src/gromacs/gmxlib/nonbonded/nb_kernel_c/nb_kernel_ElecRFCut_VdwCSTab_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_ElecRFCut_VdwCSTab_GeomW4P1_VF_c
  51  * Electrostatics interaction: ReactionField
  52  * VdW interaction:            CubicSplineTable
  53  * Geometry:                   Water4-Particle
  54  * Calculate force/pot:        PotentialAndForce
  55  */
  56 void
  57 nb_kernel_ElecRFCut_VdwCSTab_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              vfitab;
  93     real             rt,vfeps,vftabscale,Y,F,Geps,Heps2,Fp,VV,FF;
  94     real             *vftab;
  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     krf              = fr->ic->k_rf;
 110     krf2             = krf*2.0;
 111     crf              = fr->ic->c_rf;
 112     nvdwtype         = fr->ntype;
 113     vdwparam         = fr->nbfp;
 114     vdwtype          = mdatoms->typeA;
 115
 116     vftab            = kernel_data->table_vdw->data;
 117     vftabscale       = kernel_data->table_vdw->scale;
 118
 119     /* Setup water-specific parameters */
 120     inr              = nlist->iinr[0];
 121     iq1              = facel*charge[inr+1];
 122     iq2              = facel*charge[inr+2];
 123     iq3              = facel*charge[inr+3];
 124     vdwioffset0      = 2*nvdwtype*vdwtype[inr+0];
 125
 126     /* When we use explicit cutoffs the value must be identical for elec and VdW, so use elec as an arbitrary choice */
 127     rcutoff          = fr->rcoulomb;
 128     rcutoff2         = rcutoff*rcutoff;
 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             rinv00           = gmx_invsqrt(rsq00);
 214             rinv10           = gmx_invsqrt(rsq10);
 215             rinv20           = gmx_invsqrt(rsq20);
 216             rinv30           = gmx_invsqrt(rsq30);
 217
 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             r00              = rsq00*rinv00;
 231
 232             c6_00            = vdwparam[vdwioffset0+vdwjidx0];
 233             c12_00           = vdwparam[vdwioffset0+vdwjidx0+1];
 234
 235             /* Calculate table index by multiplying r with table scale and truncate to integer */
 236             rt               = r00*vftabscale;
 237             vfitab           = rt;
 238             vfeps            = rt-vfitab;
 239             vfitab           = 2*4*vfitab;
 240
 241             /* CUBIC SPLINE TABLE DISPERSION */
 242             vfitab          += 0;
 243             Y                = vftab[vfitab];
 244             F                = vftab[vfitab+1];
 245             Geps             = vfeps*vftab[vfitab+2];
 246             Heps2            = vfeps*vfeps*vftab[vfitab+3];
 247             Fp               = F+Geps+Heps2;
 248             VV               = Y+vfeps*Fp;
 249             vvdw6            = c6_00*VV;
 250             FF               = Fp+Geps+2.0*Heps2;
 251             fvdw6            = c6_00*FF;
 252
 253             /* CUBIC SPLINE TABLE REPULSION */
 254             Y                = vftab[vfitab+4];
 255             F                = vftab[vfitab+5];
 256             Geps             = vfeps*vftab[vfitab+6];
 257             Heps2            = vfeps*vfeps*vftab[vfitab+7];
 258             Fp               = F+Geps+Heps2;
 259             VV               = Y+vfeps*Fp;
 260             vvdw12           = c12_00*VV;
 261             FF               = Fp+Geps+2.0*Heps2;
 262             fvdw12           = c12_00*FF;
 263             vvdw             = vvdw12+vvdw6;
 264             fvdw             = -(fvdw6+fvdw12)*vftabscale*rinv00;
 265
 266             /* Update potential sums from outer loop */
 267             vvdwsum         += vvdw;
 268
 269             fscal            = fvdw;
 270
 271             /* Calculate temporary vectorial force */
 272             tx               = fscal*dx00;
 273             ty               = fscal*dy00;
 274             tz               = fscal*dz00;
 275
 276             /* Update vectorial force */
 277             fix0            += tx;
 278             fiy0            += ty;
 279             fiz0            += tz;
 280             f[j_coord_offset+DIM*0+XX] -= tx;
 281             f[j_coord_offset+DIM*0+YY] -= ty;
 282             f[j_coord_offset+DIM*0+ZZ] -= tz;
 283
 284             /**************************
 285              * CALCULATE INTERACTIONS *
 286              **************************/
 287
 288             if (rsq10<rcutoff2)
 289             {
 290
 291             qq10             = iq1*jq0;
 292
 293             /* REACTION-FIELD ELECTROSTATICS */
 294             velec            = qq10*(rinv10+krf*rsq10-crf);
 295             felec            = qq10*(rinv10*rinvsq10-krf2);
 296
 297             /* Update potential sums from outer loop */
 298             velecsum        += velec;
 299
 300             fscal            = felec;
 301
 302             /* Calculate temporary vectorial force */
 303             tx               = fscal*dx10;
 304             ty               = fscal*dy10;
 305             tz               = fscal*dz10;
 306
 307             /* Update vectorial force */
 308             fix1            += tx;
 309             fiy1            += ty;
 310             fiz1            += tz;
 311             f[j_coord_offset+DIM*0+XX] -= tx;
 312             f[j_coord_offset+DIM*0+YY] -= ty;
 313             f[j_coord_offset+DIM*0+ZZ] -= tz;
 314
 315             }
 316
 317             /**************************
 318              * CALCULATE INTERACTIONS *
 319              **************************/
 320
 321             if (rsq20<rcutoff2)
 322             {
 323
 324             qq20             = iq2*jq0;
 325
 326             /* REACTION-FIELD ELECTROSTATICS */
 327             velec            = qq20*(rinv20+krf*rsq20-crf);
 328             felec            = qq20*(rinv20*rinvsq20-krf2);
 329
 330             /* Update potential sums from outer loop */
 331             velecsum        += velec;
 332
 333             fscal            = felec;
 334
 335             /* Calculate temporary vectorial force */
 336             tx               = fscal*dx20;
 337             ty               = fscal*dy20;
 338             tz               = fscal*dz20;
 339
 340             /* Update vectorial force */
 341             fix2            += tx;
 342             fiy2            += ty;
 343             fiz2            += tz;
 344             f[j_coord_offset+DIM*0+XX] -= tx;
 345             f[j_coord_offset+DIM*0+YY] -= ty;
 346             f[j_coord_offset+DIM*0+ZZ] -= tz;
 347
 348             }
 349
 350             /**************************
 351              * CALCULATE INTERACTIONS *
 352              **************************/
 353
 354             if (rsq30<rcutoff2)
 355             {
 356
 357             qq30             = iq3*jq0;
 358
 359             /* REACTION-FIELD ELECTROSTATICS */
 360             velec            = qq30*(rinv30+krf*rsq30-crf);
 361             felec            = qq30*(rinv30*rinvsq30-krf2);
 362
 363             /* Update potential sums from outer loop */
 364             velecsum        += velec;
 365
 366             fscal            = felec;
 367
 368             /* Calculate temporary vectorial force */
 369             tx               = fscal*dx30;
 370             ty               = fscal*dy30;
 371             tz               = fscal*dz30;
 372
 373             /* Update vectorial force */
 374             fix3            += tx;
 375             fiy3            += ty;
 376             fiz3            += tz;
 377             f[j_coord_offset+DIM*0+XX] -= tx;
 378             f[j_coord_offset+DIM*0+YY] -= ty;
 379             f[j_coord_offset+DIM*0+ZZ] -= tz;
 380
 381             }
 382
 383             /* Inner loop uses 151 flops */
 384         }
 385         /* End of innermost loop */
 386
 387         tx = ty = tz = 0;
 388         f[i_coord_offset+DIM*0+XX] += fix0;
 389         f[i_coord_offset+DIM*0+YY] += fiy0;
 390         f[i_coord_offset+DIM*0+ZZ] += fiz0;
 391         tx                         += fix0;
 392         ty                         += fiy0;
 393         tz                         += fiz0;
 394         f[i_coord_offset+DIM*1+XX] += fix1;
 395         f[i_coord_offset+DIM*1+YY] += fiy1;
 396         f[i_coord_offset+DIM*1+ZZ] += fiz1;
 397         tx                         += fix1;
 398         ty                         += fiy1;
 399         tz                         += fiz1;
 400         f[i_coord_offset+DIM*2+XX] += fix2;
 401         f[i_coord_offset+DIM*2+YY] += fiy2;
 402         f[i_coord_offset+DIM*2+ZZ] += fiz2;
 403         tx                         += fix2;
 404         ty                         += fiy2;
 405         tz                         += fiz2;
 406         f[i_coord_offset+DIM*3+XX] += fix3;
 407         f[i_coord_offset+DIM*3+YY] += fiy3;
 408         f[i_coord_offset+DIM*3+ZZ] += fiz3;
 409         tx                         += fix3;
 410         ty                         += fiy3;
 411         tz                         += fiz3;
 412         fshift[i_shift_offset+XX]  += tx;
 413         fshift[i_shift_offset+YY]  += ty;
 414         fshift[i_shift_offset+ZZ]  += tz;
 415
 416         ggid                        = gid[iidx];
 417         /* Update potential energies */
 418         kernel_data->energygrp_elec[ggid] += velecsum;
 419         kernel_data->energygrp_vdw[ggid] += vvdwsum;
 420
 421         /* Increment number of inner iterations */
 422         inneriter                  += j_index_end - j_index_start;
 423
 424         /* Outer loop uses 41 flops */
 425     }
 426
 427     /* Increment number of outer iterations */
 428     outeriter        += nri;
 429
 430     /* Update outer/inner flops */
 431
 432     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W4_VF,outeriter*41 + inneriter*151);
 433 }
 434 /*
 435  * Gromacs nonbonded kernel:   nb_kernel_ElecRFCut_VdwCSTab_GeomW4P1_F_c
 436  * Electrostatics interaction: ReactionField
 437  * VdW interaction:            CubicSplineTable
 438  * Geometry:                   Water4-Particle
 439  * Calculate force/pot:        Force
 440  */
 441 void
 442 nb_kernel_ElecRFCut_VdwCSTab_GeomW4P1_F_c
 443                     (t_nblist                    * gmx_restrict       nlist,
 444                      rvec                        * gmx_restrict          xx,
 445                      rvec                        * gmx_restrict          ff,
 446                      t_forcerec                  * gmx_restrict          fr,
 447                      t_mdatoms                   * gmx_restrict     mdatoms,
 448                      nb_kernel_data_t gmx_unused * gmx_restrict kernel_data,
 449                      t_nrnb                      * gmx_restrict        nrnb)
 450 {
 451     int              i_shift_offset,i_coord_offset,j_coord_offset;
 452     int              j_index_start,j_index_end;
 453     int              nri,inr,ggid,iidx,jidx,jnr,outeriter,inneriter;
 454     real             shX,shY,shZ,tx,ty,tz,fscal,rcutoff,rcutoff2;
 455     int              *iinr,*jindex,*jjnr,*shiftidx,*gid;
 456     real             *shiftvec,*fshift,*x,*f;
 457     int              vdwioffset0;
 458     real             ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
 459     int              vdwioffset1;
 460     real             ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
 461     int              vdwioffset2;
 462     real             ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
 463     int              vdwioffset3;
 464     real             ix3,iy3,iz3,fix3,fiy3,fiz3,iq3,isai3;
 465     int              vdwjidx0;
 466     real             jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
 467     real             dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00,cexp1_00,cexp2_00;
 468     real             dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10,cexp1_10,cexp2_10;
 469     real             dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20,cexp1_20,cexp2_20;
 470     real             dx30,dy30,dz30,rsq30,rinv30,rinvsq30,r30,qq30,c6_30,c12_30,cexp1_30,cexp2_30;
 471     real             velec,felec,velecsum,facel,crf,krf,krf2;
 472     real             *charge;
 473     int              nvdwtype;
 474     real             rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,br,vvdwexp,sh_vdw_invrcut6;
 475     int              *vdwtype;
 476     real             *vdwparam;
 477     int              vfitab;
 478     real             rt,vfeps,vftabscale,Y,F,Geps,Heps2,Fp,VV,FF;
 479     real             *vftab;
 480
 481     x                = xx[0];
 482     f                = ff[0];
 483
 484     nri              = nlist->nri;
 485     iinr             = nlist->iinr;
 486     jindex           = nlist->jindex;
 487     jjnr             = nlist->jjnr;
 488     shiftidx         = nlist->shift;
 489     gid              = nlist->gid;
 490     shiftvec         = fr->shift_vec[0];
 491     fshift           = fr->fshift[0];
 492     facel            = fr->epsfac;
 493     charge           = mdatoms->chargeA;
 494     krf              = fr->ic->k_rf;
 495     krf2             = krf*2.0;
 496     crf              = fr->ic->c_rf;
 497     nvdwtype         = fr->ntype;
 498     vdwparam         = fr->nbfp;
 499     vdwtype          = mdatoms->typeA;
 500
 501     vftab            = kernel_data->table_vdw->data;
 502     vftabscale       = kernel_data->table_vdw->scale;
 503
 504     /* Setup water-specific parameters */
 505     inr              = nlist->iinr[0];
 506     iq1              = facel*charge[inr+1];
 507     iq2              = facel*charge[inr+2];
 508     iq3              = facel*charge[inr+3];
 509     vdwioffset0      = 2*nvdwtype*vdwtype[inr+0];
 510
 511     /* When we use explicit cutoffs the value must be identical for elec and VdW, so use elec as an arbitrary choice */
 512     rcutoff          = fr->rcoulomb;
 513     rcutoff2         = rcutoff*rcutoff;
 514
 515     outeriter        = 0;
 516     inneriter        = 0;
 517
 518     /* Start outer loop over neighborlists */
 519     for(iidx=0; iidx<nri; iidx++)
 520     {
 521         /* Load shift vector for this list */
 522         i_shift_offset   = DIM*shiftidx[iidx];
 523         shX              = shiftvec[i_shift_offset+XX];
 524         shY              = shiftvec[i_shift_offset+YY];
 525         shZ              = shiftvec[i_shift_offset+ZZ];
 526
 527         /* Load limits for loop over neighbors */
 528         j_index_start    = jindex[iidx];
 529         j_index_end      = jindex[iidx+1];
 530
 531         /* Get outer coordinate index */
 532         inr              = iinr[iidx];
 533         i_coord_offset   = DIM*inr;
 534
 535         /* Load i particle coords and add shift vector */
 536         ix0              = shX + x[i_coord_offset+DIM*0+XX];
 537         iy0              = shY + x[i_coord_offset+DIM*0+YY];
 538         iz0              = shZ + x[i_coord_offset+DIM*0+ZZ];
 539         ix1              = shX + x[i_coord_offset+DIM*1+XX];
 540         iy1              = shY + x[i_coord_offset+DIM*1+YY];
 541         iz1              = shZ + x[i_coord_offset+DIM*1+ZZ];
 542         ix2              = shX + x[i_coord_offset+DIM*2+XX];
 543         iy2              = shY + x[i_coord_offset+DIM*2+YY];
 544         iz2              = shZ + x[i_coord_offset+DIM*2+ZZ];
 545         ix3              = shX + x[i_coord_offset+DIM*3+XX];
 546         iy3              = shY + x[i_coord_offset+DIM*3+YY];
 547         iz3              = shZ + x[i_coord_offset+DIM*3+ZZ];
 548
 549         fix0             = 0.0;
 550         fiy0             = 0.0;
 551         fiz0             = 0.0;
 552         fix1             = 0.0;
 553         fiy1             = 0.0;
 554         fiz1             = 0.0;
 555         fix2             = 0.0;
 556         fiy2             = 0.0;
 557         fiz2             = 0.0;
 558         fix3             = 0.0;
 559         fiy3             = 0.0;
 560         fiz3             = 0.0;
 561
 562         /* Start inner kernel loop */
 563         for(jidx=j_index_start; jidx<j_index_end; jidx++)
 564         {
 565             /* Get j neighbor index, and coordinate index */
 566             jnr              = jjnr[jidx];
 567             j_coord_offset   = DIM*jnr;
 568
 569             /* load j atom coordinates */
 570             jx0              = x[j_coord_offset+DIM*0+XX];
 571             jy0              = x[j_coord_offset+DIM*0+YY];
 572             jz0              = x[j_coord_offset+DIM*0+ZZ];
 573
 574             /* Calculate displacement vector */
 575             dx00             = ix0 - jx0;
 576             dy00             = iy0 - jy0;
 577             dz00             = iz0 - jz0;
 578             dx10             = ix1 - jx0;
 579             dy10             = iy1 - jy0;
 580             dz10             = iz1 - jz0;
 581             dx20             = ix2 - jx0;
 582             dy20             = iy2 - jy0;
 583             dz20             = iz2 - jz0;
 584             dx30             = ix3 - jx0;
 585             dy30             = iy3 - jy0;
 586             dz30             = iz3 - jz0;
 587
 588             /* Calculate squared distance and things based on it */
 589             rsq00            = dx00*dx00+dy00*dy00+dz00*dz00;
 590             rsq10            = dx10*dx10+dy10*dy10+dz10*dz10;
 591             rsq20            = dx20*dx20+dy20*dy20+dz20*dz20;
 592             rsq30            = dx30*dx30+dy30*dy30+dz30*dz30;
 593
 594             rinv00           = gmx_invsqrt(rsq00);
 595             rinv10           = gmx_invsqrt(rsq10);
 596             rinv20           = gmx_invsqrt(rsq20);
 597             rinv30           = gmx_invsqrt(rsq30);
 598
 599             rinvsq10         = rinv10*rinv10;
 600             rinvsq20         = rinv20*rinv20;
 601             rinvsq30         = rinv30*rinv30;
 602
 603             /* Load parameters for j particles */
 604             jq0              = charge[jnr+0];
 605             vdwjidx0         = 2*vdwtype[jnr+0];
 606
 607             /**************************
 608              * CALCULATE INTERACTIONS *
 609              **************************/
 610
 611             r00              = rsq00*rinv00;
 612
 613             c6_00            = vdwparam[vdwioffset0+vdwjidx0];
 614             c12_00           = vdwparam[vdwioffset0+vdwjidx0+1];
 615
 616             /* Calculate table index by multiplying r with table scale and truncate to integer */
 617             rt               = r00*vftabscale;
 618             vfitab           = rt;
 619             vfeps            = rt-vfitab;
 620             vfitab           = 2*4*vfitab;
 621
 622             /* CUBIC SPLINE TABLE DISPERSION */
 623             vfitab          += 0;
 624             F                = vftab[vfitab+1];
 625             Geps             = vfeps*vftab[vfitab+2];
 626             Heps2            = vfeps*vfeps*vftab[vfitab+3];
 627             Fp               = F+Geps+Heps2;
 628             FF               = Fp+Geps+2.0*Heps2;
 629             fvdw6            = c6_00*FF;
 630
 631             /* CUBIC SPLINE TABLE REPULSION */
 632             F                = vftab[vfitab+5];
 633             Geps             = vfeps*vftab[vfitab+6];
 634             Heps2            = vfeps*vfeps*vftab[vfitab+7];
 635             Fp               = F+Geps+Heps2;
 636             FF               = Fp+Geps+2.0*Heps2;
 637             fvdw12           = c12_00*FF;
 638             fvdw             = -(fvdw6+fvdw12)*vftabscale*rinv00;
 639
 640             fscal            = fvdw;
 641
 642             /* Calculate temporary vectorial force */
 643             tx               = fscal*dx00;
 644             ty               = fscal*dy00;
 645             tz               = fscal*dz00;
 646
 647             /* Update vectorial force */
 648             fix0            += tx;
 649             fiy0            += ty;
 650             fiz0            += tz;
 651             f[j_coord_offset+DIM*0+XX] -= tx;
 652             f[j_coord_offset+DIM*0+YY] -= ty;
 653             f[j_coord_offset+DIM*0+ZZ] -= tz;
 654
 655             /**************************
 656              * CALCULATE INTERACTIONS *
 657              **************************/
 658
 659             if (rsq10<rcutoff2)
 660             {
 661
 662             qq10             = iq1*jq0;
 663
 664             /* REACTION-FIELD ELECTROSTATICS */
 665             felec            = qq10*(rinv10*rinvsq10-krf2);
 666
 667             fscal            = felec;
 668
 669             /* Calculate temporary vectorial force */
 670             tx               = fscal*dx10;
 671             ty               = fscal*dy10;
 672             tz               = fscal*dz10;
 673
 674             /* Update vectorial force */
 675             fix1            += tx;
 676             fiy1            += ty;
 677             fiz1            += tz;
 678             f[j_coord_offset+DIM*0+XX] -= tx;
 679             f[j_coord_offset+DIM*0+YY] -= ty;
 680             f[j_coord_offset+DIM*0+ZZ] -= tz;
 681
 682             }
 683
 684             /**************************
 685              * CALCULATE INTERACTIONS *
 686              **************************/
 687
 688             if (rsq20<rcutoff2)
 689             {
 690
 691             qq20             = iq2*jq0;
 692
 693             /* REACTION-FIELD ELECTROSTATICS */
 694             felec            = qq20*(rinv20*rinvsq20-krf2);
 695
 696             fscal            = felec;
 697
 698             /* Calculate temporary vectorial force */
 699             tx               = fscal*dx20;
 700             ty               = fscal*dy20;
 701             tz               = fscal*dz20;
 702
 703             /* Update vectorial force */
 704             fix2            += tx;
 705             fiy2            += ty;
 706             fiz2            += tz;
 707             f[j_coord_offset+DIM*0+XX] -= tx;
 708             f[j_coord_offset+DIM*0+YY] -= ty;
 709             f[j_coord_offset+DIM*0+ZZ] -= tz;
 710
 711             }
 712
 713             /**************************
 714              * CALCULATE INTERACTIONS *
 715              **************************/
 716
 717             if (rsq30<rcutoff2)
 718             {
 719
 720             qq30             = iq3*jq0;
 721
 722             /* REACTION-FIELD ELECTROSTATICS */
 723             felec            = qq30*(rinv30*rinvsq30-krf2);
 724
 725             fscal            = felec;
 726
 727             /* Calculate temporary vectorial force */
 728             tx               = fscal*dx30;
 729             ty               = fscal*dy30;
 730             tz               = fscal*dz30;
 731
 732             /* Update vectorial force */
 733             fix3            += tx;
 734             fiy3            += ty;
 735             fiz3            += tz;
 736             f[j_coord_offset+DIM*0+XX] -= tx;
 737             f[j_coord_offset+DIM*0+YY] -= ty;
 738             f[j_coord_offset+DIM*0+ZZ] -= tz;
 739
 740             }
 741
 742             /* Inner loop uses 128 flops */
 743         }
 744         /* End of innermost loop */
 745
 746         tx = ty = tz = 0;
 747         f[i_coord_offset+DIM*0+XX] += fix0;
 748         f[i_coord_offset+DIM*0+YY] += fiy0;
 749         f[i_coord_offset+DIM*0+ZZ] += fiz0;
 750         tx                         += fix0;
 751         ty                         += fiy0;
 752         tz                         += fiz0;
 753         f[i_coord_offset+DIM*1+XX] += fix1;
 754         f[i_coord_offset+DIM*1+YY] += fiy1;
 755         f[i_coord_offset+DIM*1+ZZ] += fiz1;
 756         tx                         += fix1;
 757         ty                         += fiy1;
 758         tz                         += fiz1;
 759         f[i_coord_offset+DIM*2+XX] += fix2;
 760         f[i_coord_offset+DIM*2+YY] += fiy2;
 761         f[i_coord_offset+DIM*2+ZZ] += fiz2;
 762         tx                         += fix2;
 763         ty                         += fiy2;
 764         tz                         += fiz2;
 765         f[i_coord_offset+DIM*3+XX] += fix3;
 766         f[i_coord_offset+DIM*3+YY] += fiy3;
 767         f[i_coord_offset+DIM*3+ZZ] += fiz3;
 768         tx                         += fix3;
 769         ty                         += fiy3;
 770         tz                         += fiz3;
 771         fshift[i_shift_offset+XX]  += tx;
 772         fshift[i_shift_offset+YY]  += ty;
 773         fshift[i_shift_offset+ZZ]  += tz;
 774
 775         /* Increment number of inner iterations */
 776         inneriter                  += j_index_end - j_index_start;
 777
 778         /* Outer loop uses 39 flops */
 779     }
 780
 781     /* Increment number of outer iterations */
 782     outeriter        += nri;
 783
 784     /* Update outer/inner flops */
 785
 786     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W4_F,outeriter*39 + inneriter*128);
 787 }