src/gromacs/gmxlib/nonbonded/nb_kernel_c/nb_kernel_ElecEw_VdwCSTab_GeomW3P1_c.c

   1 /*
   2  * This file is part of the GROMACS molecular simulation package.
   3  *
   4  * Copyright (c) 2012,2013,2014, by the GROMACS development team, led by
   5  * Mark Abraham, David van der Spoel, Berk Hess, and Erik Lindahl,
   6  * and including many others, as listed in the AUTHORS file in the
   7  * top-level source directory and at http://www.gromacs.org.
   8  *
   9  * GROMACS is free software; you can redistribute it and/or
  10  * modify it under the terms of the GNU Lesser General Public License
  11  * as published by the Free Software Foundation; either version 2.1
  12  * of the License, or (at your option) any later version.
  13  *
  14  * GROMACS is distributed in the hope that it will be useful,
  15  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  16  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  17  * Lesser General Public License for more details.
  18  *
  19  * You should have received a copy of the GNU Lesser General Public
  20  * License along with GROMACS; if not, see
  21  * http://www.gnu.org/licenses, or write to the Free Software Foundation,
  22  * Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02110-1301  USA.
  23  *
  24  * If you want to redistribute modifications to GROMACS, please
  25  * consider that scientific software is very special. Version
  26  * control is crucial - bugs must be traceable. We will be happy to
  27  * consider code for inclusion in the official distribution, but
  28  * derived work must not be called official GROMACS. Details are found
  29  * in the README & COPYING files - if they are missing, get the
  30  * official version at http://www.gromacs.org.
  31  *
  32  * To help us fund GROMACS development, we humbly ask that you cite
  33  * the research papers on the package. Check out http://www.gromacs.org.
  34  */
  35 /*
  36  * Note: this file was generated by the GROMACS c kernel generator.
  37  */
  38 #include "config.h"
  39
  40 #include <math.h>
  41
  42 #include "../nb_kernel.h"
  43 #include "types/simple.h"
  44 #include "gromacs/math/vec.h"
  45 #include "nrnb.h"
  46
  47 /*
  48  * Gromacs nonbonded kernel:   nb_kernel_ElecEw_VdwCSTab_GeomW3P1_VF_c
  49  * Electrostatics interaction: Ewald
  50  * VdW interaction:            CubicSplineTable
  51  * Geometry:                   Water3-Particle
  52  * Calculate force/pot:        PotentialAndForce
  53  */
  54 void
  55 nb_kernel_ElecEw_VdwCSTab_GeomW3P1_VF_c
  56                     (t_nblist                    * gmx_restrict       nlist,
  57                      rvec                        * gmx_restrict          xx,
  58                      rvec                        * gmx_restrict          ff,
  59                      t_forcerec                  * gmx_restrict          fr,
  60                      t_mdatoms                   * gmx_restrict     mdatoms,
  61                      nb_kernel_data_t gmx_unused * gmx_restrict kernel_data,
  62                      t_nrnb                      * gmx_restrict        nrnb)
  63 {
  64     int              i_shift_offset,i_coord_offset,j_coord_offset;
  65     int              j_index_start,j_index_end;
  66     int              nri,inr,ggid,iidx,jidx,jnr,outeriter,inneriter;
  67     real             shX,shY,shZ,tx,ty,tz,fscal,rcutoff,rcutoff2;
  68     int              *iinr,*jindex,*jjnr,*shiftidx,*gid;
  69     real             *shiftvec,*fshift,*x,*f;
  70     int              vdwioffset0;
  71     real             ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
  72     int              vdwioffset1;
  73     real             ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
  74     int              vdwioffset2;
  75     real             ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
  76     int              vdwjidx0;
  77     real             jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
  78     real             dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00,cexp1_00,cexp2_00;
  79     real             dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10,cexp1_10,cexp2_10;
  80     real             dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20,cexp1_20,cexp2_20;
  81     real             velec,felec,velecsum,facel,crf,krf,krf2;
  82     real             *charge;
  83     int              nvdwtype;
  84     real             rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,br,vvdwexp,sh_vdw_invrcut6;
  85     int              *vdwtype;
  86     real             *vdwparam;
  87     int              vfitab;
  88     real             rt,vfeps,vftabscale,Y,F,Geps,Heps2,Fp,VV,FF;
  89     real             *vftab;
  90     int              ewitab;
  91     real             ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace;
  92     real             *ewtab;
  93
  94     x                = xx[0];
  95     f                = ff[0];
  96
  97     nri              = nlist->nri;
  98     iinr             = nlist->iinr;
  99     jindex           = nlist->jindex;
 100     jjnr             = nlist->jjnr;
 101     shiftidx         = nlist->shift;
 102     gid              = nlist->gid;
 103     shiftvec         = fr->shift_vec[0];
 104     fshift           = fr->fshift[0];
 105     facel            = fr->epsfac;
 106     charge           = mdatoms->chargeA;
 107     nvdwtype         = fr->ntype;
 108     vdwparam         = fr->nbfp;
 109     vdwtype          = mdatoms->typeA;
 110
 111     vftab            = kernel_data->table_vdw->data;
 112     vftabscale       = kernel_data->table_vdw->scale;
 113
 114     sh_ewald         = fr->ic->sh_ewald;
 115     ewtab            = fr->ic->tabq_coul_FDV0;
 116     ewtabscale       = fr->ic->tabq_scale;
 117     ewtabhalfspace   = 0.5/ewtabscale;
 118
 119     /* Setup water-specific parameters */
 120     inr              = nlist->iinr[0];
 121     iq0              = facel*charge[inr+0];
 122     iq1              = facel*charge[inr+1];
 123     iq2              = facel*charge[inr+2];
 124     vdwioffset0      = 2*nvdwtype*vdwtype[inr+0];
 125
 126     outeriter        = 0;
 127     inneriter        = 0;
 128
 129     /* Start outer loop over neighborlists */
 130     for(iidx=0; iidx<nri; iidx++)
 131     {
 132         /* Load shift vector for this list */
 133         i_shift_offset   = DIM*shiftidx[iidx];
 134         shX              = shiftvec[i_shift_offset+XX];
 135         shY              = shiftvec[i_shift_offset+YY];
 136         shZ              = shiftvec[i_shift_offset+ZZ];
 137
 138         /* Load limits for loop over neighbors */
 139         j_index_start    = jindex[iidx];
 140         j_index_end      = jindex[iidx+1];
 141
 142         /* Get outer coordinate index */
 143         inr              = iinr[iidx];
 144         i_coord_offset   = DIM*inr;
 145
 146         /* Load i particle coords and add shift vector */
 147         ix0              = shX + x[i_coord_offset+DIM*0+XX];
 148         iy0              = shY + x[i_coord_offset+DIM*0+YY];
 149         iz0              = shZ + x[i_coord_offset+DIM*0+ZZ];
 150         ix1              = shX + x[i_coord_offset+DIM*1+XX];
 151         iy1              = shY + x[i_coord_offset+DIM*1+YY];
 152         iz1              = shZ + x[i_coord_offset+DIM*1+ZZ];
 153         ix2              = shX + x[i_coord_offset+DIM*2+XX];
 154         iy2              = shY + x[i_coord_offset+DIM*2+YY];
 155         iz2              = shZ + x[i_coord_offset+DIM*2+ZZ];
 156
 157         fix0             = 0.0;
 158         fiy0             = 0.0;
 159         fiz0             = 0.0;
 160         fix1             = 0.0;
 161         fiy1             = 0.0;
 162         fiz1             = 0.0;
 163         fix2             = 0.0;
 164         fiy2             = 0.0;
 165         fiz2             = 0.0;
 166
 167         /* Reset potential sums */
 168         velecsum         = 0.0;
 169         vvdwsum          = 0.0;
 170
 171         /* Start inner kernel loop */
 172         for(jidx=j_index_start; jidx<j_index_end; jidx++)
 173         {
 174             /* Get j neighbor index, and coordinate index */
 175             jnr              = jjnr[jidx];
 176             j_coord_offset   = DIM*jnr;
 177
 178             /* load j atom coordinates */
 179             jx0              = x[j_coord_offset+DIM*0+XX];
 180             jy0              = x[j_coord_offset+DIM*0+YY];
 181             jz0              = x[j_coord_offset+DIM*0+ZZ];
 182
 183             /* Calculate displacement vector */
 184             dx00             = ix0 - jx0;
 185             dy00             = iy0 - jy0;
 186             dz00             = iz0 - jz0;
 187             dx10             = ix1 - jx0;
 188             dy10             = iy1 - jy0;
 189             dz10             = iz1 - jz0;
 190             dx20             = ix2 - jx0;
 191             dy20             = iy2 - jy0;
 192             dz20             = iz2 - jz0;
 193
 194             /* Calculate squared distance and things based on it */
 195             rsq00            = dx00*dx00+dy00*dy00+dz00*dz00;
 196             rsq10            = dx10*dx10+dy10*dy10+dz10*dz10;
 197             rsq20            = dx20*dx20+dy20*dy20+dz20*dz20;
 198
 199             rinv00           = gmx_invsqrt(rsq00);
 200             rinv10           = gmx_invsqrt(rsq10);
 201             rinv20           = gmx_invsqrt(rsq20);
 202
 203             rinvsq00         = rinv00*rinv00;
 204             rinvsq10         = rinv10*rinv10;
 205             rinvsq20         = rinv20*rinv20;
 206
 207             /* Load parameters for j particles */
 208             jq0              = charge[jnr+0];
 209             vdwjidx0         = 2*vdwtype[jnr+0];
 210
 211             /**************************
 212              * CALCULATE INTERACTIONS *
 213              **************************/
 214
 215             r00              = rsq00*rinv00;
 216
 217             qq00             = iq0*jq0;
 218             c6_00            = vdwparam[vdwioffset0+vdwjidx0];
 219             c12_00           = vdwparam[vdwioffset0+vdwjidx0+1];
 220
 221             /* Calculate table index by multiplying r with table scale and truncate to integer */
 222             rt               = r00*vftabscale;
 223             vfitab           = rt;
 224             vfeps            = rt-vfitab;
 225             vfitab           = 2*4*vfitab;
 226
 227             /* EWALD ELECTROSTATICS */
 228
 229             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 230             ewrt             = r00*ewtabscale;
 231             ewitab           = ewrt;
 232             eweps            = ewrt-ewitab;
 233             ewitab           = 4*ewitab;
 234             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 235             velec            = qq00*(rinv00-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 236             felec            = qq00*rinv00*(rinvsq00-felec);
 237
 238             /* CUBIC SPLINE TABLE DISPERSION */
 239             vfitab          += 0;
 240             Y                = vftab[vfitab];
 241             F                = vftab[vfitab+1];
 242             Geps             = vfeps*vftab[vfitab+2];
 243             Heps2            = vfeps*vfeps*vftab[vfitab+3];
 244             Fp               = F+Geps+Heps2;
 245             VV               = Y+vfeps*Fp;
 246             vvdw6            = c6_00*VV;
 247             FF               = Fp+Geps+2.0*Heps2;
 248             fvdw6            = c6_00*FF;
 249
 250             /* CUBIC SPLINE TABLE REPULSION */
 251             Y                = vftab[vfitab+4];
 252             F                = vftab[vfitab+5];
 253             Geps             = vfeps*vftab[vfitab+6];
 254             Heps2            = vfeps*vfeps*vftab[vfitab+7];
 255             Fp               = F+Geps+Heps2;
 256             VV               = Y+vfeps*Fp;
 257             vvdw12           = c12_00*VV;
 258             FF               = Fp+Geps+2.0*Heps2;
 259             fvdw12           = c12_00*FF;
 260             vvdw             = vvdw12+vvdw6;
 261             fvdw             = -(fvdw6+fvdw12)*vftabscale*rinv00;
 262
 263             /* Update potential sums from outer loop */
 264             velecsum        += velec;
 265             vvdwsum         += vvdw;
 266
 267             fscal            = felec+fvdw;
 268
 269             /* Calculate temporary vectorial force */
 270             tx               = fscal*dx00;
 271             ty               = fscal*dy00;
 272             tz               = fscal*dz00;
 273
 274             /* Update vectorial force */
 275             fix0            += tx;
 276             fiy0            += ty;
 277             fiz0            += tz;
 278             f[j_coord_offset+DIM*0+XX] -= tx;
 279             f[j_coord_offset+DIM*0+YY] -= ty;
 280             f[j_coord_offset+DIM*0+ZZ] -= tz;
 281
 282             /**************************
 283              * CALCULATE INTERACTIONS *
 284              **************************/
 285
 286             r10              = rsq10*rinv10;
 287
 288             qq10             = iq1*jq0;
 289
 290             /* EWALD ELECTROSTATICS */
 291
 292             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 293             ewrt             = r10*ewtabscale;
 294             ewitab           = ewrt;
 295             eweps            = ewrt-ewitab;
 296             ewitab           = 4*ewitab;
 297             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 298             velec            = qq10*(rinv10-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 299             felec            = qq10*rinv10*(rinvsq10-felec);
 300
 301             /* Update potential sums from outer loop */
 302             velecsum        += velec;
 303
 304             fscal            = felec;
 305
 306             /* Calculate temporary vectorial force */
 307             tx               = fscal*dx10;
 308             ty               = fscal*dy10;
 309             tz               = fscal*dz10;
 310
 311             /* Update vectorial force */
 312             fix1            += tx;
 313             fiy1            += ty;
 314             fiz1            += tz;
 315             f[j_coord_offset+DIM*0+XX] -= tx;
 316             f[j_coord_offset+DIM*0+YY] -= ty;
 317             f[j_coord_offset+DIM*0+ZZ] -= tz;
 318
 319             /**************************
 320              * CALCULATE INTERACTIONS *
 321              **************************/
 322
 323             r20              = rsq20*rinv20;
 324
 325             qq20             = iq2*jq0;
 326
 327             /* EWALD ELECTROSTATICS */
 328
 329             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 330             ewrt             = r20*ewtabscale;
 331             ewitab           = ewrt;
 332             eweps            = ewrt-ewitab;
 333             ewitab           = 4*ewitab;
 334             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 335             velec            = qq20*(rinv20-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 336             felec            = qq20*rinv20*(rinvsq20-felec);
 337
 338             /* Update potential sums from outer loop */
 339             velecsum        += velec;
 340
 341             fscal            = felec;
 342
 343             /* Calculate temporary vectorial force */
 344             tx               = fscal*dx20;
 345             ty               = fscal*dy20;
 346             tz               = fscal*dz20;
 347
 348             /* Update vectorial force */
 349             fix2            += tx;
 350             fiy2            += ty;
 351             fiz2            += tz;
 352             f[j_coord_offset+DIM*0+XX] -= tx;
 353             f[j_coord_offset+DIM*0+YY] -= ty;
 354             f[j_coord_offset+DIM*0+ZZ] -= tz;
 355
 356             /* Inner loop uses 156 flops */
 357         }
 358         /* End of innermost loop */
 359
 360         tx = ty = tz = 0;
 361         f[i_coord_offset+DIM*0+XX] += fix0;
 362         f[i_coord_offset+DIM*0+YY] += fiy0;
 363         f[i_coord_offset+DIM*0+ZZ] += fiz0;
 364         tx                         += fix0;
 365         ty                         += fiy0;
 366         tz                         += fiz0;
 367         f[i_coord_offset+DIM*1+XX] += fix1;
 368         f[i_coord_offset+DIM*1+YY] += fiy1;
 369         f[i_coord_offset+DIM*1+ZZ] += fiz1;
 370         tx                         += fix1;
 371         ty                         += fiy1;
 372         tz                         += fiz1;
 373         f[i_coord_offset+DIM*2+XX] += fix2;
 374         f[i_coord_offset+DIM*2+YY] += fiy2;
 375         f[i_coord_offset+DIM*2+ZZ] += fiz2;
 376         tx                         += fix2;
 377         ty                         += fiy2;
 378         tz                         += fiz2;
 379         fshift[i_shift_offset+XX]  += tx;
 380         fshift[i_shift_offset+YY]  += ty;
 381         fshift[i_shift_offset+ZZ]  += tz;
 382
 383         ggid                        = gid[iidx];
 384         /* Update potential energies */
 385         kernel_data->energygrp_elec[ggid] += velecsum;
 386         kernel_data->energygrp_vdw[ggid] += vvdwsum;
 387
 388         /* Increment number of inner iterations */
 389         inneriter                  += j_index_end - j_index_start;
 390
 391         /* Outer loop uses 32 flops */
 392     }
 393
 394     /* Increment number of outer iterations */
 395     outeriter        += nri;
 396
 397     /* Update outer/inner flops */
 398
 399     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W3_VF,outeriter*32 + inneriter*156);
 400 }
 401 /*
 402  * Gromacs nonbonded kernel:   nb_kernel_ElecEw_VdwCSTab_GeomW3P1_F_c
 403  * Electrostatics interaction: Ewald
 404  * VdW interaction:            CubicSplineTable
 405  * Geometry:                   Water3-Particle
 406  * Calculate force/pot:        Force
 407  */
 408 void
 409 nb_kernel_ElecEw_VdwCSTab_GeomW3P1_F_c
 410                     (t_nblist                    * gmx_restrict       nlist,
 411                      rvec                        * gmx_restrict          xx,
 412                      rvec                        * gmx_restrict          ff,
 413                      t_forcerec                  * gmx_restrict          fr,
 414                      t_mdatoms                   * gmx_restrict     mdatoms,
 415                      nb_kernel_data_t gmx_unused * gmx_restrict kernel_data,
 416                      t_nrnb                      * gmx_restrict        nrnb)
 417 {
 418     int              i_shift_offset,i_coord_offset,j_coord_offset;
 419     int              j_index_start,j_index_end;
 420     int              nri,inr,ggid,iidx,jidx,jnr,outeriter,inneriter;
 421     real             shX,shY,shZ,tx,ty,tz,fscal,rcutoff,rcutoff2;
 422     int              *iinr,*jindex,*jjnr,*shiftidx,*gid;
 423     real             *shiftvec,*fshift,*x,*f;
 424     int              vdwioffset0;
 425     real             ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
 426     int              vdwioffset1;
 427     real             ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
 428     int              vdwioffset2;
 429     real             ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
 430     int              vdwjidx0;
 431     real             jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
 432     real             dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00,cexp1_00,cexp2_00;
 433     real             dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10,cexp1_10,cexp2_10;
 434     real             dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20,cexp1_20,cexp2_20;
 435     real             velec,felec,velecsum,facel,crf,krf,krf2;
 436     real             *charge;
 437     int              nvdwtype;
 438     real             rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,br,vvdwexp,sh_vdw_invrcut6;
 439     int              *vdwtype;
 440     real             *vdwparam;
 441     int              vfitab;
 442     real             rt,vfeps,vftabscale,Y,F,Geps,Heps2,Fp,VV,FF;
 443     real             *vftab;
 444     int              ewitab;
 445     real             ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace;
 446     real             *ewtab;
 447
 448     x                = xx[0];
 449     f                = ff[0];
 450
 451     nri              = nlist->nri;
 452     iinr             = nlist->iinr;
 453     jindex           = nlist->jindex;
 454     jjnr             = nlist->jjnr;
 455     shiftidx         = nlist->shift;
 456     gid              = nlist->gid;
 457     shiftvec         = fr->shift_vec[0];
 458     fshift           = fr->fshift[0];
 459     facel            = fr->epsfac;
 460     charge           = mdatoms->chargeA;
 461     nvdwtype         = fr->ntype;
 462     vdwparam         = fr->nbfp;
 463     vdwtype          = mdatoms->typeA;
 464
 465     vftab            = kernel_data->table_vdw->data;
 466     vftabscale       = kernel_data->table_vdw->scale;
 467
 468     sh_ewald         = fr->ic->sh_ewald;
 469     ewtab            = fr->ic->tabq_coul_F;
 470     ewtabscale       = fr->ic->tabq_scale;
 471     ewtabhalfspace   = 0.5/ewtabscale;
 472
 473     /* Setup water-specific parameters */
 474     inr              = nlist->iinr[0];
 475     iq0              = facel*charge[inr+0];
 476     iq1              = facel*charge[inr+1];
 477     iq2              = facel*charge[inr+2];
 478     vdwioffset0      = 2*nvdwtype*vdwtype[inr+0];
 479
 480     outeriter        = 0;
 481     inneriter        = 0;
 482
 483     /* Start outer loop over neighborlists */
 484     for(iidx=0; iidx<nri; iidx++)
 485     {
 486         /* Load shift vector for this list */
 487         i_shift_offset   = DIM*shiftidx[iidx];
 488         shX              = shiftvec[i_shift_offset+XX];
 489         shY              = shiftvec[i_shift_offset+YY];
 490         shZ              = shiftvec[i_shift_offset+ZZ];
 491
 492         /* Load limits for loop over neighbors */
 493         j_index_start    = jindex[iidx];
 494         j_index_end      = jindex[iidx+1];
 495
 496         /* Get outer coordinate index */
 497         inr              = iinr[iidx];
 498         i_coord_offset   = DIM*inr;
 499
 500         /* Load i particle coords and add shift vector */
 501         ix0              = shX + x[i_coord_offset+DIM*0+XX];
 502         iy0              = shY + x[i_coord_offset+DIM*0+YY];
 503         iz0              = shZ + x[i_coord_offset+DIM*0+ZZ];
 504         ix1              = shX + x[i_coord_offset+DIM*1+XX];
 505         iy1              = shY + x[i_coord_offset+DIM*1+YY];
 506         iz1              = shZ + x[i_coord_offset+DIM*1+ZZ];
 507         ix2              = shX + x[i_coord_offset+DIM*2+XX];
 508         iy2              = shY + x[i_coord_offset+DIM*2+YY];
 509         iz2              = shZ + x[i_coord_offset+DIM*2+ZZ];
 510
 511         fix0             = 0.0;
 512         fiy0             = 0.0;
 513         fiz0             = 0.0;
 514         fix1             = 0.0;
 515         fiy1             = 0.0;
 516         fiz1             = 0.0;
 517         fix2             = 0.0;
 518         fiy2             = 0.0;
 519         fiz2             = 0.0;
 520
 521         /* Start inner kernel loop */
 522         for(jidx=j_index_start; jidx<j_index_end; jidx++)
 523         {
 524             /* Get j neighbor index, and coordinate index */
 525             jnr              = jjnr[jidx];
 526             j_coord_offset   = DIM*jnr;
 527
 528             /* load j atom coordinates */
 529             jx0              = x[j_coord_offset+DIM*0+XX];
 530             jy0              = x[j_coord_offset+DIM*0+YY];
 531             jz0              = x[j_coord_offset+DIM*0+ZZ];
 532
 533             /* Calculate displacement vector */
 534             dx00             = ix0 - jx0;
 535             dy00             = iy0 - jy0;
 536             dz00             = iz0 - jz0;
 537             dx10             = ix1 - jx0;
 538             dy10             = iy1 - jy0;
 539             dz10             = iz1 - jz0;
 540             dx20             = ix2 - jx0;
 541             dy20             = iy2 - jy0;
 542             dz20             = iz2 - jz0;
 543
 544             /* Calculate squared distance and things based on it */
 545             rsq00            = dx00*dx00+dy00*dy00+dz00*dz00;
 546             rsq10            = dx10*dx10+dy10*dy10+dz10*dz10;
 547             rsq20            = dx20*dx20+dy20*dy20+dz20*dz20;
 548
 549             rinv00           = gmx_invsqrt(rsq00);
 550             rinv10           = gmx_invsqrt(rsq10);
 551             rinv20           = gmx_invsqrt(rsq20);
 552
 553             rinvsq00         = rinv00*rinv00;
 554             rinvsq10         = rinv10*rinv10;
 555             rinvsq20         = rinv20*rinv20;
 556
 557             /* Load parameters for j particles */
 558             jq0              = charge[jnr+0];
 559             vdwjidx0         = 2*vdwtype[jnr+0];
 560
 561             /**************************
 562              * CALCULATE INTERACTIONS *
 563              **************************/
 564
 565             r00              = rsq00*rinv00;
 566
 567             qq00             = iq0*jq0;
 568             c6_00            = vdwparam[vdwioffset0+vdwjidx0];
 569             c12_00           = vdwparam[vdwioffset0+vdwjidx0+1];
 570
 571             /* Calculate table index by multiplying r with table scale and truncate to integer */
 572             rt               = r00*vftabscale;
 573             vfitab           = rt;
 574             vfeps            = rt-vfitab;
 575             vfitab           = 2*4*vfitab;
 576
 577             /* EWALD ELECTROSTATICS */
 578
 579             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 580             ewrt             = r00*ewtabscale;
 581             ewitab           = ewrt;
 582             eweps            = ewrt-ewitab;
 583             felec            = (1.0-eweps)*ewtab[ewitab]+eweps*ewtab[ewitab+1];
 584             felec            = qq00*rinv00*(rinvsq00-felec);
 585
 586             /* CUBIC SPLINE TABLE DISPERSION */
 587             vfitab          += 0;
 588             F                = vftab[vfitab+1];
 589             Geps             = vfeps*vftab[vfitab+2];
 590             Heps2            = vfeps*vfeps*vftab[vfitab+3];
 591             Fp               = F+Geps+Heps2;
 592             FF               = Fp+Geps+2.0*Heps2;
 593             fvdw6            = c6_00*FF;
 594
 595             /* CUBIC SPLINE TABLE REPULSION */
 596             F                = vftab[vfitab+5];
 597             Geps             = vfeps*vftab[vfitab+6];
 598             Heps2            = vfeps*vfeps*vftab[vfitab+7];
 599             Fp               = F+Geps+Heps2;
 600             FF               = Fp+Geps+2.0*Heps2;
 601             fvdw12           = c12_00*FF;
 602             fvdw             = -(fvdw6+fvdw12)*vftabscale*rinv00;
 603
 604             fscal            = felec+fvdw;
 605
 606             /* Calculate temporary vectorial force */
 607             tx               = fscal*dx00;
 608             ty               = fscal*dy00;
 609             tz               = fscal*dz00;
 610
 611             /* Update vectorial force */
 612             fix0            += tx;
 613             fiy0            += ty;
 614             fiz0            += tz;
 615             f[j_coord_offset+DIM*0+XX] -= tx;
 616             f[j_coord_offset+DIM*0+YY] -= ty;
 617             f[j_coord_offset+DIM*0+ZZ] -= tz;
 618
 619             /**************************
 620              * CALCULATE INTERACTIONS *
 621              **************************/
 622
 623             r10              = rsq10*rinv10;
 624
 625             qq10             = iq1*jq0;
 626
 627             /* EWALD ELECTROSTATICS */
 628
 629             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 630             ewrt             = r10*ewtabscale;
 631             ewitab           = ewrt;
 632             eweps            = ewrt-ewitab;
 633             felec            = (1.0-eweps)*ewtab[ewitab]+eweps*ewtab[ewitab+1];
 634             felec            = qq10*rinv10*(rinvsq10-felec);
 635
 636             fscal            = felec;
 637
 638             /* Calculate temporary vectorial force */
 639             tx               = fscal*dx10;
 640             ty               = fscal*dy10;
 641             tz               = fscal*dz10;
 642
 643             /* Update vectorial force */
 644             fix1            += tx;
 645             fiy1            += ty;
 646             fiz1            += tz;
 647             f[j_coord_offset+DIM*0+XX] -= tx;
 648             f[j_coord_offset+DIM*0+YY] -= ty;
 649             f[j_coord_offset+DIM*0+ZZ] -= tz;
 650
 651             /**************************
 652              * CALCULATE INTERACTIONS *
 653              **************************/
 654
 655             r20              = rsq20*rinv20;
 656
 657             qq20             = iq2*jq0;
 658
 659             /* EWALD ELECTROSTATICS */
 660
 661             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 662             ewrt             = r20*ewtabscale;
 663             ewitab           = ewrt;
 664             eweps            = ewrt-ewitab;
 665             felec            = (1.0-eweps)*ewtab[ewitab]+eweps*ewtab[ewitab+1];
 666             felec            = qq20*rinv20*(rinvsq20-felec);
 667
 668             fscal            = felec;
 669
 670             /* Calculate temporary vectorial force */
 671             tx               = fscal*dx20;
 672             ty               = fscal*dy20;
 673             tz               = fscal*dz20;
 674
 675             /* Update vectorial force */
 676             fix2            += tx;
 677             fiy2            += ty;
 678             fiz2            += tz;
 679             f[j_coord_offset+DIM*0+XX] -= tx;
 680             f[j_coord_offset+DIM*0+YY] -= ty;
 681             f[j_coord_offset+DIM*0+ZZ] -= tz;
 682
 683             /* Inner loop uses 127 flops */
 684         }
 685         /* End of innermost loop */
 686
 687         tx = ty = tz = 0;
 688         f[i_coord_offset+DIM*0+XX] += fix0;
 689         f[i_coord_offset+DIM*0+YY] += fiy0;
 690         f[i_coord_offset+DIM*0+ZZ] += fiz0;
 691         tx                         += fix0;
 692         ty                         += fiy0;
 693         tz                         += fiz0;
 694         f[i_coord_offset+DIM*1+XX] += fix1;
 695         f[i_coord_offset+DIM*1+YY] += fiy1;
 696         f[i_coord_offset+DIM*1+ZZ] += fiz1;
 697         tx                         += fix1;
 698         ty                         += fiy1;
 699         tz                         += fiz1;
 700         f[i_coord_offset+DIM*2+XX] += fix2;
 701         f[i_coord_offset+DIM*2+YY] += fiy2;
 702         f[i_coord_offset+DIM*2+ZZ] += fiz2;
 703         tx                         += fix2;
 704         ty                         += fiy2;
 705         tz                         += fiz2;
 706         fshift[i_shift_offset+XX]  += tx;
 707         fshift[i_shift_offset+YY]  += ty;
 708         fshift[i_shift_offset+ZZ]  += tz;
 709
 710         /* Increment number of inner iterations */
 711         inneriter                  += j_index_end - j_index_start;
 712
 713         /* Outer loop uses 30 flops */
 714     }
 715
 716     /* Increment number of outer iterations */
 717     outeriter        += nri;
 718
 719     /* Update outer/inner flops */
 720
 721     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W3_F,outeriter*30 + inneriter*127);
 722 }