src/gromacs/gmxlib/nonbonded/nb_kernel_c/nb_kernel_ElecEwSh_VdwLJEwSh_GeomW4W4_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_GeomW4W4_VF_c
  51  * Electrostatics interaction: Ewald
  52  * VdW interaction:            LJEwald
  53  * Geometry:                   Water4-Water4
  54  * Calculate force/pot:        PotentialAndForce
  55  */
  56 void
  57 nb_kernel_ElecEwSh_VdwLJEwSh_GeomW4W4_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     int              vdwjidx1;
  83     real             jx1,jy1,jz1,fjx1,fjy1,fjz1,jq1,isaj1;
  84     int              vdwjidx2;
  85     real             jx2,jy2,jz2,fjx2,fjy2,fjz2,jq2,isaj2;
  86     int              vdwjidx3;
  87     real             jx3,jy3,jz3,fjx3,fjy3,fjz3,jq3,isaj3;
  88     real             dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00,cexp1_00,cexp2_00;
  89     real             dx11,dy11,dz11,rsq11,rinv11,rinvsq11,r11,qq11,c6_11,c12_11,cexp1_11,cexp2_11;
  90     real             dx12,dy12,dz12,rsq12,rinv12,rinvsq12,r12,qq12,c6_12,c12_12,cexp1_12,cexp2_12;
  91     real             dx13,dy13,dz13,rsq13,rinv13,rinvsq13,r13,qq13,c6_13,c12_13,cexp1_13,cexp2_13;
  92     real             dx21,dy21,dz21,rsq21,rinv21,rinvsq21,r21,qq21,c6_21,c12_21,cexp1_21,cexp2_21;
  93     real             dx22,dy22,dz22,rsq22,rinv22,rinvsq22,r22,qq22,c6_22,c12_22,cexp1_22,cexp2_22;
  94     real             dx23,dy23,dz23,rsq23,rinv23,rinvsq23,r23,qq23,c6_23,c12_23,cexp1_23,cexp2_23;
  95     real             dx31,dy31,dz31,rsq31,rinv31,rinvsq31,r31,qq31,c6_31,c12_31,cexp1_31,cexp2_31;
  96     real             dx32,dy32,dz32,rsq32,rinv32,rinvsq32,r32,qq32,c6_32,c12_32,cexp1_32,cexp2_32;
  97     real             dx33,dy33,dz33,rsq33,rinv33,rinvsq33,r33,qq33,c6_33,c12_33,cexp1_33,cexp2_33;
  98     real             velec,felec,velecsum,facel,crf,krf,krf2;
  99     real             *charge;
 100     int              nvdwtype;
 101     real             rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,br,vvdwexp,sh_vdw_invrcut6;
 102     int              *vdwtype;
 103     real             *vdwparam;
 104     real             c6grid_00;
 105     real             c6grid_11;
 106     real             c6grid_12;
 107     real             c6grid_13;
 108     real             c6grid_21;
 109     real             c6grid_22;
 110     real             c6grid_23;
 111     real             c6grid_31;
 112     real             c6grid_32;
 113     real             c6grid_33;
 114     real             ewclj,ewclj2,ewclj6,ewcljrsq,poly,exponent,sh_lj_ewald;
 115     real             *vdwgridparam;
 116     int              ewitab;
 117     real             ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace;
 118     real             *ewtab;
 119
 120     x                = xx[0];
 121     f                = ff[0];
 122
 123     nri              = nlist->nri;
 124     iinr             = nlist->iinr;
 125     jindex           = nlist->jindex;
 126     jjnr             = nlist->jjnr;
 127     shiftidx         = nlist->shift;
 128     gid              = nlist->gid;
 129     shiftvec         = fr->shift_vec[0];
 130     fshift           = fr->fshift[0];
 131     facel            = fr->epsfac;
 132     charge           = mdatoms->chargeA;
 133     nvdwtype         = fr->ntype;
 134     vdwparam         = fr->nbfp;
 135     vdwtype          = mdatoms->typeA;
 136     vdwgridparam     = fr->ljpme_c6grid;
 137     ewclj            = fr->ewaldcoeff_lj;
 138     sh_lj_ewald      = fr->ic->sh_lj_ewald;
 139     ewclj2           = ewclj*ewclj;
 140     ewclj6           = ewclj2*ewclj2*ewclj2;
 141
 142     sh_ewald         = fr->ic->sh_ewald;
 143     ewtab            = fr->ic->tabq_coul_FDV0;
 144     ewtabscale       = fr->ic->tabq_scale;
 145     ewtabhalfspace   = 0.5/ewtabscale;
 146
 147     /* Setup water-specific parameters */
 148     inr              = nlist->iinr[0];
 149     iq1              = facel*charge[inr+1];
 150     iq2              = facel*charge[inr+2];
 151     iq3              = facel*charge[inr+3];
 152     vdwioffset0      = 2*nvdwtype*vdwtype[inr+0];
 153
 154     jq1              = charge[inr+1];
 155     jq2              = charge[inr+2];
 156     jq3              = charge[inr+3];
 157     vdwjidx0         = 2*vdwtype[inr+0];
 158     c6_00            = vdwparam[vdwioffset0+vdwjidx0];
 159     c12_00           = vdwparam[vdwioffset0+vdwjidx0+1];
 160     c6grid_00        = vdwgridparam[vdwioffset0+vdwjidx0];
 161     qq11             = iq1*jq1;
 162     qq12             = iq1*jq2;
 163     qq13             = iq1*jq3;
 164     qq21             = iq2*jq1;
 165     qq22             = iq2*jq2;
 166     qq23             = iq2*jq3;
 167     qq31             = iq3*jq1;
 168     qq32             = iq3*jq2;
 169     qq33             = iq3*jq3;
 170
 171     /* When we use explicit cutoffs the value must be identical for elec and VdW, so use elec as an arbitrary choice */
 172     rcutoff          = fr->rcoulomb;
 173     rcutoff2         = rcutoff*rcutoff;
 174
 175     sh_vdw_invrcut6  = fr->ic->sh_invrc6;
 176     rvdw             = fr->rvdw;
 177
 178     outeriter        = 0;
 179     inneriter        = 0;
 180
 181     /* Start outer loop over neighborlists */
 182     for(iidx=0; iidx<nri; iidx++)
 183     {
 184         /* Load shift vector for this list */
 185         i_shift_offset   = DIM*shiftidx[iidx];
 186         shX              = shiftvec[i_shift_offset+XX];
 187         shY              = shiftvec[i_shift_offset+YY];
 188         shZ              = shiftvec[i_shift_offset+ZZ];
 189
 190         /* Load limits for loop over neighbors */
 191         j_index_start    = jindex[iidx];
 192         j_index_end      = jindex[iidx+1];
 193
 194         /* Get outer coordinate index */
 195         inr              = iinr[iidx];
 196         i_coord_offset   = DIM*inr;
 197
 198         /* Load i particle coords and add shift vector */
 199         ix0              = shX + x[i_coord_offset+DIM*0+XX];
 200         iy0              = shY + x[i_coord_offset+DIM*0+YY];
 201         iz0              = shZ + x[i_coord_offset+DIM*0+ZZ];
 202         ix1              = shX + x[i_coord_offset+DIM*1+XX];
 203         iy1              = shY + x[i_coord_offset+DIM*1+YY];
 204         iz1              = shZ + x[i_coord_offset+DIM*1+ZZ];
 205         ix2              = shX + x[i_coord_offset+DIM*2+XX];
 206         iy2              = shY + x[i_coord_offset+DIM*2+YY];
 207         iz2              = shZ + x[i_coord_offset+DIM*2+ZZ];
 208         ix3              = shX + x[i_coord_offset+DIM*3+XX];
 209         iy3              = shY + x[i_coord_offset+DIM*3+YY];
 210         iz3              = shZ + x[i_coord_offset+DIM*3+ZZ];
 211
 212         fix0             = 0.0;
 213         fiy0             = 0.0;
 214         fiz0             = 0.0;
 215         fix1             = 0.0;
 216         fiy1             = 0.0;
 217         fiz1             = 0.0;
 218         fix2             = 0.0;
 219         fiy2             = 0.0;
 220         fiz2             = 0.0;
 221         fix3             = 0.0;
 222         fiy3             = 0.0;
 223         fiz3             = 0.0;
 224
 225         /* Reset potential sums */
 226         velecsum         = 0.0;
 227         vvdwsum          = 0.0;
 228
 229         /* Start inner kernel loop */
 230         for(jidx=j_index_start; jidx<j_index_end; jidx++)
 231         {
 232             /* Get j neighbor index, and coordinate index */
 233             jnr              = jjnr[jidx];
 234             j_coord_offset   = DIM*jnr;
 235
 236             /* load j atom coordinates */
 237             jx0              = x[j_coord_offset+DIM*0+XX];
 238             jy0              = x[j_coord_offset+DIM*0+YY];
 239             jz0              = x[j_coord_offset+DIM*0+ZZ];
 240             jx1              = x[j_coord_offset+DIM*1+XX];
 241             jy1              = x[j_coord_offset+DIM*1+YY];
 242             jz1              = x[j_coord_offset+DIM*1+ZZ];
 243             jx2              = x[j_coord_offset+DIM*2+XX];
 244             jy2              = x[j_coord_offset+DIM*2+YY];
 245             jz2              = x[j_coord_offset+DIM*2+ZZ];
 246             jx3              = x[j_coord_offset+DIM*3+XX];
 247             jy3              = x[j_coord_offset+DIM*3+YY];
 248             jz3              = x[j_coord_offset+DIM*3+ZZ];
 249
 250             /* Calculate displacement vector */
 251             dx00             = ix0 - jx0;
 252             dy00             = iy0 - jy0;
 253             dz00             = iz0 - jz0;
 254             dx11             = ix1 - jx1;
 255             dy11             = iy1 - jy1;
 256             dz11             = iz1 - jz1;
 257             dx12             = ix1 - jx2;
 258             dy12             = iy1 - jy2;
 259             dz12             = iz1 - jz2;
 260             dx13             = ix1 - jx3;
 261             dy13             = iy1 - jy3;
 262             dz13             = iz1 - jz3;
 263             dx21             = ix2 - jx1;
 264             dy21             = iy2 - jy1;
 265             dz21             = iz2 - jz1;
 266             dx22             = ix2 - jx2;
 267             dy22             = iy2 - jy2;
 268             dz22             = iz2 - jz2;
 269             dx23             = ix2 - jx3;
 270             dy23             = iy2 - jy3;
 271             dz23             = iz2 - jz3;
 272             dx31             = ix3 - jx1;
 273             dy31             = iy3 - jy1;
 274             dz31             = iz3 - jz1;
 275             dx32             = ix3 - jx2;
 276             dy32             = iy3 - jy2;
 277             dz32             = iz3 - jz2;
 278             dx33             = ix3 - jx3;
 279             dy33             = iy3 - jy3;
 280             dz33             = iz3 - jz3;
 281
 282             /* Calculate squared distance and things based on it */
 283             rsq00            = dx00*dx00+dy00*dy00+dz00*dz00;
 284             rsq11            = dx11*dx11+dy11*dy11+dz11*dz11;
 285             rsq12            = dx12*dx12+dy12*dy12+dz12*dz12;
 286             rsq13            = dx13*dx13+dy13*dy13+dz13*dz13;
 287             rsq21            = dx21*dx21+dy21*dy21+dz21*dz21;
 288             rsq22            = dx22*dx22+dy22*dy22+dz22*dz22;
 289             rsq23            = dx23*dx23+dy23*dy23+dz23*dz23;
 290             rsq31            = dx31*dx31+dy31*dy31+dz31*dz31;
 291             rsq32            = dx32*dx32+dy32*dy32+dz32*dz32;
 292             rsq33            = dx33*dx33+dy33*dy33+dz33*dz33;
 293
 294             rinv00           = gmx_invsqrt(rsq00);
 295             rinv11           = gmx_invsqrt(rsq11);
 296             rinv12           = gmx_invsqrt(rsq12);
 297             rinv13           = gmx_invsqrt(rsq13);
 298             rinv21           = gmx_invsqrt(rsq21);
 299             rinv22           = gmx_invsqrt(rsq22);
 300             rinv23           = gmx_invsqrt(rsq23);
 301             rinv31           = gmx_invsqrt(rsq31);
 302             rinv32           = gmx_invsqrt(rsq32);
 303             rinv33           = gmx_invsqrt(rsq33);
 304
 305             rinvsq00         = rinv00*rinv00;
 306             rinvsq11         = rinv11*rinv11;
 307             rinvsq12         = rinv12*rinv12;
 308             rinvsq13         = rinv13*rinv13;
 309             rinvsq21         = rinv21*rinv21;
 310             rinvsq22         = rinv22*rinv22;
 311             rinvsq23         = rinv23*rinv23;
 312             rinvsq31         = rinv31*rinv31;
 313             rinvsq32         = rinv32*rinv32;
 314             rinvsq33         = rinv33*rinv33;
 315
 316             /**************************
 317              * CALCULATE INTERACTIONS *
 318              **************************/
 319
 320             if (rsq00<rcutoff2)
 321             {
 322
 323             r00              = rsq00*rinv00;
 324
 325             rinvsix          = rinvsq00*rinvsq00*rinvsq00;
 326             ewcljrsq         = ewclj2*rsq00;
 327             exponent         = exp(-ewcljrsq);
 328             poly             = exponent*(1.0 + ewcljrsq + ewcljrsq*ewcljrsq*0.5);
 329             vvdw6            = (c6_00-c6grid_00*(1.0-poly))*rinvsix;
 330             vvdw12           = c12_00*rinvsix*rinvsix;
 331             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);
 332             fvdw             = (vvdw12 - vvdw6 - c6grid_00*(1.0/6.0)*exponent*ewclj6)*rinvsq00;
 333
 334             /* Update potential sums from outer loop */
 335             vvdwsum         += vvdw;
 336
 337             fscal            = fvdw;
 338
 339             /* Calculate temporary vectorial force */
 340             tx               = fscal*dx00;
 341             ty               = fscal*dy00;
 342             tz               = fscal*dz00;
 343
 344             /* Update vectorial force */
 345             fix0            += tx;
 346             fiy0            += ty;
 347             fiz0            += tz;
 348             f[j_coord_offset+DIM*0+XX] -= tx;
 349             f[j_coord_offset+DIM*0+YY] -= ty;
 350             f[j_coord_offset+DIM*0+ZZ] -= tz;
 351
 352             }
 353
 354             /**************************
 355              * CALCULATE INTERACTIONS *
 356              **************************/
 357
 358             if (rsq11<rcutoff2)
 359             {
 360
 361             r11              = rsq11*rinv11;
 362
 363             /* EWALD ELECTROSTATICS */
 364
 365             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 366             ewrt             = r11*ewtabscale;
 367             ewitab           = ewrt;
 368             eweps            = ewrt-ewitab;
 369             ewitab           = 4*ewitab;
 370             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 371             velec            = qq11*((rinv11-sh_ewald)-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 372             felec            = qq11*rinv11*(rinvsq11-felec);
 373
 374             /* Update potential sums from outer loop */
 375             velecsum        += velec;
 376
 377             fscal            = felec;
 378
 379             /* Calculate temporary vectorial force */
 380             tx               = fscal*dx11;
 381             ty               = fscal*dy11;
 382             tz               = fscal*dz11;
 383
 384             /* Update vectorial force */
 385             fix1            += tx;
 386             fiy1            += ty;
 387             fiz1            += tz;
 388             f[j_coord_offset+DIM*1+XX] -= tx;
 389             f[j_coord_offset+DIM*1+YY] -= ty;
 390             f[j_coord_offset+DIM*1+ZZ] -= tz;
 391
 392             }
 393
 394             /**************************
 395              * CALCULATE INTERACTIONS *
 396              **************************/
 397
 398             if (rsq12<rcutoff2)
 399             {
 400
 401             r12              = rsq12*rinv12;
 402
 403             /* EWALD ELECTROSTATICS */
 404
 405             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 406             ewrt             = r12*ewtabscale;
 407             ewitab           = ewrt;
 408             eweps            = ewrt-ewitab;
 409             ewitab           = 4*ewitab;
 410             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 411             velec            = qq12*((rinv12-sh_ewald)-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 412             felec            = qq12*rinv12*(rinvsq12-felec);
 413
 414             /* Update potential sums from outer loop */
 415             velecsum        += velec;
 416
 417             fscal            = felec;
 418
 419             /* Calculate temporary vectorial force */
 420             tx               = fscal*dx12;
 421             ty               = fscal*dy12;
 422             tz               = fscal*dz12;
 423
 424             /* Update vectorial force */
 425             fix1            += tx;
 426             fiy1            += ty;
 427             fiz1            += tz;
 428             f[j_coord_offset+DIM*2+XX] -= tx;
 429             f[j_coord_offset+DIM*2+YY] -= ty;
 430             f[j_coord_offset+DIM*2+ZZ] -= tz;
 431
 432             }
 433
 434             /**************************
 435              * CALCULATE INTERACTIONS *
 436              **************************/
 437
 438             if (rsq13<rcutoff2)
 439             {
 440
 441             r13              = rsq13*rinv13;
 442
 443             /* EWALD ELECTROSTATICS */
 444
 445             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 446             ewrt             = r13*ewtabscale;
 447             ewitab           = ewrt;
 448             eweps            = ewrt-ewitab;
 449             ewitab           = 4*ewitab;
 450             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 451             velec            = qq13*((rinv13-sh_ewald)-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 452             felec            = qq13*rinv13*(rinvsq13-felec);
 453
 454             /* Update potential sums from outer loop */
 455             velecsum        += velec;
 456
 457             fscal            = felec;
 458
 459             /* Calculate temporary vectorial force */
 460             tx               = fscal*dx13;
 461             ty               = fscal*dy13;
 462             tz               = fscal*dz13;
 463
 464             /* Update vectorial force */
 465             fix1            += tx;
 466             fiy1            += ty;
 467             fiz1            += tz;
 468             f[j_coord_offset+DIM*3+XX] -= tx;
 469             f[j_coord_offset+DIM*3+YY] -= ty;
 470             f[j_coord_offset+DIM*3+ZZ] -= tz;
 471
 472             }
 473
 474             /**************************
 475              * CALCULATE INTERACTIONS *
 476              **************************/
 477
 478             if (rsq21<rcutoff2)
 479             {
 480
 481             r21              = rsq21*rinv21;
 482
 483             /* EWALD ELECTROSTATICS */
 484
 485             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 486             ewrt             = r21*ewtabscale;
 487             ewitab           = ewrt;
 488             eweps            = ewrt-ewitab;
 489             ewitab           = 4*ewitab;
 490             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 491             velec            = qq21*((rinv21-sh_ewald)-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 492             felec            = qq21*rinv21*(rinvsq21-felec);
 493
 494             /* Update potential sums from outer loop */
 495             velecsum        += velec;
 496
 497             fscal            = felec;
 498
 499             /* Calculate temporary vectorial force */
 500             tx               = fscal*dx21;
 501             ty               = fscal*dy21;
 502             tz               = fscal*dz21;
 503
 504             /* Update vectorial force */
 505             fix2            += tx;
 506             fiy2            += ty;
 507             fiz2            += tz;
 508             f[j_coord_offset+DIM*1+XX] -= tx;
 509             f[j_coord_offset+DIM*1+YY] -= ty;
 510             f[j_coord_offset+DIM*1+ZZ] -= tz;
 511
 512             }
 513
 514             /**************************
 515              * CALCULATE INTERACTIONS *
 516              **************************/
 517
 518             if (rsq22<rcutoff2)
 519             {
 520
 521             r22              = rsq22*rinv22;
 522
 523             /* EWALD ELECTROSTATICS */
 524
 525             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 526             ewrt             = r22*ewtabscale;
 527             ewitab           = ewrt;
 528             eweps            = ewrt-ewitab;
 529             ewitab           = 4*ewitab;
 530             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 531             velec            = qq22*((rinv22-sh_ewald)-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 532             felec            = qq22*rinv22*(rinvsq22-felec);
 533
 534             /* Update potential sums from outer loop */
 535             velecsum        += velec;
 536
 537             fscal            = felec;
 538
 539             /* Calculate temporary vectorial force */
 540             tx               = fscal*dx22;
 541             ty               = fscal*dy22;
 542             tz               = fscal*dz22;
 543
 544             /* Update vectorial force */
 545             fix2            += tx;
 546             fiy2            += ty;
 547             fiz2            += tz;
 548             f[j_coord_offset+DIM*2+XX] -= tx;
 549             f[j_coord_offset+DIM*2+YY] -= ty;
 550             f[j_coord_offset+DIM*2+ZZ] -= tz;
 551
 552             }
 553
 554             /**************************
 555              * CALCULATE INTERACTIONS *
 556              **************************/
 557
 558             if (rsq23<rcutoff2)
 559             {
 560
 561             r23              = rsq23*rinv23;
 562
 563             /* EWALD ELECTROSTATICS */
 564
 565             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 566             ewrt             = r23*ewtabscale;
 567             ewitab           = ewrt;
 568             eweps            = ewrt-ewitab;
 569             ewitab           = 4*ewitab;
 570             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 571             velec            = qq23*((rinv23-sh_ewald)-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 572             felec            = qq23*rinv23*(rinvsq23-felec);
 573
 574             /* Update potential sums from outer loop */
 575             velecsum        += velec;
 576
 577             fscal            = felec;
 578
 579             /* Calculate temporary vectorial force */
 580             tx               = fscal*dx23;
 581             ty               = fscal*dy23;
 582             tz               = fscal*dz23;
 583
 584             /* Update vectorial force */
 585             fix2            += tx;
 586             fiy2            += ty;
 587             fiz2            += tz;
 588             f[j_coord_offset+DIM*3+XX] -= tx;
 589             f[j_coord_offset+DIM*3+YY] -= ty;
 590             f[j_coord_offset+DIM*3+ZZ] -= tz;
 591
 592             }
 593
 594             /**************************
 595              * CALCULATE INTERACTIONS *
 596              **************************/
 597
 598             if (rsq31<rcutoff2)
 599             {
 600
 601             r31              = rsq31*rinv31;
 602
 603             /* EWALD ELECTROSTATICS */
 604
 605             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 606             ewrt             = r31*ewtabscale;
 607             ewitab           = ewrt;
 608             eweps            = ewrt-ewitab;
 609             ewitab           = 4*ewitab;
 610             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 611             velec            = qq31*((rinv31-sh_ewald)-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 612             felec            = qq31*rinv31*(rinvsq31-felec);
 613
 614             /* Update potential sums from outer loop */
 615             velecsum        += velec;
 616
 617             fscal            = felec;
 618
 619             /* Calculate temporary vectorial force */
 620             tx               = fscal*dx31;
 621             ty               = fscal*dy31;
 622             tz               = fscal*dz31;
 623
 624             /* Update vectorial force */
 625             fix3            += tx;
 626             fiy3            += ty;
 627             fiz3            += tz;
 628             f[j_coord_offset+DIM*1+XX] -= tx;
 629             f[j_coord_offset+DIM*1+YY] -= ty;
 630             f[j_coord_offset+DIM*1+ZZ] -= tz;
 631
 632             }
 633
 634             /**************************
 635              * CALCULATE INTERACTIONS *
 636              **************************/
 637
 638             if (rsq32<rcutoff2)
 639             {
 640
 641             r32              = rsq32*rinv32;
 642
 643             /* EWALD ELECTROSTATICS */
 644
 645             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 646             ewrt             = r32*ewtabscale;
 647             ewitab           = ewrt;
 648             eweps            = ewrt-ewitab;
 649             ewitab           = 4*ewitab;
 650             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 651             velec            = qq32*((rinv32-sh_ewald)-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 652             felec            = qq32*rinv32*(rinvsq32-felec);
 653
 654             /* Update potential sums from outer loop */
 655             velecsum        += velec;
 656
 657             fscal            = felec;
 658
 659             /* Calculate temporary vectorial force */
 660             tx               = fscal*dx32;
 661             ty               = fscal*dy32;
 662             tz               = fscal*dz32;
 663
 664             /* Update vectorial force */
 665             fix3            += tx;
 666             fiy3            += ty;
 667             fiz3            += tz;
 668             f[j_coord_offset+DIM*2+XX] -= tx;
 669             f[j_coord_offset+DIM*2+YY] -= ty;
 670             f[j_coord_offset+DIM*2+ZZ] -= tz;
 671
 672             }
 673
 674             /**************************
 675              * CALCULATE INTERACTIONS *
 676              **************************/
 677
 678             if (rsq33<rcutoff2)
 679             {
 680
 681             r33              = rsq33*rinv33;
 682
 683             /* EWALD ELECTROSTATICS */
 684
 685             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 686             ewrt             = r33*ewtabscale;
 687             ewitab           = ewrt;
 688             eweps            = ewrt-ewitab;
 689             ewitab           = 4*ewitab;
 690             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 691             velec            = qq33*((rinv33-sh_ewald)-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 692             felec            = qq33*rinv33*(rinvsq33-felec);
 693
 694             /* Update potential sums from outer loop */
 695             velecsum        += velec;
 696
 697             fscal            = felec;
 698
 699             /* Calculate temporary vectorial force */
 700             tx               = fscal*dx33;
 701             ty               = fscal*dy33;
 702             tz               = fscal*dz33;
 703
 704             /* Update vectorial force */
 705             fix3            += tx;
 706             fiy3            += ty;
 707             fiz3            += tz;
 708             f[j_coord_offset+DIM*3+XX] -= tx;
 709             f[j_coord_offset+DIM*3+YY] -= ty;
 710             f[j_coord_offset+DIM*3+ZZ] -= tz;
 711
 712             }
 713
 714             /* Inner loop uses 424 flops */
 715         }
 716         /* End of innermost loop */
 717
 718         tx = ty = tz = 0;
 719         f[i_coord_offset+DIM*0+XX] += fix0;
 720         f[i_coord_offset+DIM*0+YY] += fiy0;
 721         f[i_coord_offset+DIM*0+ZZ] += fiz0;
 722         tx                         += fix0;
 723         ty                         += fiy0;
 724         tz                         += fiz0;
 725         f[i_coord_offset+DIM*1+XX] += fix1;
 726         f[i_coord_offset+DIM*1+YY] += fiy1;
 727         f[i_coord_offset+DIM*1+ZZ] += fiz1;
 728         tx                         += fix1;
 729         ty                         += fiy1;
 730         tz                         += fiz1;
 731         f[i_coord_offset+DIM*2+XX] += fix2;
 732         f[i_coord_offset+DIM*2+YY] += fiy2;
 733         f[i_coord_offset+DIM*2+ZZ] += fiz2;
 734         tx                         += fix2;
 735         ty                         += fiy2;
 736         tz                         += fiz2;
 737         f[i_coord_offset+DIM*3+XX] += fix3;
 738         f[i_coord_offset+DIM*3+YY] += fiy3;
 739         f[i_coord_offset+DIM*3+ZZ] += fiz3;
 740         tx                         += fix3;
 741         ty                         += fiy3;
 742         tz                         += fiz3;
 743         fshift[i_shift_offset+XX]  += tx;
 744         fshift[i_shift_offset+YY]  += ty;
 745         fshift[i_shift_offset+ZZ]  += tz;
 746
 747         ggid                        = gid[iidx];
 748         /* Update potential energies */
 749         kernel_data->energygrp_elec[ggid] += velecsum;
 750         kernel_data->energygrp_vdw[ggid] += vvdwsum;
 751
 752         /* Increment number of inner iterations */
 753         inneriter                  += j_index_end - j_index_start;
 754
 755         /* Outer loop uses 41 flops */
 756     }
 757
 758     /* Increment number of outer iterations */
 759     outeriter        += nri;
 760
 761     /* Update outer/inner flops */
 762
 763     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W4W4_VF,outeriter*41 + inneriter*424);
 764 }
 765 /*
 766  * Gromacs nonbonded kernel:   nb_kernel_ElecEwSh_VdwLJEwSh_GeomW4W4_F_c
 767  * Electrostatics interaction: Ewald
 768  * VdW interaction:            LJEwald
 769  * Geometry:                   Water4-Water4
 770  * Calculate force/pot:        Force
 771  */
 772 void
 773 nb_kernel_ElecEwSh_VdwLJEwSh_GeomW4W4_F_c
 774                     (t_nblist                    * gmx_restrict       nlist,
 775                      rvec                        * gmx_restrict          xx,
 776                      rvec                        * gmx_restrict          ff,
 777                      t_forcerec                  * gmx_restrict          fr,
 778                      t_mdatoms                   * gmx_restrict     mdatoms,
 779                      nb_kernel_data_t gmx_unused * gmx_restrict kernel_data,
 780                      t_nrnb                      * gmx_restrict        nrnb)
 781 {
 782     int              i_shift_offset,i_coord_offset,j_coord_offset;
 783     int              j_index_start,j_index_end;
 784     int              nri,inr,ggid,iidx,jidx,jnr,outeriter,inneriter;
 785     real             shX,shY,shZ,tx,ty,tz,fscal,rcutoff,rcutoff2;
 786     int              *iinr,*jindex,*jjnr,*shiftidx,*gid;
 787     real             *shiftvec,*fshift,*x,*f;
 788     int              vdwioffset0;
 789     real             ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
 790     int              vdwioffset1;
 791     real             ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
 792     int              vdwioffset2;
 793     real             ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
 794     int              vdwioffset3;
 795     real             ix3,iy3,iz3,fix3,fiy3,fiz3,iq3,isai3;
 796     int              vdwjidx0;
 797     real             jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
 798     int              vdwjidx1;
 799     real             jx1,jy1,jz1,fjx1,fjy1,fjz1,jq1,isaj1;
 800     int              vdwjidx2;
 801     real             jx2,jy2,jz2,fjx2,fjy2,fjz2,jq2,isaj2;
 802     int              vdwjidx3;
 803     real             jx3,jy3,jz3,fjx3,fjy3,fjz3,jq3,isaj3;
 804     real             dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00,cexp1_00,cexp2_00;
 805     real             dx11,dy11,dz11,rsq11,rinv11,rinvsq11,r11,qq11,c6_11,c12_11,cexp1_11,cexp2_11;
 806     real             dx12,dy12,dz12,rsq12,rinv12,rinvsq12,r12,qq12,c6_12,c12_12,cexp1_12,cexp2_12;
 807     real             dx13,dy13,dz13,rsq13,rinv13,rinvsq13,r13,qq13,c6_13,c12_13,cexp1_13,cexp2_13;
 808     real             dx21,dy21,dz21,rsq21,rinv21,rinvsq21,r21,qq21,c6_21,c12_21,cexp1_21,cexp2_21;
 809     real             dx22,dy22,dz22,rsq22,rinv22,rinvsq22,r22,qq22,c6_22,c12_22,cexp1_22,cexp2_22;
 810     real             dx23,dy23,dz23,rsq23,rinv23,rinvsq23,r23,qq23,c6_23,c12_23,cexp1_23,cexp2_23;
 811     real             dx31,dy31,dz31,rsq31,rinv31,rinvsq31,r31,qq31,c6_31,c12_31,cexp1_31,cexp2_31;
 812     real             dx32,dy32,dz32,rsq32,rinv32,rinvsq32,r32,qq32,c6_32,c12_32,cexp1_32,cexp2_32;
 813     real             dx33,dy33,dz33,rsq33,rinv33,rinvsq33,r33,qq33,c6_33,c12_33,cexp1_33,cexp2_33;
 814     real             velec,felec,velecsum,facel,crf,krf,krf2;
 815     real             *charge;
 816     int              nvdwtype;
 817     real             rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,br,vvdwexp,sh_vdw_invrcut6;
 818     int              *vdwtype;
 819     real             *vdwparam;
 820     real             c6grid_00;
 821     real             c6grid_11;
 822     real             c6grid_12;
 823     real             c6grid_13;
 824     real             c6grid_21;
 825     real             c6grid_22;
 826     real             c6grid_23;
 827     real             c6grid_31;
 828     real             c6grid_32;
 829     real             c6grid_33;
 830     real             ewclj,ewclj2,ewclj6,ewcljrsq,poly,exponent,sh_lj_ewald;
 831     real             *vdwgridparam;
 832     int              ewitab;
 833     real             ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace;
 834     real             *ewtab;
 835
 836     x                = xx[0];
 837     f                = ff[0];
 838
 839     nri              = nlist->nri;
 840     iinr             = nlist->iinr;
 841     jindex           = nlist->jindex;
 842     jjnr             = nlist->jjnr;
 843     shiftidx         = nlist->shift;
 844     gid              = nlist->gid;
 845     shiftvec         = fr->shift_vec[0];
 846     fshift           = fr->fshift[0];
 847     facel            = fr->epsfac;
 848     charge           = mdatoms->chargeA;
 849     nvdwtype         = fr->ntype;
 850     vdwparam         = fr->nbfp;
 851     vdwtype          = mdatoms->typeA;
 852     vdwgridparam     = fr->ljpme_c6grid;
 853     ewclj            = fr->ewaldcoeff_lj;
 854     sh_lj_ewald      = fr->ic->sh_lj_ewald;
 855     ewclj2           = ewclj*ewclj;
 856     ewclj6           = ewclj2*ewclj2*ewclj2;
 857
 858     sh_ewald         = fr->ic->sh_ewald;
 859     ewtab            = fr->ic->tabq_coul_F;
 860     ewtabscale       = fr->ic->tabq_scale;
 861     ewtabhalfspace   = 0.5/ewtabscale;
 862
 863     /* Setup water-specific parameters */
 864     inr              = nlist->iinr[0];
 865     iq1              = facel*charge[inr+1];
 866     iq2              = facel*charge[inr+2];
 867     iq3              = facel*charge[inr+3];
 868     vdwioffset0      = 2*nvdwtype*vdwtype[inr+0];
 869
 870     jq1              = charge[inr+1];
 871     jq2              = charge[inr+2];
 872     jq3              = charge[inr+3];
 873     vdwjidx0         = 2*vdwtype[inr+0];
 874     c6_00            = vdwparam[vdwioffset0+vdwjidx0];
 875     c12_00           = vdwparam[vdwioffset0+vdwjidx0+1];
 876     c6grid_00        = vdwgridparam[vdwioffset0+vdwjidx0];
 877     qq11             = iq1*jq1;
 878     qq12             = iq1*jq2;
 879     qq13             = iq1*jq3;
 880     qq21             = iq2*jq1;
 881     qq22             = iq2*jq2;
 882     qq23             = iq2*jq3;
 883     qq31             = iq3*jq1;
 884     qq32             = iq3*jq2;
 885     qq33             = iq3*jq3;
 886
 887     /* When we use explicit cutoffs the value must be identical for elec and VdW, so use elec as an arbitrary choice */
 888     rcutoff          = fr->rcoulomb;
 889     rcutoff2         = rcutoff*rcutoff;
 890
 891     sh_vdw_invrcut6  = fr->ic->sh_invrc6;
 892     rvdw             = fr->rvdw;
 893
 894     outeriter        = 0;
 895     inneriter        = 0;
 896
 897     /* Start outer loop over neighborlists */
 898     for(iidx=0; iidx<nri; iidx++)
 899     {
 900         /* Load shift vector for this list */
 901         i_shift_offset   = DIM*shiftidx[iidx];
 902         shX              = shiftvec[i_shift_offset+XX];
 903         shY              = shiftvec[i_shift_offset+YY];
 904         shZ              = shiftvec[i_shift_offset+ZZ];
 905
 906         /* Load limits for loop over neighbors */
 907         j_index_start    = jindex[iidx];
 908         j_index_end      = jindex[iidx+1];
 909
 910         /* Get outer coordinate index */
 911         inr              = iinr[iidx];
 912         i_coord_offset   = DIM*inr;
 913
 914         /* Load i particle coords and add shift vector */
 915         ix0              = shX + x[i_coord_offset+DIM*0+XX];
 916         iy0              = shY + x[i_coord_offset+DIM*0+YY];
 917         iz0              = shZ + x[i_coord_offset+DIM*0+ZZ];
 918         ix1              = shX + x[i_coord_offset+DIM*1+XX];
 919         iy1              = shY + x[i_coord_offset+DIM*1+YY];
 920         iz1              = shZ + x[i_coord_offset+DIM*1+ZZ];
 921         ix2              = shX + x[i_coord_offset+DIM*2+XX];
 922         iy2              = shY + x[i_coord_offset+DIM*2+YY];
 923         iz2              = shZ + x[i_coord_offset+DIM*2+ZZ];
 924         ix3              = shX + x[i_coord_offset+DIM*3+XX];
 925         iy3              = shY + x[i_coord_offset+DIM*3+YY];
 926         iz3              = shZ + x[i_coord_offset+DIM*3+ZZ];
 927
 928         fix0             = 0.0;
 929         fiy0             = 0.0;
 930         fiz0             = 0.0;
 931         fix1             = 0.0;
 932         fiy1             = 0.0;
 933         fiz1             = 0.0;
 934         fix2             = 0.0;
 935         fiy2             = 0.0;
 936         fiz2             = 0.0;
 937         fix3             = 0.0;
 938         fiy3             = 0.0;
 939         fiz3             = 0.0;
 940
 941         /* Start inner kernel loop */
 942         for(jidx=j_index_start; jidx<j_index_end; jidx++)
 943         {
 944             /* Get j neighbor index, and coordinate index */
 945             jnr              = jjnr[jidx];
 946             j_coord_offset   = DIM*jnr;
 947
 948             /* load j atom coordinates */
 949             jx0              = x[j_coord_offset+DIM*0+XX];
 950             jy0              = x[j_coord_offset+DIM*0+YY];
 951             jz0              = x[j_coord_offset+DIM*0+ZZ];
 952             jx1              = x[j_coord_offset+DIM*1+XX];
 953             jy1              = x[j_coord_offset+DIM*1+YY];
 954             jz1              = x[j_coord_offset+DIM*1+ZZ];
 955             jx2              = x[j_coord_offset+DIM*2+XX];
 956             jy2              = x[j_coord_offset+DIM*2+YY];
 957             jz2              = x[j_coord_offset+DIM*2+ZZ];
 958             jx3              = x[j_coord_offset+DIM*3+XX];
 959             jy3              = x[j_coord_offset+DIM*3+YY];
 960             jz3              = x[j_coord_offset+DIM*3+ZZ];
 961
 962             /* Calculate displacement vector */
 963             dx00             = ix0 - jx0;
 964             dy00             = iy0 - jy0;
 965             dz00             = iz0 - jz0;
 966             dx11             = ix1 - jx1;
 967             dy11             = iy1 - jy1;
 968             dz11             = iz1 - jz1;
 969             dx12             = ix1 - jx2;
 970             dy12             = iy1 - jy2;
 971             dz12             = iz1 - jz2;
 972             dx13             = ix1 - jx3;
 973             dy13             = iy1 - jy3;
 974             dz13             = iz1 - jz3;
 975             dx21             = ix2 - jx1;
 976             dy21             = iy2 - jy1;
 977             dz21             = iz2 - jz1;
 978             dx22             = ix2 - jx2;
 979             dy22             = iy2 - jy2;
 980             dz22             = iz2 - jz2;
 981             dx23             = ix2 - jx3;
 982             dy23             = iy2 - jy3;
 983             dz23             = iz2 - jz3;
 984             dx31             = ix3 - jx1;
 985             dy31             = iy3 - jy1;
 986             dz31             = iz3 - jz1;
 987             dx32             = ix3 - jx2;
 988             dy32             = iy3 - jy2;
 989             dz32             = iz3 - jz2;
 990             dx33             = ix3 - jx3;
 991             dy33             = iy3 - jy3;
 992             dz33             = iz3 - jz3;
 993
 994             /* Calculate squared distance and things based on it */
 995             rsq00            = dx00*dx00+dy00*dy00+dz00*dz00;
 996             rsq11            = dx11*dx11+dy11*dy11+dz11*dz11;
 997             rsq12            = dx12*dx12+dy12*dy12+dz12*dz12;
 998             rsq13            = dx13*dx13+dy13*dy13+dz13*dz13;
 999             rsq21            = dx21*dx21+dy21*dy21+dz21*dz21;
1000             rsq22            = dx22*dx22+dy22*dy22+dz22*dz22;
1001             rsq23            = dx23*dx23+dy23*dy23+dz23*dz23;
1002             rsq31            = dx31*dx31+dy31*dy31+dz31*dz31;
1003             rsq32            = dx32*dx32+dy32*dy32+dz32*dz32;
1004             rsq33            = dx33*dx33+dy33*dy33+dz33*dz33;
1005
1006             rinv00           = gmx_invsqrt(rsq00);
1007             rinv11           = gmx_invsqrt(rsq11);
1008             rinv12           = gmx_invsqrt(rsq12);
1009             rinv13           = gmx_invsqrt(rsq13);
1010             rinv21           = gmx_invsqrt(rsq21);
1011             rinv22           = gmx_invsqrt(rsq22);
1012             rinv23           = gmx_invsqrt(rsq23);
1013             rinv31           = gmx_invsqrt(rsq31);
1014             rinv32           = gmx_invsqrt(rsq32);
1015             rinv33           = gmx_invsqrt(rsq33);
1016
1017             rinvsq00         = rinv00*rinv00;
1018             rinvsq11         = rinv11*rinv11;
1019             rinvsq12         = rinv12*rinv12;
1020             rinvsq13         = rinv13*rinv13;
1021             rinvsq21         = rinv21*rinv21;
1022             rinvsq22         = rinv22*rinv22;
1023             rinvsq23         = rinv23*rinv23;
1024             rinvsq31         = rinv31*rinv31;
1025             rinvsq32         = rinv32*rinv32;
1026             rinvsq33         = rinv33*rinv33;
1027
1028             /**************************
1029              * CALCULATE INTERACTIONS *
1030              **************************/
1031
1032             if (rsq00<rcutoff2)
1033             {
1034
1035             r00              = rsq00*rinv00;
1036
1037             rinvsix          = rinvsq00*rinvsq00*rinvsq00;
1038             ewcljrsq         = ewclj2*rsq00;
1039             exponent         = exp(-ewcljrsq);
1040             poly             = exponent*(1.0 + ewcljrsq + ewcljrsq*ewcljrsq*0.5);
1041             fvdw             = (((c12_00*rinvsix - c6_00 + c6grid_00*(1.0-poly))*rinvsix) - c6grid_00*(1.0/6.0)*exponent*ewclj6)*rinvsq00;
1042
1043             fscal            = fvdw;
1044
1045             /* Calculate temporary vectorial force */
1046             tx               = fscal*dx00;
1047             ty               = fscal*dy00;
1048             tz               = fscal*dz00;
1049
1050             /* Update vectorial force */
1051             fix0            += tx;
1052             fiy0            += ty;
1053             fiz0            += tz;
1054             f[j_coord_offset+DIM*0+XX] -= tx;
1055             f[j_coord_offset+DIM*0+YY] -= ty;
1056             f[j_coord_offset+DIM*0+ZZ] -= tz;
1057
1058             }
1059
1060             /**************************
1061              * CALCULATE INTERACTIONS *
1062              **************************/
1063
1064             if (rsq11<rcutoff2)
1065             {
1066
1067             r11              = rsq11*rinv11;
1068
1069             /* EWALD ELECTROSTATICS */
1070
1071             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1072             ewrt             = r11*ewtabscale;
1073             ewitab           = ewrt;
1074             eweps            = ewrt-ewitab;
1075             felec            = (1.0-eweps)*ewtab[ewitab]+eweps*ewtab[ewitab+1];
1076             felec            = qq11*rinv11*(rinvsq11-felec);
1077
1078             fscal            = felec;
1079
1080             /* Calculate temporary vectorial force */
1081             tx               = fscal*dx11;
1082             ty               = fscal*dy11;
1083             tz               = fscal*dz11;
1084
1085             /* Update vectorial force */
1086             fix1            += tx;
1087             fiy1            += ty;
1088             fiz1            += tz;
1089             f[j_coord_offset+DIM*1+XX] -= tx;
1090             f[j_coord_offset+DIM*1+YY] -= ty;
1091             f[j_coord_offset+DIM*1+ZZ] -= tz;
1092
1093             }
1094
1095             /**************************
1096              * CALCULATE INTERACTIONS *
1097              **************************/
1098
1099             if (rsq12<rcutoff2)
1100             {
1101
1102             r12              = rsq12*rinv12;
1103
1104             /* EWALD ELECTROSTATICS */
1105
1106             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1107             ewrt             = r12*ewtabscale;
1108             ewitab           = ewrt;
1109             eweps            = ewrt-ewitab;
1110             felec            = (1.0-eweps)*ewtab[ewitab]+eweps*ewtab[ewitab+1];
1111             felec            = qq12*rinv12*(rinvsq12-felec);
1112
1113             fscal            = felec;
1114
1115             /* Calculate temporary vectorial force */
1116             tx               = fscal*dx12;
1117             ty               = fscal*dy12;
1118             tz               = fscal*dz12;
1119
1120             /* Update vectorial force */
1121             fix1            += tx;
1122             fiy1            += ty;
1123             fiz1            += tz;
1124             f[j_coord_offset+DIM*2+XX] -= tx;
1125             f[j_coord_offset+DIM*2+YY] -= ty;
1126             f[j_coord_offset+DIM*2+ZZ] -= tz;
1127
1128             }
1129
1130             /**************************
1131              * CALCULATE INTERACTIONS *
1132              **************************/
1133
1134             if (rsq13<rcutoff2)
1135             {
1136
1137             r13              = rsq13*rinv13;
1138
1139             /* EWALD ELECTROSTATICS */
1140
1141             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1142             ewrt             = r13*ewtabscale;
1143             ewitab           = ewrt;
1144             eweps            = ewrt-ewitab;
1145             felec            = (1.0-eweps)*ewtab[ewitab]+eweps*ewtab[ewitab+1];
1146             felec            = qq13*rinv13*(rinvsq13-felec);
1147
1148             fscal            = felec;
1149
1150             /* Calculate temporary vectorial force */
1151             tx               = fscal*dx13;
1152             ty               = fscal*dy13;
1153             tz               = fscal*dz13;
1154
1155             /* Update vectorial force */
1156             fix1            += tx;
1157             fiy1            += ty;
1158             fiz1            += tz;
1159             f[j_coord_offset+DIM*3+XX] -= tx;
1160             f[j_coord_offset+DIM*3+YY] -= ty;
1161             f[j_coord_offset+DIM*3+ZZ] -= tz;
1162
1163             }
1164
1165             /**************************
1166              * CALCULATE INTERACTIONS *
1167              **************************/
1168
1169             if (rsq21<rcutoff2)
1170             {
1171
1172             r21              = rsq21*rinv21;
1173
1174             /* EWALD ELECTROSTATICS */
1175
1176             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1177             ewrt             = r21*ewtabscale;
1178             ewitab           = ewrt;
1179             eweps            = ewrt-ewitab;
1180             felec            = (1.0-eweps)*ewtab[ewitab]+eweps*ewtab[ewitab+1];
1181             felec            = qq21*rinv21*(rinvsq21-felec);
1182
1183             fscal            = felec;
1184
1185             /* Calculate temporary vectorial force */
1186             tx               = fscal*dx21;
1187             ty               = fscal*dy21;
1188             tz               = fscal*dz21;
1189
1190             /* Update vectorial force */
1191             fix2            += tx;
1192             fiy2            += ty;
1193             fiz2            += tz;
1194             f[j_coord_offset+DIM*1+XX] -= tx;
1195             f[j_coord_offset+DIM*1+YY] -= ty;
1196             f[j_coord_offset+DIM*1+ZZ] -= tz;
1197
1198             }
1199
1200             /**************************
1201              * CALCULATE INTERACTIONS *
1202              **************************/
1203
1204             if (rsq22<rcutoff2)
1205             {
1206
1207             r22              = rsq22*rinv22;
1208
1209             /* EWALD ELECTROSTATICS */
1210
1211             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1212             ewrt             = r22*ewtabscale;
1213             ewitab           = ewrt;
1214             eweps            = ewrt-ewitab;
1215             felec            = (1.0-eweps)*ewtab[ewitab]+eweps*ewtab[ewitab+1];
1216             felec            = qq22*rinv22*(rinvsq22-felec);
1217
1218             fscal            = felec;
1219
1220             /* Calculate temporary vectorial force */
1221             tx               = fscal*dx22;
1222             ty               = fscal*dy22;
1223             tz               = fscal*dz22;
1224
1225             /* Update vectorial force */
1226             fix2            += tx;
1227             fiy2            += ty;
1228             fiz2            += tz;
1229             f[j_coord_offset+DIM*2+XX] -= tx;
1230             f[j_coord_offset+DIM*2+YY] -= ty;
1231             f[j_coord_offset+DIM*2+ZZ] -= tz;
1232
1233             }
1234
1235             /**************************
1236              * CALCULATE INTERACTIONS *
1237              **************************/
1238
1239             if (rsq23<rcutoff2)
1240             {
1241
1242             r23              = rsq23*rinv23;
1243
1244             /* EWALD ELECTROSTATICS */
1245
1246             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1247             ewrt             = r23*ewtabscale;
1248             ewitab           = ewrt;
1249             eweps            = ewrt-ewitab;
1250             felec            = (1.0-eweps)*ewtab[ewitab]+eweps*ewtab[ewitab+1];
1251             felec            = qq23*rinv23*(rinvsq23-felec);
1252
1253             fscal            = felec;
1254
1255             /* Calculate temporary vectorial force */
1256             tx               = fscal*dx23;
1257             ty               = fscal*dy23;
1258             tz               = fscal*dz23;
1259
1260             /* Update vectorial force */
1261             fix2            += tx;
1262             fiy2            += ty;
1263             fiz2            += tz;
1264             f[j_coord_offset+DIM*3+XX] -= tx;
1265             f[j_coord_offset+DIM*3+YY] -= ty;
1266             f[j_coord_offset+DIM*3+ZZ] -= tz;
1267
1268             }
1269
1270             /**************************
1271              * CALCULATE INTERACTIONS *
1272              **************************/
1273
1274             if (rsq31<rcutoff2)
1275             {
1276
1277             r31              = rsq31*rinv31;
1278
1279             /* EWALD ELECTROSTATICS */
1280
1281             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1282             ewrt             = r31*ewtabscale;
1283             ewitab           = ewrt;
1284             eweps            = ewrt-ewitab;
1285             felec            = (1.0-eweps)*ewtab[ewitab]+eweps*ewtab[ewitab+1];
1286             felec            = qq31*rinv31*(rinvsq31-felec);
1287
1288             fscal            = felec;
1289
1290             /* Calculate temporary vectorial force */
1291             tx               = fscal*dx31;
1292             ty               = fscal*dy31;
1293             tz               = fscal*dz31;
1294
1295             /* Update vectorial force */
1296             fix3            += tx;
1297             fiy3            += ty;
1298             fiz3            += tz;
1299             f[j_coord_offset+DIM*1+XX] -= tx;
1300             f[j_coord_offset+DIM*1+YY] -= ty;
1301             f[j_coord_offset+DIM*1+ZZ] -= tz;
1302
1303             }
1304
1305             /**************************
1306              * CALCULATE INTERACTIONS *
1307              **************************/
1308
1309             if (rsq32<rcutoff2)
1310             {
1311
1312             r32              = rsq32*rinv32;
1313
1314             /* EWALD ELECTROSTATICS */
1315
1316             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1317             ewrt             = r32*ewtabscale;
1318             ewitab           = ewrt;
1319             eweps            = ewrt-ewitab;
1320             felec            = (1.0-eweps)*ewtab[ewitab]+eweps*ewtab[ewitab+1];
1321             felec            = qq32*rinv32*(rinvsq32-felec);
1322
1323             fscal            = felec;
1324
1325             /* Calculate temporary vectorial force */
1326             tx               = fscal*dx32;
1327             ty               = fscal*dy32;
1328             tz               = fscal*dz32;
1329
1330             /* Update vectorial force */
1331             fix3            += tx;
1332             fiy3            += ty;
1333             fiz3            += tz;
1334             f[j_coord_offset+DIM*2+XX] -= tx;
1335             f[j_coord_offset+DIM*2+YY] -= ty;
1336             f[j_coord_offset+DIM*2+ZZ] -= tz;
1337
1338             }
1339
1340             /**************************
1341              * CALCULATE INTERACTIONS *
1342              **************************/
1343
1344             if (rsq33<rcutoff2)
1345             {
1346
1347             r33              = rsq33*rinv33;
1348
1349             /* EWALD ELECTROSTATICS */
1350
1351             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1352             ewrt             = r33*ewtabscale;
1353             ewitab           = ewrt;
1354             eweps            = ewrt-ewitab;
1355             felec            = (1.0-eweps)*ewtab[ewitab]+eweps*ewtab[ewitab+1];
1356             felec            = qq33*rinv33*(rinvsq33-felec);
1357
1358             fscal            = felec;
1359
1360             /* Calculate temporary vectorial force */
1361             tx               = fscal*dx33;
1362             ty               = fscal*dy33;
1363             tz               = fscal*dz33;
1364
1365             /* Update vectorial force */
1366             fix3            += tx;
1367             fiy3            += ty;
1368             fiz3            += tz;
1369             f[j_coord_offset+DIM*3+XX] -= tx;
1370             f[j_coord_offset+DIM*3+YY] -= ty;
1371             f[j_coord_offset+DIM*3+ZZ] -= tz;
1372
1373             }
1374
1375             /* Inner loop uses 341 flops */
1376         }
1377         /* End of innermost loop */
1378
1379         tx = ty = tz = 0;
1380         f[i_coord_offset+DIM*0+XX] += fix0;
1381         f[i_coord_offset+DIM*0+YY] += fiy0;
1382         f[i_coord_offset+DIM*0+ZZ] += fiz0;
1383         tx                         += fix0;
1384         ty                         += fiy0;
1385         tz                         += fiz0;
1386         f[i_coord_offset+DIM*1+XX] += fix1;
1387         f[i_coord_offset+DIM*1+YY] += fiy1;
1388         f[i_coord_offset+DIM*1+ZZ] += fiz1;
1389         tx                         += fix1;
1390         ty                         += fiy1;
1391         tz                         += fiz1;
1392         f[i_coord_offset+DIM*2+XX] += fix2;
1393         f[i_coord_offset+DIM*2+YY] += fiy2;
1394         f[i_coord_offset+DIM*2+ZZ] += fiz2;
1395         tx                         += fix2;
1396         ty                         += fiy2;
1397         tz                         += fiz2;
1398         f[i_coord_offset+DIM*3+XX] += fix3;
1399         f[i_coord_offset+DIM*3+YY] += fiy3;
1400         f[i_coord_offset+DIM*3+ZZ] += fiz3;
1401         tx                         += fix3;
1402         ty                         += fiy3;
1403         tz                         += fiz3;
1404         fshift[i_shift_offset+XX]  += tx;
1405         fshift[i_shift_offset+YY]  += ty;
1406         fshift[i_shift_offset+ZZ]  += tz;
1407
1408         /* Increment number of inner iterations */
1409         inneriter                  += j_index_end - j_index_start;
1410
1411         /* Outer loop uses 39 flops */
1412     }
1413
1414     /* Increment number of outer iterations */
1415     outeriter        += nri;
1416
1417     /* Update outer/inner flops */
1418
1419     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W4W4_F,outeriter*39 + inneriter*341);
1420 }