src/gromacs/gmxlib/nonbonded/nb_kernel_c/nb_kernel_ElecEw_VdwBham_GeomW4P1_c.c

   1 /*
   2  * This file is part of the GROMACS molecular simulation package.
   3  *
   4  * Copyright (c) 2012,2013,2014, by the GROMACS development team, led by
   5  * Mark Abraham, David van der Spoel, Berk Hess, and Erik Lindahl,
   6  * and including many others, as listed in the AUTHORS file in the
   7  * top-level source directory and at http://www.gromacs.org.
   8  *
   9  * GROMACS is free software; you can redistribute it and/or
  10  * modify it under the terms of the GNU Lesser General Public License
  11  * as published by the Free Software Foundation; either version 2.1
  12  * of the License, or (at your option) any later version.
  13  *
  14  * GROMACS is distributed in the hope that it will be useful,
  15  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  16  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  17  * Lesser General Public License for more details.
  18  *
  19  * You should have received a copy of the GNU Lesser General Public
  20  * License along with GROMACS; if not, see
  21  * http://www.gnu.org/licenses, or write to the Free Software Foundation,
  22  * Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02110-1301  USA.
  23  *
  24  * If you want to redistribute modifications to GROMACS, please
  25  * consider that scientific software is very special. Version
  26  * control is crucial - bugs must be traceable. We will be happy to
  27  * consider code for inclusion in the official distribution, but
  28  * derived work must not be called official GROMACS. Details are found
  29  * in the README & COPYING files - if they are missing, get the
  30  * official version at http://www.gromacs.org.
  31  *
  32  * To help us fund GROMACS development, we humbly ask that you cite
  33  * the research papers on the package. Check out http://www.gromacs.org.
  34  */
  35 /*
  36  * Note: this file was generated by the GROMACS c kernel generator.
  37  */
  38 #include "config.h"
  39
  40 #include <math.h>
  41
  42 #include "../nb_kernel.h"
  43 #include "gromacs/legacyheaders/types/simple.h"
  44 #include "gromacs/math/vec.h"
  45 #include "gromacs/legacyheaders/nrnb.h"
  46
  47 /*
  48  * Gromacs nonbonded kernel:   nb_kernel_ElecEw_VdwBham_GeomW4P1_VF_c
  49  * Electrostatics interaction: Ewald
  50  * VdW interaction:            Buckingham
  51  * Geometry:                   Water4-Particle
  52  * Calculate force/pot:        PotentialAndForce
  53  */
  54 void
  55 nb_kernel_ElecEw_VdwBham_GeomW4P1_VF_c
  56                     (t_nblist                    * gmx_restrict       nlist,
  57                      rvec                        * gmx_restrict          xx,
  58                      rvec                        * gmx_restrict          ff,
  59                      t_forcerec                  * gmx_restrict          fr,
  60                      t_mdatoms                   * gmx_restrict     mdatoms,
  61                      nb_kernel_data_t gmx_unused * gmx_restrict kernel_data,
  62                      t_nrnb                      * gmx_restrict        nrnb)
  63 {
  64     int              i_shift_offset,i_coord_offset,j_coord_offset;
  65     int              j_index_start,j_index_end;
  66     int              nri,inr,ggid,iidx,jidx,jnr,outeriter,inneriter;
  67     real             shX,shY,shZ,tx,ty,tz,fscal,rcutoff,rcutoff2;
  68     int              *iinr,*jindex,*jjnr,*shiftidx,*gid;
  69     real             *shiftvec,*fshift,*x,*f;
  70     int              vdwioffset0;
  71     real             ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
  72     int              vdwioffset1;
  73     real             ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
  74     int              vdwioffset2;
  75     real             ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
  76     int              vdwioffset3;
  77     real             ix3,iy3,iz3,fix3,fiy3,fiz3,iq3,isai3;
  78     int              vdwjidx0;
  79     real             jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
  80     real             dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00,cexp1_00,cexp2_00;
  81     real             dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10,cexp1_10,cexp2_10;
  82     real             dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20,cexp1_20,cexp2_20;
  83     real             dx30,dy30,dz30,rsq30,rinv30,rinvsq30,r30,qq30,c6_30,c12_30,cexp1_30,cexp2_30;
  84     real             velec,felec,velecsum,facel,crf,krf,krf2;
  85     real             *charge;
  86     int              nvdwtype;
  87     real             rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,br,vvdwexp,sh_vdw_invrcut6;
  88     int              *vdwtype;
  89     real             *vdwparam;
  90     int              ewitab;
  91     real             ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace;
  92     real             *ewtab;
  93
  94     x                = xx[0];
  95     f                = ff[0];
  96
  97     nri              = nlist->nri;
  98     iinr             = nlist->iinr;
  99     jindex           = nlist->jindex;
 100     jjnr             = nlist->jjnr;
 101     shiftidx         = nlist->shift;
 102     gid              = nlist->gid;
 103     shiftvec         = fr->shift_vec[0];
 104     fshift           = fr->fshift[0];
 105     facel            = fr->epsfac;
 106     charge           = mdatoms->chargeA;
 107     nvdwtype         = fr->ntype;
 108     vdwparam         = fr->nbfp;
 109     vdwtype          = mdatoms->typeA;
 110
 111     sh_ewald         = fr->ic->sh_ewald;
 112     ewtab            = fr->ic->tabq_coul_FDV0;
 113     ewtabscale       = fr->ic->tabq_scale;
 114     ewtabhalfspace   = 0.5/ewtabscale;
 115
 116     /* Setup water-specific parameters */
 117     inr              = nlist->iinr[0];
 118     iq1              = facel*charge[inr+1];
 119     iq2              = facel*charge[inr+2];
 120     iq3              = facel*charge[inr+3];
 121     vdwioffset0      = 3*nvdwtype*vdwtype[inr+0];
 122
 123     outeriter        = 0;
 124     inneriter        = 0;
 125
 126     /* Start outer loop over neighborlists */
 127     for(iidx=0; iidx<nri; iidx++)
 128     {
 129         /* Load shift vector for this list */
 130         i_shift_offset   = DIM*shiftidx[iidx];
 131         shX              = shiftvec[i_shift_offset+XX];
 132         shY              = shiftvec[i_shift_offset+YY];
 133         shZ              = shiftvec[i_shift_offset+ZZ];
 134
 135         /* Load limits for loop over neighbors */
 136         j_index_start    = jindex[iidx];
 137         j_index_end      = jindex[iidx+1];
 138
 139         /* Get outer coordinate index */
 140         inr              = iinr[iidx];
 141         i_coord_offset   = DIM*inr;
 142
 143         /* Load i particle coords and add shift vector */
 144         ix0              = shX + x[i_coord_offset+DIM*0+XX];
 145         iy0              = shY + x[i_coord_offset+DIM*0+YY];
 146         iz0              = shZ + x[i_coord_offset+DIM*0+ZZ];
 147         ix1              = shX + x[i_coord_offset+DIM*1+XX];
 148         iy1              = shY + x[i_coord_offset+DIM*1+YY];
 149         iz1              = shZ + x[i_coord_offset+DIM*1+ZZ];
 150         ix2              = shX + x[i_coord_offset+DIM*2+XX];
 151         iy2              = shY + x[i_coord_offset+DIM*2+YY];
 152         iz2              = shZ + x[i_coord_offset+DIM*2+ZZ];
 153         ix3              = shX + x[i_coord_offset+DIM*3+XX];
 154         iy3              = shY + x[i_coord_offset+DIM*3+YY];
 155         iz3              = shZ + x[i_coord_offset+DIM*3+ZZ];
 156
 157         fix0             = 0.0;
 158         fiy0             = 0.0;
 159         fiz0             = 0.0;
 160         fix1             = 0.0;
 161         fiy1             = 0.0;
 162         fiz1             = 0.0;
 163         fix2             = 0.0;
 164         fiy2             = 0.0;
 165         fiz2             = 0.0;
 166         fix3             = 0.0;
 167         fiy3             = 0.0;
 168         fiz3             = 0.0;
 169
 170         /* Reset potential sums */
 171         velecsum         = 0.0;
 172         vvdwsum          = 0.0;
 173
 174         /* Start inner kernel loop */
 175         for(jidx=j_index_start; jidx<j_index_end; jidx++)
 176         {
 177             /* Get j neighbor index, and coordinate index */
 178             jnr              = jjnr[jidx];
 179             j_coord_offset   = DIM*jnr;
 180
 181             /* load j atom coordinates */
 182             jx0              = x[j_coord_offset+DIM*0+XX];
 183             jy0              = x[j_coord_offset+DIM*0+YY];
 184             jz0              = x[j_coord_offset+DIM*0+ZZ];
 185
 186             /* Calculate displacement vector */
 187             dx00             = ix0 - jx0;
 188             dy00             = iy0 - jy0;
 189             dz00             = iz0 - jz0;
 190             dx10             = ix1 - jx0;
 191             dy10             = iy1 - jy0;
 192             dz10             = iz1 - jz0;
 193             dx20             = ix2 - jx0;
 194             dy20             = iy2 - jy0;
 195             dz20             = iz2 - jz0;
 196             dx30             = ix3 - jx0;
 197             dy30             = iy3 - jy0;
 198             dz30             = iz3 - jz0;
 199
 200             /* Calculate squared distance and things based on it */
 201             rsq00            = dx00*dx00+dy00*dy00+dz00*dz00;
 202             rsq10            = dx10*dx10+dy10*dy10+dz10*dz10;
 203             rsq20            = dx20*dx20+dy20*dy20+dz20*dz20;
 204             rsq30            = dx30*dx30+dy30*dy30+dz30*dz30;
 205
 206             rinv00           = gmx_invsqrt(rsq00);
 207             rinv10           = gmx_invsqrt(rsq10);
 208             rinv20           = gmx_invsqrt(rsq20);
 209             rinv30           = gmx_invsqrt(rsq30);
 210
 211             rinvsq00         = rinv00*rinv00;
 212             rinvsq10         = rinv10*rinv10;
 213             rinvsq20         = rinv20*rinv20;
 214             rinvsq30         = rinv30*rinv30;
 215
 216             /* Load parameters for j particles */
 217             jq0              = charge[jnr+0];
 218             vdwjidx0         = 3*vdwtype[jnr+0];
 219
 220             /**************************
 221              * CALCULATE INTERACTIONS *
 222              **************************/
 223
 224             r00              = rsq00*rinv00;
 225
 226             c6_00            = vdwparam[vdwioffset0+vdwjidx0];
 227             cexp1_00         = vdwparam[vdwioffset0+vdwjidx0+1];
 228             cexp2_00         = vdwparam[vdwioffset0+vdwjidx0+2];
 229
 230             /* BUCKINGHAM DISPERSION/REPULSION */
 231             rinvsix          = rinvsq00*rinvsq00*rinvsq00;
 232             vvdw6            = c6_00*rinvsix;
 233             br               = cexp2_00*r00;
 234             vvdwexp          = cexp1_00*exp(-br);
 235             vvdw             = vvdwexp - vvdw6*(1.0/6.0);
 236             fvdw             = (br*vvdwexp-vvdw6)*rinvsq00;
 237
 238             /* Update potential sums from outer loop */
 239             vvdwsum         += vvdw;
 240
 241             fscal            = fvdw;
 242
 243             /* Calculate temporary vectorial force */
 244             tx               = fscal*dx00;
 245             ty               = fscal*dy00;
 246             tz               = fscal*dz00;
 247
 248             /* Update vectorial force */
 249             fix0            += tx;
 250             fiy0            += ty;
 251             fiz0            += tz;
 252             f[j_coord_offset+DIM*0+XX] -= tx;
 253             f[j_coord_offset+DIM*0+YY] -= ty;
 254             f[j_coord_offset+DIM*0+ZZ] -= tz;
 255
 256             /**************************
 257              * CALCULATE INTERACTIONS *
 258              **************************/
 259
 260             r10              = rsq10*rinv10;
 261
 262             qq10             = iq1*jq0;
 263
 264             /* EWALD ELECTROSTATICS */
 265
 266             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 267             ewrt             = r10*ewtabscale;
 268             ewitab           = ewrt;
 269             eweps            = ewrt-ewitab;
 270             ewitab           = 4*ewitab;
 271             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 272             velec            = qq10*(rinv10-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 273             felec            = qq10*rinv10*(rinvsq10-felec);
 274
 275             /* Update potential sums from outer loop */
 276             velecsum        += velec;
 277
 278             fscal            = felec;
 279
 280             /* Calculate temporary vectorial force */
 281             tx               = fscal*dx10;
 282             ty               = fscal*dy10;
 283             tz               = fscal*dz10;
 284
 285             /* Update vectorial force */
 286             fix1            += tx;
 287             fiy1            += ty;
 288             fiz1            += tz;
 289             f[j_coord_offset+DIM*0+XX] -= tx;
 290             f[j_coord_offset+DIM*0+YY] -= ty;
 291             f[j_coord_offset+DIM*0+ZZ] -= tz;
 292
 293             /**************************
 294              * CALCULATE INTERACTIONS *
 295              **************************/
 296
 297             r20              = rsq20*rinv20;
 298
 299             qq20             = iq2*jq0;
 300
 301             /* EWALD ELECTROSTATICS */
 302
 303             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 304             ewrt             = r20*ewtabscale;
 305             ewitab           = ewrt;
 306             eweps            = ewrt-ewitab;
 307             ewitab           = 4*ewitab;
 308             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 309             velec            = qq20*(rinv20-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 310             felec            = qq20*rinv20*(rinvsq20-felec);
 311
 312             /* Update potential sums from outer loop */
 313             velecsum        += velec;
 314
 315             fscal            = felec;
 316
 317             /* Calculate temporary vectorial force */
 318             tx               = fscal*dx20;
 319             ty               = fscal*dy20;
 320             tz               = fscal*dz20;
 321
 322             /* Update vectorial force */
 323             fix2            += tx;
 324             fiy2            += ty;
 325             fiz2            += tz;
 326             f[j_coord_offset+DIM*0+XX] -= tx;
 327             f[j_coord_offset+DIM*0+YY] -= ty;
 328             f[j_coord_offset+DIM*0+ZZ] -= tz;
 329
 330             /**************************
 331              * CALCULATE INTERACTIONS *
 332              **************************/
 333
 334             r30              = rsq30*rinv30;
 335
 336             qq30             = iq3*jq0;
 337
 338             /* EWALD ELECTROSTATICS */
 339
 340             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 341             ewrt             = r30*ewtabscale;
 342             ewitab           = ewrt;
 343             eweps            = ewrt-ewitab;
 344             ewitab           = 4*ewitab;
 345             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 346             velec            = qq30*(rinv30-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 347             felec            = qq30*rinv30*(rinvsq30-felec);
 348
 349             /* Update potential sums from outer loop */
 350             velecsum        += velec;
 351
 352             fscal            = felec;
 353
 354             /* Calculate temporary vectorial force */
 355             tx               = fscal*dx30;
 356             ty               = fscal*dy30;
 357             tz               = fscal*dz30;
 358
 359             /* Update vectorial force */
 360             fix3            += tx;
 361             fiy3            += ty;
 362             fiz3            += tz;
 363             f[j_coord_offset+DIM*0+XX] -= tx;
 364             f[j_coord_offset+DIM*0+YY] -= ty;
 365             f[j_coord_offset+DIM*0+ZZ] -= tz;
 366
 367             /* Inner loop uses 184 flops */
 368         }
 369         /* End of innermost loop */
 370
 371         tx = ty = tz = 0;
 372         f[i_coord_offset+DIM*0+XX] += fix0;
 373         f[i_coord_offset+DIM*0+YY] += fiy0;
 374         f[i_coord_offset+DIM*0+ZZ] += fiz0;
 375         tx                         += fix0;
 376         ty                         += fiy0;
 377         tz                         += fiz0;
 378         f[i_coord_offset+DIM*1+XX] += fix1;
 379         f[i_coord_offset+DIM*1+YY] += fiy1;
 380         f[i_coord_offset+DIM*1+ZZ] += fiz1;
 381         tx                         += fix1;
 382         ty                         += fiy1;
 383         tz                         += fiz1;
 384         f[i_coord_offset+DIM*2+XX] += fix2;
 385         f[i_coord_offset+DIM*2+YY] += fiy2;
 386         f[i_coord_offset+DIM*2+ZZ] += fiz2;
 387         tx                         += fix2;
 388         ty                         += fiy2;
 389         tz                         += fiz2;
 390         f[i_coord_offset+DIM*3+XX] += fix3;
 391         f[i_coord_offset+DIM*3+YY] += fiy3;
 392         f[i_coord_offset+DIM*3+ZZ] += fiz3;
 393         tx                         += fix3;
 394         ty                         += fiy3;
 395         tz                         += fiz3;
 396         fshift[i_shift_offset+XX]  += tx;
 397         fshift[i_shift_offset+YY]  += ty;
 398         fshift[i_shift_offset+ZZ]  += tz;
 399
 400         ggid                        = gid[iidx];
 401         /* Update potential energies */
 402         kernel_data->energygrp_elec[ggid] += velecsum;
 403         kernel_data->energygrp_vdw[ggid] += vvdwsum;
 404
 405         /* Increment number of inner iterations */
 406         inneriter                  += j_index_end - j_index_start;
 407
 408         /* Outer loop uses 41 flops */
 409     }
 410
 411     /* Increment number of outer iterations */
 412     outeriter        += nri;
 413
 414     /* Update outer/inner flops */
 415
 416     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W4_VF,outeriter*41 + inneriter*184);
 417 }
 418 /*
 419  * Gromacs nonbonded kernel:   nb_kernel_ElecEw_VdwBham_GeomW4P1_F_c
 420  * Electrostatics interaction: Ewald
 421  * VdW interaction:            Buckingham
 422  * Geometry:                   Water4-Particle
 423  * Calculate force/pot:        Force
 424  */
 425 void
 426 nb_kernel_ElecEw_VdwBham_GeomW4P1_F_c
 427                     (t_nblist                    * gmx_restrict       nlist,
 428                      rvec                        * gmx_restrict          xx,
 429                      rvec                        * gmx_restrict          ff,
 430                      t_forcerec                  * gmx_restrict          fr,
 431                      t_mdatoms                   * gmx_restrict     mdatoms,
 432                      nb_kernel_data_t gmx_unused * gmx_restrict kernel_data,
 433                      t_nrnb                      * gmx_restrict        nrnb)
 434 {
 435     int              i_shift_offset,i_coord_offset,j_coord_offset;
 436     int              j_index_start,j_index_end;
 437     int              nri,inr,ggid,iidx,jidx,jnr,outeriter,inneriter;
 438     real             shX,shY,shZ,tx,ty,tz,fscal,rcutoff,rcutoff2;
 439     int              *iinr,*jindex,*jjnr,*shiftidx,*gid;
 440     real             *shiftvec,*fshift,*x,*f;
 441     int              vdwioffset0;
 442     real             ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
 443     int              vdwioffset1;
 444     real             ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
 445     int              vdwioffset2;
 446     real             ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
 447     int              vdwioffset3;
 448     real             ix3,iy3,iz3,fix3,fiy3,fiz3,iq3,isai3;
 449     int              vdwjidx0;
 450     real             jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
 451     real             dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00,cexp1_00,cexp2_00;
 452     real             dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10,cexp1_10,cexp2_10;
 453     real             dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20,cexp1_20,cexp2_20;
 454     real             dx30,dy30,dz30,rsq30,rinv30,rinvsq30,r30,qq30,c6_30,c12_30,cexp1_30,cexp2_30;
 455     real             velec,felec,velecsum,facel,crf,krf,krf2;
 456     real             *charge;
 457     int              nvdwtype;
 458     real             rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,br,vvdwexp,sh_vdw_invrcut6;
 459     int              *vdwtype;
 460     real             *vdwparam;
 461     int              ewitab;
 462     real             ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace;
 463     real             *ewtab;
 464
 465     x                = xx[0];
 466     f                = ff[0];
 467
 468     nri              = nlist->nri;
 469     iinr             = nlist->iinr;
 470     jindex           = nlist->jindex;
 471     jjnr             = nlist->jjnr;
 472     shiftidx         = nlist->shift;
 473     gid              = nlist->gid;
 474     shiftvec         = fr->shift_vec[0];
 475     fshift           = fr->fshift[0];
 476     facel            = fr->epsfac;
 477     charge           = mdatoms->chargeA;
 478     nvdwtype         = fr->ntype;
 479     vdwparam         = fr->nbfp;
 480     vdwtype          = mdatoms->typeA;
 481
 482     sh_ewald         = fr->ic->sh_ewald;
 483     ewtab            = fr->ic->tabq_coul_F;
 484     ewtabscale       = fr->ic->tabq_scale;
 485     ewtabhalfspace   = 0.5/ewtabscale;
 486
 487     /* Setup water-specific parameters */
 488     inr              = nlist->iinr[0];
 489     iq1              = facel*charge[inr+1];
 490     iq2              = facel*charge[inr+2];
 491     iq3              = facel*charge[inr+3];
 492     vdwioffset0      = 3*nvdwtype*vdwtype[inr+0];
 493
 494     outeriter        = 0;
 495     inneriter        = 0;
 496
 497     /* Start outer loop over neighborlists */
 498     for(iidx=0; iidx<nri; iidx++)
 499     {
 500         /* Load shift vector for this list */
 501         i_shift_offset   = DIM*shiftidx[iidx];
 502         shX              = shiftvec[i_shift_offset+XX];
 503         shY              = shiftvec[i_shift_offset+YY];
 504         shZ              = shiftvec[i_shift_offset+ZZ];
 505
 506         /* Load limits for loop over neighbors */
 507         j_index_start    = jindex[iidx];
 508         j_index_end      = jindex[iidx+1];
 509
 510         /* Get outer coordinate index */
 511         inr              = iinr[iidx];
 512         i_coord_offset   = DIM*inr;
 513
 514         /* Load i particle coords and add shift vector */
 515         ix0              = shX + x[i_coord_offset+DIM*0+XX];
 516         iy0              = shY + x[i_coord_offset+DIM*0+YY];
 517         iz0              = shZ + x[i_coord_offset+DIM*0+ZZ];
 518         ix1              = shX + x[i_coord_offset+DIM*1+XX];
 519         iy1              = shY + x[i_coord_offset+DIM*1+YY];
 520         iz1              = shZ + x[i_coord_offset+DIM*1+ZZ];
 521         ix2              = shX + x[i_coord_offset+DIM*2+XX];
 522         iy2              = shY + x[i_coord_offset+DIM*2+YY];
 523         iz2              = shZ + x[i_coord_offset+DIM*2+ZZ];
 524         ix3              = shX + x[i_coord_offset+DIM*3+XX];
 525         iy3              = shY + x[i_coord_offset+DIM*3+YY];
 526         iz3              = shZ + x[i_coord_offset+DIM*3+ZZ];
 527
 528         fix0             = 0.0;
 529         fiy0             = 0.0;
 530         fiz0             = 0.0;
 531         fix1             = 0.0;
 532         fiy1             = 0.0;
 533         fiz1             = 0.0;
 534         fix2             = 0.0;
 535         fiy2             = 0.0;
 536         fiz2             = 0.0;
 537         fix3             = 0.0;
 538         fiy3             = 0.0;
 539         fiz3             = 0.0;
 540
 541         /* Start inner kernel loop */
 542         for(jidx=j_index_start; jidx<j_index_end; jidx++)
 543         {
 544             /* Get j neighbor index, and coordinate index */
 545             jnr              = jjnr[jidx];
 546             j_coord_offset   = DIM*jnr;
 547
 548             /* load j atom coordinates */
 549             jx0              = x[j_coord_offset+DIM*0+XX];
 550             jy0              = x[j_coord_offset+DIM*0+YY];
 551             jz0              = x[j_coord_offset+DIM*0+ZZ];
 552
 553             /* Calculate displacement vector */
 554             dx00             = ix0 - jx0;
 555             dy00             = iy0 - jy0;
 556             dz00             = iz0 - jz0;
 557             dx10             = ix1 - jx0;
 558             dy10             = iy1 - jy0;
 559             dz10             = iz1 - jz0;
 560             dx20             = ix2 - jx0;
 561             dy20             = iy2 - jy0;
 562             dz20             = iz2 - jz0;
 563             dx30             = ix3 - jx0;
 564             dy30             = iy3 - jy0;
 565             dz30             = iz3 - jz0;
 566
 567             /* Calculate squared distance and things based on it */
 568             rsq00            = dx00*dx00+dy00*dy00+dz00*dz00;
 569             rsq10            = dx10*dx10+dy10*dy10+dz10*dz10;
 570             rsq20            = dx20*dx20+dy20*dy20+dz20*dz20;
 571             rsq30            = dx30*dx30+dy30*dy30+dz30*dz30;
 572
 573             rinv00           = gmx_invsqrt(rsq00);
 574             rinv10           = gmx_invsqrt(rsq10);
 575             rinv20           = gmx_invsqrt(rsq20);
 576             rinv30           = gmx_invsqrt(rsq30);
 577
 578             rinvsq00         = rinv00*rinv00;
 579             rinvsq10         = rinv10*rinv10;
 580             rinvsq20         = rinv20*rinv20;
 581             rinvsq30         = rinv30*rinv30;
 582
 583             /* Load parameters for j particles */
 584             jq0              = charge[jnr+0];
 585             vdwjidx0         = 3*vdwtype[jnr+0];
 586
 587             /**************************
 588              * CALCULATE INTERACTIONS *
 589              **************************/
 590
 591             r00              = rsq00*rinv00;
 592
 593             c6_00            = vdwparam[vdwioffset0+vdwjidx0];
 594             cexp1_00         = vdwparam[vdwioffset0+vdwjidx0+1];
 595             cexp2_00         = vdwparam[vdwioffset0+vdwjidx0+2];
 596
 597             /* BUCKINGHAM DISPERSION/REPULSION */
 598             rinvsix          = rinvsq00*rinvsq00*rinvsq00;
 599             vvdw6            = c6_00*rinvsix;
 600             br               = cexp2_00*r00;
 601             vvdwexp          = cexp1_00*exp(-br);
 602             fvdw             = (br*vvdwexp-vvdw6)*rinvsq00;
 603
 604             fscal            = fvdw;
 605
 606             /* Calculate temporary vectorial force */
 607             tx               = fscal*dx00;
 608             ty               = fscal*dy00;
 609             tz               = fscal*dz00;
 610
 611             /* Update vectorial force */
 612             fix0            += tx;
 613             fiy0            += ty;
 614             fiz0            += tz;
 615             f[j_coord_offset+DIM*0+XX] -= tx;
 616             f[j_coord_offset+DIM*0+YY] -= ty;
 617             f[j_coord_offset+DIM*0+ZZ] -= tz;
 618
 619             /**************************
 620              * CALCULATE INTERACTIONS *
 621              **************************/
 622
 623             r10              = rsq10*rinv10;
 624
 625             qq10             = iq1*jq0;
 626
 627             /* EWALD ELECTROSTATICS */
 628
 629             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 630             ewrt             = r10*ewtabscale;
 631             ewitab           = ewrt;
 632             eweps            = ewrt-ewitab;
 633             felec            = (1.0-eweps)*ewtab[ewitab]+eweps*ewtab[ewitab+1];
 634             felec            = qq10*rinv10*(rinvsq10-felec);
 635
 636             fscal            = felec;
 637
 638             /* Calculate temporary vectorial force */
 639             tx               = fscal*dx10;
 640             ty               = fscal*dy10;
 641             tz               = fscal*dz10;
 642
 643             /* Update vectorial force */
 644             fix1            += tx;
 645             fiy1            += ty;
 646             fiz1            += tz;
 647             f[j_coord_offset+DIM*0+XX] -= tx;
 648             f[j_coord_offset+DIM*0+YY] -= ty;
 649             f[j_coord_offset+DIM*0+ZZ] -= tz;
 650
 651             /**************************
 652              * CALCULATE INTERACTIONS *
 653              **************************/
 654
 655             r20              = rsq20*rinv20;
 656
 657             qq20             = iq2*jq0;
 658
 659             /* EWALD ELECTROSTATICS */
 660
 661             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 662             ewrt             = r20*ewtabscale;
 663             ewitab           = ewrt;
 664             eweps            = ewrt-ewitab;
 665             felec            = (1.0-eweps)*ewtab[ewitab]+eweps*ewtab[ewitab+1];
 666             felec            = qq20*rinv20*(rinvsq20-felec);
 667
 668             fscal            = felec;
 669
 670             /* Calculate temporary vectorial force */
 671             tx               = fscal*dx20;
 672             ty               = fscal*dy20;
 673             tz               = fscal*dz20;
 674
 675             /* Update vectorial force */
 676             fix2            += tx;
 677             fiy2            += ty;
 678             fiz2            += tz;
 679             f[j_coord_offset+DIM*0+XX] -= tx;
 680             f[j_coord_offset+DIM*0+YY] -= ty;
 681             f[j_coord_offset+DIM*0+ZZ] -= tz;
 682
 683             /**************************
 684              * CALCULATE INTERACTIONS *
 685              **************************/
 686
 687             r30              = rsq30*rinv30;
 688
 689             qq30             = iq3*jq0;
 690
 691             /* EWALD ELECTROSTATICS */
 692
 693             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 694             ewrt             = r30*ewtabscale;
 695             ewitab           = ewrt;
 696             eweps            = ewrt-ewitab;
 697             felec            = (1.0-eweps)*ewtab[ewitab]+eweps*ewtab[ewitab+1];
 698             felec            = qq30*rinv30*(rinvsq30-felec);
 699
 700             fscal            = felec;
 701
 702             /* Calculate temporary vectorial force */
 703             tx               = fscal*dx30;
 704             ty               = fscal*dy30;
 705             tz               = fscal*dz30;
 706
 707             /* Update vectorial force */
 708             fix3            += tx;
 709             fiy3            += ty;
 710             fiz3            += tz;
 711             f[j_coord_offset+DIM*0+XX] -= tx;
 712             f[j_coord_offset+DIM*0+YY] -= ty;
 713             f[j_coord_offset+DIM*0+ZZ] -= tz;
 714
 715             /* Inner loop uses 160 flops */
 716         }
 717         /* End of innermost loop */
 718
 719         tx = ty = tz = 0;
 720         f[i_coord_offset+DIM*0+XX] += fix0;
 721         f[i_coord_offset+DIM*0+YY] += fiy0;
 722         f[i_coord_offset+DIM*0+ZZ] += fiz0;
 723         tx                         += fix0;
 724         ty                         += fiy0;
 725         tz                         += fiz0;
 726         f[i_coord_offset+DIM*1+XX] += fix1;
 727         f[i_coord_offset+DIM*1+YY] += fiy1;
 728         f[i_coord_offset+DIM*1+ZZ] += fiz1;
 729         tx                         += fix1;
 730         ty                         += fiy1;
 731         tz                         += fiz1;
 732         f[i_coord_offset+DIM*2+XX] += fix2;
 733         f[i_coord_offset+DIM*2+YY] += fiy2;
 734         f[i_coord_offset+DIM*2+ZZ] += fiz2;
 735         tx                         += fix2;
 736         ty                         += fiy2;
 737         tz                         += fiz2;
 738         f[i_coord_offset+DIM*3+XX] += fix3;
 739         f[i_coord_offset+DIM*3+YY] += fiy3;
 740         f[i_coord_offset+DIM*3+ZZ] += fiz3;
 741         tx                         += fix3;
 742         ty                         += fiy3;
 743         tz                         += fiz3;
 744         fshift[i_shift_offset+XX]  += tx;
 745         fshift[i_shift_offset+YY]  += ty;
 746         fshift[i_shift_offset+ZZ]  += tz;
 747
 748         /* Increment number of inner iterations */
 749         inneriter                  += j_index_end - j_index_start;
 750
 751         /* Outer loop uses 39 flops */
 752     }
 753
 754     /* Increment number of outer iterations */
 755     outeriter        += nri;
 756
 757     /* Update outer/inner flops */
 758
 759     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W4_F,outeriter*39 + inneriter*160);
 760 }