src/gromacs/gmxlib/nonbonded/nb_kernel_c/nb_kernel_ElecEwSh_VdwBhamSh_GeomW4P1_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 "gmxpre.h"
  39
  40 #include "config.h"
  41
  42 #include <math.h>
  43
  44 #include "../nb_kernel.h"
  45 #include "gromacs/legacyheaders/types/simple.h"
  46 #include "gromacs/math/vec.h"
  47 #include "gromacs/legacyheaders/nrnb.h"
  48
  49 /*
  50  * Gromacs nonbonded kernel:   nb_kernel_ElecEwSh_VdwBhamSh_GeomW4P1_VF_c
  51  * Electrostatics interaction: Ewald
  52  * VdW interaction:            Buckingham
  53  * Geometry:                   Water4-Particle
  54  * Calculate force/pot:        PotentialAndForce
  55  */
  56 void
  57 nb_kernel_ElecEwSh_VdwBhamSh_GeomW4P1_VF_c
  58                     (t_nblist                    * gmx_restrict       nlist,
  59                      rvec                        * gmx_restrict          xx,
  60                      rvec                        * gmx_restrict          ff,
  61                      t_forcerec                  * gmx_restrict          fr,
  62                      t_mdatoms                   * gmx_restrict     mdatoms,
  63                      nb_kernel_data_t gmx_unused * gmx_restrict kernel_data,
  64                      t_nrnb                      * gmx_restrict        nrnb)
  65 {
  66     int              i_shift_offset,i_coord_offset,j_coord_offset;
  67     int              j_index_start,j_index_end;
  68     int              nri,inr,ggid,iidx,jidx,jnr,outeriter,inneriter;
  69     real             shX,shY,shZ,tx,ty,tz,fscal,rcutoff,rcutoff2;
  70     int              *iinr,*jindex,*jjnr,*shiftidx,*gid;
  71     real             *shiftvec,*fshift,*x,*f;
  72     int              vdwioffset0;
  73     real             ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
  74     int              vdwioffset1;
  75     real             ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
  76     int              vdwioffset2;
  77     real             ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
  78     int              vdwioffset3;
  79     real             ix3,iy3,iz3,fix3,fiy3,fiz3,iq3,isai3;
  80     int              vdwjidx0;
  81     real             jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
  82     real             dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00,cexp1_00,cexp2_00;
  83     real             dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10,cexp1_10,cexp2_10;
  84     real             dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20,cexp1_20,cexp2_20;
  85     real             dx30,dy30,dz30,rsq30,rinv30,rinvsq30,r30,qq30,c6_30,c12_30,cexp1_30,cexp2_30;
  86     real             velec,felec,velecsum,facel,crf,krf,krf2;
  87     real             *charge;
  88     int              nvdwtype;
  89     real             rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,br,vvdwexp,sh_vdw_invrcut6;
  90     int              *vdwtype;
  91     real             *vdwparam;
  92     int              ewitab;
  93     real             ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace;
  94     real             *ewtab;
  95
  96     x                = xx[0];
  97     f                = ff[0];
  98
  99     nri              = nlist->nri;
 100     iinr             = nlist->iinr;
 101     jindex           = nlist->jindex;
 102     jjnr             = nlist->jjnr;
 103     shiftidx         = nlist->shift;
 104     gid              = nlist->gid;
 105     shiftvec         = fr->shift_vec[0];
 106     fshift           = fr->fshift[0];
 107     facel            = fr->epsfac;
 108     charge           = mdatoms->chargeA;
 109     nvdwtype         = fr->ntype;
 110     vdwparam         = fr->nbfp;
 111     vdwtype          = mdatoms->typeA;
 112
 113     sh_ewald         = fr->ic->sh_ewald;
 114     ewtab            = fr->ic->tabq_coul_FDV0;
 115     ewtabscale       = fr->ic->tabq_scale;
 116     ewtabhalfspace   = 0.5/ewtabscale;
 117
 118     /* Setup water-specific parameters */
 119     inr              = nlist->iinr[0];
 120     iq1              = facel*charge[inr+1];
 121     iq2              = facel*charge[inr+2];
 122     iq3              = facel*charge[inr+3];
 123     vdwioffset0      = 3*nvdwtype*vdwtype[inr+0];
 124
 125     /* When we use explicit cutoffs the value must be identical for elec and VdW, so use elec as an arbitrary choice */
 126     rcutoff          = fr->rcoulomb;
 127     rcutoff2         = rcutoff*rcutoff;
 128
 129     sh_vdw_invrcut6  = fr->ic->sh_invrc6;
 130     rvdw             = fr->rvdw;
 131
 132     outeriter        = 0;
 133     inneriter        = 0;
 134
 135     /* Start outer loop over neighborlists */
 136     for(iidx=0; iidx<nri; iidx++)
 137     {
 138         /* Load shift vector for this list */
 139         i_shift_offset   = DIM*shiftidx[iidx];
 140         shX              = shiftvec[i_shift_offset+XX];
 141         shY              = shiftvec[i_shift_offset+YY];
 142         shZ              = shiftvec[i_shift_offset+ZZ];
 143
 144         /* Load limits for loop over neighbors */
 145         j_index_start    = jindex[iidx];
 146         j_index_end      = jindex[iidx+1];
 147
 148         /* Get outer coordinate index */
 149         inr              = iinr[iidx];
 150         i_coord_offset   = DIM*inr;
 151
 152         /* Load i particle coords and add shift vector */
 153         ix0              = shX + x[i_coord_offset+DIM*0+XX];
 154         iy0              = shY + x[i_coord_offset+DIM*0+YY];
 155         iz0              = shZ + x[i_coord_offset+DIM*0+ZZ];
 156         ix1              = shX + x[i_coord_offset+DIM*1+XX];
 157         iy1              = shY + x[i_coord_offset+DIM*1+YY];
 158         iz1              = shZ + x[i_coord_offset+DIM*1+ZZ];
 159         ix2              = shX + x[i_coord_offset+DIM*2+XX];
 160         iy2              = shY + x[i_coord_offset+DIM*2+YY];
 161         iz2              = shZ + x[i_coord_offset+DIM*2+ZZ];
 162         ix3              = shX + x[i_coord_offset+DIM*3+XX];
 163         iy3              = shY + x[i_coord_offset+DIM*3+YY];
 164         iz3              = shZ + x[i_coord_offset+DIM*3+ZZ];
 165
 166         fix0             = 0.0;
 167         fiy0             = 0.0;
 168         fiz0             = 0.0;
 169         fix1             = 0.0;
 170         fiy1             = 0.0;
 171         fiz1             = 0.0;
 172         fix2             = 0.0;
 173         fiy2             = 0.0;
 174         fiz2             = 0.0;
 175         fix3             = 0.0;
 176         fiy3             = 0.0;
 177         fiz3             = 0.0;
 178
 179         /* Reset potential sums */
 180         velecsum         = 0.0;
 181         vvdwsum          = 0.0;
 182
 183         /* Start inner kernel loop */
 184         for(jidx=j_index_start; jidx<j_index_end; jidx++)
 185         {
 186             /* Get j neighbor index, and coordinate index */
 187             jnr              = jjnr[jidx];
 188             j_coord_offset   = DIM*jnr;
 189
 190             /* load j atom coordinates */
 191             jx0              = x[j_coord_offset+DIM*0+XX];
 192             jy0              = x[j_coord_offset+DIM*0+YY];
 193             jz0              = x[j_coord_offset+DIM*0+ZZ];
 194
 195             /* Calculate displacement vector */
 196             dx00             = ix0 - jx0;
 197             dy00             = iy0 - jy0;
 198             dz00             = iz0 - jz0;
 199             dx10             = ix1 - jx0;
 200             dy10             = iy1 - jy0;
 201             dz10             = iz1 - jz0;
 202             dx20             = ix2 - jx0;
 203             dy20             = iy2 - jy0;
 204             dz20             = iz2 - jz0;
 205             dx30             = ix3 - jx0;
 206             dy30             = iy3 - jy0;
 207             dz30             = iz3 - jz0;
 208
 209             /* Calculate squared distance and things based on it */
 210             rsq00            = dx00*dx00+dy00*dy00+dz00*dz00;
 211             rsq10            = dx10*dx10+dy10*dy10+dz10*dz10;
 212             rsq20            = dx20*dx20+dy20*dy20+dz20*dz20;
 213             rsq30            = dx30*dx30+dy30*dy30+dz30*dz30;
 214
 215             rinv00           = gmx_invsqrt(rsq00);
 216             rinv10           = gmx_invsqrt(rsq10);
 217             rinv20           = gmx_invsqrt(rsq20);
 218             rinv30           = gmx_invsqrt(rsq30);
 219
 220             rinvsq00         = rinv00*rinv00;
 221             rinvsq10         = rinv10*rinv10;
 222             rinvsq20         = rinv20*rinv20;
 223             rinvsq30         = rinv30*rinv30;
 224
 225             /* Load parameters for j particles */
 226             jq0              = charge[jnr+0];
 227             vdwjidx0         = 3*vdwtype[jnr+0];
 228
 229             /**************************
 230              * CALCULATE INTERACTIONS *
 231              **************************/
 232
 233             if (rsq00<rcutoff2)
 234             {
 235
 236             r00              = rsq00*rinv00;
 237
 238             c6_00            = vdwparam[vdwioffset0+vdwjidx0];
 239             cexp1_00         = vdwparam[vdwioffset0+vdwjidx0+1];
 240             cexp2_00         = vdwparam[vdwioffset0+vdwjidx0+2];
 241
 242             /* BUCKINGHAM DISPERSION/REPULSION */
 243             rinvsix          = rinvsq00*rinvsq00*rinvsq00;
 244             vvdw6            = c6_00*rinvsix;
 245             br               = cexp2_00*r00;
 246             vvdwexp          = cexp1_00*exp(-br);
 247             vvdw             = (vvdwexp-cexp1_00*exp(-cexp2_00*rvdw)) - (vvdw6 - c6_00*sh_vdw_invrcut6)*(1.0/6.0);
 248             fvdw             = (br*vvdwexp-vvdw6)*rinvsq00;
 249
 250             /* Update potential sums from outer loop */
 251             vvdwsum         += vvdw;
 252
 253             fscal            = fvdw;
 254
 255             /* Calculate temporary vectorial force */
 256             tx               = fscal*dx00;
 257             ty               = fscal*dy00;
 258             tz               = fscal*dz00;
 259
 260             /* Update vectorial force */
 261             fix0            += tx;
 262             fiy0            += ty;
 263             fiz0            += tz;
 264             f[j_coord_offset+DIM*0+XX] -= tx;
 265             f[j_coord_offset+DIM*0+YY] -= ty;
 266             f[j_coord_offset+DIM*0+ZZ] -= tz;
 267
 268             }
 269
 270             /**************************
 271              * CALCULATE INTERACTIONS *
 272              **************************/
 273
 274             if (rsq10<rcutoff2)
 275             {
 276
 277             r10              = rsq10*rinv10;
 278
 279             qq10             = iq1*jq0;
 280
 281             /* EWALD ELECTROSTATICS */
 282
 283             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 284             ewrt             = r10*ewtabscale;
 285             ewitab           = ewrt;
 286             eweps            = ewrt-ewitab;
 287             ewitab           = 4*ewitab;
 288             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 289             velec            = qq10*((rinv10-sh_ewald)-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 290             felec            = qq10*rinv10*(rinvsq10-felec);
 291
 292             /* Update potential sums from outer loop */
 293             velecsum        += velec;
 294
 295             fscal            = felec;
 296
 297             /* Calculate temporary vectorial force */
 298             tx               = fscal*dx10;
 299             ty               = fscal*dy10;
 300             tz               = fscal*dz10;
 301
 302             /* Update vectorial force */
 303             fix1            += tx;
 304             fiy1            += ty;
 305             fiz1            += tz;
 306             f[j_coord_offset+DIM*0+XX] -= tx;
 307             f[j_coord_offset+DIM*0+YY] -= ty;
 308             f[j_coord_offset+DIM*0+ZZ] -= tz;
 309
 310             }
 311
 312             /**************************
 313              * CALCULATE INTERACTIONS *
 314              **************************/
 315
 316             if (rsq20<rcutoff2)
 317             {
 318
 319             r20              = rsq20*rinv20;
 320
 321             qq20             = iq2*jq0;
 322
 323             /* EWALD ELECTROSTATICS */
 324
 325             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 326             ewrt             = r20*ewtabscale;
 327             ewitab           = ewrt;
 328             eweps            = ewrt-ewitab;
 329             ewitab           = 4*ewitab;
 330             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 331             velec            = qq20*((rinv20-sh_ewald)-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 332             felec            = qq20*rinv20*(rinvsq20-felec);
 333
 334             /* Update potential sums from outer loop */
 335             velecsum        += velec;
 336
 337             fscal            = felec;
 338
 339             /* Calculate temporary vectorial force */
 340             tx               = fscal*dx20;
 341             ty               = fscal*dy20;
 342             tz               = fscal*dz20;
 343
 344             /* Update vectorial force */
 345             fix2            += tx;
 346             fiy2            += ty;
 347             fiz2            += tz;
 348             f[j_coord_offset+DIM*0+XX] -= tx;
 349             f[j_coord_offset+DIM*0+YY] -= ty;
 350             f[j_coord_offset+DIM*0+ZZ] -= tz;
 351
 352             }
 353
 354             /**************************
 355              * CALCULATE INTERACTIONS *
 356              **************************/
 357
 358             if (rsq30<rcutoff2)
 359             {
 360
 361             r30              = rsq30*rinv30;
 362
 363             qq30             = iq3*jq0;
 364
 365             /* EWALD ELECTROSTATICS */
 366
 367             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 368             ewrt             = r30*ewtabscale;
 369             ewitab           = ewrt;
 370             eweps            = ewrt-ewitab;
 371             ewitab           = 4*ewitab;
 372             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 373             velec            = qq30*((rinv30-sh_ewald)-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 374             felec            = qq30*rinv30*(rinvsq30-felec);
 375
 376             /* Update potential sums from outer loop */
 377             velecsum        += velec;
 378
 379             fscal            = felec;
 380
 381             /* Calculate temporary vectorial force */
 382             tx               = fscal*dx30;
 383             ty               = fscal*dy30;
 384             tz               = fscal*dz30;
 385
 386             /* Update vectorial force */
 387             fix3            += tx;
 388             fiy3            += ty;
 389             fiz3            += tz;
 390             f[j_coord_offset+DIM*0+XX] -= tx;
 391             f[j_coord_offset+DIM*0+YY] -= ty;
 392             f[j_coord_offset+DIM*0+ZZ] -= tz;
 393
 394             }
 395
 396             /* Inner loop uses 218 flops */
 397         }
 398         /* End of innermost loop */
 399
 400         tx = ty = tz = 0;
 401         f[i_coord_offset+DIM*0+XX] += fix0;
 402         f[i_coord_offset+DIM*0+YY] += fiy0;
 403         f[i_coord_offset+DIM*0+ZZ] += fiz0;
 404         tx                         += fix0;
 405         ty                         += fiy0;
 406         tz                         += fiz0;
 407         f[i_coord_offset+DIM*1+XX] += fix1;
 408         f[i_coord_offset+DIM*1+YY] += fiy1;
 409         f[i_coord_offset+DIM*1+ZZ] += fiz1;
 410         tx                         += fix1;
 411         ty                         += fiy1;
 412         tz                         += fiz1;
 413         f[i_coord_offset+DIM*2+XX] += fix2;
 414         f[i_coord_offset+DIM*2+YY] += fiy2;
 415         f[i_coord_offset+DIM*2+ZZ] += fiz2;
 416         tx                         += fix2;
 417         ty                         += fiy2;
 418         tz                         += fiz2;
 419         f[i_coord_offset+DIM*3+XX] += fix3;
 420         f[i_coord_offset+DIM*3+YY] += fiy3;
 421         f[i_coord_offset+DIM*3+ZZ] += fiz3;
 422         tx                         += fix3;
 423         ty                         += fiy3;
 424         tz                         += fiz3;
 425         fshift[i_shift_offset+XX]  += tx;
 426         fshift[i_shift_offset+YY]  += ty;
 427         fshift[i_shift_offset+ZZ]  += tz;
 428
 429         ggid                        = gid[iidx];
 430         /* Update potential energies */
 431         kernel_data->energygrp_elec[ggid] += velecsum;
 432         kernel_data->energygrp_vdw[ggid] += vvdwsum;
 433
 434         /* Increment number of inner iterations */
 435         inneriter                  += j_index_end - j_index_start;
 436
 437         /* Outer loop uses 41 flops */
 438     }
 439
 440     /* Increment number of outer iterations */
 441     outeriter        += nri;
 442
 443     /* Update outer/inner flops */
 444
 445     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W4_VF,outeriter*41 + inneriter*218);
 446 }
 447 /*
 448  * Gromacs nonbonded kernel:   nb_kernel_ElecEwSh_VdwBhamSh_GeomW4P1_F_c
 449  * Electrostatics interaction: Ewald
 450  * VdW interaction:            Buckingham
 451  * Geometry:                   Water4-Particle
 452  * Calculate force/pot:        Force
 453  */
 454 void
 455 nb_kernel_ElecEwSh_VdwBhamSh_GeomW4P1_F_c
 456                     (t_nblist                    * gmx_restrict       nlist,
 457                      rvec                        * gmx_restrict          xx,
 458                      rvec                        * gmx_restrict          ff,
 459                      t_forcerec                  * gmx_restrict          fr,
 460                      t_mdatoms                   * gmx_restrict     mdatoms,
 461                      nb_kernel_data_t gmx_unused * gmx_restrict kernel_data,
 462                      t_nrnb                      * gmx_restrict        nrnb)
 463 {
 464     int              i_shift_offset,i_coord_offset,j_coord_offset;
 465     int              j_index_start,j_index_end;
 466     int              nri,inr,ggid,iidx,jidx,jnr,outeriter,inneriter;
 467     real             shX,shY,shZ,tx,ty,tz,fscal,rcutoff,rcutoff2;
 468     int              *iinr,*jindex,*jjnr,*shiftidx,*gid;
 469     real             *shiftvec,*fshift,*x,*f;
 470     int              vdwioffset0;
 471     real             ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
 472     int              vdwioffset1;
 473     real             ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
 474     int              vdwioffset2;
 475     real             ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
 476     int              vdwioffset3;
 477     real             ix3,iy3,iz3,fix3,fiy3,fiz3,iq3,isai3;
 478     int              vdwjidx0;
 479     real             jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
 480     real             dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00,cexp1_00,cexp2_00;
 481     real             dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10,cexp1_10,cexp2_10;
 482     real             dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20,cexp1_20,cexp2_20;
 483     real             dx30,dy30,dz30,rsq30,rinv30,rinvsq30,r30,qq30,c6_30,c12_30,cexp1_30,cexp2_30;
 484     real             velec,felec,velecsum,facel,crf,krf,krf2;
 485     real             *charge;
 486     int              nvdwtype;
 487     real             rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,br,vvdwexp,sh_vdw_invrcut6;
 488     int              *vdwtype;
 489     real             *vdwparam;
 490     int              ewitab;
 491     real             ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace;
 492     real             *ewtab;
 493
 494     x                = xx[0];
 495     f                = ff[0];
 496
 497     nri              = nlist->nri;
 498     iinr             = nlist->iinr;
 499     jindex           = nlist->jindex;
 500     jjnr             = nlist->jjnr;
 501     shiftidx         = nlist->shift;
 502     gid              = nlist->gid;
 503     shiftvec         = fr->shift_vec[0];
 504     fshift           = fr->fshift[0];
 505     facel            = fr->epsfac;
 506     charge           = mdatoms->chargeA;
 507     nvdwtype         = fr->ntype;
 508     vdwparam         = fr->nbfp;
 509     vdwtype          = mdatoms->typeA;
 510
 511     sh_ewald         = fr->ic->sh_ewald;
 512     ewtab            = fr->ic->tabq_coul_F;
 513     ewtabscale       = fr->ic->tabq_scale;
 514     ewtabhalfspace   = 0.5/ewtabscale;
 515
 516     /* Setup water-specific parameters */
 517     inr              = nlist->iinr[0];
 518     iq1              = facel*charge[inr+1];
 519     iq2              = facel*charge[inr+2];
 520     iq3              = facel*charge[inr+3];
 521     vdwioffset0      = 3*nvdwtype*vdwtype[inr+0];
 522
 523     /* When we use explicit cutoffs the value must be identical for elec and VdW, so use elec as an arbitrary choice */
 524     rcutoff          = fr->rcoulomb;
 525     rcutoff2         = rcutoff*rcutoff;
 526
 527     sh_vdw_invrcut6  = fr->ic->sh_invrc6;
 528     rvdw             = fr->rvdw;
 529
 530     outeriter        = 0;
 531     inneriter        = 0;
 532
 533     /* Start outer loop over neighborlists */
 534     for(iidx=0; iidx<nri; iidx++)
 535     {
 536         /* Load shift vector for this list */
 537         i_shift_offset   = DIM*shiftidx[iidx];
 538         shX              = shiftvec[i_shift_offset+XX];
 539         shY              = shiftvec[i_shift_offset+YY];
 540         shZ              = shiftvec[i_shift_offset+ZZ];
 541
 542         /* Load limits for loop over neighbors */
 543         j_index_start    = jindex[iidx];
 544         j_index_end      = jindex[iidx+1];
 545
 546         /* Get outer coordinate index */
 547         inr              = iinr[iidx];
 548         i_coord_offset   = DIM*inr;
 549
 550         /* Load i particle coords and add shift vector */
 551         ix0              = shX + x[i_coord_offset+DIM*0+XX];
 552         iy0              = shY + x[i_coord_offset+DIM*0+YY];
 553         iz0              = shZ + x[i_coord_offset+DIM*0+ZZ];
 554         ix1              = shX + x[i_coord_offset+DIM*1+XX];
 555         iy1              = shY + x[i_coord_offset+DIM*1+YY];
 556         iz1              = shZ + x[i_coord_offset+DIM*1+ZZ];
 557         ix2              = shX + x[i_coord_offset+DIM*2+XX];
 558         iy2              = shY + x[i_coord_offset+DIM*2+YY];
 559         iz2              = shZ + x[i_coord_offset+DIM*2+ZZ];
 560         ix3              = shX + x[i_coord_offset+DIM*3+XX];
 561         iy3              = shY + x[i_coord_offset+DIM*3+YY];
 562         iz3              = shZ + x[i_coord_offset+DIM*3+ZZ];
 563
 564         fix0             = 0.0;
 565         fiy0             = 0.0;
 566         fiz0             = 0.0;
 567         fix1             = 0.0;
 568         fiy1             = 0.0;
 569         fiz1             = 0.0;
 570         fix2             = 0.0;
 571         fiy2             = 0.0;
 572         fiz2             = 0.0;
 573         fix3             = 0.0;
 574         fiy3             = 0.0;
 575         fiz3             = 0.0;
 576
 577         /* Start inner kernel loop */
 578         for(jidx=j_index_start; jidx<j_index_end; jidx++)
 579         {
 580             /* Get j neighbor index, and coordinate index */
 581             jnr              = jjnr[jidx];
 582             j_coord_offset   = DIM*jnr;
 583
 584             /* load j atom coordinates */
 585             jx0              = x[j_coord_offset+DIM*0+XX];
 586             jy0              = x[j_coord_offset+DIM*0+YY];
 587             jz0              = x[j_coord_offset+DIM*0+ZZ];
 588
 589             /* Calculate displacement vector */
 590             dx00             = ix0 - jx0;
 591             dy00             = iy0 - jy0;
 592             dz00             = iz0 - jz0;
 593             dx10             = ix1 - jx0;
 594             dy10             = iy1 - jy0;
 595             dz10             = iz1 - jz0;
 596             dx20             = ix2 - jx0;
 597             dy20             = iy2 - jy0;
 598             dz20             = iz2 - jz0;
 599             dx30             = ix3 - jx0;
 600             dy30             = iy3 - jy0;
 601             dz30             = iz3 - jz0;
 602
 603             /* Calculate squared distance and things based on it */
 604             rsq00            = dx00*dx00+dy00*dy00+dz00*dz00;
 605             rsq10            = dx10*dx10+dy10*dy10+dz10*dz10;
 606             rsq20            = dx20*dx20+dy20*dy20+dz20*dz20;
 607             rsq30            = dx30*dx30+dy30*dy30+dz30*dz30;
 608
 609             rinv00           = gmx_invsqrt(rsq00);
 610             rinv10           = gmx_invsqrt(rsq10);
 611             rinv20           = gmx_invsqrt(rsq20);
 612             rinv30           = gmx_invsqrt(rsq30);
 613
 614             rinvsq00         = rinv00*rinv00;
 615             rinvsq10         = rinv10*rinv10;
 616             rinvsq20         = rinv20*rinv20;
 617             rinvsq30         = rinv30*rinv30;
 618
 619             /* Load parameters for j particles */
 620             jq0              = charge[jnr+0];
 621             vdwjidx0         = 3*vdwtype[jnr+0];
 622
 623             /**************************
 624              * CALCULATE INTERACTIONS *
 625              **************************/
 626
 627             if (rsq00<rcutoff2)
 628             {
 629
 630             r00              = rsq00*rinv00;
 631
 632             c6_00            = vdwparam[vdwioffset0+vdwjidx0];
 633             cexp1_00         = vdwparam[vdwioffset0+vdwjidx0+1];
 634             cexp2_00         = vdwparam[vdwioffset0+vdwjidx0+2];
 635
 636             /* BUCKINGHAM DISPERSION/REPULSION */
 637             rinvsix          = rinvsq00*rinvsq00*rinvsq00;
 638             vvdw6            = c6_00*rinvsix;
 639             br               = cexp2_00*r00;
 640             vvdwexp          = cexp1_00*exp(-br);
 641             fvdw             = (br*vvdwexp-vvdw6)*rinvsq00;
 642
 643             fscal            = fvdw;
 644
 645             /* Calculate temporary vectorial force */
 646             tx               = fscal*dx00;
 647             ty               = fscal*dy00;
 648             tz               = fscal*dz00;
 649
 650             /* Update vectorial force */
 651             fix0            += tx;
 652             fiy0            += ty;
 653             fiz0            += tz;
 654             f[j_coord_offset+DIM*0+XX] -= tx;
 655             f[j_coord_offset+DIM*0+YY] -= ty;
 656             f[j_coord_offset+DIM*0+ZZ] -= tz;
 657
 658             }
 659
 660             /**************************
 661              * CALCULATE INTERACTIONS *
 662              **************************/
 663
 664             if (rsq10<rcutoff2)
 665             {
 666
 667             r10              = rsq10*rinv10;
 668
 669             qq10             = iq1*jq0;
 670
 671             /* EWALD ELECTROSTATICS */
 672
 673             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 674             ewrt             = r10*ewtabscale;
 675             ewitab           = ewrt;
 676             eweps            = ewrt-ewitab;
 677             felec            = (1.0-eweps)*ewtab[ewitab]+eweps*ewtab[ewitab+1];
 678             felec            = qq10*rinv10*(rinvsq10-felec);
 679
 680             fscal            = felec;
 681
 682             /* Calculate temporary vectorial force */
 683             tx               = fscal*dx10;
 684             ty               = fscal*dy10;
 685             tz               = fscal*dz10;
 686
 687             /* Update vectorial force */
 688             fix1            += tx;
 689             fiy1            += ty;
 690             fiz1            += tz;
 691             f[j_coord_offset+DIM*0+XX] -= tx;
 692             f[j_coord_offset+DIM*0+YY] -= ty;
 693             f[j_coord_offset+DIM*0+ZZ] -= tz;
 694
 695             }
 696
 697             /**************************
 698              * CALCULATE INTERACTIONS *
 699              **************************/
 700
 701             if (rsq20<rcutoff2)
 702             {
 703
 704             r20              = rsq20*rinv20;
 705
 706             qq20             = iq2*jq0;
 707
 708             /* EWALD ELECTROSTATICS */
 709
 710             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 711             ewrt             = r20*ewtabscale;
 712             ewitab           = ewrt;
 713             eweps            = ewrt-ewitab;
 714             felec            = (1.0-eweps)*ewtab[ewitab]+eweps*ewtab[ewitab+1];
 715             felec            = qq20*rinv20*(rinvsq20-felec);
 716
 717             fscal            = felec;
 718
 719             /* Calculate temporary vectorial force */
 720             tx               = fscal*dx20;
 721             ty               = fscal*dy20;
 722             tz               = fscal*dz20;
 723
 724             /* Update vectorial force */
 725             fix2            += tx;
 726             fiy2            += ty;
 727             fiz2            += tz;
 728             f[j_coord_offset+DIM*0+XX] -= tx;
 729             f[j_coord_offset+DIM*0+YY] -= ty;
 730             f[j_coord_offset+DIM*0+ZZ] -= tz;
 731
 732             }
 733
 734             /**************************
 735              * CALCULATE INTERACTIONS *
 736              **************************/
 737
 738             if (rsq30<rcutoff2)
 739             {
 740
 741             r30              = rsq30*rinv30;
 742
 743             qq30             = iq3*jq0;
 744
 745             /* EWALD ELECTROSTATICS */
 746
 747             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 748             ewrt             = r30*ewtabscale;
 749             ewitab           = ewrt;
 750             eweps            = ewrt-ewitab;
 751             felec            = (1.0-eweps)*ewtab[ewitab]+eweps*ewtab[ewitab+1];
 752             felec            = qq30*rinv30*(rinvsq30-felec);
 753
 754             fscal            = felec;
 755
 756             /* Calculate temporary vectorial force */
 757             tx               = fscal*dx30;
 758             ty               = fscal*dy30;
 759             tz               = fscal*dz30;
 760
 761             /* Update vectorial force */
 762             fix3            += tx;
 763             fiy3            += ty;
 764             fiz3            += tz;
 765             f[j_coord_offset+DIM*0+XX] -= tx;
 766             f[j_coord_offset+DIM*0+YY] -= ty;
 767             f[j_coord_offset+DIM*0+ZZ] -= tz;
 768
 769             }
 770
 771             /* Inner loop uses 160 flops */
 772         }
 773         /* End of innermost loop */
 774
 775         tx = ty = tz = 0;
 776         f[i_coord_offset+DIM*0+XX] += fix0;
 777         f[i_coord_offset+DIM*0+YY] += fiy0;
 778         f[i_coord_offset+DIM*0+ZZ] += fiz0;
 779         tx                         += fix0;
 780         ty                         += fiy0;
 781         tz                         += fiz0;
 782         f[i_coord_offset+DIM*1+XX] += fix1;
 783         f[i_coord_offset+DIM*1+YY] += fiy1;
 784         f[i_coord_offset+DIM*1+ZZ] += fiz1;
 785         tx                         += fix1;
 786         ty                         += fiy1;
 787         tz                         += fiz1;
 788         f[i_coord_offset+DIM*2+XX] += fix2;
 789         f[i_coord_offset+DIM*2+YY] += fiy2;
 790         f[i_coord_offset+DIM*2+ZZ] += fiz2;
 791         tx                         += fix2;
 792         ty                         += fiy2;
 793         tz                         += fiz2;
 794         f[i_coord_offset+DIM*3+XX] += fix3;
 795         f[i_coord_offset+DIM*3+YY] += fiy3;
 796         f[i_coord_offset+DIM*3+ZZ] += fiz3;
 797         tx                         += fix3;
 798         ty                         += fiy3;
 799         tz                         += fiz3;
 800         fshift[i_shift_offset+XX]  += tx;
 801         fshift[i_shift_offset+YY]  += ty;
 802         fshift[i_shift_offset+ZZ]  += tz;
 803
 804         /* Increment number of inner iterations */
 805         inneriter                  += j_index_end - j_index_start;
 806
 807         /* Outer loop uses 39 flops */
 808     }
 809
 810     /* Increment number of outer iterations */
 811     outeriter        += nri;
 812
 813     /* Update outer/inner flops */
 814
 815     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W4_F,outeriter*39 + inneriter*160);
 816 }