src/gromacs/gmxlib/nonbonded/nb_kernel_c/nb_kernel_ElecEwSh_VdwLJEwSh_GeomW3W3_c.c

   1 /*
   2  * This file is part of the GROMACS molecular simulation package.
   3  *
   4  * Copyright (c) 2012,2013,2014, by the GROMACS development team, led by
   5  * Mark Abraham, David van der Spoel, Berk Hess, and Erik Lindahl,
   6  * and including many others, as listed in the AUTHORS file in the
   7  * top-level source directory and at http://www.gromacs.org.
   8  *
   9  * GROMACS is free software; you can redistribute it and/or
  10  * modify it under the terms of the GNU Lesser General Public License
  11  * as published by the Free Software Foundation; either version 2.1
  12  * of the License, or (at your option) any later version.
  13  *
  14  * GROMACS is distributed in the hope that it will be useful,
  15  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  16  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  17  * Lesser General Public License for more details.
  18  *
  19  * You should have received a copy of the GNU Lesser General Public
  20  * License along with GROMACS; if not, see
  21  * http://www.gnu.org/licenses, or write to the Free Software Foundation,
  22  * Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02110-1301  USA.
  23  *
  24  * If you want to redistribute modifications to GROMACS, please
  25  * consider that scientific software is very special. Version
  26  * control is crucial - bugs must be traceable. We will be happy to
  27  * consider code for inclusion in the official distribution, but
  28  * derived work must not be called official GROMACS. Details are found
  29  * in the README & COPYING files - if they are missing, get the
  30  * official version at http://www.gromacs.org.
  31  *
  32  * To help us fund GROMACS development, we humbly ask that you cite
  33  * the research papers on the package. Check out http://www.gromacs.org.
  34  */
  35 /*
  36  * Note: this file was generated by the GROMACS c kernel generator.
  37  */
  38 #include "config.h"
  39
  40 #include <math.h>
  41
  42 #include "../nb_kernel.h"
  43 #include "gromacs/legacyheaders/types/simple.h"
  44 #include "gromacs/math/vec.h"
  45 #include "gromacs/legacyheaders/nrnb.h"
  46
  47 /*
  48  * Gromacs nonbonded kernel:   nb_kernel_ElecEwSh_VdwLJEwSh_GeomW3W3_VF_c
  49  * Electrostatics interaction: Ewald
  50  * VdW interaction:            LJEwald
  51  * Geometry:                   Water3-Water3
  52  * Calculate force/pot:        PotentialAndForce
  53  */
  54 void
  55 nb_kernel_ElecEwSh_VdwLJEwSh_GeomW3W3_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     int              vdwjidx1;
  79     real             jx1,jy1,jz1,fjx1,fjy1,fjz1,jq1,isaj1;
  80     int              vdwjidx2;
  81     real             jx2,jy2,jz2,fjx2,fjy2,fjz2,jq2,isaj2;
  82     real             dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00,cexp1_00,cexp2_00;
  83     real             dx01,dy01,dz01,rsq01,rinv01,rinvsq01,r01,qq01,c6_01,c12_01,cexp1_01,cexp2_01;
  84     real             dx02,dy02,dz02,rsq02,rinv02,rinvsq02,r02,qq02,c6_02,c12_02,cexp1_02,cexp2_02;
  85     real             dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10,cexp1_10,cexp2_10;
  86     real             dx11,dy11,dz11,rsq11,rinv11,rinvsq11,r11,qq11,c6_11,c12_11,cexp1_11,cexp2_11;
  87     real             dx12,dy12,dz12,rsq12,rinv12,rinvsq12,r12,qq12,c6_12,c12_12,cexp1_12,cexp2_12;
  88     real             dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20,cexp1_20,cexp2_20;
  89     real             dx21,dy21,dz21,rsq21,rinv21,rinvsq21,r21,qq21,c6_21,c12_21,cexp1_21,cexp2_21;
  90     real             dx22,dy22,dz22,rsq22,rinv22,rinvsq22,r22,qq22,c6_22,c12_22,cexp1_22,cexp2_22;
  91     real             velec,felec,velecsum,facel,crf,krf,krf2;
  92     real             *charge;
  93     int              nvdwtype;
  94     real             rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,br,vvdwexp,sh_vdw_invrcut6;
  95     int              *vdwtype;
  96     real             *vdwparam;
  97     real             c6grid_00;
  98     real             c6grid_01;
  99     real             c6grid_02;
 100     real             c6grid_10;
 101     real             c6grid_11;
 102     real             c6grid_12;
 103     real             c6grid_20;
 104     real             c6grid_21;
 105     real             c6grid_22;
 106     real             ewclj,ewclj2,ewclj6,ewcljrsq,poly,exponent,sh_lj_ewald;
 107     real             *vdwgridparam;
 108     int              ewitab;
 109     real             ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace;
 110     real             *ewtab;
 111
 112     x                = xx[0];
 113     f                = ff[0];
 114
 115     nri              = nlist->nri;
 116     iinr             = nlist->iinr;
 117     jindex           = nlist->jindex;
 118     jjnr             = nlist->jjnr;
 119     shiftidx         = nlist->shift;
 120     gid              = nlist->gid;
 121     shiftvec         = fr->shift_vec[0];
 122     fshift           = fr->fshift[0];
 123     facel            = fr->epsfac;
 124     charge           = mdatoms->chargeA;
 125     nvdwtype         = fr->ntype;
 126     vdwparam         = fr->nbfp;
 127     vdwtype          = mdatoms->typeA;
 128     vdwgridparam     = fr->ljpme_c6grid;
 129     ewclj            = fr->ewaldcoeff_lj;
 130     sh_lj_ewald      = fr->ic->sh_lj_ewald;
 131     ewclj2           = ewclj*ewclj;
 132     ewclj6           = ewclj2*ewclj2*ewclj2;
 133
 134     sh_ewald         = fr->ic->sh_ewald;
 135     ewtab            = fr->ic->tabq_coul_FDV0;
 136     ewtabscale       = fr->ic->tabq_scale;
 137     ewtabhalfspace   = 0.5/ewtabscale;
 138
 139     /* Setup water-specific parameters */
 140     inr              = nlist->iinr[0];
 141     iq0              = facel*charge[inr+0];
 142     iq1              = facel*charge[inr+1];
 143     iq2              = facel*charge[inr+2];
 144     vdwioffset0      = 2*nvdwtype*vdwtype[inr+0];
 145
 146     jq0              = charge[inr+0];
 147     jq1              = charge[inr+1];
 148     jq2              = charge[inr+2];
 149     vdwjidx0         = 2*vdwtype[inr+0];
 150     qq00             = iq0*jq0;
 151     c6_00            = vdwparam[vdwioffset0+vdwjidx0];
 152     c12_00           = vdwparam[vdwioffset0+vdwjidx0+1];
 153     c6grid_00        = vdwgridparam[vdwioffset0+vdwjidx0];
 154     qq01             = iq0*jq1;
 155     qq02             = iq0*jq2;
 156     qq10             = iq1*jq0;
 157     qq11             = iq1*jq1;
 158     qq12             = iq1*jq2;
 159     qq20             = iq2*jq0;
 160     qq21             = iq2*jq1;
 161     qq22             = iq2*jq2;
 162
 163     /* When we use explicit cutoffs the value must be identical for elec and VdW, so use elec as an arbitrary choice */
 164     rcutoff          = fr->rcoulomb;
 165     rcutoff2         = rcutoff*rcutoff;
 166
 167     sh_vdw_invrcut6  = fr->ic->sh_invrc6;
 168     rvdw             = fr->rvdw;
 169
 170     outeriter        = 0;
 171     inneriter        = 0;
 172
 173     /* Start outer loop over neighborlists */
 174     for(iidx=0; iidx<nri; iidx++)
 175     {
 176         /* Load shift vector for this list */
 177         i_shift_offset   = DIM*shiftidx[iidx];
 178         shX              = shiftvec[i_shift_offset+XX];
 179         shY              = shiftvec[i_shift_offset+YY];
 180         shZ              = shiftvec[i_shift_offset+ZZ];
 181
 182         /* Load limits for loop over neighbors */
 183         j_index_start    = jindex[iidx];
 184         j_index_end      = jindex[iidx+1];
 185
 186         /* Get outer coordinate index */
 187         inr              = iinr[iidx];
 188         i_coord_offset   = DIM*inr;
 189
 190         /* Load i particle coords and add shift vector */
 191         ix0              = shX + x[i_coord_offset+DIM*0+XX];
 192         iy0              = shY + x[i_coord_offset+DIM*0+YY];
 193         iz0              = shZ + x[i_coord_offset+DIM*0+ZZ];
 194         ix1              = shX + x[i_coord_offset+DIM*1+XX];
 195         iy1              = shY + x[i_coord_offset+DIM*1+YY];
 196         iz1              = shZ + x[i_coord_offset+DIM*1+ZZ];
 197         ix2              = shX + x[i_coord_offset+DIM*2+XX];
 198         iy2              = shY + x[i_coord_offset+DIM*2+YY];
 199         iz2              = shZ + x[i_coord_offset+DIM*2+ZZ];
 200
 201         fix0             = 0.0;
 202         fiy0             = 0.0;
 203         fiz0             = 0.0;
 204         fix1             = 0.0;
 205         fiy1             = 0.0;
 206         fiz1             = 0.0;
 207         fix2             = 0.0;
 208         fiy2             = 0.0;
 209         fiz2             = 0.0;
 210
 211         /* Reset potential sums */
 212         velecsum         = 0.0;
 213         vvdwsum          = 0.0;
 214
 215         /* Start inner kernel loop */
 216         for(jidx=j_index_start; jidx<j_index_end; jidx++)
 217         {
 218             /* Get j neighbor index, and coordinate index */
 219             jnr              = jjnr[jidx];
 220             j_coord_offset   = DIM*jnr;
 221
 222             /* load j atom coordinates */
 223             jx0              = x[j_coord_offset+DIM*0+XX];
 224             jy0              = x[j_coord_offset+DIM*0+YY];
 225             jz0              = x[j_coord_offset+DIM*0+ZZ];
 226             jx1              = x[j_coord_offset+DIM*1+XX];
 227             jy1              = x[j_coord_offset+DIM*1+YY];
 228             jz1              = x[j_coord_offset+DIM*1+ZZ];
 229             jx2              = x[j_coord_offset+DIM*2+XX];
 230             jy2              = x[j_coord_offset+DIM*2+YY];
 231             jz2              = x[j_coord_offset+DIM*2+ZZ];
 232
 233             /* Calculate displacement vector */
 234             dx00             = ix0 - jx0;
 235             dy00             = iy0 - jy0;
 236             dz00             = iz0 - jz0;
 237             dx01             = ix0 - jx1;
 238             dy01             = iy0 - jy1;
 239             dz01             = iz0 - jz1;
 240             dx02             = ix0 - jx2;
 241             dy02             = iy0 - jy2;
 242             dz02             = iz0 - jz2;
 243             dx10             = ix1 - jx0;
 244             dy10             = iy1 - jy0;
 245             dz10             = iz1 - jz0;
 246             dx11             = ix1 - jx1;
 247             dy11             = iy1 - jy1;
 248             dz11             = iz1 - jz1;
 249             dx12             = ix1 - jx2;
 250             dy12             = iy1 - jy2;
 251             dz12             = iz1 - jz2;
 252             dx20             = ix2 - jx0;
 253             dy20             = iy2 - jy0;
 254             dz20             = iz2 - jz0;
 255             dx21             = ix2 - jx1;
 256             dy21             = iy2 - jy1;
 257             dz21             = iz2 - jz1;
 258             dx22             = ix2 - jx2;
 259             dy22             = iy2 - jy2;
 260             dz22             = iz2 - jz2;
 261
 262             /* Calculate squared distance and things based on it */
 263             rsq00            = dx00*dx00+dy00*dy00+dz00*dz00;
 264             rsq01            = dx01*dx01+dy01*dy01+dz01*dz01;
 265             rsq02            = dx02*dx02+dy02*dy02+dz02*dz02;
 266             rsq10            = dx10*dx10+dy10*dy10+dz10*dz10;
 267             rsq11            = dx11*dx11+dy11*dy11+dz11*dz11;
 268             rsq12            = dx12*dx12+dy12*dy12+dz12*dz12;
 269             rsq20            = dx20*dx20+dy20*dy20+dz20*dz20;
 270             rsq21            = dx21*dx21+dy21*dy21+dz21*dz21;
 271             rsq22            = dx22*dx22+dy22*dy22+dz22*dz22;
 272
 273             rinv00           = gmx_invsqrt(rsq00);
 274             rinv01           = gmx_invsqrt(rsq01);
 275             rinv02           = gmx_invsqrt(rsq02);
 276             rinv10           = gmx_invsqrt(rsq10);
 277             rinv11           = gmx_invsqrt(rsq11);
 278             rinv12           = gmx_invsqrt(rsq12);
 279             rinv20           = gmx_invsqrt(rsq20);
 280             rinv21           = gmx_invsqrt(rsq21);
 281             rinv22           = gmx_invsqrt(rsq22);
 282
 283             rinvsq00         = rinv00*rinv00;
 284             rinvsq01         = rinv01*rinv01;
 285             rinvsq02         = rinv02*rinv02;
 286             rinvsq10         = rinv10*rinv10;
 287             rinvsq11         = rinv11*rinv11;
 288             rinvsq12         = rinv12*rinv12;
 289             rinvsq20         = rinv20*rinv20;
 290             rinvsq21         = rinv21*rinv21;
 291             rinvsq22         = rinv22*rinv22;
 292
 293             /**************************
 294              * CALCULATE INTERACTIONS *
 295              **************************/
 296
 297             if (rsq00<rcutoff2)
 298             {
 299
 300             r00              = rsq00*rinv00;
 301
 302             /* EWALD ELECTROSTATICS */
 303
 304             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 305             ewrt             = r00*ewtabscale;
 306             ewitab           = ewrt;
 307             eweps            = ewrt-ewitab;
 308             ewitab           = 4*ewitab;
 309             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 310             velec            = qq00*((rinv00-sh_ewald)-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 311             felec            = qq00*rinv00*(rinvsq00-felec);
 312
 313             rinvsix          = rinvsq00*rinvsq00*rinvsq00;
 314             ewcljrsq         = ewclj2*rsq00;
 315             exponent         = exp(-ewcljrsq);
 316             poly             = exponent*(1.0 + ewcljrsq + ewcljrsq*ewcljrsq*0.5);
 317             vvdw6            = (c6_00-c6grid_00*(1.0-poly))*rinvsix;
 318             vvdw12           = c12_00*rinvsix*rinvsix;
 319             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);
 320             fvdw             = (vvdw12 - vvdw6 - c6grid_00*(1.0/6.0)*exponent*ewclj6)*rinvsq00;
 321
 322             /* Update potential sums from outer loop */
 323             velecsum        += velec;
 324             vvdwsum         += vvdw;
 325
 326             fscal            = felec+fvdw;
 327
 328             /* Calculate temporary vectorial force */
 329             tx               = fscal*dx00;
 330             ty               = fscal*dy00;
 331             tz               = fscal*dz00;
 332
 333             /* Update vectorial force */
 334             fix0            += tx;
 335             fiy0            += ty;
 336             fiz0            += 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             }
 342
 343             /**************************
 344              * CALCULATE INTERACTIONS *
 345              **************************/
 346
 347             if (rsq01<rcutoff2)
 348             {
 349
 350             r01              = rsq01*rinv01;
 351
 352             /* EWALD ELECTROSTATICS */
 353
 354             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 355             ewrt             = r01*ewtabscale;
 356             ewitab           = ewrt;
 357             eweps            = ewrt-ewitab;
 358             ewitab           = 4*ewitab;
 359             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 360             velec            = qq01*((rinv01-sh_ewald)-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 361             felec            = qq01*rinv01*(rinvsq01-felec);
 362
 363             /* Update potential sums from outer loop */
 364             velecsum        += velec;
 365
 366             fscal            = felec;
 367
 368             /* Calculate temporary vectorial force */
 369             tx               = fscal*dx01;
 370             ty               = fscal*dy01;
 371             tz               = fscal*dz01;
 372
 373             /* Update vectorial force */
 374             fix0            += tx;
 375             fiy0            += ty;
 376             fiz0            += tz;
 377             f[j_coord_offset+DIM*1+XX] -= tx;
 378             f[j_coord_offset+DIM*1+YY] -= ty;
 379             f[j_coord_offset+DIM*1+ZZ] -= tz;
 380
 381             }
 382
 383             /**************************
 384              * CALCULATE INTERACTIONS *
 385              **************************/
 386
 387             if (rsq02<rcutoff2)
 388             {
 389
 390             r02              = rsq02*rinv02;
 391
 392             /* EWALD ELECTROSTATICS */
 393
 394             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 395             ewrt             = r02*ewtabscale;
 396             ewitab           = ewrt;
 397             eweps            = ewrt-ewitab;
 398             ewitab           = 4*ewitab;
 399             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 400             velec            = qq02*((rinv02-sh_ewald)-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 401             felec            = qq02*rinv02*(rinvsq02-felec);
 402
 403             /* Update potential sums from outer loop */
 404             velecsum        += velec;
 405
 406             fscal            = felec;
 407
 408             /* Calculate temporary vectorial force */
 409             tx               = fscal*dx02;
 410             ty               = fscal*dy02;
 411             tz               = fscal*dz02;
 412
 413             /* Update vectorial force */
 414             fix0            += tx;
 415             fiy0            += ty;
 416             fiz0            += tz;
 417             f[j_coord_offset+DIM*2+XX] -= tx;
 418             f[j_coord_offset+DIM*2+YY] -= ty;
 419             f[j_coord_offset+DIM*2+ZZ] -= tz;
 420
 421             }
 422
 423             /**************************
 424              * CALCULATE INTERACTIONS *
 425              **************************/
 426
 427             if (rsq10<rcutoff2)
 428             {
 429
 430             r10              = rsq10*rinv10;
 431
 432             /* EWALD ELECTROSTATICS */
 433
 434             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 435             ewrt             = r10*ewtabscale;
 436             ewitab           = ewrt;
 437             eweps            = ewrt-ewitab;
 438             ewitab           = 4*ewitab;
 439             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 440             velec            = qq10*((rinv10-sh_ewald)-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 441             felec            = qq10*rinv10*(rinvsq10-felec);
 442
 443             /* Update potential sums from outer loop */
 444             velecsum        += velec;
 445
 446             fscal            = felec;
 447
 448             /* Calculate temporary vectorial force */
 449             tx               = fscal*dx10;
 450             ty               = fscal*dy10;
 451             tz               = fscal*dz10;
 452
 453             /* Update vectorial force */
 454             fix1            += tx;
 455             fiy1            += ty;
 456             fiz1            += tz;
 457             f[j_coord_offset+DIM*0+XX] -= tx;
 458             f[j_coord_offset+DIM*0+YY] -= ty;
 459             f[j_coord_offset+DIM*0+ZZ] -= tz;
 460
 461             }
 462
 463             /**************************
 464              * CALCULATE INTERACTIONS *
 465              **************************/
 466
 467             if (rsq11<rcutoff2)
 468             {
 469
 470             r11              = rsq11*rinv11;
 471
 472             /* EWALD ELECTROSTATICS */
 473
 474             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 475             ewrt             = r11*ewtabscale;
 476             ewitab           = ewrt;
 477             eweps            = ewrt-ewitab;
 478             ewitab           = 4*ewitab;
 479             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 480             velec            = qq11*((rinv11-sh_ewald)-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 481             felec            = qq11*rinv11*(rinvsq11-felec);
 482
 483             /* Update potential sums from outer loop */
 484             velecsum        += velec;
 485
 486             fscal            = felec;
 487
 488             /* Calculate temporary vectorial force */
 489             tx               = fscal*dx11;
 490             ty               = fscal*dy11;
 491             tz               = fscal*dz11;
 492
 493             /* Update vectorial force */
 494             fix1            += tx;
 495             fiy1            += ty;
 496             fiz1            += tz;
 497             f[j_coord_offset+DIM*1+XX] -= tx;
 498             f[j_coord_offset+DIM*1+YY] -= ty;
 499             f[j_coord_offset+DIM*1+ZZ] -= tz;
 500
 501             }
 502
 503             /**************************
 504              * CALCULATE INTERACTIONS *
 505              **************************/
 506
 507             if (rsq12<rcutoff2)
 508             {
 509
 510             r12              = rsq12*rinv12;
 511
 512             /* EWALD ELECTROSTATICS */
 513
 514             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 515             ewrt             = r12*ewtabscale;
 516             ewitab           = ewrt;
 517             eweps            = ewrt-ewitab;
 518             ewitab           = 4*ewitab;
 519             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 520             velec            = qq12*((rinv12-sh_ewald)-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 521             felec            = qq12*rinv12*(rinvsq12-felec);
 522
 523             /* Update potential sums from outer loop */
 524             velecsum        += velec;
 525
 526             fscal            = felec;
 527
 528             /* Calculate temporary vectorial force */
 529             tx               = fscal*dx12;
 530             ty               = fscal*dy12;
 531             tz               = fscal*dz12;
 532
 533             /* Update vectorial force */
 534             fix1            += tx;
 535             fiy1            += ty;
 536             fiz1            += tz;
 537             f[j_coord_offset+DIM*2+XX] -= tx;
 538             f[j_coord_offset+DIM*2+YY] -= ty;
 539             f[j_coord_offset+DIM*2+ZZ] -= tz;
 540
 541             }
 542
 543             /**************************
 544              * CALCULATE INTERACTIONS *
 545              **************************/
 546
 547             if (rsq20<rcutoff2)
 548             {
 549
 550             r20              = rsq20*rinv20;
 551
 552             /* EWALD ELECTROSTATICS */
 553
 554             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 555             ewrt             = r20*ewtabscale;
 556             ewitab           = ewrt;
 557             eweps            = ewrt-ewitab;
 558             ewitab           = 4*ewitab;
 559             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 560             velec            = qq20*((rinv20-sh_ewald)-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 561             felec            = qq20*rinv20*(rinvsq20-felec);
 562
 563             /* Update potential sums from outer loop */
 564             velecsum        += velec;
 565
 566             fscal            = felec;
 567
 568             /* Calculate temporary vectorial force */
 569             tx               = fscal*dx20;
 570             ty               = fscal*dy20;
 571             tz               = fscal*dz20;
 572
 573             /* Update vectorial force */
 574             fix2            += tx;
 575             fiy2            += ty;
 576             fiz2            += tz;
 577             f[j_coord_offset+DIM*0+XX] -= tx;
 578             f[j_coord_offset+DIM*0+YY] -= ty;
 579             f[j_coord_offset+DIM*0+ZZ] -= tz;
 580
 581             }
 582
 583             /**************************
 584              * CALCULATE INTERACTIONS *
 585              **************************/
 586
 587             if (rsq21<rcutoff2)
 588             {
 589
 590             r21              = rsq21*rinv21;
 591
 592             /* EWALD ELECTROSTATICS */
 593
 594             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 595             ewrt             = r21*ewtabscale;
 596             ewitab           = ewrt;
 597             eweps            = ewrt-ewitab;
 598             ewitab           = 4*ewitab;
 599             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 600             velec            = qq21*((rinv21-sh_ewald)-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 601             felec            = qq21*rinv21*(rinvsq21-felec);
 602
 603             /* Update potential sums from outer loop */
 604             velecsum        += velec;
 605
 606             fscal            = felec;
 607
 608             /* Calculate temporary vectorial force */
 609             tx               = fscal*dx21;
 610             ty               = fscal*dy21;
 611             tz               = fscal*dz21;
 612
 613             /* Update vectorial force */
 614             fix2            += tx;
 615             fiy2            += ty;
 616             fiz2            += tz;
 617             f[j_coord_offset+DIM*1+XX] -= tx;
 618             f[j_coord_offset+DIM*1+YY] -= ty;
 619             f[j_coord_offset+DIM*1+ZZ] -= tz;
 620
 621             }
 622
 623             /**************************
 624              * CALCULATE INTERACTIONS *
 625              **************************/
 626
 627             if (rsq22<rcutoff2)
 628             {
 629
 630             r22              = rsq22*rinv22;
 631
 632             /* EWALD ELECTROSTATICS */
 633
 634             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 635             ewrt             = r22*ewtabscale;
 636             ewitab           = ewrt;
 637             eweps            = ewrt-ewitab;
 638             ewitab           = 4*ewitab;
 639             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 640             velec            = qq22*((rinv22-sh_ewald)-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 641             felec            = qq22*rinv22*(rinvsq22-felec);
 642
 643             /* Update potential sums from outer loop */
 644             velecsum        += velec;
 645
 646             fscal            = felec;
 647
 648             /* Calculate temporary vectorial force */
 649             tx               = fscal*dx22;
 650             ty               = fscal*dy22;
 651             tz               = fscal*dz22;
 652
 653             /* Update vectorial force */
 654             fix2            += tx;
 655             fiy2            += ty;
 656             fiz2            += tz;
 657             f[j_coord_offset+DIM*2+XX] -= tx;
 658             f[j_coord_offset+DIM*2+YY] -= ty;
 659             f[j_coord_offset+DIM*2+ZZ] -= tz;
 660
 661             }
 662
 663             /* Inner loop uses 401 flops */
 664         }
 665         /* End of innermost loop */
 666
 667         tx = ty = tz = 0;
 668         f[i_coord_offset+DIM*0+XX] += fix0;
 669         f[i_coord_offset+DIM*0+YY] += fiy0;
 670         f[i_coord_offset+DIM*0+ZZ] += fiz0;
 671         tx                         += fix0;
 672         ty                         += fiy0;
 673         tz                         += fiz0;
 674         f[i_coord_offset+DIM*1+XX] += fix1;
 675         f[i_coord_offset+DIM*1+YY] += fiy1;
 676         f[i_coord_offset+DIM*1+ZZ] += fiz1;
 677         tx                         += fix1;
 678         ty                         += fiy1;
 679         tz                         += fiz1;
 680         f[i_coord_offset+DIM*2+XX] += fix2;
 681         f[i_coord_offset+DIM*2+YY] += fiy2;
 682         f[i_coord_offset+DIM*2+ZZ] += fiz2;
 683         tx                         += fix2;
 684         ty                         += fiy2;
 685         tz                         += fiz2;
 686         fshift[i_shift_offset+XX]  += tx;
 687         fshift[i_shift_offset+YY]  += ty;
 688         fshift[i_shift_offset+ZZ]  += tz;
 689
 690         ggid                        = gid[iidx];
 691         /* Update potential energies */
 692         kernel_data->energygrp_elec[ggid] += velecsum;
 693         kernel_data->energygrp_vdw[ggid] += vvdwsum;
 694
 695         /* Increment number of inner iterations */
 696         inneriter                  += j_index_end - j_index_start;
 697
 698         /* Outer loop uses 32 flops */
 699     }
 700
 701     /* Increment number of outer iterations */
 702     outeriter        += nri;
 703
 704     /* Update outer/inner flops */
 705
 706     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W3W3_VF,outeriter*32 + inneriter*401);
 707 }
 708 /*
 709  * Gromacs nonbonded kernel:   nb_kernel_ElecEwSh_VdwLJEwSh_GeomW3W3_F_c
 710  * Electrostatics interaction: Ewald
 711  * VdW interaction:            LJEwald
 712  * Geometry:                   Water3-Water3
 713  * Calculate force/pot:        Force
 714  */
 715 void
 716 nb_kernel_ElecEwSh_VdwLJEwSh_GeomW3W3_F_c
 717                     (t_nblist                    * gmx_restrict       nlist,
 718                      rvec                        * gmx_restrict          xx,
 719                      rvec                        * gmx_restrict          ff,
 720                      t_forcerec                  * gmx_restrict          fr,
 721                      t_mdatoms                   * gmx_restrict     mdatoms,
 722                      nb_kernel_data_t gmx_unused * gmx_restrict kernel_data,
 723                      t_nrnb                      * gmx_restrict        nrnb)
 724 {
 725     int              i_shift_offset,i_coord_offset,j_coord_offset;
 726     int              j_index_start,j_index_end;
 727     int              nri,inr,ggid,iidx,jidx,jnr,outeriter,inneriter;
 728     real             shX,shY,shZ,tx,ty,tz,fscal,rcutoff,rcutoff2;
 729     int              *iinr,*jindex,*jjnr,*shiftidx,*gid;
 730     real             *shiftvec,*fshift,*x,*f;
 731     int              vdwioffset0;
 732     real             ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
 733     int              vdwioffset1;
 734     real             ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
 735     int              vdwioffset2;
 736     real             ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
 737     int              vdwjidx0;
 738     real             jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
 739     int              vdwjidx1;
 740     real             jx1,jy1,jz1,fjx1,fjy1,fjz1,jq1,isaj1;
 741     int              vdwjidx2;
 742     real             jx2,jy2,jz2,fjx2,fjy2,fjz2,jq2,isaj2;
 743     real             dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00,cexp1_00,cexp2_00;
 744     real             dx01,dy01,dz01,rsq01,rinv01,rinvsq01,r01,qq01,c6_01,c12_01,cexp1_01,cexp2_01;
 745     real             dx02,dy02,dz02,rsq02,rinv02,rinvsq02,r02,qq02,c6_02,c12_02,cexp1_02,cexp2_02;
 746     real             dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10,cexp1_10,cexp2_10;
 747     real             dx11,dy11,dz11,rsq11,rinv11,rinvsq11,r11,qq11,c6_11,c12_11,cexp1_11,cexp2_11;
 748     real             dx12,dy12,dz12,rsq12,rinv12,rinvsq12,r12,qq12,c6_12,c12_12,cexp1_12,cexp2_12;
 749     real             dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20,cexp1_20,cexp2_20;
 750     real             dx21,dy21,dz21,rsq21,rinv21,rinvsq21,r21,qq21,c6_21,c12_21,cexp1_21,cexp2_21;
 751     real             dx22,dy22,dz22,rsq22,rinv22,rinvsq22,r22,qq22,c6_22,c12_22,cexp1_22,cexp2_22;
 752     real             velec,felec,velecsum,facel,crf,krf,krf2;
 753     real             *charge;
 754     int              nvdwtype;
 755     real             rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,br,vvdwexp,sh_vdw_invrcut6;
 756     int              *vdwtype;
 757     real             *vdwparam;
 758     real             c6grid_00;
 759     real             c6grid_01;
 760     real             c6grid_02;
 761     real             c6grid_10;
 762     real             c6grid_11;
 763     real             c6grid_12;
 764     real             c6grid_20;
 765     real             c6grid_21;
 766     real             c6grid_22;
 767     real             ewclj,ewclj2,ewclj6,ewcljrsq,poly,exponent,sh_lj_ewald;
 768     real             *vdwgridparam;
 769     int              ewitab;
 770     real             ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace;
 771     real             *ewtab;
 772
 773     x                = xx[0];
 774     f                = ff[0];
 775
 776     nri              = nlist->nri;
 777     iinr             = nlist->iinr;
 778     jindex           = nlist->jindex;
 779     jjnr             = nlist->jjnr;
 780     shiftidx         = nlist->shift;
 781     gid              = nlist->gid;
 782     shiftvec         = fr->shift_vec[0];
 783     fshift           = fr->fshift[0];
 784     facel            = fr->epsfac;
 785     charge           = mdatoms->chargeA;
 786     nvdwtype         = fr->ntype;
 787     vdwparam         = fr->nbfp;
 788     vdwtype          = mdatoms->typeA;
 789     vdwgridparam     = fr->ljpme_c6grid;
 790     ewclj            = fr->ewaldcoeff_lj;
 791     sh_lj_ewald      = fr->ic->sh_lj_ewald;
 792     ewclj2           = ewclj*ewclj;
 793     ewclj6           = ewclj2*ewclj2*ewclj2;
 794
 795     sh_ewald         = fr->ic->sh_ewald;
 796     ewtab            = fr->ic->tabq_coul_F;
 797     ewtabscale       = fr->ic->tabq_scale;
 798     ewtabhalfspace   = 0.5/ewtabscale;
 799
 800     /* Setup water-specific parameters */
 801     inr              = nlist->iinr[0];
 802     iq0              = facel*charge[inr+0];
 803     iq1              = facel*charge[inr+1];
 804     iq2              = facel*charge[inr+2];
 805     vdwioffset0      = 2*nvdwtype*vdwtype[inr+0];
 806
 807     jq0              = charge[inr+0];
 808     jq1              = charge[inr+1];
 809     jq2              = charge[inr+2];
 810     vdwjidx0         = 2*vdwtype[inr+0];
 811     qq00             = iq0*jq0;
 812     c6_00            = vdwparam[vdwioffset0+vdwjidx0];
 813     c12_00           = vdwparam[vdwioffset0+vdwjidx0+1];
 814     c6grid_00        = vdwgridparam[vdwioffset0+vdwjidx0];
 815     qq01             = iq0*jq1;
 816     qq02             = iq0*jq2;
 817     qq10             = iq1*jq0;
 818     qq11             = iq1*jq1;
 819     qq12             = iq1*jq2;
 820     qq20             = iq2*jq0;
 821     qq21             = iq2*jq1;
 822     qq22             = iq2*jq2;
 823
 824     /* When we use explicit cutoffs the value must be identical for elec and VdW, so use elec as an arbitrary choice */
 825     rcutoff          = fr->rcoulomb;
 826     rcutoff2         = rcutoff*rcutoff;
 827
 828     sh_vdw_invrcut6  = fr->ic->sh_invrc6;
 829     rvdw             = fr->rvdw;
 830
 831     outeriter        = 0;
 832     inneriter        = 0;
 833
 834     /* Start outer loop over neighborlists */
 835     for(iidx=0; iidx<nri; iidx++)
 836     {
 837         /* Load shift vector for this list */
 838         i_shift_offset   = DIM*shiftidx[iidx];
 839         shX              = shiftvec[i_shift_offset+XX];
 840         shY              = shiftvec[i_shift_offset+YY];
 841         shZ              = shiftvec[i_shift_offset+ZZ];
 842
 843         /* Load limits for loop over neighbors */
 844         j_index_start    = jindex[iidx];
 845         j_index_end      = jindex[iidx+1];
 846
 847         /* Get outer coordinate index */
 848         inr              = iinr[iidx];
 849         i_coord_offset   = DIM*inr;
 850
 851         /* Load i particle coords and add shift vector */
 852         ix0              = shX + x[i_coord_offset+DIM*0+XX];
 853         iy0              = shY + x[i_coord_offset+DIM*0+YY];
 854         iz0              = shZ + x[i_coord_offset+DIM*0+ZZ];
 855         ix1              = shX + x[i_coord_offset+DIM*1+XX];
 856         iy1              = shY + x[i_coord_offset+DIM*1+YY];
 857         iz1              = shZ + x[i_coord_offset+DIM*1+ZZ];
 858         ix2              = shX + x[i_coord_offset+DIM*2+XX];
 859         iy2              = shY + x[i_coord_offset+DIM*2+YY];
 860         iz2              = shZ + x[i_coord_offset+DIM*2+ZZ];
 861
 862         fix0             = 0.0;
 863         fiy0             = 0.0;
 864         fiz0             = 0.0;
 865         fix1             = 0.0;
 866         fiy1             = 0.0;
 867         fiz1             = 0.0;
 868         fix2             = 0.0;
 869         fiy2             = 0.0;
 870         fiz2             = 0.0;
 871
 872         /* Start inner kernel loop */
 873         for(jidx=j_index_start; jidx<j_index_end; jidx++)
 874         {
 875             /* Get j neighbor index, and coordinate index */
 876             jnr              = jjnr[jidx];
 877             j_coord_offset   = DIM*jnr;
 878
 879             /* load j atom coordinates */
 880             jx0              = x[j_coord_offset+DIM*0+XX];
 881             jy0              = x[j_coord_offset+DIM*0+YY];
 882             jz0              = x[j_coord_offset+DIM*0+ZZ];
 883             jx1              = x[j_coord_offset+DIM*1+XX];
 884             jy1              = x[j_coord_offset+DIM*1+YY];
 885             jz1              = x[j_coord_offset+DIM*1+ZZ];
 886             jx2              = x[j_coord_offset+DIM*2+XX];
 887             jy2              = x[j_coord_offset+DIM*2+YY];
 888             jz2              = x[j_coord_offset+DIM*2+ZZ];
 889
 890             /* Calculate displacement vector */
 891             dx00             = ix0 - jx0;
 892             dy00             = iy0 - jy0;
 893             dz00             = iz0 - jz0;
 894             dx01             = ix0 - jx1;
 895             dy01             = iy0 - jy1;
 896             dz01             = iz0 - jz1;
 897             dx02             = ix0 - jx2;
 898             dy02             = iy0 - jy2;
 899             dz02             = iz0 - jz2;
 900             dx10             = ix1 - jx0;
 901             dy10             = iy1 - jy0;
 902             dz10             = iz1 - jz0;
 903             dx11             = ix1 - jx1;
 904             dy11             = iy1 - jy1;
 905             dz11             = iz1 - jz1;
 906             dx12             = ix1 - jx2;
 907             dy12             = iy1 - jy2;
 908             dz12             = iz1 - jz2;
 909             dx20             = ix2 - jx0;
 910             dy20             = iy2 - jy0;
 911             dz20             = iz2 - jz0;
 912             dx21             = ix2 - jx1;
 913             dy21             = iy2 - jy1;
 914             dz21             = iz2 - jz1;
 915             dx22             = ix2 - jx2;
 916             dy22             = iy2 - jy2;
 917             dz22             = iz2 - jz2;
 918
 919             /* Calculate squared distance and things based on it */
 920             rsq00            = dx00*dx00+dy00*dy00+dz00*dz00;
 921             rsq01            = dx01*dx01+dy01*dy01+dz01*dz01;
 922             rsq02            = dx02*dx02+dy02*dy02+dz02*dz02;
 923             rsq10            = dx10*dx10+dy10*dy10+dz10*dz10;
 924             rsq11            = dx11*dx11+dy11*dy11+dz11*dz11;
 925             rsq12            = dx12*dx12+dy12*dy12+dz12*dz12;
 926             rsq20            = dx20*dx20+dy20*dy20+dz20*dz20;
 927             rsq21            = dx21*dx21+dy21*dy21+dz21*dz21;
 928             rsq22            = dx22*dx22+dy22*dy22+dz22*dz22;
 929
 930             rinv00           = gmx_invsqrt(rsq00);
 931             rinv01           = gmx_invsqrt(rsq01);
 932             rinv02           = gmx_invsqrt(rsq02);
 933             rinv10           = gmx_invsqrt(rsq10);
 934             rinv11           = gmx_invsqrt(rsq11);
 935             rinv12           = gmx_invsqrt(rsq12);
 936             rinv20           = gmx_invsqrt(rsq20);
 937             rinv21           = gmx_invsqrt(rsq21);
 938             rinv22           = gmx_invsqrt(rsq22);
 939
 940             rinvsq00         = rinv00*rinv00;
 941             rinvsq01         = rinv01*rinv01;
 942             rinvsq02         = rinv02*rinv02;
 943             rinvsq10         = rinv10*rinv10;
 944             rinvsq11         = rinv11*rinv11;
 945             rinvsq12         = rinv12*rinv12;
 946             rinvsq20         = rinv20*rinv20;
 947             rinvsq21         = rinv21*rinv21;
 948             rinvsq22         = rinv22*rinv22;
 949
 950             /**************************
 951              * CALCULATE INTERACTIONS *
 952              **************************/
 953
 954             if (rsq00<rcutoff2)
 955             {
 956
 957             r00              = rsq00*rinv00;
 958
 959             /* EWALD ELECTROSTATICS */
 960
 961             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 962             ewrt             = r00*ewtabscale;
 963             ewitab           = ewrt;
 964             eweps            = ewrt-ewitab;
 965             felec            = (1.0-eweps)*ewtab[ewitab]+eweps*ewtab[ewitab+1];
 966             felec            = qq00*rinv00*(rinvsq00-felec);
 967
 968             rinvsix          = rinvsq00*rinvsq00*rinvsq00;
 969             ewcljrsq         = ewclj2*rsq00;
 970             exponent         = exp(-ewcljrsq);
 971             poly             = exponent*(1.0 + ewcljrsq + ewcljrsq*ewcljrsq*0.5);
 972             fvdw             = (((c12_00*rinvsix - c6_00 + c6grid_00*(1.0-poly))*rinvsix) - c6grid_00*(1.0/6.0)*exponent*ewclj6)*rinvsq00;
 973
 974             fscal            = felec+fvdw;
 975
 976             /* Calculate temporary vectorial force */
 977             tx               = fscal*dx00;
 978             ty               = fscal*dy00;
 979             tz               = fscal*dz00;
 980
 981             /* Update vectorial force */
 982             fix0            += tx;
 983             fiy0            += ty;
 984             fiz0            += tz;
 985             f[j_coord_offset+DIM*0+XX] -= tx;
 986             f[j_coord_offset+DIM*0+YY] -= ty;
 987             f[j_coord_offset+DIM*0+ZZ] -= tz;
 988
 989             }
 990
 991             /**************************
 992              * CALCULATE INTERACTIONS *
 993              **************************/
 994
 995             if (rsq01<rcutoff2)
 996             {
 997
 998             r01              = rsq01*rinv01;
 999
1000             /* EWALD ELECTROSTATICS */
1001
1002             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1003             ewrt             = r01*ewtabscale;
1004             ewitab           = ewrt;
1005             eweps            = ewrt-ewitab;
1006             felec            = (1.0-eweps)*ewtab[ewitab]+eweps*ewtab[ewitab+1];
1007             felec            = qq01*rinv01*(rinvsq01-felec);
1008
1009             fscal            = felec;
1010
1011             /* Calculate temporary vectorial force */
1012             tx               = fscal*dx01;
1013             ty               = fscal*dy01;
1014             tz               = fscal*dz01;
1015
1016             /* Update vectorial force */
1017             fix0            += tx;
1018             fiy0            += ty;
1019             fiz0            += tz;
1020             f[j_coord_offset+DIM*1+XX] -= tx;
1021             f[j_coord_offset+DIM*1+YY] -= ty;
1022             f[j_coord_offset+DIM*1+ZZ] -= tz;
1023
1024             }
1025
1026             /**************************
1027              * CALCULATE INTERACTIONS *
1028              **************************/
1029
1030             if (rsq02<rcutoff2)
1031             {
1032
1033             r02              = rsq02*rinv02;
1034
1035             /* EWALD ELECTROSTATICS */
1036
1037             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1038             ewrt             = r02*ewtabscale;
1039             ewitab           = ewrt;
1040             eweps            = ewrt-ewitab;
1041             felec            = (1.0-eweps)*ewtab[ewitab]+eweps*ewtab[ewitab+1];
1042             felec            = qq02*rinv02*(rinvsq02-felec);
1043
1044             fscal            = felec;
1045
1046             /* Calculate temporary vectorial force */
1047             tx               = fscal*dx02;
1048             ty               = fscal*dy02;
1049             tz               = fscal*dz02;
1050
1051             /* Update vectorial force */
1052             fix0            += tx;
1053             fiy0            += ty;
1054             fiz0            += tz;
1055             f[j_coord_offset+DIM*2+XX] -= tx;
1056             f[j_coord_offset+DIM*2+YY] -= ty;
1057             f[j_coord_offset+DIM*2+ZZ] -= tz;
1058
1059             }
1060
1061             /**************************
1062              * CALCULATE INTERACTIONS *
1063              **************************/
1064
1065             if (rsq10<rcutoff2)
1066             {
1067
1068             r10              = rsq10*rinv10;
1069
1070             /* EWALD ELECTROSTATICS */
1071
1072             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1073             ewrt             = r10*ewtabscale;
1074             ewitab           = ewrt;
1075             eweps            = ewrt-ewitab;
1076             felec            = (1.0-eweps)*ewtab[ewitab]+eweps*ewtab[ewitab+1];
1077             felec            = qq10*rinv10*(rinvsq10-felec);
1078
1079             fscal            = felec;
1080
1081             /* Calculate temporary vectorial force */
1082             tx               = fscal*dx10;
1083             ty               = fscal*dy10;
1084             tz               = fscal*dz10;
1085
1086             /* Update vectorial force */
1087             fix1            += tx;
1088             fiy1            += ty;
1089             fiz1            += tz;
1090             f[j_coord_offset+DIM*0+XX] -= tx;
1091             f[j_coord_offset+DIM*0+YY] -= ty;
1092             f[j_coord_offset+DIM*0+ZZ] -= tz;
1093
1094             }
1095
1096             /**************************
1097              * CALCULATE INTERACTIONS *
1098              **************************/
1099
1100             if (rsq11<rcutoff2)
1101             {
1102
1103             r11              = rsq11*rinv11;
1104
1105             /* EWALD ELECTROSTATICS */
1106
1107             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1108             ewrt             = r11*ewtabscale;
1109             ewitab           = ewrt;
1110             eweps            = ewrt-ewitab;
1111             felec            = (1.0-eweps)*ewtab[ewitab]+eweps*ewtab[ewitab+1];
1112             felec            = qq11*rinv11*(rinvsq11-felec);
1113
1114             fscal            = felec;
1115
1116             /* Calculate temporary vectorial force */
1117             tx               = fscal*dx11;
1118             ty               = fscal*dy11;
1119             tz               = fscal*dz11;
1120
1121             /* Update vectorial force */
1122             fix1            += tx;
1123             fiy1            += ty;
1124             fiz1            += tz;
1125             f[j_coord_offset+DIM*1+XX] -= tx;
1126             f[j_coord_offset+DIM*1+YY] -= ty;
1127             f[j_coord_offset+DIM*1+ZZ] -= tz;
1128
1129             }
1130
1131             /**************************
1132              * CALCULATE INTERACTIONS *
1133              **************************/
1134
1135             if (rsq12<rcutoff2)
1136             {
1137
1138             r12              = rsq12*rinv12;
1139
1140             /* EWALD ELECTROSTATICS */
1141
1142             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1143             ewrt             = r12*ewtabscale;
1144             ewitab           = ewrt;
1145             eweps            = ewrt-ewitab;
1146             felec            = (1.0-eweps)*ewtab[ewitab]+eweps*ewtab[ewitab+1];
1147             felec            = qq12*rinv12*(rinvsq12-felec);
1148
1149             fscal            = felec;
1150
1151             /* Calculate temporary vectorial force */
1152             tx               = fscal*dx12;
1153             ty               = fscal*dy12;
1154             tz               = fscal*dz12;
1155
1156             /* Update vectorial force */
1157             fix1            += tx;
1158             fiy1            += ty;
1159             fiz1            += tz;
1160             f[j_coord_offset+DIM*2+XX] -= tx;
1161             f[j_coord_offset+DIM*2+YY] -= ty;
1162             f[j_coord_offset+DIM*2+ZZ] -= tz;
1163
1164             }
1165
1166             /**************************
1167              * CALCULATE INTERACTIONS *
1168              **************************/
1169
1170             if (rsq20<rcutoff2)
1171             {
1172
1173             r20              = rsq20*rinv20;
1174
1175             /* EWALD ELECTROSTATICS */
1176
1177             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1178             ewrt             = r20*ewtabscale;
1179             ewitab           = ewrt;
1180             eweps            = ewrt-ewitab;
1181             felec            = (1.0-eweps)*ewtab[ewitab]+eweps*ewtab[ewitab+1];
1182             felec            = qq20*rinv20*(rinvsq20-felec);
1183
1184             fscal            = felec;
1185
1186             /* Calculate temporary vectorial force */
1187             tx               = fscal*dx20;
1188             ty               = fscal*dy20;
1189             tz               = fscal*dz20;
1190
1191             /* Update vectorial force */
1192             fix2            += tx;
1193             fiy2            += ty;
1194             fiz2            += tz;
1195             f[j_coord_offset+DIM*0+XX] -= tx;
1196             f[j_coord_offset+DIM*0+YY] -= ty;
1197             f[j_coord_offset+DIM*0+ZZ] -= tz;
1198
1199             }
1200
1201             /**************************
1202              * CALCULATE INTERACTIONS *
1203              **************************/
1204
1205             if (rsq21<rcutoff2)
1206             {
1207
1208             r21              = rsq21*rinv21;
1209
1210             /* EWALD ELECTROSTATICS */
1211
1212             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1213             ewrt             = r21*ewtabscale;
1214             ewitab           = ewrt;
1215             eweps            = ewrt-ewitab;
1216             felec            = (1.0-eweps)*ewtab[ewitab]+eweps*ewtab[ewitab+1];
1217             felec            = qq21*rinv21*(rinvsq21-felec);
1218
1219             fscal            = felec;
1220
1221             /* Calculate temporary vectorial force */
1222             tx               = fscal*dx21;
1223             ty               = fscal*dy21;
1224             tz               = fscal*dz21;
1225
1226             /* Update vectorial force */
1227             fix2            += tx;
1228             fiy2            += ty;
1229             fiz2            += tz;
1230             f[j_coord_offset+DIM*1+XX] -= tx;
1231             f[j_coord_offset+DIM*1+YY] -= ty;
1232             f[j_coord_offset+DIM*1+ZZ] -= tz;
1233
1234             }
1235
1236             /**************************
1237              * CALCULATE INTERACTIONS *
1238              **************************/
1239
1240             if (rsq22<rcutoff2)
1241             {
1242
1243             r22              = rsq22*rinv22;
1244
1245             /* EWALD ELECTROSTATICS */
1246
1247             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1248             ewrt             = r22*ewtabscale;
1249             ewitab           = ewrt;
1250             eweps            = ewrt-ewitab;
1251             felec            = (1.0-eweps)*ewtab[ewitab]+eweps*ewtab[ewitab+1];
1252             felec            = qq22*rinv22*(rinvsq22-felec);
1253
1254             fscal            = felec;
1255
1256             /* Calculate temporary vectorial force */
1257             tx               = fscal*dx22;
1258             ty               = fscal*dy22;
1259             tz               = fscal*dz22;
1260
1261             /* Update vectorial force */
1262             fix2            += tx;
1263             fiy2            += ty;
1264             fiz2            += tz;
1265             f[j_coord_offset+DIM*2+XX] -= tx;
1266             f[j_coord_offset+DIM*2+YY] -= ty;
1267             f[j_coord_offset+DIM*2+ZZ] -= tz;
1268
1269             }
1270
1271             /* Inner loop uses 318 flops */
1272         }
1273         /* End of innermost loop */
1274
1275         tx = ty = tz = 0;
1276         f[i_coord_offset+DIM*0+XX] += fix0;
1277         f[i_coord_offset+DIM*0+YY] += fiy0;
1278         f[i_coord_offset+DIM*0+ZZ] += fiz0;
1279         tx                         += fix0;
1280         ty                         += fiy0;
1281         tz                         += fiz0;
1282         f[i_coord_offset+DIM*1+XX] += fix1;
1283         f[i_coord_offset+DIM*1+YY] += fiy1;
1284         f[i_coord_offset+DIM*1+ZZ] += fiz1;
1285         tx                         += fix1;
1286         ty                         += fiy1;
1287         tz                         += fiz1;
1288         f[i_coord_offset+DIM*2+XX] += fix2;
1289         f[i_coord_offset+DIM*2+YY] += fiy2;
1290         f[i_coord_offset+DIM*2+ZZ] += fiz2;
1291         tx                         += fix2;
1292         ty                         += fiy2;
1293         tz                         += fiz2;
1294         fshift[i_shift_offset+XX]  += tx;
1295         fshift[i_shift_offset+YY]  += ty;
1296         fshift[i_shift_offset+ZZ]  += tz;
1297
1298         /* Increment number of inner iterations */
1299         inneriter                  += j_index_end - j_index_start;
1300
1301         /* Outer loop uses 30 flops */
1302     }
1303
1304     /* Increment number of outer iterations */
1305     outeriter        += nri;
1306
1307     /* Update outer/inner flops */
1308
1309     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W3W3_F,outeriter*30 + inneriter*318);
1310 }