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 #ifdef HAVE_CONFIG_H
  39 #include <config.h>
  40 #endif
  41
  42 #include <math.h>
  43
  44 #include "../nb_kernel.h"
  45 #include "types/simple.h"
  46 #include "gromacs/math/vec.h"
  47 #include "nrnb.h"
  48
  49 /*
  50  * Gromacs nonbonded kernel:   nb_kernel_ElecEw_VdwLJEw_GeomW3P1_VF_c
  51  * Electrostatics interaction: Ewald
  52  * VdW interaction:            LJEwald
  53  * Geometry:                   Water3-Particle
  54  * Calculate force/pot:        PotentialAndForce
  55  */
  56 void
  57 nb_kernel_ElecEw_VdwLJEw_GeomW3P1_VF_c
  58                     (t_nblist                    * gmx_restrict       nlist,
  59                      rvec                        * gmx_restrict          xx,
  60                      rvec                        * gmx_restrict          ff,
  61                      t_forcerec                  * gmx_restrict          fr,
  62                      t_mdatoms                   * gmx_restrict     mdatoms,
  63                      nb_kernel_data_t gmx_unused * gmx_restrict kernel_data,
  64                      t_nrnb                      * gmx_restrict        nrnb)
  65 {
  66     int              i_shift_offset,i_coord_offset,j_coord_offset;
  67     int              j_index_start,j_index_end;
  68     int              nri,inr,ggid,iidx,jidx,jnr,outeriter,inneriter;
  69     real             shX,shY,shZ,tx,ty,tz,fscal,rcutoff,rcutoff2;
  70     int              *iinr,*jindex,*jjnr,*shiftidx,*gid;
  71     real             *shiftvec,*fshift,*x,*f;
  72     int              vdwioffset0;
  73     real             ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
  74     int              vdwioffset1;
  75     real             ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
  76     int              vdwioffset2;
  77     real             ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
  78     int              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             velec,felec,velecsum,facel,crf,krf,krf2;
  84     real             *charge;
  85     int              nvdwtype;
  86     real             rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,br,vvdwexp,sh_vdw_invrcut6;
  87     int              *vdwtype;
  88     real             *vdwparam;
  89     real             c6grid_00;
  90     real             c6grid_10;
  91     real             c6grid_20;
  92     real             ewclj,ewclj2,ewclj6,ewcljrsq,poly,exponent,sh_lj_ewald;
  93     real             *vdwgridparam;
  94     int              ewitab;
  95     real             ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace;
  96     real             *ewtab;
  97
  98     x                = xx[0];
  99     f                = ff[0];
 100
 101     nri              = nlist->nri;
 102     iinr             = nlist->iinr;
 103     jindex           = nlist->jindex;
 104     jjnr             = nlist->jjnr;
 105     shiftidx         = nlist->shift;
 106     gid              = nlist->gid;
 107     shiftvec         = fr->shift_vec[0];
 108     fshift           = fr->fshift[0];
 109     facel            = fr->epsfac;
 110     charge           = mdatoms->chargeA;
 111     nvdwtype         = fr->ntype;
 112     vdwparam         = fr->nbfp;
 113     vdwtype          = mdatoms->typeA;
 114     vdwgridparam     = fr->ljpme_c6grid;
 115     ewclj            = fr->ewaldcoeff_lj;
 116     sh_lj_ewald      = fr->ic->sh_lj_ewald;
 117     ewclj2           = ewclj*ewclj;
 118     ewclj6           = ewclj2*ewclj2*ewclj2;
 119
 120     sh_ewald         = fr->ic->sh_ewald;
 121     ewtab            = fr->ic->tabq_coul_FDV0;
 122     ewtabscale       = fr->ic->tabq_scale;
 123     ewtabhalfspace   = 0.5/ewtabscale;
 124
 125     /* Setup water-specific parameters */
 126     inr              = nlist->iinr[0];
 127     iq0              = facel*charge[inr+0];
 128     iq1              = facel*charge[inr+1];
 129     iq2              = facel*charge[inr+2];
 130     vdwioffset0      = 2*nvdwtype*vdwtype[inr+0];
 131
 132     outeriter        = 0;
 133     inneriter        = 0;
 134
 135     /* Start outer loop over neighborlists */
 136     for(iidx=0; iidx<nri; iidx++)
 137     {
 138         /* Load shift vector for this list */
 139         i_shift_offset   = DIM*shiftidx[iidx];
 140         shX              = shiftvec[i_shift_offset+XX];
 141         shY              = shiftvec[i_shift_offset+YY];
 142         shZ              = shiftvec[i_shift_offset+ZZ];
 143
 144         /* Load limits for loop over neighbors */
 145         j_index_start    = jindex[iidx];
 146         j_index_end      = jindex[iidx+1];
 147
 148         /* Get outer coordinate index */
 149         inr              = iinr[iidx];
 150         i_coord_offset   = DIM*inr;
 151
 152         /* Load i particle coords and add shift vector */
 153         ix0              = shX + x[i_coord_offset+DIM*0+XX];
 154         iy0              = shY + x[i_coord_offset+DIM*0+YY];
 155         iz0              = shZ + x[i_coord_offset+DIM*0+ZZ];
 156         ix1              = shX + x[i_coord_offset+DIM*1+XX];
 157         iy1              = shY + x[i_coord_offset+DIM*1+YY];
 158         iz1              = shZ + x[i_coord_offset+DIM*1+ZZ];
 159         ix2              = shX + x[i_coord_offset+DIM*2+XX];
 160         iy2              = shY + x[i_coord_offset+DIM*2+YY];
 161         iz2              = shZ + x[i_coord_offset+DIM*2+ZZ];
 162
 163         fix0             = 0.0;
 164         fiy0             = 0.0;
 165         fiz0             = 0.0;
 166         fix1             = 0.0;
 167         fiy1             = 0.0;
 168         fiz1             = 0.0;
 169         fix2             = 0.0;
 170         fiy2             = 0.0;
 171         fiz2             = 0.0;
 172
 173         /* Reset potential sums */
 174         velecsum         = 0.0;
 175         vvdwsum          = 0.0;
 176
 177         /* Start inner kernel loop */
 178         for(jidx=j_index_start; jidx<j_index_end; jidx++)
 179         {
 180             /* Get j neighbor index, and coordinate index */
 181             jnr              = jjnr[jidx];
 182             j_coord_offset   = DIM*jnr;
 183
 184             /* load j atom coordinates */
 185             jx0              = x[j_coord_offset+DIM*0+XX];
 186             jy0              = x[j_coord_offset+DIM*0+YY];
 187             jz0              = x[j_coord_offset+DIM*0+ZZ];
 188
 189             /* Calculate displacement vector */
 190             dx00             = ix0 - jx0;
 191             dy00             = iy0 - jy0;
 192             dz00             = iz0 - jz0;
 193             dx10             = ix1 - jx0;
 194             dy10             = iy1 - jy0;
 195             dz10             = iz1 - jz0;
 196             dx20             = ix2 - jx0;
 197             dy20             = iy2 - jy0;
 198             dz20             = iz2 - 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
 205             rinv00           = gmx_invsqrt(rsq00);
 206             rinv10           = gmx_invsqrt(rsq10);
 207             rinv20           = gmx_invsqrt(rsq20);
 208
 209             rinvsq00         = rinv00*rinv00;
 210             rinvsq10         = rinv10*rinv10;
 211             rinvsq20         = rinv20*rinv20;
 212
 213             /* Load parameters for j particles */
 214             jq0              = charge[jnr+0];
 215             vdwjidx0         = 2*vdwtype[jnr+0];
 216
 217             /**************************
 218              * CALCULATE INTERACTIONS *
 219              **************************/
 220
 221             r00              = rsq00*rinv00;
 222
 223             qq00             = iq0*jq0;
 224             c6_00            = vdwparam[vdwioffset0+vdwjidx0];
 225             c12_00           = vdwparam[vdwioffset0+vdwjidx0+1];
 226             c6grid_00        = vdwgridparam[vdwioffset0+vdwjidx0];
 227
 228             /* EWALD ELECTROSTATICS */
 229
 230             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 231             ewrt             = r00*ewtabscale;
 232             ewitab           = ewrt;
 233             eweps            = ewrt-ewitab;
 234             ewitab           = 4*ewitab;
 235             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 236             velec            = qq00*(rinv00-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 237             felec            = qq00*rinv00*(rinvsq00-felec);
 238
 239             rinvsix          = rinvsq00*rinvsq00*rinvsq00;
 240             ewcljrsq         = ewclj2*rsq00;
 241             exponent         = exp(-ewcljrsq);
 242             poly             = exponent*(1.0 + ewcljrsq + ewcljrsq*ewcljrsq*0.5);
 243             vvdw6            = (c6_00-c6grid_00*(1.0-poly))*rinvsix;
 244             vvdw12           = c12_00*rinvsix*rinvsix;
 245             vvdw             = vvdw12*(1.0/12.0) - vvdw6*(1.0/6.0);
 246             fvdw             = (vvdw12 - vvdw6 - c6grid_00*(1.0/6.0)*exponent*ewclj6)*rinvsq00;
 247
 248             /* Update potential sums from outer loop */
 249             velecsum        += velec;
 250             vvdwsum         += vvdw;
 251
 252             fscal            = felec+fvdw;
 253
 254             /* Calculate temporary vectorial force */
 255             tx               = fscal*dx00;
 256             ty               = fscal*dy00;
 257             tz               = fscal*dz00;
 258
 259             /* Update vectorial force */
 260             fix0            += tx;
 261             fiy0            += ty;
 262             fiz0            += tz;
 263             f[j_coord_offset+DIM*0+XX] -= tx;
 264             f[j_coord_offset+DIM*0+YY] -= ty;
 265             f[j_coord_offset+DIM*0+ZZ] -= tz;
 266
 267             /**************************
 268              * CALCULATE INTERACTIONS *
 269              **************************/
 270
 271             r10              = rsq10*rinv10;
 272
 273             qq10             = iq1*jq0;
 274
 275             /* EWALD ELECTROSTATICS */
 276
 277             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 278             ewrt             = r10*ewtabscale;
 279             ewitab           = ewrt;
 280             eweps            = ewrt-ewitab;
 281             ewitab           = 4*ewitab;
 282             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 283             velec            = qq10*(rinv10-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 284             felec            = qq10*rinv10*(rinvsq10-felec);
 285
 286             /* Update potential sums from outer loop */
 287             velecsum        += velec;
 288
 289             fscal            = felec;
 290
 291             /* Calculate temporary vectorial force */
 292             tx               = fscal*dx10;
 293             ty               = fscal*dy10;
 294             tz               = fscal*dz10;
 295
 296             /* Update vectorial force */
 297             fix1            += tx;
 298             fiy1            += ty;
 299             fiz1            += tz;
 300             f[j_coord_offset+DIM*0+XX] -= tx;
 301             f[j_coord_offset+DIM*0+YY] -= ty;
 302             f[j_coord_offset+DIM*0+ZZ] -= tz;
 303
 304             /**************************
 305              * CALCULATE INTERACTIONS *
 306              **************************/
 307
 308             r20              = rsq20*rinv20;
 309
 310             qq20             = iq2*jq0;
 311
 312             /* EWALD ELECTROSTATICS */
 313
 314             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 315             ewrt             = r20*ewtabscale;
 316             ewitab           = ewrt;
 317             eweps            = ewrt-ewitab;
 318             ewitab           = 4*ewitab;
 319             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 320             velec            = qq20*(rinv20-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 321             felec            = qq20*rinv20*(rinvsq20-felec);
 322
 323             /* Update potential sums from outer loop */
 324             velecsum        += velec;
 325
 326             fscal            = felec;
 327
 328             /* Calculate temporary vectorial force */
 329             tx               = fscal*dx20;
 330             ty               = fscal*dy20;
 331             tz               = fscal*dz20;
 332
 333             /* Update vectorial force */
 334             fix2            += tx;
 335             fiy2            += ty;
 336             fiz2            += tz;
 337             f[j_coord_offset+DIM*0+XX] -= tx;
 338             f[j_coord_offset+DIM*0+YY] -= ty;
 339             f[j_coord_offset+DIM*0+ZZ] -= tz;
 340
 341             /* Inner loop uses 149 flops */
 342         }
 343         /* End of innermost loop */
 344
 345         tx = ty = tz = 0;
 346         f[i_coord_offset+DIM*0+XX] += fix0;
 347         f[i_coord_offset+DIM*0+YY] += fiy0;
 348         f[i_coord_offset+DIM*0+ZZ] += fiz0;
 349         tx                         += fix0;
 350         ty                         += fiy0;
 351         tz                         += fiz0;
 352         f[i_coord_offset+DIM*1+XX] += fix1;
 353         f[i_coord_offset+DIM*1+YY] += fiy1;
 354         f[i_coord_offset+DIM*1+ZZ] += fiz1;
 355         tx                         += fix1;
 356         ty                         += fiy1;
 357         tz                         += fiz1;
 358         f[i_coord_offset+DIM*2+XX] += fix2;
 359         f[i_coord_offset+DIM*2+YY] += fiy2;
 360         f[i_coord_offset+DIM*2+ZZ] += fiz2;
 361         tx                         += fix2;
 362         ty                         += fiy2;
 363         tz                         += fiz2;
 364         fshift[i_shift_offset+XX]  += tx;
 365         fshift[i_shift_offset+YY]  += ty;
 366         fshift[i_shift_offset+ZZ]  += tz;
 367
 368         ggid                        = gid[iidx];
 369         /* Update potential energies */
 370         kernel_data->energygrp_elec[ggid] += velecsum;
 371         kernel_data->energygrp_vdw[ggid] += vvdwsum;
 372
 373         /* Increment number of inner iterations */
 374         inneriter                  += j_index_end - j_index_start;
 375
 376         /* Outer loop uses 32 flops */
 377     }
 378
 379     /* Increment number of outer iterations */
 380     outeriter        += nri;
 381
 382     /* Update outer/inner flops */
 383
 384     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W3_VF,outeriter*32 + inneriter*149);
 385 }
 386 /*
 387  * Gromacs nonbonded kernel:   nb_kernel_ElecEw_VdwLJEw_GeomW3P1_F_c
 388  * Electrostatics interaction: Ewald
 389  * VdW interaction:            LJEwald
 390  * Geometry:                   Water3-Particle
 391  * Calculate force/pot:        Force
 392  */
 393 void
 394 nb_kernel_ElecEw_VdwLJEw_GeomW3P1_F_c
 395                     (t_nblist                    * gmx_restrict       nlist,
 396                      rvec                        * gmx_restrict          xx,
 397                      rvec                        * gmx_restrict          ff,
 398                      t_forcerec                  * gmx_restrict          fr,
 399                      t_mdatoms                   * gmx_restrict     mdatoms,
 400                      nb_kernel_data_t gmx_unused * gmx_restrict kernel_data,
 401                      t_nrnb                      * gmx_restrict        nrnb)
 402 {
 403     int              i_shift_offset,i_coord_offset,j_coord_offset;
 404     int              j_index_start,j_index_end;
 405     int              nri,inr,ggid,iidx,jidx,jnr,outeriter,inneriter;
 406     real             shX,shY,shZ,tx,ty,tz,fscal,rcutoff,rcutoff2;
 407     int              *iinr,*jindex,*jjnr,*shiftidx,*gid;
 408     real             *shiftvec,*fshift,*x,*f;
 409     int              vdwioffset0;
 410     real             ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
 411     int              vdwioffset1;
 412     real             ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
 413     int              vdwioffset2;
 414     real             ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
 415     int              vdwjidx0;
 416     real             jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
 417     real             dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00,cexp1_00,cexp2_00;
 418     real             dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10,cexp1_10,cexp2_10;
 419     real             dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20,cexp1_20,cexp2_20;
 420     real             velec,felec,velecsum,facel,crf,krf,krf2;
 421     real             *charge;
 422     int              nvdwtype;
 423     real             rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,br,vvdwexp,sh_vdw_invrcut6;
 424     int              *vdwtype;
 425     real             *vdwparam;
 426     real             c6grid_00;
 427     real             c6grid_10;
 428     real             c6grid_20;
 429     real             ewclj,ewclj2,ewclj6,ewcljrsq,poly,exponent,sh_lj_ewald;
 430     real             *vdwgridparam;
 431     int              ewitab;
 432     real             ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace;
 433     real             *ewtab;
 434
 435     x                = xx[0];
 436     f                = ff[0];
 437
 438     nri              = nlist->nri;
 439     iinr             = nlist->iinr;
 440     jindex           = nlist->jindex;
 441     jjnr             = nlist->jjnr;
 442     shiftidx         = nlist->shift;
 443     gid              = nlist->gid;
 444     shiftvec         = fr->shift_vec[0];
 445     fshift           = fr->fshift[0];
 446     facel            = fr->epsfac;
 447     charge           = mdatoms->chargeA;
 448     nvdwtype         = fr->ntype;
 449     vdwparam         = fr->nbfp;
 450     vdwtype          = mdatoms->typeA;
 451     vdwgridparam     = fr->ljpme_c6grid;
 452     ewclj            = fr->ewaldcoeff_lj;
 453     sh_lj_ewald      = fr->ic->sh_lj_ewald;
 454     ewclj2           = ewclj*ewclj;
 455     ewclj6           = ewclj2*ewclj2*ewclj2;
 456
 457     sh_ewald         = fr->ic->sh_ewald;
 458     ewtab            = fr->ic->tabq_coul_F;
 459     ewtabscale       = fr->ic->tabq_scale;
 460     ewtabhalfspace   = 0.5/ewtabscale;
 461
 462     /* Setup water-specific parameters */
 463     inr              = nlist->iinr[0];
 464     iq0              = facel*charge[inr+0];
 465     iq1              = facel*charge[inr+1];
 466     iq2              = facel*charge[inr+2];
 467     vdwioffset0      = 2*nvdwtype*vdwtype[inr+0];
 468
 469     outeriter        = 0;
 470     inneriter        = 0;
 471
 472     /* Start outer loop over neighborlists */
 473     for(iidx=0; iidx<nri; iidx++)
 474     {
 475         /* Load shift vector for this list */
 476         i_shift_offset   = DIM*shiftidx[iidx];
 477         shX              = shiftvec[i_shift_offset+XX];
 478         shY              = shiftvec[i_shift_offset+YY];
 479         shZ              = shiftvec[i_shift_offset+ZZ];
 480
 481         /* Load limits for loop over neighbors */
 482         j_index_start    = jindex[iidx];
 483         j_index_end      = jindex[iidx+1];
 484
 485         /* Get outer coordinate index */
 486         inr              = iinr[iidx];
 487         i_coord_offset   = DIM*inr;
 488
 489         /* Load i particle coords and add shift vector */
 490         ix0              = shX + x[i_coord_offset+DIM*0+XX];
 491         iy0              = shY + x[i_coord_offset+DIM*0+YY];
 492         iz0              = shZ + x[i_coord_offset+DIM*0+ZZ];
 493         ix1              = shX + x[i_coord_offset+DIM*1+XX];
 494         iy1              = shY + x[i_coord_offset+DIM*1+YY];
 495         iz1              = shZ + x[i_coord_offset+DIM*1+ZZ];
 496         ix2              = shX + x[i_coord_offset+DIM*2+XX];
 497         iy2              = shY + x[i_coord_offset+DIM*2+YY];
 498         iz2              = shZ + x[i_coord_offset+DIM*2+ZZ];
 499
 500         fix0             = 0.0;
 501         fiy0             = 0.0;
 502         fiz0             = 0.0;
 503         fix1             = 0.0;
 504         fiy1             = 0.0;
 505         fiz1             = 0.0;
 506         fix2             = 0.0;
 507         fiy2             = 0.0;
 508         fiz2             = 0.0;
 509
 510         /* Start inner kernel loop */
 511         for(jidx=j_index_start; jidx<j_index_end; jidx++)
 512         {
 513             /* Get j neighbor index, and coordinate index */
 514             jnr              = jjnr[jidx];
 515             j_coord_offset   = DIM*jnr;
 516
 517             /* load j atom coordinates */
 518             jx0              = x[j_coord_offset+DIM*0+XX];
 519             jy0              = x[j_coord_offset+DIM*0+YY];
 520             jz0              = x[j_coord_offset+DIM*0+ZZ];
 521
 522             /* Calculate displacement vector */
 523             dx00             = ix0 - jx0;
 524             dy00             = iy0 - jy0;
 525             dz00             = iz0 - jz0;
 526             dx10             = ix1 - jx0;
 527             dy10             = iy1 - jy0;
 528             dz10             = iz1 - jz0;
 529             dx20             = ix2 - jx0;
 530             dy20             = iy2 - jy0;
 531             dz20             = iz2 - jz0;
 532
 533             /* Calculate squared distance and things based on it */
 534             rsq00            = dx00*dx00+dy00*dy00+dz00*dz00;
 535             rsq10            = dx10*dx10+dy10*dy10+dz10*dz10;
 536             rsq20            = dx20*dx20+dy20*dy20+dz20*dz20;
 537
 538             rinv00           = gmx_invsqrt(rsq00);
 539             rinv10           = gmx_invsqrt(rsq10);
 540             rinv20           = gmx_invsqrt(rsq20);
 541
 542             rinvsq00         = rinv00*rinv00;
 543             rinvsq10         = rinv10*rinv10;
 544             rinvsq20         = rinv20*rinv20;
 545
 546             /* Load parameters for j particles */
 547             jq0              = charge[jnr+0];
 548             vdwjidx0         = 2*vdwtype[jnr+0];
 549
 550             /**************************
 551              * CALCULATE INTERACTIONS *
 552              **************************/
 553
 554             r00              = rsq00*rinv00;
 555
 556             qq00             = iq0*jq0;
 557             c6_00            = vdwparam[vdwioffset0+vdwjidx0];
 558             c12_00           = vdwparam[vdwioffset0+vdwjidx0+1];
 559             c6grid_00        = vdwgridparam[vdwioffset0+vdwjidx0];
 560
 561             /* EWALD ELECTROSTATICS */
 562
 563             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 564             ewrt             = r00*ewtabscale;
 565             ewitab           = ewrt;
 566             eweps            = ewrt-ewitab;
 567             felec            = (1.0-eweps)*ewtab[ewitab]+eweps*ewtab[ewitab+1];
 568             felec            = qq00*rinv00*(rinvsq00-felec);
 569
 570             rinvsix          = rinvsq00*rinvsq00*rinvsq00;
 571             ewcljrsq         = ewclj2*rsq00;
 572             exponent         = exp(-ewcljrsq);
 573             poly             = exponent*(1.0 + ewcljrsq + ewcljrsq*ewcljrsq*0.5);
 574             fvdw             = (((c12_00*rinvsix - c6_00 + c6grid_00*(1.0-poly))*rinvsix) - c6grid_00*(1.0/6.0)*exponent*ewclj6)*rinvsq00;
 575
 576             fscal            = felec+fvdw;
 577
 578             /* Calculate temporary vectorial force */
 579             tx               = fscal*dx00;
 580             ty               = fscal*dy00;
 581             tz               = fscal*dz00;
 582
 583             /* Update vectorial force */
 584             fix0            += tx;
 585             fiy0            += ty;
 586             fiz0            += tz;
 587             f[j_coord_offset+DIM*0+XX] -= tx;
 588             f[j_coord_offset+DIM*0+YY] -= ty;
 589             f[j_coord_offset+DIM*0+ZZ] -= tz;
 590
 591             /**************************
 592              * CALCULATE INTERACTIONS *
 593              **************************/
 594
 595             r10              = rsq10*rinv10;
 596
 597             qq10             = iq1*jq0;
 598
 599             /* EWALD ELECTROSTATICS */
 600
 601             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 602             ewrt             = r10*ewtabscale;
 603             ewitab           = ewrt;
 604             eweps            = ewrt-ewitab;
 605             felec            = (1.0-eweps)*ewtab[ewitab]+eweps*ewtab[ewitab+1];
 606             felec            = qq10*rinv10*(rinvsq10-felec);
 607
 608             fscal            = felec;
 609
 610             /* Calculate temporary vectorial force */
 611             tx               = fscal*dx10;
 612             ty               = fscal*dy10;
 613             tz               = fscal*dz10;
 614
 615             /* Update vectorial force */
 616             fix1            += tx;
 617             fiy1            += ty;
 618             fiz1            += tz;
 619             f[j_coord_offset+DIM*0+XX] -= tx;
 620             f[j_coord_offset+DIM*0+YY] -= ty;
 621             f[j_coord_offset+DIM*0+ZZ] -= tz;
 622
 623             /**************************
 624              * CALCULATE INTERACTIONS *
 625              **************************/
 626
 627             r20              = rsq20*rinv20;
 628
 629             qq20             = iq2*jq0;
 630
 631             /* EWALD ELECTROSTATICS */
 632
 633             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 634             ewrt             = r20*ewtabscale;
 635             ewitab           = ewrt;
 636             eweps            = ewrt-ewitab;
 637             felec            = (1.0-eweps)*ewtab[ewitab]+eweps*ewtab[ewitab+1];
 638             felec            = qq20*rinv20*(rinvsq20-felec);
 639
 640             fscal            = felec;
 641
 642             /* Calculate temporary vectorial force */
 643             tx               = fscal*dx20;
 644             ty               = fscal*dy20;
 645             tz               = fscal*dz20;
 646
 647             /* Update vectorial force */
 648             fix2            += tx;
 649             fiy2            += ty;
 650             fiz2            += tz;
 651             f[j_coord_offset+DIM*0+XX] -= tx;
 652             f[j_coord_offset+DIM*0+YY] -= ty;
 653             f[j_coord_offset+DIM*0+ZZ] -= tz;
 654
 655             /* Inner loop uses 123 flops */
 656         }
 657         /* End of innermost loop */
 658
 659         tx = ty = tz = 0;
 660         f[i_coord_offset+DIM*0+XX] += fix0;
 661         f[i_coord_offset+DIM*0+YY] += fiy0;
 662         f[i_coord_offset+DIM*0+ZZ] += fiz0;
 663         tx                         += fix0;
 664         ty                         += fiy0;
 665         tz                         += fiz0;
 666         f[i_coord_offset+DIM*1+XX] += fix1;
 667         f[i_coord_offset+DIM*1+YY] += fiy1;
 668         f[i_coord_offset+DIM*1+ZZ] += fiz1;
 669         tx                         += fix1;
 670         ty                         += fiy1;
 671         tz                         += fiz1;
 672         f[i_coord_offset+DIM*2+XX] += fix2;
 673         f[i_coord_offset+DIM*2+YY] += fiy2;
 674         f[i_coord_offset+DIM*2+ZZ] += fiz2;
 675         tx                         += fix2;
 676         ty                         += fiy2;
 677         tz                         += fiz2;
 678         fshift[i_shift_offset+XX]  += tx;
 679         fshift[i_shift_offset+YY]  += ty;
 680         fshift[i_shift_offset+ZZ]  += tz;
 681
 682         /* Increment number of inner iterations */
 683         inneriter                  += j_index_end - j_index_start;
 684
 685         /* Outer loop uses 30 flops */
 686     }
 687
 688     /* Increment number of outer iterations */
 689     outeriter        += nri;
 690
 691     /* Update outer/inner flops */
 692
 693     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W3_F,outeriter*30 + inneriter*123);
 694 }