src/gmxlib/nonbonded/nb_kernel_c/nb_kernel_ElecEwSh_VdwBhamSh_GeomW4P1_c.c

   1 /*
   2  * Note: this file was generated by the Gromacs c kernel generator.
   3  *
   4  *                This source code is part of
   5  *
   6  *                 G   R   O   M   A   C   S
   7  *
   8  * Copyright (c) 2001-2012, The GROMACS Development Team
   9  *
  10  * Gromacs is a library for molecular simulation and trajectory analysis,
  11  * written by Erik Lindahl, David van der Spoel, Berk Hess, and others - for
  12  * a full list of developers and information, check out http://www.gromacs.org
  13  *
  14  * This program is free software; you can redistribute it and/or modify it under
  15  * the terms of the GNU Lesser General Public License as published by the Free
  16  * Software Foundation; either version 2 of the License, or (at your option) any
  17  * later version.
  18  *
  19  * To help fund GROMACS development, we humbly ask that you cite
  20  * the papers people have written on it - you can find them on the website.
  21  */
  22 #ifdef HAVE_CONFIG_H
  23 #include <config.h>
  24 #endif
  25
  26 #include <math.h>
  27
  28 #include "../nb_kernel.h"
  29 #include "types/simple.h"
  30 #include "vec.h"
  31 #include "nrnb.h"
  32
  33 /*
  34  * Gromacs nonbonded kernel:   nb_kernel_ElecEwSh_VdwBhamSh_GeomW4P1_VF_c
  35  * Electrostatics interaction: Ewald
  36  * VdW interaction:            Buckingham
  37  * Geometry:                   Water4-Particle
  38  * Calculate force/pot:        PotentialAndForce
  39  */
  40 void
  41 nb_kernel_ElecEwSh_VdwBhamSh_GeomW4P1_VF_c
  42                     (t_nblist * gmx_restrict                nlist,
  43                      rvec * gmx_restrict                    xx,
  44                      rvec * gmx_restrict                    ff,
  45                      t_forcerec * gmx_restrict              fr,
  46                      t_mdatoms * gmx_restrict               mdatoms,
  47                      nb_kernel_data_t * gmx_restrict        kernel_data,
  48                      t_nrnb * gmx_restrict                  nrnb)
  49 {
  50     int              i_shift_offset,i_coord_offset,j_coord_offset;
  51     int              j_index_start,j_index_end;
  52     int              nri,inr,ggid,iidx,jidx,jnr,outeriter,inneriter;
  53     real             shX,shY,shZ,tx,ty,tz,fscal,rcutoff,rcutoff2;
  54     int              *iinr,*jindex,*jjnr,*shiftidx,*gid;
  55     real             *shiftvec,*fshift,*x,*f;
  56     int              vdwioffset0;
  57     real             ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
  58     int              vdwioffset1;
  59     real             ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
  60     int              vdwioffset2;
  61     real             ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
  62     int              vdwioffset3;
  63     real             ix3,iy3,iz3,fix3,fiy3,fiz3,iq3,isai3;
  64     int              vdwjidx0;
  65     real             jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
  66     real             dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00,cexp1_00,cexp2_00;
  67     real             dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10,cexp1_10,cexp2_10;
  68     real             dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20,cexp1_20,cexp2_20;
  69     real             dx30,dy30,dz30,rsq30,rinv30,rinvsq30,r30,qq30,c6_30,c12_30,cexp1_30,cexp2_30;
  70     real             velec,felec,velecsum,facel,crf,krf,krf2;
  71     real             *charge;
  72     int              nvdwtype;
  73     real             rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,br,vvdwexp,sh_vdw_invrcut6;
  74     int              *vdwtype;
  75     real             *vdwparam;
  76     int              ewitab;
  77     real             ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace;
  78     real             *ewtab;
  79
  80     x                = xx[0];
  81     f                = ff[0];
  82
  83     nri              = nlist->nri;
  84     iinr             = nlist->iinr;
  85     jindex           = nlist->jindex;
  86     jjnr             = nlist->jjnr;
  87     shiftidx         = nlist->shift;
  88     gid              = nlist->gid;
  89     shiftvec         = fr->shift_vec[0];
  90     fshift           = fr->fshift[0];
  91     facel            = fr->epsfac;
  92     charge           = mdatoms->chargeA;
  93     nvdwtype         = fr->ntype;
  94     vdwparam         = fr->nbfp;
  95     vdwtype          = mdatoms->typeA;
  96
  97     sh_ewald         = fr->ic->sh_ewald;
  98     ewtab            = fr->ic->tabq_coul_FDV0;
  99     ewtabscale       = fr->ic->tabq_scale;
 100     ewtabhalfspace   = 0.5/ewtabscale;
 101
 102     /* Setup water-specific parameters */
 103     inr              = nlist->iinr[0];
 104     iq1              = facel*charge[inr+1];
 105     iq2              = facel*charge[inr+2];
 106     iq3              = facel*charge[inr+3];
 107     vdwioffset0      = 3*nvdwtype*vdwtype[inr+0];
 108
 109     /* When we use explicit cutoffs the value must be identical for elec and VdW, so use elec as an arbitrary choice */
 110     rcutoff          = fr->rcoulomb;
 111     rcutoff2         = rcutoff*rcutoff;
 112
 113     sh_vdw_invrcut6  = fr->ic->sh_invrc6;
 114     rvdw             = fr->rvdw;
 115
 116     outeriter        = 0;
 117     inneriter        = 0;
 118
 119     /* Start outer loop over neighborlists */
 120     for(iidx=0; iidx<nri; iidx++)
 121     {
 122         /* Load shift vector for this list */
 123         i_shift_offset   = DIM*shiftidx[iidx];
 124         shX              = shiftvec[i_shift_offset+XX];
 125         shY              = shiftvec[i_shift_offset+YY];
 126         shZ              = shiftvec[i_shift_offset+ZZ];
 127
 128         /* Load limits for loop over neighbors */
 129         j_index_start    = jindex[iidx];
 130         j_index_end      = jindex[iidx+1];
 131
 132         /* Get outer coordinate index */
 133         inr              = iinr[iidx];
 134         i_coord_offset   = DIM*inr;
 135
 136         /* Load i particle coords and add shift vector */
 137         ix0              = shX + x[i_coord_offset+DIM*0+XX];
 138         iy0              = shY + x[i_coord_offset+DIM*0+YY];
 139         iz0              = shZ + x[i_coord_offset+DIM*0+ZZ];
 140         ix1              = shX + x[i_coord_offset+DIM*1+XX];
 141         iy1              = shY + x[i_coord_offset+DIM*1+YY];
 142         iz1              = shZ + x[i_coord_offset+DIM*1+ZZ];
 143         ix2              = shX + x[i_coord_offset+DIM*2+XX];
 144         iy2              = shY + x[i_coord_offset+DIM*2+YY];
 145         iz2              = shZ + x[i_coord_offset+DIM*2+ZZ];
 146         ix3              = shX + x[i_coord_offset+DIM*3+XX];
 147         iy3              = shY + x[i_coord_offset+DIM*3+YY];
 148         iz3              = shZ + x[i_coord_offset+DIM*3+ZZ];
 149
 150         fix0             = 0.0;
 151         fiy0             = 0.0;
 152         fiz0             = 0.0;
 153         fix1             = 0.0;
 154         fiy1             = 0.0;
 155         fiz1             = 0.0;
 156         fix2             = 0.0;
 157         fiy2             = 0.0;
 158         fiz2             = 0.0;
 159         fix3             = 0.0;
 160         fiy3             = 0.0;
 161         fiz3             = 0.0;
 162
 163         /* Reset potential sums */
 164         velecsum         = 0.0;
 165         vvdwsum          = 0.0;
 166
 167         /* Start inner kernel loop */
 168         for(jidx=j_index_start; jidx<j_index_end; jidx++)
 169         {
 170             /* Get j neighbor index, and coordinate index */
 171             jnr              = jjnr[jidx];
 172             j_coord_offset   = DIM*jnr;
 173
 174             /* load j atom coordinates */
 175             jx0              = x[j_coord_offset+DIM*0+XX];
 176             jy0              = x[j_coord_offset+DIM*0+YY];
 177             jz0              = x[j_coord_offset+DIM*0+ZZ];
 178
 179             /* Calculate displacement vector */
 180             dx00             = ix0 - jx0;
 181             dy00             = iy0 - jy0;
 182             dz00             = iz0 - jz0;
 183             dx10             = ix1 - jx0;
 184             dy10             = iy1 - jy0;
 185             dz10             = iz1 - jz0;
 186             dx20             = ix2 - jx0;
 187             dy20             = iy2 - jy0;
 188             dz20             = iz2 - jz0;
 189             dx30             = ix3 - jx0;
 190             dy30             = iy3 - jy0;
 191             dz30             = iz3 - jz0;
 192
 193             /* Calculate squared distance and things based on it */
 194             rsq00            = dx00*dx00+dy00*dy00+dz00*dz00;
 195             rsq10            = dx10*dx10+dy10*dy10+dz10*dz10;
 196             rsq20            = dx20*dx20+dy20*dy20+dz20*dz20;
 197             rsq30            = dx30*dx30+dy30*dy30+dz30*dz30;
 198
 199             rinv00           = gmx_invsqrt(rsq00);
 200             rinv10           = gmx_invsqrt(rsq10);
 201             rinv20           = gmx_invsqrt(rsq20);
 202             rinv30           = gmx_invsqrt(rsq30);
 203
 204             rinvsq00         = rinv00*rinv00;
 205             rinvsq10         = rinv10*rinv10;
 206             rinvsq20         = rinv20*rinv20;
 207             rinvsq30         = rinv30*rinv30;
 208
 209             /* Load parameters for j particles */
 210             jq0              = charge[jnr+0];
 211             vdwjidx0         = 3*vdwtype[jnr+0];
 212
 213             /**************************
 214              * CALCULATE INTERACTIONS *
 215              **************************/
 216
 217             if (rsq00<rcutoff2)
 218             {
 219
 220             r00              = rsq00*rinv00;
 221
 222             c6_00            = vdwparam[vdwioffset0+vdwjidx0];
 223             cexp1_00         = vdwparam[vdwioffset0+vdwjidx0+1];
 224             cexp2_00         = vdwparam[vdwioffset0+vdwjidx0+2];
 225
 226             /* BUCKINGHAM DISPERSION/REPULSION */
 227             rinvsix          = rinvsq00*rinvsq00*rinvsq00;
 228             vvdw6            = c6_00*rinvsix;
 229             br               = cexp2_00*r00;
 230             vvdwexp          = cexp1_00*exp(-br);
 231             vvdw             = (vvdwexp-cexp1_00*exp(-cexp2_00*rvdw)) - (vvdw6 - c6_00*sh_vdw_invrcut6)*(1.0/6.0);
 232             fvdw             = (br*vvdwexp-vvdw6)*rinvsq00;
 233
 234             /* Update potential sums from outer loop */
 235             vvdwsum         += vvdw;
 236
 237             fscal            = fvdw;
 238
 239             /* Calculate temporary vectorial force */
 240             tx               = fscal*dx00;
 241             ty               = fscal*dy00;
 242             tz               = fscal*dz00;
 243
 244             /* Update vectorial force */
 245             fix0            += tx;
 246             fiy0            += ty;
 247             fiz0            += tz;
 248             f[j_coord_offset+DIM*0+XX] -= tx;
 249             f[j_coord_offset+DIM*0+YY] -= ty;
 250             f[j_coord_offset+DIM*0+ZZ] -= tz;
 251
 252             }
 253
 254             /**************************
 255              * CALCULATE INTERACTIONS *
 256              **************************/
 257
 258             if (rsq10<rcutoff2)
 259             {
 260
 261             r10              = rsq10*rinv10;
 262
 263             qq10             = iq1*jq0;
 264
 265             /* EWALD ELECTROSTATICS */
 266
 267             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 268             ewrt             = r10*ewtabscale;
 269             ewitab           = ewrt;
 270             eweps            = ewrt-ewitab;
 271             ewitab           = 4*ewitab;
 272             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 273             velec            = qq10*((rinv10-sh_ewald)-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 274             felec            = qq10*rinv10*(rinvsq10-felec);
 275
 276             /* Update potential sums from outer loop */
 277             velecsum        += velec;
 278
 279             fscal            = felec;
 280
 281             /* Calculate temporary vectorial force */
 282             tx               = fscal*dx10;
 283             ty               = fscal*dy10;
 284             tz               = fscal*dz10;
 285
 286             /* Update vectorial force */
 287             fix1            += tx;
 288             fiy1            += ty;
 289             fiz1            += tz;
 290             f[j_coord_offset+DIM*0+XX] -= tx;
 291             f[j_coord_offset+DIM*0+YY] -= ty;
 292             f[j_coord_offset+DIM*0+ZZ] -= tz;
 293
 294             }
 295
 296             /**************************
 297              * CALCULATE INTERACTIONS *
 298              **************************/
 299
 300             if (rsq20<rcutoff2)
 301             {
 302
 303             r20              = rsq20*rinv20;
 304
 305             qq20             = iq2*jq0;
 306
 307             /* EWALD ELECTROSTATICS */
 308
 309             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 310             ewrt             = r20*ewtabscale;
 311             ewitab           = ewrt;
 312             eweps            = ewrt-ewitab;
 313             ewitab           = 4*ewitab;
 314             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 315             velec            = qq20*((rinv20-sh_ewald)-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 316             felec            = qq20*rinv20*(rinvsq20-felec);
 317
 318             /* Update potential sums from outer loop */
 319             velecsum        += velec;
 320
 321             fscal            = felec;
 322
 323             /* Calculate temporary vectorial force */
 324             tx               = fscal*dx20;
 325             ty               = fscal*dy20;
 326             tz               = fscal*dz20;
 327
 328             /* Update vectorial force */
 329             fix2            += tx;
 330             fiy2            += ty;
 331             fiz2            += tz;
 332             f[j_coord_offset+DIM*0+XX] -= tx;
 333             f[j_coord_offset+DIM*0+YY] -= ty;
 334             f[j_coord_offset+DIM*0+ZZ] -= tz;
 335
 336             }
 337
 338             /**************************
 339              * CALCULATE INTERACTIONS *
 340              **************************/
 341
 342             if (rsq30<rcutoff2)
 343             {
 344
 345             r30              = rsq30*rinv30;
 346
 347             qq30             = iq3*jq0;
 348
 349             /* EWALD ELECTROSTATICS */
 350
 351             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 352             ewrt             = r30*ewtabscale;
 353             ewitab           = ewrt;
 354             eweps            = ewrt-ewitab;
 355             ewitab           = 4*ewitab;
 356             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 357             velec            = qq30*((rinv30-sh_ewald)-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 358             felec            = qq30*rinv30*(rinvsq30-felec);
 359
 360             /* Update potential sums from outer loop */
 361             velecsum        += velec;
 362
 363             fscal            = felec;
 364
 365             /* Calculate temporary vectorial force */
 366             tx               = fscal*dx30;
 367             ty               = fscal*dy30;
 368             tz               = fscal*dz30;
 369
 370             /* Update vectorial force */
 371             fix3            += tx;
 372             fiy3            += ty;
 373             fiz3            += tz;
 374             f[j_coord_offset+DIM*0+XX] -= tx;
 375             f[j_coord_offset+DIM*0+YY] -= ty;
 376             f[j_coord_offset+DIM*0+ZZ] -= tz;
 377
 378             }
 379
 380             /* Inner loop uses 218 flops */
 381         }
 382         /* End of innermost loop */
 383
 384         tx = ty = tz = 0;
 385         f[i_coord_offset+DIM*0+XX] += fix0;
 386         f[i_coord_offset+DIM*0+YY] += fiy0;
 387         f[i_coord_offset+DIM*0+ZZ] += fiz0;
 388         tx                         += fix0;
 389         ty                         += fiy0;
 390         tz                         += fiz0;
 391         f[i_coord_offset+DIM*1+XX] += fix1;
 392         f[i_coord_offset+DIM*1+YY] += fiy1;
 393         f[i_coord_offset+DIM*1+ZZ] += fiz1;
 394         tx                         += fix1;
 395         ty                         += fiy1;
 396         tz                         += fiz1;
 397         f[i_coord_offset+DIM*2+XX] += fix2;
 398         f[i_coord_offset+DIM*2+YY] += fiy2;
 399         f[i_coord_offset+DIM*2+ZZ] += fiz2;
 400         tx                         += fix2;
 401         ty                         += fiy2;
 402         tz                         += fiz2;
 403         f[i_coord_offset+DIM*3+XX] += fix3;
 404         f[i_coord_offset+DIM*3+YY] += fiy3;
 405         f[i_coord_offset+DIM*3+ZZ] += fiz3;
 406         tx                         += fix3;
 407         ty                         += fiy3;
 408         tz                         += fiz3;
 409         fshift[i_shift_offset+XX]  += tx;
 410         fshift[i_shift_offset+YY]  += ty;
 411         fshift[i_shift_offset+ZZ]  += tz;
 412
 413         ggid                        = gid[iidx];
 414         /* Update potential energies */
 415         kernel_data->energygrp_elec[ggid] += velecsum;
 416         kernel_data->energygrp_vdw[ggid] += vvdwsum;
 417
 418         /* Increment number of inner iterations */
 419         inneriter                  += j_index_end - j_index_start;
 420
 421         /* Outer loop uses 41 flops */
 422     }
 423
 424     /* Increment number of outer iterations */
 425     outeriter        += nri;
 426
 427     /* Update outer/inner flops */
 428
 429     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W4_VF,outeriter*41 + inneriter*218);
 430 }
 431 /*
 432  * Gromacs nonbonded kernel:   nb_kernel_ElecEwSh_VdwBhamSh_GeomW4P1_F_c
 433  * Electrostatics interaction: Ewald
 434  * VdW interaction:            Buckingham
 435  * Geometry:                   Water4-Particle
 436  * Calculate force/pot:        Force
 437  */
 438 void
 439 nb_kernel_ElecEwSh_VdwBhamSh_GeomW4P1_F_c
 440                     (t_nblist * gmx_restrict                nlist,
 441                      rvec * gmx_restrict                    xx,
 442                      rvec * gmx_restrict                    ff,
 443                      t_forcerec * gmx_restrict              fr,
 444                      t_mdatoms * gmx_restrict               mdatoms,
 445                      nb_kernel_data_t * gmx_restrict        kernel_data,
 446                      t_nrnb * gmx_restrict                  nrnb)
 447 {
 448     int              i_shift_offset,i_coord_offset,j_coord_offset;
 449     int              j_index_start,j_index_end;
 450     int              nri,inr,ggid,iidx,jidx,jnr,outeriter,inneriter;
 451     real             shX,shY,shZ,tx,ty,tz,fscal,rcutoff,rcutoff2;
 452     int              *iinr,*jindex,*jjnr,*shiftidx,*gid;
 453     real             *shiftvec,*fshift,*x,*f;
 454     int              vdwioffset0;
 455     real             ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
 456     int              vdwioffset1;
 457     real             ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
 458     int              vdwioffset2;
 459     real             ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
 460     int              vdwioffset3;
 461     real             ix3,iy3,iz3,fix3,fiy3,fiz3,iq3,isai3;
 462     int              vdwjidx0;
 463     real             jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
 464     real             dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00,cexp1_00,cexp2_00;
 465     real             dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10,cexp1_10,cexp2_10;
 466     real             dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20,cexp1_20,cexp2_20;
 467     real             dx30,dy30,dz30,rsq30,rinv30,rinvsq30,r30,qq30,c6_30,c12_30,cexp1_30,cexp2_30;
 468     real             velec,felec,velecsum,facel,crf,krf,krf2;
 469     real             *charge;
 470     int              nvdwtype;
 471     real             rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,br,vvdwexp,sh_vdw_invrcut6;
 472     int              *vdwtype;
 473     real             *vdwparam;
 474     int              ewitab;
 475     real             ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace;
 476     real             *ewtab;
 477
 478     x                = xx[0];
 479     f                = ff[0];
 480
 481     nri              = nlist->nri;
 482     iinr             = nlist->iinr;
 483     jindex           = nlist->jindex;
 484     jjnr             = nlist->jjnr;
 485     shiftidx         = nlist->shift;
 486     gid              = nlist->gid;
 487     shiftvec         = fr->shift_vec[0];
 488     fshift           = fr->fshift[0];
 489     facel            = fr->epsfac;
 490     charge           = mdatoms->chargeA;
 491     nvdwtype         = fr->ntype;
 492     vdwparam         = fr->nbfp;
 493     vdwtype          = mdatoms->typeA;
 494
 495     sh_ewald         = fr->ic->sh_ewald;
 496     ewtab            = fr->ic->tabq_coul_F;
 497     ewtabscale       = fr->ic->tabq_scale;
 498     ewtabhalfspace   = 0.5/ewtabscale;
 499
 500     /* Setup water-specific parameters */
 501     inr              = nlist->iinr[0];
 502     iq1              = facel*charge[inr+1];
 503     iq2              = facel*charge[inr+2];
 504     iq3              = facel*charge[inr+3];
 505     vdwioffset0      = 3*nvdwtype*vdwtype[inr+0];
 506
 507     /* When we use explicit cutoffs the value must be identical for elec and VdW, so use elec as an arbitrary choice */
 508     rcutoff          = fr->rcoulomb;
 509     rcutoff2         = rcutoff*rcutoff;
 510
 511     sh_vdw_invrcut6  = fr->ic->sh_invrc6;
 512     rvdw             = fr->rvdw;
 513
 514     outeriter        = 0;
 515     inneriter        = 0;
 516
 517     /* Start outer loop over neighborlists */
 518     for(iidx=0; iidx<nri; iidx++)
 519     {
 520         /* Load shift vector for this list */
 521         i_shift_offset   = DIM*shiftidx[iidx];
 522         shX              = shiftvec[i_shift_offset+XX];
 523         shY              = shiftvec[i_shift_offset+YY];
 524         shZ              = shiftvec[i_shift_offset+ZZ];
 525
 526         /* Load limits for loop over neighbors */
 527         j_index_start    = jindex[iidx];
 528         j_index_end      = jindex[iidx+1];
 529
 530         /* Get outer coordinate index */
 531         inr              = iinr[iidx];
 532         i_coord_offset   = DIM*inr;
 533
 534         /* Load i particle coords and add shift vector */
 535         ix0              = shX + x[i_coord_offset+DIM*0+XX];
 536         iy0              = shY + x[i_coord_offset+DIM*0+YY];
 537         iz0              = shZ + x[i_coord_offset+DIM*0+ZZ];
 538         ix1              = shX + x[i_coord_offset+DIM*1+XX];
 539         iy1              = shY + x[i_coord_offset+DIM*1+YY];
 540         iz1              = shZ + x[i_coord_offset+DIM*1+ZZ];
 541         ix2              = shX + x[i_coord_offset+DIM*2+XX];
 542         iy2              = shY + x[i_coord_offset+DIM*2+YY];
 543         iz2              = shZ + x[i_coord_offset+DIM*2+ZZ];
 544         ix3              = shX + x[i_coord_offset+DIM*3+XX];
 545         iy3              = shY + x[i_coord_offset+DIM*3+YY];
 546         iz3              = shZ + x[i_coord_offset+DIM*3+ZZ];
 547
 548         fix0             = 0.0;
 549         fiy0             = 0.0;
 550         fiz0             = 0.0;
 551         fix1             = 0.0;
 552         fiy1             = 0.0;
 553         fiz1             = 0.0;
 554         fix2             = 0.0;
 555         fiy2             = 0.0;
 556         fiz2             = 0.0;
 557         fix3             = 0.0;
 558         fiy3             = 0.0;
 559         fiz3             = 0.0;
 560
 561         /* Start inner kernel loop */
 562         for(jidx=j_index_start; jidx<j_index_end; jidx++)
 563         {
 564             /* Get j neighbor index, and coordinate index */
 565             jnr              = jjnr[jidx];
 566             j_coord_offset   = DIM*jnr;
 567
 568             /* load j atom coordinates */
 569             jx0              = x[j_coord_offset+DIM*0+XX];
 570             jy0              = x[j_coord_offset+DIM*0+YY];
 571             jz0              = x[j_coord_offset+DIM*0+ZZ];
 572
 573             /* Calculate displacement vector */
 574             dx00             = ix0 - jx0;
 575             dy00             = iy0 - jy0;
 576             dz00             = iz0 - jz0;
 577             dx10             = ix1 - jx0;
 578             dy10             = iy1 - jy0;
 579             dz10             = iz1 - jz0;
 580             dx20             = ix2 - jx0;
 581             dy20             = iy2 - jy0;
 582             dz20             = iz2 - jz0;
 583             dx30             = ix3 - jx0;
 584             dy30             = iy3 - jy0;
 585             dz30             = iz3 - jz0;
 586
 587             /* Calculate squared distance and things based on it */
 588             rsq00            = dx00*dx00+dy00*dy00+dz00*dz00;
 589             rsq10            = dx10*dx10+dy10*dy10+dz10*dz10;
 590             rsq20            = dx20*dx20+dy20*dy20+dz20*dz20;
 591             rsq30            = dx30*dx30+dy30*dy30+dz30*dz30;
 592
 593             rinv00           = gmx_invsqrt(rsq00);
 594             rinv10           = gmx_invsqrt(rsq10);
 595             rinv20           = gmx_invsqrt(rsq20);
 596             rinv30           = gmx_invsqrt(rsq30);
 597
 598             rinvsq00         = rinv00*rinv00;
 599             rinvsq10         = rinv10*rinv10;
 600             rinvsq20         = rinv20*rinv20;
 601             rinvsq30         = rinv30*rinv30;
 602
 603             /* Load parameters for j particles */
 604             jq0              = charge[jnr+0];
 605             vdwjidx0         = 3*vdwtype[jnr+0];
 606
 607             /**************************
 608              * CALCULATE INTERACTIONS *
 609              **************************/
 610
 611             if (rsq00<rcutoff2)
 612             {
 613
 614             r00              = rsq00*rinv00;
 615
 616             c6_00            = vdwparam[vdwioffset0+vdwjidx0];
 617             cexp1_00         = vdwparam[vdwioffset0+vdwjidx0+1];
 618             cexp2_00         = vdwparam[vdwioffset0+vdwjidx0+2];
 619
 620             /* BUCKINGHAM DISPERSION/REPULSION */
 621             rinvsix          = rinvsq00*rinvsq00*rinvsq00;
 622             vvdw6            = c6_00*rinvsix;
 623             br               = cexp2_00*r00;
 624             vvdwexp          = cexp1_00*exp(-br);
 625             fvdw             = (br*vvdwexp-vvdw6)*rinvsq00;
 626
 627             fscal            = fvdw;
 628
 629             /* Calculate temporary vectorial force */
 630             tx               = fscal*dx00;
 631             ty               = fscal*dy00;
 632             tz               = fscal*dz00;
 633
 634             /* Update vectorial force */
 635             fix0            += tx;
 636             fiy0            += ty;
 637             fiz0            += tz;
 638             f[j_coord_offset+DIM*0+XX] -= tx;
 639             f[j_coord_offset+DIM*0+YY] -= ty;
 640             f[j_coord_offset+DIM*0+ZZ] -= tz;
 641
 642             }
 643
 644             /**************************
 645              * CALCULATE INTERACTIONS *
 646              **************************/
 647
 648             if (rsq10<rcutoff2)
 649             {
 650
 651             r10              = rsq10*rinv10;
 652
 653             qq10             = iq1*jq0;
 654
 655             /* EWALD ELECTROSTATICS */
 656
 657             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 658             ewrt             = r10*ewtabscale;
 659             ewitab           = ewrt;
 660             eweps            = ewrt-ewitab;
 661             felec            = (1.0-eweps)*ewtab[ewitab]+eweps*ewtab[ewitab+1];
 662             felec            = qq10*rinv10*(rinvsq10-felec);
 663
 664             fscal            = felec;
 665
 666             /* Calculate temporary vectorial force */
 667             tx               = fscal*dx10;
 668             ty               = fscal*dy10;
 669             tz               = fscal*dz10;
 670
 671             /* Update vectorial force */
 672             fix1            += tx;
 673             fiy1            += ty;
 674             fiz1            += tz;
 675             f[j_coord_offset+DIM*0+XX] -= tx;
 676             f[j_coord_offset+DIM*0+YY] -= ty;
 677             f[j_coord_offset+DIM*0+ZZ] -= tz;
 678
 679             }
 680
 681             /**************************
 682              * CALCULATE INTERACTIONS *
 683              **************************/
 684
 685             if (rsq20<rcutoff2)
 686             {
 687
 688             r20              = rsq20*rinv20;
 689
 690             qq20             = iq2*jq0;
 691
 692             /* EWALD ELECTROSTATICS */
 693
 694             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 695             ewrt             = r20*ewtabscale;
 696             ewitab           = ewrt;
 697             eweps            = ewrt-ewitab;
 698             felec            = (1.0-eweps)*ewtab[ewitab]+eweps*ewtab[ewitab+1];
 699             felec            = qq20*rinv20*(rinvsq20-felec);
 700
 701             fscal            = felec;
 702
 703             /* Calculate temporary vectorial force */
 704             tx               = fscal*dx20;
 705             ty               = fscal*dy20;
 706             tz               = fscal*dz20;
 707
 708             /* Update vectorial force */
 709             fix2            += tx;
 710             fiy2            += ty;
 711             fiz2            += tz;
 712             f[j_coord_offset+DIM*0+XX] -= tx;
 713             f[j_coord_offset+DIM*0+YY] -= ty;
 714             f[j_coord_offset+DIM*0+ZZ] -= tz;
 715
 716             }
 717
 718             /**************************
 719              * CALCULATE INTERACTIONS *
 720              **************************/
 721
 722             if (rsq30<rcutoff2)
 723             {
 724
 725             r30              = rsq30*rinv30;
 726
 727             qq30             = iq3*jq0;
 728
 729             /* EWALD ELECTROSTATICS */
 730
 731             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 732             ewrt             = r30*ewtabscale;
 733             ewitab           = ewrt;
 734             eweps            = ewrt-ewitab;
 735             felec            = (1.0-eweps)*ewtab[ewitab]+eweps*ewtab[ewitab+1];
 736             felec            = qq30*rinv30*(rinvsq30-felec);
 737
 738             fscal            = felec;
 739
 740             /* Calculate temporary vectorial force */
 741             tx               = fscal*dx30;
 742             ty               = fscal*dy30;
 743             tz               = fscal*dz30;
 744
 745             /* Update vectorial force */
 746             fix3            += tx;
 747             fiy3            += ty;
 748             fiz3            += tz;
 749             f[j_coord_offset+DIM*0+XX] -= tx;
 750             f[j_coord_offset+DIM*0+YY] -= ty;
 751             f[j_coord_offset+DIM*0+ZZ] -= tz;
 752
 753             }
 754
 755             /* Inner loop uses 160 flops */
 756         }
 757         /* End of innermost loop */
 758
 759         tx = ty = tz = 0;
 760         f[i_coord_offset+DIM*0+XX] += fix0;
 761         f[i_coord_offset+DIM*0+YY] += fiy0;
 762         f[i_coord_offset+DIM*0+ZZ] += fiz0;
 763         tx                         += fix0;
 764         ty                         += fiy0;
 765         tz                         += fiz0;
 766         f[i_coord_offset+DIM*1+XX] += fix1;
 767         f[i_coord_offset+DIM*1+YY] += fiy1;
 768         f[i_coord_offset+DIM*1+ZZ] += fiz1;
 769         tx                         += fix1;
 770         ty                         += fiy1;
 771         tz                         += fiz1;
 772         f[i_coord_offset+DIM*2+XX] += fix2;
 773         f[i_coord_offset+DIM*2+YY] += fiy2;
 774         f[i_coord_offset+DIM*2+ZZ] += fiz2;
 775         tx                         += fix2;
 776         ty                         += fiy2;
 777         tz                         += fiz2;
 778         f[i_coord_offset+DIM*3+XX] += fix3;
 779         f[i_coord_offset+DIM*3+YY] += fiy3;
 780         f[i_coord_offset+DIM*3+ZZ] += fiz3;
 781         tx                         += fix3;
 782         ty                         += fiy3;
 783         tz                         += fiz3;
 784         fshift[i_shift_offset+XX]  += tx;
 785         fshift[i_shift_offset+YY]  += ty;
 786         fshift[i_shift_offset+ZZ]  += tz;
 787
 788         /* Increment number of inner iterations */
 789         inneriter                  += j_index_end - j_index_start;
 790
 791         /* Outer loop uses 39 flops */
 792     }
 793
 794     /* Increment number of outer iterations */
 795     outeriter        += nri;
 796
 797     /* Update outer/inner flops */
 798
 799     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W4_F,outeriter*39 + inneriter*160);
 800 }