src/gromacs/gmxlib/nonbonded/nb_kernel_c/nb_kernel_ElecCSTab_VdwBham_GeomW4P1_c.c

   1 /*
   2  * Note: this file was generated by the Gromacs c kernel generator.
   3  *
   4  *                This source code is part of
   5  *
   6  *                 G   R   O   M   A   C   S
   7  *
   8  * Copyright (c) 2001-2012, The GROMACS Development Team
   9  *
  10  * Gromacs is a library for molecular simulation and trajectory analysis,
  11  * written by Erik Lindahl, David van der Spoel, Berk Hess, and others - for
  12  * a full list of developers and information, check out http://www.gromacs.org
  13  *
  14  * This program is free software; you can redistribute it and/or modify it under
  15  * the terms of the GNU Lesser General Public License as published by the Free
  16  * Software Foundation; either version 2 of the License, or (at your option) any
  17  * later version.
  18  *
  19  * To help fund GROMACS development, we humbly ask that you cite
  20  * the papers people have written on it - you can find them on the website.
  21  */
  22 #ifdef HAVE_CONFIG_H
  23 #include <config.h>
  24 #endif
  25
  26 #include <math.h>
  27
  28 #include "../nb_kernel.h"
  29 #include "types/simple.h"
  30 #include "vec.h"
  31 #include "nrnb.h"
  32
  33 /*
  34  * Gromacs nonbonded kernel:   nb_kernel_ElecCSTab_VdwBham_GeomW4P1_VF_c
  35  * Electrostatics interaction: CubicSplineTable
  36  * VdW interaction:            Buckingham
  37  * Geometry:                   Water4-Particle
  38  * Calculate force/pot:        PotentialAndForce
  39  */
  40 void
  41 nb_kernel_ElecCSTab_VdwBham_GeomW4P1_VF_c
  42                     (t_nblist * gmx_restrict                nlist,
  43                      rvec * gmx_restrict                    xx,
  44                      rvec * gmx_restrict                    ff,
  45                      t_forcerec * gmx_restrict              fr,
  46                      t_mdatoms * gmx_restrict               mdatoms,
  47                      nb_kernel_data_t * gmx_restrict        kernel_data,
  48                      t_nrnb * gmx_restrict                  nrnb)
  49 {
  50     int              i_shift_offset,i_coord_offset,j_coord_offset;
  51     int              j_index_start,j_index_end;
  52     int              nri,inr,ggid,iidx,jidx,jnr,outeriter,inneriter;
  53     real             shX,shY,shZ,tx,ty,tz,fscal,rcutoff,rcutoff2;
  54     int              *iinr,*jindex,*jjnr,*shiftidx,*gid;
  55     real             *shiftvec,*fshift,*x,*f;
  56     int              vdwioffset0;
  57     real             ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
  58     int              vdwioffset1;
  59     real             ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
  60     int              vdwioffset2;
  61     real             ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
  62     int              vdwioffset3;
  63     real             ix3,iy3,iz3,fix3,fiy3,fiz3,iq3,isai3;
  64     int              vdwjidx0;
  65     real             jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
  66     real             dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00,cexp1_00,cexp2_00;
  67     real             dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10,cexp1_10,cexp2_10;
  68     real             dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20,cexp1_20,cexp2_20;
  69     real             dx30,dy30,dz30,rsq30,rinv30,rinvsq30,r30,qq30,c6_30,c12_30,cexp1_30,cexp2_30;
  70     real             velec,felec,velecsum,facel,crf,krf,krf2;
  71     real             *charge;
  72     int              nvdwtype;
  73     real             rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,br,vvdwexp,sh_vdw_invrcut6;
  74     int              *vdwtype;
  75     real             *vdwparam;
  76     int              vfitab;
  77     real             rt,vfeps,vftabscale,Y,F,Geps,Heps2,Fp,VV,FF;
  78     real             *vftab;
  79
  80     x                = xx[0];
  81     f                = ff[0];
  82
  83     nri              = nlist->nri;
  84     iinr             = nlist->iinr;
  85     jindex           = nlist->jindex;
  86     jjnr             = nlist->jjnr;
  87     shiftidx         = nlist->shift;
  88     gid              = nlist->gid;
  89     shiftvec         = fr->shift_vec[0];
  90     fshift           = fr->fshift[0];
  91     facel            = fr->epsfac;
  92     charge           = mdatoms->chargeA;
  93     nvdwtype         = fr->ntype;
  94     vdwparam         = fr->nbfp;
  95     vdwtype          = mdatoms->typeA;
  96
  97     vftab            = kernel_data->table_elec->data;
  98     vftabscale       = kernel_data->table_elec->scale;
  99
 100     /* Setup water-specific parameters */
 101     inr              = nlist->iinr[0];
 102     iq1              = facel*charge[inr+1];
 103     iq2              = facel*charge[inr+2];
 104     iq3              = facel*charge[inr+3];
 105     vdwioffset0      = 3*nvdwtype*vdwtype[inr+0];
 106
 107     outeriter        = 0;
 108     inneriter        = 0;
 109
 110     /* Start outer loop over neighborlists */
 111     for(iidx=0; iidx<nri; iidx++)
 112     {
 113         /* Load shift vector for this list */
 114         i_shift_offset   = DIM*shiftidx[iidx];
 115         shX              = shiftvec[i_shift_offset+XX];
 116         shY              = shiftvec[i_shift_offset+YY];
 117         shZ              = shiftvec[i_shift_offset+ZZ];
 118
 119         /* Load limits for loop over neighbors */
 120         j_index_start    = jindex[iidx];
 121         j_index_end      = jindex[iidx+1];
 122
 123         /* Get outer coordinate index */
 124         inr              = iinr[iidx];
 125         i_coord_offset   = DIM*inr;
 126
 127         /* Load i particle coords and add shift vector */
 128         ix0              = shX + x[i_coord_offset+DIM*0+XX];
 129         iy0              = shY + x[i_coord_offset+DIM*0+YY];
 130         iz0              = shZ + x[i_coord_offset+DIM*0+ZZ];
 131         ix1              = shX + x[i_coord_offset+DIM*1+XX];
 132         iy1              = shY + x[i_coord_offset+DIM*1+YY];
 133         iz1              = shZ + x[i_coord_offset+DIM*1+ZZ];
 134         ix2              = shX + x[i_coord_offset+DIM*2+XX];
 135         iy2              = shY + x[i_coord_offset+DIM*2+YY];
 136         iz2              = shZ + x[i_coord_offset+DIM*2+ZZ];
 137         ix3              = shX + x[i_coord_offset+DIM*3+XX];
 138         iy3              = shY + x[i_coord_offset+DIM*3+YY];
 139         iz3              = shZ + x[i_coord_offset+DIM*3+ZZ];
 140
 141         fix0             = 0.0;
 142         fiy0             = 0.0;
 143         fiz0             = 0.0;
 144         fix1             = 0.0;
 145         fiy1             = 0.0;
 146         fiz1             = 0.0;
 147         fix2             = 0.0;
 148         fiy2             = 0.0;
 149         fiz2             = 0.0;
 150         fix3             = 0.0;
 151         fiy3             = 0.0;
 152         fiz3             = 0.0;
 153
 154         /* Reset potential sums */
 155         velecsum         = 0.0;
 156         vvdwsum          = 0.0;
 157
 158         /* Start inner kernel loop */
 159         for(jidx=j_index_start; jidx<j_index_end; jidx++)
 160         {
 161             /* Get j neighbor index, and coordinate index */
 162             jnr              = jjnr[jidx];
 163             j_coord_offset   = DIM*jnr;
 164
 165             /* load j atom coordinates */
 166             jx0              = x[j_coord_offset+DIM*0+XX];
 167             jy0              = x[j_coord_offset+DIM*0+YY];
 168             jz0              = x[j_coord_offset+DIM*0+ZZ];
 169
 170             /* Calculate displacement vector */
 171             dx00             = ix0 - jx0;
 172             dy00             = iy0 - jy0;
 173             dz00             = iz0 - jz0;
 174             dx10             = ix1 - jx0;
 175             dy10             = iy1 - jy0;
 176             dz10             = iz1 - jz0;
 177             dx20             = ix2 - jx0;
 178             dy20             = iy2 - jy0;
 179             dz20             = iz2 - jz0;
 180             dx30             = ix3 - jx0;
 181             dy30             = iy3 - jy0;
 182             dz30             = iz3 - jz0;
 183
 184             /* Calculate squared distance and things based on it */
 185             rsq00            = dx00*dx00+dy00*dy00+dz00*dz00;
 186             rsq10            = dx10*dx10+dy10*dy10+dz10*dz10;
 187             rsq20            = dx20*dx20+dy20*dy20+dz20*dz20;
 188             rsq30            = dx30*dx30+dy30*dy30+dz30*dz30;
 189
 190             rinv00           = gmx_invsqrt(rsq00);
 191             rinv10           = gmx_invsqrt(rsq10);
 192             rinv20           = gmx_invsqrt(rsq20);
 193             rinv30           = gmx_invsqrt(rsq30);
 194
 195             rinvsq00         = rinv00*rinv00;
 196
 197             /* Load parameters for j particles */
 198             jq0              = charge[jnr+0];
 199             vdwjidx0         = 3*vdwtype[jnr+0];
 200
 201             /**************************
 202              * CALCULATE INTERACTIONS *
 203              **************************/
 204
 205             r00              = rsq00*rinv00;
 206
 207             c6_00            = vdwparam[vdwioffset0+vdwjidx0];
 208             cexp1_00         = vdwparam[vdwioffset0+vdwjidx0+1];
 209             cexp2_00         = vdwparam[vdwioffset0+vdwjidx0+2];
 210
 211             /* BUCKINGHAM DISPERSION/REPULSION */
 212             rinvsix          = rinvsq00*rinvsq00*rinvsq00;
 213             vvdw6            = c6_00*rinvsix;
 214             br               = cexp2_00*r00;
 215             vvdwexp          = cexp1_00*exp(-br);
 216             vvdw             = vvdwexp - vvdw6*(1.0/6.0);
 217             fvdw             = (br*vvdwexp-vvdw6)*rinvsq00;
 218
 219             /* Update potential sums from outer loop */
 220             vvdwsum         += vvdw;
 221
 222             fscal            = fvdw;
 223
 224             /* Calculate temporary vectorial force */
 225             tx               = fscal*dx00;
 226             ty               = fscal*dy00;
 227             tz               = fscal*dz00;
 228
 229             /* Update vectorial force */
 230             fix0            += tx;
 231             fiy0            += ty;
 232             fiz0            += tz;
 233             f[j_coord_offset+DIM*0+XX] -= tx;
 234             f[j_coord_offset+DIM*0+YY] -= ty;
 235             f[j_coord_offset+DIM*0+ZZ] -= tz;
 236
 237             /**************************
 238              * CALCULATE INTERACTIONS *
 239              **************************/
 240
 241             r10              = rsq10*rinv10;
 242
 243             qq10             = iq1*jq0;
 244
 245             /* Calculate table index by multiplying r with table scale and truncate to integer */
 246             rt               = r10*vftabscale;
 247             vfitab           = rt;
 248             vfeps            = rt-vfitab;
 249             vfitab           = 1*4*vfitab;
 250
 251             /* CUBIC SPLINE TABLE ELECTROSTATICS */
 252             Y                = vftab[vfitab];
 253             F                = vftab[vfitab+1];
 254             Geps             = vfeps*vftab[vfitab+2];
 255             Heps2            = vfeps*vfeps*vftab[vfitab+3];
 256             Fp               = F+Geps+Heps2;
 257             VV               = Y+vfeps*Fp;
 258             velec            = qq10*VV;
 259             FF               = Fp+Geps+2.0*Heps2;
 260             felec            = -qq10*FF*vftabscale*rinv10;
 261
 262             /* Update potential sums from outer loop */
 263             velecsum        += velec;
 264
 265             fscal            = felec;
 266
 267             /* Calculate temporary vectorial force */
 268             tx               = fscal*dx10;
 269             ty               = fscal*dy10;
 270             tz               = fscal*dz10;
 271
 272             /* Update vectorial force */
 273             fix1            += tx;
 274             fiy1            += ty;
 275             fiz1            += tz;
 276             f[j_coord_offset+DIM*0+XX] -= tx;
 277             f[j_coord_offset+DIM*0+YY] -= ty;
 278             f[j_coord_offset+DIM*0+ZZ] -= tz;
 279
 280             /**************************
 281              * CALCULATE INTERACTIONS *
 282              **************************/
 283
 284             r20              = rsq20*rinv20;
 285
 286             qq20             = iq2*jq0;
 287
 288             /* Calculate table index by multiplying r with table scale and truncate to integer */
 289             rt               = r20*vftabscale;
 290             vfitab           = rt;
 291             vfeps            = rt-vfitab;
 292             vfitab           = 1*4*vfitab;
 293
 294             /* CUBIC SPLINE TABLE ELECTROSTATICS */
 295             Y                = vftab[vfitab];
 296             F                = vftab[vfitab+1];
 297             Geps             = vfeps*vftab[vfitab+2];
 298             Heps2            = vfeps*vfeps*vftab[vfitab+3];
 299             Fp               = F+Geps+Heps2;
 300             VV               = Y+vfeps*Fp;
 301             velec            = qq20*VV;
 302             FF               = Fp+Geps+2.0*Heps2;
 303             felec            = -qq20*FF*vftabscale*rinv20;
 304
 305             /* Update potential sums from outer loop */
 306             velecsum        += velec;
 307
 308             fscal            = felec;
 309
 310             /* Calculate temporary vectorial force */
 311             tx               = fscal*dx20;
 312             ty               = fscal*dy20;
 313             tz               = fscal*dz20;
 314
 315             /* Update vectorial force */
 316             fix2            += tx;
 317             fiy2            += ty;
 318             fiz2            += tz;
 319             f[j_coord_offset+DIM*0+XX] -= tx;
 320             f[j_coord_offset+DIM*0+YY] -= ty;
 321             f[j_coord_offset+DIM*0+ZZ] -= tz;
 322
 323             /**************************
 324              * CALCULATE INTERACTIONS *
 325              **************************/
 326
 327             r30              = rsq30*rinv30;
 328
 329             qq30             = iq3*jq0;
 330
 331             /* Calculate table index by multiplying r with table scale and truncate to integer */
 332             rt               = r30*vftabscale;
 333             vfitab           = rt;
 334             vfeps            = rt-vfitab;
 335             vfitab           = 1*4*vfitab;
 336
 337             /* CUBIC SPLINE TABLE ELECTROSTATICS */
 338             Y                = vftab[vfitab];
 339             F                = vftab[vfitab+1];
 340             Geps             = vfeps*vftab[vfitab+2];
 341             Heps2            = vfeps*vfeps*vftab[vfitab+3];
 342             Fp               = F+Geps+Heps2;
 343             VV               = Y+vfeps*Fp;
 344             velec            = qq30*VV;
 345             FF               = Fp+Geps+2.0*Heps2;
 346             felec            = -qq30*FF*vftabscale*rinv30;
 347
 348             /* Update potential sums from outer loop */
 349             velecsum        += velec;
 350
 351             fscal            = felec;
 352
 353             /* Calculate temporary vectorial force */
 354             tx               = fscal*dx30;
 355             ty               = fscal*dy30;
 356             tz               = fscal*dz30;
 357
 358             /* Update vectorial force */
 359             fix3            += tx;
 360             fiy3            += ty;
 361             fiz3            += tz;
 362             f[j_coord_offset+DIM*0+XX] -= tx;
 363             f[j_coord_offset+DIM*0+YY] -= ty;
 364             f[j_coord_offset+DIM*0+ZZ] -= tz;
 365
 366             /* Inner loop uses 187 flops */
 367         }
 368         /* End of innermost loop */
 369
 370         tx = ty = tz = 0;
 371         f[i_coord_offset+DIM*0+XX] += fix0;
 372         f[i_coord_offset+DIM*0+YY] += fiy0;
 373         f[i_coord_offset+DIM*0+ZZ] += fiz0;
 374         tx                         += fix0;
 375         ty                         += fiy0;
 376         tz                         += fiz0;
 377         f[i_coord_offset+DIM*1+XX] += fix1;
 378         f[i_coord_offset+DIM*1+YY] += fiy1;
 379         f[i_coord_offset+DIM*1+ZZ] += fiz1;
 380         tx                         += fix1;
 381         ty                         += fiy1;
 382         tz                         += fiz1;
 383         f[i_coord_offset+DIM*2+XX] += fix2;
 384         f[i_coord_offset+DIM*2+YY] += fiy2;
 385         f[i_coord_offset+DIM*2+ZZ] += fiz2;
 386         tx                         += fix2;
 387         ty                         += fiy2;
 388         tz                         += fiz2;
 389         f[i_coord_offset+DIM*3+XX] += fix3;
 390         f[i_coord_offset+DIM*3+YY] += fiy3;
 391         f[i_coord_offset+DIM*3+ZZ] += fiz3;
 392         tx                         += fix3;
 393         ty                         += fiy3;
 394         tz                         += fiz3;
 395         fshift[i_shift_offset+XX]  += tx;
 396         fshift[i_shift_offset+YY]  += ty;
 397         fshift[i_shift_offset+ZZ]  += tz;
 398
 399         ggid                        = gid[iidx];
 400         /* Update potential energies */
 401         kernel_data->energygrp_elec[ggid] += velecsum;
 402         kernel_data->energygrp_vdw[ggid] += vvdwsum;
 403
 404         /* Increment number of inner iterations */
 405         inneriter                  += j_index_end - j_index_start;
 406
 407         /* Outer loop uses 41 flops */
 408     }
 409
 410     /* Increment number of outer iterations */
 411     outeriter        += nri;
 412
 413     /* Update outer/inner flops */
 414
 415     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W4_VF,outeriter*41 + inneriter*187);
 416 }
 417 /*
 418  * Gromacs nonbonded kernel:   nb_kernel_ElecCSTab_VdwBham_GeomW4P1_F_c
 419  * Electrostatics interaction: CubicSplineTable
 420  * VdW interaction:            Buckingham
 421  * Geometry:                   Water4-Particle
 422  * Calculate force/pot:        Force
 423  */
 424 void
 425 nb_kernel_ElecCSTab_VdwBham_GeomW4P1_F_c
 426                     (t_nblist * gmx_restrict                nlist,
 427                      rvec * gmx_restrict                    xx,
 428                      rvec * gmx_restrict                    ff,
 429                      t_forcerec * gmx_restrict              fr,
 430                      t_mdatoms * gmx_restrict               mdatoms,
 431                      nb_kernel_data_t * gmx_restrict        kernel_data,
 432                      t_nrnb * gmx_restrict                  nrnb)
 433 {
 434     int              i_shift_offset,i_coord_offset,j_coord_offset;
 435     int              j_index_start,j_index_end;
 436     int              nri,inr,ggid,iidx,jidx,jnr,outeriter,inneriter;
 437     real             shX,shY,shZ,tx,ty,tz,fscal,rcutoff,rcutoff2;
 438     int              *iinr,*jindex,*jjnr,*shiftidx,*gid;
 439     real             *shiftvec,*fshift,*x,*f;
 440     int              vdwioffset0;
 441     real             ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
 442     int              vdwioffset1;
 443     real             ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
 444     int              vdwioffset2;
 445     real             ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
 446     int              vdwioffset3;
 447     real             ix3,iy3,iz3,fix3,fiy3,fiz3,iq3,isai3;
 448     int              vdwjidx0;
 449     real             jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
 450     real             dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00,cexp1_00,cexp2_00;
 451     real             dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10,cexp1_10,cexp2_10;
 452     real             dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20,cexp1_20,cexp2_20;
 453     real             dx30,dy30,dz30,rsq30,rinv30,rinvsq30,r30,qq30,c6_30,c12_30,cexp1_30,cexp2_30;
 454     real             velec,felec,velecsum,facel,crf,krf,krf2;
 455     real             *charge;
 456     int              nvdwtype;
 457     real             rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,br,vvdwexp,sh_vdw_invrcut6;
 458     int              *vdwtype;
 459     real             *vdwparam;
 460     int              vfitab;
 461     real             rt,vfeps,vftabscale,Y,F,Geps,Heps2,Fp,VV,FF;
 462     real             *vftab;
 463
 464     x                = xx[0];
 465     f                = ff[0];
 466
 467     nri              = nlist->nri;
 468     iinr             = nlist->iinr;
 469     jindex           = nlist->jindex;
 470     jjnr             = nlist->jjnr;
 471     shiftidx         = nlist->shift;
 472     gid              = nlist->gid;
 473     shiftvec         = fr->shift_vec[0];
 474     fshift           = fr->fshift[0];
 475     facel            = fr->epsfac;
 476     charge           = mdatoms->chargeA;
 477     nvdwtype         = fr->ntype;
 478     vdwparam         = fr->nbfp;
 479     vdwtype          = mdatoms->typeA;
 480
 481     vftab            = kernel_data->table_elec->data;
 482     vftabscale       = kernel_data->table_elec->scale;
 483
 484     /* Setup water-specific parameters */
 485     inr              = nlist->iinr[0];
 486     iq1              = facel*charge[inr+1];
 487     iq2              = facel*charge[inr+2];
 488     iq3              = facel*charge[inr+3];
 489     vdwioffset0      = 3*nvdwtype*vdwtype[inr+0];
 490
 491     outeriter        = 0;
 492     inneriter        = 0;
 493
 494     /* Start outer loop over neighborlists */
 495     for(iidx=0; iidx<nri; iidx++)
 496     {
 497         /* Load shift vector for this list */
 498         i_shift_offset   = DIM*shiftidx[iidx];
 499         shX              = shiftvec[i_shift_offset+XX];
 500         shY              = shiftvec[i_shift_offset+YY];
 501         shZ              = shiftvec[i_shift_offset+ZZ];
 502
 503         /* Load limits for loop over neighbors */
 504         j_index_start    = jindex[iidx];
 505         j_index_end      = jindex[iidx+1];
 506
 507         /* Get outer coordinate index */
 508         inr              = iinr[iidx];
 509         i_coord_offset   = DIM*inr;
 510
 511         /* Load i particle coords and add shift vector */
 512         ix0              = shX + x[i_coord_offset+DIM*0+XX];
 513         iy0              = shY + x[i_coord_offset+DIM*0+YY];
 514         iz0              = shZ + x[i_coord_offset+DIM*0+ZZ];
 515         ix1              = shX + x[i_coord_offset+DIM*1+XX];
 516         iy1              = shY + x[i_coord_offset+DIM*1+YY];
 517         iz1              = shZ + x[i_coord_offset+DIM*1+ZZ];
 518         ix2              = shX + x[i_coord_offset+DIM*2+XX];
 519         iy2              = shY + x[i_coord_offset+DIM*2+YY];
 520         iz2              = shZ + x[i_coord_offset+DIM*2+ZZ];
 521         ix3              = shX + x[i_coord_offset+DIM*3+XX];
 522         iy3              = shY + x[i_coord_offset+DIM*3+YY];
 523         iz3              = shZ + x[i_coord_offset+DIM*3+ZZ];
 524
 525         fix0             = 0.0;
 526         fiy0             = 0.0;
 527         fiz0             = 0.0;
 528         fix1             = 0.0;
 529         fiy1             = 0.0;
 530         fiz1             = 0.0;
 531         fix2             = 0.0;
 532         fiy2             = 0.0;
 533         fiz2             = 0.0;
 534         fix3             = 0.0;
 535         fiy3             = 0.0;
 536         fiz3             = 0.0;
 537
 538         /* Start inner kernel loop */
 539         for(jidx=j_index_start; jidx<j_index_end; jidx++)
 540         {
 541             /* Get j neighbor index, and coordinate index */
 542             jnr              = jjnr[jidx];
 543             j_coord_offset   = DIM*jnr;
 544
 545             /* load j atom coordinates */
 546             jx0              = x[j_coord_offset+DIM*0+XX];
 547             jy0              = x[j_coord_offset+DIM*0+YY];
 548             jz0              = x[j_coord_offset+DIM*0+ZZ];
 549
 550             /* Calculate displacement vector */
 551             dx00             = ix0 - jx0;
 552             dy00             = iy0 - jy0;
 553             dz00             = iz0 - jz0;
 554             dx10             = ix1 - jx0;
 555             dy10             = iy1 - jy0;
 556             dz10             = iz1 - jz0;
 557             dx20             = ix2 - jx0;
 558             dy20             = iy2 - jy0;
 559             dz20             = iz2 - jz0;
 560             dx30             = ix3 - jx0;
 561             dy30             = iy3 - jy0;
 562             dz30             = iz3 - jz0;
 563
 564             /* Calculate squared distance and things based on it */
 565             rsq00            = dx00*dx00+dy00*dy00+dz00*dz00;
 566             rsq10            = dx10*dx10+dy10*dy10+dz10*dz10;
 567             rsq20            = dx20*dx20+dy20*dy20+dz20*dz20;
 568             rsq30            = dx30*dx30+dy30*dy30+dz30*dz30;
 569
 570             rinv00           = gmx_invsqrt(rsq00);
 571             rinv10           = gmx_invsqrt(rsq10);
 572             rinv20           = gmx_invsqrt(rsq20);
 573             rinv30           = gmx_invsqrt(rsq30);
 574
 575             rinvsq00         = rinv00*rinv00;
 576
 577             /* Load parameters for j particles */
 578             jq0              = charge[jnr+0];
 579             vdwjidx0         = 3*vdwtype[jnr+0];
 580
 581             /**************************
 582              * CALCULATE INTERACTIONS *
 583              **************************/
 584
 585             r00              = rsq00*rinv00;
 586
 587             c6_00            = vdwparam[vdwioffset0+vdwjidx0];
 588             cexp1_00         = vdwparam[vdwioffset0+vdwjidx0+1];
 589             cexp2_00         = vdwparam[vdwioffset0+vdwjidx0+2];
 590
 591             /* BUCKINGHAM DISPERSION/REPULSION */
 592             rinvsix          = rinvsq00*rinvsq00*rinvsq00;
 593             vvdw6            = c6_00*rinvsix;
 594             br               = cexp2_00*r00;
 595             vvdwexp          = cexp1_00*exp(-br);
 596             fvdw             = (br*vvdwexp-vvdw6)*rinvsq00;
 597
 598             fscal            = fvdw;
 599
 600             /* Calculate temporary vectorial force */
 601             tx               = fscal*dx00;
 602             ty               = fscal*dy00;
 603             tz               = fscal*dz00;
 604
 605             /* Update vectorial force */
 606             fix0            += tx;
 607             fiy0            += ty;
 608             fiz0            += tz;
 609             f[j_coord_offset+DIM*0+XX] -= tx;
 610             f[j_coord_offset+DIM*0+YY] -= ty;
 611             f[j_coord_offset+DIM*0+ZZ] -= tz;
 612
 613             /**************************
 614              * CALCULATE INTERACTIONS *
 615              **************************/
 616
 617             r10              = rsq10*rinv10;
 618
 619             qq10             = iq1*jq0;
 620
 621             /* Calculate table index by multiplying r with table scale and truncate to integer */
 622             rt               = r10*vftabscale;
 623             vfitab           = rt;
 624             vfeps            = rt-vfitab;
 625             vfitab           = 1*4*vfitab;
 626
 627             /* CUBIC SPLINE TABLE ELECTROSTATICS */
 628             F                = vftab[vfitab+1];
 629             Geps             = vfeps*vftab[vfitab+2];
 630             Heps2            = vfeps*vfeps*vftab[vfitab+3];
 631             Fp               = F+Geps+Heps2;
 632             FF               = Fp+Geps+2.0*Heps2;
 633             felec            = -qq10*FF*vftabscale*rinv10;
 634
 635             fscal            = felec;
 636
 637             /* Calculate temporary vectorial force */
 638             tx               = fscal*dx10;
 639             ty               = fscal*dy10;
 640             tz               = fscal*dz10;
 641
 642             /* Update vectorial force */
 643             fix1            += tx;
 644             fiy1            += ty;
 645             fiz1            += tz;
 646             f[j_coord_offset+DIM*0+XX] -= tx;
 647             f[j_coord_offset+DIM*0+YY] -= ty;
 648             f[j_coord_offset+DIM*0+ZZ] -= tz;
 649
 650             /**************************
 651              * CALCULATE INTERACTIONS *
 652              **************************/
 653
 654             r20              = rsq20*rinv20;
 655
 656             qq20             = iq2*jq0;
 657
 658             /* Calculate table index by multiplying r with table scale and truncate to integer */
 659             rt               = r20*vftabscale;
 660             vfitab           = rt;
 661             vfeps            = rt-vfitab;
 662             vfitab           = 1*4*vfitab;
 663
 664             /* CUBIC SPLINE TABLE ELECTROSTATICS */
 665             F                = vftab[vfitab+1];
 666             Geps             = vfeps*vftab[vfitab+2];
 667             Heps2            = vfeps*vfeps*vftab[vfitab+3];
 668             Fp               = F+Geps+Heps2;
 669             FF               = Fp+Geps+2.0*Heps2;
 670             felec            = -qq20*FF*vftabscale*rinv20;
 671
 672             fscal            = felec;
 673
 674             /* Calculate temporary vectorial force */
 675             tx               = fscal*dx20;
 676             ty               = fscal*dy20;
 677             tz               = fscal*dz20;
 678
 679             /* Update vectorial force */
 680             fix2            += tx;
 681             fiy2            += ty;
 682             fiz2            += tz;
 683             f[j_coord_offset+DIM*0+XX] -= tx;
 684             f[j_coord_offset+DIM*0+YY] -= ty;
 685             f[j_coord_offset+DIM*0+ZZ] -= tz;
 686
 687             /**************************
 688              * CALCULATE INTERACTIONS *
 689              **************************/
 690
 691             r30              = rsq30*rinv30;
 692
 693             qq30             = iq3*jq0;
 694
 695             /* Calculate table index by multiplying r with table scale and truncate to integer */
 696             rt               = r30*vftabscale;
 697             vfitab           = rt;
 698             vfeps            = rt-vfitab;
 699             vfitab           = 1*4*vfitab;
 700
 701             /* CUBIC SPLINE TABLE ELECTROSTATICS */
 702             F                = vftab[vfitab+1];
 703             Geps             = vfeps*vftab[vfitab+2];
 704             Heps2            = vfeps*vfeps*vftab[vfitab+3];
 705             Fp               = F+Geps+Heps2;
 706             FF               = Fp+Geps+2.0*Heps2;
 707             felec            = -qq30*FF*vftabscale*rinv30;
 708
 709             fscal            = felec;
 710
 711             /* Calculate temporary vectorial force */
 712             tx               = fscal*dx30;
 713             ty               = fscal*dy30;
 714             tz               = fscal*dz30;
 715
 716             /* Update vectorial force */
 717             fix3            += tx;
 718             fiy3            += ty;
 719             fiz3            += tz;
 720             f[j_coord_offset+DIM*0+XX] -= tx;
 721             f[j_coord_offset+DIM*0+YY] -= ty;
 722             f[j_coord_offset+DIM*0+ZZ] -= tz;
 723
 724             /* Inner loop uses 172 flops */
 725         }
 726         /* End of innermost loop */
 727
 728         tx = ty = tz = 0;
 729         f[i_coord_offset+DIM*0+XX] += fix0;
 730         f[i_coord_offset+DIM*0+YY] += fiy0;
 731         f[i_coord_offset+DIM*0+ZZ] += fiz0;
 732         tx                         += fix0;
 733         ty                         += fiy0;
 734         tz                         += fiz0;
 735         f[i_coord_offset+DIM*1+XX] += fix1;
 736         f[i_coord_offset+DIM*1+YY] += fiy1;
 737         f[i_coord_offset+DIM*1+ZZ] += fiz1;
 738         tx                         += fix1;
 739         ty                         += fiy1;
 740         tz                         += fiz1;
 741         f[i_coord_offset+DIM*2+XX] += fix2;
 742         f[i_coord_offset+DIM*2+YY] += fiy2;
 743         f[i_coord_offset+DIM*2+ZZ] += fiz2;
 744         tx                         += fix2;
 745         ty                         += fiy2;
 746         tz                         += fiz2;
 747         f[i_coord_offset+DIM*3+XX] += fix3;
 748         f[i_coord_offset+DIM*3+YY] += fiy3;
 749         f[i_coord_offset+DIM*3+ZZ] += fiz3;
 750         tx                         += fix3;
 751         ty                         += fiy3;
 752         tz                         += fiz3;
 753         fshift[i_shift_offset+XX]  += tx;
 754         fshift[i_shift_offset+YY]  += ty;
 755         fshift[i_shift_offset+ZZ]  += tz;
 756
 757         /* Increment number of inner iterations */
 758         inneriter                  += j_index_end - j_index_start;
 759
 760         /* Outer loop uses 39 flops */
 761     }
 762
 763     /* Increment number of outer iterations */
 764     outeriter        += nri;
 765
 766     /* Update outer/inner flops */
 767
 768     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W4_F,outeriter*39 + inneriter*172);
 769 }