src/gromacs/gmxlib/nonbonded/nb_kernel_c/nb_kernel_ElecEwSh_VdwLJSh_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 "config.h"
  39
  40 #include <math.h>
  41
  42 #include "../nb_kernel.h"
  43 #include "gromacs/legacyheaders/types/simple.h"
  44 #include "gromacs/math/vec.h"
  45 #include "gromacs/legacyheaders/nrnb.h"
  46
  47 /*
  48  * Gromacs nonbonded kernel:   nb_kernel_ElecEwSh_VdwLJSh_GeomW4P1_VF_c
  49  * Electrostatics interaction: Ewald
  50  * VdW interaction:            LennardJones
  51  * Geometry:                   Water4-Particle
  52  * Calculate force/pot:        PotentialAndForce
  53  */
  54 void
  55 nb_kernel_ElecEwSh_VdwLJSh_GeomW4P1_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              vdwioffset3;
  77     real             ix3,iy3,iz3,fix3,fiy3,fiz3,iq3,isai3;
  78     int              vdwjidx0;
  79     real             jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
  80     real             dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00,cexp1_00,cexp2_00;
  81     real             dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10,cexp1_10,cexp2_10;
  82     real             dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20,cexp1_20,cexp2_20;
  83     real             dx30,dy30,dz30,rsq30,rinv30,rinvsq30,r30,qq30,c6_30,c12_30,cexp1_30,cexp2_30;
  84     real             velec,felec,velecsum,facel,crf,krf,krf2;
  85     real             *charge;
  86     int              nvdwtype;
  87     real             rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,br,vvdwexp,sh_vdw_invrcut6;
  88     int              *vdwtype;
  89     real             *vdwparam;
  90     int              ewitab;
  91     real             ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace;
  92     real             *ewtab;
  93
  94     x                = xx[0];
  95     f                = ff[0];
  96
  97     nri              = nlist->nri;
  98     iinr             = nlist->iinr;
  99     jindex           = nlist->jindex;
 100     jjnr             = nlist->jjnr;
 101     shiftidx         = nlist->shift;
 102     gid              = nlist->gid;
 103     shiftvec         = fr->shift_vec[0];
 104     fshift           = fr->fshift[0];
 105     facel            = fr->epsfac;
 106     charge           = mdatoms->chargeA;
 107     nvdwtype         = fr->ntype;
 108     vdwparam         = fr->nbfp;
 109     vdwtype          = mdatoms->typeA;
 110
 111     sh_ewald         = fr->ic->sh_ewald;
 112     ewtab            = fr->ic->tabq_coul_FDV0;
 113     ewtabscale       = fr->ic->tabq_scale;
 114     ewtabhalfspace   = 0.5/ewtabscale;
 115
 116     /* Setup water-specific parameters */
 117     inr              = nlist->iinr[0];
 118     iq1              = facel*charge[inr+1];
 119     iq2              = facel*charge[inr+2];
 120     iq3              = facel*charge[inr+3];
 121     vdwioffset0      = 2*nvdwtype*vdwtype[inr+0];
 122
 123     /* When we use explicit cutoffs the value must be identical for elec and VdW, so use elec as an arbitrary choice */
 124     rcutoff          = fr->rcoulomb;
 125     rcutoff2         = rcutoff*rcutoff;
 126
 127     sh_vdw_invrcut6  = fr->ic->sh_invrc6;
 128     rvdw             = fr->rvdw;
 129
 130     outeriter        = 0;
 131     inneriter        = 0;
 132
 133     /* Start outer loop over neighborlists */
 134     for(iidx=0; iidx<nri; iidx++)
 135     {
 136         /* Load shift vector for this list */
 137         i_shift_offset   = DIM*shiftidx[iidx];
 138         shX              = shiftvec[i_shift_offset+XX];
 139         shY              = shiftvec[i_shift_offset+YY];
 140         shZ              = shiftvec[i_shift_offset+ZZ];
 141
 142         /* Load limits for loop over neighbors */
 143         j_index_start    = jindex[iidx];
 144         j_index_end      = jindex[iidx+1];
 145
 146         /* Get outer coordinate index */
 147         inr              = iinr[iidx];
 148         i_coord_offset   = DIM*inr;
 149
 150         /* Load i particle coords and add shift vector */
 151         ix0              = shX + x[i_coord_offset+DIM*0+XX];
 152         iy0              = shY + x[i_coord_offset+DIM*0+YY];
 153         iz0              = shZ + x[i_coord_offset+DIM*0+ZZ];
 154         ix1              = shX + x[i_coord_offset+DIM*1+XX];
 155         iy1              = shY + x[i_coord_offset+DIM*1+YY];
 156         iz1              = shZ + x[i_coord_offset+DIM*1+ZZ];
 157         ix2              = shX + x[i_coord_offset+DIM*2+XX];
 158         iy2              = shY + x[i_coord_offset+DIM*2+YY];
 159         iz2              = shZ + x[i_coord_offset+DIM*2+ZZ];
 160         ix3              = shX + x[i_coord_offset+DIM*3+XX];
 161         iy3              = shY + x[i_coord_offset+DIM*3+YY];
 162         iz3              = shZ + x[i_coord_offset+DIM*3+ZZ];
 163
 164         fix0             = 0.0;
 165         fiy0             = 0.0;
 166         fiz0             = 0.0;
 167         fix1             = 0.0;
 168         fiy1             = 0.0;
 169         fiz1             = 0.0;
 170         fix2             = 0.0;
 171         fiy2             = 0.0;
 172         fiz2             = 0.0;
 173         fix3             = 0.0;
 174         fiy3             = 0.0;
 175         fiz3             = 0.0;
 176
 177         /* Reset potential sums */
 178         velecsum         = 0.0;
 179         vvdwsum          = 0.0;
 180
 181         /* Start inner kernel loop */
 182         for(jidx=j_index_start; jidx<j_index_end; jidx++)
 183         {
 184             /* Get j neighbor index, and coordinate index */
 185             jnr              = jjnr[jidx];
 186             j_coord_offset   = DIM*jnr;
 187
 188             /* load j atom coordinates */
 189             jx0              = x[j_coord_offset+DIM*0+XX];
 190             jy0              = x[j_coord_offset+DIM*0+YY];
 191             jz0              = x[j_coord_offset+DIM*0+ZZ];
 192
 193             /* Calculate displacement vector */
 194             dx00             = ix0 - jx0;
 195             dy00             = iy0 - jy0;
 196             dz00             = iz0 - jz0;
 197             dx10             = ix1 - jx0;
 198             dy10             = iy1 - jy0;
 199             dz10             = iz1 - jz0;
 200             dx20             = ix2 - jx0;
 201             dy20             = iy2 - jy0;
 202             dz20             = iz2 - jz0;
 203             dx30             = ix3 - jx0;
 204             dy30             = iy3 - jy0;
 205             dz30             = iz3 - jz0;
 206
 207             /* Calculate squared distance and things based on it */
 208             rsq00            = dx00*dx00+dy00*dy00+dz00*dz00;
 209             rsq10            = dx10*dx10+dy10*dy10+dz10*dz10;
 210             rsq20            = dx20*dx20+dy20*dy20+dz20*dz20;
 211             rsq30            = dx30*dx30+dy30*dy30+dz30*dz30;
 212
 213             rinv10           = gmx_invsqrt(rsq10);
 214             rinv20           = gmx_invsqrt(rsq20);
 215             rinv30           = gmx_invsqrt(rsq30);
 216
 217             rinvsq00         = 1.0/rsq00;
 218             rinvsq10         = rinv10*rinv10;
 219             rinvsq20         = rinv20*rinv20;
 220             rinvsq30         = rinv30*rinv30;
 221
 222             /* Load parameters for j particles */
 223             jq0              = charge[jnr+0];
 224             vdwjidx0         = 2*vdwtype[jnr+0];
 225
 226             /**************************
 227              * CALCULATE INTERACTIONS *
 228              **************************/
 229
 230             if (rsq00<rcutoff2)
 231             {
 232
 233             c6_00            = vdwparam[vdwioffset0+vdwjidx0];
 234             c12_00           = vdwparam[vdwioffset0+vdwjidx0+1];
 235
 236             /* LENNARD-JONES DISPERSION/REPULSION */
 237
 238             rinvsix          = rinvsq00*rinvsq00*rinvsq00;
 239             vvdw6            = c6_00*rinvsix;
 240             vvdw12           = c12_00*rinvsix*rinvsix;
 241             vvdw             = (vvdw12 - c12_00*sh_vdw_invrcut6*sh_vdw_invrcut6)*(1.0/12.0) - (vvdw6 - c6_00*sh_vdw_invrcut6)*(1.0/6.0);
 242             fvdw             = (vvdw12-vvdw6)*rinvsq00;
 243
 244             /* Update potential sums from outer loop */
 245             vvdwsum         += vvdw;
 246
 247             fscal            = fvdw;
 248
 249             /* Calculate temporary vectorial force */
 250             tx               = fscal*dx00;
 251             ty               = fscal*dy00;
 252             tz               = fscal*dz00;
 253
 254             /* Update vectorial force */
 255             fix0            += tx;
 256             fiy0            += ty;
 257             fiz0            += tz;
 258             f[j_coord_offset+DIM*0+XX] -= tx;
 259             f[j_coord_offset+DIM*0+YY] -= ty;
 260             f[j_coord_offset+DIM*0+ZZ] -= tz;
 261
 262             }
 263
 264             /**************************
 265              * CALCULATE INTERACTIONS *
 266              **************************/
 267
 268             if (rsq10<rcutoff2)
 269             {
 270
 271             r10              = rsq10*rinv10;
 272
 273             qq10             = iq1*jq0;
 274
 275             /* EWALD ELECTROSTATICS */
 276
 277             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 278             ewrt             = r10*ewtabscale;
 279             ewitab           = ewrt;
 280             eweps            = ewrt-ewitab;
 281             ewitab           = 4*ewitab;
 282             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 283             velec            = qq10*((rinv10-sh_ewald)-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 284             felec            = qq10*rinv10*(rinvsq10-felec);
 285
 286             /* Update potential sums from outer loop */
 287             velecsum        += velec;
 288
 289             fscal            = felec;
 290
 291             /* Calculate temporary vectorial force */
 292             tx               = fscal*dx10;
 293             ty               = fscal*dy10;
 294             tz               = fscal*dz10;
 295
 296             /* Update vectorial force */
 297             fix1            += tx;
 298             fiy1            += ty;
 299             fiz1            += tz;
 300             f[j_coord_offset+DIM*0+XX] -= tx;
 301             f[j_coord_offset+DIM*0+YY] -= ty;
 302             f[j_coord_offset+DIM*0+ZZ] -= tz;
 303
 304             }
 305
 306             /**************************
 307              * CALCULATE INTERACTIONS *
 308              **************************/
 309
 310             if (rsq20<rcutoff2)
 311             {
 312
 313             r20              = rsq20*rinv20;
 314
 315             qq20             = iq2*jq0;
 316
 317             /* EWALD ELECTROSTATICS */
 318
 319             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 320             ewrt             = r20*ewtabscale;
 321             ewitab           = ewrt;
 322             eweps            = ewrt-ewitab;
 323             ewitab           = 4*ewitab;
 324             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 325             velec            = qq20*((rinv20-sh_ewald)-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 326             felec            = qq20*rinv20*(rinvsq20-felec);
 327
 328             /* Update potential sums from outer loop */
 329             velecsum        += velec;
 330
 331             fscal            = felec;
 332
 333             /* Calculate temporary vectorial force */
 334             tx               = fscal*dx20;
 335             ty               = fscal*dy20;
 336             tz               = fscal*dz20;
 337
 338             /* Update vectorial force */
 339             fix2            += tx;
 340             fiy2            += ty;
 341             fiz2            += tz;
 342             f[j_coord_offset+DIM*0+XX] -= tx;
 343             f[j_coord_offset+DIM*0+YY] -= ty;
 344             f[j_coord_offset+DIM*0+ZZ] -= tz;
 345
 346             }
 347
 348             /**************************
 349              * CALCULATE INTERACTIONS *
 350              **************************/
 351
 352             if (rsq30<rcutoff2)
 353             {
 354
 355             r30              = rsq30*rinv30;
 356
 357             qq30             = iq3*jq0;
 358
 359             /* EWALD ELECTROSTATICS */
 360
 361             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 362             ewrt             = r30*ewtabscale;
 363             ewitab           = ewrt;
 364             eweps            = ewrt-ewitab;
 365             ewitab           = 4*ewitab;
 366             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 367             velec            = qq30*((rinv30-sh_ewald)-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 368             felec            = qq30*rinv30*(rinvsq30-felec);
 369
 370             /* Update potential sums from outer loop */
 371             velecsum        += velec;
 372
 373             fscal            = felec;
 374
 375             /* Calculate temporary vectorial force */
 376             tx               = fscal*dx30;
 377             ty               = fscal*dy30;
 378             tz               = fscal*dz30;
 379
 380             /* Update vectorial force */
 381             fix3            += tx;
 382             fiy3            += ty;
 383             fiz3            += tz;
 384             f[j_coord_offset+DIM*0+XX] -= tx;
 385             f[j_coord_offset+DIM*0+YY] -= ty;
 386             f[j_coord_offset+DIM*0+ZZ] -= tz;
 387
 388             }
 389
 390             /* Inner loop uses 163 flops */
 391         }
 392         /* End of innermost loop */
 393
 394         tx = ty = tz = 0;
 395         f[i_coord_offset+DIM*0+XX] += fix0;
 396         f[i_coord_offset+DIM*0+YY] += fiy0;
 397         f[i_coord_offset+DIM*0+ZZ] += fiz0;
 398         tx                         += fix0;
 399         ty                         += fiy0;
 400         tz                         += fiz0;
 401         f[i_coord_offset+DIM*1+XX] += fix1;
 402         f[i_coord_offset+DIM*1+YY] += fiy1;
 403         f[i_coord_offset+DIM*1+ZZ] += fiz1;
 404         tx                         += fix1;
 405         ty                         += fiy1;
 406         tz                         += fiz1;
 407         f[i_coord_offset+DIM*2+XX] += fix2;
 408         f[i_coord_offset+DIM*2+YY] += fiy2;
 409         f[i_coord_offset+DIM*2+ZZ] += fiz2;
 410         tx                         += fix2;
 411         ty                         += fiy2;
 412         tz                         += fiz2;
 413         f[i_coord_offset+DIM*3+XX] += fix3;
 414         f[i_coord_offset+DIM*3+YY] += fiy3;
 415         f[i_coord_offset+DIM*3+ZZ] += fiz3;
 416         tx                         += fix3;
 417         ty                         += fiy3;
 418         tz                         += fiz3;
 419         fshift[i_shift_offset+XX]  += tx;
 420         fshift[i_shift_offset+YY]  += ty;
 421         fshift[i_shift_offset+ZZ]  += tz;
 422
 423         ggid                        = gid[iidx];
 424         /* Update potential energies */
 425         kernel_data->energygrp_elec[ggid] += velecsum;
 426         kernel_data->energygrp_vdw[ggid] += vvdwsum;
 427
 428         /* Increment number of inner iterations */
 429         inneriter                  += j_index_end - j_index_start;
 430
 431         /* Outer loop uses 41 flops */
 432     }
 433
 434     /* Increment number of outer iterations */
 435     outeriter        += nri;
 436
 437     /* Update outer/inner flops */
 438
 439     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W4_VF,outeriter*41 + inneriter*163);
 440 }
 441 /*
 442  * Gromacs nonbonded kernel:   nb_kernel_ElecEwSh_VdwLJSh_GeomW4P1_F_c
 443  * Electrostatics interaction: Ewald
 444  * VdW interaction:            LennardJones
 445  * Geometry:                   Water4-Particle
 446  * Calculate force/pot:        Force
 447  */
 448 void
 449 nb_kernel_ElecEwSh_VdwLJSh_GeomW4P1_F_c
 450                     (t_nblist                    * gmx_restrict       nlist,
 451                      rvec                        * gmx_restrict          xx,
 452                      rvec                        * gmx_restrict          ff,
 453                      t_forcerec                  * gmx_restrict          fr,
 454                      t_mdatoms                   * gmx_restrict     mdatoms,
 455                      nb_kernel_data_t gmx_unused * gmx_restrict kernel_data,
 456                      t_nrnb                      * gmx_restrict        nrnb)
 457 {
 458     int              i_shift_offset,i_coord_offset,j_coord_offset;
 459     int              j_index_start,j_index_end;
 460     int              nri,inr,ggid,iidx,jidx,jnr,outeriter,inneriter;
 461     real             shX,shY,shZ,tx,ty,tz,fscal,rcutoff,rcutoff2;
 462     int              *iinr,*jindex,*jjnr,*shiftidx,*gid;
 463     real             *shiftvec,*fshift,*x,*f;
 464     int              vdwioffset0;
 465     real             ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
 466     int              vdwioffset1;
 467     real             ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
 468     int              vdwioffset2;
 469     real             ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
 470     int              vdwioffset3;
 471     real             ix3,iy3,iz3,fix3,fiy3,fiz3,iq3,isai3;
 472     int              vdwjidx0;
 473     real             jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
 474     real             dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00,cexp1_00,cexp2_00;
 475     real             dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10,cexp1_10,cexp2_10;
 476     real             dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20,cexp1_20,cexp2_20;
 477     real             dx30,dy30,dz30,rsq30,rinv30,rinvsq30,r30,qq30,c6_30,c12_30,cexp1_30,cexp2_30;
 478     real             velec,felec,velecsum,facel,crf,krf,krf2;
 479     real             *charge;
 480     int              nvdwtype;
 481     real             rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,br,vvdwexp,sh_vdw_invrcut6;
 482     int              *vdwtype;
 483     real             *vdwparam;
 484     int              ewitab;
 485     real             ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace;
 486     real             *ewtab;
 487
 488     x                = xx[0];
 489     f                = ff[0];
 490
 491     nri              = nlist->nri;
 492     iinr             = nlist->iinr;
 493     jindex           = nlist->jindex;
 494     jjnr             = nlist->jjnr;
 495     shiftidx         = nlist->shift;
 496     gid              = nlist->gid;
 497     shiftvec         = fr->shift_vec[0];
 498     fshift           = fr->fshift[0];
 499     facel            = fr->epsfac;
 500     charge           = mdatoms->chargeA;
 501     nvdwtype         = fr->ntype;
 502     vdwparam         = fr->nbfp;
 503     vdwtype          = mdatoms->typeA;
 504
 505     sh_ewald         = fr->ic->sh_ewald;
 506     ewtab            = fr->ic->tabq_coul_F;
 507     ewtabscale       = fr->ic->tabq_scale;
 508     ewtabhalfspace   = 0.5/ewtabscale;
 509
 510     /* Setup water-specific parameters */
 511     inr              = nlist->iinr[0];
 512     iq1              = facel*charge[inr+1];
 513     iq2              = facel*charge[inr+2];
 514     iq3              = facel*charge[inr+3];
 515     vdwioffset0      = 2*nvdwtype*vdwtype[inr+0];
 516
 517     /* When we use explicit cutoffs the value must be identical for elec and VdW, so use elec as an arbitrary choice */
 518     rcutoff          = fr->rcoulomb;
 519     rcutoff2         = rcutoff*rcutoff;
 520
 521     sh_vdw_invrcut6  = fr->ic->sh_invrc6;
 522     rvdw             = fr->rvdw;
 523
 524     outeriter        = 0;
 525     inneriter        = 0;
 526
 527     /* Start outer loop over neighborlists */
 528     for(iidx=0; iidx<nri; iidx++)
 529     {
 530         /* Load shift vector for this list */
 531         i_shift_offset   = DIM*shiftidx[iidx];
 532         shX              = shiftvec[i_shift_offset+XX];
 533         shY              = shiftvec[i_shift_offset+YY];
 534         shZ              = shiftvec[i_shift_offset+ZZ];
 535
 536         /* Load limits for loop over neighbors */
 537         j_index_start    = jindex[iidx];
 538         j_index_end      = jindex[iidx+1];
 539
 540         /* Get outer coordinate index */
 541         inr              = iinr[iidx];
 542         i_coord_offset   = DIM*inr;
 543
 544         /* Load i particle coords and add shift vector */
 545         ix0              = shX + x[i_coord_offset+DIM*0+XX];
 546         iy0              = shY + x[i_coord_offset+DIM*0+YY];
 547         iz0              = shZ + x[i_coord_offset+DIM*0+ZZ];
 548         ix1              = shX + x[i_coord_offset+DIM*1+XX];
 549         iy1              = shY + x[i_coord_offset+DIM*1+YY];
 550         iz1              = shZ + x[i_coord_offset+DIM*1+ZZ];
 551         ix2              = shX + x[i_coord_offset+DIM*2+XX];
 552         iy2              = shY + x[i_coord_offset+DIM*2+YY];
 553         iz2              = shZ + x[i_coord_offset+DIM*2+ZZ];
 554         ix3              = shX + x[i_coord_offset+DIM*3+XX];
 555         iy3              = shY + x[i_coord_offset+DIM*3+YY];
 556         iz3              = shZ + x[i_coord_offset+DIM*3+ZZ];
 557
 558         fix0             = 0.0;
 559         fiy0             = 0.0;
 560         fiz0             = 0.0;
 561         fix1             = 0.0;
 562         fiy1             = 0.0;
 563         fiz1             = 0.0;
 564         fix2             = 0.0;
 565         fiy2             = 0.0;
 566         fiz2             = 0.0;
 567         fix3             = 0.0;
 568         fiy3             = 0.0;
 569         fiz3             = 0.0;
 570
 571         /* Start inner kernel loop */
 572         for(jidx=j_index_start; jidx<j_index_end; jidx++)
 573         {
 574             /* Get j neighbor index, and coordinate index */
 575             jnr              = jjnr[jidx];
 576             j_coord_offset   = DIM*jnr;
 577
 578             /* load j atom coordinates */
 579             jx0              = x[j_coord_offset+DIM*0+XX];
 580             jy0              = x[j_coord_offset+DIM*0+YY];
 581             jz0              = x[j_coord_offset+DIM*0+ZZ];
 582
 583             /* Calculate displacement vector */
 584             dx00             = ix0 - jx0;
 585             dy00             = iy0 - jy0;
 586             dz00             = iz0 - jz0;
 587             dx10             = ix1 - jx0;
 588             dy10             = iy1 - jy0;
 589             dz10             = iz1 - jz0;
 590             dx20             = ix2 - jx0;
 591             dy20             = iy2 - jy0;
 592             dz20             = iz2 - jz0;
 593             dx30             = ix3 - jx0;
 594             dy30             = iy3 - jy0;
 595             dz30             = iz3 - jz0;
 596
 597             /* Calculate squared distance and things based on it */
 598             rsq00            = dx00*dx00+dy00*dy00+dz00*dz00;
 599             rsq10            = dx10*dx10+dy10*dy10+dz10*dz10;
 600             rsq20            = dx20*dx20+dy20*dy20+dz20*dz20;
 601             rsq30            = dx30*dx30+dy30*dy30+dz30*dz30;
 602
 603             rinv10           = gmx_invsqrt(rsq10);
 604             rinv20           = gmx_invsqrt(rsq20);
 605             rinv30           = gmx_invsqrt(rsq30);
 606
 607             rinvsq00         = 1.0/rsq00;
 608             rinvsq10         = rinv10*rinv10;
 609             rinvsq20         = rinv20*rinv20;
 610             rinvsq30         = rinv30*rinv30;
 611
 612             /* Load parameters for j particles */
 613             jq0              = charge[jnr+0];
 614             vdwjidx0         = 2*vdwtype[jnr+0];
 615
 616             /**************************
 617              * CALCULATE INTERACTIONS *
 618              **************************/
 619
 620             if (rsq00<rcutoff2)
 621             {
 622
 623             c6_00            = vdwparam[vdwioffset0+vdwjidx0];
 624             c12_00           = vdwparam[vdwioffset0+vdwjidx0+1];
 625
 626             /* LENNARD-JONES DISPERSION/REPULSION */
 627
 628             rinvsix          = rinvsq00*rinvsq00*rinvsq00;
 629             fvdw             = (c12_00*rinvsix-c6_00)*rinvsix*rinvsq00;
 630
 631             fscal            = fvdw;
 632
 633             /* Calculate temporary vectorial force */
 634             tx               = fscal*dx00;
 635             ty               = fscal*dy00;
 636             tz               = fscal*dz00;
 637
 638             /* Update vectorial force */
 639             fix0            += tx;
 640             fiy0            += ty;
 641             fiz0            += tz;
 642             f[j_coord_offset+DIM*0+XX] -= tx;
 643             f[j_coord_offset+DIM*0+YY] -= ty;
 644             f[j_coord_offset+DIM*0+ZZ] -= tz;
 645
 646             }
 647
 648             /**************************
 649              * CALCULATE INTERACTIONS *
 650              **************************/
 651
 652             if (rsq10<rcutoff2)
 653             {
 654
 655             r10              = rsq10*rinv10;
 656
 657             qq10             = iq1*jq0;
 658
 659             /* EWALD ELECTROSTATICS */
 660
 661             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 662             ewrt             = r10*ewtabscale;
 663             ewitab           = ewrt;
 664             eweps            = ewrt-ewitab;
 665             felec            = (1.0-eweps)*ewtab[ewitab]+eweps*ewtab[ewitab+1];
 666             felec            = qq10*rinv10*(rinvsq10-felec);
 667
 668             fscal            = felec;
 669
 670             /* Calculate temporary vectorial force */
 671             tx               = fscal*dx10;
 672             ty               = fscal*dy10;
 673             tz               = fscal*dz10;
 674
 675             /* Update vectorial force */
 676             fix1            += tx;
 677             fiy1            += ty;
 678             fiz1            += tz;
 679             f[j_coord_offset+DIM*0+XX] -= tx;
 680             f[j_coord_offset+DIM*0+YY] -= ty;
 681             f[j_coord_offset+DIM*0+ZZ] -= tz;
 682
 683             }
 684
 685             /**************************
 686              * CALCULATE INTERACTIONS *
 687              **************************/
 688
 689             if (rsq20<rcutoff2)
 690             {
 691
 692             r20              = rsq20*rinv20;
 693
 694             qq20             = iq2*jq0;
 695
 696             /* EWALD ELECTROSTATICS */
 697
 698             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 699             ewrt             = r20*ewtabscale;
 700             ewitab           = ewrt;
 701             eweps            = ewrt-ewitab;
 702             felec            = (1.0-eweps)*ewtab[ewitab]+eweps*ewtab[ewitab+1];
 703             felec            = qq20*rinv20*(rinvsq20-felec);
 704
 705             fscal            = felec;
 706
 707             /* Calculate temporary vectorial force */
 708             tx               = fscal*dx20;
 709             ty               = fscal*dy20;
 710             tz               = fscal*dz20;
 711
 712             /* Update vectorial force */
 713             fix2            += tx;
 714             fiy2            += ty;
 715             fiz2            += tz;
 716             f[j_coord_offset+DIM*0+XX] -= tx;
 717             f[j_coord_offset+DIM*0+YY] -= ty;
 718             f[j_coord_offset+DIM*0+ZZ] -= tz;
 719
 720             }
 721
 722             /**************************
 723              * CALCULATE INTERACTIONS *
 724              **************************/
 725
 726             if (rsq30<rcutoff2)
 727             {
 728
 729             r30              = rsq30*rinv30;
 730
 731             qq30             = iq3*jq0;
 732
 733             /* EWALD ELECTROSTATICS */
 734
 735             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 736             ewrt             = r30*ewtabscale;
 737             ewitab           = ewrt;
 738             eweps            = ewrt-ewitab;
 739             felec            = (1.0-eweps)*ewtab[ewitab]+eweps*ewtab[ewitab+1];
 740             felec            = qq30*rinv30*(rinvsq30-felec);
 741
 742             fscal            = felec;
 743
 744             /* Calculate temporary vectorial force */
 745             tx               = fscal*dx30;
 746             ty               = fscal*dy30;
 747             tz               = fscal*dz30;
 748
 749             /* Update vectorial force */
 750             fix3            += tx;
 751             fiy3            += ty;
 752             fiz3            += tz;
 753             f[j_coord_offset+DIM*0+XX] -= tx;
 754             f[j_coord_offset+DIM*0+YY] -= ty;
 755             f[j_coord_offset+DIM*0+ZZ] -= tz;
 756
 757             }
 758
 759             /* Inner loop uses 129 flops */
 760         }
 761         /* End of innermost loop */
 762
 763         tx = ty = tz = 0;
 764         f[i_coord_offset+DIM*0+XX] += fix0;
 765         f[i_coord_offset+DIM*0+YY] += fiy0;
 766         f[i_coord_offset+DIM*0+ZZ] += fiz0;
 767         tx                         += fix0;
 768         ty                         += fiy0;
 769         tz                         += fiz0;
 770         f[i_coord_offset+DIM*1+XX] += fix1;
 771         f[i_coord_offset+DIM*1+YY] += fiy1;
 772         f[i_coord_offset+DIM*1+ZZ] += fiz1;
 773         tx                         += fix1;
 774         ty                         += fiy1;
 775         tz                         += fiz1;
 776         f[i_coord_offset+DIM*2+XX] += fix2;
 777         f[i_coord_offset+DIM*2+YY] += fiy2;
 778         f[i_coord_offset+DIM*2+ZZ] += fiz2;
 779         tx                         += fix2;
 780         ty                         += fiy2;
 781         tz                         += fiz2;
 782         f[i_coord_offset+DIM*3+XX] += fix3;
 783         f[i_coord_offset+DIM*3+YY] += fiy3;
 784         f[i_coord_offset+DIM*3+ZZ] += fiz3;
 785         tx                         += fix3;
 786         ty                         += fiy3;
 787         tz                         += fiz3;
 788         fshift[i_shift_offset+XX]  += tx;
 789         fshift[i_shift_offset+YY]  += ty;
 790         fshift[i_shift_offset+ZZ]  += tz;
 791
 792         /* Increment number of inner iterations */
 793         inneriter                  += j_index_end - j_index_start;
 794
 795         /* Outer loop uses 39 flops */
 796     }
 797
 798     /* Increment number of outer iterations */
 799     outeriter        += nri;
 800
 801     /* Update outer/inner flops */
 802
 803     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W4_F,outeriter*39 + inneriter*129);
 804 }