src/gmxlib/nonbonded/nb_kernel_c/nb_kernel_ElecRFCut_VdwLJSw_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_ElecRFCut_VdwLJSw_GeomW4P1_VF_c
  35  * Electrostatics interaction: ReactionField
  36  * VdW interaction:            LennardJones
  37  * Geometry:                   Water4-Particle
  38  * Calculate force/pot:        PotentialAndForce
  39  */
  40 void
  41 nb_kernel_ElecRFCut_VdwLJSw_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     real             rswitch,swV3,swV4,swV5,swF2,swF3,swF4,d,d2,sw,dsw;
  77
  78     x                = xx[0];
  79     f                = ff[0];
  80
  81     nri              = nlist->nri;
  82     iinr             = nlist->iinr;
  83     jindex           = nlist->jindex;
  84     jjnr             = nlist->jjnr;
  85     shiftidx         = nlist->shift;
  86     gid              = nlist->gid;
  87     shiftvec         = fr->shift_vec[0];
  88     fshift           = fr->fshift[0];
  89     facel            = fr->epsfac;
  90     charge           = mdatoms->chargeA;
  91     krf              = fr->ic->k_rf;
  92     krf2             = krf*2.0;
  93     crf              = fr->ic->c_rf;
  94     nvdwtype         = fr->ntype;
  95     vdwparam         = fr->nbfp;
  96     vdwtype          = mdatoms->typeA;
  97
  98     /* Setup water-specific parameters */
  99     inr              = nlist->iinr[0];
 100     iq1              = facel*charge[inr+1];
 101     iq2              = facel*charge[inr+2];
 102     iq3              = facel*charge[inr+3];
 103     vdwioffset0      = 2*nvdwtype*vdwtype[inr+0];
 104
 105     /* When we use explicit cutoffs the value must be identical for elec and VdW, so use elec as an arbitrary choice */
 106     rcutoff          = fr->rcoulomb;
 107     rcutoff2         = rcutoff*rcutoff;
 108
 109     rswitch          = fr->rvdw_switch;
 110     /* Setup switch parameters */
 111     d                = rcutoff-rswitch;
 112     swV3             = -10.0/(d*d*d);
 113     swV4             =  15.0/(d*d*d*d);
 114     swV5             =  -6.0/(d*d*d*d*d);
 115     swF2             = -30.0/(d*d*d);
 116     swF3             =  60.0/(d*d*d*d);
 117     swF4             = -30.0/(d*d*d*d*d);
 118
 119     outeriter        = 0;
 120     inneriter        = 0;
 121
 122     /* Start outer loop over neighborlists */
 123     for(iidx=0; iidx<nri; iidx++)
 124     {
 125         /* Load shift vector for this list */
 126         i_shift_offset   = DIM*shiftidx[iidx];
 127         shX              = shiftvec[i_shift_offset+XX];
 128         shY              = shiftvec[i_shift_offset+YY];
 129         shZ              = shiftvec[i_shift_offset+ZZ];
 130
 131         /* Load limits for loop over neighbors */
 132         j_index_start    = jindex[iidx];
 133         j_index_end      = jindex[iidx+1];
 134
 135         /* Get outer coordinate index */
 136         inr              = iinr[iidx];
 137         i_coord_offset   = DIM*inr;
 138
 139         /* Load i particle coords and add shift vector */
 140         ix0              = shX + x[i_coord_offset+DIM*0+XX];
 141         iy0              = shY + x[i_coord_offset+DIM*0+YY];
 142         iz0              = shZ + x[i_coord_offset+DIM*0+ZZ];
 143         ix1              = shX + x[i_coord_offset+DIM*1+XX];
 144         iy1              = shY + x[i_coord_offset+DIM*1+YY];
 145         iz1              = shZ + x[i_coord_offset+DIM*1+ZZ];
 146         ix2              = shX + x[i_coord_offset+DIM*2+XX];
 147         iy2              = shY + x[i_coord_offset+DIM*2+YY];
 148         iz2              = shZ + x[i_coord_offset+DIM*2+ZZ];
 149         ix3              = shX + x[i_coord_offset+DIM*3+XX];
 150         iy3              = shY + x[i_coord_offset+DIM*3+YY];
 151         iz3              = shZ + x[i_coord_offset+DIM*3+ZZ];
 152
 153         fix0             = 0.0;
 154         fiy0             = 0.0;
 155         fiz0             = 0.0;
 156         fix1             = 0.0;
 157         fiy1             = 0.0;
 158         fiz1             = 0.0;
 159         fix2             = 0.0;
 160         fiy2             = 0.0;
 161         fiz2             = 0.0;
 162         fix3             = 0.0;
 163         fiy3             = 0.0;
 164         fiz3             = 0.0;
 165
 166         /* Reset potential sums */
 167         velecsum         = 0.0;
 168         vvdwsum          = 0.0;
 169
 170         /* Start inner kernel loop */
 171         for(jidx=j_index_start; jidx<j_index_end; jidx++)
 172         {
 173             /* Get j neighbor index, and coordinate index */
 174             jnr              = jjnr[jidx];
 175             j_coord_offset   = DIM*jnr;
 176
 177             /* load j atom coordinates */
 178             jx0              = x[j_coord_offset+DIM*0+XX];
 179             jy0              = x[j_coord_offset+DIM*0+YY];
 180             jz0              = x[j_coord_offset+DIM*0+ZZ];
 181
 182             /* Calculate displacement vector */
 183             dx00             = ix0 - jx0;
 184             dy00             = iy0 - jy0;
 185             dz00             = iz0 - jz0;
 186             dx10             = ix1 - jx0;
 187             dy10             = iy1 - jy0;
 188             dz10             = iz1 - jz0;
 189             dx20             = ix2 - jx0;
 190             dy20             = iy2 - jy0;
 191             dz20             = iz2 - jz0;
 192             dx30             = ix3 - jx0;
 193             dy30             = iy3 - jy0;
 194             dz30             = iz3 - jz0;
 195
 196             /* Calculate squared distance and things based on it */
 197             rsq00            = dx00*dx00+dy00*dy00+dz00*dz00;
 198             rsq10            = dx10*dx10+dy10*dy10+dz10*dz10;
 199             rsq20            = dx20*dx20+dy20*dy20+dz20*dz20;
 200             rsq30            = dx30*dx30+dy30*dy30+dz30*dz30;
 201
 202             rinv00           = gmx_invsqrt(rsq00);
 203             rinv10           = gmx_invsqrt(rsq10);
 204             rinv20           = gmx_invsqrt(rsq20);
 205             rinv30           = gmx_invsqrt(rsq30);
 206
 207             rinvsq00         = rinv00*rinv00;
 208             rinvsq10         = rinv10*rinv10;
 209             rinvsq20         = rinv20*rinv20;
 210             rinvsq30         = rinv30*rinv30;
 211
 212             /* Load parameters for j particles */
 213             jq0              = charge[jnr+0];
 214             vdwjidx0         = 2*vdwtype[jnr+0];
 215
 216             /**************************
 217              * CALCULATE INTERACTIONS *
 218              **************************/
 219
 220             if (rsq00<rcutoff2)
 221             {
 222
 223             r00              = rsq00*rinv00;
 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             d                = r00-rswitch;
 237             d                = (d>0.0) ? d : 0.0;
 238             d2               = d*d;
 239             sw               = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
 240
 241             dsw              = d2*(swF2+d*(swF3+d*swF4));
 242
 243             /* Evaluate switch function */
 244             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 245             fvdw             = fvdw*sw - rinv00*vvdw*dsw;
 246             vvdw            *= sw;
 247
 248             /* Update potential sums from outer loop */
 249             vvdwsum         += vvdw;
 250
 251             fscal            = fvdw;
 252
 253             /* Calculate temporary vectorial force */
 254             tx               = fscal*dx00;
 255             ty               = fscal*dy00;
 256             tz               = fscal*dz00;
 257
 258             /* Update vectorial force */
 259             fix0            += tx;
 260             fiy0            += ty;
 261             fiz0            += tz;
 262             f[j_coord_offset+DIM*0+XX] -= tx;
 263             f[j_coord_offset+DIM*0+YY] -= ty;
 264             f[j_coord_offset+DIM*0+ZZ] -= tz;
 265
 266             }
 267
 268             /**************************
 269              * CALCULATE INTERACTIONS *
 270              **************************/
 271
 272             if (rsq10<rcutoff2)
 273             {
 274
 275             qq10             = iq1*jq0;
 276
 277             /* REACTION-FIELD ELECTROSTATICS */
 278             velec            = qq10*(rinv10+krf*rsq10-crf);
 279             felec            = qq10*(rinv10*rinvsq10-krf2);
 280
 281             /* Update potential sums from outer loop */
 282             velecsum        += velec;
 283
 284             fscal            = felec;
 285
 286             /* Calculate temporary vectorial force */
 287             tx               = fscal*dx10;
 288             ty               = fscal*dy10;
 289             tz               = fscal*dz10;
 290
 291             /* Update vectorial force */
 292             fix1            += tx;
 293             fiy1            += ty;
 294             fiz1            += tz;
 295             f[j_coord_offset+DIM*0+XX] -= tx;
 296             f[j_coord_offset+DIM*0+YY] -= ty;
 297             f[j_coord_offset+DIM*0+ZZ] -= tz;
 298
 299             }
 300
 301             /**************************
 302              * CALCULATE INTERACTIONS *
 303              **************************/
 304
 305             if (rsq20<rcutoff2)
 306             {
 307
 308             qq20             = iq2*jq0;
 309
 310             /* REACTION-FIELD ELECTROSTATICS */
 311             velec            = qq20*(rinv20+krf*rsq20-crf);
 312             felec            = qq20*(rinv20*rinvsq20-krf2);
 313
 314             /* Update potential sums from outer loop */
 315             velecsum        += velec;
 316
 317             fscal            = felec;
 318
 319             /* Calculate temporary vectorial force */
 320             tx               = fscal*dx20;
 321             ty               = fscal*dy20;
 322             tz               = fscal*dz20;
 323
 324             /* Update vectorial force */
 325             fix2            += tx;
 326             fiy2            += ty;
 327             fiz2            += tz;
 328             f[j_coord_offset+DIM*0+XX] -= tx;
 329             f[j_coord_offset+DIM*0+YY] -= ty;
 330             f[j_coord_offset+DIM*0+ZZ] -= tz;
 331
 332             }
 333
 334             /**************************
 335              * CALCULATE INTERACTIONS *
 336              **************************/
 337
 338             if (rsq30<rcutoff2)
 339             {
 340
 341             qq30             = iq3*jq0;
 342
 343             /* REACTION-FIELD ELECTROSTATICS */
 344             velec            = qq30*(rinv30+krf*rsq30-crf);
 345             felec            = qq30*(rinv30*rinvsq30-krf2);
 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             }
 366
 367             /* Inner loop uses 149 flops */
 368         }
 369         /* End of innermost loop */
 370
 371         tx = ty = tz = 0;
 372         f[i_coord_offset+DIM*0+XX] += fix0;
 373         f[i_coord_offset+DIM*0+YY] += fiy0;
 374         f[i_coord_offset+DIM*0+ZZ] += fiz0;
 375         tx                         += fix0;
 376         ty                         += fiy0;
 377         tz                         += fiz0;
 378         f[i_coord_offset+DIM*1+XX] += fix1;
 379         f[i_coord_offset+DIM*1+YY] += fiy1;
 380         f[i_coord_offset+DIM*1+ZZ] += fiz1;
 381         tx                         += fix1;
 382         ty                         += fiy1;
 383         tz                         += fiz1;
 384         f[i_coord_offset+DIM*2+XX] += fix2;
 385         f[i_coord_offset+DIM*2+YY] += fiy2;
 386         f[i_coord_offset+DIM*2+ZZ] += fiz2;
 387         tx                         += fix2;
 388         ty                         += fiy2;
 389         tz                         += fiz2;
 390         f[i_coord_offset+DIM*3+XX] += fix3;
 391         f[i_coord_offset+DIM*3+YY] += fiy3;
 392         f[i_coord_offset+DIM*3+ZZ] += fiz3;
 393         tx                         += fix3;
 394         ty                         += fiy3;
 395         tz                         += fiz3;
 396         fshift[i_shift_offset+XX]  += tx;
 397         fshift[i_shift_offset+YY]  += ty;
 398         fshift[i_shift_offset+ZZ]  += tz;
 399
 400         ggid                        = gid[iidx];
 401         /* Update potential energies */
 402         kernel_data->energygrp_elec[ggid] += velecsum;
 403         kernel_data->energygrp_vdw[ggid] += vvdwsum;
 404
 405         /* Increment number of inner iterations */
 406         inneriter                  += j_index_end - j_index_start;
 407
 408         /* Outer loop uses 41 flops */
 409     }
 410
 411     /* Increment number of outer iterations */
 412     outeriter        += nri;
 413
 414     /* Update outer/inner flops */
 415
 416     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W4_VF,outeriter*41 + inneriter*149);
 417 }
 418 /*
 419  * Gromacs nonbonded kernel:   nb_kernel_ElecRFCut_VdwLJSw_GeomW4P1_F_c
 420  * Electrostatics interaction: ReactionField
 421  * VdW interaction:            LennardJones
 422  * Geometry:                   Water4-Particle
 423  * Calculate force/pot:        Force
 424  */
 425 void
 426 nb_kernel_ElecRFCut_VdwLJSw_GeomW4P1_F_c
 427                     (t_nblist * gmx_restrict                nlist,
 428                      rvec * gmx_restrict                    xx,
 429                      rvec * gmx_restrict                    ff,
 430                      t_forcerec * gmx_restrict              fr,
 431                      t_mdatoms * gmx_restrict               mdatoms,
 432                      nb_kernel_data_t * gmx_restrict        kernel_data,
 433                      t_nrnb * gmx_restrict                  nrnb)
 434 {
 435     int              i_shift_offset,i_coord_offset,j_coord_offset;
 436     int              j_index_start,j_index_end;
 437     int              nri,inr,ggid,iidx,jidx,jnr,outeriter,inneriter;
 438     real             shX,shY,shZ,tx,ty,tz,fscal,rcutoff,rcutoff2;
 439     int              *iinr,*jindex,*jjnr,*shiftidx,*gid;
 440     real             *shiftvec,*fshift,*x,*f;
 441     int              vdwioffset0;
 442     real             ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
 443     int              vdwioffset1;
 444     real             ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
 445     int              vdwioffset2;
 446     real             ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
 447     int              vdwioffset3;
 448     real             ix3,iy3,iz3,fix3,fiy3,fiz3,iq3,isai3;
 449     int              vdwjidx0;
 450     real             jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
 451     real             dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00,cexp1_00,cexp2_00;
 452     real             dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10,cexp1_10,cexp2_10;
 453     real             dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20,cexp1_20,cexp2_20;
 454     real             dx30,dy30,dz30,rsq30,rinv30,rinvsq30,r30,qq30,c6_30,c12_30,cexp1_30,cexp2_30;
 455     real             velec,felec,velecsum,facel,crf,krf,krf2;
 456     real             *charge;
 457     int              nvdwtype;
 458     real             rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,br,vvdwexp,sh_vdw_invrcut6;
 459     int              *vdwtype;
 460     real             *vdwparam;
 461     real             rswitch,swV3,swV4,swV5,swF2,swF3,swF4,d,d2,sw,dsw;
 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     krf              = fr->ic->k_rf;
 477     krf2             = krf*2.0;
 478     crf              = fr->ic->c_rf;
 479     nvdwtype         = fr->ntype;
 480     vdwparam         = fr->nbfp;
 481     vdwtype          = mdatoms->typeA;
 482
 483     /* Setup water-specific parameters */
 484     inr              = nlist->iinr[0];
 485     iq1              = facel*charge[inr+1];
 486     iq2              = facel*charge[inr+2];
 487     iq3              = facel*charge[inr+3];
 488     vdwioffset0      = 2*nvdwtype*vdwtype[inr+0];
 489
 490     /* When we use explicit cutoffs the value must be identical for elec and VdW, so use elec as an arbitrary choice */
 491     rcutoff          = fr->rcoulomb;
 492     rcutoff2         = rcutoff*rcutoff;
 493
 494     rswitch          = fr->rvdw_switch;
 495     /* Setup switch parameters */
 496     d                = rcutoff-rswitch;
 497     swV3             = -10.0/(d*d*d);
 498     swV4             =  15.0/(d*d*d*d);
 499     swV5             =  -6.0/(d*d*d*d*d);
 500     swF2             = -30.0/(d*d*d);
 501     swF3             =  60.0/(d*d*d*d);
 502     swF4             = -30.0/(d*d*d*d*d);
 503
 504     outeriter        = 0;
 505     inneriter        = 0;
 506
 507     /* Start outer loop over neighborlists */
 508     for(iidx=0; iidx<nri; iidx++)
 509     {
 510         /* Load shift vector for this list */
 511         i_shift_offset   = DIM*shiftidx[iidx];
 512         shX              = shiftvec[i_shift_offset+XX];
 513         shY              = shiftvec[i_shift_offset+YY];
 514         shZ              = shiftvec[i_shift_offset+ZZ];
 515
 516         /* Load limits for loop over neighbors */
 517         j_index_start    = jindex[iidx];
 518         j_index_end      = jindex[iidx+1];
 519
 520         /* Get outer coordinate index */
 521         inr              = iinr[iidx];
 522         i_coord_offset   = DIM*inr;
 523
 524         /* Load i particle coords and add shift vector */
 525         ix0              = shX + x[i_coord_offset+DIM*0+XX];
 526         iy0              = shY + x[i_coord_offset+DIM*0+YY];
 527         iz0              = shZ + x[i_coord_offset+DIM*0+ZZ];
 528         ix1              = shX + x[i_coord_offset+DIM*1+XX];
 529         iy1              = shY + x[i_coord_offset+DIM*1+YY];
 530         iz1              = shZ + x[i_coord_offset+DIM*1+ZZ];
 531         ix2              = shX + x[i_coord_offset+DIM*2+XX];
 532         iy2              = shY + x[i_coord_offset+DIM*2+YY];
 533         iz2              = shZ + x[i_coord_offset+DIM*2+ZZ];
 534         ix3              = shX + x[i_coord_offset+DIM*3+XX];
 535         iy3              = shY + x[i_coord_offset+DIM*3+YY];
 536         iz3              = shZ + x[i_coord_offset+DIM*3+ZZ];
 537
 538         fix0             = 0.0;
 539         fiy0             = 0.0;
 540         fiz0             = 0.0;
 541         fix1             = 0.0;
 542         fiy1             = 0.0;
 543         fiz1             = 0.0;
 544         fix2             = 0.0;
 545         fiy2             = 0.0;
 546         fiz2             = 0.0;
 547         fix3             = 0.0;
 548         fiy3             = 0.0;
 549         fiz3             = 0.0;
 550
 551         /* Start inner kernel loop */
 552         for(jidx=j_index_start; jidx<j_index_end; jidx++)
 553         {
 554             /* Get j neighbor index, and coordinate index */
 555             jnr              = jjnr[jidx];
 556             j_coord_offset   = DIM*jnr;
 557
 558             /* load j atom coordinates */
 559             jx0              = x[j_coord_offset+DIM*0+XX];
 560             jy0              = x[j_coord_offset+DIM*0+YY];
 561             jz0              = x[j_coord_offset+DIM*0+ZZ];
 562
 563             /* Calculate displacement vector */
 564             dx00             = ix0 - jx0;
 565             dy00             = iy0 - jy0;
 566             dz00             = iz0 - jz0;
 567             dx10             = ix1 - jx0;
 568             dy10             = iy1 - jy0;
 569             dz10             = iz1 - jz0;
 570             dx20             = ix2 - jx0;
 571             dy20             = iy2 - jy0;
 572             dz20             = iz2 - jz0;
 573             dx30             = ix3 - jx0;
 574             dy30             = iy3 - jy0;
 575             dz30             = iz3 - jz0;
 576
 577             /* Calculate squared distance and things based on it */
 578             rsq00            = dx00*dx00+dy00*dy00+dz00*dz00;
 579             rsq10            = dx10*dx10+dy10*dy10+dz10*dz10;
 580             rsq20            = dx20*dx20+dy20*dy20+dz20*dz20;
 581             rsq30            = dx30*dx30+dy30*dy30+dz30*dz30;
 582
 583             rinv00           = gmx_invsqrt(rsq00);
 584             rinv10           = gmx_invsqrt(rsq10);
 585             rinv20           = gmx_invsqrt(rsq20);
 586             rinv30           = gmx_invsqrt(rsq30);
 587
 588             rinvsq00         = rinv00*rinv00;
 589             rinvsq10         = rinv10*rinv10;
 590             rinvsq20         = rinv20*rinv20;
 591             rinvsq30         = rinv30*rinv30;
 592
 593             /* Load parameters for j particles */
 594             jq0              = charge[jnr+0];
 595             vdwjidx0         = 2*vdwtype[jnr+0];
 596
 597             /**************************
 598              * CALCULATE INTERACTIONS *
 599              **************************/
 600
 601             if (rsq00<rcutoff2)
 602             {
 603
 604             r00              = rsq00*rinv00;
 605
 606             c6_00            = vdwparam[vdwioffset0+vdwjidx0];
 607             c12_00           = vdwparam[vdwioffset0+vdwjidx0+1];
 608
 609             /* LENNARD-JONES DISPERSION/REPULSION */
 610
 611             rinvsix          = rinvsq00*rinvsq00*rinvsq00;
 612             vvdw6            = c6_00*rinvsix;
 613             vvdw12           = c12_00*rinvsix*rinvsix;
 614             vvdw             = vvdw12*(1.0/12.0) - vvdw6*(1.0/6.0);
 615             fvdw             = (vvdw12-vvdw6)*rinvsq00;
 616
 617             d                = r00-rswitch;
 618             d                = (d>0.0) ? d : 0.0;
 619             d2               = d*d;
 620             sw               = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
 621
 622             dsw              = d2*(swF2+d*(swF3+d*swF4));
 623
 624             /* Evaluate switch function */
 625             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 626             fvdw             = fvdw*sw - rinv00*vvdw*dsw;
 627
 628             fscal            = fvdw;
 629
 630             /* Calculate temporary vectorial force */
 631             tx               = fscal*dx00;
 632             ty               = fscal*dy00;
 633             tz               = fscal*dz00;
 634
 635             /* Update vectorial force */
 636             fix0            += tx;
 637             fiy0            += ty;
 638             fiz0            += 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
 645             /**************************
 646              * CALCULATE INTERACTIONS *
 647              **************************/
 648
 649             if (rsq10<rcutoff2)
 650             {
 651
 652             qq10             = iq1*jq0;
 653
 654             /* REACTION-FIELD ELECTROSTATICS */
 655             felec            = qq10*(rinv10*rinvsq10-krf2);
 656
 657             fscal            = felec;
 658
 659             /* Calculate temporary vectorial force */
 660             tx               = fscal*dx10;
 661             ty               = fscal*dy10;
 662             tz               = fscal*dz10;
 663
 664             /* Update vectorial force */
 665             fix1            += tx;
 666             fiy1            += ty;
 667             fiz1            += tz;
 668             f[j_coord_offset+DIM*0+XX] -= tx;
 669             f[j_coord_offset+DIM*0+YY] -= ty;
 670             f[j_coord_offset+DIM*0+ZZ] -= tz;
 671
 672             }
 673
 674             /**************************
 675              * CALCULATE INTERACTIONS *
 676              **************************/
 677
 678             if (rsq20<rcutoff2)
 679             {
 680
 681             qq20             = iq2*jq0;
 682
 683             /* REACTION-FIELD ELECTROSTATICS */
 684             felec            = qq20*(rinv20*rinvsq20-krf2);
 685
 686             fscal            = felec;
 687
 688             /* Calculate temporary vectorial force */
 689             tx               = fscal*dx20;
 690             ty               = fscal*dy20;
 691             tz               = fscal*dz20;
 692
 693             /* Update vectorial force */
 694             fix2            += tx;
 695             fiy2            += ty;
 696             fiz2            += tz;
 697             f[j_coord_offset+DIM*0+XX] -= tx;
 698             f[j_coord_offset+DIM*0+YY] -= ty;
 699             f[j_coord_offset+DIM*0+ZZ] -= tz;
 700
 701             }
 702
 703             /**************************
 704              * CALCULATE INTERACTIONS *
 705              **************************/
 706
 707             if (rsq30<rcutoff2)
 708             {
 709
 710             qq30             = iq3*jq0;
 711
 712             /* REACTION-FIELD ELECTROSTATICS */
 713             felec            = qq30*(rinv30*rinvsq30-krf2);
 714
 715             fscal            = felec;
 716
 717             /* Calculate temporary vectorial force */
 718             tx               = fscal*dx30;
 719             ty               = fscal*dy30;
 720             tz               = fscal*dz30;
 721
 722             /* Update vectorial force */
 723             fix3            += tx;
 724             fiy3            += ty;
 725             fiz3            += tz;
 726             f[j_coord_offset+DIM*0+XX] -= tx;
 727             f[j_coord_offset+DIM*0+YY] -= ty;
 728             f[j_coord_offset+DIM*0+ZZ] -= tz;
 729
 730             }
 731
 732             /* Inner loop uses 132 flops */
 733         }
 734         /* End of innermost loop */
 735
 736         tx = ty = tz = 0;
 737         f[i_coord_offset+DIM*0+XX] += fix0;
 738         f[i_coord_offset+DIM*0+YY] += fiy0;
 739         f[i_coord_offset+DIM*0+ZZ] += fiz0;
 740         tx                         += fix0;
 741         ty                         += fiy0;
 742         tz                         += fiz0;
 743         f[i_coord_offset+DIM*1+XX] += fix1;
 744         f[i_coord_offset+DIM*1+YY] += fiy1;
 745         f[i_coord_offset+DIM*1+ZZ] += fiz1;
 746         tx                         += fix1;
 747         ty                         += fiy1;
 748         tz                         += fiz1;
 749         f[i_coord_offset+DIM*2+XX] += fix2;
 750         f[i_coord_offset+DIM*2+YY] += fiy2;
 751         f[i_coord_offset+DIM*2+ZZ] += fiz2;
 752         tx                         += fix2;
 753         ty                         += fiy2;
 754         tz                         += fiz2;
 755         f[i_coord_offset+DIM*3+XX] += fix3;
 756         f[i_coord_offset+DIM*3+YY] += fiy3;
 757         f[i_coord_offset+DIM*3+ZZ] += fiz3;
 758         tx                         += fix3;
 759         ty                         += fiy3;
 760         tz                         += fiz3;
 761         fshift[i_shift_offset+XX]  += tx;
 762         fshift[i_shift_offset+YY]  += ty;
 763         fshift[i_shift_offset+ZZ]  += tz;
 764
 765         /* Increment number of inner iterations */
 766         inneriter                  += j_index_end - j_index_start;
 767
 768         /* Outer loop uses 39 flops */
 769     }
 770
 771     /* Increment number of outer iterations */
 772     outeriter        += nri;
 773
 774     /* Update outer/inner flops */
 775
 776     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W4_F,outeriter*39 + inneriter*132);
 777 }