src/gromacs/gmxlib/nonbonded/nb_kernel_c/nb_kernel_ElecEwSh_VdwLJEwSh_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 "gmxpre.h"
  39
  40 #include "config.h"
  41
  42 #include <math.h>
  43
  44 #include "../nb_kernel.h"
  45 #include "gromacs/legacyheaders/types/simple.h"
  46 #include "gromacs/math/vec.h"
  47 #include "gromacs/legacyheaders/nrnb.h"
  48
  49 /*
  50  * Gromacs nonbonded kernel:   nb_kernel_ElecEwSh_VdwLJEwSh_GeomW4P1_VF_c
  51  * Electrostatics interaction: Ewald
  52  * VdW interaction:            LJEwald
  53  * Geometry:                   Water4-Particle
  54  * Calculate force/pot:        PotentialAndForce
  55  */
  56 void
  57 nb_kernel_ElecEwSh_VdwLJEwSh_GeomW4P1_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              vdwioffset3;
  79     real             ix3,iy3,iz3,fix3,fiy3,fiz3,iq3,isai3;
  80     int              vdwjidx0;
  81     real             jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
  82     real             dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00,cexp1_00,cexp2_00;
  83     real             dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10,cexp1_10,cexp2_10;
  84     real             dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20,cexp1_20,cexp2_20;
  85     real             dx30,dy30,dz30,rsq30,rinv30,rinvsq30,r30,qq30,c6_30,c12_30,cexp1_30,cexp2_30;
  86     real             velec,felec,velecsum,facel,crf,krf,krf2;
  87     real             *charge;
  88     int              nvdwtype;
  89     real             rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,br,vvdwexp,sh_vdw_invrcut6;
  90     int              *vdwtype;
  91     real             *vdwparam;
  92     real             c6grid_00;
  93     real             c6grid_10;
  94     real             c6grid_20;
  95     real             c6grid_30;
  96     real             ewclj,ewclj2,ewclj6,ewcljrsq,poly,exponent,sh_lj_ewald;
  97     real             *vdwgridparam;
  98     int              ewitab;
  99     real             ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace;
 100     real             *ewtab;
 101
 102     x                = xx[0];
 103     f                = ff[0];
 104
 105     nri              = nlist->nri;
 106     iinr             = nlist->iinr;
 107     jindex           = nlist->jindex;
 108     jjnr             = nlist->jjnr;
 109     shiftidx         = nlist->shift;
 110     gid              = nlist->gid;
 111     shiftvec         = fr->shift_vec[0];
 112     fshift           = fr->fshift[0];
 113     facel            = fr->epsfac;
 114     charge           = mdatoms->chargeA;
 115     nvdwtype         = fr->ntype;
 116     vdwparam         = fr->nbfp;
 117     vdwtype          = mdatoms->typeA;
 118     vdwgridparam     = fr->ljpme_c6grid;
 119     ewclj            = fr->ewaldcoeff_lj;
 120     sh_lj_ewald      = fr->ic->sh_lj_ewald;
 121     ewclj2           = ewclj*ewclj;
 122     ewclj6           = ewclj2*ewclj2*ewclj2;
 123
 124     sh_ewald         = fr->ic->sh_ewald;
 125     ewtab            = fr->ic->tabq_coul_FDV0;
 126     ewtabscale       = fr->ic->tabq_scale;
 127     ewtabhalfspace   = 0.5/ewtabscale;
 128
 129     /* Setup water-specific parameters */
 130     inr              = nlist->iinr[0];
 131     iq1              = facel*charge[inr+1];
 132     iq2              = facel*charge[inr+2];
 133     iq3              = facel*charge[inr+3];
 134     vdwioffset0      = 2*nvdwtype*vdwtype[inr+0];
 135
 136     /* When we use explicit cutoffs the value must be identical for elec and VdW, so use elec as an arbitrary choice */
 137     rcutoff          = fr->rcoulomb;
 138     rcutoff2         = rcutoff*rcutoff;
 139
 140     sh_vdw_invrcut6  = fr->ic->sh_invrc6;
 141     rvdw             = fr->rvdw;
 142
 143     outeriter        = 0;
 144     inneriter        = 0;
 145
 146     /* Start outer loop over neighborlists */
 147     for(iidx=0; iidx<nri; iidx++)
 148     {
 149         /* Load shift vector for this list */
 150         i_shift_offset   = DIM*shiftidx[iidx];
 151         shX              = shiftvec[i_shift_offset+XX];
 152         shY              = shiftvec[i_shift_offset+YY];
 153         shZ              = shiftvec[i_shift_offset+ZZ];
 154
 155         /* Load limits for loop over neighbors */
 156         j_index_start    = jindex[iidx];
 157         j_index_end      = jindex[iidx+1];
 158
 159         /* Get outer coordinate index */
 160         inr              = iinr[iidx];
 161         i_coord_offset   = DIM*inr;
 162
 163         /* Load i particle coords and add shift vector */
 164         ix0              = shX + x[i_coord_offset+DIM*0+XX];
 165         iy0              = shY + x[i_coord_offset+DIM*0+YY];
 166         iz0              = shZ + x[i_coord_offset+DIM*0+ZZ];
 167         ix1              = shX + x[i_coord_offset+DIM*1+XX];
 168         iy1              = shY + x[i_coord_offset+DIM*1+YY];
 169         iz1              = shZ + x[i_coord_offset+DIM*1+ZZ];
 170         ix2              = shX + x[i_coord_offset+DIM*2+XX];
 171         iy2              = shY + x[i_coord_offset+DIM*2+YY];
 172         iz2              = shZ + x[i_coord_offset+DIM*2+ZZ];
 173         ix3              = shX + x[i_coord_offset+DIM*3+XX];
 174         iy3              = shY + x[i_coord_offset+DIM*3+YY];
 175         iz3              = shZ + x[i_coord_offset+DIM*3+ZZ];
 176
 177         fix0             = 0.0;
 178         fiy0             = 0.0;
 179         fiz0             = 0.0;
 180         fix1             = 0.0;
 181         fiy1             = 0.0;
 182         fiz1             = 0.0;
 183         fix2             = 0.0;
 184         fiy2             = 0.0;
 185         fiz2             = 0.0;
 186         fix3             = 0.0;
 187         fiy3             = 0.0;
 188         fiz3             = 0.0;
 189
 190         /* Reset potential sums */
 191         velecsum         = 0.0;
 192         vvdwsum          = 0.0;
 193
 194         /* Start inner kernel loop */
 195         for(jidx=j_index_start; jidx<j_index_end; jidx++)
 196         {
 197             /* Get j neighbor index, and coordinate index */
 198             jnr              = jjnr[jidx];
 199             j_coord_offset   = DIM*jnr;
 200
 201             /* load j atom coordinates */
 202             jx0              = x[j_coord_offset+DIM*0+XX];
 203             jy0              = x[j_coord_offset+DIM*0+YY];
 204             jz0              = x[j_coord_offset+DIM*0+ZZ];
 205
 206             /* Calculate displacement vector */
 207             dx00             = ix0 - jx0;
 208             dy00             = iy0 - jy0;
 209             dz00             = iz0 - jz0;
 210             dx10             = ix1 - jx0;
 211             dy10             = iy1 - jy0;
 212             dz10             = iz1 - jz0;
 213             dx20             = ix2 - jx0;
 214             dy20             = iy2 - jy0;
 215             dz20             = iz2 - jz0;
 216             dx30             = ix3 - jx0;
 217             dy30             = iy3 - jy0;
 218             dz30             = iz3 - jz0;
 219
 220             /* Calculate squared distance and things based on it */
 221             rsq00            = dx00*dx00+dy00*dy00+dz00*dz00;
 222             rsq10            = dx10*dx10+dy10*dy10+dz10*dz10;
 223             rsq20            = dx20*dx20+dy20*dy20+dz20*dz20;
 224             rsq30            = dx30*dx30+dy30*dy30+dz30*dz30;
 225
 226             rinv00           = gmx_invsqrt(rsq00);
 227             rinv10           = gmx_invsqrt(rsq10);
 228             rinv20           = gmx_invsqrt(rsq20);
 229             rinv30           = gmx_invsqrt(rsq30);
 230
 231             rinvsq00         = rinv00*rinv00;
 232             rinvsq10         = rinv10*rinv10;
 233             rinvsq20         = rinv20*rinv20;
 234             rinvsq30         = rinv30*rinv30;
 235
 236             /* Load parameters for j particles */
 237             jq0              = charge[jnr+0];
 238             vdwjidx0         = 2*vdwtype[jnr+0];
 239
 240             /**************************
 241              * CALCULATE INTERACTIONS *
 242              **************************/
 243
 244             if (rsq00<rcutoff2)
 245             {
 246
 247             r00              = rsq00*rinv00;
 248
 249             c6_00            = vdwparam[vdwioffset0+vdwjidx0];
 250             c12_00           = vdwparam[vdwioffset0+vdwjidx0+1];
 251             c6grid_00        = vdwgridparam[vdwioffset0+vdwjidx0];
 252
 253             rinvsix          = rinvsq00*rinvsq00*rinvsq00;
 254             ewcljrsq         = ewclj2*rsq00;
 255             exponent         = exp(-ewcljrsq);
 256             poly             = exponent*(1.0 + ewcljrsq + ewcljrsq*ewcljrsq*0.5);
 257             vvdw6            = (c6_00-c6grid_00*(1.0-poly))*rinvsix;
 258             vvdw12           = c12_00*rinvsix*rinvsix;
 259             vvdw             = (vvdw12 - c12_00*sh_vdw_invrcut6*sh_vdw_invrcut6)*(1.0/12.0) - (vvdw6 - c6_00*sh_vdw_invrcut6 - c6grid_00*sh_lj_ewald)*(1.0/6.0);
 260             fvdw             = (vvdw12 - vvdw6 - c6grid_00*(1.0/6.0)*exponent*ewclj6)*rinvsq00;
 261
 262             /* Update potential sums from outer loop */
 263             vvdwsum         += vvdw;
 264
 265             fscal            = fvdw;
 266
 267             /* Calculate temporary vectorial force */
 268             tx               = fscal*dx00;
 269             ty               = fscal*dy00;
 270             tz               = fscal*dz00;
 271
 272             /* Update vectorial force */
 273             fix0            += tx;
 274             fiy0            += ty;
 275             fiz0            += tz;
 276             f[j_coord_offset+DIM*0+XX] -= tx;
 277             f[j_coord_offset+DIM*0+YY] -= ty;
 278             f[j_coord_offset+DIM*0+ZZ] -= tz;
 279
 280             }
 281
 282             /**************************
 283              * CALCULATE INTERACTIONS *
 284              **************************/
 285
 286             if (rsq10<rcutoff2)
 287             {
 288
 289             r10              = rsq10*rinv10;
 290
 291             qq10             = iq1*jq0;
 292
 293             /* EWALD ELECTROSTATICS */
 294
 295             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 296             ewrt             = r10*ewtabscale;
 297             ewitab           = ewrt;
 298             eweps            = ewrt-ewitab;
 299             ewitab           = 4*ewitab;
 300             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 301             velec            = qq10*((rinv10-sh_ewald)-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 302             felec            = qq10*rinv10*(rinvsq10-felec);
 303
 304             /* Update potential sums from outer loop */
 305             velecsum        += velec;
 306
 307             fscal            = felec;
 308
 309             /* Calculate temporary vectorial force */
 310             tx               = fscal*dx10;
 311             ty               = fscal*dy10;
 312             tz               = fscal*dz10;
 313
 314             /* Update vectorial force */
 315             fix1            += tx;
 316             fiy1            += ty;
 317             fiz1            += tz;
 318             f[j_coord_offset+DIM*0+XX] -= tx;
 319             f[j_coord_offset+DIM*0+YY] -= ty;
 320             f[j_coord_offset+DIM*0+ZZ] -= tz;
 321
 322             }
 323
 324             /**************************
 325              * CALCULATE INTERACTIONS *
 326              **************************/
 327
 328             if (rsq20<rcutoff2)
 329             {
 330
 331             r20              = rsq20*rinv20;
 332
 333             qq20             = iq2*jq0;
 334
 335             /* EWALD ELECTROSTATICS */
 336
 337             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 338             ewrt             = r20*ewtabscale;
 339             ewitab           = ewrt;
 340             eweps            = ewrt-ewitab;
 341             ewitab           = 4*ewitab;
 342             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 343             velec            = qq20*((rinv20-sh_ewald)-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 344             felec            = qq20*rinv20*(rinvsq20-felec);
 345
 346             /* Update potential sums from outer loop */
 347             velecsum        += velec;
 348
 349             fscal            = felec;
 350
 351             /* Calculate temporary vectorial force */
 352             tx               = fscal*dx20;
 353             ty               = fscal*dy20;
 354             tz               = fscal*dz20;
 355
 356             /* Update vectorial force */
 357             fix2            += tx;
 358             fiy2            += ty;
 359             fiz2            += tz;
 360             f[j_coord_offset+DIM*0+XX] -= tx;
 361             f[j_coord_offset+DIM*0+YY] -= ty;
 362             f[j_coord_offset+DIM*0+ZZ] -= tz;
 363
 364             }
 365
 366             /**************************
 367              * CALCULATE INTERACTIONS *
 368              **************************/
 369
 370             if (rsq30<rcutoff2)
 371             {
 372
 373             r30              = rsq30*rinv30;
 374
 375             qq30             = iq3*jq0;
 376
 377             /* EWALD ELECTROSTATICS */
 378
 379             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 380             ewrt             = r30*ewtabscale;
 381             ewitab           = ewrt;
 382             eweps            = ewrt-ewitab;
 383             ewitab           = 4*ewitab;
 384             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 385             velec            = qq30*((rinv30-sh_ewald)-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 386             felec            = qq30*rinv30*(rinvsq30-felec);
 387
 388             /* Update potential sums from outer loop */
 389             velecsum        += velec;
 390
 391             fscal            = felec;
 392
 393             /* Calculate temporary vectorial force */
 394             tx               = fscal*dx30;
 395             ty               = fscal*dy30;
 396             tz               = fscal*dz30;
 397
 398             /* Update vectorial force */
 399             fix3            += tx;
 400             fiy3            += ty;
 401             fiz3            += tz;
 402             f[j_coord_offset+DIM*0+XX] -= tx;
 403             f[j_coord_offset+DIM*0+YY] -= ty;
 404             f[j_coord_offset+DIM*0+ZZ] -= tz;
 405
 406             }
 407
 408             /* Inner loop uses 181 flops */
 409         }
 410         /* End of innermost loop */
 411
 412         tx = ty = tz = 0;
 413         f[i_coord_offset+DIM*0+XX] += fix0;
 414         f[i_coord_offset+DIM*0+YY] += fiy0;
 415         f[i_coord_offset+DIM*0+ZZ] += fiz0;
 416         tx                         += fix0;
 417         ty                         += fiy0;
 418         tz                         += fiz0;
 419         f[i_coord_offset+DIM*1+XX] += fix1;
 420         f[i_coord_offset+DIM*1+YY] += fiy1;
 421         f[i_coord_offset+DIM*1+ZZ] += fiz1;
 422         tx                         += fix1;
 423         ty                         += fiy1;
 424         tz                         += fiz1;
 425         f[i_coord_offset+DIM*2+XX] += fix2;
 426         f[i_coord_offset+DIM*2+YY] += fiy2;
 427         f[i_coord_offset+DIM*2+ZZ] += fiz2;
 428         tx                         += fix2;
 429         ty                         += fiy2;
 430         tz                         += fiz2;
 431         f[i_coord_offset+DIM*3+XX] += fix3;
 432         f[i_coord_offset+DIM*3+YY] += fiy3;
 433         f[i_coord_offset+DIM*3+ZZ] += fiz3;
 434         tx                         += fix3;
 435         ty                         += fiy3;
 436         tz                         += fiz3;
 437         fshift[i_shift_offset+XX]  += tx;
 438         fshift[i_shift_offset+YY]  += ty;
 439         fshift[i_shift_offset+ZZ]  += tz;
 440
 441         ggid                        = gid[iidx];
 442         /* Update potential energies */
 443         kernel_data->energygrp_elec[ggid] += velecsum;
 444         kernel_data->energygrp_vdw[ggid] += vvdwsum;
 445
 446         /* Increment number of inner iterations */
 447         inneriter                  += j_index_end - j_index_start;
 448
 449         /* Outer loop uses 41 flops */
 450     }
 451
 452     /* Increment number of outer iterations */
 453     outeriter        += nri;
 454
 455     /* Update outer/inner flops */
 456
 457     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W4_VF,outeriter*41 + inneriter*181);
 458 }
 459 /*
 460  * Gromacs nonbonded kernel:   nb_kernel_ElecEwSh_VdwLJEwSh_GeomW4P1_F_c
 461  * Electrostatics interaction: Ewald
 462  * VdW interaction:            LJEwald
 463  * Geometry:                   Water4-Particle
 464  * Calculate force/pot:        Force
 465  */
 466 void
 467 nb_kernel_ElecEwSh_VdwLJEwSh_GeomW4P1_F_c
 468                     (t_nblist                    * gmx_restrict       nlist,
 469                      rvec                        * gmx_restrict          xx,
 470                      rvec                        * gmx_restrict          ff,
 471                      t_forcerec                  * gmx_restrict          fr,
 472                      t_mdatoms                   * gmx_restrict     mdatoms,
 473                      nb_kernel_data_t gmx_unused * gmx_restrict kernel_data,
 474                      t_nrnb                      * gmx_restrict        nrnb)
 475 {
 476     int              i_shift_offset,i_coord_offset,j_coord_offset;
 477     int              j_index_start,j_index_end;
 478     int              nri,inr,ggid,iidx,jidx,jnr,outeriter,inneriter;
 479     real             shX,shY,shZ,tx,ty,tz,fscal,rcutoff,rcutoff2;
 480     int              *iinr,*jindex,*jjnr,*shiftidx,*gid;
 481     real             *shiftvec,*fshift,*x,*f;
 482     int              vdwioffset0;
 483     real             ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
 484     int              vdwioffset1;
 485     real             ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
 486     int              vdwioffset2;
 487     real             ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
 488     int              vdwioffset3;
 489     real             ix3,iy3,iz3,fix3,fiy3,fiz3,iq3,isai3;
 490     int              vdwjidx0;
 491     real             jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
 492     real             dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00,cexp1_00,cexp2_00;
 493     real             dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10,cexp1_10,cexp2_10;
 494     real             dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20,cexp1_20,cexp2_20;
 495     real             dx30,dy30,dz30,rsq30,rinv30,rinvsq30,r30,qq30,c6_30,c12_30,cexp1_30,cexp2_30;
 496     real             velec,felec,velecsum,facel,crf,krf,krf2;
 497     real             *charge;
 498     int              nvdwtype;
 499     real             rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,br,vvdwexp,sh_vdw_invrcut6;
 500     int              *vdwtype;
 501     real             *vdwparam;
 502     real             c6grid_00;
 503     real             c6grid_10;
 504     real             c6grid_20;
 505     real             c6grid_30;
 506     real             ewclj,ewclj2,ewclj6,ewcljrsq,poly,exponent,sh_lj_ewald;
 507     real             *vdwgridparam;
 508     int              ewitab;
 509     real             ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace;
 510     real             *ewtab;
 511
 512     x                = xx[0];
 513     f                = ff[0];
 514
 515     nri              = nlist->nri;
 516     iinr             = nlist->iinr;
 517     jindex           = nlist->jindex;
 518     jjnr             = nlist->jjnr;
 519     shiftidx         = nlist->shift;
 520     gid              = nlist->gid;
 521     shiftvec         = fr->shift_vec[0];
 522     fshift           = fr->fshift[0];
 523     facel            = fr->epsfac;
 524     charge           = mdatoms->chargeA;
 525     nvdwtype         = fr->ntype;
 526     vdwparam         = fr->nbfp;
 527     vdwtype          = mdatoms->typeA;
 528     vdwgridparam     = fr->ljpme_c6grid;
 529     ewclj            = fr->ewaldcoeff_lj;
 530     sh_lj_ewald      = fr->ic->sh_lj_ewald;
 531     ewclj2           = ewclj*ewclj;
 532     ewclj6           = ewclj2*ewclj2*ewclj2;
 533
 534     sh_ewald         = fr->ic->sh_ewald;
 535     ewtab            = fr->ic->tabq_coul_F;
 536     ewtabscale       = fr->ic->tabq_scale;
 537     ewtabhalfspace   = 0.5/ewtabscale;
 538
 539     /* Setup water-specific parameters */
 540     inr              = nlist->iinr[0];
 541     iq1              = facel*charge[inr+1];
 542     iq2              = facel*charge[inr+2];
 543     iq3              = facel*charge[inr+3];
 544     vdwioffset0      = 2*nvdwtype*vdwtype[inr+0];
 545
 546     /* When we use explicit cutoffs the value must be identical for elec and VdW, so use elec as an arbitrary choice */
 547     rcutoff          = fr->rcoulomb;
 548     rcutoff2         = rcutoff*rcutoff;
 549
 550     sh_vdw_invrcut6  = fr->ic->sh_invrc6;
 551     rvdw             = fr->rvdw;
 552
 553     outeriter        = 0;
 554     inneriter        = 0;
 555
 556     /* Start outer loop over neighborlists */
 557     for(iidx=0; iidx<nri; iidx++)
 558     {
 559         /* Load shift vector for this list */
 560         i_shift_offset   = DIM*shiftidx[iidx];
 561         shX              = shiftvec[i_shift_offset+XX];
 562         shY              = shiftvec[i_shift_offset+YY];
 563         shZ              = shiftvec[i_shift_offset+ZZ];
 564
 565         /* Load limits for loop over neighbors */
 566         j_index_start    = jindex[iidx];
 567         j_index_end      = jindex[iidx+1];
 568
 569         /* Get outer coordinate index */
 570         inr              = iinr[iidx];
 571         i_coord_offset   = DIM*inr;
 572
 573         /* Load i particle coords and add shift vector */
 574         ix0              = shX + x[i_coord_offset+DIM*0+XX];
 575         iy0              = shY + x[i_coord_offset+DIM*0+YY];
 576         iz0              = shZ + x[i_coord_offset+DIM*0+ZZ];
 577         ix1              = shX + x[i_coord_offset+DIM*1+XX];
 578         iy1              = shY + x[i_coord_offset+DIM*1+YY];
 579         iz1              = shZ + x[i_coord_offset+DIM*1+ZZ];
 580         ix2              = shX + x[i_coord_offset+DIM*2+XX];
 581         iy2              = shY + x[i_coord_offset+DIM*2+YY];
 582         iz2              = shZ + x[i_coord_offset+DIM*2+ZZ];
 583         ix3              = shX + x[i_coord_offset+DIM*3+XX];
 584         iy3              = shY + x[i_coord_offset+DIM*3+YY];
 585         iz3              = shZ + x[i_coord_offset+DIM*3+ZZ];
 586
 587         fix0             = 0.0;
 588         fiy0             = 0.0;
 589         fiz0             = 0.0;
 590         fix1             = 0.0;
 591         fiy1             = 0.0;
 592         fiz1             = 0.0;
 593         fix2             = 0.0;
 594         fiy2             = 0.0;
 595         fiz2             = 0.0;
 596         fix3             = 0.0;
 597         fiy3             = 0.0;
 598         fiz3             = 0.0;
 599
 600         /* Start inner kernel loop */
 601         for(jidx=j_index_start; jidx<j_index_end; jidx++)
 602         {
 603             /* Get j neighbor index, and coordinate index */
 604             jnr              = jjnr[jidx];
 605             j_coord_offset   = DIM*jnr;
 606
 607             /* load j atom coordinates */
 608             jx0              = x[j_coord_offset+DIM*0+XX];
 609             jy0              = x[j_coord_offset+DIM*0+YY];
 610             jz0              = x[j_coord_offset+DIM*0+ZZ];
 611
 612             /* Calculate displacement vector */
 613             dx00             = ix0 - jx0;
 614             dy00             = iy0 - jy0;
 615             dz00             = iz0 - jz0;
 616             dx10             = ix1 - jx0;
 617             dy10             = iy1 - jy0;
 618             dz10             = iz1 - jz0;
 619             dx20             = ix2 - jx0;
 620             dy20             = iy2 - jy0;
 621             dz20             = iz2 - jz0;
 622             dx30             = ix3 - jx0;
 623             dy30             = iy3 - jy0;
 624             dz30             = iz3 - jz0;
 625
 626             /* Calculate squared distance and things based on it */
 627             rsq00            = dx00*dx00+dy00*dy00+dz00*dz00;
 628             rsq10            = dx10*dx10+dy10*dy10+dz10*dz10;
 629             rsq20            = dx20*dx20+dy20*dy20+dz20*dz20;
 630             rsq30            = dx30*dx30+dy30*dy30+dz30*dz30;
 631
 632             rinv00           = gmx_invsqrt(rsq00);
 633             rinv10           = gmx_invsqrt(rsq10);
 634             rinv20           = gmx_invsqrt(rsq20);
 635             rinv30           = gmx_invsqrt(rsq30);
 636
 637             rinvsq00         = rinv00*rinv00;
 638             rinvsq10         = rinv10*rinv10;
 639             rinvsq20         = rinv20*rinv20;
 640             rinvsq30         = rinv30*rinv30;
 641
 642             /* Load parameters for j particles */
 643             jq0              = charge[jnr+0];
 644             vdwjidx0         = 2*vdwtype[jnr+0];
 645
 646             /**************************
 647              * CALCULATE INTERACTIONS *
 648              **************************/
 649
 650             if (rsq00<rcutoff2)
 651             {
 652
 653             r00              = rsq00*rinv00;
 654
 655             c6_00            = vdwparam[vdwioffset0+vdwjidx0];
 656             c12_00           = vdwparam[vdwioffset0+vdwjidx0+1];
 657             c6grid_00        = vdwgridparam[vdwioffset0+vdwjidx0];
 658
 659             rinvsix          = rinvsq00*rinvsq00*rinvsq00;
 660             ewcljrsq         = ewclj2*rsq00;
 661             exponent         = exp(-ewcljrsq);
 662             poly             = exponent*(1.0 + ewcljrsq + ewcljrsq*ewcljrsq*0.5);
 663             fvdw             = (((c12_00*rinvsix - c6_00 + c6grid_00*(1.0-poly))*rinvsix) - c6grid_00*(1.0/6.0)*exponent*ewclj6)*rinvsq00;
 664
 665             fscal            = fvdw;
 666
 667             /* Calculate temporary vectorial force */
 668             tx               = fscal*dx00;
 669             ty               = fscal*dy00;
 670             tz               = fscal*dz00;
 671
 672             /* Update vectorial force */
 673             fix0            += tx;
 674             fiy0            += ty;
 675             fiz0            += tz;
 676             f[j_coord_offset+DIM*0+XX] -= tx;
 677             f[j_coord_offset+DIM*0+YY] -= ty;
 678             f[j_coord_offset+DIM*0+ZZ] -= tz;
 679
 680             }
 681
 682             /**************************
 683              * CALCULATE INTERACTIONS *
 684              **************************/
 685
 686             if (rsq10<rcutoff2)
 687             {
 688
 689             r10              = rsq10*rinv10;
 690
 691             qq10             = iq1*jq0;
 692
 693             /* EWALD ELECTROSTATICS */
 694
 695             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 696             ewrt             = r10*ewtabscale;
 697             ewitab           = ewrt;
 698             eweps            = ewrt-ewitab;
 699             felec            = (1.0-eweps)*ewtab[ewitab]+eweps*ewtab[ewitab+1];
 700             felec            = qq10*rinv10*(rinvsq10-felec);
 701
 702             fscal            = felec;
 703
 704             /* Calculate temporary vectorial force */
 705             tx               = fscal*dx10;
 706             ty               = fscal*dy10;
 707             tz               = fscal*dz10;
 708
 709             /* Update vectorial force */
 710             fix1            += tx;
 711             fiy1            += ty;
 712             fiz1            += tz;
 713             f[j_coord_offset+DIM*0+XX] -= tx;
 714             f[j_coord_offset+DIM*0+YY] -= ty;
 715             f[j_coord_offset+DIM*0+ZZ] -= tz;
 716
 717             }
 718
 719             /**************************
 720              * CALCULATE INTERACTIONS *
 721              **************************/
 722
 723             if (rsq20<rcutoff2)
 724             {
 725
 726             r20              = rsq20*rinv20;
 727
 728             qq20             = iq2*jq0;
 729
 730             /* EWALD ELECTROSTATICS */
 731
 732             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 733             ewrt             = r20*ewtabscale;
 734             ewitab           = ewrt;
 735             eweps            = ewrt-ewitab;
 736             felec            = (1.0-eweps)*ewtab[ewitab]+eweps*ewtab[ewitab+1];
 737             felec            = qq20*rinv20*(rinvsq20-felec);
 738
 739             fscal            = felec;
 740
 741             /* Calculate temporary vectorial force */
 742             tx               = fscal*dx20;
 743             ty               = fscal*dy20;
 744             tz               = fscal*dz20;
 745
 746             /* Update vectorial force */
 747             fix2            += tx;
 748             fiy2            += ty;
 749             fiz2            += tz;
 750             f[j_coord_offset+DIM*0+XX] -= tx;
 751             f[j_coord_offset+DIM*0+YY] -= ty;
 752             f[j_coord_offset+DIM*0+ZZ] -= tz;
 753
 754             }
 755
 756             /**************************
 757              * CALCULATE INTERACTIONS *
 758              **************************/
 759
 760             if (rsq30<rcutoff2)
 761             {
 762
 763             r30              = rsq30*rinv30;
 764
 765             qq30             = iq3*jq0;
 766
 767             /* EWALD ELECTROSTATICS */
 768
 769             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 770             ewrt             = r30*ewtabscale;
 771             ewitab           = ewrt;
 772             eweps            = ewrt-ewitab;
 773             felec            = (1.0-eweps)*ewtab[ewitab]+eweps*ewtab[ewitab+1];
 774             felec            = qq30*rinv30*(rinvsq30-felec);
 775
 776             fscal            = felec;
 777
 778             /* Calculate temporary vectorial force */
 779             tx               = fscal*dx30;
 780             ty               = fscal*dy30;
 781             tz               = fscal*dz30;
 782
 783             /* Update vectorial force */
 784             fix3            += tx;
 785             fiy3            += ty;
 786             fiz3            += tz;
 787             f[j_coord_offset+DIM*0+XX] -= tx;
 788             f[j_coord_offset+DIM*0+YY] -= ty;
 789             f[j_coord_offset+DIM*0+ZZ] -= tz;
 790
 791             }
 792
 793             /* Inner loop uses 146 flops */
 794         }
 795         /* End of innermost loop */
 796
 797         tx = ty = tz = 0;
 798         f[i_coord_offset+DIM*0+XX] += fix0;
 799         f[i_coord_offset+DIM*0+YY] += fiy0;
 800         f[i_coord_offset+DIM*0+ZZ] += fiz0;
 801         tx                         += fix0;
 802         ty                         += fiy0;
 803         tz                         += fiz0;
 804         f[i_coord_offset+DIM*1+XX] += fix1;
 805         f[i_coord_offset+DIM*1+YY] += fiy1;
 806         f[i_coord_offset+DIM*1+ZZ] += fiz1;
 807         tx                         += fix1;
 808         ty                         += fiy1;
 809         tz                         += fiz1;
 810         f[i_coord_offset+DIM*2+XX] += fix2;
 811         f[i_coord_offset+DIM*2+YY] += fiy2;
 812         f[i_coord_offset+DIM*2+ZZ] += fiz2;
 813         tx                         += fix2;
 814         ty                         += fiy2;
 815         tz                         += fiz2;
 816         f[i_coord_offset+DIM*3+XX] += fix3;
 817         f[i_coord_offset+DIM*3+YY] += fiy3;
 818         f[i_coord_offset+DIM*3+ZZ] += fiz3;
 819         tx                         += fix3;
 820         ty                         += fiy3;
 821         tz                         += fiz3;
 822         fshift[i_shift_offset+XX]  += tx;
 823         fshift[i_shift_offset+YY]  += ty;
 824         fshift[i_shift_offset+ZZ]  += tz;
 825
 826         /* Increment number of inner iterations */
 827         inneriter                  += j_index_end - j_index_start;
 828
 829         /* Outer loop uses 39 flops */
 830     }
 831
 832     /* Increment number of outer iterations */
 833     outeriter        += nri;
 834
 835     /* Update outer/inner flops */
 836
 837     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W4_F,outeriter*39 + inneriter*146);
 838 }