx_enerdata_t *enerd, t_fcdata *fcd, |
810 | real *lambda, t_graph *graph, |
811 | t_forcerec *fr, interaction_const_t *ic, |
812 | gmx_vsite_t *vsite, rvec mu_tot, |
813 | double t, FILE *field, gmx_edsam_t ed, |
814 | gmx_bool bBornRadii, |
815 | int flags) |
816 | { |
817 | int cg0, cg1, i, j; |
818 | int start, homenr; |
819 | int nb_kernel_type; |
820 | double mu[2*DIM3]; |
821 | gmx_bool bSepDVDL, bStateChanged, bNS, bFillGrid, bCalcCGCM, bBS; |
822 | gmx_bool bDoLongRange, bDoForces, bSepLRF, bUseGPU, bUseOrEmulGPU; |
823 | gmx_bool bDiffKernels = FALSE0; |
824 | matrix boxs; |
825 | rvec vzero, box_diag; |
826 | real e, v, dvdl; |
827 | float cycles_pme, cycles_force, cycles_wait_gpu; |
828 | nonbonded_verlet_t *nbv; |
829 | |
830 | cycles_force = 0; |
831 | cycles_wait_gpu = 0; |
832 | nbv = fr->nbv; |
833 | nb_kernel_type = fr->nbv->grp[0].kernel_type; |
834 | |
835 | start = 0; |
836 | homenr = mdatoms->homenr; |
837 | |
838 | bSepDVDL = (fr->bSepDVDL && do_per_step(step, inputrec->nstlog)); |
839 | |
840 | clear_mat(vir_force); |
841 | |
842 | cg0 = 0; |
Value stored to 'cg0' is never read | |
843 | if (DOMAINDECOMP(cr)(((cr)->dd != ((void*)0)) && ((cr)->nnodes > 1))) |
844 | { |
845 | cg1 = cr->dd->ncg_tot; |
846 | } |
847 | else |
848 | { |
849 | cg1 = top->cgs.nr; |
850 | } |
851 | if (fr->n_tpi > 0) |
852 | { |
853 | cg1--; |
854 | } |
855 | |
856 | bStateChanged = (flags & GMX_FORCE_STATECHANGED(1<<0)); |
857 | bNS = (flags & GMX_FORCE_NS(1<<2)) && (fr->bAllvsAll == FALSE0); |
858 | bFillGrid = (bNS && bStateChanged); |
859 | bCalcCGCM = (bFillGrid && !DOMAINDECOMP(cr)(((cr)->dd != ((void*)0)) && ((cr)->nnodes > 1))); |
860 | bDoLongRange = (fr->bTwinRange && bNS && (flags & GMX_FORCE_DO_LR(1<<11))); |
861 | bDoForces = (flags & GMX_FORCE_FORCES(1<<7)); |
862 | bSepLRF = (bDoLongRange && bDoForces && (flags & GMX_FORCE_SEPLRF(1<<5))); |
863 | bUseGPU = fr->nbv->bUseGPU; |
864 | bUseOrEmulGPU = bUseGPU || (nbv->grp[0].kernel_type == nbnxnk8x8x8_PlainC); |
865 | |
866 | if (bStateChanged) |
867 | { |
868 | update_forcerec(fr, box); |
869 | |
870 | if (NEED_MUTOT(*inputrec)(((*inputrec).coulombtype == eelEWALD || (((*inputrec).coulombtype ) == eelPME || ((*inputrec).coulombtype) == eelPMESWITCH || ( (*inputrec).coulombtype) == eelPMEUSER || ((*inputrec).coulombtype ) == eelPMEUSERSWITCH || ((*inputrec).coulombtype) == eelP3M_AD )) && ((*inputrec).ewald_geometry == eewg3DC || (*inputrec ).epsilon_surface != 0))) |
871 | { |
872 | /* Calculate total (local) dipole moment in a temporary common array. |
873 | * This makes it possible to sum them over nodes faster. |
874 | */ |
875 | calc_mu(start, homenr, |
876 | x, mdatoms->chargeA, mdatoms->chargeB, mdatoms->nChargePerturbed, |
877 | mu, mu+DIM3); |
878 | } |
879 | } |
880 | |
881 | if (fr->ePBC != epbcNONE) |
882 | { |
883 | /* Compute shift vectors every step, |
884 | * because of pressure coupling or box deformation! |
885 | */ |
886 | if ((flags & GMX_FORCE_DYNAMICBOX(1<<1)) && bStateChanged) |
887 | { |
888 | calc_shifts(box, fr->shift_vec); |
889 | } |
890 | |
891 | if (bCalcCGCM) |
892 | { |
893 | put_atoms_in_box_omp(fr->ePBC, box, homenr, x); |
894 | inc_nrnb(nrnb, eNR_SHIFTX, homenr)(nrnb)->n[eNR_SHIFTX] += homenr; |
895 | } |
896 | else if (EI_ENERGY_MINIMIZATION(inputrec->eI)((inputrec->eI) == eiSteep || (inputrec->eI) == eiCG || (inputrec->eI) == eiLBFGS) && graph) |
897 | { |
898 | unshift_self(graph, box, x); |
899 | } |
900 | } |
901 | |
902 | nbnxn_atomdata_copy_shiftvec(flags & GMX_FORCE_DYNAMICBOX(1<<1), |
903 | fr->shift_vec, nbv->grp[0].nbat); |
904 | |
905 | #ifdef GMX_MPI |
906 | if (!(cr->duty & DUTY_PME(1<<1))) |
907 | { |
908 | /* Send particle coordinates to the pme nodes. |
909 | * Since this is only implemented for domain decomposition |
910 | * and domain decomposition does not use the graph, |
911 | * we do not need to worry about shifting. |
912 | */ |
913 | |
914 | int pme_flags = 0; |
915 | |
916 | wallcycle_start(wcycle, ewcPP_PMESENDX); |
917 | |
918 | bBS = (inputrec->nwall == 2); |
919 | if (bBS) |
920 | { |
921 | copy_mat(box, boxs); |
922 | svmul(inputrec->wall_ewald_zfac, boxs[ZZ2], boxs[ZZ2]); |
923 | } |
924 | |
925 | if (EEL_PME(fr->eeltype)((fr->eeltype) == eelPME || (fr->eeltype) == eelPMESWITCH || (fr->eeltype) == eelPMEUSER || (fr->eeltype) == eelPMEUSERSWITCH || (fr->eeltype) == eelP3M_AD)) |
926 | { |
927 | pme_flags |= GMX_PME_DO_COULOMB(1<<13); |
928 | } |
929 | |
930 | if (EVDW_PME(fr->vdwtype)((fr->vdwtype) == evdwPME)) |
931 | { |
932 | pme_flags |= GMX_PME_DO_LJ(1<<14); |
933 | } |
934 | |
935 | gmx_pme_send_coordinates(cr, bBS ? boxs : box, x, |
936 | mdatoms->nChargePerturbed, mdatoms->nTypePerturbed, lambda[efptCOUL], lambda[efptVDW], |
937 | (flags & (GMX_FORCE_VIRIAL(1<<8) | GMX_FORCE_ENERGY(1<<9))), |
938 | pme_flags, step); |
939 | |
940 | wallcycle_stop(wcycle, ewcPP_PMESENDX); |
941 | } |
942 | #endif /* GMX_MPI */ |
943 | |
944 | /* do gridding for pair search */ |
945 | if (bNS) |
946 | { |
947 | if (graph && bStateChanged) |
948 | { |
949 | /* Calculate intramolecular shift vectors to make molecules whole */ |
950 | mk_mshift(fplog, graph, fr->ePBC, box, x); |
951 | } |
952 | |
953 | clear_rvec(vzero); |
954 | box_diag[XX0] = box[XX0][XX0]; |
955 | box_diag[YY1] = box[YY1][YY1]; |
956 | box_diag[ZZ2] = box[ZZ2][ZZ2]; |
957 | |
958 | wallcycle_start(wcycle, ewcNS); |
959 | if (!fr->bDomDec) |
960 | { |
961 | wallcycle_sub_start(wcycle, ewcsNBS_GRID_LOCAL); |
962 | nbnxn_put_on_grid(nbv->nbs, fr->ePBC, box, |
963 | 0, vzero, box_diag, |
964 | 0, mdatoms->homenr, -1, fr->cginfo, x, |
965 | 0, NULL((void*)0), |
966 | nbv->grp[eintLocal].kernel_type, |
967 | nbv->grp[eintLocal].nbat); |
968 | wallcycle_sub_stop(wcycle, ewcsNBS_GRID_LOCAL); |
969 | } |
970 | else |
971 | { |
972 | wallcycle_sub_start(wcycle, ewcsNBS_GRID_NONLOCAL); |
973 | nbnxn_put_on_grid_nonlocal(nbv->nbs, domdec_zones(cr->dd), |
974 | fr->cginfo, x, |
975 | nbv->grp[eintNonlocal].kernel_type, |
976 | nbv->grp[eintNonlocal].nbat); |
977 | wallcycle_sub_stop(wcycle, ewcsNBS_GRID_NONLOCAL); |
978 | } |
979 | |
980 | if (nbv->ngrp == 1 || |
981 | nbv->grp[eintNonlocal].nbat == nbv->grp[eintLocal].nbat) |
982 | { |
983 | nbnxn_atomdata_set(nbv->grp[eintLocal].nbat, eatAll, |
984 | nbv->nbs, mdatoms, fr->cginfo); |
985 | } |
986 | else |
987 | { |
988 | nbnxn_atomdata_set(nbv->grp[eintLocal].nbat, eatLocal, |
989 | nbv->nbs, mdatoms, fr->cginfo); |
990 | nbnxn_atomdata_set(nbv->grp[eintNonlocal].nbat, eatAll, |
991 | nbv->nbs, mdatoms, fr->cginfo); |
992 | } |
993 | wallcycle_stop(wcycle, ewcNS); |
994 | } |
995 | |
996 | /* initialize the GPU atom data and copy shift vector */ |
997 | if (bUseGPU) |
998 | { |
999 | if (bNS) |
1000 | { |
1001 | wallcycle_start_nocount(wcycle, ewcLAUNCH_GPU_NB); |
1002 | nbnxn_cuda_init_atomdata(nbv->cu_nbv, nbv->grp[eintLocal].nbat); |
1003 | wallcycle_stop(wcycle, ewcLAUNCH_GPU_NB); |
1004 | } |
1005 | |
1006 | wallcycle_start_nocount(wcycle, ewcLAUNCH_GPU_NB); |
1007 | nbnxn_cuda_upload_shiftvec(nbv->cu_nbv, nbv->grp[eintLocal].nbat); |
1008 | wallcycle_stop(wcycle, ewcLAUNCH_GPU_NB); |
1009 | } |
1010 | |
1011 | /* do local pair search */ |
1012 | if (bNS) |
1013 | { |
1014 | wallcycle_start_nocount(wcycle, ewcNS); |
1015 | wallcycle_sub_start(wcycle, ewcsNBS_SEARCH_LOCAL); |
1016 | nbnxn_make_pairlist(nbv->nbs, nbv->grp[eintLocal].nbat, |
1017 | &top->excls, |
1018 | ic->rlist, |
1019 | nbv->min_ci_balanced, |
1020 | &nbv->grp[eintLocal].nbl_lists, |
1021 | eintLocal, |
1022 | nbv->grp[eintLocal].kernel_type, |
1023 | nrnb); |
1024 | wallcycle_sub_stop(wcycle, ewcsNBS_SEARCH_LOCAL); |
1025 | |
1026 | if (bUseGPU) |
1027 | { |
1028 | /* initialize local pair-list on the GPU */ |
1029 | nbnxn_cuda_init_pairlist(nbv->cu_nbv, |
1030 | nbv->grp[eintLocal].nbl_lists.nbl[0], |
1031 | eintLocal); |
1032 | } |
1033 | wallcycle_stop(wcycle, ewcNS); |
1034 | } |
1035 | else |
1036 | { |
1037 | wallcycle_start(wcycle, ewcNB_XF_BUF_OPS); |
1038 | wallcycle_sub_start(wcycle, ewcsNB_X_BUF_OPS); |
1039 | nbnxn_atomdata_copy_x_to_nbat_x(nbv->nbs, eatLocal, FALSE0, x, |
1040 | nbv->grp[eintLocal].nbat); |
1041 | wallcycle_sub_stop(wcycle, ewcsNB_X_BUF_OPS); |
1042 | wallcycle_stop(wcycle, ewcNB_XF_BUF_OPS); |
1043 | } |
1044 | |
1045 | if (bUseGPU) |
1046 | { |
1047 | wallcycle_start(wcycle, ewcLAUNCH_GPU_NB); |
1048 | /* launch local nonbonded F on GPU */ |
1049 | do_nb_verlet(fr, ic, enerd, flags, eintLocal, enbvClearFNo, |
1050 | nrnb, wcycle); |
1051 | wallcycle_stop(wcycle, ewcLAUNCH_GPU_NB); |
1052 | } |
1053 | |
1054 | /* Communicate coordinates and sum dipole if necessary + |
1055 | do non-local pair search */ |
1056 | if (DOMAINDECOMP(cr)(((cr)->dd != ((void*)0)) && ((cr)->nnodes > 1))) |
1057 | { |
1058 | bDiffKernels = (nbv->grp[eintNonlocal].kernel_type != |
1059 | nbv->grp[eintLocal].kernel_type); |
1060 | |
1061 | if (bDiffKernels) |
1062 | { |
1063 | /* With GPU+CPU non-bonded calculations we need to copy |
1064 | * the local coordinates to the non-local nbat struct |
1065 | * (in CPU format) as the non-local kernel call also |
1066 | * calculates the local - non-local interactions. |
1067 | */ |
1068 | wallcycle_start(wcycle, ewcNB_XF_BUF_OPS); |
1069 | wallcycle_sub_start(wcycle, ewcsNB_X_BUF_OPS); |
1070 | nbnxn_atomdata_copy_x_to_nbat_x(nbv->nbs, eatLocal, TRUE1, x, |
1071 | nbv->grp[eintNonlocal].nbat); |
1072 | wallcycle_sub_stop(wcycle, ewcsNB_X_BUF_OPS); |
1073 | wallcycle_stop(wcycle, ewcNB_XF_BUF_OPS); |
1074 | } |
1075 | |
1076 | if (bNS) |
1077 | { |
1078 | wallcycle_start_nocount(wcycle, ewcNS); |
1079 | wallcycle_sub_start(wcycle, ewcsNBS_SEARCH_NONLOCAL); |
1080 | |
1081 | if (bDiffKernels) |
1082 | { |
1083 | nbnxn_grid_add_simple(nbv->nbs, nbv->grp[eintNonlocal].nbat); |
1084 | } |
1085 | |
1086 | nbnxn_make_pairlist(nbv->nbs, nbv->grp[eintNonlocal].nbat, |
1087 | &top->excls, |
1088 | ic->rlist, |
1089 | nbv->min_ci_balanced, |
1090 | &nbv->grp[eintNonlocal].nbl_lists, |
1091 | eintNonlocal, |
1092 | nbv->grp[eintNonlocal].kernel_type, |
1093 | nrnb); |
1094 | |
1095 | wallcycle_sub_stop(wcycle, ewcsNBS_SEARCH_NONLOCAL); |
1096 | |
1097 | if (nbv->grp[eintNonlocal].kernel_type == nbnxnk8x8x8_CUDA) |
1098 | { |
1099 | /* initialize non-local pair-list on the GPU */ |
1100 | nbnxn_cuda_init_pairlist(nbv->cu_nbv, |
1101 | nbv->grp[eintNonlocal].nbl_lists.nbl[0], |
1102 | eintNonlocal); |
1103 | } |
1104 | wallcycle_stop(wcycle, ewcNS); |
1105 | } |
1106 | else |
1107 | { |
1108 | wallcycle_start(wcycle, ewcMOVEX); |
1109 | dd_move_x(cr->dd, box, x); |
1110 | |
1111 | /* When we don't need the total dipole we sum it in global_stat */ |
1112 | if (bStateChanged && NEED_MUTOT(*inputrec)(((*inputrec).coulombtype == eelEWALD || (((*inputrec).coulombtype ) == eelPME || ((*inputrec).coulombtype) == eelPMESWITCH || ( (*inputrec).coulombtype) == eelPMEUSER || ((*inputrec).coulombtype ) == eelPMEUSERSWITCH || ((*inputrec).coulombtype) == eelP3M_AD )) && ((*inputrec).ewald_geometry == eewg3DC || (*inputrec ).epsilon_surface != 0))) |
1113 | { |
1114 | gmx_sumd(2*DIM3, mu, cr); |
1115 | } |
1116 | wallcycle_stop(wcycle, ewcMOVEX); |
1117 | |
1118 | wallcycle_start(wcycle, ewcNB_XF_BUF_OPS); |
1119 | wallcycle_sub_start(wcycle, ewcsNB_X_BUF_OPS); |
1120 | nbnxn_atomdata_copy_x_to_nbat_x(nbv->nbs, eatNonlocal, FALSE0, x, |
1121 | nbv->grp[eintNonlocal].nbat); |
1122 | wallcycle_sub_stop(wcycle, ewcsNB_X_BUF_OPS); |
1123 | cycles_force += wallcycle_stop(wcycle, ewcNB_XF_BUF_OPS); |
1124 | } |
1125 | |
1126 | if (bUseGPU && !bDiffKernels) |
1127 | { |
1128 | wallcycle_start(wcycle, ewcLAUNCH_GPU_NB); |
1129 | /* launch non-local nonbonded F on GPU */ |
1130 | do_nb_verlet(fr, ic, enerd, flags, eintNonlocal, enbvClearFNo, |
1131 | nrnb, wcycle); |
1132 | cycles_force += wallcycle_stop(wcycle, ewcLAUNCH_GPU_NB); |
1133 | } |
1134 | } |
1135 | |
1136 | if (bUseGPU) |
1137 | { |
1138 | /* launch D2H copy-back F */ |
1139 | wallcycle_start_nocount(wcycle, ewcLAUNCH_GPU_NB); |
1140 | if (DOMAINDECOMP(cr)(((cr)->dd != ((void*)0)) && ((cr)->nnodes > 1)) && !bDiffKernels) |
1141 | { |
1142 | nbnxn_cuda_launch_cpyback(nbv->cu_nbv, nbv->grp[eintNonlocal].nbat, |
1143 | flags, eatNonlocal); |
1144 | } |
1145 | nbnxn_cuda_launch_cpyback(nbv->cu_nbv, nbv->grp[eintLocal].nbat, |
1146 | flags, eatLocal); |
1147 | cycles_force += wallcycle_stop(wcycle, ewcLAUNCH_GPU_NB); |
1148 | } |
1149 | |
1150 | if (bStateChanged && NEED_MUTOT(*inputrec)(((*inputrec).coulombtype == eelEWALD || (((*inputrec).coulombtype ) == eelPME || ((*inputrec).coulombtype) == eelPMESWITCH || ( (*inputrec).coulombtype) == eelPMEUSER || ((*inputrec).coulombtype ) == eelPMEUSERSWITCH || ((*inputrec).coulombtype) == eelP3M_AD )) && ((*inputrec).ewald_geometry == eewg3DC || (*inputrec ).epsilon_surface != 0))) |
1151 | { |
1152 | if (PAR(cr)((cr)->nnodes > 1)) |
1153 | { |
1154 | gmx_sumd(2*DIM3, mu, cr); |
1155 | } |
1156 | |
1157 | for (i = 0; i < 2; i++) |
1158 | { |
1159 | for (j = 0; j < DIM3; j++) |
1160 | { |
1161 | fr->mu_tot[i][j] = mu[i*DIM3 + j]; |
1162 | } |
1163 | } |
1164 | } |
1165 | if (fr->efep == efepNO) |
1166 | { |
1167 | copy_rvec(fr->mu_tot[0], mu_tot); |
1168 | } |
1169 | else |
1170 | { |
1171 | for (j = 0; j < DIM3; j++) |
1172 | { |
1173 | mu_tot[j] = |
1174 | (1.0 - lambda[efptCOUL])*fr->mu_tot[0][j] + |
1175 | lambda[efptCOUL]*fr->mu_tot[1][j]; |
1176 | } |
1177 | } |
1178 | |
1179 | /* Reset energies */ |
1180 | reset_enerdata(fr, bNS, enerd, MASTER(cr)(((cr)->nodeid == 0) || !((cr)->nnodes > 1))); |
1181 | clear_rvecs(SHIFTS((2*1 +1)*(2*1 +1)*(2*2 +1)), fr->fshift); |
1182 | |
1183 | if (DOMAINDECOMP(cr)(((cr)->dd != ((void*)0)) && ((cr)->nnodes > 1)) && !(cr->duty & DUTY_PME(1<<1))) |
1184 | { |
1185 | wallcycle_start(wcycle, ewcPPDURINGPME); |
1186 | dd_force_flop_start(cr->dd, nrnb); |
1187 | } |
1188 | |
1189 | if (inputrec->bRot) |
1190 | { |
1191 | /* Enforced rotation has its own cycle counter that starts after the collective |
1192 | * coordinates have been communicated. It is added to ddCyclF to allow |
1193 | * for proper load-balancing */ |
1194 | wallcycle_start(wcycle, ewcROT); |
1195 | do_rotation(cr, inputrec, box, x, t, step, wcycle, bNS); |
1196 | wallcycle_stop(wcycle, ewcROT); |
1197 | } |
1198 | |
1199 | /* Start the force cycle counter. |
1200 | * This counter is stopped in do_forcelow_level. |
1201 | * No parallel communication should occur while this counter is running, |
1202 | * since that will interfere with the dynamic load balancing. |
1203 | */ |
1204 | wallcycle_start(wcycle, ewcFORCE); |
1205 | if (bDoForces) |
1206 | { |
1207 | /* Reset forces for which the virial is calculated separately: |
1208 | * PME/Ewald forces if necessary */ |
1209 | if (fr->bF_NoVirSum) |
1210 | { |
1211 | if (flags & GMX_FORCE_VIRIAL(1<<8)) |
1212 | { |
1213 | fr->f_novirsum = fr->f_novirsum_alloc; |
1214 | if (fr->bDomDec) |
1215 | { |
1216 | clear_rvecs(fr->f_novirsum_n, fr->f_novirsum); |
1217 | } |
1218 | else |
1219 | { |
1220 | clear_rvecs(homenr, fr->f_novirsum+start); |
1221 | } |
1222 | } |
1223 | else |
1224 | { |
1225 | /* We are not calculating the pressure so we do not need |
1226 | * a separate array for forces that do not contribute |
1227 | * to the pressure. |
1228 | */ |
1229 | fr->f_novirsum = f; |
1230 | } |
1231 | } |
1232 | |
1233 | /* Clear the short- and long-range forces */ |
1234 | clear_rvecs(fr->natoms_force_constr, f); |
1235 | if (bSepLRF && do_per_step(step, inputrec->nstcalclr)) |
1236 | { |
1237 | clear_rvecs(fr->natoms_force_constr, fr->f_twin); |
1238 | } |
1239 | |
1240 | clear_rvec(fr->vir_diag_posres); |
1241 | } |
1242 | |
1243 | if (inputrec->ePull == epullCONSTRAINT) |
1244 | { |
1245 | clear_pull_forces(inputrec->pull); |
1246 | } |
1247 | |
1248 | /* We calculate the non-bonded forces, when done on the CPU, here. |
1249 | * We do this before calling do_force_lowlevel, as in there bondeds |
1250 | * forces are calculated before PME, which does communication. |
1251 | * With this order, non-bonded and bonded force calculation imbalance |
1252 | * can be balanced out by the domain decomposition load balancing. |
1253 | */ |
1254 | |
1255 | if (!bUseOrEmulGPU) |
1256 | { |
1257 | /* Maybe we should move this into do_force_lowlevel */ |
1258 | do_nb_verlet(fr, ic, enerd, flags, eintLocal, enbvClearFYes, |
1259 | nrnb, wcycle); |
1260 | } |
1261 | |
1262 | if (fr->efep != efepNO) |
1263 | { |
1264 | /* Calculate the local and non-local free energy interactions here. |
1265 | * Happens here on the CPU both with and without GPU. |
1266 | */ |
1267 | if (fr->nbv->grp[eintLocal].nbl_lists.nbl_fep[0]->nrj > 0) |
1268 | { |
1269 | do_nb_verlet_fep(&fr->nbv->grp[eintLocal].nbl_lists, |
1270 | fr, x, f, mdatoms, |
1271 | inputrec->fepvals, lambda, |
1272 | enerd, flags, nrnb, wcycle); |
1273 | } |
1274 | |
1275 | if (DOMAINDECOMP(cr)(((cr)->dd != ((void*)0)) && ((cr)->nnodes > 1)) && |
1276 | fr->nbv->grp[eintNonlocal].nbl_lists.nbl_fep[0]->nrj > 0) |
1277 | { |
1278 | do_nb_verlet_fep(&fr->nbv->grp[eintNonlocal].nbl_lists, |
1279 | fr, x, f, mdatoms, |
1280 | inputrec->fepvals, lambda, |
1281 | enerd, flags, nrnb, wcycle); |
1282 | } |
1283 | } |
1284 | |
1285 | if (!bUseOrEmulGPU || bDiffKernels) |
1286 | { |
1287 | int aloc; |
1288 | |
1289 | if (DOMAINDECOMP(cr)(((cr)->dd != ((void*)0)) && ((cr)->nnodes > 1))) |
1290 | { |
1291 | do_nb_verlet(fr, ic, enerd, flags, eintNonlocal, |
1292 | bDiffKernels ? enbvClearFYes : enbvClearFNo, |
1293 | nrnb, wcycle); |
1294 | } |
1295 | |
1296 | if (!bUseOrEmulGPU) |
1297 | { |
1298 | aloc = eintLocal; |
1299 | } |
1300 | else |
1301 | { |
1302 | aloc = eintNonlocal; |
1303 | } |
1304 | |
1305 | /* Add all the non-bonded force to the normal force array. |
1306 | * This can be split into a local a non-local part when overlapping |
1307 | * communication with calculation with domain decomposition. |
1308 | */ |
1309 | cycles_force += wallcycle_stop(wcycle, ewcFORCE); |
1310 | wallcycle_start(wcycle, ewcNB_XF_BUF_OPS); |
1311 | wallcycle_sub_start(wcycle, ewcsNB_F_BUF_OPS); |
1312 | nbnxn_atomdata_add_nbat_f_to_f(nbv->nbs, eatAll, nbv->grp[aloc].nbat, f); |
1313 | wallcycle_sub_stop(wcycle, ewcsNB_F_BUF_OPS); |
1314 | cycles_force += wallcycle_stop(wcycle, ewcNB_XF_BUF_OPS); |
1315 | wallcycle_start_nocount(wcycle, ewcFORCE); |
1316 | |
1317 | /* if there are multiple fshift output buffers reduce them */ |
1318 | if ((flags & GMX_FORCE_VIRIAL(1<<8)) && |
1319 | nbv->grp[aloc].nbl_lists.nnbl > 1) |
1320 | { |
1321 | nbnxn_atomdata_add_nbat_fshift_to_fshift(nbv->grp[aloc].nbat, |
1322 | fr->fshift); |
1323 | } |
1324 | } |
1325 | |
1326 | /* update QMMMrec, if necessary */ |
1327 | if (fr->bQMMM) |
1328 | { |
1329 | update_QMMMrec(cr, fr, x, mdatoms, box, top); |
1330 | } |
1331 | |
1332 | if ((flags & GMX_FORCE_BONDED(1<<4)) && top->idef.il[F_POSRES].nr > 0) |
1333 | { |
1334 | posres_wrapper(fplog, flags, bSepDVDL, inputrec, nrnb, top, box, x, |
1335 | enerd, lambda, fr); |
1336 | } |
1337 | |
1338 | if ((flags & GMX_FORCE_BONDED(1<<4)) && top->idef.il[F_FBPOSRES].nr > 0) |
1339 | { |
1340 | fbposres_wrapper(inputrec, nrnb, top, box, x, enerd, fr); |
1341 | } |
1342 | |
1343 | /* Compute the bonded and non-bonded energies and optionally forces */ |
1344 | do_force_lowlevel(fplog, step, fr, inputrec, &(top->idef), |
1345 | cr, nrnb, wcycle, mdatoms, |
1346 | x, hist, f, bSepLRF ? fr->f_twin : f, enerd, fcd, top, fr->born, |
1347 | &(top->atomtypes), bBornRadii, box, |
1348 | inputrec->fepvals, lambda, graph, &(top->excls), fr->mu_tot, |
1349 | flags, &cycles_pme); |
1350 | |
1351 | if (bSepLRF) |
1352 | { |
1353 | if (do_per_step(step, inputrec->nstcalclr)) |
1354 | { |
1355 | /* Add the long range forces to the short range forces */ |
1356 | for (i = 0; i < fr->natoms_force_constr; i++) |
1357 | { |
1358 | rvec_add(fr->f_twin[i], f[i], f[i]); |
1359 | } |
1360 | } |
1361 | } |
1362 | |
1363 | cycles_force += wallcycle_stop(wcycle, ewcFORCE); |
1364 | |
1365 | if (ed) |
1366 | { |
1367 | do_flood(cr, inputrec, x, f, ed, box, step, bNS); |
1368 | } |
1369 | |
1370 | if (bUseOrEmulGPU && !bDiffKernels) |
1371 | { |
1372 | /* wait for non-local forces (or calculate in emulation mode) */ |
1373 | if (DOMAINDECOMP(cr)(((cr)->dd != ((void*)0)) && ((cr)->nnodes > 1))) |
1374 | { |
1375 | if (bUseGPU) |
1376 | { |
1377 | float cycles_tmp; |
1378 | |
1379 | wallcycle_start(wcycle, ewcWAIT_GPU_NB_NL); |
1380 | nbnxn_cuda_wait_gpu(nbv->cu_nbv, |
1381 | nbv->grp[eintNonlocal].nbat, |
1382 | flags, eatNonlocal, |
1383 | enerd->grpp.ener[egLJSR], enerd->grpp.ener[egCOULSR], |
1384 | fr->fshift); |
1385 | cycles_tmp = wallcycle_stop(wcycle, ewcWAIT_GPU_NB_NL); |
1386 | cycles_wait_gpu += cycles_tmp; |
1387 | cycles_force += cycles_tmp; |
1388 | } |
1389 | else |
1390 | { |
1391 | wallcycle_start_nocount(wcycle, ewcFORCE); |
1392 | do_nb_verlet(fr, ic, enerd, flags, eintNonlocal, enbvClearFYes, |
1393 | nrnb, wcycle); |
1394 | cycles_force += wallcycle_stop(wcycle, ewcFORCE); |
1395 | } |
1396 | wallcycle_start(wcycle, ewcNB_XF_BUF_OPS); |
1397 | wallcycle_sub_start(wcycle, ewcsNB_F_BUF_OPS); |
1398 | /* skip the reduction if there was no non-local work to do */ |
1399 | if (nbv->grp[eintLocal].nbl_lists.nbl[0]->nsci > 0) |
1400 | { |
1401 | nbnxn_atomdata_add_nbat_f_to_f(nbv->nbs, eatNonlocal, |
1402 | nbv->grp[eintNonlocal].nbat, f); |
1403 | } |
1404 | wallcycle_sub_stop(wcycle, ewcsNB_F_BUF_OPS); |
1405 | cycles_force += wallcycle_stop(wcycle, ewcNB_XF_BUF_OPS); |
1406 | } |
1407 | } |
1408 | |
1409 | if (bDoForces && DOMAINDECOMP(cr)(((cr)->dd != ((void*)0)) && ((cr)->nnodes > 1))) |
1410 | { |
1411 | /* Communicate the forces */ |
1412 | wallcycle_start(wcycle, ewcMOVEF); |
1413 | dd_move_f(cr->dd, f, fr->fshift); |
1414 | /* Do we need to communicate the separate force array |
1415 | * for terms that do not contribute to the single sum virial? |
1416 | * Position restraints and electric fields do not introduce |
1417 | * inter-cg forces, only full electrostatics methods do. |
1418 | * When we do not calculate the virial, fr->f_novirsum = f, |
1419 | * so we have already communicated these forces. |
1420 | */ |
1421 | if (EEL_FULL(fr->eeltype)((((fr->eeltype) == eelPME || (fr->eeltype) == eelPMESWITCH || (fr->eeltype) == eelPMEUSER || (fr->eeltype) == eelPMEUSERSWITCH || (fr->eeltype) == eelP3M_AD) || (fr->eeltype) == eelEWALD ) || (fr->eeltype) == eelPOISSON) && cr->dd->n_intercg_excl && |
1422 | (flags & GMX_FORCE_VIRIAL(1<<8))) |
1423 | { |
1424 | dd_move_f(cr->dd, fr->f_novirsum, NULL((void*)0)); |
1425 | } |
1426 | if (bSepLRF) |
1427 | { |
1428 | /* We should not update the shift forces here, |
1429 | * since f_twin is already included in f. |
1430 | */ |
1431 | dd_move_f(cr->dd, fr->f_twin, NULL((void*)0)); |
1432 | } |
1433 | wallcycle_stop(wcycle, ewcMOVEF); |
1434 | } |
1435 | |
1436 | if (bUseOrEmulGPU) |
1437 | { |
1438 | /* wait for local forces (or calculate in emulation mode) */ |
1439 | if (bUseGPU) |
1440 | { |
1441 | wallcycle_start(wcycle, ewcWAIT_GPU_NB_L); |
1442 | nbnxn_cuda_wait_gpu(nbv->cu_nbv, |
1443 | nbv->grp[eintLocal].nbat, |
1444 | flags, eatLocal, |
1445 | enerd->grpp.ener[egLJSR], enerd->grpp.ener[egCOULSR], |
1446 | fr->fshift); |
1447 | cycles_wait_gpu += wallcycle_stop(wcycle, ewcWAIT_GPU_NB_L); |
1448 | |
1449 | /* now clear the GPU outputs while we finish the step on the CPU */ |
1450 | |
1451 | wallcycle_start_nocount(wcycle, ewcLAUNCH_GPU_NB); |
1452 | nbnxn_cuda_clear_outputs(nbv->cu_nbv, flags); |
1453 | wallcycle_stop(wcycle, ewcLAUNCH_GPU_NB); |
1454 | } |
1455 | else |
1456 | { |
1457 | wallcycle_start_nocount(wcycle, ewcFORCE); |
1458 | do_nb_verlet(fr, ic, enerd, flags, eintLocal, |
1459 | DOMAINDECOMP(cr)(((cr)->dd != ((void*)0)) && ((cr)->nnodes > 1)) ? enbvClearFNo : enbvClearFYes, |
1460 | nrnb, wcycle); |
1461 | wallcycle_stop(wcycle, ewcFORCE); |
1462 | } |
1463 | wallcycle_start(wcycle, ewcNB_XF_BUF_OPS); |
1464 | wallcycle_sub_start(wcycle, ewcsNB_F_BUF_OPS); |
1465 | if (nbv->grp[eintLocal].nbl_lists.nbl[0]->nsci > 0) |
1466 | { |
1467 | /* skip the reduction if there was no non-local work to do */ |
1468 | nbnxn_atomdata_add_nbat_f_to_f(nbv->nbs, eatLocal, |
1469 | nbv->grp[eintLocal].nbat, f); |
1470 | } |
1471 | wallcycle_sub_stop(wcycle, ewcsNB_F_BUF_OPS); |
1472 | wallcycle_stop(wcycle, ewcNB_XF_BUF_OPS); |
1473 | } |
1474 | |
1475 | if (DOMAINDECOMP(cr)(((cr)->dd != ((void*)0)) && ((cr)->nnodes > 1))) |
1476 | { |
1477 | dd_force_flop_stop(cr->dd, nrnb); |
1478 | if (wcycle) |
1479 | { |
1480 | dd_cycles_add(cr->dd, cycles_force-cycles_pme, ddCyclF); |
1481 | if (bUseGPU) |
1482 | { |
1483 | dd_cycles_add(cr->dd, cycles_wait_gpu, ddCyclWaitGPU); |
1484 | } |
1485 | } |
1486 | } |
1487 | |
1488 | if (bDoForces) |
1489 | { |
1490 | if (IR_ELEC_FIELD(*inputrec)((*inputrec).ex[0].n > 0 || (*inputrec).ex[1].n > 0 || ( *inputrec).ex[2].n > 0)) |
1491 | { |
1492 | /* Compute forces due to electric field */ |
1493 | calc_f_el(MASTER(cr)(((cr)->nodeid == 0) || !((cr)->nnodes > 1)) ? field : NULL((void*)0), |
1494 | start, homenr, mdatoms->chargeA, fr->f_novirsum, |
1495 | inputrec->ex, inputrec->et, t); |
1496 | } |
1497 | |
1498 | /* If we have NoVirSum forces, but we do not calculate the virial, |
1499 | * we sum fr->f_novirum=f later. |
1500 | */ |
1501 | if (vsite && !(fr->bF_NoVirSum && !(flags & GMX_FORCE_VIRIAL(1<<8)))) |
1502 | { |
1503 | wallcycle_start(wcycle, ewcVSITESPREAD); |
1504 | spread_vsite_f(vsite, x, f, fr->fshift, FALSE0, NULL((void*)0), nrnb, |
1505 | &top->idef, fr->ePBC, fr->bMolPBC, graph, box, cr); |
1506 | wallcycle_stop(wcycle, ewcVSITESPREAD); |
1507 | |
1508 | if (bSepLRF) |
1509 | { |
1510 | wallcycle_start(wcycle, ewcVSITESPREAD); |
1511 | spread_vsite_f(vsite, x, fr->f_twin, NULL((void*)0), FALSE0, NULL((void*)0), |
1512 | nrnb, |
1513 | &top->idef, fr->ePBC, fr->bMolPBC, graph, box, cr); |
1514 | wallcycle_stop(wcycle, ewcVSITESPREAD); |
1515 | } |
1516 | } |
1517 | |
1518 | if (flags & GMX_FORCE_VIRIAL(1<<8)) |
1519 | { |
1520 | /* Calculation of the virial must be done after vsites! */ |
1521 | calc_virial(0, mdatoms->homenr, x, f, |
1522 | vir_force, graph, box, nrnb, fr, inputrec->ePBC); |
1523 | } |
1524 | } |
1525 | |
1526 | if (inputrec->ePull == epullUMBRELLA || inputrec->ePull == epullCONST_F) |
1527 | { |
1528 | pull_potential_wrapper(fplog, bSepDVDL, cr, inputrec, box, x, |
1529 | f, vir_force, mdatoms, enerd, lambda, t); |
1530 | } |
1531 | |
1532 | /* Add the forces from enforced rotation potentials (if any) */ |
1533 | if (inputrec->bRot) |
1534 | { |
1535 | wallcycle_start(wcycle, ewcROTadd); |
1536 | enerd->term[F_COM_PULL] += add_rot_forces(inputrec->rot, f, cr, step, t); |
1537 | wallcycle_stop(wcycle, ewcROTadd); |
1538 | } |
1539 | |
1540 | /* Add forces from interactive molecular dynamics (IMD), if bIMD == TRUE. */ |
1541 | IMD_apply_forces(inputrec->bIMD, inputrec->imd, cr, f, wcycle); |
1542 | |
1543 | if (PAR(cr)((cr)->nnodes > 1) && !(cr->duty & DUTY_PME(1<<1))) |
1544 | { |
1545 | /* In case of node-splitting, the PP nodes receive the long-range |
1546 | * forces, virial and energy from the PME nodes here. |
1547 | */ |
1548 | pme_receive_force_ener(fplog, bSepDVDL, cr, wcycle, enerd, fr); |
1549 | } |
1550 | |
1551 | if (bDoForces) |
1552 | { |
1553 | post_process_forces(cr, step, nrnb, wcycle, |
1554 | top, box, x, f, vir_force, mdatoms, graph, fr, vsite, |
1555 | flags); |
1556 | } |
1557 | |
1558 | /* Sum the potential energy terms from group contributions */ |
1559 | sum_epot(&(enerd->grpp), enerd->term); |
1560 | } |
1561 | |
1562 | void do_force_cutsGROUP(FILE *fplog, t_commrec *cr, |
1563 | t_inputrec *inputrec, |
1564 | gmx_int64_t step, t_nrnb *nrnb, gmx_wallcycle_t wcycle, |
1565 | gmx_localtop_t *top, |
1566 | gmx_groups_t *groups, |
1567 | matrix box, rvec x[], history_t *hist, |
1568 | rvec f[], |
1569 | tensor vir_force, |
1570 | t_mdatoms *mdatoms, |
1571 | gmx_enerdata_t *enerd, t_fcdata *fcd, |
1572 | real *lambda, t_graph *graph, |
1573 | t_forcerec *fr, gmx_vsite_t *vsite, rvec mu_tot, |
1574 | double t, FILE *field, gmx_edsam_t ed, |
1575 | gmx_bool bBornRadii, |
1576 | int flags) |
1577 | { |
1578 | int cg0, cg1, i, j; |
1579 | int start, homenr; |
1580 | double mu[2*DIM3]; |
1581 | gmx_bool bSepDVDL, bStateChanged, bNS, bFillGrid, bCalcCGCM, bBS; |
1582 | gmx_bool bDoLongRangeNS, bDoForces, bDoPotential, bSepLRF; |
1583 | gmx_bool bDoAdressWF; |
1584 | matrix boxs; |
1585 | rvec vzero, box_diag; |
1586 | real e, v, dvdlambda[efptNR]; |
1587 | t_pbc pbc; |
1588 | float cycles_pme, cycles_force; |
1589 | |
1590 | start = 0; |
1591 | homenr = mdatoms->homenr; |
1592 | |
1593 | bSepDVDL = (fr->bSepDVDL && do_per_step(step, inputrec->nstlog)); |
1594 | |
1595 | clear_mat(vir_force); |
1596 | |
1597 | cg0 = 0; |
1598 | if (DOMAINDECOMP(cr)(((cr)->dd != ((void*)0)) && ((cr)->nnodes > 1))) |
1599 | { |
1600 | cg1 = cr->dd->ncg_tot; |
1601 | } |
1602 | else |
1603 | { |
1604 | cg1 = top->cgs.nr; |
1605 | } |
1606 | if (fr->n_tpi > 0) |
1607 | { |
1608 | cg1--; |
1609 | } |
1610 | |
1611 | bStateChanged = (flags & GMX_FORCE_STATECHANGED(1<<0)); |
1612 | bNS = (flags & GMX_FORCE_NS(1<<2)) && (fr->bAllvsAll == FALSE0); |
1613 | /* Should we update the long-range neighborlists at this step? */ |
1614 | bDoLongRangeNS = fr->bTwinRange && bNS; |
1615 | /* Should we perform the long-range nonbonded evaluation inside the neighborsearching? */ |
1616 | bFillGrid = (bNS && bStateChanged); |
1617 | bCalcCGCM = (bFillGrid && !DOMAINDECOMP(cr)(((cr)->dd != ((void*)0)) && ((cr)->nnodes > 1))); |
1618 | bDoForces = (flags & GMX_FORCE_FORCES(1<<7)); |
1619 | bDoPotential = (flags & GMX_FORCE_ENERGY(1<<9)); |
1620 | bSepLRF = ((inputrec->nstcalclr > 1) && bDoForces && |
1621 | (flags & GMX_FORCE_SEPLRF(1<<5)) && (flags & GMX_FORCE_DO_LR(1<<11))); |
1622 | |
1623 | /* should probably move this to the forcerec since it doesn't change */ |
1624 | bDoAdressWF = ((fr->adress_type != eAdressOff)); |
1625 | |
1626 | if (bStateChanged) |
1627 | { |
1628 | update_forcerec(fr, box); |
1629 | |
1630 | if (NEED_MUTOT(*inputrec)(((*inputrec).coulombtype == eelEWALD || (((*inputrec).coulombtype ) == eelPME || ((*inputrec).coulombtype) == eelPMESWITCH || ( (*inputrec).coulombtype) == eelPMEUSER || ((*inputrec).coulombtype ) == eelPMEUSERSWITCH || ((*inputrec).coulombtype) == eelP3M_AD )) && ((*inputrec).ewald_geometry == eewg3DC || (*inputrec ).epsilon_surface != 0))) |
1631 | { |
1632 | /* Calculate total (local) dipole moment in a temporary common array. |
1633 | * This makes it possible to sum them over nodes faster. |
1634 | */ |
1635 | calc_mu(start, homenr, |
1636 | x, mdatoms->chargeA, mdatoms->chargeB, mdatoms->nChargePerturbed, |
1637 | mu, mu+DIM3); |
1638 | } |
1639 | } |
1640 | |
1641 | if (fr->ePBC != epbcNONE) |
1642 | { |
1643 | /* Compute shift vectors every step, |
1644 | * because of pressure coupling or box deformation! |
1645 | */ |
1646 | if ((flags & GMX_FORCE_DYNAMICBOX(1<<1)) && bStateChanged) |
1647 | { |
1648 | calc_shifts(box, fr->shift_vec); |
1649 | } |
1650 | |
1651 | if (bCalcCGCM) |
1652 | { |
1653 | put_charge_groups_in_box(fplog, cg0, cg1, fr->ePBC, box, |
1654 | &(top->cgs), x, fr->cg_cm); |
1655 | inc_nrnb(nrnb, eNR_CGCM, homenr)(nrnb)->n[eNR_CGCM] += homenr; |
1656 | inc_nrnb(nrnb, eNR_RESETX, cg1-cg0)(nrnb)->n[eNR_RESETX] += cg1-cg0; |
1657 | } |
1658 | else if (EI_ENERGY_MINIMIZATION(inputrec->eI)((inputrec->eI) == eiSteep || (inputrec->eI) == eiCG || (inputrec->eI) == eiLBFGS) && graph) |
1659 | { |
1660 | unshift_self(graph, box, x); |
1661 | } |
1662 | } |
1663 | else if (bCalcCGCM) |
1664 | { |
1665 | calc_cgcm(fplog, cg0, cg1, &(top->cgs), x, fr->cg_cm); |
1666 | inc_nrnb(nrnb, eNR_CGCM, homenr)(nrnb)->n[eNR_CGCM] += homenr; |
1667 | } |
1668 | |
1669 | if (bCalcCGCM && gmx_debug_at) |
1670 | { |
1671 | pr_rvecs(debug, 0, "cgcm", fr->cg_cm, top->cgs.nr); |
1672 | } |
1673 | |
1674 | #ifdef GMX_MPI |
1675 | if (!(cr->duty & DUTY_PME(1<<1))) |
1676 | { |
1677 | /* Send particle coordinates to the pme nodes. |
1678 | * Since this is only implemented for domain decomposition |
1679 | * and domain decomposition does not use the graph, |
1680 | * we do not need to worry about shifting. |
1681 | */ |
1682 | |
1683 | int pme_flags = 0; |
1684 | |
1685 | wallcycle_start(wcycle, ewcPP_PMESENDX); |
1686 | |
1687 | bBS = (inputrec->nwall == 2); |
1688 | if (bBS) |
1689 | { |
1690 | copy_mat(box, boxs); |
1691 | svmul(inputrec->wall_ewald_zfac, boxs[ZZ2], boxs[ZZ2]); |
1692 | } |
1693 | |
1694 | if (EEL_PME(fr->eeltype)((fr->eeltype) == eelPME || (fr->eeltype) == eelPMESWITCH || (fr->eeltype) == eelPMEUSER || (fr->eeltype) == eelPMEUSERSWITCH || (fr->eeltype) == eelP3M_AD)) |
1695 | { |
1696 | pme_flags |= GMX_PME_DO_COULOMB(1<<13); |
1697 | } |
1698 | |
1699 | if (EVDW_PME(fr->vdwtype)((fr->vdwtype) == evdwPME)) |
1700 | { |
1701 | pme_flags |= GMX_PME_DO_LJ(1<<14); |
1702 | } |
1703 | |
1704 | gmx_pme_send_coordinates(cr, bBS ? boxs : box, x, |
1705 | mdatoms->nChargePerturbed, mdatoms->nTypePerturbed, lambda[efptCOUL], lambda[efptVDW], |
1706 | (flags & (GMX_FORCE_VIRIAL(1<<8) | GMX_FORCE_ENERGY(1<<9))), |
1707 | pme_flags, step); |
1708 | |
1709 | wallcycle_stop(wcycle, ewcPP_PMESENDX); |
1710 | } |
1711 | #endif /* GMX_MPI */ |
1712 | |
1713 | /* Communicate coordinates and sum dipole if necessary */ |
1714 | if (DOMAINDECOMP(cr)(((cr)->dd != ((void*)0)) && ((cr)->nnodes > 1))) |
1715 | { |
1716 | wallcycle_start(wcycle, ewcMOVEX); |
1717 | dd_move_x(cr->dd, box, x); |
1718 | wallcycle_stop(wcycle, ewcMOVEX); |
1719 | } |
1720 | |
1721 | /* update adress weight beforehand */ |
1722 | if (bStateChanged && bDoAdressWF) |
1723 | { |
1724 | /* need pbc for adress weight calculation with pbc_dx */ |
1725 | set_pbc(&pbc, inputrec->ePBC, box); |
1726 | if (fr->adress_site == eAdressSITEcog) |
1727 | { |
1728 | update_adress_weights_cog(top->idef.iparams, top->idef.il, x, fr, mdatoms, |
1729 | inputrec->ePBC == epbcNONE ? NULL((void*)0) : &pbc); |
1730 | } |
1731 | else if (fr->adress_site == eAdressSITEcom) |
1732 | { |
1733 | update_adress_weights_com(fplog, cg0, cg1, &(top->cgs), x, fr, mdatoms, |
1734 | inputrec->ePBC == epbcNONE ? NULL((void*)0) : &pbc); |
1735 | } |
1736 | else if (fr->adress_site == eAdressSITEatomatom) |
1737 | { |
1738 | update_adress_weights_atom_per_atom(cg0, cg1, &(top->cgs), x, fr, mdatoms, |
1739 | inputrec->ePBC == epbcNONE ? NULL((void*)0) : &pbc); |
1740 | } |
1741 | else |
1742 | { |
1743 | update_adress_weights_atom(cg0, cg1, &(top->cgs), x, fr, mdatoms, |
1744 | inputrec->ePBC == epbcNONE ? NULL((void*)0) : &pbc); |
1745 | } |
1746 | } |
1747 | |
1748 | if (NEED_MUTOT(*inputrec)(((*inputrec).coulombtype == eelEWALD || (((*inputrec).coulombtype ) == eelPME || ((*inputrec).coulombtype) == eelPMESWITCH || ( (*inputrec).coulombtype) == eelPMEUSER || ((*inputrec).coulombtype ) == eelPMEUSERSWITCH || ((*inputrec).coulombtype) == eelP3M_AD )) && ((*inputrec).ewald_geometry == eewg3DC || (*inputrec ).epsilon_surface != 0))) |
1749 | { |
1750 | |
1751 | if (bStateChanged) |
1752 | { |
1753 | if (PAR(cr)((cr)->nnodes > 1)) |
1754 | { |
1755 | gmx_sumd(2*DIM3, mu, cr); |
1756 | } |
1757 | for (i = 0; i < 2; i++) |
1758 | { |
1759 | for (j = 0; j < DIM3; j++) |
1760 | { |
1761 | fr->mu_tot[i][j] = mu[i*DIM3 + j]; |
1762 | } |
1763 | } |
1764 | } |
1765 | if (fr->efep == efepNO) |
1766 | { |
1767 | copy_rvec(fr->mu_tot[0], mu_tot); |
1768 | } |
1769 | else |
1770 | { |
1771 | for (j = 0; j < DIM3; j++) |
1772 | { |
1773 | mu_tot[j] = |
1774 | (1.0 - lambda[efptCOUL])*fr->mu_tot[0][j] + lambda[efptCOUL]*fr->mu_tot[1][j]; |
1775 | } |
1776 | } |
1777 | } |
1778 | |
1779 | /* Reset energies */ |
1780 | reset_enerdata(fr, bNS, enerd, MASTER(cr)(((cr)->nodeid == 0) || !((cr)->nnodes > 1))); |
1781 | clear_rvecs(SHIFTS((2*1 +1)*(2*1 +1)*(2*2 +1)), fr->fshift); |
1782 | |
1783 | if (bNS) |
1784 | { |
1785 | wallcycle_start(wcycle, ewcNS); |
1786 | |
1787 | if (graph && bStateChanged) |
1788 | { |
1789 | /* Calculate intramolecular shift vectors to make molecules whole */ |
1790 | mk_mshift(fplog, graph, fr->ePBC, box, x); |
1791 | } |
1792 | |
1793 | /* Do the actual neighbour searching */ |
1794 | ns(fplog, fr, box, |
1795 | groups, top, mdatoms, |
1796 | cr, nrnb, bFillGrid, |
1797 | bDoLongRangeNS); |
1798 | |
1799 | wallcycle_stop(wcycle, ewcNS); |
1800 | } |
1801 | |
1802 | if (inputrec->implicit_solvent && bNS) |
1803 | { |
1804 | make_gb_nblist(cr, inputrec->gb_algorithm, |
1805 | x, box, fr, &top->idef, graph, fr->born); |
1806 | } |
1807 | |
1808 | if (DOMAINDECOMP(cr)(((cr)->dd != ((void*)0)) && ((cr)->nnodes > 1)) && !(cr->duty & DUTY_PME(1<<1))) |
1809 | { |
1810 | wallcycle_start(wcycle, ewcPPDURINGPME); |
1811 | dd_force_flop_start(cr->dd, nrnb); |
1812 | } |
1813 | |
1814 | if (inputrec->bRot) |
1815 | { |
1816 | /* Enforced rotation has its own cycle counter that starts after the collective |
1817 | * coordinates have been communicated. It is added to ddCyclF to allow |
1818 | * for proper load-balancing */ |
1819 | wallcycle_start(wcycle, ewcROT); |
1820 | do_rotation(cr, inputrec, box, x, t, step, wcycle, bNS); |
1821 | wallcycle_stop(wcycle, ewcROT); |
1822 | } |
1823 | |
1824 | /* Start the force cycle counter. |
1825 | * This counter is stopped in do_forcelow_level. |
1826 | * No parallel communication should occur while this counter is running, |
1827 | * since that will interfere with the dynamic load balancing. |
1828 | */ |
1829 | wallcycle_start(wcycle, ewcFORCE); |
1830 | |
1831 | if (bDoForces) |
1832 | { |
1833 | /* Reset forces for which the virial is calculated separately: |
1834 | * PME/Ewald forces if necessary */ |
1835 | if (fr->bF_NoVirSum) |
1836 | { |
1837 | if (flags & GMX_FORCE_VIRIAL(1<<8)) |
1838 | { |
1839 | fr->f_novirsum = fr->f_novirsum_alloc; |
1840 | if (fr->bDomDec) |
1841 | { |
1842 | clear_rvecs(fr->f_novirsum_n, fr->f_novirsum); |
1843 | } |
1844 | else |
1845 | { |
1846 | clear_rvecs(homenr, fr->f_novirsum+start); |
1847 | } |
1848 | } |
1849 | else |
1850 | { |
1851 | /* We are not calculating the pressure so we do not need |
1852 | * a separate array for forces that do not contribute |
1853 | * to the pressure. |
1854 | */ |
1855 | fr->f_novirsum = f; |
1856 | } |
1857 | } |
1858 | |
1859 | /* Clear the short- and long-range forces */ |
1860 | clear_rvecs(fr->natoms_force_constr, f); |
1861 | if (bSepLRF && do_per_step(step, inputrec->nstcalclr)) |
1862 | { |
1863 | clear_rvecs(fr->natoms_force_constr, fr->f_twin); |
1864 | } |
1865 | |
1866 | clear_rvec(fr->vir_diag_posres); |
1867 | } |
1868 | if (inputrec->ePull == epullCONSTRAINT) |
1869 | { |
1870 | clear_pull_forces(inputrec->pull); |
1871 | } |
1872 | |
1873 | /* update QMMMrec, if necessary */ |
1874 | if (fr->bQMMM) |
1875 | { |
1876 | update_QMMMrec(cr, fr, x, mdatoms, box, top); |
1877 | } |
1878 | |
1879 | if ((flags & GMX_FORCE_BONDED(1<<4)) && top->idef.il[F_POSRES].nr > 0) |
1880 | { |
1881 | posres_wrapper(fplog, flags, bSepDVDL, inputrec, nrnb, top, box, x, |
1882 | enerd, lambda, fr); |
1883 | } |
1884 | |
1885 | if ((flags & GMX_FORCE_BONDED(1<<4)) && top->idef.il[F_FBPOSRES].nr > 0) |
1886 | { |
1887 | fbposres_wrapper(inputrec, nrnb, top, box, x, enerd, fr); |
1888 | } |
1889 | |
1890 | /* Compute the bonded and non-bonded energies and optionally forces */ |
1891 | do_force_lowlevel(fplog, step, fr, inputrec, &(top->idef), |
1892 | cr, nrnb, wcycle, mdatoms, |
1893 | x, hist, f, bSepLRF ? fr->f_twin : f, enerd, fcd, top, fr->born, |
1894 | &(top->atomtypes), bBornRadii, box, |
1895 | inputrec->fepvals, lambda, |
1896 | graph, &(top->excls), fr->mu_tot, |
1897 | flags, |
1898 | &cycles_pme); |
1899 | |
1900 | if (bSepLRF) |
1901 | { |
1902 | if (do_per_step(step, inputrec->nstcalclr)) |
1903 | { |
1904 | /* Add the long range forces to the short range forces */ |
1905 | for (i = 0; i < fr->natoms_force_constr; i++) |
1906 | { |
1907 | rvec_add(fr->f_twin[i], f[i], f[i]); |
1908 | } |
1909 | } |
1910 | } |
1911 | |
1912 | cycles_force = wallcycle_stop(wcycle, ewcFORCE); |
1913 | |
1914 | if (ed) |
1915 | { |
1916 | do_flood(cr, inputrec, x, f, ed, box, step, bNS); |
1917 | } |
1918 | |
1919 | if (DOMAINDECOMP(cr)(((cr)->dd != ((void*)0)) && ((cr)->nnodes > 1))) |
1920 | { |
1921 | dd_force_flop_stop(cr->dd, nrnb); |
1922 | if (wcycle) |
1923 | { |
1924 | dd_cycles_add(cr->dd, cycles_force-cycles_pme, ddCyclF); |
1925 | } |
1926 | } |
1927 | |
1928 | if (bDoForces) |
1929 | { |
1930 | if (IR_ELEC_FIELD(*inputrec)((*inputrec).ex[0].n > 0 || (*inputrec).ex[1].n > 0 || ( *inputrec).ex[2].n > 0)) |
1931 | { |
1932 | /* Compute forces due to electric field */ |
1933 | calc_f_el(MASTER(cr)(((cr)->nodeid == 0) || !((cr)->nnodes > 1)) ? field : NULL((void*)0), |
1934 | start, homenr, mdatoms->chargeA, fr->f_novirsum, |
1935 | inputrec->ex, inputrec->et, t); |
1936 | } |
1937 | |
1938 | if (bDoAdressWF && fr->adress_icor == eAdressICThermoForce) |
1939 | { |
1940 | /* Compute thermodynamic force in hybrid AdResS region */ |
1941 | adress_thermo_force(start, homenr, &(top->cgs), x, fr->f_novirsum, fr, mdatoms, |
1942 | inputrec->ePBC == epbcNONE ? NULL((void*)0) : &pbc); |
1943 | } |
1944 | |
1945 | /* Communicate the forces */ |
1946 | if (DOMAINDECOMP(cr)(((cr)->dd != ((void*)0)) && ((cr)->nnodes > 1))) |
1947 | { |
1948 | wallcycle_start(wcycle, ewcMOVEF); |
1949 | dd_move_f(cr->dd, f, fr->fshift); |
1950 | /* Do we need to communicate the separate force array |
1951 | * for terms that do not contribute to the single sum virial? |
1952 | * Position restraints and electric fields do not introduce |
1953 | * inter-cg forces, only full electrostatics methods do. |
1954 | * When we do not calculate the virial, fr->f_novirsum = f, |
1955 | * so we have already communicated these forces. |
1956 | */ |
1957 | if (EEL_FULL(fr->eeltype)((((fr->eeltype) == eelPME || (fr->eeltype) == eelPMESWITCH || (fr->eeltype) == eelPMEUSER || (fr->eeltype) == eelPMEUSERSWITCH || (fr->eeltype) == eelP3M_AD) || (fr->eeltype) == eelEWALD ) || (fr->eeltype) == eelPOISSON) && cr->dd->n_intercg_excl && |
1958 | (flags & GMX_FORCE_VIRIAL(1<<8))) |
1959 | { |
1960 | dd_move_f(cr->dd, fr->f_novirsum, NULL((void*)0)); |
1961 | } |
1962 | if (bSepLRF) |
1963 | { |
1964 | /* We should not update the shift forces here, |
1965 | * since f_twin is already included in f. |
1966 | */ |
1967 | dd_move_f(cr->dd, fr->f_twin, NULL((void*)0)); |
1968 | } |
1969 | wallcycle_stop(wcycle, ewcMOVEF); |
1970 | } |
1971 | |
1972 | /* If we have NoVirSum forces, but we do not calculate the virial, |
1973 | * we sum fr->f_novirum=f later. |
1974 | */ |
1975 | if (vsite && !(fr->bF_NoVirSum && !(flags & GMX_FORCE_VIRIAL(1<<8)))) |
1976 | { |
1977 | wallcycle_start(wcycle, ewcVSITESPREAD); |
1978 | spread_vsite_f(vsite, x, f, fr->fshift, FALSE0, NULL((void*)0), nrnb, |
1979 | &top->idef, fr->ePBC, fr->bMolPBC, graph, box, cr); |
1980 | wallcycle_stop(wcycle, ewcVSITESPREAD); |
1981 | |
1982 | if (bSepLRF) |
1983 | { |
1984 | wallcycle_start(wcycle, ewcVSITESPREAD); |
1985 | spread_vsite_f(vsite, x, fr->f_twin, NULL((void*)0), FALSE0, NULL((void*)0), |
1986 | nrnb, |
1987 | &top->idef, fr->ePBC, fr->bMolPBC, graph, box, cr); |
1988 | wallcycle_stop(wcycle, ewcVSITESPREAD); |
1989 | } |
1990 | } |
1991 | |
1992 | if (flags & GMX_FORCE_VIRIAL(1<<8)) |
1993 | { |
1994 | /* Calculation of the virial must be done after vsites! */ |
1995 | calc_virial(0, mdatoms->homenr, x, f, |
1996 | vir_force, graph, box, nrnb, fr, inputrec->ePBC); |
1997 | } |
1998 | } |
1999 | |
2000 | if (inputrec->ePull == epullUMBRELLA || inputrec->ePull == epullCONST_F) |
2001 | { |
2002 | pull_potential_wrapper(fplog, bSepDVDL, cr, inputrec, box, x, |
2003 | f, vir_force, mdatoms, enerd, lambda, t); |
2004 | } |
2005 | |
2006 | /* Add the forces from enforced rotation potentials (if any) */ |
2007 | if (inputrec->bRot) |
2008 | { |
2009 | wallcycle_start(wcycle, ewcROTadd); |
2010 | enerd->term[F_COM_PULL] += add_rot_forces(inputrec->rot, f, cr, step, t); |
2011 | wallcycle_stop(wcycle, ewcROTadd); |
2012 | } |
2013 | |
2014 | /* Add forces from interactive molecular dynamics (IMD), if bIMD == TRUE. */ |
2015 | IMD_apply_forces(inputrec->bIMD, inputrec->imd, cr, f, wcycle); |
2016 | |
2017 | if (PAR(cr)((cr)->nnodes > 1) && !(cr->duty & DUTY_PME(1<<1))) |
2018 | { |
2019 | /* In case of node-splitting, the PP nodes receive the long-range |
2020 | * forces, virial and energy from the PME nodes here. |
2021 | */ |
2022 | pme_receive_force_ener(fplog, bSepDVDL, cr, wcycle, enerd, fr); |
2023 | } |
2024 | |
2025 | if (bDoForces) |
2026 | { |
2027 | post_process_forces(cr, step, nrnb, wcycle, |
2028 | top, box, x, f, vir_force, mdatoms, graph, fr, vsite, |
2029 | flags); |
2030 | } |
2031 | |
2032 | /* Sum the potential energy terms from group contributions */ |
2033 | sum_epot(&(enerd->grpp), enerd->term); |
2034 | } |
2035 | |
2036 | void do_force(FILE *fplog, t_commrec *cr, |
2037 | t_inputrec *inputrec, |
2038 | gmx_int64_t step, t_nrnb *nrnb, gmx_wallcycle_t wcycle, |
2039 | gmx_localtop_t *top, |
2040 | gmx_groups_t *groups, |
2041 | matrix box, rvec x[], history_t *hist, |
2042 | rvec f[], |
2043 | tensor vir_force, |
2044 | t_mdatoms *mdatoms, |
2045 | gmx_enerdata_t *enerd, t_fcdata *fcd, |
2046 | real *lambda, t_graph *graph, |
2047 | t_forcerec *fr, |
2048 | gmx_vsite_t *vsite, rvec mu_tot, |
2049 | double t, FILE *field, gmx_edsam_t ed, |
2050 | gmx_bool bBornRadii, |
2051 | int flags) |
2052 | { |
2053 | /* modify force flag if not doing nonbonded */ |
2054 | if (!fr->bNonbonded) |
2055 | { |
2056 | flags &= ~GMX_FORCE_NONBONDED(1<<6); |
2057 | } |
2058 | |
2059 | switch (inputrec->cutoff_scheme) |
2060 | { |
2061 | case ecutsVERLET: |
2062 | do_force_cutsVERLET(fplog, cr, inputrec, |
2063 | step, nrnb, wcycle, |
2064 | top, |
2065 | groups, |
2066 | box, x, hist, |
2067 | f, vir_force, |
2068 | mdatoms, |
2069 | enerd, fcd, |
2070 | lambda, graph, |
2071 | fr, fr->ic, |
2072 | vsite, mu_tot, |
2073 | t, field, ed, |
2074 | bBornRadii, |
2075 | flags); |
2076 | break; |
2077 | case ecutsGROUP: |
2078 | do_force_cutsGROUP(fplog, cr, inputrec, |
2079 | step, nrnb, wcycle, |
2080 | top, |
2081 | groups, |
2082 | box, x, hist, |
2083 | f, vir_force, |
2084 | mdatoms, |
2085 | enerd, fcd, |
2086 | lambda, graph, |
2087 | fr, vsite, mu_tot, |
2088 | t, field, ed, |
2089 | bBornRadii, |
2090 | flags); |
2091 | break; |
2092 | default: |
2093 | gmx_incons("Invalid cut-off scheme passed!")_gmx_error("incons", "Invalid cut-off scheme passed!", "/home/alexxy/Develop/gromacs/src/gromacs/mdlib/sim_util.c" , 2093); |
2094 | } |
2095 | } |
2096 | |
2097 | |
2098 | void do_constrain_first(FILE *fplog, gmx_constr_t constr, |
2099 | t_inputrec *ir, t_mdatoms *md, |
2100 | t_state *state, t_commrec *cr, t_nrnb *nrnb, |
2101 | t_forcerec *fr, gmx_localtop_t *top) |
2102 | { |
2103 | int i, m, start, end; |
2104 | gmx_int64_t step; |
2105 | real dt = ir->delta_t; |
2106 | real dvdl_dum; |
2107 | rvec *savex; |
2108 | |
2109 | snew(savex, state->natoms)(savex) = save_calloc("savex", "/home/alexxy/Develop/gromacs/src/gromacs/mdlib/sim_util.c" , 2109, (state->natoms), sizeof(*(savex))); |
2110 | |
2111 | start = 0; |
2112 | end = md->homenr; |
2113 | |
2114 | if (debug) |
2115 | { |
2116 | fprintf(debug, "vcm: start=%d, homenr=%d, end=%d\n", |
2117 | start, md->homenr, end); |
2118 | } |
2119 | /* Do a first constrain to reset particles... */ |
2120 | step = ir->init_step; |
2121 | if (fplog) |
2122 | { |
2123 | char buf[STEPSTRSIZE22]; |
2124 | fprintf(fplog, "\nConstraining the starting coordinates (step %s)\n", |
2125 | gmx_step_str(step, buf)); |
2126 | } |
2127 | dvdl_dum = 0; |
2128 | |
2129 | /* constrain the current position */ |
2130 | constrain(NULL((void*)0), TRUE1, FALSE0, constr, &(top->idef), |
2131 | ir, NULL((void*)0), cr, step, 0, md, |
2132 | state->x, state->x, NULL((void*)0), |
2133 | fr->bMolPBC, state->box, |
2134 | state->lambda[efptBONDED], &dvdl_dum, |
2135 | NULL((void*)0), NULL((void*)0), nrnb, econqCoord, |
2136 | ir->epc == epcMTTK, state->veta, state->veta); |
2137 | if (EI_VV(ir->eI)((ir->eI) == eiVV || (ir->eI) == eiVVAK)) |
2138 | { |
2139 | /* constrain the inital velocity, and save it */ |
2140 | /* also may be useful if we need the ekin from the halfstep for velocity verlet */ |
2141 | /* might not yet treat veta correctly */ |
2142 | constrain(NULL((void*)0), TRUE1, FALSE0, constr, &(top->idef), |
2143 | ir, NULL((void*)0), cr, step, 0, md, |
2144 | state->x, state->v, state->v, |
2145 | fr->bMolPBC, state->box, |
2146 | state->lambda[efptBONDED], &dvdl_dum, |
2147 | NULL((void*)0), NULL((void*)0), nrnb, econqVeloc, |
2148 | ir->epc == epcMTTK, state->veta, state->veta); |
2149 | } |
2150 | /* constrain the inital velocities at t-dt/2 */ |
2151 | if (EI_STATE_VELOCITY(ir->eI)(((ir->eI) == eiMD || ((ir->eI) == eiVV || (ir->eI) == eiVVAK)) || ((ir->eI) == eiSD1 || (ir->eI) == eiSD2)) && ir->eI != eiVV) |
2152 | { |
2153 | for (i = start; (i < end); i++) |
2154 | { |
2155 | for (m = 0; (m < DIM3); m++) |
2156 | { |
2157 | /* Reverse the velocity */ |
2158 | state->v[i][m] = -state->v[i][m]; |
2159 | /* Store the position at t-dt in buf */ |
2160 | savex[i][m] = state->x[i][m] + dt*state->v[i][m]; |
2161 | } |
2162 | } |
2163 | /* Shake the positions at t=-dt with the positions at t=0 |
2164 | * as reference coordinates. |
2165 | */ |
2166 | if (fplog) |
2167 | { |
2168 | char buf[STEPSTRSIZE22]; |
2169 | fprintf(fplog, "\nConstraining the coordinates at t0-dt (step %s)\n", |
2170 | gmx_step_str(step, buf)); |
2171 | } |
2172 | dvdl_dum = 0; |
2173 | constrain(NULL((void*)0), TRUE1, FALSE0, constr, &(top->idef), |
2174 | ir, NULL((void*)0), cr, step, -1, md, |
2175 | state->x, savex, NULL((void*)0), |
2176 | fr->bMolPBC, state->box, |
2177 | state->lambda[efptBONDED], &dvdl_dum, |
2178 | state->v, NULL((void*)0), nrnb, econqCoord, |
2179 | ir->epc == epcMTTK, state->veta, state->veta); |
2180 | |
2181 | for (i = start; i < end; i++) |
2182 | { |
2183 | for (m = 0; m < DIM3; m++) |
2184 | { |
2185 | /* Re-reverse the velocities */ |
2186 | state->v[i][m] = -state->v[i][m]; |
2187 | } |
2188 | } |
2189 | } |
2190 | sfree(savex)save_free("savex", "/home/alexxy/Develop/gromacs/src/gromacs/mdlib/sim_util.c" , 2190, (savex)); |
2191 | } |
2192 | |
2193 | |
2194 | static void |
2195 | integrate_table(real vdwtab[], real scale, int offstart, int rstart, int rend, |
2196 | double *enerout, double *virout) |
2197 | { |
2198 | double enersum, virsum; |
2199 | double invscale, invscale2, invscale3; |
2200 | double r, ea, eb, ec, pa, pb, pc, pd; |
2201 | double y0, f, g, h; |
2202 | int ri, offset, tabfactor; |
2203 | |
2204 | invscale = 1.0/scale; |
2205 | invscale2 = invscale*invscale; |
2206 | invscale3 = invscale*invscale2; |
2207 | |
2208 | /* Following summation derived from cubic spline definition, |
2209 | * Numerical Recipies in C, second edition, p. 113-116. Exact for |
2210 | * the cubic spline. We first calculate the negative of the |
2211 | * energy from rvdw to rvdw_switch, assuming that g(r)=1, and then |
2212 | * add the more standard, abrupt cutoff correction to that result, |
2213 | * yielding the long-range correction for a switched function. We |
2214 | * perform both the pressure and energy loops at the same time for |
2215 | * simplicity, as the computational cost is low. */ |
2216 | |
2217 | if (offstart == 0) |
2218 | { |
2219 | /* Since the dispersion table has been scaled down a factor |
2220 | * 6.0 and the repulsion a factor 12.0 to compensate for the |
2221 | * c6/c12 parameters inside nbfp[] being scaled up (to save |
2222 | * flops in kernels), we need to correct for this. |
2223 | */ |
2224 | tabfactor = 6.0; |
2225 | } |
2226 | else |
2227 | { |
2228 | tabfactor = 12.0; |
2229 | } |
2230 | |
2231 | enersum = 0.0; |
2232 | virsum = 0.0; |
2233 | for (ri = rstart; ri < rend; ++ri) |
2234 | { |
2235 | r = ri*invscale; |
2236 | ea = invscale3; |
2237 | eb = 2.0*invscale2*r; |
2238 | ec = invscale*r*r; |
2239 | |
2240 | pa = invscale3; |
2241 | pb = 3.0*invscale2*r; |
2242 | pc = 3.0*invscale*r*r; |
2243 | pd = r*r*r; |
2244 | |
2245 | /* this "8" is from the packing in the vdwtab array - perhaps |
2246 | should be defined? */ |
2247 | |
2248 | offset = 8*ri + offstart; |
2249 | y0 = vdwtab[offset]; |
2250 | f = vdwtab[offset+1]; |
2251 | g = vdwtab[offset+2]; |
2252 | h = vdwtab[offset+3]; |
2253 | |
2254 | enersum += y0*(ea/3 + eb/2 + ec) + f*(ea/4 + eb/3 + ec/2) + g*(ea/5 + eb/4 + ec/3) + h*(ea/6 + eb/5 + ec/4); |
2255 | virsum += f*(pa/4 + pb/3 + pc/2 + pd) + 2*g*(pa/5 + pb/4 + pc/3 + pd/2) + 3*h*(pa/6 + pb/5 + pc/4 + pd/3); |
2256 | } |
2257 | *enerout = 4.0*M_PI3.14159265358979323846*enersum*tabfactor; |
2258 | *virout = 4.0*M_PI3.14159265358979323846*virsum*tabfactor; |
2259 | } |
2260 | |
2261 | void calc_enervirdiff(FILE *fplog, int eDispCorr, t_forcerec *fr) |
2262 | { |
2263 | double eners[2], virs[2], enersum, virsum, y0, f, g, h; |
2264 | double r0, r1, r, rc3, rc9, ea, eb, ec, pa, pb, pc, pd; |
2265 | double invscale, invscale2, invscale3; |
2266 | int ri0, ri1, ri, i, offstart, offset; |
2267 | real scale, *vdwtab, tabfactor, tmp; |
2268 | |
2269 | fr->enershiftsix = 0; |
2270 | fr->enershifttwelve = 0; |
2271 | fr->enerdiffsix = 0; |
2272 | fr->enerdifftwelve = 0; |
2273 | fr->virdiffsix = 0; |
2274 | fr->virdifftwelve = 0; |
2275 | |
2276 | if (eDispCorr != edispcNO) |
2277 | { |
2278 | for (i = 0; i < 2; i++) |
2279 | { |
2280 | eners[i] = 0; |
2281 | virs[i] = 0; |
2282 | } |
2283 | if (fr->vdwtype == evdwSWITCH || fr->vdwtype == evdwSHIFT || |
2284 | fr->vdw_modifier == eintmodPOTSWITCH || |
2285 | fr->vdw_modifier == eintmodFORCESWITCH) |
2286 | { |
2287 | if (fr->rvdw_switch == 0) |
2288 | { |
2289 | gmx_fatal(FARGS0, "/home/alexxy/Develop/gromacs/src/gromacs/mdlib/sim_util.c" , 2289, |
2290 | "With dispersion correction rvdw-switch can not be zero " |
2291 | "for vdw-type = %s", evdw_names[fr->vdwtype]); |
2292 | } |
2293 | |
2294 | scale = fr->nblists[0].table_elec_vdw.scale; |
2295 | vdwtab = fr->nblists[0].table_vdw.data; |
2296 | |
2297 | /* Round the cut-offs to exact table values for precision */ |
2298 | ri0 = floor(fr->rvdw_switch*scale); |
2299 | ri1 = ceil(fr->rvdw*scale); |
2300 | r0 = ri0/scale; |
2301 | r1 = ri1/scale; |
2302 | rc3 = r0*r0*r0; |
2303 | rc9 = rc3*rc3*rc3; |
2304 | |
2305 | if (fr->vdwtype == evdwSHIFT || |
2306 | fr->vdw_modifier == eintmodFORCESWITCH) |
2307 | { |
2308 | /* Determine the constant energy shift below rvdw_switch. |
2309 | * Table has a scale factor since we have scaled it down to compensate |
2310 | * for scaling-up c6/c12 with the derivative factors to save flops in analytical kernels. |
2311 | */ |
2312 | fr->enershiftsix = (real)(-1.0/(rc3*rc3)) - 6.0*vdwtab[8*ri0]; |
2313 | fr->enershifttwelve = (real)( 1.0/(rc9*rc3)) - 12.0*vdwtab[8*ri0 + 4]; |
2314 | } |
2315 | /* Add the constant part from 0 to rvdw_switch. |
2316 | * This integration from 0 to rvdw_switch overcounts the number |
2317 | * of interactions by 1, as it also counts the self interaction. |
2318 | * We will correct for this later. |
2319 | */ |
2320 | eners[0] += 4.0*M_PI3.14159265358979323846*fr->enershiftsix*rc3/3.0; |
2321 | eners[1] += 4.0*M_PI3.14159265358979323846*fr->enershifttwelve*rc3/3.0; |
2322 | for (i = 0; i < 2; i++) |
2323 | { |
2324 | enersum = 0; |
2325 | virsum = 0; |
2326 | integrate_table(vdwtab, scale, (i == 0 ? 0 : 4), ri0, ri1, &enersum, &virsum); |
2327 | eners[i] -= enersum; |
2328 | virs[i] -= virsum; |
2329 | } |
2330 | |
2331 | /* now add the correction for rvdw_switch to infinity */ |
2332 | eners[0] += -4.0*M_PI3.14159265358979323846/(3.0*rc3); |
2333 | eners[1] += 4.0*M_PI3.14159265358979323846/(9.0*rc9); |
2334 | virs[0] += 8.0*M_PI3.14159265358979323846/rc3; |
2335 | virs[1] += -16.0*M_PI3.14159265358979323846/(3.0*rc9); |
2336 | } |
2337 | else if (fr->vdwtype == evdwCUT || |
2338 | EVDW_PME(fr->vdwtype)((fr->vdwtype) == evdwPME) || |
2339 | fr->vdwtype == evdwUSER) |
2340 | { |
2341 | if (fr->vdwtype == evdwUSER && fplog) |
2342 | { |
2343 | fprintf(fplog, |
2344 | "WARNING: using dispersion correction with user tables\n"); |
2345 | } |
2346 | |
2347 | /* Note that with LJ-PME, the dispersion correction is multiplied |
2348 | * by the difference between the actual C6 and the value of C6 |
2349 | * that would produce the combination rule. |
2350 | * This means the normal energy and virial difference formulas |
2351 | * can be used here. |
2352 | */ |
2353 | |
2354 | rc3 = fr->rvdw*fr->rvdw*fr->rvdw; |
2355 | rc9 = rc3*rc3*rc3; |
2356 | /* Contribution beyond the cut-off */ |
2357 | eners[0] += -4.0*M_PI3.14159265358979323846/(3.0*rc3); |
2358 | eners[1] += 4.0*M_PI3.14159265358979323846/(9.0*rc9); |
2359 | if (fr->vdw_modifier == eintmodPOTSHIFT) |
2360 | { |
2361 | /* Contribution within the cut-off */ |
2362 | eners[0] += -4.0*M_PI3.14159265358979323846/(3.0*rc3); |
2363 | eners[1] += 4.0*M_PI3.14159265358979323846/(3.0*rc9); |
2364 | } |
2365 | /* Contribution beyond the cut-off */ |
2366 | virs[0] += 8.0*M_PI3.14159265358979323846/rc3; |
2367 | virs[1] += -16.0*M_PI3.14159265358979323846/(3.0*rc9); |
2368 | } |
2369 | else |
2370 | { |
2371 | gmx_fatal(FARGS0, "/home/alexxy/Develop/gromacs/src/gromacs/mdlib/sim_util.c" , 2371, |
2372 | "Dispersion correction is not implemented for vdw-type = %s", |
2373 | evdw_names[fr->vdwtype]); |
2374 | } |
2375 | |
2376 | /* TODO: remove this code once we have group LJ-PME kernels |
2377 | * that calculate the exact, full LJ param C6/r^6 within the cut-off, |
2378 | * as the current nbnxn kernels do. |
2379 | */ |
2380 | if (fr->vdwtype == evdwPME && fr->cutoff_scheme == ecutsGROUP) |
2381 | { |
2382 | /* Calculate self-interaction coefficient (assuming that |
2383 | * the reciprocal-space contribution is constant in the |
2384 | * region that contributes to the self-interaction). |
2385 | */ |
2386 | fr->enershiftsix = pow(fr->ewaldcoeff_lj, 6) / 6.0; |
2387 | |
2388 | eners[0] += -pow(sqrt(M_PI3.14159265358979323846)*fr->ewaldcoeff_lj, 3)/3.0; |
2389 | virs[0] += pow(sqrt(M_PI3.14159265358979323846)*fr->ewaldcoeff_lj, 3); |
2390 | } |
2391 | |
2392 | fr->enerdiffsix = eners[0]; |
2393 | fr->enerdifftwelve = eners[1]; |
2394 | /* The 0.5 is due to the Gromacs definition of the virial */ |
2395 | fr->virdiffsix = 0.5*virs[0]; |
2396 | fr->virdifftwelve = 0.5*virs[1]; |
2397 | } |
2398 | } |
2399 | |
2400 | void calc_dispcorr(FILE *fplog, t_inputrec *ir, t_forcerec *fr, |
2401 | gmx_int64_t step, int natoms, |
2402 | matrix box, real lambda, tensor pres, tensor virial, |
2403 | real *prescorr, real *enercorr, real *dvdlcorr) |
2404 | { |
2405 | gmx_bool bCorrAll, bCorrPres; |
2406 | real dvdlambda, invvol, dens, ninter, avcsix, avctwelve, enerdiff, svir = 0, spres = 0; |
2407 | int m; |
2408 | |
2409 | *prescorr = 0; |
2410 | *enercorr = 0; |
2411 | *dvdlcorr = 0; |
2412 | |
2413 | clear_mat(virial); |
2414 | clear_mat(pres); |
2415 | |
2416 | if (ir->eDispCorr != edispcNO) |
2417 | { |
2418 | bCorrAll = (ir->eDispCorr == edispcAllEner || |
2419 | ir->eDispCorr == edispcAllEnerPres); |
2420 | bCorrPres = (ir->eDispCorr == edispcEnerPres || |
2421 | ir->eDispCorr == edispcAllEnerPres); |
2422 | |
2423 | invvol = 1/det(box); |
2424 | if (fr->n_tpi) |
2425 | { |
2426 | /* Only correct for the interactions with the inserted molecule */ |
2427 | dens = (natoms - fr->n_tpi)*invvol; |
2428 | ninter = fr->n_tpi; |
2429 | } |
2430 | else |
2431 | { |
2432 | dens = natoms*invvol; |
2433 | ninter = 0.5*natoms; |
2434 | } |
2435 | |
2436 | if (ir->efep == efepNO) |
2437 | { |
2438 | avcsix = fr->avcsix[0]; |
2439 | avctwelve = fr->avctwelve[0]; |
2440 | } |
2441 | else |
2442 | { |
2443 | avcsix = (1 - lambda)*fr->avcsix[0] + lambda*fr->avcsix[1]; |
2444 | avctwelve = (1 - lambda)*fr->avctwelve[0] + lambda*fr->avctwelve[1]; |
2445 | } |
2446 | |
2447 | enerdiff = ninter*(dens*fr->enerdiffsix - fr->enershiftsix); |
2448 | *enercorr += avcsix*enerdiff; |
2449 | dvdlambda = 0.0; |
2450 | if (ir->efep != efepNO) |
2451 | { |
2452 | dvdlambda += (fr->avcsix[1] - fr->avcsix[0])*enerdiff; |
2453 | } |
2454 | if (bCorrAll) |
2455 | { |
2456 | enerdiff = ninter*(dens*fr->enerdifftwelve - fr->enershifttwelve); |
2457 | *enercorr += avctwelve*enerdiff; |
2458 | if (fr->efep != efepNO) |
2459 | { |
2460 | dvdlambda += (fr->avctwelve[1] - fr->avctwelve[0])*enerdiff; |
2461 | } |
2462 | } |
2463 | |
2464 | if (bCorrPres) |
2465 | { |
2466 | svir = ninter*dens*avcsix*fr->virdiffsix/3.0; |
2467 | if (ir->eDispCorr == edispcAllEnerPres) |
2468 | { |
2469 | svir += ninter*dens*avctwelve*fr->virdifftwelve/3.0; |
2470 | } |
2471 | /* The factor 2 is because of the Gromacs virial definition */ |
2472 | spres = -2.0*invvol*svir*PRESFAC(16.6054); |
2473 | |
2474 | for (m = 0; m < DIM3; m++) |
2475 | { |
2476 | virial[m][m] += svir; |
2477 | pres[m][m] += spres; |
2478 | } |
2479 | *prescorr += spres; |
2480 | } |
2481 | |
2482 | /* Can't currently control when it prints, for now, just print when degugging */ |
2483 | if (debug) |
2484 | { |
2485 | if (bCorrAll) |
2486 | { |
2487 | fprintf(debug, "Long Range LJ corr.: <C6> %10.4e, <C12> %10.4e\n", |
2488 | avcsix, avctwelve); |
2489 | } |
2490 | if (bCorrPres) |
2491 | { |
2492 | fprintf(debug, |
2493 | "Long Range LJ corr.: Epot %10g, Pres: %10g, Vir: %10g\n", |
2494 | *enercorr, spres, svir); |
2495 | } |
2496 | else |
2497 | { |
2498 | fprintf(debug, "Long Range LJ corr.: Epot %10g\n", *enercorr); |
2499 | } |
2500 | } |
2501 | |
2502 | if (fr->bSepDVDL && do_per_step(step, ir->nstlog)) |
2503 | { |
2504 | gmx_print_sepdvdl(fplog, "Dispersion correction", *enercorr, dvdlambda); |
2505 | } |
2506 | if (fr->efep != efepNO) |
2507 | { |
2508 | *dvdlcorr += dvdlambda; |
2509 | } |
2510 | } |
2511 | } |
2512 | |
2513 | void do_pbc_first(FILE *fplog, matrix box, t_forcerec *fr, |
2514 | t_graph *graph, rvec x[]) |
2515 | { |
2516 | if (fplog) |
2517 | { |
2518 | fprintf(fplog, "Removing pbc first time\n"); |
2519 | } |
2520 | calc_shifts(box, fr->shift_vec); |
2521 | if (graph) |
2522 | { |
2523 | mk_mshift(fplog, graph, fr->ePBC, box, x); |
2524 | if (gmx_debug_at) |
2525 | { |
2526 | p_graph(debug, "do_pbc_first 1", graph); |
2527 | } |
2528 | shift_self(graph, box, x); |
2529 | /* By doing an extra mk_mshift the molecules that are broken |
2530 | * because they were e.g. imported from another software |
2531 | * will be made whole again. Such are the healing powers |
2532 | * of GROMACS. |
2533 | */ |
2534 | mk_mshift(fplog, graph, fr->ePBC, box, x); |
2535 | if (gmx_debug_at) |
2536 | { |
2537 | p_graph(debug, "do_pbc_first 2", graph); |
2538 | } |
2539 | } |
2540 | if (fplog) |
2541 | { |
2542 | fprintf(fplog, "Done rmpbc\n"); |
2543 | } |
2544 | } |
2545 | |
2546 | static void low_do_pbc_mtop(FILE *fplog, int ePBC, matrix box, |
2547 | gmx_mtop_t *mtop, rvec x[], |
2548 | gmx_bool bFirst) |
2549 | { |
2550 | t_graph *graph; |
2551 | int mb, as, mol; |
2552 | gmx_molblock_t *molb; |
2553 | |
2554 | if (bFirst && fplog) |
2555 | { |
2556 | fprintf(fplog, "Removing pbc first time\n"); |
2557 | } |
2558 | |
2559 | snew(graph, 1)(graph) = save_calloc("graph", "/home/alexxy/Develop/gromacs/src/gromacs/mdlib/sim_util.c" , 2559, (1), sizeof(*(graph))); |
2560 | as = 0; |
2561 | for (mb = 0; mb < mtop->nmolblock; mb++) |
2562 | { |
2563 | molb = &mtop->molblock[mb]; |
2564 | if (molb->natoms_mol == 1 || |
2565 | (!bFirst && mtop->moltype[molb->type].cgs.nr == 1)) |
2566 | { |
2567 | /* Just one atom or charge group in the molecule, no PBC required */ |
2568 | as += molb->nmol*molb->natoms_mol; |
2569 | } |
2570 | else |
2571 | { |
2572 | /* Pass NULL iso fplog to avoid graph prints for each molecule type */ |
2573 | mk_graph_ilist(NULL((void*)0), mtop->moltype[molb->type].ilist, |
2574 | 0, molb->natoms_mol, FALSE0, FALSE0, graph); |
2575 | |
2576 | for (mol = 0; mol < molb->nmol; mol++) |
2577 | { |
2578 | mk_mshift(fplog, graph, ePBC, box, x+as); |
2579 | |
2580 | shift_self(graph, box, x+as); |
2581 | /* The molecule is whole now. |
2582 | * We don't need the second mk_mshift call as in do_pbc_first, |
2583 | * since we no longer need this graph. |
2584 | */ |
2585 | |
2586 | as += molb->natoms_mol; |
2587 | } |
2588 | done_graph(graph); |
2589 | } |
2590 | } |
2591 | sfree(graph)save_free("graph", "/home/alexxy/Develop/gromacs/src/gromacs/mdlib/sim_util.c" , 2591, (graph)); |
2592 | } |
2593 | |
2594 | void do_pbc_first_mtop(FILE *fplog, int ePBC, matrix box, |
2595 | gmx_mtop_t *mtop, rvec x[]) |
2596 | { |
2597 | low_do_pbc_mtop(fplog, ePBC, box, mtop, x, TRUE1); |
2598 | } |
2599 | |
2600 | void do_pbc_mtop(FILE *fplog, int ePBC, matrix box, |
2601 | gmx_mtop_t *mtop, rvec x[]) |
2602 | { |
2603 | low_do_pbc_mtop(fplog, ePBC, box, mtop, x, FALSE0); |
2604 | } |
2605 | |
2606 | void finish_run(FILE *fplog, t_commrec *cr, |
2607 | t_inputrec *inputrec, |
2608 | t_nrnb nrnb[], gmx_wallcycle_t wcycle, |
2609 | gmx_walltime_accounting_t walltime_accounting, |
2610 | wallclock_gpu_t *gputimes, |
2611 | gmx_bool bWriteStat) |
2612 | { |
2613 | int i, j; |
2614 | t_nrnb *nrnb_tot = NULL((void*)0); |
2615 | real delta_t; |
2616 | double nbfs, mflop; |
2617 | double elapsed_time, |
2618 | elapsed_time_over_all_ranks, |
2619 | elapsed_time_over_all_threads, |
2620 | elapsed_time_over_all_threads_over_all_ranks; |
2621 | wallcycle_sum(cr, wcycle); |
2622 | |
2623 | if (cr->nnodes > 1) |
2624 | { |
2625 | snew(nrnb_tot, 1)(nrnb_tot) = save_calloc("nrnb_tot", "/home/alexxy/Develop/gromacs/src/gromacs/mdlib/sim_util.c" , 2625, (1), sizeof(*(nrnb_tot))); |
2626 | #ifdef GMX_MPI |
2627 | MPI_AllreducetMPI_Allreduce(nrnb->n, nrnb_tot->n, eNRNB, MPI_DOUBLETMPI_DOUBLE, MPI_SUMTMPI_SUM, |
2628 | cr->mpi_comm_mysim); |
2629 | #endif |
2630 | } |
2631 | else |
2632 | { |
2633 | nrnb_tot = nrnb; |
2634 | } |
2635 | |
2636 | elapsed_time = walltime_accounting_get_elapsed_time(walltime_accounting); |
2637 | elapsed_time_over_all_ranks = elapsed_time; |
2638 | elapsed_time_over_all_threads = walltime_accounting_get_elapsed_time_over_all_threads(walltime_accounting); |
2639 | elapsed_time_over_all_threads_over_all_ranks = elapsed_time_over_all_threads; |
2640 | #ifdef GMX_MPI |
2641 | if (cr->nnodes > 1) |
2642 | { |
2643 | /* reduce elapsed_time over all MPI ranks in the current simulation */ |
2644 | MPI_AllreducetMPI_Allreduce(&elapsed_time, |
2645 | &elapsed_time_over_all_ranks, |
2646 | 1, MPI_DOUBLETMPI_DOUBLE, MPI_SUMTMPI_SUM, |
2647 | cr->mpi_comm_mysim); |
2648 | elapsed_time_over_all_ranks /= cr->nnodes; |
2649 | /* Reduce elapsed_time_over_all_threads over all MPI ranks in the |
2650 | * current simulation. */ |
2651 | MPI_AllreducetMPI_Allreduce(&elapsed_time_over_all_threads, |
2652 | &elapsed_time_over_all_threads_over_all_ranks, |
2653 | 1, MPI_DOUBLETMPI_DOUBLE, MPI_SUMTMPI_SUM, |
2654 | cr->mpi_comm_mysim); |
2655 | } |
2656 | #endif |
2657 | |
2658 | if (SIMMASTER(cr)(((((cr)->nodeid == 0) || !((cr)->nnodes > 1)) && ((cr)->duty & (1<<0))) || !((cr)->nnodes > 1))) |
2659 | { |
2660 | print_flop(fplog, nrnb_tot, &nbfs, &mflop); |
2661 | } |
2662 | if (cr->nnodes > 1) |
2663 | { |
2664 | sfree(nrnb_tot)save_free("nrnb_tot", "/home/alexxy/Develop/gromacs/src/gromacs/mdlib/sim_util.c" , 2664, (nrnb_tot)); |
2665 | } |
2666 | |
2667 | if ((cr->duty & DUTY_PP(1<<0)) && DOMAINDECOMP(cr)(((cr)->dd != ((void*)0)) && ((cr)->nnodes > 1))) |
2668 | { |
2669 | print_dd_statistics(cr, inputrec, fplog); |
2670 | } |
2671 | |
2672 | if (SIMMASTER(cr)(((((cr)->nodeid == 0) || !((cr)->nnodes > 1)) && ((cr)->duty & (1<<0))) || !((cr)->nnodes > 1))) |
2673 | { |
2674 | wallcycle_print(fplog, cr->nnodes, cr->npmenodes, |
2675 | elapsed_time_over_all_ranks, |
2676 | wcycle, gputimes); |
2677 | |
2678 | if (EI_DYNAMICS(inputrec->eI)(((inputrec->eI) == eiMD || ((inputrec->eI) == eiVV || ( inputrec->eI) == eiVVAK)) || ((inputrec->eI) == eiSD1 || (inputrec->eI) == eiSD2) || (inputrec->eI) == eiBD)) |
2679 | { |
2680 | delta_t = inputrec->delta_t; |
2681 | } |
2682 | else |
2683 | { |
2684 | delta_t = 0; |
2685 | } |
2686 | |
2687 | if (fplog) |
2688 | { |
2689 | print_perf(fplog, elapsed_time_over_all_threads_over_all_ranks, |
2690 | elapsed_time_over_all_ranks, |
2691 | walltime_accounting_get_nsteps_done(walltime_accounting), |
2692 | delta_t, nbfs, mflop); |
2693 | } |
2694 | if (bWriteStat) |
2695 | { |
2696 | print_perf(stderrstderr, elapsed_time_over_all_threads_over_all_ranks, |
2697 | elapsed_time_over_all_ranks, |
2698 | walltime_accounting_get_nsteps_done(walltime_accounting), |
2699 | delta_t, nbfs, mflop); |
2700 | } |
2701 | } |
2702 | } |
2703 | |
2704 | extern void initialize_lambdas(FILE *fplog, t_inputrec *ir, int *fep_state, real *lambda, double *lam0) |
2705 | { |
2706 | /* this function works, but could probably use a logic rewrite to keep all the different |
2707 | types of efep straight. */ |
2708 | |
2709 | int i; |
2710 | t_lambda *fep = ir->fepvals; |
2711 | |
2712 | if ((ir->efep == efepNO) && (ir->bSimTemp == FALSE0)) |
2713 | { |
2714 | for (i = 0; i < efptNR; i++) |
2715 | { |
2716 | lambda[i] = 0.0; |
2717 | if (lam0) |
2718 | { |
2719 | lam0[i] = 0.0; |
2720 | } |
2721 | } |
2722 | return; |
2723 | } |
2724 | else |
2725 | { |
2726 | *fep_state = fep->init_fep_state; /* this might overwrite the checkpoint |
2727 | if checkpoint is set -- a kludge is in for now |
2728 | to prevent this.*/ |
2729 | for (i = 0; i < efptNR; i++) |
2730 | { |
2731 | /* overwrite lambda state with init_lambda for now for backwards compatibility */ |
2732 | if (fep->init_lambda >= 0) /* if it's -1, it was never initializd */ |
2733 | { |
2734 | lambda[i] = fep->init_lambda; |
2735 | if (lam0) |
2736 | { |
2737 | lam0[i] = lambda[i]; |
2738 | } |
2739 | } |
2740 | else |
2741 | { |
2742 | lambda[i] = fep->all_lambda[i][*fep_state]; |
2743 | if (lam0) |
2744 | { |
2745 | lam0[i] = lambda[i]; |
2746 | } |
2747 | } |
2748 | } |
2749 | if (ir->bSimTemp) |
2750 | { |
2751 | /* need to rescale control temperatures to match current state */ |
2752 | for (i = 0; i < ir->opts.ngtc; i++) |
2753 | { |
2754 | if (ir->opts.ref_t[i] > 0) |
2755 | { |
2756 | ir->opts.ref_t[i] = ir->simtempvals->temperatures[*fep_state]; |
2757 | } |
2758 | } |
2759 | } |
2760 | } |
2761 | |
2762 | /* Send to the log the information on the current lambdas */ |
2763 | if (fplog != NULL((void*)0)) |
2764 | { |
2765 | fprintf(fplog, "Initial vector of lambda components:[ "); |
2766 | for (i = 0; i < efptNR; i++) |
2767 | { |
2768 | fprintf(fplog, "%10.4f ", lambda[i]); |
2769 | } |
2770 | fprintf(fplog, "]\n"); |
2771 | } |
2772 | return; |
2773 | } |
2774 | |
2775 | |
2776 | void init_md(FILE *fplog, |
2777 | t_commrec *cr, t_inputrec *ir, const output_env_t oenv, |
2778 | double *t, double *t0, |
2779 | real *lambda, int *fep_state, double *lam0, |
2780 | t_nrnb *nrnb, gmx_mtop_t *mtop, |
2781 | gmx_update_t *upd, |
2782 | int nfile, const t_filenm fnm[], |
2783 | gmx_mdoutf_t *outf, t_mdebin **mdebin, |
2784 | tensor force_vir, tensor shake_vir, rvec mu_tot, |
2785 | gmx_bool *bSimAnn, t_vcm **vcm, unsigned long Flags) |
2786 | { |
2787 | int i, j, n; |
2788 | real tmpt, mod; |
2789 | |
2790 | /* Initial values */ |
2791 | *t = *t0 = ir->init_t; |
2792 | |
2793 | *bSimAnn = FALSE0; |
2794 | for (i = 0; i < ir->opts.ngtc; i++) |
2795 | { |
2796 | /* set bSimAnn if any group is being annealed */ |
2797 | if (ir->opts.annealing[i] != eannNO) |
2798 | { |
2799 | *bSimAnn = TRUE1; |
2800 | } |
2801 | } |
2802 | if (*bSimAnn) |
2803 | { |
2804 | update_annealing_target_temp(&(ir->opts), ir->init_t); |
2805 | } |
2806 | |
2807 | /* Initialize lambda variables */ |
2808 | initialize_lambdas(fplog, ir, fep_state, lambda, lam0); |
2809 | |
2810 | if (upd) |
2811 | { |
2812 | *upd = init_update(ir); |
2813 | } |
2814 | |
2815 | |
2816 | if (vcm != NULL((void*)0)) |
2817 | { |
2818 | *vcm = init_vcm(fplog, &mtop->groups, ir); |
2819 | } |
2820 | |
2821 | if (EI_DYNAMICS(ir->eI)(((ir->eI) == eiMD || ((ir->eI) == eiVV || (ir->eI) == eiVVAK)) || ((ir->eI) == eiSD1 || (ir->eI) == eiSD2) || (ir->eI) == eiBD) && !(Flags & MD_APPENDFILES(1<<15))) |
2822 | { |
2823 | if (ir->etc == etcBERENDSEN) |
2824 | { |
2825 | please_cite(fplog, "Berendsen84a"); |
2826 | } |
2827 | if (ir->etc == etcVRESCALE) |
2828 | { |
2829 | please_cite(fplog, "Bussi2007a"); |
2830 | } |
2831 | } |
2832 | |
2833 | init_nrnb(nrnb); |
2834 | |
2835 | if (nfile != -1) |
2836 | { |
2837 | *outf = init_mdoutf(fplog, nfile, fnm, Flags, cr, ir, mtop, oenv); |
2838 | |
2839 | *mdebin = init_mdebin((Flags & MD_APPENDFILES(1<<15)) ? NULL((void*)0) : mdoutf_get_fp_ene(*outf), |
2840 | mtop, ir, mdoutf_get_fp_dhdl(*outf)); |
2841 | } |
2842 | |
2843 | if (ir->bAdress) |
2844 | { |
2845 | please_cite(fplog, "Fritsch12"); |
2846 | please_cite(fplog, "Junghans10"); |
2847 | } |
2848 | /* Initiate variables */ |
2849 | clear_mat(force_vir); |
2850 | clear_mat(shake_vir); |
2851 | clear_rvec(mu_tot); |
2852 | |
2853 | debug_gmx(); |
2854 | } |