src/gmxlib/nonbonded/nb_kernel_c/nb_kernel_ElecEw_VdwBham_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_ElecEw_VdwBham_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_ElecEw_VdwBham_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     outeriter        = 0;
 110     inneriter        = 0;
 111
 112     /* Start outer loop over neighborlists */
 113     for(iidx=0; iidx<nri; iidx++)
 114     {
 115         /* Load shift vector for this list */
 116         i_shift_offset   = DIM*shiftidx[iidx];
 117         shX              = shiftvec[i_shift_offset+XX];
 118         shY              = shiftvec[i_shift_offset+YY];
 119         shZ              = shiftvec[i_shift_offset+ZZ];
 120
 121         /* Load limits for loop over neighbors */
 122         j_index_start    = jindex[iidx];
 123         j_index_end      = jindex[iidx+1];
 124
 125         /* Get outer coordinate index */
 126         inr              = iinr[iidx];
 127         i_coord_offset   = DIM*inr;
 128
 129         /* Load i particle coords and add shift vector */
 130         ix0              = shX + x[i_coord_offset+DIM*0+XX];
 131         iy0              = shY + x[i_coord_offset+DIM*0+YY];
 132         iz0              = shZ + x[i_coord_offset+DIM*0+ZZ];
 133         ix1              = shX + x[i_coord_offset+DIM*1+XX];
 134         iy1              = shY + x[i_coord_offset+DIM*1+YY];
 135         iz1              = shZ + x[i_coord_offset+DIM*1+ZZ];
 136         ix2              = shX + x[i_coord_offset+DIM*2+XX];
 137         iy2              = shY + x[i_coord_offset+DIM*2+YY];
 138         iz2              = shZ + x[i_coord_offset+DIM*2+ZZ];
 139         ix3              = shX + x[i_coord_offset+DIM*3+XX];
 140         iy3              = shY + x[i_coord_offset+DIM*3+YY];
 141         iz3              = shZ + x[i_coord_offset+DIM*3+ZZ];
 142
 143         fix0             = 0.0;
 144         fiy0             = 0.0;
 145         fiz0             = 0.0;
 146         fix1             = 0.0;
 147         fiy1             = 0.0;
 148         fiz1             = 0.0;
 149         fix2             = 0.0;
 150         fiy2             = 0.0;
 151         fiz2             = 0.0;
 152         fix3             = 0.0;
 153         fiy3             = 0.0;
 154         fiz3             = 0.0;
 155
 156         /* Reset potential sums */
 157         velecsum         = 0.0;
 158         vvdwsum          = 0.0;
 159
 160         /* Start inner kernel loop */
 161         for(jidx=j_index_start; jidx<j_index_end; jidx++)
 162         {
 163             /* Get j neighbor index, and coordinate index */
 164             jnr              = jjnr[jidx];
 165             j_coord_offset   = DIM*jnr;
 166
 167             /* load j atom coordinates */
 168             jx0              = x[j_coord_offset+DIM*0+XX];
 169             jy0              = x[j_coord_offset+DIM*0+YY];
 170             jz0              = x[j_coord_offset+DIM*0+ZZ];
 171
 172             /* Calculate displacement vector */
 173             dx00             = ix0 - jx0;
 174             dy00             = iy0 - jy0;
 175             dz00             = iz0 - jz0;
 176             dx10             = ix1 - jx0;
 177             dy10             = iy1 - jy0;
 178             dz10             = iz1 - jz0;
 179             dx20             = ix2 - jx0;
 180             dy20             = iy2 - jy0;
 181             dz20             = iz2 - jz0;
 182             dx30             = ix3 - jx0;
 183             dy30             = iy3 - jy0;
 184             dz30             = iz3 - jz0;
 185
 186             /* Calculate squared distance and things based on it */
 187             rsq00            = dx00*dx00+dy00*dy00+dz00*dz00;
 188             rsq10            = dx10*dx10+dy10*dy10+dz10*dz10;
 189             rsq20            = dx20*dx20+dy20*dy20+dz20*dz20;
 190             rsq30            = dx30*dx30+dy30*dy30+dz30*dz30;
 191
 192             rinv00           = gmx_invsqrt(rsq00);
 193             rinv10           = gmx_invsqrt(rsq10);
 194             rinv20           = gmx_invsqrt(rsq20);
 195             rinv30           = gmx_invsqrt(rsq30);
 196
 197             rinvsq00         = rinv00*rinv00;
 198             rinvsq10         = rinv10*rinv10;
 199             rinvsq20         = rinv20*rinv20;
 200             rinvsq30         = rinv30*rinv30;
 201
 202             /* Load parameters for j particles */
 203             jq0              = charge[jnr+0];
 204             vdwjidx0         = 3*vdwtype[jnr+0];
 205
 206             /**************************
 207              * CALCULATE INTERACTIONS *
 208              **************************/
 209
 210             r00              = rsq00*rinv00;
 211
 212             c6_00            = vdwparam[vdwioffset0+vdwjidx0];
 213             cexp1_00         = vdwparam[vdwioffset0+vdwjidx0+1];
 214             cexp2_00         = vdwparam[vdwioffset0+vdwjidx0+2];
 215
 216             /* BUCKINGHAM DISPERSION/REPULSION */
 217             rinvsix          = rinvsq00*rinvsq00*rinvsq00;
 218             vvdw6            = c6_00*rinvsix;
 219             br               = cexp2_00*r00;
 220             vvdwexp          = cexp1_00*exp(-br);
 221             vvdw             = vvdwexp - vvdw6*(1.0/6.0);
 222             fvdw             = (br*vvdwexp-vvdw6)*rinvsq00;
 223
 224             /* Update potential sums from outer loop */
 225             vvdwsum         += vvdw;
 226
 227             fscal            = fvdw;
 228
 229             /* Calculate temporary vectorial force */
 230             tx               = fscal*dx00;
 231             ty               = fscal*dy00;
 232             tz               = fscal*dz00;
 233
 234             /* Update vectorial force */
 235             fix0            += tx;
 236             fiy0            += ty;
 237             fiz0            += tz;
 238             f[j_coord_offset+DIM*0+XX] -= tx;
 239             f[j_coord_offset+DIM*0+YY] -= ty;
 240             f[j_coord_offset+DIM*0+ZZ] -= tz;
 241
 242             /**************************
 243              * CALCULATE INTERACTIONS *
 244              **************************/
 245
 246             r10              = rsq10*rinv10;
 247
 248             qq10             = iq1*jq0;
 249
 250             /* EWALD ELECTROSTATICS */
 251
 252             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 253             ewrt             = r10*ewtabscale;
 254             ewitab           = ewrt;
 255             eweps            = ewrt-ewitab;
 256             ewitab           = 4*ewitab;
 257             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 258             velec            = qq10*(rinv10-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 259             felec            = qq10*rinv10*(rinvsq10-felec);
 260
 261             /* Update potential sums from outer loop */
 262             velecsum        += velec;
 263
 264             fscal            = felec;
 265
 266             /* Calculate temporary vectorial force */
 267             tx               = fscal*dx10;
 268             ty               = fscal*dy10;
 269             tz               = fscal*dz10;
 270
 271             /* Update vectorial force */
 272             fix1            += tx;
 273             fiy1            += ty;
 274             fiz1            += tz;
 275             f[j_coord_offset+DIM*0+XX] -= tx;
 276             f[j_coord_offset+DIM*0+YY] -= ty;
 277             f[j_coord_offset+DIM*0+ZZ] -= tz;
 278
 279             /**************************
 280              * CALCULATE INTERACTIONS *
 281              **************************/
 282
 283             r20              = rsq20*rinv20;
 284
 285             qq20             = iq2*jq0;
 286
 287             /* EWALD ELECTROSTATICS */
 288
 289             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 290             ewrt             = r20*ewtabscale;
 291             ewitab           = ewrt;
 292             eweps            = ewrt-ewitab;
 293             ewitab           = 4*ewitab;
 294             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 295             velec            = qq20*(rinv20-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 296             felec            = qq20*rinv20*(rinvsq20-felec);
 297
 298             /* Update potential sums from outer loop */
 299             velecsum        += velec;
 300
 301             fscal            = felec;
 302
 303             /* Calculate temporary vectorial force */
 304             tx               = fscal*dx20;
 305             ty               = fscal*dy20;
 306             tz               = fscal*dz20;
 307
 308             /* Update vectorial force */
 309             fix2            += tx;
 310             fiy2            += ty;
 311             fiz2            += tz;
 312             f[j_coord_offset+DIM*0+XX] -= tx;
 313             f[j_coord_offset+DIM*0+YY] -= ty;
 314             f[j_coord_offset+DIM*0+ZZ] -= tz;
 315
 316             /**************************
 317              * CALCULATE INTERACTIONS *
 318              **************************/
 319
 320             r30              = rsq30*rinv30;
 321
 322             qq30             = iq3*jq0;
 323
 324             /* EWALD ELECTROSTATICS */
 325
 326             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 327             ewrt             = r30*ewtabscale;
 328             ewitab           = ewrt;
 329             eweps            = ewrt-ewitab;
 330             ewitab           = 4*ewitab;
 331             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 332             velec            = qq30*(rinv30-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 333             felec            = qq30*rinv30*(rinvsq30-felec);
 334
 335             /* Update potential sums from outer loop */
 336             velecsum        += velec;
 337
 338             fscal            = felec;
 339
 340             /* Calculate temporary vectorial force */
 341             tx               = fscal*dx30;
 342             ty               = fscal*dy30;
 343             tz               = fscal*dz30;
 344
 345             /* Update vectorial force */
 346             fix3            += tx;
 347             fiy3            += ty;
 348             fiz3            += tz;
 349             f[j_coord_offset+DIM*0+XX] -= tx;
 350             f[j_coord_offset+DIM*0+YY] -= ty;
 351             f[j_coord_offset+DIM*0+ZZ] -= tz;
 352
 353             /* Inner loop uses 184 flops */
 354         }
 355         /* End of innermost loop */
 356
 357         tx = ty = tz = 0;
 358         f[i_coord_offset+DIM*0+XX] += fix0;
 359         f[i_coord_offset+DIM*0+YY] += fiy0;
 360         f[i_coord_offset+DIM*0+ZZ] += fiz0;
 361         tx                         += fix0;
 362         ty                         += fiy0;
 363         tz                         += fiz0;
 364         f[i_coord_offset+DIM*1+XX] += fix1;
 365         f[i_coord_offset+DIM*1+YY] += fiy1;
 366         f[i_coord_offset+DIM*1+ZZ] += fiz1;
 367         tx                         += fix1;
 368         ty                         += fiy1;
 369         tz                         += fiz1;
 370         f[i_coord_offset+DIM*2+XX] += fix2;
 371         f[i_coord_offset+DIM*2+YY] += fiy2;
 372         f[i_coord_offset+DIM*2+ZZ] += fiz2;
 373         tx                         += fix2;
 374         ty                         += fiy2;
 375         tz                         += fiz2;
 376         f[i_coord_offset+DIM*3+XX] += fix3;
 377         f[i_coord_offset+DIM*3+YY] += fiy3;
 378         f[i_coord_offset+DIM*3+ZZ] += fiz3;
 379         tx                         += fix3;
 380         ty                         += fiy3;
 381         tz                         += fiz3;
 382         fshift[i_shift_offset+XX]  += tx;
 383         fshift[i_shift_offset+YY]  += ty;
 384         fshift[i_shift_offset+ZZ]  += tz;
 385
 386         ggid                        = gid[iidx];
 387         /* Update potential energies */
 388         kernel_data->energygrp_elec[ggid] += velecsum;
 389         kernel_data->energygrp_vdw[ggid] += vvdwsum;
 390
 391         /* Increment number of inner iterations */
 392         inneriter                  += j_index_end - j_index_start;
 393
 394         /* Outer loop uses 41 flops */
 395     }
 396
 397     /* Increment number of outer iterations */
 398     outeriter        += nri;
 399
 400     /* Update outer/inner flops */
 401
 402     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W4_VF,outeriter*41 + inneriter*184);
 403 }
 404 /*
 405  * Gromacs nonbonded kernel:   nb_kernel_ElecEw_VdwBham_GeomW4P1_F_c
 406  * Electrostatics interaction: Ewald
 407  * VdW interaction:            Buckingham
 408  * Geometry:                   Water4-Particle
 409  * Calculate force/pot:        Force
 410  */
 411 void
 412 nb_kernel_ElecEw_VdwBham_GeomW4P1_F_c
 413                     (t_nblist * gmx_restrict                nlist,
 414                      rvec * gmx_restrict                    xx,
 415                      rvec * gmx_restrict                    ff,
 416                      t_forcerec * gmx_restrict              fr,
 417                      t_mdatoms * gmx_restrict               mdatoms,
 418                      nb_kernel_data_t * gmx_restrict        kernel_data,
 419                      t_nrnb * gmx_restrict                  nrnb)
 420 {
 421     int              i_shift_offset,i_coord_offset,j_coord_offset;
 422     int              j_index_start,j_index_end;
 423     int              nri,inr,ggid,iidx,jidx,jnr,outeriter,inneriter;
 424     real             shX,shY,shZ,tx,ty,tz,fscal,rcutoff,rcutoff2;
 425     int              *iinr,*jindex,*jjnr,*shiftidx,*gid;
 426     real             *shiftvec,*fshift,*x,*f;
 427     int              vdwioffset0;
 428     real             ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
 429     int              vdwioffset1;
 430     real             ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
 431     int              vdwioffset2;
 432     real             ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
 433     int              vdwioffset3;
 434     real             ix3,iy3,iz3,fix3,fiy3,fiz3,iq3,isai3;
 435     int              vdwjidx0;
 436     real             jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
 437     real             dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00,cexp1_00,cexp2_00;
 438     real             dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10,cexp1_10,cexp2_10;
 439     real             dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20,cexp1_20,cexp2_20;
 440     real             dx30,dy30,dz30,rsq30,rinv30,rinvsq30,r30,qq30,c6_30,c12_30,cexp1_30,cexp2_30;
 441     real             velec,felec,velecsum,facel,crf,krf,krf2;
 442     real             *charge;
 443     int              nvdwtype;
 444     real             rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,br,vvdwexp,sh_vdw_invrcut6;
 445     int              *vdwtype;
 446     real             *vdwparam;
 447     int              ewitab;
 448     real             ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace;
 449     real             *ewtab;
 450
 451     x                = xx[0];
 452     f                = ff[0];
 453
 454     nri              = nlist->nri;
 455     iinr             = nlist->iinr;
 456     jindex           = nlist->jindex;
 457     jjnr             = nlist->jjnr;
 458     shiftidx         = nlist->shift;
 459     gid              = nlist->gid;
 460     shiftvec         = fr->shift_vec[0];
 461     fshift           = fr->fshift[0];
 462     facel            = fr->epsfac;
 463     charge           = mdatoms->chargeA;
 464     nvdwtype         = fr->ntype;
 465     vdwparam         = fr->nbfp;
 466     vdwtype          = mdatoms->typeA;
 467
 468     sh_ewald         = fr->ic->sh_ewald;
 469     ewtab            = fr->ic->tabq_coul_F;
 470     ewtabscale       = fr->ic->tabq_scale;
 471     ewtabhalfspace   = 0.5/ewtabscale;
 472
 473     /* Setup water-specific parameters */
 474     inr              = nlist->iinr[0];
 475     iq1              = facel*charge[inr+1];
 476     iq2              = facel*charge[inr+2];
 477     iq3              = facel*charge[inr+3];
 478     vdwioffset0      = 3*nvdwtype*vdwtype[inr+0];
 479
 480     outeriter        = 0;
 481     inneriter        = 0;
 482
 483     /* Start outer loop over neighborlists */
 484     for(iidx=0; iidx<nri; iidx++)
 485     {
 486         /* Load shift vector for this list */
 487         i_shift_offset   = DIM*shiftidx[iidx];
 488         shX              = shiftvec[i_shift_offset+XX];
 489         shY              = shiftvec[i_shift_offset+YY];
 490         shZ              = shiftvec[i_shift_offset+ZZ];
 491
 492         /* Load limits for loop over neighbors */
 493         j_index_start    = jindex[iidx];
 494         j_index_end      = jindex[iidx+1];
 495
 496         /* Get outer coordinate index */
 497         inr              = iinr[iidx];
 498         i_coord_offset   = DIM*inr;
 499
 500         /* Load i particle coords and add shift vector */
 501         ix0              = shX + x[i_coord_offset+DIM*0+XX];
 502         iy0              = shY + x[i_coord_offset+DIM*0+YY];
 503         iz0              = shZ + x[i_coord_offset+DIM*0+ZZ];
 504         ix1              = shX + x[i_coord_offset+DIM*1+XX];
 505         iy1              = shY + x[i_coord_offset+DIM*1+YY];
 506         iz1              = shZ + x[i_coord_offset+DIM*1+ZZ];
 507         ix2              = shX + x[i_coord_offset+DIM*2+XX];
 508         iy2              = shY + x[i_coord_offset+DIM*2+YY];
 509         iz2              = shZ + x[i_coord_offset+DIM*2+ZZ];
 510         ix3              = shX + x[i_coord_offset+DIM*3+XX];
 511         iy3              = shY + x[i_coord_offset+DIM*3+YY];
 512         iz3              = shZ + x[i_coord_offset+DIM*3+ZZ];
 513
 514         fix0             = 0.0;
 515         fiy0             = 0.0;
 516         fiz0             = 0.0;
 517         fix1             = 0.0;
 518         fiy1             = 0.0;
 519         fiz1             = 0.0;
 520         fix2             = 0.0;
 521         fiy2             = 0.0;
 522         fiz2             = 0.0;
 523         fix3             = 0.0;
 524         fiy3             = 0.0;
 525         fiz3             = 0.0;
 526
 527         /* Start inner kernel loop */
 528         for(jidx=j_index_start; jidx<j_index_end; jidx++)
 529         {
 530             /* Get j neighbor index, and coordinate index */
 531             jnr              = jjnr[jidx];
 532             j_coord_offset   = DIM*jnr;
 533
 534             /* load j atom coordinates */
 535             jx0              = x[j_coord_offset+DIM*0+XX];
 536             jy0              = x[j_coord_offset+DIM*0+YY];
 537             jz0              = x[j_coord_offset+DIM*0+ZZ];
 538
 539             /* Calculate displacement vector */
 540             dx00             = ix0 - jx0;
 541             dy00             = iy0 - jy0;
 542             dz00             = iz0 - jz0;
 543             dx10             = ix1 - jx0;
 544             dy10             = iy1 - jy0;
 545             dz10             = iz1 - jz0;
 546             dx20             = ix2 - jx0;
 547             dy20             = iy2 - jy0;
 548             dz20             = iz2 - jz0;
 549             dx30             = ix3 - jx0;
 550             dy30             = iy3 - jy0;
 551             dz30             = iz3 - jz0;
 552
 553             /* Calculate squared distance and things based on it */
 554             rsq00            = dx00*dx00+dy00*dy00+dz00*dz00;
 555             rsq10            = dx10*dx10+dy10*dy10+dz10*dz10;
 556             rsq20            = dx20*dx20+dy20*dy20+dz20*dz20;
 557             rsq30            = dx30*dx30+dy30*dy30+dz30*dz30;
 558
 559             rinv00           = gmx_invsqrt(rsq00);
 560             rinv10           = gmx_invsqrt(rsq10);
 561             rinv20           = gmx_invsqrt(rsq20);
 562             rinv30           = gmx_invsqrt(rsq30);
 563
 564             rinvsq00         = rinv00*rinv00;
 565             rinvsq10         = rinv10*rinv10;
 566             rinvsq20         = rinv20*rinv20;
 567             rinvsq30         = rinv30*rinv30;
 568
 569             /* Load parameters for j particles */
 570             jq0              = charge[jnr+0];
 571             vdwjidx0         = 3*vdwtype[jnr+0];
 572
 573             /**************************
 574              * CALCULATE INTERACTIONS *
 575              **************************/
 576
 577             r00              = rsq00*rinv00;
 578
 579             c6_00            = vdwparam[vdwioffset0+vdwjidx0];
 580             cexp1_00         = vdwparam[vdwioffset0+vdwjidx0+1];
 581             cexp2_00         = vdwparam[vdwioffset0+vdwjidx0+2];
 582
 583             /* BUCKINGHAM DISPERSION/REPULSION */
 584             rinvsix          = rinvsq00*rinvsq00*rinvsq00;
 585             vvdw6            = c6_00*rinvsix;
 586             br               = cexp2_00*r00;
 587             vvdwexp          = cexp1_00*exp(-br);
 588             fvdw             = (br*vvdwexp-vvdw6)*rinvsq00;
 589
 590             fscal            = fvdw;
 591
 592             /* Calculate temporary vectorial force */
 593             tx               = fscal*dx00;
 594             ty               = fscal*dy00;
 595             tz               = fscal*dz00;
 596
 597             /* Update vectorial force */
 598             fix0            += tx;
 599             fiy0            += ty;
 600             fiz0            += tz;
 601             f[j_coord_offset+DIM*0+XX] -= tx;
 602             f[j_coord_offset+DIM*0+YY] -= ty;
 603             f[j_coord_offset+DIM*0+ZZ] -= tz;
 604
 605             /**************************
 606              * CALCULATE INTERACTIONS *
 607              **************************/
 608
 609             r10              = rsq10*rinv10;
 610
 611             qq10             = iq1*jq0;
 612
 613             /* EWALD ELECTROSTATICS */
 614
 615             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 616             ewrt             = r10*ewtabscale;
 617             ewitab           = ewrt;
 618             eweps            = ewrt-ewitab;
 619             felec            = (1.0-eweps)*ewtab[ewitab]+eweps*ewtab[ewitab+1];
 620             felec            = qq10*rinv10*(rinvsq10-felec);
 621
 622             fscal            = felec;
 623
 624             /* Calculate temporary vectorial force */
 625             tx               = fscal*dx10;
 626             ty               = fscal*dy10;
 627             tz               = fscal*dz10;
 628
 629             /* Update vectorial force */
 630             fix1            += tx;
 631             fiy1            += ty;
 632             fiz1            += tz;
 633             f[j_coord_offset+DIM*0+XX] -= tx;
 634             f[j_coord_offset+DIM*0+YY] -= ty;
 635             f[j_coord_offset+DIM*0+ZZ] -= tz;
 636
 637             /**************************
 638              * CALCULATE INTERACTIONS *
 639              **************************/
 640
 641             r20              = rsq20*rinv20;
 642
 643             qq20             = iq2*jq0;
 644
 645             /* EWALD ELECTROSTATICS */
 646
 647             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 648             ewrt             = r20*ewtabscale;
 649             ewitab           = ewrt;
 650             eweps            = ewrt-ewitab;
 651             felec            = (1.0-eweps)*ewtab[ewitab]+eweps*ewtab[ewitab+1];
 652             felec            = qq20*rinv20*(rinvsq20-felec);
 653
 654             fscal            = felec;
 655
 656             /* Calculate temporary vectorial force */
 657             tx               = fscal*dx20;
 658             ty               = fscal*dy20;
 659             tz               = fscal*dz20;
 660
 661             /* Update vectorial force */
 662             fix2            += tx;
 663             fiy2            += ty;
 664             fiz2            += tz;
 665             f[j_coord_offset+DIM*0+XX] -= tx;
 666             f[j_coord_offset+DIM*0+YY] -= ty;
 667             f[j_coord_offset+DIM*0+ZZ] -= tz;
 668
 669             /**************************
 670              * CALCULATE INTERACTIONS *
 671              **************************/
 672
 673             r30              = rsq30*rinv30;
 674
 675             qq30             = iq3*jq0;
 676
 677             /* EWALD ELECTROSTATICS */
 678
 679             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 680             ewrt             = r30*ewtabscale;
 681             ewitab           = ewrt;
 682             eweps            = ewrt-ewitab;
 683             felec            = (1.0-eweps)*ewtab[ewitab]+eweps*ewtab[ewitab+1];
 684             felec            = qq30*rinv30*(rinvsq30-felec);
 685
 686             fscal            = felec;
 687
 688             /* Calculate temporary vectorial force */
 689             tx               = fscal*dx30;
 690             ty               = fscal*dy30;
 691             tz               = fscal*dz30;
 692
 693             /* Update vectorial force */
 694             fix3            += tx;
 695             fiy3            += ty;
 696             fiz3            += 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             /* Inner loop uses 160 flops */
 702         }
 703         /* End of innermost loop */
 704
 705         tx = ty = tz = 0;
 706         f[i_coord_offset+DIM*0+XX] += fix0;
 707         f[i_coord_offset+DIM*0+YY] += fiy0;
 708         f[i_coord_offset+DIM*0+ZZ] += fiz0;
 709         tx                         += fix0;
 710         ty                         += fiy0;
 711         tz                         += fiz0;
 712         f[i_coord_offset+DIM*1+XX] += fix1;
 713         f[i_coord_offset+DIM*1+YY] += fiy1;
 714         f[i_coord_offset+DIM*1+ZZ] += fiz1;
 715         tx                         += fix1;
 716         ty                         += fiy1;
 717         tz                         += fiz1;
 718         f[i_coord_offset+DIM*2+XX] += fix2;
 719         f[i_coord_offset+DIM*2+YY] += fiy2;
 720         f[i_coord_offset+DIM*2+ZZ] += fiz2;
 721         tx                         += fix2;
 722         ty                         += fiy2;
 723         tz                         += fiz2;
 724         f[i_coord_offset+DIM*3+XX] += fix3;
 725         f[i_coord_offset+DIM*3+YY] += fiy3;
 726         f[i_coord_offset+DIM*3+ZZ] += fiz3;
 727         tx                         += fix3;
 728         ty                         += fiy3;
 729         tz                         += fiz3;
 730         fshift[i_shift_offset+XX]  += tx;
 731         fshift[i_shift_offset+YY]  += ty;
 732         fshift[i_shift_offset+ZZ]  += tz;
 733
 734         /* Increment number of inner iterations */
 735         inneriter                  += j_index_end - j_index_start;
 736
 737         /* Outer loop uses 39 flops */
 738     }
 739
 740     /* Increment number of outer iterations */
 741     outeriter        += nri;
 742
 743     /* Update outer/inner flops */
 744
 745     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W4_F,outeriter*39 + inneriter*160);
 746 }