src/gromacs/gmxlib/nonbonded/nb_kernel_c/nb_kernel_ElecEwSh_VdwLJEwSh_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 "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_VdwLJEwSh_GeomW3P1_VF_c
  51  * Electrostatics interaction: Ewald
  52  * VdW interaction:            LJEwald
  53  * Geometry:                   Water3-Particle
  54  * Calculate force/pot:        PotentialAndForce
  55  */
  56 void
  57 nb_kernel_ElecEwSh_VdwLJEwSh_GeomW3P1_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              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             velec,felec,velecsum,facel,crf,krf,krf2;
  84     real             *charge;
  85     int              nvdwtype;
  86     real             rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,br,vvdwexp,sh_vdw_invrcut6;
  87     int              *vdwtype;
  88     real             *vdwparam;
  89     real             c6grid_00;
  90     real             c6grid_10;
  91     real             c6grid_20;
  92     real             ewclj,ewclj2,ewclj6,ewcljrsq,poly,exponent,sh_lj_ewald;
  93     real             *vdwgridparam;
  94     int              ewitab;
  95     real             ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace;
  96     real             *ewtab;
  97
  98     x                = xx[0];
  99     f                = ff[0];
 100
 101     nri              = nlist->nri;
 102     iinr             = nlist->iinr;
 103     jindex           = nlist->jindex;
 104     jjnr             = nlist->jjnr;
 105     shiftidx         = nlist->shift;
 106     gid              = nlist->gid;
 107     shiftvec         = fr->shift_vec[0];
 108     fshift           = fr->fshift[0];
 109     facel            = fr->epsfac;
 110     charge           = mdatoms->chargeA;
 111     nvdwtype         = fr->ntype;
 112     vdwparam         = fr->nbfp;
 113     vdwtype          = mdatoms->typeA;
 114     vdwgridparam     = fr->ljpme_c6grid;
 115     ewclj            = fr->ewaldcoeff_lj;
 116     sh_lj_ewald      = fr->ic->sh_lj_ewald;
 117     ewclj2           = ewclj*ewclj;
 118     ewclj6           = ewclj2*ewclj2*ewclj2;
 119
 120     sh_ewald         = fr->ic->sh_ewald;
 121     ewtab            = fr->ic->tabq_coul_FDV0;
 122     ewtabscale       = fr->ic->tabq_scale;
 123     ewtabhalfspace   = 0.5/ewtabscale;
 124
 125     /* Setup water-specific parameters */
 126     inr              = nlist->iinr[0];
 127     iq0              = facel*charge[inr+0];
 128     iq1              = facel*charge[inr+1];
 129     iq2              = facel*charge[inr+2];
 130     vdwioffset0      = 2*nvdwtype*vdwtype[inr+0];
 131
 132     /* When we use explicit cutoffs the value must be identical for elec and VdW, so use elec as an arbitrary choice */
 133     rcutoff          = fr->rcoulomb;
 134     rcutoff2         = rcutoff*rcutoff;
 135
 136     sh_vdw_invrcut6  = fr->ic->sh_invrc6;
 137     rvdw             = fr->rvdw;
 138
 139     outeriter        = 0;
 140     inneriter        = 0;
 141
 142     /* Start outer loop over neighborlists */
 143     for(iidx=0; iidx<nri; iidx++)
 144     {
 145         /* Load shift vector for this list */
 146         i_shift_offset   = DIM*shiftidx[iidx];
 147         shX              = shiftvec[i_shift_offset+XX];
 148         shY              = shiftvec[i_shift_offset+YY];
 149         shZ              = shiftvec[i_shift_offset+ZZ];
 150
 151         /* Load limits for loop over neighbors */
 152         j_index_start    = jindex[iidx];
 153         j_index_end      = jindex[iidx+1];
 154
 155         /* Get outer coordinate index */
 156         inr              = iinr[iidx];
 157         i_coord_offset   = DIM*inr;
 158
 159         /* Load i particle coords and add shift vector */
 160         ix0              = shX + x[i_coord_offset+DIM*0+XX];
 161         iy0              = shY + x[i_coord_offset+DIM*0+YY];
 162         iz0              = shZ + x[i_coord_offset+DIM*0+ZZ];
 163         ix1              = shX + x[i_coord_offset+DIM*1+XX];
 164         iy1              = shY + x[i_coord_offset+DIM*1+YY];
 165         iz1              = shZ + x[i_coord_offset+DIM*1+ZZ];
 166         ix2              = shX + x[i_coord_offset+DIM*2+XX];
 167         iy2              = shY + x[i_coord_offset+DIM*2+YY];
 168         iz2              = shZ + x[i_coord_offset+DIM*2+ZZ];
 169
 170         fix0             = 0.0;
 171         fiy0             = 0.0;
 172         fiz0             = 0.0;
 173         fix1             = 0.0;
 174         fiy1             = 0.0;
 175         fiz1             = 0.0;
 176         fix2             = 0.0;
 177         fiy2             = 0.0;
 178         fiz2             = 0.0;
 179
 180         /* Reset potential sums */
 181         velecsum         = 0.0;
 182         vvdwsum          = 0.0;
 183
 184         /* Start inner kernel loop */
 185         for(jidx=j_index_start; jidx<j_index_end; jidx++)
 186         {
 187             /* Get j neighbor index, and coordinate index */
 188             jnr              = jjnr[jidx];
 189             j_coord_offset   = DIM*jnr;
 190
 191             /* load j atom coordinates */
 192             jx0              = x[j_coord_offset+DIM*0+XX];
 193             jy0              = x[j_coord_offset+DIM*0+YY];
 194             jz0              = x[j_coord_offset+DIM*0+ZZ];
 195
 196             /* Calculate displacement vector */
 197             dx00             = ix0 - jx0;
 198             dy00             = iy0 - jy0;
 199             dz00             = iz0 - jz0;
 200             dx10             = ix1 - jx0;
 201             dy10             = iy1 - jy0;
 202             dz10             = iz1 - jz0;
 203             dx20             = ix2 - jx0;
 204             dy20             = iy2 - jy0;
 205             dz20             = iz2 - 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
 212             rinv00           = gmx_invsqrt(rsq00);
 213             rinv10           = gmx_invsqrt(rsq10);
 214             rinv20           = gmx_invsqrt(rsq20);
 215
 216             rinvsq00         = rinv00*rinv00;
 217             rinvsq10         = rinv10*rinv10;
 218             rinvsq20         = rinv20*rinv20;
 219
 220             /* Load parameters for j particles */
 221             jq0              = charge[jnr+0];
 222             vdwjidx0         = 2*vdwtype[jnr+0];
 223
 224             /**************************
 225              * CALCULATE INTERACTIONS *
 226              **************************/
 227
 228             if (rsq00<rcutoff2)
 229             {
 230
 231             r00              = rsq00*rinv00;
 232
 233             qq00             = iq0*jq0;
 234             c6_00            = vdwparam[vdwioffset0+vdwjidx0];
 235             c12_00           = vdwparam[vdwioffset0+vdwjidx0+1];
 236             c6grid_00        = vdwgridparam[vdwioffset0+vdwjidx0];
 237
 238             /* EWALD ELECTROSTATICS */
 239
 240             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 241             ewrt             = r00*ewtabscale;
 242             ewitab           = ewrt;
 243             eweps            = ewrt-ewitab;
 244             ewitab           = 4*ewitab;
 245             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 246             velec            = qq00*((rinv00-sh_ewald)-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 247             felec            = qq00*rinv00*(rinvsq00-felec);
 248
 249             rinvsix          = rinvsq00*rinvsq00*rinvsq00;
 250             ewcljrsq         = ewclj2*rsq00;
 251             exponent         = exp(-ewcljrsq);
 252             poly             = exponent*(1.0 + ewcljrsq + ewcljrsq*ewcljrsq*0.5);
 253             vvdw6            = (c6_00-c6grid_00*(1.0-poly))*rinvsix;
 254             vvdw12           = c12_00*rinvsix*rinvsix;
 255             vvdw             = (vvdw12 - c12_00*sh_vdw_invrcut6*sh_vdw_invrcut6)*(1.0/12.0) - (vvdw6 - c6_00*sh_vdw_invrcut6 - c6grid_00*sh_lj_ewald)*(1.0/6.0);
 256             fvdw             = (vvdw12 - vvdw6 - c6grid_00*(1.0/6.0)*exponent*ewclj6)*rinvsq00;
 257
 258             /* Update potential sums from outer loop */
 259             velecsum        += velec;
 260             vvdwsum         += vvdw;
 261
 262             fscal            = felec+fvdw;
 263
 264             /* Calculate temporary vectorial force */
 265             tx               = fscal*dx00;
 266             ty               = fscal*dy00;
 267             tz               = fscal*dz00;
 268
 269             /* Update vectorial force */
 270             fix0            += tx;
 271             fiy0            += ty;
 272             fiz0            += tz;
 273             f[j_coord_offset+DIM*0+XX] -= tx;
 274             f[j_coord_offset+DIM*0+YY] -= ty;
 275             f[j_coord_offset+DIM*0+ZZ] -= tz;
 276
 277             }
 278
 279             /**************************
 280              * CALCULATE INTERACTIONS *
 281              **************************/
 282
 283             if (rsq10<rcutoff2)
 284             {
 285
 286             r10              = rsq10*rinv10;
 287
 288             qq10             = iq1*jq0;
 289
 290             /* EWALD ELECTROSTATICS */
 291
 292             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 293             ewrt             = r10*ewtabscale;
 294             ewitab           = ewrt;
 295             eweps            = ewrt-ewitab;
 296             ewitab           = 4*ewitab;
 297             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 298             velec            = qq10*((rinv10-sh_ewald)-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 299             felec            = qq10*rinv10*(rinvsq10-felec);
 300
 301             /* Update potential sums from outer loop */
 302             velecsum        += velec;
 303
 304             fscal            = felec;
 305
 306             /* Calculate temporary vectorial force */
 307             tx               = fscal*dx10;
 308             ty               = fscal*dy10;
 309             tz               = fscal*dz10;
 310
 311             /* Update vectorial force */
 312             fix1            += tx;
 313             fiy1            += ty;
 314             fiz1            += tz;
 315             f[j_coord_offset+DIM*0+XX] -= tx;
 316             f[j_coord_offset+DIM*0+YY] -= ty;
 317             f[j_coord_offset+DIM*0+ZZ] -= tz;
 318
 319             }
 320
 321             /**************************
 322              * CALCULATE INTERACTIONS *
 323              **************************/
 324
 325             if (rsq20<rcutoff2)
 326             {
 327
 328             r20              = rsq20*rinv20;
 329
 330             qq20             = iq2*jq0;
 331
 332             /* EWALD ELECTROSTATICS */
 333
 334             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 335             ewrt             = r20*ewtabscale;
 336             ewitab           = ewrt;
 337             eweps            = ewrt-ewitab;
 338             ewitab           = 4*ewitab;
 339             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 340             velec            = qq20*((rinv20-sh_ewald)-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 341             felec            = qq20*rinv20*(rinvsq20-felec);
 342
 343             /* Update potential sums from outer loop */
 344             velecsum        += velec;
 345
 346             fscal            = felec;
 347
 348             /* Calculate temporary vectorial force */
 349             tx               = fscal*dx20;
 350             ty               = fscal*dy20;
 351             tz               = fscal*dz20;
 352
 353             /* Update vectorial force */
 354             fix2            += tx;
 355             fiy2            += ty;
 356             fiz2            += tz;
 357             f[j_coord_offset+DIM*0+XX] -= tx;
 358             f[j_coord_offset+DIM*0+YY] -= ty;
 359             f[j_coord_offset+DIM*0+ZZ] -= tz;
 360
 361             }
 362
 363             /* Inner loop uses 158 flops */
 364         }
 365         /* End of innermost loop */
 366
 367         tx = ty = tz = 0;
 368         f[i_coord_offset+DIM*0+XX] += fix0;
 369         f[i_coord_offset+DIM*0+YY] += fiy0;
 370         f[i_coord_offset+DIM*0+ZZ] += fiz0;
 371         tx                         += fix0;
 372         ty                         += fiy0;
 373         tz                         += fiz0;
 374         f[i_coord_offset+DIM*1+XX] += fix1;
 375         f[i_coord_offset+DIM*1+YY] += fiy1;
 376         f[i_coord_offset+DIM*1+ZZ] += fiz1;
 377         tx                         += fix1;
 378         ty                         += fiy1;
 379         tz                         += fiz1;
 380         f[i_coord_offset+DIM*2+XX] += fix2;
 381         f[i_coord_offset+DIM*2+YY] += fiy2;
 382         f[i_coord_offset+DIM*2+ZZ] += fiz2;
 383         tx                         += fix2;
 384         ty                         += fiy2;
 385         tz                         += fiz2;
 386         fshift[i_shift_offset+XX]  += tx;
 387         fshift[i_shift_offset+YY]  += ty;
 388         fshift[i_shift_offset+ZZ]  += tz;
 389
 390         ggid                        = gid[iidx];
 391         /* Update potential energies */
 392         kernel_data->energygrp_elec[ggid] += velecsum;
 393         kernel_data->energygrp_vdw[ggid] += vvdwsum;
 394
 395         /* Increment number of inner iterations */
 396         inneriter                  += j_index_end - j_index_start;
 397
 398         /* Outer loop uses 32 flops */
 399     }
 400
 401     /* Increment number of outer iterations */
 402     outeriter        += nri;
 403
 404     /* Update outer/inner flops */
 405
 406     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W3_VF,outeriter*32 + inneriter*158);
 407 }
 408 /*
 409  * Gromacs nonbonded kernel:   nb_kernel_ElecEwSh_VdwLJEwSh_GeomW3P1_F_c
 410  * Electrostatics interaction: Ewald
 411  * VdW interaction:            LJEwald
 412  * Geometry:                   Water3-Particle
 413  * Calculate force/pot:        Force
 414  */
 415 void
 416 nb_kernel_ElecEwSh_VdwLJEwSh_GeomW3P1_F_c
 417                     (t_nblist                    * gmx_restrict       nlist,
 418                      rvec                        * gmx_restrict          xx,
 419                      rvec                        * gmx_restrict          ff,
 420                      t_forcerec                  * gmx_restrict          fr,
 421                      t_mdatoms                   * gmx_restrict     mdatoms,
 422                      nb_kernel_data_t gmx_unused * gmx_restrict kernel_data,
 423                      t_nrnb                      * gmx_restrict        nrnb)
 424 {
 425     int              i_shift_offset,i_coord_offset,j_coord_offset;
 426     int              j_index_start,j_index_end;
 427     int              nri,inr,ggid,iidx,jidx,jnr,outeriter,inneriter;
 428     real             shX,shY,shZ,tx,ty,tz,fscal,rcutoff,rcutoff2;
 429     int              *iinr,*jindex,*jjnr,*shiftidx,*gid;
 430     real             *shiftvec,*fshift,*x,*f;
 431     int              vdwioffset0;
 432     real             ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
 433     int              vdwioffset1;
 434     real             ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
 435     int              vdwioffset2;
 436     real             ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
 437     int              vdwjidx0;
 438     real             jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
 439     real             dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00,cexp1_00,cexp2_00;
 440     real             dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10,cexp1_10,cexp2_10;
 441     real             dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20,cexp1_20,cexp2_20;
 442     real             velec,felec,velecsum,facel,crf,krf,krf2;
 443     real             *charge;
 444     int              nvdwtype;
 445     real             rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,br,vvdwexp,sh_vdw_invrcut6;
 446     int              *vdwtype;
 447     real             *vdwparam;
 448     real             c6grid_00;
 449     real             c6grid_10;
 450     real             c6grid_20;
 451     real             ewclj,ewclj2,ewclj6,ewcljrsq,poly,exponent,sh_lj_ewald;
 452     real             *vdwgridparam;
 453     int              ewitab;
 454     real             ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace;
 455     real             *ewtab;
 456
 457     x                = xx[0];
 458     f                = ff[0];
 459
 460     nri              = nlist->nri;
 461     iinr             = nlist->iinr;
 462     jindex           = nlist->jindex;
 463     jjnr             = nlist->jjnr;
 464     shiftidx         = nlist->shift;
 465     gid              = nlist->gid;
 466     shiftvec         = fr->shift_vec[0];
 467     fshift           = fr->fshift[0];
 468     facel            = fr->epsfac;
 469     charge           = mdatoms->chargeA;
 470     nvdwtype         = fr->ntype;
 471     vdwparam         = fr->nbfp;
 472     vdwtype          = mdatoms->typeA;
 473     vdwgridparam     = fr->ljpme_c6grid;
 474     ewclj            = fr->ewaldcoeff_lj;
 475     sh_lj_ewald      = fr->ic->sh_lj_ewald;
 476     ewclj2           = ewclj*ewclj;
 477     ewclj6           = ewclj2*ewclj2*ewclj2;
 478
 479     sh_ewald         = fr->ic->sh_ewald;
 480     ewtab            = fr->ic->tabq_coul_F;
 481     ewtabscale       = fr->ic->tabq_scale;
 482     ewtabhalfspace   = 0.5/ewtabscale;
 483
 484     /* Setup water-specific parameters */
 485     inr              = nlist->iinr[0];
 486     iq0              = facel*charge[inr+0];
 487     iq1              = facel*charge[inr+1];
 488     iq2              = facel*charge[inr+2];
 489     vdwioffset0      = 2*nvdwtype*vdwtype[inr+0];
 490
 491     /* When we use explicit cutoffs the value must be identical for elec and VdW, so use elec as an arbitrary choice */
 492     rcutoff          = fr->rcoulomb;
 493     rcutoff2         = rcutoff*rcutoff;
 494
 495     sh_vdw_invrcut6  = fr->ic->sh_invrc6;
 496     rvdw             = fr->rvdw;
 497
 498     outeriter        = 0;
 499     inneriter        = 0;
 500
 501     /* Start outer loop over neighborlists */
 502     for(iidx=0; iidx<nri; iidx++)
 503     {
 504         /* Load shift vector for this list */
 505         i_shift_offset   = DIM*shiftidx[iidx];
 506         shX              = shiftvec[i_shift_offset+XX];
 507         shY              = shiftvec[i_shift_offset+YY];
 508         shZ              = shiftvec[i_shift_offset+ZZ];
 509
 510         /* Load limits for loop over neighbors */
 511         j_index_start    = jindex[iidx];
 512         j_index_end      = jindex[iidx+1];
 513
 514         /* Get outer coordinate index */
 515         inr              = iinr[iidx];
 516         i_coord_offset   = DIM*inr;
 517
 518         /* Load i particle coords and add shift vector */
 519         ix0              = shX + x[i_coord_offset+DIM*0+XX];
 520         iy0              = shY + x[i_coord_offset+DIM*0+YY];
 521         iz0              = shZ + x[i_coord_offset+DIM*0+ZZ];
 522         ix1              = shX + x[i_coord_offset+DIM*1+XX];
 523         iy1              = shY + x[i_coord_offset+DIM*1+YY];
 524         iz1              = shZ + x[i_coord_offset+DIM*1+ZZ];
 525         ix2              = shX + x[i_coord_offset+DIM*2+XX];
 526         iy2              = shY + x[i_coord_offset+DIM*2+YY];
 527         iz2              = shZ + x[i_coord_offset+DIM*2+ZZ];
 528
 529         fix0             = 0.0;
 530         fiy0             = 0.0;
 531         fiz0             = 0.0;
 532         fix1             = 0.0;
 533         fiy1             = 0.0;
 534         fiz1             = 0.0;
 535         fix2             = 0.0;
 536         fiy2             = 0.0;
 537         fiz2             = 0.0;
 538
 539         /* Start inner kernel loop */
 540         for(jidx=j_index_start; jidx<j_index_end; jidx++)
 541         {
 542             /* Get j neighbor index, and coordinate index */
 543             jnr              = jjnr[jidx];
 544             j_coord_offset   = DIM*jnr;
 545
 546             /* load j atom coordinates */
 547             jx0              = x[j_coord_offset+DIM*0+XX];
 548             jy0              = x[j_coord_offset+DIM*0+YY];
 549             jz0              = x[j_coord_offset+DIM*0+ZZ];
 550
 551             /* Calculate displacement vector */
 552             dx00             = ix0 - jx0;
 553             dy00             = iy0 - jy0;
 554             dz00             = iz0 - jz0;
 555             dx10             = ix1 - jx0;
 556             dy10             = iy1 - jy0;
 557             dz10             = iz1 - jz0;
 558             dx20             = ix2 - jx0;
 559             dy20             = iy2 - jy0;
 560             dz20             = iz2 - jz0;
 561
 562             /* Calculate squared distance and things based on it */
 563             rsq00            = dx00*dx00+dy00*dy00+dz00*dz00;
 564             rsq10            = dx10*dx10+dy10*dy10+dz10*dz10;
 565             rsq20            = dx20*dx20+dy20*dy20+dz20*dz20;
 566
 567             rinv00           = gmx_invsqrt(rsq00);
 568             rinv10           = gmx_invsqrt(rsq10);
 569             rinv20           = gmx_invsqrt(rsq20);
 570
 571             rinvsq00         = rinv00*rinv00;
 572             rinvsq10         = rinv10*rinv10;
 573             rinvsq20         = rinv20*rinv20;
 574
 575             /* Load parameters for j particles */
 576             jq0              = charge[jnr+0];
 577             vdwjidx0         = 2*vdwtype[jnr+0];
 578
 579             /**************************
 580              * CALCULATE INTERACTIONS *
 581              **************************/
 582
 583             if (rsq00<rcutoff2)
 584             {
 585
 586             r00              = rsq00*rinv00;
 587
 588             qq00             = iq0*jq0;
 589             c6_00            = vdwparam[vdwioffset0+vdwjidx0];
 590             c12_00           = vdwparam[vdwioffset0+vdwjidx0+1];
 591             c6grid_00        = vdwgridparam[vdwioffset0+vdwjidx0];
 592
 593             /* EWALD ELECTROSTATICS */
 594
 595             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 596             ewrt             = r00*ewtabscale;
 597             ewitab           = ewrt;
 598             eweps            = ewrt-ewitab;
 599             felec            = (1.0-eweps)*ewtab[ewitab]+eweps*ewtab[ewitab+1];
 600             felec            = qq00*rinv00*(rinvsq00-felec);
 601
 602             rinvsix          = rinvsq00*rinvsq00*rinvsq00;
 603             ewcljrsq         = ewclj2*rsq00;
 604             exponent         = exp(-ewcljrsq);
 605             poly             = exponent*(1.0 + ewcljrsq + ewcljrsq*ewcljrsq*0.5);
 606             fvdw             = (((c12_00*rinvsix - c6_00 + c6grid_00*(1.0-poly))*rinvsix) - c6grid_00*(1.0/6.0)*exponent*ewclj6)*rinvsq00;
 607
 608             fscal            = felec+fvdw;
 609
 610             /* Calculate temporary vectorial force */
 611             tx               = fscal*dx00;
 612             ty               = fscal*dy00;
 613             tz               = fscal*dz00;
 614
 615             /* Update vectorial force */
 616             fix0            += tx;
 617             fiy0            += ty;
 618             fiz0            += tz;
 619             f[j_coord_offset+DIM*0+XX] -= tx;
 620             f[j_coord_offset+DIM*0+YY] -= ty;
 621             f[j_coord_offset+DIM*0+ZZ] -= tz;
 622
 623             }
 624
 625             /**************************
 626              * CALCULATE INTERACTIONS *
 627              **************************/
 628
 629             if (rsq10<rcutoff2)
 630             {
 631
 632             r10              = rsq10*rinv10;
 633
 634             qq10             = iq1*jq0;
 635
 636             /* EWALD ELECTROSTATICS */
 637
 638             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 639             ewrt             = r10*ewtabscale;
 640             ewitab           = ewrt;
 641             eweps            = ewrt-ewitab;
 642             felec            = (1.0-eweps)*ewtab[ewitab]+eweps*ewtab[ewitab+1];
 643             felec            = qq10*rinv10*(rinvsq10-felec);
 644
 645             fscal            = felec;
 646
 647             /* Calculate temporary vectorial force */
 648             tx               = fscal*dx10;
 649             ty               = fscal*dy10;
 650             tz               = fscal*dz10;
 651
 652             /* Update vectorial force */
 653             fix1            += tx;
 654             fiy1            += ty;
 655             fiz1            += tz;
 656             f[j_coord_offset+DIM*0+XX] -= tx;
 657             f[j_coord_offset+DIM*0+YY] -= ty;
 658             f[j_coord_offset+DIM*0+ZZ] -= tz;
 659
 660             }
 661
 662             /**************************
 663              * CALCULATE INTERACTIONS *
 664              **************************/
 665
 666             if (rsq20<rcutoff2)
 667             {
 668
 669             r20              = rsq20*rinv20;
 670
 671             qq20             = iq2*jq0;
 672
 673             /* EWALD ELECTROSTATICS */
 674
 675             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 676             ewrt             = r20*ewtabscale;
 677             ewitab           = ewrt;
 678             eweps            = ewrt-ewitab;
 679             felec            = (1.0-eweps)*ewtab[ewitab]+eweps*ewtab[ewitab+1];
 680             felec            = qq20*rinv20*(rinvsq20-felec);
 681
 682             fscal            = felec;
 683
 684             /* Calculate temporary vectorial force */
 685             tx               = fscal*dx20;
 686             ty               = fscal*dy20;
 687             tz               = fscal*dz20;
 688
 689             /* Update vectorial force */
 690             fix2            += tx;
 691             fiy2            += ty;
 692             fiz2            += tz;
 693             f[j_coord_offset+DIM*0+XX] -= tx;
 694             f[j_coord_offset+DIM*0+YY] -= ty;
 695             f[j_coord_offset+DIM*0+ZZ] -= tz;
 696
 697             }
 698
 699             /* Inner loop uses 123 flops */
 700         }
 701         /* End of innermost loop */
 702
 703         tx = ty = tz = 0;
 704         f[i_coord_offset+DIM*0+XX] += fix0;
 705         f[i_coord_offset+DIM*0+YY] += fiy0;
 706         f[i_coord_offset+DIM*0+ZZ] += fiz0;
 707         tx                         += fix0;
 708         ty                         += fiy0;
 709         tz                         += fiz0;
 710         f[i_coord_offset+DIM*1+XX] += fix1;
 711         f[i_coord_offset+DIM*1+YY] += fiy1;
 712         f[i_coord_offset+DIM*1+ZZ] += fiz1;
 713         tx                         += fix1;
 714         ty                         += fiy1;
 715         tz                         += fiz1;
 716         f[i_coord_offset+DIM*2+XX] += fix2;
 717         f[i_coord_offset+DIM*2+YY] += fiy2;
 718         f[i_coord_offset+DIM*2+ZZ] += fiz2;
 719         tx                         += fix2;
 720         ty                         += fiy2;
 721         tz                         += fiz2;
 722         fshift[i_shift_offset+XX]  += tx;
 723         fshift[i_shift_offset+YY]  += ty;
 724         fshift[i_shift_offset+ZZ]  += tz;
 725
 726         /* Increment number of inner iterations */
 727         inneriter                  += j_index_end - j_index_start;
 728
 729         /* Outer loop uses 30 flops */
 730     }
 731
 732     /* Increment number of outer iterations */
 733     outeriter        += nri;
 734
 735     /* Update outer/inner flops */
 736
 737     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W3_F,outeriter*30 + inneriter*123);
 738 }