src/gromacs/gmxlib/nrnb.cpp

   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) 2013,2014,2015,2017,2018 by the GROMACS development team.
   7  * Copyright (c) 2019,2020, by the GROMACS development team, led by
   8  * Mark Abraham, David van der Spoel, Berk Hess, and Erik Lindahl,
   9  * and including many others, as listed in the AUTHORS file in the
  10  * top-level source directory and at http://www.gromacs.org.
  11  *
  12  * GROMACS is free software; you can redistribute it and/or
  13  * modify it under the terms of the GNU Lesser General Public License
  14  * as published by the Free Software Foundation; either version 2.1
  15  * of the License, or (at your option) any later version.
  16  *
  17  * GROMACS is distributed in the hope that it will be useful,
  18  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  19  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  20  * Lesser General Public License for more details.
  21  *
  22  * You should have received a copy of the GNU Lesser General Public
  23  * License along with GROMACS; if not, see
  24  * http://www.gnu.org/licenses, or write to the Free Software Foundation,
  25  * Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02110-1301  USA.
  26  *
  27  * If you want to redistribute modifications to GROMACS, please
  28  * consider that scientific software is very special. Version
  29  * control is crucial - bugs must be traceable. We will be happy to
  30  * consider code for inclusion in the official distribution, but
  31  * derived work must not be called official GROMACS. Details are found
  32  * in the README & COPYING files - if they are missing, get the
  33  * official version at http://www.gromacs.org.
  34  *
  35  * To help us fund GROMACS development, we humbly ask that you cite
  36  * the research papers on the package. Check out http://www.gromacs.org.
  37  */
  38 #include "gmxpre.h"
  39
  40 #include "nrnb.h"
  41
  42 #include <cstdlib>
  43 #include <cstring>
  44
  45 #include <algorithm>
  46 #include <vector>
  47
  48 #include "gromacs/mdtypes/commrec.h"
  49 #include "gromacs/mdtypes/md_enums.h"
  50 #include "gromacs/utility/arraysize.h"
  51
  52 typedef struct
  53 {
  54     const char* name;
  55     int         flop;
  56 } t_nrnb_data;
  57
  58
  59 static const t_nrnb_data nbdata[eNRNB] = {
  60     /* These are re-used for different NB kernels, since there are so many.
  61      * The actual number of flops is set dynamically.
  62      */
  63     { "NB VdW [V&F]", 1 },
  64     { "NB VdW [F]", 1 },
  65     { "NB Elec. [V&F]", 1 },
  66     { "NB Elec. [F]", 1 },
  67     { "NB Elec. [W3,V&F]", 1 },
  68     { "NB Elec. [W3,F]", 1 },
  69     { "NB Elec. [W3-W3,V&F]", 1 },
  70     { "NB Elec. [W3-W3,F]", 1 },
  71     { "NB Elec. [W4,V&F]", 1 },
  72     { "NB Elec. [W4,F]", 1 },
  73     { "NB Elec. [W4-W4,V&F]", 1 },
  74     { "NB Elec. [W4-W4,F]", 1 },
  75     { "NB VdW & Elec. [V&F]", 1 },
  76     { "NB VdW & Elec. [F]", 1 },
  77     { "NB VdW & Elec. [W3,V&F]", 1 },
  78     { "NB VdW & Elec. [W3,F]", 1 },
  79     { "NB VdW & Elec. [W3-W3,V&F]", 1 },
  80     { "NB VdW & Elec. [W3-W3,F]", 1 },
  81     { "NB VdW & Elec. [W4,V&F]", 1 },
  82     { "NB VdW & Elec. [W4,F]", 1 },
  83     { "NB VdW & Elec. [W4-W4,V&F]", 1 },
  84     { "NB VdW & Elec. [W4-W4,F]", 1 },
  85
  86     { "NB Generic kernel", 1 },
  87     { "NB Generic charge grp kernel", 1 },
  88     { "NB Free energy kernel", 1 },
  89
  90     { "Pair Search distance check", 9 }, /* nbnxn pair dist. check */
  91     /* nbnxn kernel flops are based on inner-loops without exclusion checks.
  92      * Plain Coulomb runs through the RF kernels, except with GPUs.
  93      * invsqrt is counted as 6 flops: 1 for _mm_rsqt_ps + 5 for iteration.
  94      * The flops are equal for plain-C, x86 SIMD and GPUs, except for:
  95      * - plain-C kernel uses one flop more for Coulomb-only (F) than listed
  96      * - x86 SIMD LJ geom-comb.rule kernels (fastest) use 2 more flops
  97      * - x86 SIMD LJ LB-comb.rule kernels (fast) use 3 (8 for F+E) more flops
  98      * - GPU always does exclusions, which requires 2-4 flops, but as invsqrt
  99      *   is always counted as 6 flops, this roughly compensates.
 100      */
 101     { "NxN RF Elec. + LJ [F]", 38 }, /* nbnxn kernel LJ+RF, no ener */
 102     { "NxN RF Elec. + LJ [V&F]", 54 },
 103     { "NxN QSTab Elec. + LJ [F]", 41 }, /* nbnxn kernel LJ+tab, no en */
 104     { "NxN QSTab Elec. + LJ [V&F]", 59 },
 105     { "NxN Ewald Elec. + LJ [F]", 66 }, /* nbnxn kernel LJ+Ewald, no en */
 106     { "NxN Ewald Elec. + LJ [V&F]", 107 },
 107     { "NxN LJ [F]", 33 }, /* nbnxn kernel LJ, no ener */
 108     { "NxN LJ [V&F]", 43 },
 109     { "NxN RF Electrostatics [F]", 31 }, /* nbnxn kernel RF, no ener */
 110     { "NxN RF Electrostatics [V&F]", 36 },
 111     { "NxN QSTab Elec. [F]", 34 }, /* nbnxn kernel tab, no ener */
 112     { "NxN QSTab Elec. [V&F]", 41 },
 113     { "NxN Ewald Elec. [F]", 61 }, /* nbnxn kernel Ewald, no ener */
 114     { "NxN Ewald Elec. [V&F]", 84 },
 115     /* The switch function flops should be added to the LJ kernels above */
 116     { "NxN LJ add F-switch [F]", 12 }, /* extra cost for LJ F-switch */
 117     { "NxN LJ add F-switch [V&F]", 22 },
 118     { "NxN LJ add P-switch [F]", 27 }, /* extra cost for LJ P-switch */
 119     { "NxN LJ add P-switch [V&F]", 20 },
 120     { "NxN LJ add LJ Ewald [F]", 36 }, /* extra cost for LJ Ewald */
 121     { "NxN LJ add LJ Ewald [V&F]", 33 },
 122     { "1,4 nonbonded interactions", 90 },
 123     { "Calc Weights", 36 },
 124     { "Spread Q", 6 },
 125     { "Spread Q Bspline", 2 },
 126     { "Gather F", 23 },
 127     { "Gather F Bspline", 6 },
 128     { "3D-FFT", 8 },
 129     { "Convolution", 4 },
 130     { "Solve PME", 64 },
 131     { "NS-Pairs", 21 },
 132     { "Reset In Box", 3 },
 133     { "Shift-X", 6 },
 134     { "CG-CoM", 3 },
 135     { "Sum Forces", 1 },
 136     { "Bonds", 59 },
 137     { "G96Bonds", 44 },
 138     { "FENE Bonds", 58 },
 139     { "Tab. Bonds", 62 },
 140     { "Restraint Potential", 86 },
 141     { "Linear Angles", 57 },
 142     { "Angles", 168 },
 143     { "G96Angles", 150 },
 144     { "Quartic Angles", 160 },
 145     { "Tab. Angles", 169 },
 146     { "Propers", 229 },
 147     { "Impropers", 208 },
 148     { "RB-Dihedrals", 247 },
 149     { "Four. Dihedrals", 247 },
 150     { "Tab. Dihedrals", 227 },
 151     { "Dist. Restr.", 200 },
 152     { "Orient. Restr.", 200 },
 153     { "Dihedral Restr.", 200 },
 154     { "Pos. Restr.", 50 },
 155     { "Flat-bottom posres", 50 },
 156     { "Angle Restr.", 191 },
 157     { "Angle Restr. Z", 164 },
 158     { "Morse Potent.", 83 },
 159     { "Cubic Bonds", 54 },
 160     { "Walls", 31 },
 161     { "Polarization", 59 },
 162     { "Anharmonic Polarization", 72 },
 163     { "Water Pol.", 62 },
 164     { "Thole Pol.", 296 },
 165     { "Virial", 18 },
 166     { "Update", 31 },
 167     { "Ext.ens. Update", 54 },
 168     { "Stop-CM", 10 },
 169     { "P-Coupling", 6 },
 170     { "Calc-Ekin", 27 },
 171     { "Lincs", 60 },
 172     { "Lincs-Mat", 4 },
 173     { "Shake", 30 },
 174     { "Constraint-V", 9 },
 175     { "Shake-Init", 10 },
 176     { "Constraint-Vir", 24 },
 177     { "Settle", 370 },
 178     { "Virtual Site 1", 1 },
 179     { "Virtual Site 2", 23 },
 180     { "Virtual Site 2fd", 63 },
 181     { "Virtual Site 3", 37 },
 182     { "Virtual Site 3fd", 95 },
 183     { "Virtual Site 3fad", 176 },
 184     { "Virtual Site 3out", 87 },
 185     { "Virtual Site 4fd", 110 },
 186     { "Virtual Site 4fdn", 254 },
 187     { "Virtual Site N", 15 },
 188     { "CMAP", 1700 }, // Estimate!
 189     { "Urey-Bradley", 183 },
 190     { "Cross-Bond-Bond", 163 },
 191     { "Cross-Bond-Angle", 163 }
 192 };
 193
 194 static void pr_two(FILE* out, int c, int i)
 195 {
 196     if (i < 10)
 197     {
 198         fprintf(out, "%c0%1d", c, i);
 199     }
 200     else
 201     {
 202         fprintf(out, "%c%2d", c, i);
 203     }
 204 }
 205
 206 static void pr_difftime(FILE* out, double dt)
 207 {
 208     int      ndays, nhours, nmins, nsecs;
 209     gmx_bool bPrint, bPrinted;
 210
 211     ndays    = static_cast<int>(dt / (24 * 3600));
 212     dt       = dt - 24 * 3600 * ndays;
 213     nhours   = static_cast<int>(dt / 3600);
 214     dt       = dt - 3600 * nhours;
 215     nmins    = static_cast<int>(dt / 60);
 216     dt       = dt - nmins * 60;
 217     nsecs    = static_cast<int>(dt);
 218     bPrint   = (ndays > 0);
 219     bPrinted = bPrint;
 220     if (bPrint)
 221     {
 222         fprintf(out, "%d", ndays);
 223     }
 224     bPrint = bPrint || (nhours > 0);
 225     if (bPrint)
 226     {
 227         if (bPrinted)
 228         {
 229             pr_two(out, 'd', nhours);
 230         }
 231         else
 232         {
 233             fprintf(out, "%d", nhours);
 234         }
 235     }
 236     bPrinted = bPrinted || bPrint;
 237     bPrint   = bPrint || (nmins > 0);
 238     if (bPrint)
 239     {
 240         if (bPrinted)
 241         {
 242             pr_two(out, 'h', nmins);
 243         }
 244         else
 245         {
 246             fprintf(out, "%d", nmins);
 247         }
 248     }
 249     bPrinted = bPrinted || bPrint;
 250     if (bPrinted)
 251     {
 252         pr_two(out, ':', nsecs);
 253     }
 254     else
 255     {
 256         fprintf(out, "%ds", nsecs);
 257     }
 258     fprintf(out, "\n");
 259 }
 260
 261 void clear_nrnb(t_nrnb* nrnb)
 262 {
 263     int i;
 264
 265     for (i = 0; (i < eNRNB); i++)
 266     {
 267         nrnb->n[i] = 0.0;
 268     }
 269 }
 270
 271 void add_nrnb(t_nrnb* dest, t_nrnb* s1, t_nrnb* s2)
 272 {
 273     int i;
 274
 275     for (i = 0; (i < eNRNB); i++)
 276     {
 277         dest->n[i] = s1->n[i] + s2->n[i];
 278     }
 279 }
 280
 281 void print_nrnb(FILE* out, t_nrnb* nrnb)
 282 {
 283     int i;
 284
 285     for (i = 0; (i < eNRNB); i++)
 286     {
 287         if (nrnb->n[i] > 0)
 288         {
 289             fprintf(out, " %-26s %10.0f.\n", nbdata[i].name, nrnb->n[i]);
 290         }
 291     }
 292 }
 293
 294 void _inc_nrnb(t_nrnb* nrnb, int enr, int inc, char gmx_unused* file, int gmx_unused line)
 295 {
 296     nrnb->n[enr] += inc;
 297 #ifdef DEBUG_NRNB
 298     printf("nrnb %15s(%2d) incremented with %8d from file %s line %d\n", nbdata[enr].name, enr, inc,
 299            file, line);
 300 #endif
 301 }
 302
 303 /* Returns in enr is the index of a full nbnxn VdW kernel */
 304 static gmx_bool nrnb_is_nbnxn_vdw_kernel(int enr)
 305 {
 306     return (enr >= eNR_NBNXN_LJ_RF && enr <= eNR_NBNXN_LJ_E);
 307 }
 308
 309 /* Returns in enr is the index of an nbnxn kernel addition (LJ modification) */
 310 static gmx_bool nrnb_is_nbnxn_kernel_addition(int enr)
 311 {
 312     return (enr >= eNR_NBNXN_ADD_LJ_FSW && enr <= eNR_NBNXN_ADD_LJ_EWALD_E);
 313 }
 314
 315 void print_flop(FILE* out, t_nrnb* nrnb, double* nbfs, double* mflop)
 316 {
 317     int         i, j;
 318     double      mni, frac, tfrac, tflop;
 319     const char* myline =
 320             "-----------------------------------------------------------------------------";
 321
 322     *nbfs = 0.0;
 323     for (i = 0; (i < eNR_NBKERNEL_TOTAL_NR); i++)
 324     {
 325         if (std::strstr(nbdata[i].name, "W3-W3") != nullptr)
 326         {
 327             *nbfs += 9e-6 * nrnb->n[i];
 328         }
 329         else if (std::strstr(nbdata[i].name, "W3") != nullptr)
 330         {
 331             *nbfs += 3e-6 * nrnb->n[i];
 332         }
 333         else if (std::strstr(nbdata[i].name, "W4-W4") != nullptr)
 334         {
 335             *nbfs += 10e-6 * nrnb->n[i];
 336         }
 337         else if (std::strstr(nbdata[i].name, "W4") != nullptr)
 338         {
 339             *nbfs += 4e-6 * nrnb->n[i];
 340         }
 341         else
 342         {
 343             *nbfs += 1e-6 * nrnb->n[i];
 344         }
 345     }
 346     tflop = 0;
 347     for (i = 0; (i < eNRNB); i++)
 348     {
 349         tflop += 1e-6 * nrnb->n[i] * nbdata[i].flop;
 350     }
 351
 352     if (tflop == 0)
 353     {
 354         fprintf(out, "No MEGA Flopsen this time\n");
 355         return;
 356     }
 357     if (out)
 358     {
 359         fprintf(out, "\n\tM E G A - F L O P S   A C C O U N T I N G\n\n");
 360     }
 361
 362     if (out)
 363     {
 364         fprintf(out, " NB=Group-cutoff nonbonded kernels    NxN=N-by-N cluster Verlet kernels\n");
 365         fprintf(out, " RF=Reaction-Field  VdW=Van der Waals  QSTab=quadratic-spline table\n");
 366         fprintf(out, " W3=SPC/TIP3p  W4=TIP4p (single or pairs)\n");
 367         fprintf(out, " V&F=Potential and force  V=Potential only  F=Force only\n\n");
 368
 369         fprintf(out, " %-32s %16s %15s  %7s\n", "Computing:", "M-Number", "M-Flops", "% Flops");
 370         fprintf(out, "%s\n", myline);
 371     }
 372     *mflop = 0.0;
 373     tfrac  = 0.0;
 374     for (i = 0; (i < eNRNB); i++)
 375     {
 376         mni = 1e-6 * nrnb->n[i];
 377         /* Skip empty entries and nbnxn additional flops,
 378          * which have been added to the kernel entry.
 379          */
 380         if (mni > 0 && !nrnb_is_nbnxn_kernel_addition(i))
 381         {
 382             int flop;
 383
 384             flop = nbdata[i].flop;
 385             if (nrnb_is_nbnxn_vdw_kernel(i))
 386             {
 387                 /* Possibly add the cost of an LJ switch/Ewald function */
 388                 for (j = eNR_NBNXN_ADD_LJ_FSW; j <= eNR_NBNXN_ADD_LJ_EWALD; j += 2)
 389                 {
 390                     int e_kernel_add;
 391
 392                     /* Select the force or energy flop count */
 393                     e_kernel_add = j + ((i - eNR_NBNXN_LJ_RF) % 2);
 394
 395                     if (nrnb->n[e_kernel_add] > 0)
 396                     {
 397                         flop += nbdata[e_kernel_add].flop;
 398                     }
 399                 }
 400             }
 401             *mflop += mni * flop;
 402             frac = 100.0 * mni * flop / tflop;
 403             tfrac += frac;
 404             if (out != nullptr)
 405             {
 406                 fprintf(out, " %-32s %16.6f %15.3f  %6.1f\n", nbdata[i].name, mni, mni * flop, frac);
 407             }
 408         }
 409     }
 410     if (out)
 411     {
 412         fprintf(out, "%s\n", myline);
 413         fprintf(out, " %-32s %16s %15.3f  %6.1f\n", "Total", "", *mflop, tfrac);
 414         fprintf(out, "%s\n\n", myline);
 415
 416         if (nrnb->n[eNR_NBKERNEL_GENERIC] > 0)
 417         {
 418             fprintf(out,
 419                     "WARNING: Using the slow generic C kernel. This is fine if you are\n"
 420                     "comparing different implementations or MD software. Routine\n"
 421                     "simulations should use a different non-bonded setup for much better\n"
 422                     "performance.\n\n");
 423         }
 424     }
 425 }
 426
 427 void print_perf(FILE*   out,
 428                 double  time_per_thread,
 429                 double  time_per_node,
 430                 int64_t nsteps,
 431                 double  delta_t,
 432                 double  nbfs,
 433                 double  mflop)
 434 {
 435     double wallclocktime;
 436
 437     fprintf(out, "\n");
 438
 439     if (time_per_node > 0)
 440     {
 441         fprintf(out, "%12s %12s %12s %10s\n", "", "Core t (s)", "Wall t (s)", "(%)");
 442         fprintf(out, "%12s %12.3f %12.3f %10.1f\n", "Time:", time_per_thread, time_per_node,
 443                 100.0 * time_per_thread / time_per_node);
 444         /* only print day-hour-sec format if time_per_node is more than 30 min */
 445         if (time_per_node > 30 * 60)
 446         {
 447             fprintf(out, "%12s %12s", "", "");
 448             pr_difftime(out, time_per_node);
 449         }
 450         if (delta_t > 0)
 451         {
 452             mflop         = mflop / time_per_node;
 453             wallclocktime = nsteps * delta_t;
 454
 455             if (getenv("GMX_DETAILED_PERF_STATS") == nullptr)
 456             {
 457                 fprintf(out, "%12s %12s %12s\n", "", "(ns/day)", "(hour/ns)");
 458                 fprintf(out, "%12s %12.3f %12.3f\n", "Performance:", wallclocktime * 24 * 3.6 / time_per_node,
 459                         1000 * time_per_node / (3600 * wallclocktime));
 460             }
 461             else
 462             {
 463                 fprintf(out, "%12s %12s %12s %12s %12s\n", "", "(Mnbf/s)",
 464                         (mflop > 1000) ? "(GFlops)" : "(MFlops)", "(ns/day)", "(hour/ns)");
 465                 fprintf(out, "%12s %12.3f %12.3f %12.3f %12.3f\n", "Performance:", nbfs / time_per_node,
 466                         (mflop > 1000) ? (mflop / 1000) : mflop, wallclocktime * 24 * 3.6 / time_per_node,
 467                         1000 * time_per_node / (3600 * wallclocktime));
 468             }
 469         }
 470         else
 471         {
 472             if (getenv("GMX_DETAILED_PERF_STATS") == nullptr)
 473             {
 474                 fprintf(out, "%12s %14s\n", "", "(steps/hour)");
 475                 fprintf(out, "%12s %14.1f\n", "Performance:", nsteps * 3600.0 / time_per_node);
 476             }
 477             else
 478             {
 479                 fprintf(out, "%12s %12s %12s %14s\n", "", "(Mnbf/s)",
 480                         (mflop > 1000) ? "(GFlops)" : "(MFlops)", "(steps/hour)");
 481                 fprintf(out, "%12s %12.3f %12.3f %14.1f\n", "Performance:", nbfs / time_per_node,
 482                         (mflop > 1000) ? (mflop / 1000) : mflop, nsteps * 3600.0 / time_per_node);
 483             }
 484         }
 485     }
 486 }
 487
 488 int cost_nrnb(int enr)
 489 {
 490     return nbdata[enr].flop;
 491 }
 492
 493 const char* nrnb_str(int enr)
 494 {
 495     return nbdata[enr].name;
 496 }
 497
 498 static const int force_index[] = {
 499     eNR_BONDS,  eNR_ANGLES, eNR_PROPER, eNR_IMPROPER, eNR_RB,
 500     eNR_DISRES, eNR_ORIRES, eNR_POSRES, eNR_FBPOSRES, eNR_NS,
 501 };
 502 #define NFORCE_INDEX asize(force_index)
 503
 504 static const int constr_index[] = { eNR_SHAKE,  eNR_SHAKE_RIJ,  eNR_SETTLE,  eNR_UPDATE,
 505                                     eNR_PCOUPL, eNR_CONSTR_VIR, eNR_CONSTR_V };
 506 #define NCONSTR_INDEX asize(constr_index)
 507
 508 static double pr_av(FILE* log, t_commrec* cr, double fav, const double ftot[], const char* title)
 509 {
 510     int    i, perc;
 511     double dperc, unb;
 512
 513     unb = 0;
 514     if (fav > 0)
 515     {
 516         fav /= cr->nnodes - cr->npmenodes;
 517         fprintf(log, "\n %-26s", title);
 518         for (i = 0; (i < cr->nnodes); i++)
 519         {
 520             dperc = (100.0 * ftot[i]) / fav;
 521             unb   = std::max(unb, dperc);
 522             perc  = static_cast<int>(dperc);
 523             fprintf(log, "%3d ", perc);
 524         }
 525         if (unb > 0)
 526         {
 527             perc = static_cast<int>(10000.0 / unb);
 528             fprintf(log, "%6d%%\n\n", perc);
 529         }
 530         else
 531         {
 532             fprintf(log, "\n\n");
 533         }
 534     }
 535     return unb;
 536 }
 537
 538 void pr_load(FILE* log, t_commrec* cr, t_nrnb nrnb[])
 539 {
 540     t_nrnb av;
 541
 542     std::vector<double> ftot(cr->nnodes);
 543     std::vector<double> stot(cr->nnodes);
 544     for (int i = 0; (i < cr->nnodes); i++)
 545     {
 546         add_nrnb(&av, &av, &(nrnb[i]));
 547         /* Cost due to forces */
 548         for (int j = 0; (j < eNR_NBKERNEL_TOTAL_NR); j++)
 549         {
 550             ftot[i] += nrnb[i].n[j] * cost_nrnb(j);
 551         }
 552         for (int j = 0; (j < NFORCE_INDEX); j++)
 553         {
 554             ftot[i] += nrnb[i].n[force_index[j]] * cost_nrnb(force_index[j]);
 555         }
 556         /* Due to shake */
 557         for (int j = 0; (j < NCONSTR_INDEX); j++)
 558         {
 559             stot[i] += nrnb[i].n[constr_index[j]] * cost_nrnb(constr_index[j]);
 560         }
 561     }
 562     for (int j = 0; (j < eNRNB); j++)
 563     {
 564         av.n[j] = av.n[j] / static_cast<double>(cr->nnodes - cr->npmenodes);
 565     }
 566
 567     fprintf(log, "\nDetailed load balancing info in percentage of average\n");
 568
 569     fprintf(log, " Type                 RANK:");
 570     for (int i = 0; (i < cr->nnodes); i++)
 571     {
 572         fprintf(log, "%3d ", i);
 573     }
 574     fprintf(log, "Scaling\n");
 575     fprintf(log, "---------------------------");
 576     for (int i = 0; (i < cr->nnodes); i++)
 577     {
 578         fprintf(log, "----");
 579     }
 580     fprintf(log, "-------\n");
 581
 582     for (int j = 0; (j < eNRNB); j++)
 583     {
 584         double unb = 100.0;
 585         if (av.n[j] > 0)
 586         {
 587             double dperc;
 588             int    perc;
 589             fprintf(log, " %-26s", nrnb_str(j));
 590             for (int i = 0; (i < cr->nnodes); i++)
 591             {
 592                 dperc = (100.0 * nrnb[i].n[j]) / av.n[j];
 593                 unb   = std::max(unb, dperc);
 594                 perc  = static_cast<int>(dperc);
 595                 fprintf(log, "%3d ", perc);
 596             }
 597             if (unb > 0)
 598             {
 599                 perc = static_cast<int>(10000.0 / unb);
 600                 fprintf(log, "%6d%%\n", perc);
 601             }
 602             else
 603             {
 604                 fprintf(log, "\n");
 605             }
 606         }
 607     }
 608     double fav = 0;
 609     double sav = 0;
 610     for (int i = 0; (i < cr->nnodes); i++)
 611     {
 612         fav += ftot[i];
 613         sav += stot[i];
 614     }
 615     double uf = pr_av(log, cr, fav, ftot.data(), "Total Force");
 616     double us = pr_av(log, cr, sav, stot.data(), "Total Constr.");
 617
 618     double unb = (uf * fav + us * sav) / (fav + sav);
 619     if (unb > 0)
 620     {
 621         unb = 10000.0 / unb;
 622         fprintf(log, "\nTotal Scaling: %.0f%% of max performance\n\n", unb);
 623     }
 624 }