src/gromacs/gmxlib/nonbonded/nb_kernel_c/nb_kernel_ElecEw_VdwLJ_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 "types/simple.h"
  44 #include "gromacs/math/vec.h"
  45 #include "nrnb.h"
  46
  47 /*
  48  * Gromacs nonbonded kernel:   nb_kernel_ElecEw_VdwLJ_GeomW4P1_VF_c
  49  * Electrostatics interaction: Ewald
  50  * VdW interaction:            LennardJones
  51  * Geometry:                   Water4-Particle
  52  * Calculate force/pot:        PotentialAndForce
  53  */
  54 void
  55 nb_kernel_ElecEw_VdwLJ_GeomW4P1_VF_c
  56                     (t_nblist                    * gmx_restrict       nlist,
  57                      rvec                        * gmx_restrict          xx,
  58                      rvec                        * gmx_restrict          ff,
  59                      t_forcerec                  * gmx_restrict          fr,
  60                      t_mdatoms                   * gmx_restrict     mdatoms,
  61                      nb_kernel_data_t gmx_unused * gmx_restrict kernel_data,
  62                      t_nrnb                      * gmx_restrict        nrnb)
  63 {
  64     int              i_shift_offset,i_coord_offset,j_coord_offset;
  65     int              j_index_start,j_index_end;
  66     int              nri,inr,ggid,iidx,jidx,jnr,outeriter,inneriter;
  67     real             shX,shY,shZ,tx,ty,tz,fscal,rcutoff,rcutoff2;
  68     int              *iinr,*jindex,*jjnr,*shiftidx,*gid;
  69     real             *shiftvec,*fshift,*x,*f;
  70     int              vdwioffset0;
  71     real             ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
  72     int              vdwioffset1;
  73     real             ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
  74     int              vdwioffset2;
  75     real             ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
  76     int              vdwioffset3;
  77     real             ix3,iy3,iz3,fix3,fiy3,fiz3,iq3,isai3;
  78     int              vdwjidx0;
  79     real             jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
  80     real             dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00,cexp1_00,cexp2_00;
  81     real             dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10,cexp1_10,cexp2_10;
  82     real             dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20,cexp1_20,cexp2_20;
  83     real             dx30,dy30,dz30,rsq30,rinv30,rinvsq30,r30,qq30,c6_30,c12_30,cexp1_30,cexp2_30;
  84     real             velec,felec,velecsum,facel,crf,krf,krf2;
  85     real             *charge;
  86     int              nvdwtype;
  87     real             rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,br,vvdwexp,sh_vdw_invrcut6;
  88     int              *vdwtype;
  89     real             *vdwparam;
  90     int              ewitab;
  91     real             ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace;
  92     real             *ewtab;
  93
  94     x                = xx[0];
  95     f                = ff[0];
  96
  97     nri              = nlist->nri;
  98     iinr             = nlist->iinr;
  99     jindex           = nlist->jindex;
 100     jjnr             = nlist->jjnr;
 101     shiftidx         = nlist->shift;
 102     gid              = nlist->gid;
 103     shiftvec         = fr->shift_vec[0];
 104     fshift           = fr->fshift[0];
 105     facel            = fr->epsfac;
 106     charge           = mdatoms->chargeA;
 107     nvdwtype         = fr->ntype;
 108     vdwparam         = fr->nbfp;
 109     vdwtype          = mdatoms->typeA;
 110
 111     sh_ewald         = fr->ic->sh_ewald;
 112     ewtab            = fr->ic->tabq_coul_FDV0;
 113     ewtabscale       = fr->ic->tabq_scale;
 114     ewtabhalfspace   = 0.5/ewtabscale;
 115
 116     /* Setup water-specific parameters */
 117     inr              = nlist->iinr[0];
 118     iq1              = facel*charge[inr+1];
 119     iq2              = facel*charge[inr+2];
 120     iq3              = facel*charge[inr+3];
 121     vdwioffset0      = 2*nvdwtype*vdwtype[inr+0];
 122
 123     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             rinv10           = gmx_invsqrt(rsq10);
 207             rinv20           = gmx_invsqrt(rsq20);
 208             rinv30           = gmx_invsqrt(rsq30);
 209
 210             rinvsq00         = 1.0/rsq00;
 211             rinvsq10         = rinv10*rinv10;
 212             rinvsq20         = rinv20*rinv20;
 213             rinvsq30         = rinv30*rinv30;
 214
 215             /* Load parameters for j particles */
 216             jq0              = charge[jnr+0];
 217             vdwjidx0         = 2*vdwtype[jnr+0];
 218
 219             /**************************
 220              * CALCULATE INTERACTIONS *
 221              **************************/
 222
 223             c6_00            = vdwparam[vdwioffset0+vdwjidx0];
 224             c12_00           = vdwparam[vdwioffset0+vdwjidx0+1];
 225
 226             /* LENNARD-JONES DISPERSION/REPULSION */
 227
 228             rinvsix          = rinvsq00*rinvsq00*rinvsq00;
 229             vvdw6            = c6_00*rinvsix;
 230             vvdw12           = c12_00*rinvsix*rinvsix;
 231             vvdw             = vvdw12*(1.0/12.0) - vvdw6*(1.0/6.0);
 232             fvdw             = (vvdw12-vvdw6)*rinvsq00;
 233
 234             /* Update potential sums from outer loop */
 235             vvdwsum         += vvdw;
 236
 237             fscal            = fvdw;
 238
 239             /* Calculate temporary vectorial force */
 240             tx               = fscal*dx00;
 241             ty               = fscal*dy00;
 242             tz               = fscal*dz00;
 243
 244             /* Update vectorial force */
 245             fix0            += tx;
 246             fiy0            += ty;
 247             fiz0            += tz;
 248             f[j_coord_offset+DIM*0+XX] -= tx;
 249             f[j_coord_offset+DIM*0+YY] -= ty;
 250             f[j_coord_offset+DIM*0+ZZ] -= tz;
 251
 252             /**************************
 253              * CALCULATE INTERACTIONS *
 254              **************************/
 255
 256             r10              = rsq10*rinv10;
 257
 258             qq10             = iq1*jq0;
 259
 260             /* EWALD ELECTROSTATICS */
 261
 262             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 263             ewrt             = r10*ewtabscale;
 264             ewitab           = ewrt;
 265             eweps            = ewrt-ewitab;
 266             ewitab           = 4*ewitab;
 267             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 268             velec            = qq10*(rinv10-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 269             felec            = qq10*rinv10*(rinvsq10-felec);
 270
 271             /* Update potential sums from outer loop */
 272             velecsum        += velec;
 273
 274             fscal            = felec;
 275
 276             /* Calculate temporary vectorial force */
 277             tx               = fscal*dx10;
 278             ty               = fscal*dy10;
 279             tz               = fscal*dz10;
 280
 281             /* Update vectorial force */
 282             fix1            += tx;
 283             fiy1            += ty;
 284             fiz1            += tz;
 285             f[j_coord_offset+DIM*0+XX] -= tx;
 286             f[j_coord_offset+DIM*0+YY] -= ty;
 287             f[j_coord_offset+DIM*0+ZZ] -= tz;
 288
 289             /**************************
 290              * CALCULATE INTERACTIONS *
 291              **************************/
 292
 293             r20              = rsq20*rinv20;
 294
 295             qq20             = iq2*jq0;
 296
 297             /* EWALD ELECTROSTATICS */
 298
 299             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 300             ewrt             = r20*ewtabscale;
 301             ewitab           = ewrt;
 302             eweps            = ewrt-ewitab;
 303             ewitab           = 4*ewitab;
 304             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 305             velec            = qq20*(rinv20-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 306             felec            = qq20*rinv20*(rinvsq20-felec);
 307
 308             /* Update potential sums from outer loop */
 309             velecsum        += velec;
 310
 311             fscal            = felec;
 312
 313             /* Calculate temporary vectorial force */
 314             tx               = fscal*dx20;
 315             ty               = fscal*dy20;
 316             tz               = fscal*dz20;
 317
 318             /* Update vectorial force */
 319             fix2            += tx;
 320             fiy2            += ty;
 321             fiz2            += tz;
 322             f[j_coord_offset+DIM*0+XX] -= tx;
 323             f[j_coord_offset+DIM*0+YY] -= ty;
 324             f[j_coord_offset+DIM*0+ZZ] -= tz;
 325
 326             /**************************
 327              * CALCULATE INTERACTIONS *
 328              **************************/
 329
 330             r30              = rsq30*rinv30;
 331
 332             qq30             = iq3*jq0;
 333
 334             /* EWALD ELECTROSTATICS */
 335
 336             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 337             ewrt             = r30*ewtabscale;
 338             ewitab           = ewrt;
 339             eweps            = ewrt-ewitab;
 340             ewitab           = 4*ewitab;
 341             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 342             velec            = qq30*(rinv30-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 343             felec            = qq30*rinv30*(rinvsq30-felec);
 344
 345             /* Update potential sums from outer loop */
 346             velecsum        += velec;
 347
 348             fscal            = felec;
 349
 350             /* Calculate temporary vectorial force */
 351             tx               = fscal*dx30;
 352             ty               = fscal*dy30;
 353             tz               = fscal*dz30;
 354
 355             /* Update vectorial force */
 356             fix3            += tx;
 357             fiy3            += ty;
 358             fiz3            += tz;
 359             f[j_coord_offset+DIM*0+XX] -= tx;
 360             f[j_coord_offset+DIM*0+YY] -= ty;
 361             f[j_coord_offset+DIM*0+ZZ] -= tz;
 362
 363             /* Inner loop uses 155 flops */
 364         }
 365         /* End of innermost loop */
 366
 367         tx = ty = tz = 0;
 368         f[i_coord_offset+DIM*0+XX] += fix0;
 369         f[i_coord_offset+DIM*0+YY] += fiy0;
 370         f[i_coord_offset+DIM*0+ZZ] += fiz0;
 371         tx                         += fix0;
 372         ty                         += fiy0;
 373         tz                         += fiz0;
 374         f[i_coord_offset+DIM*1+XX] += fix1;
 375         f[i_coord_offset+DIM*1+YY] += fiy1;
 376         f[i_coord_offset+DIM*1+ZZ] += fiz1;
 377         tx                         += fix1;
 378         ty                         += fiy1;
 379         tz                         += fiz1;
 380         f[i_coord_offset+DIM*2+XX] += fix2;
 381         f[i_coord_offset+DIM*2+YY] += fiy2;
 382         f[i_coord_offset+DIM*2+ZZ] += fiz2;
 383         tx                         += fix2;
 384         ty                         += fiy2;
 385         tz                         += fiz2;
 386         f[i_coord_offset+DIM*3+XX] += fix3;
 387         f[i_coord_offset+DIM*3+YY] += fiy3;
 388         f[i_coord_offset+DIM*3+ZZ] += fiz3;
 389         tx                         += fix3;
 390         ty                         += fiy3;
 391         tz                         += fiz3;
 392         fshift[i_shift_offset+XX]  += tx;
 393         fshift[i_shift_offset+YY]  += ty;
 394         fshift[i_shift_offset+ZZ]  += tz;
 395
 396         ggid                        = gid[iidx];
 397         /* Update potential energies */
 398         kernel_data->energygrp_elec[ggid] += velecsum;
 399         kernel_data->energygrp_vdw[ggid] += vvdwsum;
 400
 401         /* Increment number of inner iterations */
 402         inneriter                  += j_index_end - j_index_start;
 403
 404         /* Outer loop uses 41 flops */
 405     }
 406
 407     /* Increment number of outer iterations */
 408     outeriter        += nri;
 409
 410     /* Update outer/inner flops */
 411
 412     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W4_VF,outeriter*41 + inneriter*155);
 413 }
 414 /*
 415  * Gromacs nonbonded kernel:   nb_kernel_ElecEw_VdwLJ_GeomW4P1_F_c
 416  * Electrostatics interaction: Ewald
 417  * VdW interaction:            LennardJones
 418  * Geometry:                   Water4-Particle
 419  * Calculate force/pot:        Force
 420  */
 421 void
 422 nb_kernel_ElecEw_VdwLJ_GeomW4P1_F_c
 423                     (t_nblist                    * gmx_restrict       nlist,
 424                      rvec                        * gmx_restrict          xx,
 425                      rvec                        * gmx_restrict          ff,
 426                      t_forcerec                  * gmx_restrict          fr,
 427                      t_mdatoms                   * gmx_restrict     mdatoms,
 428                      nb_kernel_data_t gmx_unused * gmx_restrict kernel_data,
 429                      t_nrnb                      * gmx_restrict        nrnb)
 430 {
 431     int              i_shift_offset,i_coord_offset,j_coord_offset;
 432     int              j_index_start,j_index_end;
 433     int              nri,inr,ggid,iidx,jidx,jnr,outeriter,inneriter;
 434     real             shX,shY,shZ,tx,ty,tz,fscal,rcutoff,rcutoff2;
 435     int              *iinr,*jindex,*jjnr,*shiftidx,*gid;
 436     real             *shiftvec,*fshift,*x,*f;
 437     int              vdwioffset0;
 438     real             ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
 439     int              vdwioffset1;
 440     real             ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
 441     int              vdwioffset2;
 442     real             ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
 443     int              vdwioffset3;
 444     real             ix3,iy3,iz3,fix3,fiy3,fiz3,iq3,isai3;
 445     int              vdwjidx0;
 446     real             jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
 447     real             dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00,cexp1_00,cexp2_00;
 448     real             dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10,cexp1_10,cexp2_10;
 449     real             dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20,cexp1_20,cexp2_20;
 450     real             dx30,dy30,dz30,rsq30,rinv30,rinvsq30,r30,qq30,c6_30,c12_30,cexp1_30,cexp2_30;
 451     real             velec,felec,velecsum,facel,crf,krf,krf2;
 452     real             *charge;
 453     int              nvdwtype;
 454     real             rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,br,vvdwexp,sh_vdw_invrcut6;
 455     int              *vdwtype;
 456     real             *vdwparam;
 457     int              ewitab;
 458     real             ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace;
 459     real             *ewtab;
 460
 461     x                = xx[0];
 462     f                = ff[0];
 463
 464     nri              = nlist->nri;
 465     iinr             = nlist->iinr;
 466     jindex           = nlist->jindex;
 467     jjnr             = nlist->jjnr;
 468     shiftidx         = nlist->shift;
 469     gid              = nlist->gid;
 470     shiftvec         = fr->shift_vec[0];
 471     fshift           = fr->fshift[0];
 472     facel            = fr->epsfac;
 473     charge           = mdatoms->chargeA;
 474     nvdwtype         = fr->ntype;
 475     vdwparam         = fr->nbfp;
 476     vdwtype          = mdatoms->typeA;
 477
 478     sh_ewald         = fr->ic->sh_ewald;
 479     ewtab            = fr->ic->tabq_coul_F;
 480     ewtabscale       = fr->ic->tabq_scale;
 481     ewtabhalfspace   = 0.5/ewtabscale;
 482
 483     /* Setup water-specific parameters */
 484     inr              = nlist->iinr[0];
 485     iq1              = facel*charge[inr+1];
 486     iq2              = facel*charge[inr+2];
 487     iq3              = facel*charge[inr+3];
 488     vdwioffset0      = 2*nvdwtype*vdwtype[inr+0];
 489
 490     outeriter        = 0;
 491     inneriter        = 0;
 492
 493     /* Start outer loop over neighborlists */
 494     for(iidx=0; iidx<nri; iidx++)
 495     {
 496         /* Load shift vector for this list */
 497         i_shift_offset   = DIM*shiftidx[iidx];
 498         shX              = shiftvec[i_shift_offset+XX];
 499         shY              = shiftvec[i_shift_offset+YY];
 500         shZ              = shiftvec[i_shift_offset+ZZ];
 501
 502         /* Load limits for loop over neighbors */
 503         j_index_start    = jindex[iidx];
 504         j_index_end      = jindex[iidx+1];
 505
 506         /* Get outer coordinate index */
 507         inr              = iinr[iidx];
 508         i_coord_offset   = DIM*inr;
 509
 510         /* Load i particle coords and add shift vector */
 511         ix0              = shX + x[i_coord_offset+DIM*0+XX];
 512         iy0              = shY + x[i_coord_offset+DIM*0+YY];
 513         iz0              = shZ + x[i_coord_offset+DIM*0+ZZ];
 514         ix1              = shX + x[i_coord_offset+DIM*1+XX];
 515         iy1              = shY + x[i_coord_offset+DIM*1+YY];
 516         iz1              = shZ + x[i_coord_offset+DIM*1+ZZ];
 517         ix2              = shX + x[i_coord_offset+DIM*2+XX];
 518         iy2              = shY + x[i_coord_offset+DIM*2+YY];
 519         iz2              = shZ + x[i_coord_offset+DIM*2+ZZ];
 520         ix3              = shX + x[i_coord_offset+DIM*3+XX];
 521         iy3              = shY + x[i_coord_offset+DIM*3+YY];
 522         iz3              = shZ + x[i_coord_offset+DIM*3+ZZ];
 523
 524         fix0             = 0.0;
 525         fiy0             = 0.0;
 526         fiz0             = 0.0;
 527         fix1             = 0.0;
 528         fiy1             = 0.0;
 529         fiz1             = 0.0;
 530         fix2             = 0.0;
 531         fiy2             = 0.0;
 532         fiz2             = 0.0;
 533         fix3             = 0.0;
 534         fiy3             = 0.0;
 535         fiz3             = 0.0;
 536
 537         /* Start inner kernel loop */
 538         for(jidx=j_index_start; jidx<j_index_end; jidx++)
 539         {
 540             /* Get j neighbor index, and coordinate index */
 541             jnr              = jjnr[jidx];
 542             j_coord_offset   = DIM*jnr;
 543
 544             /* load j atom coordinates */
 545             jx0              = x[j_coord_offset+DIM*0+XX];
 546             jy0              = x[j_coord_offset+DIM*0+YY];
 547             jz0              = x[j_coord_offset+DIM*0+ZZ];
 548
 549             /* Calculate displacement vector */
 550             dx00             = ix0 - jx0;
 551             dy00             = iy0 - jy0;
 552             dz00             = iz0 - jz0;
 553             dx10             = ix1 - jx0;
 554             dy10             = iy1 - jy0;
 555             dz10             = iz1 - jz0;
 556             dx20             = ix2 - jx0;
 557             dy20             = iy2 - jy0;
 558             dz20             = iz2 - jz0;
 559             dx30             = ix3 - jx0;
 560             dy30             = iy3 - jy0;
 561             dz30             = iz3 - jz0;
 562
 563             /* Calculate squared distance and things based on it */
 564             rsq00            = dx00*dx00+dy00*dy00+dz00*dz00;
 565             rsq10            = dx10*dx10+dy10*dy10+dz10*dz10;
 566             rsq20            = dx20*dx20+dy20*dy20+dz20*dz20;
 567             rsq30            = dx30*dx30+dy30*dy30+dz30*dz30;
 568
 569             rinv10           = gmx_invsqrt(rsq10);
 570             rinv20           = gmx_invsqrt(rsq20);
 571             rinv30           = gmx_invsqrt(rsq30);
 572
 573             rinvsq00         = 1.0/rsq00;
 574             rinvsq10         = rinv10*rinv10;
 575             rinvsq20         = rinv20*rinv20;
 576             rinvsq30         = rinv30*rinv30;
 577
 578             /* Load parameters for j particles */
 579             jq0              = charge[jnr+0];
 580             vdwjidx0         = 2*vdwtype[jnr+0];
 581
 582             /**************************
 583              * CALCULATE INTERACTIONS *
 584              **************************/
 585
 586             c6_00            = vdwparam[vdwioffset0+vdwjidx0];
 587             c12_00           = vdwparam[vdwioffset0+vdwjidx0+1];
 588
 589             /* LENNARD-JONES DISPERSION/REPULSION */
 590
 591             rinvsix          = rinvsq00*rinvsq00*rinvsq00;
 592             fvdw             = (c12_00*rinvsix-c6_00)*rinvsix*rinvsq00;
 593
 594             fscal            = fvdw;
 595
 596             /* Calculate temporary vectorial force */
 597             tx               = fscal*dx00;
 598             ty               = fscal*dy00;
 599             tz               = fscal*dz00;
 600
 601             /* Update vectorial force */
 602             fix0            += tx;
 603             fiy0            += ty;
 604             fiz0            += tz;
 605             f[j_coord_offset+DIM*0+XX] -= tx;
 606             f[j_coord_offset+DIM*0+YY] -= ty;
 607             f[j_coord_offset+DIM*0+ZZ] -= tz;
 608
 609             /**************************
 610              * CALCULATE INTERACTIONS *
 611              **************************/
 612
 613             r10              = rsq10*rinv10;
 614
 615             qq10             = iq1*jq0;
 616
 617             /* EWALD ELECTROSTATICS */
 618
 619             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 620             ewrt             = r10*ewtabscale;
 621             ewitab           = ewrt;
 622             eweps            = ewrt-ewitab;
 623             felec            = (1.0-eweps)*ewtab[ewitab]+eweps*ewtab[ewitab+1];
 624             felec            = qq10*rinv10*(rinvsq10-felec);
 625
 626             fscal            = felec;
 627
 628             /* Calculate temporary vectorial force */
 629             tx               = fscal*dx10;
 630             ty               = fscal*dy10;
 631             tz               = fscal*dz10;
 632
 633             /* Update vectorial force */
 634             fix1            += tx;
 635             fiy1            += ty;
 636             fiz1            += tz;
 637             f[j_coord_offset+DIM*0+XX] -= tx;
 638             f[j_coord_offset+DIM*0+YY] -= ty;
 639             f[j_coord_offset+DIM*0+ZZ] -= tz;
 640
 641             /**************************
 642              * CALCULATE INTERACTIONS *
 643              **************************/
 644
 645             r20              = rsq20*rinv20;
 646
 647             qq20             = iq2*jq0;
 648
 649             /* EWALD ELECTROSTATICS */
 650
 651             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 652             ewrt             = r20*ewtabscale;
 653             ewitab           = ewrt;
 654             eweps            = ewrt-ewitab;
 655             felec            = (1.0-eweps)*ewtab[ewitab]+eweps*ewtab[ewitab+1];
 656             felec            = qq20*rinv20*(rinvsq20-felec);
 657
 658             fscal            = felec;
 659
 660             /* Calculate temporary vectorial force */
 661             tx               = fscal*dx20;
 662             ty               = fscal*dy20;
 663             tz               = fscal*dz20;
 664
 665             /* Update vectorial force */
 666             fix2            += tx;
 667             fiy2            += ty;
 668             fiz2            += tz;
 669             f[j_coord_offset+DIM*0+XX] -= tx;
 670             f[j_coord_offset+DIM*0+YY] -= ty;
 671             f[j_coord_offset+DIM*0+ZZ] -= tz;
 672
 673             /**************************
 674              * CALCULATE INTERACTIONS *
 675              **************************/
 676
 677             r30              = rsq30*rinv30;
 678
 679             qq30             = iq3*jq0;
 680
 681             /* EWALD ELECTROSTATICS */
 682
 683             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 684             ewrt             = r30*ewtabscale;
 685             ewitab           = ewrt;
 686             eweps            = ewrt-ewitab;
 687             felec            = (1.0-eweps)*ewtab[ewitab]+eweps*ewtab[ewitab+1];
 688             felec            = qq30*rinv30*(rinvsq30-felec);
 689
 690             fscal            = felec;
 691
 692             /* Calculate temporary vectorial force */
 693             tx               = fscal*dx30;
 694             ty               = fscal*dy30;
 695             tz               = fscal*dz30;
 696
 697             /* Update vectorial force */
 698             fix3            += tx;
 699             fiy3            += ty;
 700             fiz3            += tz;
 701             f[j_coord_offset+DIM*0+XX] -= tx;
 702             f[j_coord_offset+DIM*0+YY] -= ty;
 703             f[j_coord_offset+DIM*0+ZZ] -= tz;
 704
 705             /* Inner loop uses 129 flops */
 706         }
 707         /* End of innermost loop */
 708
 709         tx = ty = tz = 0;
 710         f[i_coord_offset+DIM*0+XX] += fix0;
 711         f[i_coord_offset+DIM*0+YY] += fiy0;
 712         f[i_coord_offset+DIM*0+ZZ] += fiz0;
 713         tx                         += fix0;
 714         ty                         += fiy0;
 715         tz                         += fiz0;
 716         f[i_coord_offset+DIM*1+XX] += fix1;
 717         f[i_coord_offset+DIM*1+YY] += fiy1;
 718         f[i_coord_offset+DIM*1+ZZ] += fiz1;
 719         tx                         += fix1;
 720         ty                         += fiy1;
 721         tz                         += fiz1;
 722         f[i_coord_offset+DIM*2+XX] += fix2;
 723         f[i_coord_offset+DIM*2+YY] += fiy2;
 724         f[i_coord_offset+DIM*2+ZZ] += fiz2;
 725         tx                         += fix2;
 726         ty                         += fiy2;
 727         tz                         += fiz2;
 728         f[i_coord_offset+DIM*3+XX] += fix3;
 729         f[i_coord_offset+DIM*3+YY] += fiy3;
 730         f[i_coord_offset+DIM*3+ZZ] += fiz3;
 731         tx                         += fix3;
 732         ty                         += fiy3;
 733         tz                         += fiz3;
 734         fshift[i_shift_offset+XX]  += tx;
 735         fshift[i_shift_offset+YY]  += ty;
 736         fshift[i_shift_offset+ZZ]  += tz;
 737
 738         /* Increment number of inner iterations */
 739         inneriter                  += j_index_end - j_index_start;
 740
 741         /* Outer loop uses 39 flops */
 742     }
 743
 744     /* Increment number of outer iterations */
 745     outeriter        += nri;
 746
 747     /* Update outer/inner flops */
 748
 749     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W4_F,outeriter*39 + inneriter*129);
 750 }