src/gromacs/gmxlib/nonbonded/nb_kernel_c/nb_kernel_ElecEw_VdwLJ_GeomW4P1_c.c

   1 /*
   2  * This file is part of the GROMACS molecular simulation package.
   3  *
   4  * Copyright (c) 2012,2013,2014, by the GROMACS development team, led by
   5  * Mark Abraham, David van der Spoel, Berk Hess, and Erik Lindahl,
   6  * and including many others, as listed in the AUTHORS file in the
   7  * top-level source directory and at http://www.gromacs.org.
   8  *
   9  * GROMACS is free software; you can redistribute it and/or
  10  * modify it under the terms of the GNU Lesser General Public License
  11  * as published by the Free Software Foundation; either version 2.1
  12  * of the License, or (at your option) any later version.
  13  *
  14  * GROMACS is distributed in the hope that it will be useful,
  15  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  16  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  17  * Lesser General Public License for more details.
  18  *
  19  * You should have received a copy of the GNU Lesser General Public
  20  * License along with GROMACS; if not, see
  21  * http://www.gnu.org/licenses, or write to the Free Software Foundation,
  22  * Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02110-1301  USA.
  23  *
  24  * If you want to redistribute modifications to GROMACS, please
  25  * consider that scientific software is very special. Version
  26  * control is crucial - bugs must be traceable. We will be happy to
  27  * consider code for inclusion in the official distribution, but
  28  * derived work must not be called official GROMACS. Details are found
  29  * in the README & COPYING files - if they are missing, get the
  30  * official version at http://www.gromacs.org.
  31  *
  32  * To help us fund GROMACS development, we humbly ask that you cite
  33  * the research papers on the package. Check out http://www.gromacs.org.
  34  */
  35 /*
  36  * Note: this file was generated by the GROMACS c kernel generator.
  37  */
  38 #include "gmxpre.h"
  39
  40 #include "config.h"
  41
  42 #include <math.h>
  43
  44 #include "../nb_kernel.h"
  45 #include "gromacs/legacyheaders/types/simple.h"
  46 #include "gromacs/math/vec.h"
  47 #include "gromacs/legacyheaders/nrnb.h"
  48
  49 /*
  50  * Gromacs nonbonded kernel:   nb_kernel_ElecEw_VdwLJ_GeomW4P1_VF_c
  51  * Electrostatics interaction: Ewald
  52  * VdW interaction:            LennardJones
  53  * Geometry:                   Water4-Particle
  54  * Calculate force/pot:        PotentialAndForce
  55  */
  56 void
  57 nb_kernel_ElecEw_VdwLJ_GeomW4P1_VF_c
  58                     (t_nblist                    * gmx_restrict       nlist,
  59                      rvec                        * gmx_restrict          xx,
  60                      rvec                        * gmx_restrict          ff,
  61                      t_forcerec                  * gmx_restrict          fr,
  62                      t_mdatoms                   * gmx_restrict     mdatoms,
  63                      nb_kernel_data_t gmx_unused * gmx_restrict kernel_data,
  64                      t_nrnb                      * gmx_restrict        nrnb)
  65 {
  66     int              i_shift_offset,i_coord_offset,j_coord_offset;
  67     int              j_index_start,j_index_end;
  68     int              nri,inr,ggid,iidx,jidx,jnr,outeriter,inneriter;
  69     real             shX,shY,shZ,tx,ty,tz,fscal,rcutoff,rcutoff2;
  70     int              *iinr,*jindex,*jjnr,*shiftidx,*gid;
  71     real             *shiftvec,*fshift,*x,*f;
  72     int              vdwioffset0;
  73     real             ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
  74     int              vdwioffset1;
  75     real             ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
  76     int              vdwioffset2;
  77     real             ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
  78     int              vdwioffset3;
  79     real             ix3,iy3,iz3,fix3,fiy3,fiz3,iq3,isai3;
  80     int              vdwjidx0;
  81     real             jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
  82     real             dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00,cexp1_00,cexp2_00;
  83     real             dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10,cexp1_10,cexp2_10;
  84     real             dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20,cexp1_20,cexp2_20;
  85     real             dx30,dy30,dz30,rsq30,rinv30,rinvsq30,r30,qq30,c6_30,c12_30,cexp1_30,cexp2_30;
  86     real             velec,felec,velecsum,facel,crf,krf,krf2;
  87     real             *charge;
  88     int              nvdwtype;
  89     real             rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,br,vvdwexp,sh_vdw_invrcut6;
  90     int              *vdwtype;
  91     real             *vdwparam;
  92     int              ewitab;
  93     real             ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace;
  94     real             *ewtab;
  95
  96     x                = xx[0];
  97     f                = ff[0];
  98
  99     nri              = nlist->nri;
 100     iinr             = nlist->iinr;
 101     jindex           = nlist->jindex;
 102     jjnr             = nlist->jjnr;
 103     shiftidx         = nlist->shift;
 104     gid              = nlist->gid;
 105     shiftvec         = fr->shift_vec[0];
 106     fshift           = fr->fshift[0];
 107     facel            = fr->epsfac;
 108     charge           = mdatoms->chargeA;
 109     nvdwtype         = fr->ntype;
 110     vdwparam         = fr->nbfp;
 111     vdwtype          = mdatoms->typeA;
 112
 113     sh_ewald         = fr->ic->sh_ewald;
 114     ewtab            = fr->ic->tabq_coul_FDV0;
 115     ewtabscale       = fr->ic->tabq_scale;
 116     ewtabhalfspace   = 0.5/ewtabscale;
 117
 118     /* Setup water-specific parameters */
 119     inr              = nlist->iinr[0];
 120     iq1              = facel*charge[inr+1];
 121     iq2              = facel*charge[inr+2];
 122     iq3              = facel*charge[inr+3];
 123     vdwioffset0      = 2*nvdwtype*vdwtype[inr+0];
 124
 125     outeriter        = 0;
 126     inneriter        = 0;
 127
 128     /* Start outer loop over neighborlists */
 129     for(iidx=0; iidx<nri; iidx++)
 130     {
 131         /* Load shift vector for this list */
 132         i_shift_offset   = DIM*shiftidx[iidx];
 133         shX              = shiftvec[i_shift_offset+XX];
 134         shY              = shiftvec[i_shift_offset+YY];
 135         shZ              = shiftvec[i_shift_offset+ZZ];
 136
 137         /* Load limits for loop over neighbors */
 138         j_index_start    = jindex[iidx];
 139         j_index_end      = jindex[iidx+1];
 140
 141         /* Get outer coordinate index */
 142         inr              = iinr[iidx];
 143         i_coord_offset   = DIM*inr;
 144
 145         /* Load i particle coords and add shift vector */
 146         ix0              = shX + x[i_coord_offset+DIM*0+XX];
 147         iy0              = shY + x[i_coord_offset+DIM*0+YY];
 148         iz0              = shZ + x[i_coord_offset+DIM*0+ZZ];
 149         ix1              = shX + x[i_coord_offset+DIM*1+XX];
 150         iy1              = shY + x[i_coord_offset+DIM*1+YY];
 151         iz1              = shZ + x[i_coord_offset+DIM*1+ZZ];
 152         ix2              = shX + x[i_coord_offset+DIM*2+XX];
 153         iy2              = shY + x[i_coord_offset+DIM*2+YY];
 154         iz2              = shZ + x[i_coord_offset+DIM*2+ZZ];
 155         ix3              = shX + x[i_coord_offset+DIM*3+XX];
 156         iy3              = shY + x[i_coord_offset+DIM*3+YY];
 157         iz3              = shZ + x[i_coord_offset+DIM*3+ZZ];
 158
 159         fix0             = 0.0;
 160         fiy0             = 0.0;
 161         fiz0             = 0.0;
 162         fix1             = 0.0;
 163         fiy1             = 0.0;
 164         fiz1             = 0.0;
 165         fix2             = 0.0;
 166         fiy2             = 0.0;
 167         fiz2             = 0.0;
 168         fix3             = 0.0;
 169         fiy3             = 0.0;
 170         fiz3             = 0.0;
 171
 172         /* Reset potential sums */
 173         velecsum         = 0.0;
 174         vvdwsum          = 0.0;
 175
 176         /* Start inner kernel loop */
 177         for(jidx=j_index_start; jidx<j_index_end; jidx++)
 178         {
 179             /* Get j neighbor index, and coordinate index */
 180             jnr              = jjnr[jidx];
 181             j_coord_offset   = DIM*jnr;
 182
 183             /* load j atom coordinates */
 184             jx0              = x[j_coord_offset+DIM*0+XX];
 185             jy0              = x[j_coord_offset+DIM*0+YY];
 186             jz0              = x[j_coord_offset+DIM*0+ZZ];
 187
 188             /* Calculate displacement vector */
 189             dx00             = ix0 - jx0;
 190             dy00             = iy0 - jy0;
 191             dz00             = iz0 - jz0;
 192             dx10             = ix1 - jx0;
 193             dy10             = iy1 - jy0;
 194             dz10             = iz1 - jz0;
 195             dx20             = ix2 - jx0;
 196             dy20             = iy2 - jy0;
 197             dz20             = iz2 - jz0;
 198             dx30             = ix3 - jx0;
 199             dy30             = iy3 - jy0;
 200             dz30             = iz3 - jz0;
 201
 202             /* Calculate squared distance and things based on it */
 203             rsq00            = dx00*dx00+dy00*dy00+dz00*dz00;
 204             rsq10            = dx10*dx10+dy10*dy10+dz10*dz10;
 205             rsq20            = dx20*dx20+dy20*dy20+dz20*dz20;
 206             rsq30            = dx30*dx30+dy30*dy30+dz30*dz30;
 207
 208             rinv10           = gmx_invsqrt(rsq10);
 209             rinv20           = gmx_invsqrt(rsq20);
 210             rinv30           = gmx_invsqrt(rsq30);
 211
 212             rinvsq00         = 1.0/rsq00;
 213             rinvsq10         = rinv10*rinv10;
 214             rinvsq20         = rinv20*rinv20;
 215             rinvsq30         = rinv30*rinv30;
 216
 217             /* Load parameters for j particles */
 218             jq0              = charge[jnr+0];
 219             vdwjidx0         = 2*vdwtype[jnr+0];
 220
 221             /**************************
 222              * CALCULATE INTERACTIONS *
 223              **************************/
 224
 225             c6_00            = vdwparam[vdwioffset0+vdwjidx0];
 226             c12_00           = vdwparam[vdwioffset0+vdwjidx0+1];
 227
 228             /* LENNARD-JONES DISPERSION/REPULSION */
 229
 230             rinvsix          = rinvsq00*rinvsq00*rinvsq00;
 231             vvdw6            = c6_00*rinvsix;
 232             vvdw12           = c12_00*rinvsix*rinvsix;
 233             vvdw             = vvdw12*(1.0/12.0) - vvdw6*(1.0/6.0);
 234             fvdw             = (vvdw12-vvdw6)*rinvsq00;
 235
 236             /* Update potential sums from outer loop */
 237             vvdwsum         += vvdw;
 238
 239             fscal            = fvdw;
 240
 241             /* Calculate temporary vectorial force */
 242             tx               = fscal*dx00;
 243             ty               = fscal*dy00;
 244             tz               = fscal*dz00;
 245
 246             /* Update vectorial force */
 247             fix0            += tx;
 248             fiy0            += ty;
 249             fiz0            += tz;
 250             f[j_coord_offset+DIM*0+XX] -= tx;
 251             f[j_coord_offset+DIM*0+YY] -= ty;
 252             f[j_coord_offset+DIM*0+ZZ] -= tz;
 253
 254             /**************************
 255              * CALCULATE INTERACTIONS *
 256              **************************/
 257
 258             r10              = rsq10*rinv10;
 259
 260             qq10             = iq1*jq0;
 261
 262             /* EWALD ELECTROSTATICS */
 263
 264             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 265             ewrt             = r10*ewtabscale;
 266             ewitab           = ewrt;
 267             eweps            = ewrt-ewitab;
 268             ewitab           = 4*ewitab;
 269             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 270             velec            = qq10*(rinv10-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 271             felec            = qq10*rinv10*(rinvsq10-felec);
 272
 273             /* Update potential sums from outer loop */
 274             velecsum        += velec;
 275
 276             fscal            = felec;
 277
 278             /* Calculate temporary vectorial force */
 279             tx               = fscal*dx10;
 280             ty               = fscal*dy10;
 281             tz               = fscal*dz10;
 282
 283             /* Update vectorial force */
 284             fix1            += tx;
 285             fiy1            += ty;
 286             fiz1            += tz;
 287             f[j_coord_offset+DIM*0+XX] -= tx;
 288             f[j_coord_offset+DIM*0+YY] -= ty;
 289             f[j_coord_offset+DIM*0+ZZ] -= tz;
 290
 291             /**************************
 292              * CALCULATE INTERACTIONS *
 293              **************************/
 294
 295             r20              = rsq20*rinv20;
 296
 297             qq20             = iq2*jq0;
 298
 299             /* EWALD ELECTROSTATICS */
 300
 301             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 302             ewrt             = r20*ewtabscale;
 303             ewitab           = ewrt;
 304             eweps            = ewrt-ewitab;
 305             ewitab           = 4*ewitab;
 306             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 307             velec            = qq20*(rinv20-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 308             felec            = qq20*rinv20*(rinvsq20-felec);
 309
 310             /* Update potential sums from outer loop */
 311             velecsum        += velec;
 312
 313             fscal            = felec;
 314
 315             /* Calculate temporary vectorial force */
 316             tx               = fscal*dx20;
 317             ty               = fscal*dy20;
 318             tz               = fscal*dz20;
 319
 320             /* Update vectorial force */
 321             fix2            += tx;
 322             fiy2            += ty;
 323             fiz2            += tz;
 324             f[j_coord_offset+DIM*0+XX] -= tx;
 325             f[j_coord_offset+DIM*0+YY] -= ty;
 326             f[j_coord_offset+DIM*0+ZZ] -= tz;
 327
 328             /**************************
 329              * CALCULATE INTERACTIONS *
 330              **************************/
 331
 332             r30              = rsq30*rinv30;
 333
 334             qq30             = iq3*jq0;
 335
 336             /* EWALD ELECTROSTATICS */
 337
 338             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 339             ewrt             = r30*ewtabscale;
 340             ewitab           = ewrt;
 341             eweps            = ewrt-ewitab;
 342             ewitab           = 4*ewitab;
 343             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 344             velec            = qq30*(rinv30-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 345             felec            = qq30*rinv30*(rinvsq30-felec);
 346
 347             /* Update potential sums from outer loop */
 348             velecsum        += velec;
 349
 350             fscal            = felec;
 351
 352             /* Calculate temporary vectorial force */
 353             tx               = fscal*dx30;
 354             ty               = fscal*dy30;
 355             tz               = fscal*dz30;
 356
 357             /* Update vectorial force */
 358             fix3            += tx;
 359             fiy3            += ty;
 360             fiz3            += tz;
 361             f[j_coord_offset+DIM*0+XX] -= tx;
 362             f[j_coord_offset+DIM*0+YY] -= ty;
 363             f[j_coord_offset+DIM*0+ZZ] -= tz;
 364
 365             /* Inner loop uses 155 flops */
 366         }
 367         /* End of innermost loop */
 368
 369         tx = ty = tz = 0;
 370         f[i_coord_offset+DIM*0+XX] += fix0;
 371         f[i_coord_offset+DIM*0+YY] += fiy0;
 372         f[i_coord_offset+DIM*0+ZZ] += fiz0;
 373         tx                         += fix0;
 374         ty                         += fiy0;
 375         tz                         += fiz0;
 376         f[i_coord_offset+DIM*1+XX] += fix1;
 377         f[i_coord_offset+DIM*1+YY] += fiy1;
 378         f[i_coord_offset+DIM*1+ZZ] += fiz1;
 379         tx                         += fix1;
 380         ty                         += fiy1;
 381         tz                         += fiz1;
 382         f[i_coord_offset+DIM*2+XX] += fix2;
 383         f[i_coord_offset+DIM*2+YY] += fiy2;
 384         f[i_coord_offset+DIM*2+ZZ] += fiz2;
 385         tx                         += fix2;
 386         ty                         += fiy2;
 387         tz                         += fiz2;
 388         f[i_coord_offset+DIM*3+XX] += fix3;
 389         f[i_coord_offset+DIM*3+YY] += fiy3;
 390         f[i_coord_offset+DIM*3+ZZ] += fiz3;
 391         tx                         += fix3;
 392         ty                         += fiy3;
 393         tz                         += fiz3;
 394         fshift[i_shift_offset+XX]  += tx;
 395         fshift[i_shift_offset+YY]  += ty;
 396         fshift[i_shift_offset+ZZ]  += tz;
 397
 398         ggid                        = gid[iidx];
 399         /* Update potential energies */
 400         kernel_data->energygrp_elec[ggid] += velecsum;
 401         kernel_data->energygrp_vdw[ggid] += vvdwsum;
 402
 403         /* Increment number of inner iterations */
 404         inneriter                  += j_index_end - j_index_start;
 405
 406         /* Outer loop uses 41 flops */
 407     }
 408
 409     /* Increment number of outer iterations */
 410     outeriter        += nri;
 411
 412     /* Update outer/inner flops */
 413
 414     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W4_VF,outeriter*41 + inneriter*155);
 415 }
 416 /*
 417  * Gromacs nonbonded kernel:   nb_kernel_ElecEw_VdwLJ_GeomW4P1_F_c
 418  * Electrostatics interaction: Ewald
 419  * VdW interaction:            LennardJones
 420  * Geometry:                   Water4-Particle
 421  * Calculate force/pot:        Force
 422  */
 423 void
 424 nb_kernel_ElecEw_VdwLJ_GeomW4P1_F_c
 425                     (t_nblist                    * gmx_restrict       nlist,
 426                      rvec                        * gmx_restrict          xx,
 427                      rvec                        * gmx_restrict          ff,
 428                      t_forcerec                  * gmx_restrict          fr,
 429                      t_mdatoms                   * gmx_restrict     mdatoms,
 430                      nb_kernel_data_t gmx_unused * gmx_restrict kernel_data,
 431                      t_nrnb                      * gmx_restrict        nrnb)
 432 {
 433     int              i_shift_offset,i_coord_offset,j_coord_offset;
 434     int              j_index_start,j_index_end;
 435     int              nri,inr,ggid,iidx,jidx,jnr,outeriter,inneriter;
 436     real             shX,shY,shZ,tx,ty,tz,fscal,rcutoff,rcutoff2;
 437     int              *iinr,*jindex,*jjnr,*shiftidx,*gid;
 438     real             *shiftvec,*fshift,*x,*f;
 439     int              vdwioffset0;
 440     real             ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
 441     int              vdwioffset1;
 442     real             ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
 443     int              vdwioffset2;
 444     real             ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
 445     int              vdwioffset3;
 446     real             ix3,iy3,iz3,fix3,fiy3,fiz3,iq3,isai3;
 447     int              vdwjidx0;
 448     real             jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
 449     real             dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00,cexp1_00,cexp2_00;
 450     real             dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10,cexp1_10,cexp2_10;
 451     real             dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20,cexp1_20,cexp2_20;
 452     real             dx30,dy30,dz30,rsq30,rinv30,rinvsq30,r30,qq30,c6_30,c12_30,cexp1_30,cexp2_30;
 453     real             velec,felec,velecsum,facel,crf,krf,krf2;
 454     real             *charge;
 455     int              nvdwtype;
 456     real             rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,br,vvdwexp,sh_vdw_invrcut6;
 457     int              *vdwtype;
 458     real             *vdwparam;
 459     int              ewitab;
 460     real             ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace;
 461     real             *ewtab;
 462
 463     x                = xx[0];
 464     f                = ff[0];
 465
 466     nri              = nlist->nri;
 467     iinr             = nlist->iinr;
 468     jindex           = nlist->jindex;
 469     jjnr             = nlist->jjnr;
 470     shiftidx         = nlist->shift;
 471     gid              = nlist->gid;
 472     shiftvec         = fr->shift_vec[0];
 473     fshift           = fr->fshift[0];
 474     facel            = fr->epsfac;
 475     charge           = mdatoms->chargeA;
 476     nvdwtype         = fr->ntype;
 477     vdwparam         = fr->nbfp;
 478     vdwtype          = mdatoms->typeA;
 479
 480     sh_ewald         = fr->ic->sh_ewald;
 481     ewtab            = fr->ic->tabq_coul_F;
 482     ewtabscale       = fr->ic->tabq_scale;
 483     ewtabhalfspace   = 0.5/ewtabscale;
 484
 485     /* Setup water-specific parameters */
 486     inr              = nlist->iinr[0];
 487     iq1              = facel*charge[inr+1];
 488     iq2              = facel*charge[inr+2];
 489     iq3              = facel*charge[inr+3];
 490     vdwioffset0      = 2*nvdwtype*vdwtype[inr+0];
 491
 492     outeriter        = 0;
 493     inneriter        = 0;
 494
 495     /* Start outer loop over neighborlists */
 496     for(iidx=0; iidx<nri; iidx++)
 497     {
 498         /* Load shift vector for this list */
 499         i_shift_offset   = DIM*shiftidx[iidx];
 500         shX              = shiftvec[i_shift_offset+XX];
 501         shY              = shiftvec[i_shift_offset+YY];
 502         shZ              = shiftvec[i_shift_offset+ZZ];
 503
 504         /* Load limits for loop over neighbors */
 505         j_index_start    = jindex[iidx];
 506         j_index_end      = jindex[iidx+1];
 507
 508         /* Get outer coordinate index */
 509         inr              = iinr[iidx];
 510         i_coord_offset   = DIM*inr;
 511
 512         /* Load i particle coords and add shift vector */
 513         ix0              = shX + x[i_coord_offset+DIM*0+XX];
 514         iy0              = shY + x[i_coord_offset+DIM*0+YY];
 515         iz0              = shZ + x[i_coord_offset+DIM*0+ZZ];
 516         ix1              = shX + x[i_coord_offset+DIM*1+XX];
 517         iy1              = shY + x[i_coord_offset+DIM*1+YY];
 518         iz1              = shZ + x[i_coord_offset+DIM*1+ZZ];
 519         ix2              = shX + x[i_coord_offset+DIM*2+XX];
 520         iy2              = shY + x[i_coord_offset+DIM*2+YY];
 521         iz2              = shZ + x[i_coord_offset+DIM*2+ZZ];
 522         ix3              = shX + x[i_coord_offset+DIM*3+XX];
 523         iy3              = shY + x[i_coord_offset+DIM*3+YY];
 524         iz3              = shZ + x[i_coord_offset+DIM*3+ZZ];
 525
 526         fix0             = 0.0;
 527         fiy0             = 0.0;
 528         fiz0             = 0.0;
 529         fix1             = 0.0;
 530         fiy1             = 0.0;
 531         fiz1             = 0.0;
 532         fix2             = 0.0;
 533         fiy2             = 0.0;
 534         fiz2             = 0.0;
 535         fix3             = 0.0;
 536         fiy3             = 0.0;
 537         fiz3             = 0.0;
 538
 539         /* Start inner kernel loop */
 540         for(jidx=j_index_start; jidx<j_index_end; jidx++)
 541         {
 542             /* Get j neighbor index, and coordinate index */
 543             jnr              = jjnr[jidx];
 544             j_coord_offset   = DIM*jnr;
 545
 546             /* load j atom coordinates */
 547             jx0              = x[j_coord_offset+DIM*0+XX];
 548             jy0              = x[j_coord_offset+DIM*0+YY];
 549             jz0              = x[j_coord_offset+DIM*0+ZZ];
 550
 551             /* Calculate displacement vector */
 552             dx00             = ix0 - jx0;
 553             dy00             = iy0 - jy0;
 554             dz00             = iz0 - jz0;
 555             dx10             = ix1 - jx0;
 556             dy10             = iy1 - jy0;
 557             dz10             = iz1 - jz0;
 558             dx20             = ix2 - jx0;
 559             dy20             = iy2 - jy0;
 560             dz20             = iz2 - jz0;
 561             dx30             = ix3 - jx0;
 562             dy30             = iy3 - jy0;
 563             dz30             = iz3 - jz0;
 564
 565             /* Calculate squared distance and things based on it */
 566             rsq00            = dx00*dx00+dy00*dy00+dz00*dz00;
 567             rsq10            = dx10*dx10+dy10*dy10+dz10*dz10;
 568             rsq20            = dx20*dx20+dy20*dy20+dz20*dz20;
 569             rsq30            = dx30*dx30+dy30*dy30+dz30*dz30;
 570
 571             rinv10           = gmx_invsqrt(rsq10);
 572             rinv20           = gmx_invsqrt(rsq20);
 573             rinv30           = gmx_invsqrt(rsq30);
 574
 575             rinvsq00         = 1.0/rsq00;
 576             rinvsq10         = rinv10*rinv10;
 577             rinvsq20         = rinv20*rinv20;
 578             rinvsq30         = rinv30*rinv30;
 579
 580             /* Load parameters for j particles */
 581             jq0              = charge[jnr+0];
 582             vdwjidx0         = 2*vdwtype[jnr+0];
 583
 584             /**************************
 585              * CALCULATE INTERACTIONS *
 586              **************************/
 587
 588             c6_00            = vdwparam[vdwioffset0+vdwjidx0];
 589             c12_00           = vdwparam[vdwioffset0+vdwjidx0+1];
 590
 591             /* LENNARD-JONES DISPERSION/REPULSION */
 592
 593             rinvsix          = rinvsq00*rinvsq00*rinvsq00;
 594             fvdw             = (c12_00*rinvsix-c6_00)*rinvsix*rinvsq00;
 595
 596             fscal            = fvdw;
 597
 598             /* Calculate temporary vectorial force */
 599             tx               = fscal*dx00;
 600             ty               = fscal*dy00;
 601             tz               = fscal*dz00;
 602
 603             /* Update vectorial force */
 604             fix0            += tx;
 605             fiy0            += ty;
 606             fiz0            += tz;
 607             f[j_coord_offset+DIM*0+XX] -= tx;
 608             f[j_coord_offset+DIM*0+YY] -= ty;
 609             f[j_coord_offset+DIM*0+ZZ] -= tz;
 610
 611             /**************************
 612              * CALCULATE INTERACTIONS *
 613              **************************/
 614
 615             r10              = rsq10*rinv10;
 616
 617             qq10             = iq1*jq0;
 618
 619             /* EWALD ELECTROSTATICS */
 620
 621             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 622             ewrt             = r10*ewtabscale;
 623             ewitab           = ewrt;
 624             eweps            = ewrt-ewitab;
 625             felec            = (1.0-eweps)*ewtab[ewitab]+eweps*ewtab[ewitab+1];
 626             felec            = qq10*rinv10*(rinvsq10-felec);
 627
 628             fscal            = felec;
 629
 630             /* Calculate temporary vectorial force */
 631             tx               = fscal*dx10;
 632             ty               = fscal*dy10;
 633             tz               = fscal*dz10;
 634
 635             /* Update vectorial force */
 636             fix1            += tx;
 637             fiy1            += ty;
 638             fiz1            += tz;
 639             f[j_coord_offset+DIM*0+XX] -= tx;
 640             f[j_coord_offset+DIM*0+YY] -= ty;
 641             f[j_coord_offset+DIM*0+ZZ] -= tz;
 642
 643             /**************************
 644              * CALCULATE INTERACTIONS *
 645              **************************/
 646
 647             r20              = rsq20*rinv20;
 648
 649             qq20             = iq2*jq0;
 650
 651             /* EWALD ELECTROSTATICS */
 652
 653             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 654             ewrt             = r20*ewtabscale;
 655             ewitab           = ewrt;
 656             eweps            = ewrt-ewitab;
 657             felec            = (1.0-eweps)*ewtab[ewitab]+eweps*ewtab[ewitab+1];
 658             felec            = qq20*rinv20*(rinvsq20-felec);
 659
 660             fscal            = felec;
 661
 662             /* Calculate temporary vectorial force */
 663             tx               = fscal*dx20;
 664             ty               = fscal*dy20;
 665             tz               = fscal*dz20;
 666
 667             /* Update vectorial force */
 668             fix2            += tx;
 669             fiy2            += ty;
 670             fiz2            += tz;
 671             f[j_coord_offset+DIM*0+XX] -= tx;
 672             f[j_coord_offset+DIM*0+YY] -= ty;
 673             f[j_coord_offset+DIM*0+ZZ] -= tz;
 674
 675             /**************************
 676              * CALCULATE INTERACTIONS *
 677              **************************/
 678
 679             r30              = rsq30*rinv30;
 680
 681             qq30             = iq3*jq0;
 682
 683             /* EWALD ELECTROSTATICS */
 684
 685             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 686             ewrt             = r30*ewtabscale;
 687             ewitab           = ewrt;
 688             eweps            = ewrt-ewitab;
 689             felec            = (1.0-eweps)*ewtab[ewitab]+eweps*ewtab[ewitab+1];
 690             felec            = qq30*rinv30*(rinvsq30-felec);
 691
 692             fscal            = felec;
 693
 694             /* Calculate temporary vectorial force */
 695             tx               = fscal*dx30;
 696             ty               = fscal*dy30;
 697             tz               = fscal*dz30;
 698
 699             /* Update vectorial force */
 700             fix3            += tx;
 701             fiy3            += ty;
 702             fiz3            += tz;
 703             f[j_coord_offset+DIM*0+XX] -= tx;
 704             f[j_coord_offset+DIM*0+YY] -= ty;
 705             f[j_coord_offset+DIM*0+ZZ] -= tz;
 706
 707             /* Inner loop uses 129 flops */
 708         }
 709         /* End of innermost loop */
 710
 711         tx = ty = tz = 0;
 712         f[i_coord_offset+DIM*0+XX] += fix0;
 713         f[i_coord_offset+DIM*0+YY] += fiy0;
 714         f[i_coord_offset+DIM*0+ZZ] += fiz0;
 715         tx                         += fix0;
 716         ty                         += fiy0;
 717         tz                         += fiz0;
 718         f[i_coord_offset+DIM*1+XX] += fix1;
 719         f[i_coord_offset+DIM*1+YY] += fiy1;
 720         f[i_coord_offset+DIM*1+ZZ] += fiz1;
 721         tx                         += fix1;
 722         ty                         += fiy1;
 723         tz                         += fiz1;
 724         f[i_coord_offset+DIM*2+XX] += fix2;
 725         f[i_coord_offset+DIM*2+YY] += fiy2;
 726         f[i_coord_offset+DIM*2+ZZ] += fiz2;
 727         tx                         += fix2;
 728         ty                         += fiy2;
 729         tz                         += fiz2;
 730         f[i_coord_offset+DIM*3+XX] += fix3;
 731         f[i_coord_offset+DIM*3+YY] += fiy3;
 732         f[i_coord_offset+DIM*3+ZZ] += fiz3;
 733         tx                         += fix3;
 734         ty                         += fiy3;
 735         tz                         += fiz3;
 736         fshift[i_shift_offset+XX]  += tx;
 737         fshift[i_shift_offset+YY]  += ty;
 738         fshift[i_shift_offset+ZZ]  += tz;
 739
 740         /* Increment number of inner iterations */
 741         inneriter                  += j_index_end - j_index_start;
 742
 743         /* Outer loop uses 39 flops */
 744     }
 745
 746     /* Increment number of outer iterations */
 747     outeriter        += nri;
 748
 749     /* Update outer/inner flops */
 750
 751     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W4_F,outeriter*39 + inneriter*129);
 752 }