src/gromacs/gmxlib/nonbonded/nb_kernel_c/nb_kernel_ElecEwSh_VdwBhamSh_GeomW3P1_c.c

   1 /*
   2  * This file is part of the GROMACS molecular simulation package.
   3  *
   4  * Copyright (c) 2012,2013,2014, by the GROMACS development team, led by
   5  * Mark Abraham, David van der Spoel, Berk Hess, and Erik Lindahl,
   6  * and including many others, as listed in the AUTHORS file in the
   7  * top-level source directory and at http://www.gromacs.org.
   8  *
   9  * GROMACS is free software; you can redistribute it and/or
  10  * modify it under the terms of the GNU Lesser General Public License
  11  * as published by the Free Software Foundation; either version 2.1
  12  * of the License, or (at your option) any later version.
  13  *
  14  * GROMACS is distributed in the hope that it will be useful,
  15  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  16  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  17  * Lesser General Public License for more details.
  18  *
  19  * You should have received a copy of the GNU Lesser General Public
  20  * License along with GROMACS; if not, see
  21  * http://www.gnu.org/licenses, or write to the Free Software Foundation,
  22  * Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02110-1301  USA.
  23  *
  24  * If you want to redistribute modifications to GROMACS, please
  25  * consider that scientific software is very special. Version
  26  * control is crucial - bugs must be traceable. We will be happy to
  27  * consider code for inclusion in the official distribution, but
  28  * derived work must not be called official GROMACS. Details are found
  29  * in the README & COPYING files - if they are missing, get the
  30  * official version at http://www.gromacs.org.
  31  *
  32  * To help us fund GROMACS development, we humbly ask that you cite
  33  * the research papers on the package. Check out http://www.gromacs.org.
  34  */
  35 /*
  36  * Note: this file was generated by the GROMACS c kernel generator.
  37  */
  38 #include "config.h"
  39
  40 #include <math.h>
  41
  42 #include "../nb_kernel.h"
  43 #include "types/simple.h"
  44 #include "gromacs/math/vec.h"
  45 #include "nrnb.h"
  46
  47 /*
  48  * Gromacs nonbonded kernel:   nb_kernel_ElecEwSh_VdwBhamSh_GeomW3P1_VF_c
  49  * Electrostatics interaction: Ewald
  50  * VdW interaction:            Buckingham
  51  * Geometry:                   Water3-Particle
  52  * Calculate force/pot:        PotentialAndForce
  53  */
  54 void
  55 nb_kernel_ElecEwSh_VdwBhamSh_GeomW3P1_VF_c
  56                     (t_nblist                    * gmx_restrict       nlist,
  57                      rvec                        * gmx_restrict          xx,
  58                      rvec                        * gmx_restrict          ff,
  59                      t_forcerec                  * gmx_restrict          fr,
  60                      t_mdatoms                   * gmx_restrict     mdatoms,
  61                      nb_kernel_data_t gmx_unused * gmx_restrict kernel_data,
  62                      t_nrnb                      * gmx_restrict        nrnb)
  63 {
  64     int              i_shift_offset,i_coord_offset,j_coord_offset;
  65     int              j_index_start,j_index_end;
  66     int              nri,inr,ggid,iidx,jidx,jnr,outeriter,inneriter;
  67     real             shX,shY,shZ,tx,ty,tz,fscal,rcutoff,rcutoff2;
  68     int              *iinr,*jindex,*jjnr,*shiftidx,*gid;
  69     real             *shiftvec,*fshift,*x,*f;
  70     int              vdwioffset0;
  71     real             ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
  72     int              vdwioffset1;
  73     real             ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
  74     int              vdwioffset2;
  75     real             ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
  76     int              vdwjidx0;
  77     real             jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
  78     real             dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00,cexp1_00,cexp2_00;
  79     real             dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10,cexp1_10,cexp2_10;
  80     real             dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20,cexp1_20,cexp2_20;
  81     real             velec,felec,velecsum,facel,crf,krf,krf2;
  82     real             *charge;
  83     int              nvdwtype;
  84     real             rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,br,vvdwexp,sh_vdw_invrcut6;
  85     int              *vdwtype;
  86     real             *vdwparam;
  87     int              ewitab;
  88     real             ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace;
  89     real             *ewtab;
  90
  91     x                = xx[0];
  92     f                = ff[0];
  93
  94     nri              = nlist->nri;
  95     iinr             = nlist->iinr;
  96     jindex           = nlist->jindex;
  97     jjnr             = nlist->jjnr;
  98     shiftidx         = nlist->shift;
  99     gid              = nlist->gid;
 100     shiftvec         = fr->shift_vec[0];
 101     fshift           = fr->fshift[0];
 102     facel            = fr->epsfac;
 103     charge           = mdatoms->chargeA;
 104     nvdwtype         = fr->ntype;
 105     vdwparam         = fr->nbfp;
 106     vdwtype          = mdatoms->typeA;
 107
 108     sh_ewald         = fr->ic->sh_ewald;
 109     ewtab            = fr->ic->tabq_coul_FDV0;
 110     ewtabscale       = fr->ic->tabq_scale;
 111     ewtabhalfspace   = 0.5/ewtabscale;
 112
 113     /* Setup water-specific parameters */
 114     inr              = nlist->iinr[0];
 115     iq0              = facel*charge[inr+0];
 116     iq1              = facel*charge[inr+1];
 117     iq2              = facel*charge[inr+2];
 118     vdwioffset0      = 3*nvdwtype*vdwtype[inr+0];
 119
 120     /* When we use explicit cutoffs the value must be identical for elec and VdW, so use elec as an arbitrary choice */
 121     rcutoff          = fr->rcoulomb;
 122     rcutoff2         = rcutoff*rcutoff;
 123
 124     sh_vdw_invrcut6  = fr->ic->sh_invrc6;
 125     rvdw             = fr->rvdw;
 126
 127     outeriter        = 0;
 128     inneriter        = 0;
 129
 130     /* Start outer loop over neighborlists */
 131     for(iidx=0; iidx<nri; iidx++)
 132     {
 133         /* Load shift vector for this list */
 134         i_shift_offset   = DIM*shiftidx[iidx];
 135         shX              = shiftvec[i_shift_offset+XX];
 136         shY              = shiftvec[i_shift_offset+YY];
 137         shZ              = shiftvec[i_shift_offset+ZZ];
 138
 139         /* Load limits for loop over neighbors */
 140         j_index_start    = jindex[iidx];
 141         j_index_end      = jindex[iidx+1];
 142
 143         /* Get outer coordinate index */
 144         inr              = iinr[iidx];
 145         i_coord_offset   = DIM*inr;
 146
 147         /* Load i particle coords and add shift vector */
 148         ix0              = shX + x[i_coord_offset+DIM*0+XX];
 149         iy0              = shY + x[i_coord_offset+DIM*0+YY];
 150         iz0              = shZ + x[i_coord_offset+DIM*0+ZZ];
 151         ix1              = shX + x[i_coord_offset+DIM*1+XX];
 152         iy1              = shY + x[i_coord_offset+DIM*1+YY];
 153         iz1              = shZ + x[i_coord_offset+DIM*1+ZZ];
 154         ix2              = shX + x[i_coord_offset+DIM*2+XX];
 155         iy2              = shY + x[i_coord_offset+DIM*2+YY];
 156         iz2              = shZ + x[i_coord_offset+DIM*2+ZZ];
 157
 158         fix0             = 0.0;
 159         fiy0             = 0.0;
 160         fiz0             = 0.0;
 161         fix1             = 0.0;
 162         fiy1             = 0.0;
 163         fiz1             = 0.0;
 164         fix2             = 0.0;
 165         fiy2             = 0.0;
 166         fiz2             = 0.0;
 167
 168         /* Reset potential sums */
 169         velecsum         = 0.0;
 170         vvdwsum          = 0.0;
 171
 172         /* Start inner kernel loop */
 173         for(jidx=j_index_start; jidx<j_index_end; jidx++)
 174         {
 175             /* Get j neighbor index, and coordinate index */
 176             jnr              = jjnr[jidx];
 177             j_coord_offset   = DIM*jnr;
 178
 179             /* load j atom coordinates */
 180             jx0              = x[j_coord_offset+DIM*0+XX];
 181             jy0              = x[j_coord_offset+DIM*0+YY];
 182             jz0              = x[j_coord_offset+DIM*0+ZZ];
 183
 184             /* Calculate displacement vector */
 185             dx00             = ix0 - jx0;
 186             dy00             = iy0 - jy0;
 187             dz00             = iz0 - jz0;
 188             dx10             = ix1 - jx0;
 189             dy10             = iy1 - jy0;
 190             dz10             = iz1 - jz0;
 191             dx20             = ix2 - jx0;
 192             dy20             = iy2 - jy0;
 193             dz20             = iz2 - jz0;
 194
 195             /* Calculate squared distance and things based on it */
 196             rsq00            = dx00*dx00+dy00*dy00+dz00*dz00;
 197             rsq10            = dx10*dx10+dy10*dy10+dz10*dz10;
 198             rsq20            = dx20*dx20+dy20*dy20+dz20*dz20;
 199
 200             rinv00           = gmx_invsqrt(rsq00);
 201             rinv10           = gmx_invsqrt(rsq10);
 202             rinv20           = gmx_invsqrt(rsq20);
 203
 204             rinvsq00         = rinv00*rinv00;
 205             rinvsq10         = rinv10*rinv10;
 206             rinvsq20         = rinv20*rinv20;
 207
 208             /* Load parameters for j particles */
 209             jq0              = charge[jnr+0];
 210             vdwjidx0         = 3*vdwtype[jnr+0];
 211
 212             /**************************
 213              * CALCULATE INTERACTIONS *
 214              **************************/
 215
 216             if (rsq00<rcutoff2)
 217             {
 218
 219             r00              = rsq00*rinv00;
 220
 221             qq00             = iq0*jq0;
 222             c6_00            = vdwparam[vdwioffset0+vdwjidx0];
 223             cexp1_00         = vdwparam[vdwioffset0+vdwjidx0+1];
 224             cexp2_00         = vdwparam[vdwioffset0+vdwjidx0+2];
 225
 226             /* EWALD ELECTROSTATICS */
 227
 228             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 229             ewrt             = r00*ewtabscale;
 230             ewitab           = ewrt;
 231             eweps            = ewrt-ewitab;
 232             ewitab           = 4*ewitab;
 233             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 234             velec            = qq00*((rinv00-sh_ewald)-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 235             felec            = qq00*rinv00*(rinvsq00-felec);
 236
 237             /* BUCKINGHAM DISPERSION/REPULSION */
 238             rinvsix          = rinvsq00*rinvsq00*rinvsq00;
 239             vvdw6            = c6_00*rinvsix;
 240             br               = cexp2_00*r00;
 241             vvdwexp          = cexp1_00*exp(-br);
 242             vvdw             = (vvdwexp-cexp1_00*exp(-cexp2_00*rvdw)) - (vvdw6 - c6_00*sh_vdw_invrcut6)*(1.0/6.0);
 243             fvdw             = (br*vvdwexp-vvdw6)*rinvsq00;
 244
 245             /* Update potential sums from outer loop */
 246             velecsum        += velec;
 247             vvdwsum         += vvdw;
 248
 249             fscal            = felec+fvdw;
 250
 251             /* Calculate temporary vectorial force */
 252             tx               = fscal*dx00;
 253             ty               = fscal*dy00;
 254             tz               = fscal*dz00;
 255
 256             /* Update vectorial force */
 257             fix0            += tx;
 258             fiy0            += ty;
 259             fiz0            += tz;
 260             f[j_coord_offset+DIM*0+XX] -= tx;
 261             f[j_coord_offset+DIM*0+YY] -= ty;
 262             f[j_coord_offset+DIM*0+ZZ] -= tz;
 263
 264             }
 265
 266             /**************************
 267              * CALCULATE INTERACTIONS *
 268              **************************/
 269
 270             if (rsq10<rcutoff2)
 271             {
 272
 273             r10              = rsq10*rinv10;
 274
 275             qq10             = iq1*jq0;
 276
 277             /* EWALD ELECTROSTATICS */
 278
 279             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 280             ewrt             = r10*ewtabscale;
 281             ewitab           = ewrt;
 282             eweps            = ewrt-ewitab;
 283             ewitab           = 4*ewitab;
 284             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 285             velec            = qq10*((rinv10-sh_ewald)-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 286             felec            = qq10*rinv10*(rinvsq10-felec);
 287
 288             /* Update potential sums from outer loop */
 289             velecsum        += velec;
 290
 291             fscal            = felec;
 292
 293             /* Calculate temporary vectorial force */
 294             tx               = fscal*dx10;
 295             ty               = fscal*dy10;
 296             tz               = fscal*dz10;
 297
 298             /* Update vectorial force */
 299             fix1            += tx;
 300             fiy1            += ty;
 301             fiz1            += tz;
 302             f[j_coord_offset+DIM*0+XX] -= tx;
 303             f[j_coord_offset+DIM*0+YY] -= ty;
 304             f[j_coord_offset+DIM*0+ZZ] -= tz;
 305
 306             }
 307
 308             /**************************
 309              * CALCULATE INTERACTIONS *
 310              **************************/
 311
 312             if (rsq20<rcutoff2)
 313             {
 314
 315             r20              = rsq20*rinv20;
 316
 317             qq20             = iq2*jq0;
 318
 319             /* EWALD ELECTROSTATICS */
 320
 321             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 322             ewrt             = r20*ewtabscale;
 323             ewitab           = ewrt;
 324             eweps            = ewrt-ewitab;
 325             ewitab           = 4*ewitab;
 326             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 327             velec            = qq20*((rinv20-sh_ewald)-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 328             felec            = qq20*rinv20*(rinvsq20-felec);
 329
 330             /* Update potential sums from outer loop */
 331             velecsum        += velec;
 332
 333             fscal            = felec;
 334
 335             /* Calculate temporary vectorial force */
 336             tx               = fscal*dx20;
 337             ty               = fscal*dy20;
 338             tz               = fscal*dz20;
 339
 340             /* Update vectorial force */
 341             fix2            += tx;
 342             fiy2            += ty;
 343             fiz2            += tz;
 344             f[j_coord_offset+DIM*0+XX] -= tx;
 345             f[j_coord_offset+DIM*0+YY] -= ty;
 346             f[j_coord_offset+DIM*0+ZZ] -= tz;
 347
 348             }
 349
 350             /* Inner loop uses 195 flops */
 351         }
 352         /* End of innermost loop */
 353
 354         tx = ty = tz = 0;
 355         f[i_coord_offset+DIM*0+XX] += fix0;
 356         f[i_coord_offset+DIM*0+YY] += fiy0;
 357         f[i_coord_offset+DIM*0+ZZ] += fiz0;
 358         tx                         += fix0;
 359         ty                         += fiy0;
 360         tz                         += fiz0;
 361         f[i_coord_offset+DIM*1+XX] += fix1;
 362         f[i_coord_offset+DIM*1+YY] += fiy1;
 363         f[i_coord_offset+DIM*1+ZZ] += fiz1;
 364         tx                         += fix1;
 365         ty                         += fiy1;
 366         tz                         += fiz1;
 367         f[i_coord_offset+DIM*2+XX] += fix2;
 368         f[i_coord_offset+DIM*2+YY] += fiy2;
 369         f[i_coord_offset+DIM*2+ZZ] += fiz2;
 370         tx                         += fix2;
 371         ty                         += fiy2;
 372         tz                         += fiz2;
 373         fshift[i_shift_offset+XX]  += tx;
 374         fshift[i_shift_offset+YY]  += ty;
 375         fshift[i_shift_offset+ZZ]  += tz;
 376
 377         ggid                        = gid[iidx];
 378         /* Update potential energies */
 379         kernel_data->energygrp_elec[ggid] += velecsum;
 380         kernel_data->energygrp_vdw[ggid] += vvdwsum;
 381
 382         /* Increment number of inner iterations */
 383         inneriter                  += j_index_end - j_index_start;
 384
 385         /* Outer loop uses 32 flops */
 386     }
 387
 388     /* Increment number of outer iterations */
 389     outeriter        += nri;
 390
 391     /* Update outer/inner flops */
 392
 393     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W3_VF,outeriter*32 + inneriter*195);
 394 }
 395 /*
 396  * Gromacs nonbonded kernel:   nb_kernel_ElecEwSh_VdwBhamSh_GeomW3P1_F_c
 397  * Electrostatics interaction: Ewald
 398  * VdW interaction:            Buckingham
 399  * Geometry:                   Water3-Particle
 400  * Calculate force/pot:        Force
 401  */
 402 void
 403 nb_kernel_ElecEwSh_VdwBhamSh_GeomW3P1_F_c
 404                     (t_nblist                    * gmx_restrict       nlist,
 405                      rvec                        * gmx_restrict          xx,
 406                      rvec                        * gmx_restrict          ff,
 407                      t_forcerec                  * gmx_restrict          fr,
 408                      t_mdatoms                   * gmx_restrict     mdatoms,
 409                      nb_kernel_data_t gmx_unused * gmx_restrict kernel_data,
 410                      t_nrnb                      * gmx_restrict        nrnb)
 411 {
 412     int              i_shift_offset,i_coord_offset,j_coord_offset;
 413     int              j_index_start,j_index_end;
 414     int              nri,inr,ggid,iidx,jidx,jnr,outeriter,inneriter;
 415     real             shX,shY,shZ,tx,ty,tz,fscal,rcutoff,rcutoff2;
 416     int              *iinr,*jindex,*jjnr,*shiftidx,*gid;
 417     real             *shiftvec,*fshift,*x,*f;
 418     int              vdwioffset0;
 419     real             ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
 420     int              vdwioffset1;
 421     real             ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
 422     int              vdwioffset2;
 423     real             ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
 424     int              vdwjidx0;
 425     real             jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
 426     real             dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00,cexp1_00,cexp2_00;
 427     real             dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10,cexp1_10,cexp2_10;
 428     real             dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20,cexp1_20,cexp2_20;
 429     real             velec,felec,velecsum,facel,crf,krf,krf2;
 430     real             *charge;
 431     int              nvdwtype;
 432     real             rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,br,vvdwexp,sh_vdw_invrcut6;
 433     int              *vdwtype;
 434     real             *vdwparam;
 435     int              ewitab;
 436     real             ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace;
 437     real             *ewtab;
 438
 439     x                = xx[0];
 440     f                = ff[0];
 441
 442     nri              = nlist->nri;
 443     iinr             = nlist->iinr;
 444     jindex           = nlist->jindex;
 445     jjnr             = nlist->jjnr;
 446     shiftidx         = nlist->shift;
 447     gid              = nlist->gid;
 448     shiftvec         = fr->shift_vec[0];
 449     fshift           = fr->fshift[0];
 450     facel            = fr->epsfac;
 451     charge           = mdatoms->chargeA;
 452     nvdwtype         = fr->ntype;
 453     vdwparam         = fr->nbfp;
 454     vdwtype          = mdatoms->typeA;
 455
 456     sh_ewald         = fr->ic->sh_ewald;
 457     ewtab            = fr->ic->tabq_coul_F;
 458     ewtabscale       = fr->ic->tabq_scale;
 459     ewtabhalfspace   = 0.5/ewtabscale;
 460
 461     /* Setup water-specific parameters */
 462     inr              = nlist->iinr[0];
 463     iq0              = facel*charge[inr+0];
 464     iq1              = facel*charge[inr+1];
 465     iq2              = facel*charge[inr+2];
 466     vdwioffset0      = 3*nvdwtype*vdwtype[inr+0];
 467
 468     /* When we use explicit cutoffs the value must be identical for elec and VdW, so use elec as an arbitrary choice */
 469     rcutoff          = fr->rcoulomb;
 470     rcutoff2         = rcutoff*rcutoff;
 471
 472     sh_vdw_invrcut6  = fr->ic->sh_invrc6;
 473     rvdw             = fr->rvdw;
 474
 475     outeriter        = 0;
 476     inneriter        = 0;
 477
 478     /* Start outer loop over neighborlists */
 479     for(iidx=0; iidx<nri; iidx++)
 480     {
 481         /* Load shift vector for this list */
 482         i_shift_offset   = DIM*shiftidx[iidx];
 483         shX              = shiftvec[i_shift_offset+XX];
 484         shY              = shiftvec[i_shift_offset+YY];
 485         shZ              = shiftvec[i_shift_offset+ZZ];
 486
 487         /* Load limits for loop over neighbors */
 488         j_index_start    = jindex[iidx];
 489         j_index_end      = jindex[iidx+1];
 490
 491         /* Get outer coordinate index */
 492         inr              = iinr[iidx];
 493         i_coord_offset   = DIM*inr;
 494
 495         /* Load i particle coords and add shift vector */
 496         ix0              = shX + x[i_coord_offset+DIM*0+XX];
 497         iy0              = shY + x[i_coord_offset+DIM*0+YY];
 498         iz0              = shZ + x[i_coord_offset+DIM*0+ZZ];
 499         ix1              = shX + x[i_coord_offset+DIM*1+XX];
 500         iy1              = shY + x[i_coord_offset+DIM*1+YY];
 501         iz1              = shZ + x[i_coord_offset+DIM*1+ZZ];
 502         ix2              = shX + x[i_coord_offset+DIM*2+XX];
 503         iy2              = shY + x[i_coord_offset+DIM*2+YY];
 504         iz2              = shZ + x[i_coord_offset+DIM*2+ZZ];
 505
 506         fix0             = 0.0;
 507         fiy0             = 0.0;
 508         fiz0             = 0.0;
 509         fix1             = 0.0;
 510         fiy1             = 0.0;
 511         fiz1             = 0.0;
 512         fix2             = 0.0;
 513         fiy2             = 0.0;
 514         fiz2             = 0.0;
 515
 516         /* Start inner kernel loop */
 517         for(jidx=j_index_start; jidx<j_index_end; jidx++)
 518         {
 519             /* Get j neighbor index, and coordinate index */
 520             jnr              = jjnr[jidx];
 521             j_coord_offset   = DIM*jnr;
 522
 523             /* load j atom coordinates */
 524             jx0              = x[j_coord_offset+DIM*0+XX];
 525             jy0              = x[j_coord_offset+DIM*0+YY];
 526             jz0              = x[j_coord_offset+DIM*0+ZZ];
 527
 528             /* Calculate displacement vector */
 529             dx00             = ix0 - jx0;
 530             dy00             = iy0 - jy0;
 531             dz00             = iz0 - jz0;
 532             dx10             = ix1 - jx0;
 533             dy10             = iy1 - jy0;
 534             dz10             = iz1 - jz0;
 535             dx20             = ix2 - jx0;
 536             dy20             = iy2 - jy0;
 537             dz20             = iz2 - jz0;
 538
 539             /* Calculate squared distance and things based on it */
 540             rsq00            = dx00*dx00+dy00*dy00+dz00*dz00;
 541             rsq10            = dx10*dx10+dy10*dy10+dz10*dz10;
 542             rsq20            = dx20*dx20+dy20*dy20+dz20*dz20;
 543
 544             rinv00           = gmx_invsqrt(rsq00);
 545             rinv10           = gmx_invsqrt(rsq10);
 546             rinv20           = gmx_invsqrt(rsq20);
 547
 548             rinvsq00         = rinv00*rinv00;
 549             rinvsq10         = rinv10*rinv10;
 550             rinvsq20         = rinv20*rinv20;
 551
 552             /* Load parameters for j particles */
 553             jq0              = charge[jnr+0];
 554             vdwjidx0         = 3*vdwtype[jnr+0];
 555
 556             /**************************
 557              * CALCULATE INTERACTIONS *
 558              **************************/
 559
 560             if (rsq00<rcutoff2)
 561             {
 562
 563             r00              = rsq00*rinv00;
 564
 565             qq00             = iq0*jq0;
 566             c6_00            = vdwparam[vdwioffset0+vdwjidx0];
 567             cexp1_00         = vdwparam[vdwioffset0+vdwjidx0+1];
 568             cexp2_00         = vdwparam[vdwioffset0+vdwjidx0+2];
 569
 570             /* EWALD ELECTROSTATICS */
 571
 572             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 573             ewrt             = r00*ewtabscale;
 574             ewitab           = ewrt;
 575             eweps            = ewrt-ewitab;
 576             felec            = (1.0-eweps)*ewtab[ewitab]+eweps*ewtab[ewitab+1];
 577             felec            = qq00*rinv00*(rinvsq00-felec);
 578
 579             /* BUCKINGHAM DISPERSION/REPULSION */
 580             rinvsix          = rinvsq00*rinvsq00*rinvsq00;
 581             vvdw6            = c6_00*rinvsix;
 582             br               = cexp2_00*r00;
 583             vvdwexp          = cexp1_00*exp(-br);
 584             fvdw             = (br*vvdwexp-vvdw6)*rinvsq00;
 585
 586             fscal            = felec+fvdw;
 587
 588             /* Calculate temporary vectorial force */
 589             tx               = fscal*dx00;
 590             ty               = fscal*dy00;
 591             tz               = fscal*dz00;
 592
 593             /* Update vectorial force */
 594             fix0            += tx;
 595             fiy0            += ty;
 596             fiz0            += tz;
 597             f[j_coord_offset+DIM*0+XX] -= tx;
 598             f[j_coord_offset+DIM*0+YY] -= ty;
 599             f[j_coord_offset+DIM*0+ZZ] -= tz;
 600
 601             }
 602
 603             /**************************
 604              * CALCULATE INTERACTIONS *
 605              **************************/
 606
 607             if (rsq10<rcutoff2)
 608             {
 609
 610             r10              = rsq10*rinv10;
 611
 612             qq10             = iq1*jq0;
 613
 614             /* EWALD ELECTROSTATICS */
 615
 616             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 617             ewrt             = r10*ewtabscale;
 618             ewitab           = ewrt;
 619             eweps            = ewrt-ewitab;
 620             felec            = (1.0-eweps)*ewtab[ewitab]+eweps*ewtab[ewitab+1];
 621             felec            = qq10*rinv10*(rinvsq10-felec);
 622
 623             fscal            = felec;
 624
 625             /* Calculate temporary vectorial force */
 626             tx               = fscal*dx10;
 627             ty               = fscal*dy10;
 628             tz               = fscal*dz10;
 629
 630             /* Update vectorial force */
 631             fix1            += tx;
 632             fiy1            += ty;
 633             fiz1            += tz;
 634             f[j_coord_offset+DIM*0+XX] -= tx;
 635             f[j_coord_offset+DIM*0+YY] -= ty;
 636             f[j_coord_offset+DIM*0+ZZ] -= tz;
 637
 638             }
 639
 640             /**************************
 641              * CALCULATE INTERACTIONS *
 642              **************************/
 643
 644             if (rsq20<rcutoff2)
 645             {
 646
 647             r20              = rsq20*rinv20;
 648
 649             qq20             = iq2*jq0;
 650
 651             /* EWALD ELECTROSTATICS */
 652
 653             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 654             ewrt             = r20*ewtabscale;
 655             ewitab           = ewrt;
 656             eweps            = ewrt-ewitab;
 657             felec            = (1.0-eweps)*ewtab[ewitab]+eweps*ewtab[ewitab+1];
 658             felec            = qq20*rinv20*(rinvsq20-felec);
 659
 660             fscal            = felec;
 661
 662             /* Calculate temporary vectorial force */
 663             tx               = fscal*dx20;
 664             ty               = fscal*dy20;
 665             tz               = fscal*dz20;
 666
 667             /* Update vectorial force */
 668             fix2            += tx;
 669             fiy2            += ty;
 670             fiz2            += tz;
 671             f[j_coord_offset+DIM*0+XX] -= tx;
 672             f[j_coord_offset+DIM*0+YY] -= ty;
 673             f[j_coord_offset+DIM*0+ZZ] -= tz;
 674
 675             }
 676
 677             /* Inner loop uses 137 flops */
 678         }
 679         /* End of innermost loop */
 680
 681         tx = ty = tz = 0;
 682         f[i_coord_offset+DIM*0+XX] += fix0;
 683         f[i_coord_offset+DIM*0+YY] += fiy0;
 684         f[i_coord_offset+DIM*0+ZZ] += fiz0;
 685         tx                         += fix0;
 686         ty                         += fiy0;
 687         tz                         += fiz0;
 688         f[i_coord_offset+DIM*1+XX] += fix1;
 689         f[i_coord_offset+DIM*1+YY] += fiy1;
 690         f[i_coord_offset+DIM*1+ZZ] += fiz1;
 691         tx                         += fix1;
 692         ty                         += fiy1;
 693         tz                         += fiz1;
 694         f[i_coord_offset+DIM*2+XX] += fix2;
 695         f[i_coord_offset+DIM*2+YY] += fiy2;
 696         f[i_coord_offset+DIM*2+ZZ] += fiz2;
 697         tx                         += fix2;
 698         ty                         += fiy2;
 699         tz                         += fiz2;
 700         fshift[i_shift_offset+XX]  += tx;
 701         fshift[i_shift_offset+YY]  += ty;
 702         fshift[i_shift_offset+ZZ]  += tz;
 703
 704         /* Increment number of inner iterations */
 705         inneriter                  += j_index_end - j_index_start;
 706
 707         /* Outer loop uses 30 flops */
 708     }
 709
 710     /* Increment number of outer iterations */
 711     outeriter        += nri;
 712
 713     /* Update outer/inner flops */
 714
 715     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W3_F,outeriter*30 + inneriter*137);
 716 }