3 * This source code is part of
7 * GROningen MAchine for Chemical Simulations
9 * Written by David van der Spoel, Erik Lindahl, Berk Hess, and others.
10 * Copyright (c) 1991-2000, University of Groningen, The Netherlands.
11 * Copyright (c) 2001-2008, The GROMACS development team,
12 * check out http://www.gromacs.org for more information.
14 * This program is free software; you can redistribute it and/or
15 * modify it under the terms of the GNU General Public License
16 * as published by the Free Software Foundation; either version 2
17 * of the License, or (at your option) any later version.
19 * If you want to redistribute modifications, please consider that
20 * scientific software is very special. Version control is crucial -
21 * bugs must be traceable. We will be happy to consider code for
22 * inclusion in the official distribution, but derived work must not
23 * be called official GROMACS. Details are found in the README & COPYING
24 * files - if they are missing, get the official version at www.gromacs.org.
26 * To help us fund GROMACS development, we humbly ask that you cite
27 * the papers on the package - you can find them in the top README file.
29 * For more info, check our website at http://www.gromacs.org
32 * Gallium Rubidium Oxygen Manganese Argon Carbon Silicon
40 #ifdef HAVE_SYS_TIME_H
56 #include "checkpoint.h"
63 ddnoSEL, ddnoINTERLEAVE, ddnoPP_PME, ddnoCARTESIAN, ddnoNR
66 /* Enum for situations that can occur during log file parsing, the
67 * corresponding string entries can be found in do_the_tests() in
68 * const char* ParseLog[] */
74 eParselogResetProblem,
85 int nPMEnodes; /* number of PME-only nodes used in this test */
86 int nx, ny, nz; /* DD grid */
87 int guessPME; /* if nPMEnodes == -1, this is the guessed number of PME nodes */
88 double *Gcycles; /* This can contain more than one value if doing multiple tests */
92 float *PME_f_load; /* PME mesh/force load average*/
93 float PME_f_load_Av; /* Average average ;) ... */
94 char *mdrun_cmd_line; /* Mdrun command line used for this test */
100 int nr_inputfiles; /* The number of tpr and mdp input files */
101 gmx_large_int_t orig_sim_steps; /* Number of steps to be done in the real simulation */
102 real *rcoulomb; /* The coulomb radii [0...nr_inputfiles] */
103 real *rvdw; /* The vdW radii */
104 real *rlist; /* Neighbourlist cutoff radius */
106 int *nkx, *nky, *nkz;
107 real *fsx, *fsy, *fsz; /* Fourierspacing in x,y,z dimension */
111 static void sep_line(FILE *fp)
113 fprintf(fp, "\n------------------------------------------------------------\n");
117 /* Wrapper for system calls */
118 static int gmx_system_call(char *command)
121 gmx_fatal(FARGS,"No calls to system(3) supported on this platform. Attempted to call:\n'%s'\n",command);
123 return ( system(command) );
128 /* Check if string starts with substring */
129 static gmx_bool str_starts(const char *string, const char *substring)
131 return ( strncmp(string, substring, strlen(substring)) == 0);
135 static void cleandata(t_perf *perfdata, int test_nr)
137 perfdata->Gcycles[test_nr] = 0.0;
138 perfdata->ns_per_day[test_nr] = 0.0;
139 perfdata->PME_f_load[test_nr] = 0.0;
145 static gmx_bool is_equal(real a, real b)
147 real diff, eps=1.0e-7;
152 if (diff < 0.0) diff = -diff;
161 static void finalize(const char *fn_out)
167 fp = fopen(fn_out,"r");
168 fprintf(stdout,"\n\n");
170 while( fgets(buf,STRLEN-1,fp) != NULL )
172 fprintf(stdout,"%s",buf);
175 fprintf(stdout,"\n\n");
180 enum {eFoundNothing, eFoundDDStr, eFoundAccountingStr, eFoundCycleStr};
182 static int parse_logfile(const char *logfile, const char *errfile,
183 t_perf *perfdata, int test_nr, int presteps, gmx_large_int_t cpt_steps,
187 char line[STRLEN], dumstring[STRLEN], dumstring2[STRLEN];
188 const char matchstrdd[]="Domain decomposition grid";
189 const char matchstrcr[]="resetting all time and cycle counters";
190 const char matchstrbal[]="Average PME mesh/force load:";
191 const char matchstring[]="R E A L C Y C L E A N D T I M E A C C O U N T I N G";
192 const char errSIG[]="signal, stopping at the next";
195 float dum1,dum2,dum3;
197 gmx_large_int_t resetsteps=-1;
198 gmx_bool bFoundResetStr = FALSE;
199 gmx_bool bResetChecked = FALSE;
202 if (!gmx_fexist(logfile))
204 fprintf(stderr, "WARNING: Could not find logfile %s.\n", logfile);
205 cleandata(perfdata, test_nr);
206 return eParselogNotFound;
209 fp = fopen(logfile, "r");
210 perfdata->PME_f_load[test_nr] = -1.0;
211 perfdata->guessPME = -1;
213 iFound = eFoundNothing;
215 iFound = eFoundDDStr; /* Skip some case statements */
217 while (fgets(line, STRLEN, fp) != NULL)
219 /* Remove leading spaces */
222 /* Check for TERM and INT signals from user: */
223 if ( strstr(line, errSIG) != NULL )
226 cleandata(perfdata, test_nr);
227 return eParselogTerm;
230 /* Check whether cycle resetting worked */
231 if (presteps > 0 && !bFoundResetStr)
233 if (strstr(line, matchstrcr) != NULL)
235 sprintf(dumstring, "Step %s", gmx_large_int_pfmt);
236 sscanf(line, dumstring, &resetsteps);
237 bFoundResetStr = TRUE;
238 if (resetsteps == presteps+cpt_steps)
240 bResetChecked = TRUE;
244 sprintf(dumstring , gmx_large_int_pfmt, resetsteps);
245 sprintf(dumstring2, gmx_large_int_pfmt, presteps+cpt_steps);
246 fprintf(stderr, "WARNING: Time step counters were reset at step %s,\n"
247 " though they were supposed to be reset at step %s!\n",
248 dumstring, dumstring2);
253 /* Look for strings that appear in a certain order in the log file: */
257 /* Look for domain decomp grid and separate PME nodes: */
258 if (str_starts(line, matchstrdd))
260 sscanf(line, "Domain decomposition grid %d x %d x %d, separate PME nodes %d",
261 &(perfdata->nx), &(perfdata->ny), &(perfdata->nz), &npme);
262 if (perfdata->nPMEnodes == -1)
263 perfdata->guessPME = npme;
264 else if (perfdata->nPMEnodes != npme)
265 gmx_fatal(FARGS, "PME nodes from command line and output file are not identical");
266 iFound = eFoundDDStr;
268 /* Catch a few errors that might have occured: */
269 else if (str_starts(line, "There is no domain decomposition for"))
272 return eParselogNoDDGrid;
274 else if (str_starts(line, "reading tpx file"))
277 return eParselogTPXVersion;
279 else if (str_starts(line, "The -dd or -npme option request a parallel simulation"))
282 return eParselogNotParallel;
286 /* Look for PME mesh/force balance (not necessarily present, though) */
287 if (str_starts(line, matchstrbal))
288 sscanf(&line[strlen(matchstrbal)], "%f", &(perfdata->PME_f_load[test_nr]));
289 /* Look for matchstring */
290 if (str_starts(line, matchstring))
291 iFound = eFoundAccountingStr;
293 case eFoundAccountingStr:
294 /* Already found matchstring - look for cycle data */
295 if (str_starts(line, "Total "))
297 sscanf(line,"Total %d %lf",&procs,&(perfdata->Gcycles[test_nr]));
298 iFound = eFoundCycleStr;
302 /* Already found cycle data - look for remaining performance info and return */
303 if (str_starts(line, "Performance:"))
305 sscanf(line,"%s %f %f %f %f", dumstring, &dum1, &dum2, &(perfdata->ns_per_day[test_nr]), &dum3);
307 if (bResetChecked || presteps == 0)
310 return eParselogResetProblem;
316 /* Close the log file */
319 /* Check why there is no performance data in the log file.
320 * Did a fatal errors occur? */
321 if (gmx_fexist(errfile))
323 fp = fopen(errfile, "r");
324 while (fgets(line, STRLEN, fp) != NULL)
326 if ( str_starts(line, "Fatal error:") )
328 if (fgets(line, STRLEN, fp) != NULL)
329 fprintf(stderr, "\nWARNING: A fatal error has occured during this benchmark:\n"
332 cleandata(perfdata, test_nr);
333 return eParselogFatal;
340 fprintf(stderr, "WARNING: Could not find stderr file %s.\n", errfile);
343 /* Giving up ... we could not find out why there is no performance data in
345 fprintf(stdout, "No performance data in log file.\n");
346 cleandata(perfdata, test_nr);
348 return eParselogNoPerfData;
352 static gmx_bool analyze_data(
361 int *index_tpr, /* OUT: Nr of mdp file with best settings */
362 int *npme_optimal) /* OUT: Optimal number of PME nodes */
365 int line=0, line_win=-1;
366 int k_win=-1, i_win=-1, winPME;
367 double s=0.0; /* standard deviation */
370 char str_PME_f_load[13];
371 gmx_bool bCanUseOrigTPR;
372 gmx_bool bRefinedCoul, bRefinedVdW, bRefinedGrid;
378 fprintf(fp, "Summary of successful runs:\n");
379 fprintf(fp, "Line tpr PME nodes Gcycles Av. Std.dev. ns/day PME/f");
381 fprintf(fp, " DD grid");
386 for (k=0; k<ntprs; k++)
388 for (i=0; i<ntests; i++)
390 /* Select the right dataset: */
391 pd = &(perfdata[k][i]);
393 pd->Gcycles_Av = 0.0;
394 pd->PME_f_load_Av = 0.0;
395 pd->ns_per_day_Av = 0.0;
397 if (pd->nPMEnodes == -1)
398 sprintf(strbuf, "(%3d)", pd->guessPME);
400 sprintf(strbuf, " ");
402 /* Get the average run time of a setting */
403 for (j=0; j<nrepeats; j++)
405 pd->Gcycles_Av += pd->Gcycles[j];
406 pd->PME_f_load_Av += pd->PME_f_load[j];
408 pd->Gcycles_Av /= nrepeats;
409 pd->PME_f_load_Av /= nrepeats;
411 for (j=0; j<nrepeats; j++)
413 if (pd->ns_per_day[j] > 0.0)
414 pd->ns_per_day_Av += pd->ns_per_day[j];
417 /* Somehow the performance number was not aquired for this run,
418 * therefor set the average to some negative value: */
419 pd->ns_per_day_Av = -1.0f*nrepeats;
423 pd->ns_per_day_Av /= nrepeats;
426 if (pd->PME_f_load_Av > 0.0)
427 sprintf(str_PME_f_load, "%12.3f", pd->PME_f_load_Av);
429 sprintf(str_PME_f_load, "%s", " - ");
432 /* We assume we had a successful run if both averages are positive */
433 if (pd->Gcycles_Av > 0.0 && pd->ns_per_day_Av > 0.0)
435 /* Output statistics if repeats were done */
438 /* Calculate the standard deviation */
440 for (j=0; j<nrepeats; j++)
441 s += pow( pd->Gcycles[j] - pd->Gcycles_Av, 2 );
445 fprintf(fp, "%4d %3d %4d%s %12.3f %12.3f %12.3f %s",
446 line, k, pd->nPMEnodes, strbuf, pd->Gcycles_Av, s,
447 pd->ns_per_day_Av, str_PME_f_load);
449 fprintf(fp, " %3d %3d %3d", pd->nx, pd->ny, pd->nz);
452 /* Store the index of the best run found so far in 'winner': */
453 if ( (k_win == -1) || (pd->Gcycles_Av < perfdata[k_win][i_win].Gcycles_Av) )
465 gmx_fatal(FARGS, "None of the runs was successful! Check %s for problems.", fn);
469 winPME = perfdata[k_win][i_win].nPMEnodes;
473 /* We stuck to a fixed number of PME-only nodes */
474 sprintf(strbuf, "settings No. %d", k_win);
478 /* We have optimized the number of PME-only nodes */
480 sprintf(strbuf, "%s", "the automatic number of PME nodes");
482 sprintf(strbuf, "%d PME nodes", winPME);
484 fprintf(fp, "Best performance was achieved with %s", strbuf);
485 if ((nrepeats > 1) && (ntests > 1))
486 fprintf(fp, " (see line %d)", line_win);
489 /* Only mention settings if they were modified: */
490 bRefinedCoul = !is_equal(info->rcoulomb[k_win], info->rcoulomb[0]);
491 bRefinedVdW = !is_equal(info->rvdw[k_win] , info->rvdw[0] );
492 bRefinedGrid = !(info->nkx[k_win] == info->nkx[0] &&
493 info->nky[k_win] == info->nky[0] &&
494 info->nkz[k_win] == info->nkz[0]);
496 if (bRefinedCoul || bRefinedVdW || bRefinedGrid)
498 fprintf(fp, "Optimized PME settings:\n");
499 bCanUseOrigTPR = FALSE;
503 bCanUseOrigTPR = TRUE;
507 fprintf(fp, " New Coulomb radius: %f nm (was %f nm)\n", info->rcoulomb[k_win], info->rcoulomb[0]);
510 fprintf(fp, " New Van der Waals radius: %f nm (was %f nm)\n", info->rvdw[k_win], info->rvdw[0]);
513 fprintf(fp, " New Fourier grid xyz: %d %d %d (was %d %d %d)\n", info->nkx[k_win], info->nky[k_win], info->nkz[k_win],
514 info->nkx[0], info->nky[0], info->nkz[0]);
516 if (bCanUseOrigTPR && ntprs > 1)
517 fprintf(fp, "and original PME settings.\n");
521 /* Return the index of the mdp file that showed the highest performance
522 * and the optimal number of PME nodes */
524 *npme_optimal = winPME;
526 return bCanUseOrigTPR;
530 /* Get the commands we need to set up the runs from environment variables */
531 static void get_program_paths(gmx_bool bThreads, char *cmd_mpirun[], char cmd_np[],
532 char *cmd_mdrun[], int repeats)
539 const char def_mpirun[] = "mpirun";
540 const char def_mdrun[] = "mdrun";
541 const char filename[] = "benchtest.log";
542 const char match_mpi[] = "NNODES=";
543 const char match_mdrun[]= "Program: ";
544 const char empty_mpirun[] = "";
545 gmx_bool bMdrun = FALSE;
546 gmx_bool bMPI = FALSE;
549 /* Get the commands we need to set up the runs from environment variables */
552 if ( (cp = getenv("MPIRUN")) != NULL)
553 *cmd_mpirun = strdup(cp);
555 *cmd_mpirun = strdup(def_mpirun);
559 *cmd_mpirun = strdup(empty_mpirun);
562 if ( (cp = getenv("MDRUN" )) != NULL )
563 *cmd_mdrun = strdup(cp);
565 *cmd_mdrun = strdup(def_mdrun);
568 /* If no simulations have to be performed, we are done here */
572 /* Run a small test to see whether mpirun + mdrun work */
573 fprintf(stdout, "Making sure that mdrun can be executed. ");
576 snew(command, strlen(*cmd_mdrun) + strlen(cmd_np) + strlen(filename) + 50);
577 sprintf(command, "%s%s-version -maxh 0.001 1> %s 2>&1", *cmd_mdrun, cmd_np, filename);
581 snew(command, strlen(*cmd_mpirun) + strlen(cmd_np) + strlen(*cmd_mdrun) + strlen(filename) + 50);
582 sprintf(command, "%s%s%s -version -maxh 0.001 1> %s 2>&1", *cmd_mpirun, cmd_np, *cmd_mdrun, filename);
584 fprintf(stdout, "Trying '%s' ... ", command);
585 make_backup(filename);
586 gmx_system_call(command);
588 /* Check if we find the characteristic string in the output: */
589 if (!gmx_fexist(filename))
590 gmx_fatal(FARGS, "Output from test run could not be found.");
592 fp = fopen(filename, "r");
593 /* We need to scan the whole output file, since sometimes the queuing system
594 * also writes stuff to stdout/err */
597 cp2=fgets(line, STRLEN, fp);
600 if ( str_starts(line, match_mdrun) )
602 if ( str_starts(line, match_mpi) )
612 gmx_fatal(FARGS, "Need a threaded version of mdrun. This one\n"
614 "seems to have been compiled with MPI instead.",
622 gmx_fatal(FARGS, "Need an MPI-enabled version of mdrun. This one\n"
624 "seems to have been compiled without MPI support.",
631 gmx_fatal(FARGS, "Cannot execute mdrun. Please check %s for problems!",
635 fprintf(stdout, "passed.\n");
643 static void launch_simulation(
644 gmx_bool bLaunch, /* Should the simulation be launched? */
645 FILE *fp, /* General log file */
646 gmx_bool bThreads, /* whether to use threads */
647 char *cmd_mpirun, /* Command for mpirun */
648 char *cmd_np, /* Switch for -np or -nt or empty */
649 char *cmd_mdrun, /* Command for mdrun */
650 char *args_for_mdrun, /* Arguments for mdrun */
651 const char *simulation_tpr, /* This tpr will be simulated */
652 int nnodes, /* Number of nodes to run on */
653 int nPMEnodes) /* Number of PME nodes to use */
658 /* Make enough space for the system call command,
659 * (100 extra chars for -npme ... etc. options should suffice): */
660 snew(command, strlen(cmd_mpirun)+strlen(cmd_mdrun)+strlen(cmd_np)+strlen(args_for_mdrun)+strlen(simulation_tpr)+100);
662 /* Note that the -passall options requires args_for_mdrun to be at the end
663 * of the command line string */
666 sprintf(command, "%s%s-npme %d -s %s %s",
667 cmd_mdrun, cmd_np, nPMEnodes, simulation_tpr, args_for_mdrun);
671 sprintf(command, "%s%s%s -npme %d -s %s %s",
672 cmd_mpirun, cmd_np, cmd_mdrun, nPMEnodes, simulation_tpr, args_for_mdrun);
675 fprintf(fp, "%s this command line to launch the simulation:\n\n%s", bLaunch? "Using":"Please use", command);
679 /* Now the real thing! */
682 fprintf(stdout, "\nLaunching simulation with best parameters now.\nExecuting '%s'", command);
685 gmx_system_call(command);
691 static void modify_PMEsettings(
692 gmx_large_int_t simsteps, /* Set this value as number of time steps */
693 const char *fn_best_tpr, /* tpr file with the best performance */
694 const char *fn_sim_tpr) /* name of tpr file to be launched */
702 read_tpx_state(fn_best_tpr,ir,&state,NULL,&mtop);
704 /* Set nsteps to the right value */
705 ir->nsteps = simsteps;
707 /* Write the tpr file which will be launched */
708 sprintf(buf, "Writing optimized simulation file %s with nsteps=%s.\n", fn_sim_tpr, gmx_large_int_pfmt);
709 fprintf(stdout,buf,ir->nsteps);
711 write_tpx_state(fn_sim_tpr,ir,&state,&mtop);
719 int nkx, nky, nkz; /* Fourier grid */
720 real fac; /* actual scaling factor */
721 real fs; /* Fourierspacing */
725 static void copy_grid(t_pmegrid *ingrid, t_pmegrid *outgrid)
727 outgrid->nkx = ingrid->nkx;
728 outgrid->nky = ingrid->nky;
729 outgrid->nkz = ingrid->nkz;
730 outgrid->fac = ingrid->fac;
731 outgrid->fs = ingrid->fs;
734 /* Removes entry 'index' from the t_pmegrid list */
735 static void remove_from_list(
736 t_pmegrid gridlist[],
737 int *nlist, /* Length of the list */
738 int index) /* Index to remove from the list */
743 for (j = index; j < (*nlist - 1); j++)
745 copy_grid(&gridlist[j+1], &gridlist[j]);
751 /* Returns the index of the least necessary grid in the list.
753 * This is the one where the following conditions hold for the scaling factor:
754 * 1. this grid has the smallest distance to its neighboring grid (distance
755 * measured by fac) -> this condition is true for two grids at the same time
756 * 2. this grid (of the two) has the smaller distance to the other neighboring
759 * The extreme grids (the ones with the smallest and largest
760 * scaling factor) are never thrown away.
762 static int get_throwaway_index(
763 t_pmegrid gridlist[],
764 int nlist) /* Length of the list */
767 real dist,mindist,d_left,d_right;
770 /* Find the two factors with the smallest mutual distance */
771 mindist = GMX_FLOAT_MAX;
772 for (i = 1; i < nlist; i++)
774 dist = fabs(gridlist[i].fac - gridlist[i-1].fac);
781 /* index and index-1 have the smallest mutual distance */
784 /* Never return the first index, i.e. index == 0 */
789 d_left = fabs(gridlist[index-1].fac - gridlist[index-2].fac);
790 d_right = fabs(gridlist[index+1].fac - gridlist[index ].fac);
792 /* Return the left index if its distance to its left neighbor is shorter
793 * than the distance of the right index to its right neighbor */
794 if (d_left < d_right)
797 /* Never return the last index */
798 if (index == nlist-1)
805 static void make_grid_list(
806 real fmin, /* minimum scaling factor (downscaling fac) */
807 real fmax, /* maximum scaling factor (upscaling fac) */
808 int *ntprs, /* Number of tpr files to test */
809 matrix box, /* The box */
810 t_pmegrid *griduse[], /* List of grids that have to be tested */
811 real fs) /* Requested fourierspacing at a scaling factor
814 real req_fac,act_fac=0,act_fs,eps;
816 int i,jmin=0,d,ngridall=0;
817 int nkx=0,nky=0,nkz=0;
818 int nkx_old=0,nky_old=0,nkz_old=0;
820 int gridalloc,excess;
823 /* Determine length of triclinic box vectors */
828 box_size[d] += box[d][i]*box[d][i];
829 box_size[d] = sqrt(box_size[d]);
833 snew(gridall, gridalloc);
835 fprintf(stdout, "Possible PME grid settings (apart from input file settings):\n");
836 fprintf(stdout, " nkx nky nkz max spacing scaling factor\n");
838 /* eps should provide a fine enough spacing not to miss any grid */
840 eps = (fmax-fmin)/(100.0*(*ntprs - 1));
842 eps = 1.0/max( (*griduse)[0].nkz, max( (*griduse)[0].nkx, (*griduse)[0].nky ) );
844 for (req_fac = fmin; act_fac < fmax; req_fac += eps)
849 calc_grid(NULL,box,fs*req_fac,&nkx,&nky,&nkz);
850 act_fs = max(box_size[XX]/nkx,max(box_size[YY]/nky,box_size[ZZ]/nkz));
852 if ( ! ( nkx==nkx_old && nky==nky_old && nkz==nkz_old ) /* Exclude if grid is already in list */
853 && ! ( nkx==(*griduse)[0].nkx && nky==(*griduse)[0].nky && nkz==(*griduse)[0].nkz ) ) /* Exclude input file grid */
855 /* We found a new grid that will do */
859 gridall[ngridall].nkx = nkx;
860 gridall[ngridall].nky = nky;
861 gridall[ngridall].nkz = nkz;
862 gridall[ngridall].fac = act_fac;
863 gridall[ngridall].fs = act_fs;
864 fprintf(stdout, "%5d%5d%5d %12f %12f\n",nkx,nky,nkz,act_fs,act_fac);
866 if (ngridall >= gridalloc)
869 srenew(gridall, gridalloc);
874 /* excess is the number of grids we have to get rid of */
875 excess = ngridall+1 - *ntprs;
877 /* If we found less grids than tpr files were requested, simply test all
878 * the grid there are ... */
881 fprintf(stdout, "NOTE: You requested %d tpr files, but apart from the input file grid only the\n"
882 " above listed %d suitable PME grids were found. Will test all suitable settings.\n",
890 /* We can only keep the input tpr file plus one extra tpr file.
891 * We make that choice depending on the values of -upfac and -downfac */
894 /* Keep the one with the -downfac as scaling factor. This is already
895 * stored in gridall[0] */
900 /* Keep the one with -upfac as scaling factor */
901 copy_grid(&(gridall[ngridall-1]), &(gridall[0]));
907 /* From the grid list throw away the unnecessary ones (keep the extremes) */
908 for (i = 0; i < excess; i++)
910 /* Get the index of the least necessary grid from the list ... */
911 jmin = get_throwaway_index(gridall, ngridall);
912 /* ... and remove the grid from the list */
913 remove_from_list(gridall, &ngridall, jmin);
918 /* The remaining list contains the grids we want to test */
919 for (i=1; i < *ntprs; i++)
920 copy_grid(&(gridall[i-1]), &((*griduse)[i]));
926 #define EPME_SWITCHED(e) ((e) == eelPMESWITCH || (e) == eelPMEUSERSWITCH)
928 /* Make additional TPR files with more computational load for the
929 * direct space processors: */
930 static void make_benchmark_tprs(
931 const char *fn_sim_tpr, /* READ : User-provided tpr file */
932 char *fn_bench_tprs[], /* WRITE: Names of benchmark tpr files */
933 gmx_large_int_t benchsteps, /* Number of time steps for benchmark runs */
934 gmx_large_int_t statesteps, /* Step counter in checkpoint file */
935 real upfac, /* Scale rcoulomb inbetween downfac and upfac */
937 int *ntprs, /* No. of TPRs to write, each with a different
938 rcoulomb and fourierspacing. If not enough
939 grids are found, ntprs is reduced accordingly */
940 real fourierspacing, /* Basic fourierspacing from tpr input file */
941 t_inputinfo *info, /* Contains information about mdp file options */
942 FILE *fp) /* Write the output here */
948 real nlist_buffer; /* Thickness of the buffer regions for PME-switch potentials: */
951 gmx_bool bNote = FALSE;
952 t_pmegrid *pmegrid=NULL; /* Grid settings for the PME grids to test */
955 sprintf(buf, "Making benchmark tpr file%s with %s time step%s",
956 *ntprs > 1? "s":"", gmx_large_int_pfmt, benchsteps>1?"s":"");
957 fprintf(stdout, buf, benchsteps);
960 sprintf(buf, " (adding %s steps from checkpoint file)", gmx_large_int_pfmt);
961 fprintf(stdout, buf, statesteps);
962 benchsteps += statesteps;
964 fprintf(stdout, ".\n");
968 read_tpx_state(fn_sim_tpr,ir,&state,NULL,&mtop);
970 /* Check if some kind of PME was chosen */
971 if (EEL_PME(ir->coulombtype) == FALSE)
972 gmx_fatal(FARGS, "Can only do optimizations for simulations with %s electrostatics.",
975 /* Check if rcoulomb == rlist, which is necessary for plain PME. */
976 if ( (eelPME == ir->coulombtype) && !(ir->rcoulomb == ir->rlist) )
978 gmx_fatal(FARGS, "%s requires rcoulomb (%f) to be equal to rlist (%f).",
979 EELTYPE(eelPME), ir->rcoulomb, ir->rlist);
981 /* For other PME types, rcoulomb is allowed to be smaller than rlist */
982 else if (ir->rcoulomb > ir->rlist)
984 gmx_fatal(FARGS, "%s requires rcoulomb (%f) to be equal to or smaller than rlist (%f)",
985 EELTYPE(ir->coulombtype), ir->rcoulomb, ir->rlist);
988 /* Reduce the number of steps for the benchmarks */
989 info->orig_sim_steps = ir->nsteps;
990 ir->nsteps = benchsteps;
992 /* For PME-switch potentials, keep the radial distance of the buffer region */
993 nlist_buffer = ir->rlist - ir->rcoulomb;
995 /* Determine length of triclinic box vectors */
1000 box_size[d] += state.box[d][i]*state.box[d][i];
1001 box_size[d] = sqrt(box_size[d]);
1004 /* Reconstruct fourierspacing per dimension from the number of grid points and box size */
1005 info->fsx[0] = box_size[XX]/ir->nkx;
1006 info->fsy[0] = box_size[YY]/ir->nky;
1007 info->fsz[0] = box_size[ZZ]/ir->nkz;
1009 /* Put the input grid as first entry into the grid list */
1010 snew(pmegrid, *ntprs);
1011 pmegrid[0].fac = 1.00;
1012 pmegrid[0].nkx = ir->nkx;
1013 pmegrid[0].nky = ir->nky;
1014 pmegrid[0].nkz = ir->nkz;
1015 pmegrid[0].fs = max(box_size[ZZ]/ir->nkz, max(box_size[XX]/ir->nkx, box_size[YY]/ir->nky));
1017 /* If no value for the fourierspacing was provided on the command line, we
1018 * use the reconstruction from the tpr file */
1019 if (fourierspacing <= 0)
1021 fourierspacing = pmegrid[0].fs;
1024 fprintf(stdout, "Calculating PME grid points on the basis of a fourierspacing of %f nm\n", fourierspacing);
1026 /* Print information about settings of which some are potentially modified: */
1027 fprintf(fp, " Coulomb type : %s\n", EELTYPE(ir->coulombtype));
1028 fprintf(fp, " Grid spacing x y z : %f %f %f\n",
1029 box_size[XX]/ir->nkx, box_size[YY]/ir->nky, box_size[ZZ]/ir->nkz);
1030 fprintf(fp, " Van der Waals type : %s\n", EVDWTYPE(ir->vdwtype));
1031 if (EVDW_SWITCHED(ir->vdwtype))
1032 fprintf(fp, " rvdw_switch : %f nm\n", ir->rvdw_switch);
1033 if (EPME_SWITCHED(ir->coulombtype))
1034 fprintf(fp, " rlist : %f nm\n", ir->rlist);
1035 if (ir->rlistlong != max_cutoff(ir->rvdw,ir->rcoulomb))
1036 fprintf(fp, " rlistlong : %f nm\n", ir->rlistlong);
1038 /* Print a descriptive line about the tpr settings tested */
1039 fprintf(fp, "\nWill try these real/reciprocal workload settings:\n");
1040 fprintf(fp, " No. scaling rcoulomb");
1041 fprintf(fp, " nkx nky nkz");
1042 fprintf(fp, " spacing");
1043 if (evdwCUT == ir->vdwtype)
1044 fprintf(fp, " rvdw");
1045 if (EPME_SWITCHED(ir->coulombtype))
1046 fprintf(fp, " rlist");
1047 if ( ir->rlistlong != max_cutoff(ir->rlist,max_cutoff(ir->rvdw,ir->rcoulomb)) )
1048 fprintf(fp, " rlistlong");
1049 fprintf(fp, " tpr file\n");
1052 /* Get useful PME grids if requested, the actual number of entries is
1053 * returned in npmegrid */
1055 make_grid_list(downfac, upfac, ntprs, state.box, &pmegrid, fourierspacing);
1057 /* Loop to create the requested number of tpr input files */
1058 for (j = 0; j < *ntprs; j++)
1060 /* The first one is the provided tpr file, just need to modify the number
1061 * of steps, so skip the following block */
1064 /* Scale the Coulomb radius */
1065 ir->rcoulomb = info->rcoulomb[0]*pmegrid[j].fac;
1067 /* Adjust other radii since various conditions neet to be fulfilled */
1068 if (eelPME == ir->coulombtype)
1070 /* plain PME, rcoulomb must be equal to rlist */
1071 ir->rlist = ir->rcoulomb;
1075 /* rlist must be >= rcoulomb, we keep the size of the buffer region */
1076 ir->rlist = ir->rcoulomb + nlist_buffer;
1079 if (evdwCUT == ir->vdwtype)
1081 /* For vdw cutoff, rvdw >= rlist */
1082 ir->rvdw = max(info->rvdw[0], ir->rlist);
1085 ir->rlistlong = max_cutoff(ir->rlist,max_cutoff(ir->rvdw,ir->rcoulomb));
1087 /* Copy the optimal grid dimensions to ir */
1088 ir->nkx = pmegrid[j].nkx;
1089 ir->nky = pmegrid[j].nky;
1090 ir->nkz = pmegrid[j].nkz;
1092 } /* end of "if (j != 0)" */
1094 /* for j==0: Save the original settings
1095 * for j >0: Save modified radii and fourier grids */
1096 info->rcoulomb[j] = ir->rcoulomb;
1097 info->rvdw[j] = ir->rvdw;
1098 info->nkx[j] = ir->nkx;
1099 info->nky[j] = ir->nky;
1100 info->nkz[j] = ir->nkz;
1101 info->rlist[j] = ir->rlist;
1102 info->rlistlong[j] = ir->rlistlong;
1103 info->fsx[j] = pmegrid[j].fs;
1104 info->fsy[j] = pmegrid[j].fs;
1105 info->fsz[j] = pmegrid[j].fs;
1107 /* Write the benchmark tpr file */
1108 strncpy(fn_bench_tprs[j],fn_sim_tpr,strlen(fn_sim_tpr)-strlen(".tpr"));
1109 sprintf(buf, "_bench%.2d.tpr", j);
1110 strcat(fn_bench_tprs[j], buf);
1111 fprintf(stdout,"Writing benchmark tpr %s with nsteps=", fn_bench_tprs[j]);
1112 fprintf(stdout, gmx_large_int_pfmt, ir->nsteps);
1114 fprintf(stdout,", scaling factor %f\n", pmegrid[j].fac);
1116 fprintf(stdout,", unmodified settings\n");
1118 write_tpx_state(fn_bench_tprs[j],ir,&state,&mtop);
1120 /* Write information about modified tpr settings to log file */
1122 fprintf(fp, "%4d%10s%10f", j, "-input-", ir->rcoulomb);
1124 fprintf(fp, "%4d%10f%10f", j, pmegrid[j].fac, ir->rcoulomb);
1125 fprintf(fp, "%5d%5d%5d", ir->nkx, ir->nky, ir->nkz);
1126 fprintf(fp, " %9f ", info->fsx[j]);
1127 if (evdwCUT == ir->vdwtype)
1128 fprintf(fp, "%10f", ir->rvdw);
1129 if (EPME_SWITCHED(ir->coulombtype))
1130 fprintf(fp, "%10f", ir->rlist);
1131 if ( info->rlistlong[0] != max_cutoff(info->rlist[0],max_cutoff(info->rvdw[0],info->rcoulomb[0])) )
1132 fprintf(fp, "%10f", ir->rlistlong);
1133 fprintf(fp, " %-14s\n",fn_bench_tprs[j]);
1135 /* Make it clear to the user that some additional settings were modified */
1136 if ( !is_equal(ir->rvdw , info->rvdw[0])
1137 || !is_equal(ir->rlistlong, info->rlistlong[0]) )
1143 fprintf(fp, "\nNote that in addition to rcoulomb and the fourier grid\n"
1144 "also other input settings were changed (see table above).\n"
1145 "Please check if the modified settings are appropriate.\n");
1152 /* Whether these files are written depends on tpr (or mdp) settings,
1153 * not on mdrun command line options! */
1154 static gmx_bool tpr_triggers_file(const char *opt)
1156 if ( (0 == strcmp(opt, "-ro"))
1157 || (0 == strcmp(opt, "-ra"))
1158 || (0 == strcmp(opt, "-rt"))
1159 || (0 == strcmp(opt, "-rs"))
1160 || (0 == strcmp(opt, "-pf"))
1161 || (0 == strcmp(opt, "-px")) )
1168 /* Rename the files we want to keep to some meaningful filename and
1169 * delete the rest */
1170 static void cleanup(const t_filenm *fnm, int nfile, int k, int nnodes,
1171 int nPMEnodes, int nr, gmx_bool bKeepStderr)
1173 char numstring[STRLEN];
1174 char newfilename[STRLEN];
1175 const char *fn=NULL;
1180 fprintf(stdout, "Cleaning up, deleting benchmark temp files ...\n");
1182 for (i=0; i<nfile; i++)
1184 opt = (char *)fnm[i].opt;
1185 if ( strcmp(opt, "-p") == 0 )
1187 /* do nothing; keep this file */
1190 else if (strcmp(opt, "-bg") == 0)
1192 /* Give the log file a nice name so one can later see which parameters were used */
1193 numstring[0] = '\0';
1195 sprintf(numstring, "_%d", nr);
1196 sprintf(newfilename, "%s_no%d_np%d_npme%d%s", opt2fn("-bg",nfile,fnm), k, nnodes, nPMEnodes, numstring);
1197 if (gmx_fexist(opt2fn("-bg",nfile,fnm)))
1199 fprintf(stdout, "renaming log file to %s\n", newfilename);
1200 make_backup(newfilename);
1201 rename(opt2fn("-bg",nfile,fnm), newfilename);
1204 else if (strcmp(opt, "-err") == 0)
1206 /* This file contains the output of stderr. We want to keep it in
1207 * cases where there have been problems. */
1208 fn = opt2fn(opt, nfile, fnm);
1209 numstring[0] = '\0';
1211 sprintf(numstring, "_%d", nr);
1212 sprintf(newfilename, "%s_no%d_np%d_npme%d%s", fn, k, nnodes, nPMEnodes, numstring);
1217 fprintf(stdout, "Saving stderr output in %s\n", newfilename);
1218 make_backup(newfilename);
1219 rename(fn, newfilename);
1223 fprintf(stdout, "Deleting %s\n", fn);
1228 /* Delete the files which are created for each benchmark run: (options -b*) */
1229 else if ( ( (0 == strncmp(opt, "-b", 2)) && (opt2bSet(opt,nfile,fnm) || !is_optional(&fnm[i])) )
1230 || tpr_triggers_file(opt) )
1232 fn = opt2fn(opt, nfile, fnm);
1235 fprintf(stdout, "Deleting %s\n", fn);
1243 /* Returns the largest common factor of n1 and n2 */
1244 static int largest_common_factor(int n1, int n2)
1249 for (factor=nmax; factor > 0; factor--)
1251 if ( 0==(n1 % factor) && 0==(n2 % factor) )
1256 return 0; /* one for the compiler */
1259 enum {eNpmeAuto, eNpmeAll, eNpmeReduced, eNpmeSubset, eNpmeNr};
1261 /* Create a list of numbers of PME nodes to test */
1262 static void make_npme_list(
1263 const char *npmevalues_opt, /* Make a complete list with all
1264 * possibilities or a short list that keeps only
1265 * reasonable numbers of PME nodes */
1266 int *nentries, /* Number of entries we put in the nPMEnodes list */
1267 int *nPMEnodes[], /* Each entry contains the value for -npme */
1268 int nnodes, /* Total number of nodes to do the tests on */
1269 int minPMEnodes, /* Minimum number of PME nodes */
1270 int maxPMEnodes) /* Maximum number of PME nodes */
1273 int min_factor=1; /* We request that npp and npme have this minimal
1274 * largest common factor (depends on npp) */
1275 int nlistmax; /* Max. list size */
1276 int nlist; /* Actual number of entries in list */
1280 /* Do we need to check all possible values for -npme or is a reduced list enough? */
1281 if ( 0 == strcmp(npmevalues_opt, "all") )
1285 else if ( 0 == strcmp(npmevalues_opt, "subset") )
1287 eNPME = eNpmeSubset;
1289 else if ( 0 == strcmp(npmevalues_opt, "auto") )
1293 else if (nnodes < 128)
1294 eNPME = eNpmeReduced;
1296 eNPME = eNpmeSubset;
1300 gmx_fatal(FARGS, "Unknown option for -npme in make_npme_list");
1303 /* Calculate how many entries we could possibly have (in case of -npme all) */
1306 nlistmax = maxPMEnodes - minPMEnodes + 3;
1307 if (0 == minPMEnodes)
1313 /* Now make the actual list which is at most of size nlist */
1314 snew(*nPMEnodes, nlistmax);
1315 nlist = 0; /* start counting again, now the real entries in the list */
1316 for (i = 0; i < nlistmax - 2; i++)
1318 npme = maxPMEnodes - i;
1329 /* For 2d PME we want a common largest factor of at least the cube
1330 * root of the number of PP nodes */
1331 min_factor = (int) pow(npp, 1.0/3.0);
1334 gmx_fatal(FARGS, "Unknown option for eNPME in make_npme_list");
1337 if (largest_common_factor(npp, npme) >= min_factor)
1339 (*nPMEnodes)[nlist] = npme;
1343 /* We always test 0 PME nodes and the automatic number */
1344 *nentries = nlist + 2;
1345 (*nPMEnodes)[nlist ] = 0;
1346 (*nPMEnodes)[nlist+1] = -1;
1348 fprintf(stderr, "Will try the following %d different values for -npme:\n", *nentries);
1349 for (i=0; i<*nentries-1; i++)
1350 fprintf(stderr, "%d, ", (*nPMEnodes)[i]);
1351 fprintf(stderr, "and %d (auto).\n", (*nPMEnodes)[*nentries-1]);
1355 /* Allocate memory to store the performance data */
1356 static void init_perfdata(t_perf *perfdata[], int ntprs, int datasets, int repeats)
1361 for (k=0; k<ntprs; k++)
1363 snew(perfdata[k], datasets);
1364 for (i=0; i<datasets; i++)
1366 for (j=0; j<repeats; j++)
1368 snew(perfdata[k][i].Gcycles , repeats);
1369 snew(perfdata[k][i].ns_per_day, repeats);
1370 snew(perfdata[k][i].PME_f_load, repeats);
1377 static void do_the_tests(
1378 FILE *fp, /* General g_tune_pme output file */
1379 char **tpr_names, /* Filenames of the input files to test */
1380 int maxPMEnodes, /* Max fraction of nodes to use for PME */
1381 int minPMEnodes, /* Min fraction of nodes to use for PME */
1382 int npme_fixed, /* If >= -1, test fixed number of PME
1384 const char *npmevalues_opt, /* Which -npme values should be tested */
1385 t_perf **perfdata, /* Here the performace data is stored */
1386 int *pmeentries, /* Entries in the nPMEnodes list */
1387 int repeats, /* Repeat each test this often */
1388 int nnodes, /* Total number of nodes = nPP + nPME */
1389 int nr_tprs, /* Total number of tpr files to test */
1390 gmx_bool bThreads, /* Threads or MPI? */
1391 char *cmd_mpirun, /* mpirun command string */
1392 char *cmd_np, /* "-np", "-n", whatever mpirun needs */
1393 char *cmd_mdrun, /* mdrun command string */
1394 char *cmd_args_bench, /* arguments for mdrun in a string */
1395 const t_filenm *fnm, /* List of filenames from command line */
1396 int nfile, /* Number of files specified on the cmdl. */
1397 int sim_part, /* For checkpointing */
1398 int presteps, /* DLB equilibration steps, is checked */
1399 gmx_large_int_t cpt_steps) /* Time step counter in the checkpoint */
1401 int i,nr,k,ret,count=0,totaltests;
1402 int *nPMEnodes=NULL;
1405 char *command, *cmd_stub;
1407 gmx_bool bResetProblem=FALSE;
1410 /* This string array corresponds to the eParselog enum type at the start
1412 const char* ParseLog[] = {"OK.",
1413 "Logfile not found!",
1414 "No timings, logfile truncated?",
1415 "Run was terminated.",
1416 "Counters were not reset properly.",
1417 "No DD grid found for these settings.",
1418 "TPX version conflict!",
1419 "mdrun was not started in parallel!",
1420 "A fatal error occured!" };
1421 char str_PME_f_load[13];
1424 /* Allocate space for the mdrun command line. 100 extra characters should
1425 be more than enough for the -npme etcetera arguments */
1426 cmdline_length = strlen(cmd_mpirun)
1429 + strlen(cmd_args_bench)
1430 + strlen(tpr_names[0]) + 100;
1431 snew(command , cmdline_length);
1432 snew(cmd_stub, cmdline_length);
1434 /* Construct the part of the command line that stays the same for all tests: */
1437 sprintf(cmd_stub, "%s%s", cmd_mdrun, cmd_np);
1441 sprintf(cmd_stub, "%s%s%s ", cmd_mpirun, cmd_np, cmd_mdrun);
1444 /* Create a list of numbers of PME nodes to test */
1445 if (npme_fixed < -1)
1447 make_npme_list(npmevalues_opt, pmeentries, &nPMEnodes,
1448 nnodes, minPMEnodes, maxPMEnodes);
1454 nPMEnodes[0] = npme_fixed;
1455 fprintf(stderr, "Will use a fixed number of %d PME-only nodes.\n", nPMEnodes[0]);
1460 fprintf(fp, "\nNo benchmarks done since number of repeats (-r) is 0.\n");
1462 finalize(opt2fn("-p", nfile, fnm));
1466 /* Allocate one dataset for each tpr input file: */
1467 init_perfdata(perfdata, nr_tprs, *pmeentries, repeats);
1469 /*****************************************/
1470 /* Main loop over all tpr files to test: */
1471 /*****************************************/
1472 totaltests = nr_tprs*(*pmeentries)*repeats;
1473 for (k=0; k<nr_tprs;k++)
1475 fprintf(fp, "\nIndividual timings for input file %d (%s):\n", k, tpr_names[k]);
1476 fprintf(fp, "PME nodes Gcycles ns/day PME/f Remark\n");
1477 /* Loop over various numbers of PME nodes: */
1478 for (i = 0; i < *pmeentries; i++)
1480 pd = &perfdata[k][i];
1482 /* Loop over the repeats for each scenario: */
1483 for (nr = 0; nr < repeats; nr++)
1485 pd->nPMEnodes = nPMEnodes[i];
1487 /* Add -npme and -s to the command line and save it. Note that
1488 * the -passall (if set) options requires cmd_args_bench to be
1489 * at the end of the command line string */
1490 snew(pd->mdrun_cmd_line, cmdline_length);
1491 sprintf(pd->mdrun_cmd_line, "%s-npme %d -s %s %s",
1492 cmd_stub, pd->nPMEnodes, tpr_names[k], cmd_args_bench);
1494 /* Do a benchmark simulation: */
1496 sprintf(buf, ", pass %d/%d", nr+1, repeats);
1499 fprintf(stdout, "\n=== Progress %2.0f%%, tpr %d/%d, run %d/%d%s:\n",
1500 (100.0*count)/totaltests,
1501 k+1, nr_tprs, i+1, *pmeentries, buf);
1502 make_backup(opt2fn("-err",nfile,fnm));
1503 sprintf(command, "%s 1> /dev/null 2>%s", pd->mdrun_cmd_line, opt2fn("-err",nfile,fnm));
1504 fprintf(stdout, "%s\n", pd->mdrun_cmd_line);
1505 gmx_system_call(command);
1507 /* Collect the performance data from the log file; also check stderr
1508 * for fatal errors */
1509 ret = parse_logfile(opt2fn("-bg",nfile,fnm), opt2fn("-err",nfile,fnm),
1510 pd, nr, presteps, cpt_steps, nnodes);
1511 if ((presteps > 0) && (ret == eParselogResetProblem))
1512 bResetProblem = TRUE;
1514 if (-1 == pd->nPMEnodes)
1515 sprintf(buf, "(%3d)", pd->guessPME);
1520 if (pd->PME_f_load[nr] > 0.0)
1521 sprintf(str_PME_f_load, "%12.3f", pd->PME_f_load[nr]);
1523 sprintf(str_PME_f_load, "%s", " - ");
1525 /* Write the data we got to disk */
1526 fprintf(fp, "%4d%s %12.3f %12.3f %s %s", pd->nPMEnodes,
1527 buf, pd->Gcycles[nr], pd->ns_per_day[nr], str_PME_f_load, ParseLog[ret]);
1528 if (! (ret==eParselogOK || ret==eParselogNoDDGrid || ret==eParselogNotFound) )
1529 fprintf(fp, " Check %s file for problems.", ret==eParselogFatal? "err":"log");
1534 /* Do some cleaning up and delete the files we do not need any more */
1535 cleanup(fnm, nfile, k, nnodes, pd->nPMEnodes, nr, ret==eParselogFatal);
1537 /* If the first run with this number of processors already failed, do not try again: */
1538 if (pd->Gcycles[0] <= 0.0 && repeats > 1)
1540 fprintf(stdout, "Skipping remaining passes of unsuccessful setting, see log file for details.\n");
1541 count += repeats-(nr+1);
1544 } /* end of repeats loop */
1545 } /* end of -npme loop */
1546 } /* end of tpr file loop */
1550 fprintf(fp, "WARNING: The cycle and time step counters could not be reset\n"
1551 "properly. The reason could be that mpirun did not manage to\n"
1552 "export the environment variable GMX_RESET_COUNTER. You might\n"
1553 "have to give a special switch to mpirun for that.\n"
1554 "Alternatively, you can manually set GMX_RESET_COUNTER to the\n"
1555 "value normally provided by -presteps.");
1563 static void check_input(
1569 real maxPMEfraction,
1570 real minPMEfraction,
1572 real fourierspacing,
1573 gmx_large_int_t bench_nsteps,
1574 const t_filenm *fnm,
1581 /* Make sure the input file exists */
1582 if (!gmx_fexist(opt2fn("-s",nfile,fnm)))
1583 gmx_fatal(FARGS, "File %s not found.", opt2fn("-s",nfile,fnm));
1585 /* Make sure that the checkpoint file is not overwritten during benchmarking */
1586 if ( (0 == strcmp(opt2fn("-cpi",nfile,fnm), opt2fn("-bcpo",nfile,fnm)) ) && (sim_part > 1) )
1587 gmx_fatal(FARGS, "Checkpoint input (-cpi) and benchmark checkpoint output (-bcpo) files must not be identical.\n"
1588 "The checkpoint input file must not be overwritten during the benchmarks.\n");
1590 /* Make sure that repeats is >= 0 (if == 0, only write tpr files) */
1592 gmx_fatal(FARGS, "Number of repeats < 0!");
1594 /* Check number of nodes */
1596 gmx_fatal(FARGS, "Number of nodes/threads must be a positive integer.");
1598 /* Automatically choose -ntpr if not set */
1605 fprintf(stderr, "Will test %d tpr file%s.\n", *ntprs, *ntprs==1?"":"s");
1610 fprintf(stderr, "Note: Choose ntpr>1 to shift PME load between real and reciprocal space.\n");
1613 if ( is_equal(*downfac,*upfac) && (*ntprs > 2) && !(fourierspacing > 0.0))
1615 fprintf(stderr, "WARNING: Resetting -ntpr to 2 since both scaling factors are the same.\n"
1616 "Please choose upfac unequal to downfac to test various PME grid settings\n");
1620 /* Check whether max and min fraction are within required values */
1621 if (maxPMEfraction > 0.5 || maxPMEfraction < 0)
1622 gmx_fatal(FARGS, "-max must be between 0 and 0.5");
1623 if (minPMEfraction > 0.5 || minPMEfraction < 0)
1624 gmx_fatal(FARGS, "-min must be between 0 and 0.5");
1625 if (maxPMEfraction < minPMEfraction)
1626 gmx_fatal(FARGS, "-max must be larger or equal to -min");
1628 /* Check whether the number of steps - if it was set - has a reasonable value */
1629 if (bench_nsteps < 0)
1630 gmx_fatal(FARGS, "Number of steps must be positive.");
1632 if (bench_nsteps > 10000 || bench_nsteps < 100)
1634 fprintf(stderr, "WARNING: steps=");
1635 fprintf(stderr, gmx_large_int_pfmt, bench_nsteps);
1636 fprintf(stderr, ". Are you sure you want to perform so %s steps for each benchmark?\n", (bench_nsteps < 100)? "few" : "many");
1641 gmx_fatal(FARGS, "Cannot have a negative number of presteps.\n");
1644 if (*upfac <= 0.0 || *downfac <= 0.0 || *downfac > *upfac)
1645 gmx_fatal(FARGS, "Both scaling factors must be larger than zero and upper\n"
1646 "scaling limit (%f) must be larger than lower limit (%f).",
1649 if (*downfac < 0.75 || *upfac > 1.25)
1650 fprintf(stderr, "WARNING: Applying extreme scaling factor. I hope you know what you are doing.\n");
1652 if (fourierspacing < 0)
1653 gmx_fatal(FARGS, "Please choose a positive value for fourierspacing.");
1655 /* Make shure that the scaling factor options are compatible with the number of tprs */
1656 if ( (*ntprs < 3) && ( opt2parg_bSet("-upfac",npargs,pa) || opt2parg_bSet("-downfac",npargs,pa) ) )
1658 if (opt2parg_bSet("-upfac",npargs,pa) && opt2parg_bSet("-downfac",npargs,pa) && !is_equal(*upfac,*downfac))
1660 fprintf(stderr, "NOTE: Specify -ntpr > 2 for both scaling factors to take effect.\n"
1661 "(upfac=%f, downfac=%f)\n", *upfac, *downfac);
1663 if (opt2parg_bSet("-upfac",npargs,pa))
1665 if (opt2parg_bSet("-downfac",npargs,pa))
1668 if ( (2 == *ntprs) && (opt2parg_bSet("-downfac",npargs,pa)) && !is_equal(*downfac, 1.0))
1673 /* If a fixed number of PME nodes is set we do rcoulomb and PME gird tuning
1674 * only. We need to check whether the requested number of PME-only nodes
1676 if (npme_fixed > -1)
1678 /* No more than 50% of all nodes can be assigned as PME-only nodes. */
1679 if (2*npme_fixed > nnodes)
1681 gmx_fatal(FARGS, "Cannot have more than %d PME-only nodes for a total of %d nodes (you chose %d).\n",
1682 nnodes/2, nnodes, npme_fixed);
1684 if ((npme_fixed > 0) && (5*npme_fixed < nnodes))
1686 fprintf(stderr, "WARNING: Only %g percent of the nodes are assigned as PME-only nodes.\n",
1687 100.0*((real)npme_fixed / (real)nnodes));
1689 if (opt2parg_bSet("-min",npargs,pa) || opt2parg_bSet("-max",npargs,pa))
1691 fprintf(stderr, "NOTE: The -min, -max, and -npme options have no effect when a\n"
1692 " fixed number of PME-only nodes is requested with -fix.\n");
1698 /* Returns TRUE when "opt" is a switch for g_tune_pme itself */
1699 static gmx_bool is_main_switch(char *opt)
1701 if ( (0 == strcmp(opt,"-s" ))
1702 || (0 == strcmp(opt,"-p" ))
1703 || (0 == strcmp(opt,"-launch" ))
1704 || (0 == strcmp(opt,"-r" ))
1705 || (0 == strcmp(opt,"-ntpr" ))
1706 || (0 == strcmp(opt,"-max" ))
1707 || (0 == strcmp(opt,"-min" ))
1708 || (0 == strcmp(opt,"-upfac" ))
1709 || (0 == strcmp(opt,"-downfac" ))
1710 || (0 == strcmp(opt,"-fix" ))
1711 || (0 == strcmp(opt,"-four" ))
1712 || (0 == strcmp(opt,"-steps" ))
1713 || (0 == strcmp(opt,"-simsteps" ))
1714 || (0 == strcmp(opt,"-resetstep"))
1715 || (0 == strcmp(opt,"-so" ))
1716 || (0 == strcmp(opt,"-npstring" ))
1717 || (0 == strcmp(opt,"-npme" ))
1718 || (0 == strcmp(opt,"-passall" )) )
1725 /* Returns TRUE when "opt" is needed at launch time */
1726 static gmx_bool is_launch_option(char *opt, gmx_bool bSet)
1735 /* Returns TRUE when "opt" is needed at launch time */
1736 static gmx_bool is_launch_file(char *opt, gmx_bool bSet)
1738 /* We need all options that were set on the command line
1739 * and that do not start with -b */
1740 if (0 == strncmp(opt,"-b", 2))
1750 /* Returns TRUE when "opt" gives an option needed for the benchmarks runs */
1751 static gmx_bool is_bench_option(char *opt, gmx_bool bSet)
1753 /* If option is set, we might need it for the benchmarks.
1754 * This includes -cpi */
1757 if ( (0 == strcmp(opt, "-append" ))
1758 || (0 == strcmp(opt, "-maxh" ))
1759 || (0 == strcmp(opt, "-deffnm" ))
1760 || (0 == strcmp(opt, "-resethway"))
1761 || (0 == strcmp(opt, "-cpnum" )) )
1771 /* Returns TRUE when "opt" defines a file which is needed for the benchmarks runs */
1772 static gmx_bool is_bench_file(char *opt, gmx_bool bSet, gmx_bool bOptional, gmx_bool bIsOutput)
1774 /* All options starting with "-b" are for _b_enchmark files exclusively */
1775 if (0 == strncmp(opt,"-b", 2))
1777 if (!bOptional || bSet)
1787 if (bSet) /* These are additional input files like -cpi -ei */
1795 /* Adds 'buf' to 'str' */
1796 static void add_to_string(char **str, char *buf)
1801 len = strlen(*str) + strlen(buf) + 1;
1807 /* Create the command line for the benchmark as well as for the real run */
1808 static void create_command_line_snippets(
1815 const char *procstring, /* How to pass the number of processors to $MPIRUN */
1816 char *cmd_np[], /* Actual command line snippet, e.g. '-np <N>' */
1817 char *cmd_args_bench[], /* command line arguments for benchmark runs */
1818 char *cmd_args_launch[], /* command line arguments for simulation run */
1819 char extra_args[]) /* Add this to the end of the command line */
1825 #define BUFLENGTH 255
1826 char buf[BUFLENGTH];
1827 char strbuf[BUFLENGTH];
1828 char strbuf2[BUFLENGTH];
1832 np_or_nt=strdup("-nt");
1834 np_or_nt=strdup("-np");
1836 /* strlen needs at least '\0' as a string: */
1837 snew(*cmd_args_bench ,1);
1838 snew(*cmd_args_launch,1);
1839 *cmd_args_launch[0]='\0';
1840 *cmd_args_bench[0] ='\0';
1843 /*******************************************/
1844 /* 1. Process other command line arguments */
1845 /*******************************************/
1846 for (i=0; i<npargs; i++)
1848 /* What command line switch are we currently processing: */
1849 opt = (char *)pa[i].option;
1850 /* Skip options not meant for mdrun */
1851 if (!is_main_switch(opt))
1853 /* Boolean arguments need to be generated in the -[no]argname format */
1854 if (pa[i].type == etBOOL)
1856 sprintf(strbuf,"-%s%s ",*pa[i].u.b ? "" : "no",opt+1);
1860 /* Print it to a string buffer, strip away trailing whitespaces that pa_val also returns: */
1861 sprintf(strbuf2, "%s", pa_val(&pa[i],buf,BUFLENGTH));
1863 sprintf(strbuf, "%s %s ", opt, strbuf2);
1866 /* We need the -np (or -nt) switch in a separate buffer - whether or not it was set! */
1867 if (0 == strcmp(opt,np_or_nt))
1869 if (strcmp(procstring, "none")==0 && !bThreads)
1871 /* Omit -np <N> entirely */
1873 sprintf(*cmd_np, " ");
1877 /* This is the normal case with -np <N> */
1878 snew(*cmd_np, strlen(procstring)+strlen(strbuf2)+4);
1879 sprintf(*cmd_np, " %s %s ", bThreads? "-nt" : procstring, strbuf2);
1884 if (is_bench_option(opt,pa[i].bSet))
1885 add_to_string(cmd_args_bench, strbuf);
1887 if (is_launch_option(opt,pa[i].bSet))
1888 add_to_string(cmd_args_launch, strbuf);
1894 /* Add equilibration steps to benchmark options */
1895 sprintf(strbuf, "-resetstep %d ", presteps);
1896 add_to_string(cmd_args_bench, strbuf);
1899 /********************/
1900 /* 2. Process files */
1901 /********************/
1902 for (i=0; i<nfile; i++)
1904 opt = (char *)fnm[i].opt;
1905 name = opt2fn(opt,nfile,fnm);
1907 /* Strbuf contains the options, now let's sort out where we need that */
1908 sprintf(strbuf, "%s %s ", opt, name);
1910 /* Skip options not meant for mdrun */
1911 if (!is_main_switch(opt))
1914 if ( is_bench_file(opt, opt2bSet(opt,nfile,fnm), is_optional(&fnm[i]), is_output(&fnm[i])) )
1916 /* All options starting with -b* need the 'b' removed,
1917 * therefore overwrite strbuf */
1918 if (0 == strncmp(opt, "-b", 2))
1919 sprintf(strbuf, "-%s %s ", &opt[2], name);
1921 add_to_string(cmd_args_bench, strbuf);
1924 if ( is_launch_file(opt,opt2bSet(opt,nfile,fnm)) )
1925 add_to_string(cmd_args_launch, strbuf);
1929 add_to_string(cmd_args_bench , extra_args);
1930 add_to_string(cmd_args_launch, extra_args);
1935 /* Set option opt */
1936 static void setopt(const char *opt,int nfile,t_filenm fnm[])
1940 for(i=0; (i<nfile); i++)
1941 if (strcmp(opt,fnm[i].opt)==0)
1942 fnm[i].flag |= ffSET;
1946 static void couple_files_options(int nfile, t_filenm fnm[])
1949 gmx_bool bSet,bBench;
1954 for (i=0; i<nfile; i++)
1956 opt = (char *)fnm[i].opt;
1957 bSet = ((fnm[i].flag & ffSET) != 0);
1958 bBench = (0 == strncmp(opt,"-b", 2));
1960 /* Check optional files */
1961 /* If e.g. -eo is set, then -beo also needs to be set */
1962 if (is_optional(&fnm[i]) && bSet && !bBench)
1964 sprintf(buf, "-b%s", &opt[1]);
1965 setopt(buf,nfile,fnm);
1967 /* If -beo is set, then -eo also needs to be! */
1968 if (is_optional(&fnm[i]) && bSet && bBench)
1970 sprintf(buf, "-%s", &opt[2]);
1971 setopt(buf,nfile,fnm);
1977 static double gettime()
1979 #ifdef HAVE_GETTIMEOFDAY
1983 gettimeofday(&t,NULL);
1985 seconds = (double) t.tv_sec + 1e-6*(double)t.tv_usec;
1991 seconds = time(NULL);
1998 #define BENCHSTEPS (1000)
2000 int gmx_tune_pme(int argc,char *argv[])
2002 const char *desc[] = {
2003 "For a given number [TT]-np[tt] or [TT]-nt[tt] of processors/threads, this program systematically",
2004 "times [TT]mdrun[tt] with various numbers of PME-only nodes and determines",
2005 "which setting is fastest. It will also test whether performance can",
2006 "be enhanced by shifting load from the reciprocal to the real space",
2007 "part of the Ewald sum. ",
2008 "Simply pass your [TT].tpr[tt] file to [TT]g_tune_pme[tt] together with other options",
2009 "for [TT]mdrun[tt] as needed.[PAR]",
2010 "Which executables are used can be set in the environment variables",
2011 "MPIRUN and MDRUN. If these are not present, 'mpirun' and 'mdrun'",
2012 "will be used as defaults. Note that for certain MPI frameworks you",
2013 "need to provide a machine- or hostfile. This can also be passed",
2014 "via the MPIRUN variable, e.g.[PAR]",
2015 "[TT]export MPIRUN=\"/usr/local/mpirun -machinefile hosts\"[tt][PAR]",
2016 "Please call [TT]g_tune_pme[tt] with the normal options you would pass to",
2017 "[TT]mdrun[tt] and add [TT]-np[tt] for the number of processors to perform the",
2018 "tests on, or [TT]-nt[tt] for the number of threads. You can also add [TT]-r[tt]",
2019 "to repeat each test several times to get better statistics. [PAR]",
2020 "[TT]g_tune_pme[tt] can test various real space / reciprocal space workloads",
2021 "for you. With [TT]-ntpr[tt] you control how many extra [TT].tpr[tt] files will be",
2022 "written with enlarged cutoffs and smaller fourier grids respectively.",
2023 "Typically, the first test (number 0) will be with the settings from the input",
2024 "[TT].tpr[tt] file; the last test (number [TT]ntpr[tt]) will have cutoffs multiplied",
2025 "by (and at the same time fourier grid dimensions divided by) the scaling",
2026 "factor [TT]-fac[tt] (default 1.2). The remaining [TT].tpr[tt] files will have about ",
2027 "equally-spaced values in between these extremes. [BB]Note[bb] that you can set [TT]-ntpr[tt] to 1",
2028 "if you just want to find the optimal number of PME-only nodes; in that case",
2029 "your input [TT].tpr[tt] file will remain unchanged.[PAR]",
2030 "For the benchmark runs, the default of 1000 time steps should suffice for most",
2031 "MD systems. The dynamic load balancing needs about 100 time steps",
2032 "to adapt to local load imbalances, therefore the time step counters",
2033 "are by default reset after 100 steps. For large systems",
2034 "(>1M atoms) you may have to set [TT]-resetstep[tt] to a higher value.",
2035 "From the 'DD' load imbalance entries in the md.log output file you",
2036 "can tell after how many steps the load is sufficiently balanced. Example call:[PAR]"
2037 "[TT]g_tune_pme -np 64 -s protein.tpr -launch[tt][PAR]",
2038 "After calling [TT]mdrun[tt] several times, detailed performance information",
2039 "is available in the output file [TT]perf.out.[tt] ",
2040 "[BB]Note[bb] that during the benchmarks, a couple of temporary files are written",
2041 "(options [TT]-b[tt]*), these will be automatically deleted after each test.[PAR]",
2042 "If you want the simulation to be started automatically with the",
2043 "optimized parameters, use the command line option [TT]-launch[tt].[PAR]",
2048 int pmeentries=0; /* How many values for -npme do we actually test for each tpr file */
2049 real maxPMEfraction=0.50;
2050 real minPMEfraction=0.25;
2051 int maxPMEnodes, minPMEnodes;
2052 int npme_fixed=-2; /* If >= -1, use only this number
2053 * of PME-only nodes */
2054 real downfac=1.0,upfac=1.2;
2056 real fs=0.0; /* 0 indicates: not set by the user */
2057 gmx_large_int_t bench_nsteps=BENCHSTEPS;
2058 gmx_large_int_t new_sim_nsteps=-1; /* -1 indicates: not set by the user */
2059 gmx_large_int_t cpt_steps=0; /* Step counter in .cpt input file */
2060 int presteps=100; /* Do a full cycle reset after presteps steps */
2061 gmx_bool bOverwrite=FALSE, bKeepTPR;
2062 gmx_bool bLaunch=FALSE;
2063 gmx_bool bPassAll=FALSE;
2064 char *ExtraArgs=NULL;
2065 char **tpr_names=NULL;
2066 const char *simulation_tpr=NULL;
2067 int best_npme, best_tpr;
2068 int sim_part = 1; /* For benchmarks with checkpoint files */
2070 /* Default program names if nothing else is found */
2071 char *cmd_mpirun=NULL, *cmd_mdrun=NULL;
2072 char *cmd_args_bench, *cmd_args_launch;
2075 t_perf **perfdata=NULL;
2081 /* Print out how long the tuning took */
2084 static t_filenm fnm[] = {
2086 { efOUT, "-p", "perf", ffWRITE },
2087 { efLOG, "-err", "errors", ffWRITE },
2088 { efTPX, "-so", "tuned", ffWRITE },
2090 { efTPX, NULL, NULL, ffREAD },
2091 { efTRN, "-o", NULL, ffWRITE },
2092 { efXTC, "-x", NULL, ffOPTWR },
2093 { efCPT, "-cpi", NULL, ffOPTRD },
2094 { efCPT, "-cpo", NULL, ffOPTWR },
2095 { efSTO, "-c", "confout", ffWRITE },
2096 { efEDR, "-e", "ener", ffWRITE },
2097 { efLOG, "-g", "md", ffWRITE },
2098 { efXVG, "-dhdl", "dhdl", ffOPTWR },
2099 { efXVG, "-field", "field", ffOPTWR },
2100 { efXVG, "-table", "table", ffOPTRD },
2101 { efXVG, "-tablep", "tablep", ffOPTRD },
2102 { efXVG, "-tableb", "table", ffOPTRD },
2103 { efTRX, "-rerun", "rerun", ffOPTRD },
2104 { efXVG, "-tpi", "tpi", ffOPTWR },
2105 { efXVG, "-tpid", "tpidist", ffOPTWR },
2106 { efEDI, "-ei", "sam", ffOPTRD },
2107 { efEDO, "-eo", "sam", ffOPTWR },
2108 { efGCT, "-j", "wham", ffOPTRD },
2109 { efGCT, "-jo", "bam", ffOPTWR },
2110 { efXVG, "-ffout", "gct", ffOPTWR },
2111 { efXVG, "-devout", "deviatie", ffOPTWR },
2112 { efXVG, "-runav", "runaver", ffOPTWR },
2113 { efXVG, "-px", "pullx", ffOPTWR },
2114 { efXVG, "-pf", "pullf", ffOPTWR },
2115 { efXVG, "-ro", "rotation", ffOPTWR },
2116 { efLOG, "-ra", "rotangles",ffOPTWR },
2117 { efLOG, "-rs", "rotslabs", ffOPTWR },
2118 { efLOG, "-rt", "rottorque",ffOPTWR },
2119 { efMTX, "-mtx", "nm", ffOPTWR },
2120 { efNDX, "-dn", "dipole", ffOPTWR },
2121 /* Output files that are deleted after each benchmark run */
2122 { efTRN, "-bo", "bench", ffWRITE },
2123 { efXTC, "-bx", "bench", ffWRITE },
2124 { efCPT, "-bcpo", "bench", ffWRITE },
2125 { efSTO, "-bc", "bench", ffWRITE },
2126 { efEDR, "-be", "bench", ffWRITE },
2127 { efLOG, "-bg", "bench", ffWRITE },
2128 { efEDO, "-beo", "bench", ffOPTWR },
2129 { efXVG, "-bdhdl", "benchdhdl",ffOPTWR },
2130 { efXVG, "-bfield", "benchfld" ,ffOPTWR },
2131 { efXVG, "-btpi", "benchtpi", ffOPTWR },
2132 { efXVG, "-btpid", "benchtpid",ffOPTWR },
2133 { efGCT, "-bjo", "bench", ffOPTWR },
2134 { efXVG, "-bffout", "benchgct", ffOPTWR },
2135 { efXVG, "-bdevout","benchdev", ffOPTWR },
2136 { efXVG, "-brunav", "benchrnav",ffOPTWR },
2137 { efXVG, "-bpx", "benchpx", ffOPTWR },
2138 { efXVG, "-bpf", "benchpf", ffOPTWR },
2139 { efXVG, "-bro", "benchrot", ffOPTWR },
2140 { efLOG, "-bra", "benchrota",ffOPTWR },
2141 { efLOG, "-brs", "benchrots",ffOPTWR },
2142 { efLOG, "-brt", "benchrott",ffOPTWR },
2143 { efMTX, "-bmtx", "benchn", ffOPTWR },
2144 { efNDX, "-bdn", "bench", ffOPTWR }
2147 /* Command line options of mdrun */
2148 gmx_bool bDDBondCheck = TRUE;
2149 gmx_bool bDDBondComm = TRUE;
2150 gmx_bool bVerbose = FALSE;
2151 gmx_bool bCompact = TRUE;
2152 gmx_bool bSepPot = FALSE;
2153 gmx_bool bRerunVSite = FALSE;
2154 gmx_bool bIonize = FALSE;
2155 gmx_bool bConfout = TRUE;
2156 gmx_bool bReproducible = FALSE;
2157 gmx_bool bThreads = FALSE;
2160 int nstglobalcomm=-1;
2162 int repl_ex_seed=-1;
2166 const char *ddno_opt[ddnoNR+1] =
2167 { NULL, "interleave", "pp_pme", "cartesian", NULL };
2168 const char *dddlb_opt[] =
2169 { NULL, "auto", "no", "yes", NULL };
2170 const char *procstring[] =
2171 { NULL, "-np", "-n", "none", NULL };
2172 const char *npmevalues_opt[] =
2173 { NULL, "auto", "all", "subset", NULL };
2174 real rdd=0.0,rconstr=0.0,dlb_scale=0.8,pforce=-1;
2175 char *ddcsx=NULL,*ddcsy=NULL,*ddcsz=NULL;
2177 #define STD_CPT_PERIOD (15.0)
2178 real cpt_period=STD_CPT_PERIOD,max_hours=-1;
2179 gmx_bool bAppendFiles=TRUE;
2180 gmx_bool bKeepAndNumCPT=FALSE;
2181 gmx_bool bResetCountersHalfWay=FALSE;
2182 output_env_t oenv=NULL;
2185 /***********************/
2186 /* g_tune_pme options: */
2187 /***********************/
2188 { "-np", FALSE, etINT, {&nnodes},
2189 "Number of nodes to run the tests on (must be > 2 for separate PME nodes)" },
2190 { "-npstring", FALSE, etENUM, {procstring},
2191 "Specify the number of processors to [TT]$MPIRUN[tt] using this string"},
2192 { "-passall", FALSE, etBOOL, {&bPassAll},
2193 "HIDDENPut arguments unknown to [TT]mdrun[tt] at the end of the command line. Can be used for debugging purposes. "},
2194 { "-nt", FALSE, etINT, {&nthreads},
2195 "Number of threads to run the tests on (turns MPI & mpirun off)"},
2196 { "-r", FALSE, etINT, {&repeats},
2197 "Repeat each test this often" },
2198 { "-max", FALSE, etREAL, {&maxPMEfraction},
2199 "Max fraction of PME nodes to test with" },
2200 { "-min", FALSE, etREAL, {&minPMEfraction},
2201 "Min fraction of PME nodes to test with" },
2202 { "-npme", FALSE, etENUM, {npmevalues_opt},
2203 "Benchmark all possible values for [TT]-npme[tt] or just the subset that is expected to perform well"},
2204 { "-fix", FALSE, etINT, {&npme_fixed},
2205 "If >= -1, do not vary the number of PME-only nodes, instead use this fixed value and only vary rcoulomb and the PME grid spacing."},
2206 { "-upfac", FALSE, etREAL, {&upfac},
2207 "Upper limit for rcoulomb scaling factor (Note that rcoulomb upscaling results in fourier grid downscaling)" },
2208 { "-downfac", FALSE, etREAL, {&downfac},
2209 "Lower limit for rcoulomb scaling factor" },
2210 { "-ntpr", FALSE, etINT, {&ntprs},
2211 "Number of [TT].tpr[tt] files to benchmark. Create this many files with scaling factors ranging from 1.0 to fac. If < 1, automatically choose the number of [TT].tpr[tt] files to test" },
2212 { "-four", FALSE, etREAL, {&fs},
2213 "Use this fourierspacing value instead of the grid found in the [TT].tpr[tt] input file. (Spacing applies to a scaling factor of 1.0 if multiple [TT].tpr[tt] files are written)" },
2214 { "-steps", FALSE, etGMX_LARGE_INT, {&bench_nsteps},
2215 "Take timings for this many steps in the benchmark runs" },
2216 { "-resetstep",FALSE, etINT, {&presteps},
2217 "Let dlb equilibrate this many steps before timings are taken (reset cycle counters after this many steps)" },
2218 { "-simsteps", FALSE, etGMX_LARGE_INT, {&new_sim_nsteps},
2219 "If non-negative, perform this many steps in the real run (overwrites nsteps from [TT].tpr[tt], add [TT].cpt[tt] steps)" },
2220 { "-launch", FALSE, etBOOL, {&bLaunch},
2221 "Launch the real simulation after optimization" },
2222 /******************/
2223 /* mdrun options: */
2224 /******************/
2225 { "-deffnm", FALSE, etSTR, {&deffnm},
2226 "Set the default filename for all file options at launch time" },
2227 { "-ddorder", FALSE, etENUM, {ddno_opt},
2229 { "-ddcheck", FALSE, etBOOL, {&bDDBondCheck},
2230 "Check for all bonded interactions with DD" },
2231 { "-ddbondcomm",FALSE, etBOOL, {&bDDBondComm},
2232 "HIDDENUse special bonded atom communication when [TT]-rdd[tt] > cut-off" },
2233 { "-rdd", FALSE, etREAL, {&rdd},
2234 "The maximum distance for bonded interactions with DD (nm), 0 is determine from initial coordinates" },
2235 { "-rcon", FALSE, etREAL, {&rconstr},
2236 "Maximum distance for P-LINCS (nm), 0 is estimate" },
2237 { "-dlb", FALSE, etENUM, {dddlb_opt},
2238 "Dynamic load balancing (with DD)" },
2239 { "-dds", FALSE, etREAL, {&dlb_scale},
2240 "Minimum allowed dlb scaling of the DD cell size" },
2241 { "-ddcsx", FALSE, etSTR, {&ddcsx},
2242 "HIDDENThe DD cell sizes in x" },
2243 { "-ddcsy", FALSE, etSTR, {&ddcsy},
2244 "HIDDENThe DD cell sizes in y" },
2245 { "-ddcsz", FALSE, etSTR, {&ddcsz},
2246 "HIDDENThe DD cell sizes in z" },
2247 { "-gcom", FALSE, etINT, {&nstglobalcomm},
2248 "Global communication frequency" },
2249 { "-v", FALSE, etBOOL, {&bVerbose},
2250 "Be loud and noisy" },
2251 { "-compact", FALSE, etBOOL, {&bCompact},
2252 "Write a compact log file" },
2253 { "-seppot", FALSE, etBOOL, {&bSepPot},
2254 "Write separate V and dVdl terms for each interaction type and node to the log file(s)" },
2255 { "-pforce", FALSE, etREAL, {&pforce},
2256 "Print all forces larger than this (kJ/mol nm)" },
2257 { "-reprod", FALSE, etBOOL, {&bReproducible},
2258 "Try to avoid optimizations that affect binary reproducibility" },
2259 { "-cpt", FALSE, etREAL, {&cpt_period},
2260 "Checkpoint interval (minutes)" },
2261 { "-cpnum", FALSE, etBOOL, {&bKeepAndNumCPT},
2262 "Keep and number checkpoint files" },
2263 { "-append", FALSE, etBOOL, {&bAppendFiles},
2264 "Append to previous output files when continuing from checkpoint instead of adding the simulation part number to all file names (for launch only)" },
2265 { "-maxh", FALSE, etREAL, {&max_hours},
2266 "Terminate after 0.99 times this time (hours)" },
2267 { "-multi", FALSE, etINT, {&nmultisim},
2268 "Do multiple simulations in parallel" },
2269 { "-replex", FALSE, etINT, {&repl_ex_nst},
2270 "Attempt replica exchange every # steps" },
2271 { "-reseed", FALSE, etINT, {&repl_ex_seed},
2272 "Seed for replica exchange, -1 is generate a seed" },
2273 { "-rerunvsite", FALSE, etBOOL, {&bRerunVSite},
2274 "HIDDENRecalculate virtual site coordinates with [TT]-rerun[tt]" },
2275 { "-ionize", FALSE, etBOOL, {&bIonize},
2276 "Do a simulation including the effect of an X-ray bombardment on your system" },
2277 { "-confout", FALSE, etBOOL, {&bConfout},
2278 "HIDDENWrite the last configuration with [TT]-c[tt] and force checkpointing at the last step" },
2279 { "-stepout", FALSE, etINT, {&nstepout},
2280 "HIDDENFrequency of writing the remaining runtime" },
2281 { "-resethway", FALSE, etBOOL, {&bResetCountersHalfWay},
2282 "HIDDENReset the cycle counters after half the number of steps or halfway [TT]-maxh[tt] (launch only)" }
2286 #define NFILE asize(fnm)
2288 CopyRight(stderr,argv[0]);
2290 seconds = gettime();
2292 parse_common_args(&argc,argv,PCA_NOEXIT_ON_ARGS,
2293 NFILE,fnm,asize(pa),pa,asize(desc),desc,
2296 /* Store the remaining unparsed command line entries in a string */
2298 ExtraArgs[0] = '\0';
2299 for (i=1; i<argc; i++) /* argc will now be 1 if everything was understood */
2301 add_to_string(&ExtraArgs, argv[i]);
2302 add_to_string(&ExtraArgs, " ");
2304 if ( !bPassAll && (ExtraArgs[0] != '\0') )
2306 fprintf(stderr, "\nWARNING: The following arguments you provided have no effect:\n"
2308 "Use the -passall option to force them to appear on the command lines\n"
2309 "for the benchmark simulations%s.\n\n",
2310 ExtraArgs, bLaunch? " and at launch time" : "");
2313 if (opt2parg_bSet("-nt",asize(pa),pa))
2316 if (opt2parg_bSet("-npstring",asize(pa),pa))
2317 fprintf(stderr, "WARNING: -npstring has no effect when using threads.\n");
2320 gmx_fatal(FARGS, "Can't run multi-threaded MPI simulation yet!");
2321 /* and now we just set this; a bit of an ugly hack*/
2324 /* Automatically set -beo options if -eo is set etc. */
2325 couple_files_options(NFILE,fnm);
2327 /* Construct the command line arguments for benchmark runs
2328 * as well as for the simulation run
2330 create_command_line_snippets(bThreads,presteps,NFILE,fnm,asize(pa),pa,procstring[0],
2331 &cmd_np, &cmd_args_bench, &cmd_args_launch,
2332 bPassAll? ExtraArgs : (char *)"");
2334 /* Read in checkpoint file if requested */
2336 if(opt2bSet("-cpi",NFILE,fnm))
2339 cr->duty=DUTY_PP; /* makes the following routine happy */
2340 read_checkpoint_simulation_part(opt2fn("-cpi",NFILE,fnm),
2341 &sim_part,&cpt_steps,cr,
2342 FALSE,NFILE,fnm,NULL,NULL);
2345 /* sim_part will now be 1 if no checkpoint file was found */
2347 gmx_fatal(FARGS, "Checkpoint file %s not found!", opt2fn("-cpi",
2352 /* Open performance output file and write header info */
2353 fp = ffopen(opt2fn("-p",NFILE,fnm),"w");
2355 /* Make a quick consistency check of command line parameters */
2356 check_input(nnodes, repeats, &ntprs, &upfac, &downfac,
2357 maxPMEfraction, minPMEfraction, npme_fixed,
2358 fs, bench_nsteps, fnm, NFILE, sim_part, presteps,
2361 /* Determine the maximum and minimum number of PME nodes to test,
2362 * the actual list of settings is build in do_the_tests(). */
2363 if ((nnodes > 2) && (npme_fixed < -1))
2365 maxPMEnodes = floor(maxPMEfraction*nnodes);
2366 minPMEnodes = max(floor(minPMEfraction*nnodes), 0);
2367 fprintf(stdout, "Will try runs with %d ", minPMEnodes);
2368 if (maxPMEnodes != minPMEnodes)
2369 fprintf(stdout, "- %d ", maxPMEnodes);
2370 fprintf(stdout, "PME-only nodes.\n Note that the automatic number of PME-only nodes and no separate PME nodes are always tested.\n");
2378 /* Get the commands we need to set up the runs from environment variables */
2379 get_program_paths(bThreads, &cmd_mpirun, cmd_np, &cmd_mdrun, repeats);
2381 /* Print some header info to file */
2383 fprintf(fp, "\n P E R F O R M A N C E R E S U L T S\n");
2385 fprintf(fp, "%s for Gromacs %s\n", ShortProgram(),GromacsVersion());
2388 fprintf(fp, "Number of nodes : %d\n", nnodes);
2389 fprintf(fp, "The mpirun command is : %s\n", cmd_mpirun);
2390 if ( strcmp(procstring[0], "none") != 0)
2391 fprintf(fp, "Passing # of nodes via : %s\n", procstring[0]);
2393 fprintf(fp, "Not setting number of nodes in system call\n");
2396 fprintf(fp, "Number of threads : %d\n", nnodes);
2398 fprintf(fp, "The mdrun command is : %s\n", cmd_mdrun);
2399 fprintf(fp, "mdrun args benchmarks : %s\n", cmd_args_bench);
2400 fprintf(fp, "Benchmark steps : ");
2401 fprintf(fp, gmx_large_int_pfmt, bench_nsteps);
2403 fprintf(fp, "dlb equilibration steps : %d\n", presteps);
2406 fprintf(fp, "Checkpoint time step : ");
2407 fprintf(fp, gmx_large_int_pfmt, cpt_steps);
2411 fprintf(fp, "mdrun args at launchtime: %s\n", cmd_args_launch);
2412 if (!bPassAll && ExtraArgs[0] != '\0')
2413 fprintf(fp, "Unused arguments : %s\n", ExtraArgs);
2414 if (new_sim_nsteps >= 0)
2417 fprintf(stderr, "Note: Simulation input file %s will have ", opt2fn("-so",NFILE,fnm));
2418 fprintf(stderr, gmx_large_int_pfmt, new_sim_nsteps+cpt_steps);
2419 fprintf(stderr, " steps.\n");
2420 fprintf(fp, "Simulation steps : ");
2421 fprintf(fp, gmx_large_int_pfmt, new_sim_nsteps);
2425 fprintf(fp, "Repeats for each test : %d\n", repeats);
2429 fprintf(fp, "Requested grid spacing : %f (This will be the grid spacing at a scaling factor of 1.0)\n", fs);
2432 if (npme_fixed >= -1)
2434 fprintf(fp, "Fixing -npme at : %d\n", npme_fixed);
2437 fprintf(fp, "Input file : %s\n", opt2fn("-s",NFILE,fnm));
2439 /* Allocate memory for the inputinfo struct: */
2441 info->nr_inputfiles = ntprs;
2442 for (i=0; i<ntprs; i++)
2444 snew(info->rcoulomb , ntprs);
2445 snew(info->rvdw , ntprs);
2446 snew(info->rlist , ntprs);
2447 snew(info->rlistlong, ntprs);
2448 snew(info->nkx , ntprs);
2449 snew(info->nky , ntprs);
2450 snew(info->nkz , ntprs);
2451 snew(info->fsx , ntprs);
2452 snew(info->fsy , ntprs);
2453 snew(info->fsz , ntprs);
2455 /* Make alternative tpr files to test: */
2456 snew(tpr_names, ntprs);
2457 for (i=0; i<ntprs; i++)
2458 snew(tpr_names[i], STRLEN);
2460 /* It can be that ntprs is reduced by make_benchmark_tprs if not enough
2461 * different grids could be found. */
2462 make_benchmark_tprs(opt2fn("-s",NFILE,fnm), tpr_names, bench_nsteps+presteps,
2463 cpt_steps, upfac, downfac, &ntprs, fs, info, fp);
2466 /********************************************************************************/
2467 /* Main loop over all scenarios we need to test: tpr files, PME nodes, repeats */
2468 /********************************************************************************/
2469 snew(perfdata, ntprs);
2470 do_the_tests(fp, tpr_names, maxPMEnodes, minPMEnodes, npme_fixed, npmevalues_opt[0], perfdata, &pmeentries,
2471 repeats, nnodes, ntprs, bThreads, cmd_mpirun, cmd_np, cmd_mdrun,
2472 cmd_args_bench, fnm, NFILE, sim_part, presteps, cpt_steps);
2474 fprintf(fp, "\nTuning took%8.1f minutes.\n", (gettime()-seconds)/60.0);
2476 /* Analyse the results and give a suggestion for optimal settings: */
2477 bKeepTPR = analyze_data(fp, opt2fn("-p", NFILE, fnm), perfdata, nnodes, ntprs, pmeentries,
2478 repeats, info, &best_tpr, &best_npme);
2480 /* Take the best-performing tpr file and enlarge nsteps to original value */
2481 if ( bKeepTPR && !bOverwrite && !(fs > 0.0) )
2483 simulation_tpr = opt2fn("-s",NFILE,fnm);
2487 simulation_tpr = opt2fn("-so",NFILE,fnm);
2488 modify_PMEsettings(bOverwrite? (new_sim_nsteps+cpt_steps) :
2489 info->orig_sim_steps, tpr_names[best_tpr],
2493 /* Now start the real simulation if the user requested it ... */
2494 launch_simulation(bLaunch, fp, bThreads, cmd_mpirun, cmd_np, cmd_mdrun,
2495 cmd_args_launch, simulation_tpr, nnodes, best_npme);
2498 /* ... or simply print the performance results to screen: */
2500 finalize(opt2fn("-p", NFILE, fnm));