src/gromacs/gmxlib/nonbonded/nb_kernel_c/nb_kernel_ElecEw_VdwLJEw_GeomW3P1_c.c

   1 /*
   2  * This file is part of the GROMACS molecular simulation package.
   3  *
   4  * Copyright (c) 2012,2013,2014, by the GROMACS development team, led by
   5  * Mark Abraham, David van der Spoel, Berk Hess, and Erik Lindahl,
   6  * and including many others, as listed in the AUTHORS file in the
   7  * top-level source directory and at http://www.gromacs.org.
   8  *
   9  * GROMACS is free software; you can redistribute it and/or
  10  * modify it under the terms of the GNU Lesser General Public License
  11  * as published by the Free Software Foundation; either version 2.1
  12  * of the License, or (at your option) any later version.
  13  *
  14  * GROMACS is distributed in the hope that it will be useful,
  15  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  16  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  17  * Lesser General Public License for more details.
  18  *
  19  * You should have received a copy of the GNU Lesser General Public
  20  * License along with GROMACS; if not, see
  21  * http://www.gnu.org/licenses, or write to the Free Software Foundation,
  22  * Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02110-1301  USA.
  23  *
  24  * If you want to redistribute modifications to GROMACS, please
  25  * consider that scientific software is very special. Version
  26  * control is crucial - bugs must be traceable. We will be happy to
  27  * consider code for inclusion in the official distribution, but
  28  * derived work must not be called official GROMACS. Details are found
  29  * in the README & COPYING files - if they are missing, get the
  30  * official version at http://www.gromacs.org.
  31  *
  32  * To help us fund GROMACS development, we humbly ask that you cite
  33  * the research papers on the package. Check out http://www.gromacs.org.
  34  */
  35 /*
  36  * Note: this file was generated by the GROMACS c kernel generator.
  37  */
  38 #include "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_VdwLJEw_GeomW3P1_VF_c
  49  * Electrostatics interaction: Ewald
  50  * VdW interaction:            LJEwald
  51  * Geometry:                   Water3-Particle
  52  * Calculate force/pot:        PotentialAndForce
  53  */
  54 void
  55 nb_kernel_ElecEw_VdwLJEw_GeomW3P1_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              vdwjidx0;
  77     real             jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
  78     real             dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00,cexp1_00,cexp2_00;
  79     real             dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10,cexp1_10,cexp2_10;
  80     real             dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20,cexp1_20,cexp2_20;
  81     real             velec,felec,velecsum,facel,crf,krf,krf2;
  82     real             *charge;
  83     int              nvdwtype;
  84     real             rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,br,vvdwexp,sh_vdw_invrcut6;
  85     int              *vdwtype;
  86     real             *vdwparam;
  87     real             c6grid_00;
  88     real             c6grid_10;
  89     real             c6grid_20;
  90     real             ewclj,ewclj2,ewclj6,ewcljrsq,poly,exponent,sh_lj_ewald;
  91     real             *vdwgridparam;
  92     int              ewitab;
  93     real             ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace;
  94     real             *ewtab;
  95
  96     x                = xx[0];
  97     f                = ff[0];
  98
  99     nri              = nlist->nri;
 100     iinr             = nlist->iinr;
 101     jindex           = nlist->jindex;
 102     jjnr             = nlist->jjnr;
 103     shiftidx         = nlist->shift;
 104     gid              = nlist->gid;
 105     shiftvec         = fr->shift_vec[0];
 106     fshift           = fr->fshift[0];
 107     facel            = fr->epsfac;
 108     charge           = mdatoms->chargeA;
 109     nvdwtype         = fr->ntype;
 110     vdwparam         = fr->nbfp;
 111     vdwtype          = mdatoms->typeA;
 112     vdwgridparam     = fr->ljpme_c6grid;
 113     ewclj            = fr->ewaldcoeff_lj;
 114     sh_lj_ewald      = fr->ic->sh_lj_ewald;
 115     ewclj2           = ewclj*ewclj;
 116     ewclj6           = ewclj2*ewclj2*ewclj2;
 117
 118     sh_ewald         = fr->ic->sh_ewald;
 119     ewtab            = fr->ic->tabq_coul_FDV0;
 120     ewtabscale       = fr->ic->tabq_scale;
 121     ewtabhalfspace   = 0.5/ewtabscale;
 122
 123     /* Setup water-specific parameters */
 124     inr              = nlist->iinr[0];
 125     iq0              = facel*charge[inr+0];
 126     iq1              = facel*charge[inr+1];
 127     iq2              = facel*charge[inr+2];
 128     vdwioffset0      = 2*nvdwtype*vdwtype[inr+0];
 129
 130     outeriter        = 0;
 131     inneriter        = 0;
 132
 133     /* Start outer loop over neighborlists */
 134     for(iidx=0; iidx<nri; iidx++)
 135     {
 136         /* Load shift vector for this list */
 137         i_shift_offset   = DIM*shiftidx[iidx];
 138         shX              = shiftvec[i_shift_offset+XX];
 139         shY              = shiftvec[i_shift_offset+YY];
 140         shZ              = shiftvec[i_shift_offset+ZZ];
 141
 142         /* Load limits for loop over neighbors */
 143         j_index_start    = jindex[iidx];
 144         j_index_end      = jindex[iidx+1];
 145
 146         /* Get outer coordinate index */
 147         inr              = iinr[iidx];
 148         i_coord_offset   = DIM*inr;
 149
 150         /* Load i particle coords and add shift vector */
 151         ix0              = shX + x[i_coord_offset+DIM*0+XX];
 152         iy0              = shY + x[i_coord_offset+DIM*0+YY];
 153         iz0              = shZ + x[i_coord_offset+DIM*0+ZZ];
 154         ix1              = shX + x[i_coord_offset+DIM*1+XX];
 155         iy1              = shY + x[i_coord_offset+DIM*1+YY];
 156         iz1              = shZ + x[i_coord_offset+DIM*1+ZZ];
 157         ix2              = shX + x[i_coord_offset+DIM*2+XX];
 158         iy2              = shY + x[i_coord_offset+DIM*2+YY];
 159         iz2              = shZ + x[i_coord_offset+DIM*2+ZZ];
 160
 161         fix0             = 0.0;
 162         fiy0             = 0.0;
 163         fiz0             = 0.0;
 164         fix1             = 0.0;
 165         fiy1             = 0.0;
 166         fiz1             = 0.0;
 167         fix2             = 0.0;
 168         fiy2             = 0.0;
 169         fiz2             = 0.0;
 170
 171         /* Reset potential sums */
 172         velecsum         = 0.0;
 173         vvdwsum          = 0.0;
 174
 175         /* Start inner kernel loop */
 176         for(jidx=j_index_start; jidx<j_index_end; jidx++)
 177         {
 178             /* Get j neighbor index, and coordinate index */
 179             jnr              = jjnr[jidx];
 180             j_coord_offset   = DIM*jnr;
 181
 182             /* load j atom coordinates */
 183             jx0              = x[j_coord_offset+DIM*0+XX];
 184             jy0              = x[j_coord_offset+DIM*0+YY];
 185             jz0              = x[j_coord_offset+DIM*0+ZZ];
 186
 187             /* Calculate displacement vector */
 188             dx00             = ix0 - jx0;
 189             dy00             = iy0 - jy0;
 190             dz00             = iz0 - jz0;
 191             dx10             = ix1 - jx0;
 192             dy10             = iy1 - jy0;
 193             dz10             = iz1 - jz0;
 194             dx20             = ix2 - jx0;
 195             dy20             = iy2 - jy0;
 196             dz20             = iz2 - jz0;
 197
 198             /* Calculate squared distance and things based on it */
 199             rsq00            = dx00*dx00+dy00*dy00+dz00*dz00;
 200             rsq10            = dx10*dx10+dy10*dy10+dz10*dz10;
 201             rsq20            = dx20*dx20+dy20*dy20+dz20*dz20;
 202
 203             rinv00           = gmx_invsqrt(rsq00);
 204             rinv10           = gmx_invsqrt(rsq10);
 205             rinv20           = gmx_invsqrt(rsq20);
 206
 207             rinvsq00         = rinv00*rinv00;
 208             rinvsq10         = rinv10*rinv10;
 209             rinvsq20         = rinv20*rinv20;
 210
 211             /* Load parameters for j particles */
 212             jq0              = charge[jnr+0];
 213             vdwjidx0         = 2*vdwtype[jnr+0];
 214
 215             /**************************
 216              * CALCULATE INTERACTIONS *
 217              **************************/
 218
 219             r00              = rsq00*rinv00;
 220
 221             qq00             = iq0*jq0;
 222             c6_00            = vdwparam[vdwioffset0+vdwjidx0];
 223             c12_00           = vdwparam[vdwioffset0+vdwjidx0+1];
 224             c6grid_00        = vdwgridparam[vdwioffset0+vdwjidx0];
 225
 226             /* EWALD ELECTROSTATICS */
 227
 228             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 229             ewrt             = r00*ewtabscale;
 230             ewitab           = ewrt;
 231             eweps            = ewrt-ewitab;
 232             ewitab           = 4*ewitab;
 233             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 234             velec            = qq00*(rinv00-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 235             felec            = qq00*rinv00*(rinvsq00-felec);
 236
 237             rinvsix          = rinvsq00*rinvsq00*rinvsq00;
 238             ewcljrsq         = ewclj2*rsq00;
 239             exponent         = exp(-ewcljrsq);
 240             poly             = exponent*(1.0 + ewcljrsq + ewcljrsq*ewcljrsq*0.5);
 241             vvdw6            = (c6_00-c6grid_00*(1.0-poly))*rinvsix;
 242             vvdw12           = c12_00*rinvsix*rinvsix;
 243             vvdw             = vvdw12*(1.0/12.0) - vvdw6*(1.0/6.0);
 244             fvdw             = (vvdw12 - vvdw6 - c6grid_00*(1.0/6.0)*exponent*ewclj6)*rinvsq00;
 245
 246             /* Update potential sums from outer loop */
 247             velecsum        += velec;
 248             vvdwsum         += vvdw;
 249
 250             fscal            = felec+fvdw;
 251
 252             /* Calculate temporary vectorial force */
 253             tx               = fscal*dx00;
 254             ty               = fscal*dy00;
 255             tz               = fscal*dz00;
 256
 257             /* Update vectorial force */
 258             fix0            += tx;
 259             fiy0            += ty;
 260             fiz0            += tz;
 261             f[j_coord_offset+DIM*0+XX] -= tx;
 262             f[j_coord_offset+DIM*0+YY] -= ty;
 263             f[j_coord_offset+DIM*0+ZZ] -= tz;
 264
 265             /**************************
 266              * CALCULATE INTERACTIONS *
 267              **************************/
 268
 269             r10              = rsq10*rinv10;
 270
 271             qq10             = iq1*jq0;
 272
 273             /* EWALD ELECTROSTATICS */
 274
 275             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 276             ewrt             = r10*ewtabscale;
 277             ewitab           = ewrt;
 278             eweps            = ewrt-ewitab;
 279             ewitab           = 4*ewitab;
 280             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 281             velec            = qq10*(rinv10-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 282             felec            = qq10*rinv10*(rinvsq10-felec);
 283
 284             /* Update potential sums from outer loop */
 285             velecsum        += velec;
 286
 287             fscal            = felec;
 288
 289             /* Calculate temporary vectorial force */
 290             tx               = fscal*dx10;
 291             ty               = fscal*dy10;
 292             tz               = fscal*dz10;
 293
 294             /* Update vectorial force */
 295             fix1            += tx;
 296             fiy1            += ty;
 297             fiz1            += tz;
 298             f[j_coord_offset+DIM*0+XX] -= tx;
 299             f[j_coord_offset+DIM*0+YY] -= ty;
 300             f[j_coord_offset+DIM*0+ZZ] -= tz;
 301
 302             /**************************
 303              * CALCULATE INTERACTIONS *
 304              **************************/
 305
 306             r20              = rsq20*rinv20;
 307
 308             qq20             = iq2*jq0;
 309
 310             /* EWALD ELECTROSTATICS */
 311
 312             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 313             ewrt             = r20*ewtabscale;
 314             ewitab           = ewrt;
 315             eweps            = ewrt-ewitab;
 316             ewitab           = 4*ewitab;
 317             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 318             velec            = qq20*(rinv20-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 319             felec            = qq20*rinv20*(rinvsq20-felec);
 320
 321             /* Update potential sums from outer loop */
 322             velecsum        += velec;
 323
 324             fscal            = felec;
 325
 326             /* Calculate temporary vectorial force */
 327             tx               = fscal*dx20;
 328             ty               = fscal*dy20;
 329             tz               = fscal*dz20;
 330
 331             /* Update vectorial force */
 332             fix2            += tx;
 333             fiy2            += ty;
 334             fiz2            += tz;
 335             f[j_coord_offset+DIM*0+XX] -= tx;
 336             f[j_coord_offset+DIM*0+YY] -= ty;
 337             f[j_coord_offset+DIM*0+ZZ] -= tz;
 338
 339             /* Inner loop uses 149 flops */
 340         }
 341         /* End of innermost loop */
 342
 343         tx = ty = tz = 0;
 344         f[i_coord_offset+DIM*0+XX] += fix0;
 345         f[i_coord_offset+DIM*0+YY] += fiy0;
 346         f[i_coord_offset+DIM*0+ZZ] += fiz0;
 347         tx                         += fix0;
 348         ty                         += fiy0;
 349         tz                         += fiz0;
 350         f[i_coord_offset+DIM*1+XX] += fix1;
 351         f[i_coord_offset+DIM*1+YY] += fiy1;
 352         f[i_coord_offset+DIM*1+ZZ] += fiz1;
 353         tx                         += fix1;
 354         ty                         += fiy1;
 355         tz                         += fiz1;
 356         f[i_coord_offset+DIM*2+XX] += fix2;
 357         f[i_coord_offset+DIM*2+YY] += fiy2;
 358         f[i_coord_offset+DIM*2+ZZ] += fiz2;
 359         tx                         += fix2;
 360         ty                         += fiy2;
 361         tz                         += fiz2;
 362         fshift[i_shift_offset+XX]  += tx;
 363         fshift[i_shift_offset+YY]  += ty;
 364         fshift[i_shift_offset+ZZ]  += tz;
 365
 366         ggid                        = gid[iidx];
 367         /* Update potential energies */
 368         kernel_data->energygrp_elec[ggid] += velecsum;
 369         kernel_data->energygrp_vdw[ggid] += vvdwsum;
 370
 371         /* Increment number of inner iterations */
 372         inneriter                  += j_index_end - j_index_start;
 373
 374         /* Outer loop uses 32 flops */
 375     }
 376
 377     /* Increment number of outer iterations */
 378     outeriter        += nri;
 379
 380     /* Update outer/inner flops */
 381
 382     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W3_VF,outeriter*32 + inneriter*149);
 383 }
 384 /*
 385  * Gromacs nonbonded kernel:   nb_kernel_ElecEw_VdwLJEw_GeomW3P1_F_c
 386  * Electrostatics interaction: Ewald
 387  * VdW interaction:            LJEwald
 388  * Geometry:                   Water3-Particle
 389  * Calculate force/pot:        Force
 390  */
 391 void
 392 nb_kernel_ElecEw_VdwLJEw_GeomW3P1_F_c
 393                     (t_nblist                    * gmx_restrict       nlist,
 394                      rvec                        * gmx_restrict          xx,
 395                      rvec                        * gmx_restrict          ff,
 396                      t_forcerec                  * gmx_restrict          fr,
 397                      t_mdatoms                   * gmx_restrict     mdatoms,
 398                      nb_kernel_data_t gmx_unused * gmx_restrict kernel_data,
 399                      t_nrnb                      * gmx_restrict        nrnb)
 400 {
 401     int              i_shift_offset,i_coord_offset,j_coord_offset;
 402     int              j_index_start,j_index_end;
 403     int              nri,inr,ggid,iidx,jidx,jnr,outeriter,inneriter;
 404     real             shX,shY,shZ,tx,ty,tz,fscal,rcutoff,rcutoff2;
 405     int              *iinr,*jindex,*jjnr,*shiftidx,*gid;
 406     real             *shiftvec,*fshift,*x,*f;
 407     int              vdwioffset0;
 408     real             ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
 409     int              vdwioffset1;
 410     real             ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
 411     int              vdwioffset2;
 412     real             ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
 413     int              vdwjidx0;
 414     real             jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
 415     real             dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00,cexp1_00,cexp2_00;
 416     real             dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10,cexp1_10,cexp2_10;
 417     real             dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20,cexp1_20,cexp2_20;
 418     real             velec,felec,velecsum,facel,crf,krf,krf2;
 419     real             *charge;
 420     int              nvdwtype;
 421     real             rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,br,vvdwexp,sh_vdw_invrcut6;
 422     int              *vdwtype;
 423     real             *vdwparam;
 424     real             c6grid_00;
 425     real             c6grid_10;
 426     real             c6grid_20;
 427     real             ewclj,ewclj2,ewclj6,ewcljrsq,poly,exponent,sh_lj_ewald;
 428     real             *vdwgridparam;
 429     int              ewitab;
 430     real             ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace;
 431     real             *ewtab;
 432
 433     x                = xx[0];
 434     f                = ff[0];
 435
 436     nri              = nlist->nri;
 437     iinr             = nlist->iinr;
 438     jindex           = nlist->jindex;
 439     jjnr             = nlist->jjnr;
 440     shiftidx         = nlist->shift;
 441     gid              = nlist->gid;
 442     shiftvec         = fr->shift_vec[0];
 443     fshift           = fr->fshift[0];
 444     facel            = fr->epsfac;
 445     charge           = mdatoms->chargeA;
 446     nvdwtype         = fr->ntype;
 447     vdwparam         = fr->nbfp;
 448     vdwtype          = mdatoms->typeA;
 449     vdwgridparam     = fr->ljpme_c6grid;
 450     ewclj            = fr->ewaldcoeff_lj;
 451     sh_lj_ewald      = fr->ic->sh_lj_ewald;
 452     ewclj2           = ewclj*ewclj;
 453     ewclj6           = ewclj2*ewclj2*ewclj2;
 454
 455     sh_ewald         = fr->ic->sh_ewald;
 456     ewtab            = fr->ic->tabq_coul_F;
 457     ewtabscale       = fr->ic->tabq_scale;
 458     ewtabhalfspace   = 0.5/ewtabscale;
 459
 460     /* Setup water-specific parameters */
 461     inr              = nlist->iinr[0];
 462     iq0              = facel*charge[inr+0];
 463     iq1              = facel*charge[inr+1];
 464     iq2              = facel*charge[inr+2];
 465     vdwioffset0      = 2*nvdwtype*vdwtype[inr+0];
 466
 467     outeriter        = 0;
 468     inneriter        = 0;
 469
 470     /* Start outer loop over neighborlists */
 471     for(iidx=0; iidx<nri; iidx++)
 472     {
 473         /* Load shift vector for this list */
 474         i_shift_offset   = DIM*shiftidx[iidx];
 475         shX              = shiftvec[i_shift_offset+XX];
 476         shY              = shiftvec[i_shift_offset+YY];
 477         shZ              = shiftvec[i_shift_offset+ZZ];
 478
 479         /* Load limits for loop over neighbors */
 480         j_index_start    = jindex[iidx];
 481         j_index_end      = jindex[iidx+1];
 482
 483         /* Get outer coordinate index */
 484         inr              = iinr[iidx];
 485         i_coord_offset   = DIM*inr;
 486
 487         /* Load i particle coords and add shift vector */
 488         ix0              = shX + x[i_coord_offset+DIM*0+XX];
 489         iy0              = shY + x[i_coord_offset+DIM*0+YY];
 490         iz0              = shZ + x[i_coord_offset+DIM*0+ZZ];
 491         ix1              = shX + x[i_coord_offset+DIM*1+XX];
 492         iy1              = shY + x[i_coord_offset+DIM*1+YY];
 493         iz1              = shZ + x[i_coord_offset+DIM*1+ZZ];
 494         ix2              = shX + x[i_coord_offset+DIM*2+XX];
 495         iy2              = shY + x[i_coord_offset+DIM*2+YY];
 496         iz2              = shZ + x[i_coord_offset+DIM*2+ZZ];
 497
 498         fix0             = 0.0;
 499         fiy0             = 0.0;
 500         fiz0             = 0.0;
 501         fix1             = 0.0;
 502         fiy1             = 0.0;
 503         fiz1             = 0.0;
 504         fix2             = 0.0;
 505         fiy2             = 0.0;
 506         fiz2             = 0.0;
 507
 508         /* Start inner kernel loop */
 509         for(jidx=j_index_start; jidx<j_index_end; jidx++)
 510         {
 511             /* Get j neighbor index, and coordinate index */
 512             jnr              = jjnr[jidx];
 513             j_coord_offset   = DIM*jnr;
 514
 515             /* load j atom coordinates */
 516             jx0              = x[j_coord_offset+DIM*0+XX];
 517             jy0              = x[j_coord_offset+DIM*0+YY];
 518             jz0              = x[j_coord_offset+DIM*0+ZZ];
 519
 520             /* Calculate displacement vector */
 521             dx00             = ix0 - jx0;
 522             dy00             = iy0 - jy0;
 523             dz00             = iz0 - jz0;
 524             dx10             = ix1 - jx0;
 525             dy10             = iy1 - jy0;
 526             dz10             = iz1 - jz0;
 527             dx20             = ix2 - jx0;
 528             dy20             = iy2 - jy0;
 529             dz20             = iz2 - jz0;
 530
 531             /* Calculate squared distance and things based on it */
 532             rsq00            = dx00*dx00+dy00*dy00+dz00*dz00;
 533             rsq10            = dx10*dx10+dy10*dy10+dz10*dz10;
 534             rsq20            = dx20*dx20+dy20*dy20+dz20*dz20;
 535
 536             rinv00           = gmx_invsqrt(rsq00);
 537             rinv10           = gmx_invsqrt(rsq10);
 538             rinv20           = gmx_invsqrt(rsq20);
 539
 540             rinvsq00         = rinv00*rinv00;
 541             rinvsq10         = rinv10*rinv10;
 542             rinvsq20         = rinv20*rinv20;
 543
 544             /* Load parameters for j particles */
 545             jq0              = charge[jnr+0];
 546             vdwjidx0         = 2*vdwtype[jnr+0];
 547
 548             /**************************
 549              * CALCULATE INTERACTIONS *
 550              **************************/
 551
 552             r00              = rsq00*rinv00;
 553
 554             qq00             = iq0*jq0;
 555             c6_00            = vdwparam[vdwioffset0+vdwjidx0];
 556             c12_00           = vdwparam[vdwioffset0+vdwjidx0+1];
 557             c6grid_00        = vdwgridparam[vdwioffset0+vdwjidx0];
 558
 559             /* EWALD ELECTROSTATICS */
 560
 561             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 562             ewrt             = r00*ewtabscale;
 563             ewitab           = ewrt;
 564             eweps            = ewrt-ewitab;
 565             felec            = (1.0-eweps)*ewtab[ewitab]+eweps*ewtab[ewitab+1];
 566             felec            = qq00*rinv00*(rinvsq00-felec);
 567
 568             rinvsix          = rinvsq00*rinvsq00*rinvsq00;
 569             ewcljrsq         = ewclj2*rsq00;
 570             exponent         = exp(-ewcljrsq);
 571             poly             = exponent*(1.0 + ewcljrsq + ewcljrsq*ewcljrsq*0.5);
 572             fvdw             = (((c12_00*rinvsix - c6_00 + c6grid_00*(1.0-poly))*rinvsix) - c6grid_00*(1.0/6.0)*exponent*ewclj6)*rinvsq00;
 573
 574             fscal            = felec+fvdw;
 575
 576             /* Calculate temporary vectorial force */
 577             tx               = fscal*dx00;
 578             ty               = fscal*dy00;
 579             tz               = fscal*dz00;
 580
 581             /* Update vectorial force */
 582             fix0            += tx;
 583             fiy0            += ty;
 584             fiz0            += tz;
 585             f[j_coord_offset+DIM*0+XX] -= tx;
 586             f[j_coord_offset+DIM*0+YY] -= ty;
 587             f[j_coord_offset+DIM*0+ZZ] -= tz;
 588
 589             /**************************
 590              * CALCULATE INTERACTIONS *
 591              **************************/
 592
 593             r10              = rsq10*rinv10;
 594
 595             qq10             = iq1*jq0;
 596
 597             /* EWALD ELECTROSTATICS */
 598
 599             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 600             ewrt             = r10*ewtabscale;
 601             ewitab           = ewrt;
 602             eweps            = ewrt-ewitab;
 603             felec            = (1.0-eweps)*ewtab[ewitab]+eweps*ewtab[ewitab+1];
 604             felec            = qq10*rinv10*(rinvsq10-felec);
 605
 606             fscal            = felec;
 607
 608             /* Calculate temporary vectorial force */
 609             tx               = fscal*dx10;
 610             ty               = fscal*dy10;
 611             tz               = fscal*dz10;
 612
 613             /* Update vectorial force */
 614             fix1            += tx;
 615             fiy1            += ty;
 616             fiz1            += tz;
 617             f[j_coord_offset+DIM*0+XX] -= tx;
 618             f[j_coord_offset+DIM*0+YY] -= ty;
 619             f[j_coord_offset+DIM*0+ZZ] -= tz;
 620
 621             /**************************
 622              * CALCULATE INTERACTIONS *
 623              **************************/
 624
 625             r20              = rsq20*rinv20;
 626
 627             qq20             = iq2*jq0;
 628
 629             /* EWALD ELECTROSTATICS */
 630
 631             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 632             ewrt             = r20*ewtabscale;
 633             ewitab           = ewrt;
 634             eweps            = ewrt-ewitab;
 635             felec            = (1.0-eweps)*ewtab[ewitab]+eweps*ewtab[ewitab+1];
 636             felec            = qq20*rinv20*(rinvsq20-felec);
 637
 638             fscal            = felec;
 639
 640             /* Calculate temporary vectorial force */
 641             tx               = fscal*dx20;
 642             ty               = fscal*dy20;
 643             tz               = fscal*dz20;
 644
 645             /* Update vectorial force */
 646             fix2            += tx;
 647             fiy2            += ty;
 648             fiz2            += tz;
 649             f[j_coord_offset+DIM*0+XX] -= tx;
 650             f[j_coord_offset+DIM*0+YY] -= ty;
 651             f[j_coord_offset+DIM*0+ZZ] -= tz;
 652
 653             /* Inner loop uses 123 flops */
 654         }
 655         /* End of innermost loop */
 656
 657         tx = ty = tz = 0;
 658         f[i_coord_offset+DIM*0+XX] += fix0;
 659         f[i_coord_offset+DIM*0+YY] += fiy0;
 660         f[i_coord_offset+DIM*0+ZZ] += fiz0;
 661         tx                         += fix0;
 662         ty                         += fiy0;
 663         tz                         += fiz0;
 664         f[i_coord_offset+DIM*1+XX] += fix1;
 665         f[i_coord_offset+DIM*1+YY] += fiy1;
 666         f[i_coord_offset+DIM*1+ZZ] += fiz1;
 667         tx                         += fix1;
 668         ty                         += fiy1;
 669         tz                         += fiz1;
 670         f[i_coord_offset+DIM*2+XX] += fix2;
 671         f[i_coord_offset+DIM*2+YY] += fiy2;
 672         f[i_coord_offset+DIM*2+ZZ] += fiz2;
 673         tx                         += fix2;
 674         ty                         += fiy2;
 675         tz                         += fiz2;
 676         fshift[i_shift_offset+XX]  += tx;
 677         fshift[i_shift_offset+YY]  += ty;
 678         fshift[i_shift_offset+ZZ]  += tz;
 679
 680         /* Increment number of inner iterations */
 681         inneriter                  += j_index_end - j_index_start;
 682
 683         /* Outer loop uses 30 flops */
 684     }
 685
 686     /* Increment number of outer iterations */
 687     outeriter        += nri;
 688
 689     /* Update outer/inner flops */
 690
 691     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W3_F,outeriter*30 + inneriter*123);
 692 }