src/gmxlib/nonbonded/nb_kernel_c/nb_kernel_ElecEwSh_VdwLJSh_GeomW4P1_c.c

   1 /*
   2  * Note: this file was generated by the Gromacs c kernel generator.
   3  *
   4  *                This source code is part of
   5  *
   6  *                 G   R   O   M   A   C   S
   7  *
   8  * Copyright (c) 2001-2012, The GROMACS Development Team
   9  *
  10  * Gromacs is a library for molecular simulation and trajectory analysis,
  11  * written by Erik Lindahl, David van der Spoel, Berk Hess, and others - for
  12  * a full list of developers and information, check out http://www.gromacs.org
  13  *
  14  * This program is free software; you can redistribute it and/or modify it under
  15  * the terms of the GNU Lesser General Public License as published by the Free
  16  * Software Foundation; either version 2 of the License, or (at your option) any
  17  * later version.
  18  *
  19  * To help fund GROMACS development, we humbly ask that you cite
  20  * the papers people have written on it - you can find them on the website.
  21  */
  22 #ifdef HAVE_CONFIG_H
  23 #include <config.h>
  24 #endif
  25
  26 #include <math.h>
  27
  28 #include "../nb_kernel.h"
  29 #include "types/simple.h"
  30 #include "vec.h"
  31 #include "nrnb.h"
  32
  33 /*
  34  * Gromacs nonbonded kernel:   nb_kernel_ElecEwSh_VdwLJSh_GeomW4P1_VF_c
  35  * Electrostatics interaction: Ewald
  36  * VdW interaction:            LennardJones
  37  * Geometry:                   Water4-Particle
  38  * Calculate force/pot:        PotentialAndForce
  39  */
  40 void
  41 nb_kernel_ElecEwSh_VdwLJSh_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      = 2*nvdwtype*vdwtype[inr+0];
 108
 109     /* When we use explicit cutoffs the value must be identical for elec and VdW, so use elec as an arbitrary choice */
 110     rcutoff          = fr->rcoulomb;
 111     rcutoff2         = rcutoff*rcutoff;
 112
 113     sh_vdw_invrcut6  = fr->ic->sh_invrc6;
 114     rvdw             = fr->rvdw;
 115
 116     outeriter        = 0;
 117     inneriter        = 0;
 118
 119     /* Start outer loop over neighborlists */
 120     for(iidx=0; iidx<nri; iidx++)
 121     {
 122         /* Load shift vector for this list */
 123         i_shift_offset   = DIM*shiftidx[iidx];
 124         shX              = shiftvec[i_shift_offset+XX];
 125         shY              = shiftvec[i_shift_offset+YY];
 126         shZ              = shiftvec[i_shift_offset+ZZ];
 127
 128         /* Load limits for loop over neighbors */
 129         j_index_start    = jindex[iidx];
 130         j_index_end      = jindex[iidx+1];
 131
 132         /* Get outer coordinate index */
 133         inr              = iinr[iidx];
 134         i_coord_offset   = DIM*inr;
 135
 136         /* Load i particle coords and add shift vector */
 137         ix0              = shX + x[i_coord_offset+DIM*0+XX];
 138         iy0              = shY + x[i_coord_offset+DIM*0+YY];
 139         iz0              = shZ + x[i_coord_offset+DIM*0+ZZ];
 140         ix1              = shX + x[i_coord_offset+DIM*1+XX];
 141         iy1              = shY + x[i_coord_offset+DIM*1+YY];
 142         iz1              = shZ + x[i_coord_offset+DIM*1+ZZ];
 143         ix2              = shX + x[i_coord_offset+DIM*2+XX];
 144         iy2              = shY + x[i_coord_offset+DIM*2+YY];
 145         iz2              = shZ + x[i_coord_offset+DIM*2+ZZ];
 146         ix3              = shX + x[i_coord_offset+DIM*3+XX];
 147         iy3              = shY + x[i_coord_offset+DIM*3+YY];
 148         iz3              = shZ + x[i_coord_offset+DIM*3+ZZ];
 149
 150         fix0             = 0.0;
 151         fiy0             = 0.0;
 152         fiz0             = 0.0;
 153         fix1             = 0.0;
 154         fiy1             = 0.0;
 155         fiz1             = 0.0;
 156         fix2             = 0.0;
 157         fiy2             = 0.0;
 158         fiz2             = 0.0;
 159         fix3             = 0.0;
 160         fiy3             = 0.0;
 161         fiz3             = 0.0;
 162
 163         /* Reset potential sums */
 164         velecsum         = 0.0;
 165         vvdwsum          = 0.0;
 166
 167         /* Start inner kernel loop */
 168         for(jidx=j_index_start; jidx<j_index_end; jidx++)
 169         {
 170             /* Get j neighbor index, and coordinate index */
 171             jnr              = jjnr[jidx];
 172             j_coord_offset   = DIM*jnr;
 173
 174             /* load j atom coordinates */
 175             jx0              = x[j_coord_offset+DIM*0+XX];
 176             jy0              = x[j_coord_offset+DIM*0+YY];
 177             jz0              = x[j_coord_offset+DIM*0+ZZ];
 178
 179             /* Calculate displacement vector */
 180             dx00             = ix0 - jx0;
 181             dy00             = iy0 - jy0;
 182             dz00             = iz0 - jz0;
 183             dx10             = ix1 - jx0;
 184             dy10             = iy1 - jy0;
 185             dz10             = iz1 - jz0;
 186             dx20             = ix2 - jx0;
 187             dy20             = iy2 - jy0;
 188             dz20             = iz2 - jz0;
 189             dx30             = ix3 - jx0;
 190             dy30             = iy3 - jy0;
 191             dz30             = iz3 - jz0;
 192
 193             /* Calculate squared distance and things based on it */
 194             rsq00            = dx00*dx00+dy00*dy00+dz00*dz00;
 195             rsq10            = dx10*dx10+dy10*dy10+dz10*dz10;
 196             rsq20            = dx20*dx20+dy20*dy20+dz20*dz20;
 197             rsq30            = dx30*dx30+dy30*dy30+dz30*dz30;
 198
 199             rinv10           = gmx_invsqrt(rsq10);
 200             rinv20           = gmx_invsqrt(rsq20);
 201             rinv30           = gmx_invsqrt(rsq30);
 202
 203             rinvsq00         = 1.0/rsq00;
 204             rinvsq10         = rinv10*rinv10;
 205             rinvsq20         = rinv20*rinv20;
 206             rinvsq30         = rinv30*rinv30;
 207
 208             /* Load parameters for j particles */
 209             jq0              = charge[jnr+0];
 210             vdwjidx0         = 2*vdwtype[jnr+0];
 211
 212             /**************************
 213              * CALCULATE INTERACTIONS *
 214              **************************/
 215
 216             if (rsq00<rcutoff2)
 217             {
 218
 219             c6_00            = vdwparam[vdwioffset0+vdwjidx0];
 220             c12_00           = vdwparam[vdwioffset0+vdwjidx0+1];
 221
 222             /* LENNARD-JONES DISPERSION/REPULSION */
 223
 224             rinvsix          = rinvsq00*rinvsq00*rinvsq00;
 225             vvdw6            = c6_00*rinvsix;
 226             vvdw12           = c12_00*rinvsix*rinvsix;
 227             vvdw             = (vvdw12 - c12_00*sh_vdw_invrcut6*sh_vdw_invrcut6)*(1.0/12.0) - (vvdw6 - c6_00*sh_vdw_invrcut6)*(1.0/6.0);
 228             fvdw             = (vvdw12-vvdw6)*rinvsq00;
 229
 230             /* Update potential sums from outer loop */
 231             vvdwsum         += vvdw;
 232
 233             fscal            = fvdw;
 234
 235             /* Calculate temporary vectorial force */
 236             tx               = fscal*dx00;
 237             ty               = fscal*dy00;
 238             tz               = fscal*dz00;
 239
 240             /* Update vectorial force */
 241             fix0            += tx;
 242             fiy0            += ty;
 243             fiz0            += tz;
 244             f[j_coord_offset+DIM*0+XX] -= tx;
 245             f[j_coord_offset+DIM*0+YY] -= ty;
 246             f[j_coord_offset+DIM*0+ZZ] -= tz;
 247
 248             }
 249
 250             /**************************
 251              * CALCULATE INTERACTIONS *
 252              **************************/
 253
 254             if (rsq10<rcutoff2)
 255             {
 256
 257             r10              = rsq10*rinv10;
 258
 259             qq10             = iq1*jq0;
 260
 261             /* EWALD ELECTROSTATICS */
 262
 263             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 264             ewrt             = r10*ewtabscale;
 265             ewitab           = ewrt;
 266             eweps            = ewrt-ewitab;
 267             ewitab           = 4*ewitab;
 268             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 269             velec            = qq10*((rinv10-sh_ewald)-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 270             felec            = qq10*rinv10*(rinvsq10-felec);
 271
 272             /* Update potential sums from outer loop */
 273             velecsum        += velec;
 274
 275             fscal            = felec;
 276
 277             /* Calculate temporary vectorial force */
 278             tx               = fscal*dx10;
 279             ty               = fscal*dy10;
 280             tz               = fscal*dz10;
 281
 282             /* Update vectorial force */
 283             fix1            += tx;
 284             fiy1            += ty;
 285             fiz1            += tz;
 286             f[j_coord_offset+DIM*0+XX] -= tx;
 287             f[j_coord_offset+DIM*0+YY] -= ty;
 288             f[j_coord_offset+DIM*0+ZZ] -= tz;
 289
 290             }
 291
 292             /**************************
 293              * CALCULATE INTERACTIONS *
 294              **************************/
 295
 296             if (rsq20<rcutoff2)
 297             {
 298
 299             r20              = rsq20*rinv20;
 300
 301             qq20             = iq2*jq0;
 302
 303             /* EWALD ELECTROSTATICS */
 304
 305             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 306             ewrt             = r20*ewtabscale;
 307             ewitab           = ewrt;
 308             eweps            = ewrt-ewitab;
 309             ewitab           = 4*ewitab;
 310             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 311             velec            = qq20*((rinv20-sh_ewald)-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 312             felec            = qq20*rinv20*(rinvsq20-felec);
 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             r30              = rsq30*rinv30;
 342
 343             qq30             = iq3*jq0;
 344
 345             /* EWALD ELECTROSTATICS */
 346
 347             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 348             ewrt             = r30*ewtabscale;
 349             ewitab           = ewrt;
 350             eweps            = ewrt-ewitab;
 351             ewitab           = 4*ewitab;
 352             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 353             velec            = qq30*((rinv30-sh_ewald)-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 354             felec            = qq30*rinv30*(rinvsq30-felec);
 355
 356             /* Update potential sums from outer loop */
 357             velecsum        += velec;
 358
 359             fscal            = felec;
 360
 361             /* Calculate temporary vectorial force */
 362             tx               = fscal*dx30;
 363             ty               = fscal*dy30;
 364             tz               = fscal*dz30;
 365
 366             /* Update vectorial force */
 367             fix3            += tx;
 368             fiy3            += ty;
 369             fiz3            += tz;
 370             f[j_coord_offset+DIM*0+XX] -= tx;
 371             f[j_coord_offset+DIM*0+YY] -= ty;
 372             f[j_coord_offset+DIM*0+ZZ] -= tz;
 373
 374             }
 375
 376             /* Inner loop uses 163 flops */
 377         }
 378         /* End of innermost loop */
 379
 380         tx = ty = tz = 0;
 381         f[i_coord_offset+DIM*0+XX] += fix0;
 382         f[i_coord_offset+DIM*0+YY] += fiy0;
 383         f[i_coord_offset+DIM*0+ZZ] += fiz0;
 384         tx                         += fix0;
 385         ty                         += fiy0;
 386         tz                         += fiz0;
 387         f[i_coord_offset+DIM*1+XX] += fix1;
 388         f[i_coord_offset+DIM*1+YY] += fiy1;
 389         f[i_coord_offset+DIM*1+ZZ] += fiz1;
 390         tx                         += fix1;
 391         ty                         += fiy1;
 392         tz                         += fiz1;
 393         f[i_coord_offset+DIM*2+XX] += fix2;
 394         f[i_coord_offset+DIM*2+YY] += fiy2;
 395         f[i_coord_offset+DIM*2+ZZ] += fiz2;
 396         tx                         += fix2;
 397         ty                         += fiy2;
 398         tz                         += fiz2;
 399         f[i_coord_offset+DIM*3+XX] += fix3;
 400         f[i_coord_offset+DIM*3+YY] += fiy3;
 401         f[i_coord_offset+DIM*3+ZZ] += fiz3;
 402         tx                         += fix3;
 403         ty                         += fiy3;
 404         tz                         += fiz3;
 405         fshift[i_shift_offset+XX]  += tx;
 406         fshift[i_shift_offset+YY]  += ty;
 407         fshift[i_shift_offset+ZZ]  += tz;
 408
 409         ggid                        = gid[iidx];
 410         /* Update potential energies */
 411         kernel_data->energygrp_elec[ggid] += velecsum;
 412         kernel_data->energygrp_vdw[ggid] += vvdwsum;
 413
 414         /* Increment number of inner iterations */
 415         inneriter                  += j_index_end - j_index_start;
 416
 417         /* Outer loop uses 41 flops */
 418     }
 419
 420     /* Increment number of outer iterations */
 421     outeriter        += nri;
 422
 423     /* Update outer/inner flops */
 424
 425     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W4_VF,outeriter*41 + inneriter*163);
 426 }
 427 /*
 428  * Gromacs nonbonded kernel:   nb_kernel_ElecEwSh_VdwLJSh_GeomW4P1_F_c
 429  * Electrostatics interaction: Ewald
 430  * VdW interaction:            LennardJones
 431  * Geometry:                   Water4-Particle
 432  * Calculate force/pot:        Force
 433  */
 434 void
 435 nb_kernel_ElecEwSh_VdwLJSh_GeomW4P1_F_c
 436                     (t_nblist * gmx_restrict                nlist,
 437                      rvec * gmx_restrict                    xx,
 438                      rvec * gmx_restrict                    ff,
 439                      t_forcerec * gmx_restrict              fr,
 440                      t_mdatoms * gmx_restrict               mdatoms,
 441                      nb_kernel_data_t * gmx_restrict        kernel_data,
 442                      t_nrnb * gmx_restrict                  nrnb)
 443 {
 444     int              i_shift_offset,i_coord_offset,j_coord_offset;
 445     int              j_index_start,j_index_end;
 446     int              nri,inr,ggid,iidx,jidx,jnr,outeriter,inneriter;
 447     real             shX,shY,shZ,tx,ty,tz,fscal,rcutoff,rcutoff2;
 448     int              *iinr,*jindex,*jjnr,*shiftidx,*gid;
 449     real             *shiftvec,*fshift,*x,*f;
 450     int              vdwioffset0;
 451     real             ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
 452     int              vdwioffset1;
 453     real             ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
 454     int              vdwioffset2;
 455     real             ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
 456     int              vdwioffset3;
 457     real             ix3,iy3,iz3,fix3,fiy3,fiz3,iq3,isai3;
 458     int              vdwjidx0;
 459     real             jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
 460     real             dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00,cexp1_00,cexp2_00;
 461     real             dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10,cexp1_10,cexp2_10;
 462     real             dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20,cexp1_20,cexp2_20;
 463     real             dx30,dy30,dz30,rsq30,rinv30,rinvsq30,r30,qq30,c6_30,c12_30,cexp1_30,cexp2_30;
 464     real             velec,felec,velecsum,facel,crf,krf,krf2;
 465     real             *charge;
 466     int              nvdwtype;
 467     real             rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,br,vvdwexp,sh_vdw_invrcut6;
 468     int              *vdwtype;
 469     real             *vdwparam;
 470     int              ewitab;
 471     real             ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace;
 472     real             *ewtab;
 473
 474     x                = xx[0];
 475     f                = ff[0];
 476
 477     nri              = nlist->nri;
 478     iinr             = nlist->iinr;
 479     jindex           = nlist->jindex;
 480     jjnr             = nlist->jjnr;
 481     shiftidx         = nlist->shift;
 482     gid              = nlist->gid;
 483     shiftvec         = fr->shift_vec[0];
 484     fshift           = fr->fshift[0];
 485     facel            = fr->epsfac;
 486     charge           = mdatoms->chargeA;
 487     nvdwtype         = fr->ntype;
 488     vdwparam         = fr->nbfp;
 489     vdwtype          = mdatoms->typeA;
 490
 491     sh_ewald         = fr->ic->sh_ewald;
 492     ewtab            = fr->ic->tabq_coul_F;
 493     ewtabscale       = fr->ic->tabq_scale;
 494     ewtabhalfspace   = 0.5/ewtabscale;
 495
 496     /* Setup water-specific parameters */
 497     inr              = nlist->iinr[0];
 498     iq1              = facel*charge[inr+1];
 499     iq2              = facel*charge[inr+2];
 500     iq3              = facel*charge[inr+3];
 501     vdwioffset0      = 2*nvdwtype*vdwtype[inr+0];
 502
 503     /* When we use explicit cutoffs the value must be identical for elec and VdW, so use elec as an arbitrary choice */
 504     rcutoff          = fr->rcoulomb;
 505     rcutoff2         = rcutoff*rcutoff;
 506
 507     sh_vdw_invrcut6  = fr->ic->sh_invrc6;
 508     rvdw             = fr->rvdw;
 509
 510     outeriter        = 0;
 511     inneriter        = 0;
 512
 513     /* Start outer loop over neighborlists */
 514     for(iidx=0; iidx<nri; iidx++)
 515     {
 516         /* Load shift vector for this list */
 517         i_shift_offset   = DIM*shiftidx[iidx];
 518         shX              = shiftvec[i_shift_offset+XX];
 519         shY              = shiftvec[i_shift_offset+YY];
 520         shZ              = shiftvec[i_shift_offset+ZZ];
 521
 522         /* Load limits for loop over neighbors */
 523         j_index_start    = jindex[iidx];
 524         j_index_end      = jindex[iidx+1];
 525
 526         /* Get outer coordinate index */
 527         inr              = iinr[iidx];
 528         i_coord_offset   = DIM*inr;
 529
 530         /* Load i particle coords and add shift vector */
 531         ix0              = shX + x[i_coord_offset+DIM*0+XX];
 532         iy0              = shY + x[i_coord_offset+DIM*0+YY];
 533         iz0              = shZ + x[i_coord_offset+DIM*0+ZZ];
 534         ix1              = shX + x[i_coord_offset+DIM*1+XX];
 535         iy1              = shY + x[i_coord_offset+DIM*1+YY];
 536         iz1              = shZ + x[i_coord_offset+DIM*1+ZZ];
 537         ix2              = shX + x[i_coord_offset+DIM*2+XX];
 538         iy2              = shY + x[i_coord_offset+DIM*2+YY];
 539         iz2              = shZ + x[i_coord_offset+DIM*2+ZZ];
 540         ix3              = shX + x[i_coord_offset+DIM*3+XX];
 541         iy3              = shY + x[i_coord_offset+DIM*3+YY];
 542         iz3              = shZ + x[i_coord_offset+DIM*3+ZZ];
 543
 544         fix0             = 0.0;
 545         fiy0             = 0.0;
 546         fiz0             = 0.0;
 547         fix1             = 0.0;
 548         fiy1             = 0.0;
 549         fiz1             = 0.0;
 550         fix2             = 0.0;
 551         fiy2             = 0.0;
 552         fiz2             = 0.0;
 553         fix3             = 0.0;
 554         fiy3             = 0.0;
 555         fiz3             = 0.0;
 556
 557         /* Start inner kernel loop */
 558         for(jidx=j_index_start; jidx<j_index_end; jidx++)
 559         {
 560             /* Get j neighbor index, and coordinate index */
 561             jnr              = jjnr[jidx];
 562             j_coord_offset   = DIM*jnr;
 563
 564             /* load j atom coordinates */
 565             jx0              = x[j_coord_offset+DIM*0+XX];
 566             jy0              = x[j_coord_offset+DIM*0+YY];
 567             jz0              = x[j_coord_offset+DIM*0+ZZ];
 568
 569             /* Calculate displacement vector */
 570             dx00             = ix0 - jx0;
 571             dy00             = iy0 - jy0;
 572             dz00             = iz0 - jz0;
 573             dx10             = ix1 - jx0;
 574             dy10             = iy1 - jy0;
 575             dz10             = iz1 - jz0;
 576             dx20             = ix2 - jx0;
 577             dy20             = iy2 - jy0;
 578             dz20             = iz2 - jz0;
 579             dx30             = ix3 - jx0;
 580             dy30             = iy3 - jy0;
 581             dz30             = iz3 - jz0;
 582
 583             /* Calculate squared distance and things based on it */
 584             rsq00            = dx00*dx00+dy00*dy00+dz00*dz00;
 585             rsq10            = dx10*dx10+dy10*dy10+dz10*dz10;
 586             rsq20            = dx20*dx20+dy20*dy20+dz20*dz20;
 587             rsq30            = dx30*dx30+dy30*dy30+dz30*dz30;
 588
 589             rinv10           = gmx_invsqrt(rsq10);
 590             rinv20           = gmx_invsqrt(rsq20);
 591             rinv30           = gmx_invsqrt(rsq30);
 592
 593             rinvsq00         = 1.0/rsq00;
 594             rinvsq10         = rinv10*rinv10;
 595             rinvsq20         = rinv20*rinv20;
 596             rinvsq30         = rinv30*rinv30;
 597
 598             /* Load parameters for j particles */
 599             jq0              = charge[jnr+0];
 600             vdwjidx0         = 2*vdwtype[jnr+0];
 601
 602             /**************************
 603              * CALCULATE INTERACTIONS *
 604              **************************/
 605
 606             if (rsq00<rcutoff2)
 607             {
 608
 609             c6_00            = vdwparam[vdwioffset0+vdwjidx0];
 610             c12_00           = vdwparam[vdwioffset0+vdwjidx0+1];
 611
 612             /* LENNARD-JONES DISPERSION/REPULSION */
 613
 614             rinvsix          = rinvsq00*rinvsq00*rinvsq00;
 615             fvdw             = (c12_00*rinvsix-c6_00)*rinvsix*rinvsq00;
 616
 617             fscal            = fvdw;
 618
 619             /* Calculate temporary vectorial force */
 620             tx               = fscal*dx00;
 621             ty               = fscal*dy00;
 622             tz               = fscal*dz00;
 623
 624             /* Update vectorial force */
 625             fix0            += tx;
 626             fiy0            += ty;
 627             fiz0            += tz;
 628             f[j_coord_offset+DIM*0+XX] -= tx;
 629             f[j_coord_offset+DIM*0+YY] -= ty;
 630             f[j_coord_offset+DIM*0+ZZ] -= tz;
 631
 632             }
 633
 634             /**************************
 635              * CALCULATE INTERACTIONS *
 636              **************************/
 637
 638             if (rsq10<rcutoff2)
 639             {
 640
 641             r10              = rsq10*rinv10;
 642
 643             qq10             = iq1*jq0;
 644
 645             /* EWALD ELECTROSTATICS */
 646
 647             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 648             ewrt             = r10*ewtabscale;
 649             ewitab           = ewrt;
 650             eweps            = ewrt-ewitab;
 651             felec            = (1.0-eweps)*ewtab[ewitab]+eweps*ewtab[ewitab+1];
 652             felec            = qq10*rinv10*(rinvsq10-felec);
 653
 654             fscal            = felec;
 655
 656             /* Calculate temporary vectorial force */
 657             tx               = fscal*dx10;
 658             ty               = fscal*dy10;
 659             tz               = fscal*dz10;
 660
 661             /* Update vectorial force */
 662             fix1            += tx;
 663             fiy1            += ty;
 664             fiz1            += 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
 671             /**************************
 672              * CALCULATE INTERACTIONS *
 673              **************************/
 674
 675             if (rsq20<rcutoff2)
 676             {
 677
 678             r20              = rsq20*rinv20;
 679
 680             qq20             = iq2*jq0;
 681
 682             /* EWALD ELECTROSTATICS */
 683
 684             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 685             ewrt             = r20*ewtabscale;
 686             ewitab           = ewrt;
 687             eweps            = ewrt-ewitab;
 688             felec            = (1.0-eweps)*ewtab[ewitab]+eweps*ewtab[ewitab+1];
 689             felec            = qq20*rinv20*(rinvsq20-felec);
 690
 691             fscal            = felec;
 692
 693             /* Calculate temporary vectorial force */
 694             tx               = fscal*dx20;
 695             ty               = fscal*dy20;
 696             tz               = fscal*dz20;
 697
 698             /* Update vectorial force */
 699             fix2            += tx;
 700             fiy2            += ty;
 701             fiz2            += tz;
 702             f[j_coord_offset+DIM*0+XX] -= tx;
 703             f[j_coord_offset+DIM*0+YY] -= ty;
 704             f[j_coord_offset+DIM*0+ZZ] -= tz;
 705
 706             }
 707
 708             /**************************
 709              * CALCULATE INTERACTIONS *
 710              **************************/
 711
 712             if (rsq30<rcutoff2)
 713             {
 714
 715             r30              = rsq30*rinv30;
 716
 717             qq30             = iq3*jq0;
 718
 719             /* EWALD ELECTROSTATICS */
 720
 721             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 722             ewrt             = r30*ewtabscale;
 723             ewitab           = ewrt;
 724             eweps            = ewrt-ewitab;
 725             felec            = (1.0-eweps)*ewtab[ewitab]+eweps*ewtab[ewitab+1];
 726             felec            = qq30*rinv30*(rinvsq30-felec);
 727
 728             fscal            = felec;
 729
 730             /* Calculate temporary vectorial force */
 731             tx               = fscal*dx30;
 732             ty               = fscal*dy30;
 733             tz               = fscal*dz30;
 734
 735             /* Update vectorial force */
 736             fix3            += tx;
 737             fiy3            += ty;
 738             fiz3            += tz;
 739             f[j_coord_offset+DIM*0+XX] -= tx;
 740             f[j_coord_offset+DIM*0+YY] -= ty;
 741             f[j_coord_offset+DIM*0+ZZ] -= tz;
 742
 743             }
 744
 745             /* Inner loop uses 129 flops */
 746         }
 747         /* End of innermost loop */
 748
 749         tx = ty = tz = 0;
 750         f[i_coord_offset+DIM*0+XX] += fix0;
 751         f[i_coord_offset+DIM*0+YY] += fiy0;
 752         f[i_coord_offset+DIM*0+ZZ] += fiz0;
 753         tx                         += fix0;
 754         ty                         += fiy0;
 755         tz                         += fiz0;
 756         f[i_coord_offset+DIM*1+XX] += fix1;
 757         f[i_coord_offset+DIM*1+YY] += fiy1;
 758         f[i_coord_offset+DIM*1+ZZ] += fiz1;
 759         tx                         += fix1;
 760         ty                         += fiy1;
 761         tz                         += fiz1;
 762         f[i_coord_offset+DIM*2+XX] += fix2;
 763         f[i_coord_offset+DIM*2+YY] += fiy2;
 764         f[i_coord_offset+DIM*2+ZZ] += fiz2;
 765         tx                         += fix2;
 766         ty                         += fiy2;
 767         tz                         += fiz2;
 768         f[i_coord_offset+DIM*3+XX] += fix3;
 769         f[i_coord_offset+DIM*3+YY] += fiy3;
 770         f[i_coord_offset+DIM*3+ZZ] += fiz3;
 771         tx                         += fix3;
 772         ty                         += fiy3;
 773         tz                         += fiz3;
 774         fshift[i_shift_offset+XX]  += tx;
 775         fshift[i_shift_offset+YY]  += ty;
 776         fshift[i_shift_offset+ZZ]  += tz;
 777
 778         /* Increment number of inner iterations */
 779         inneriter                  += j_index_end - j_index_start;
 780
 781         /* Outer loop uses 39 flops */
 782     }
 783
 784     /* Increment number of outer iterations */
 785     outeriter        += nri;
 786
 787     /* Update outer/inner flops */
 788
 789     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W4_F,outeriter*39 + inneriter*129);
 790 }