src/gromacs/gmxlib/nonbonded/nb_kernel_c/nb_kernel_ElecEwSw_VdwLJSw_GeomW4P1_c.c

   1 /*
   2  * This file is part of the GROMACS molecular simulation package.
   3  *
   4  * Copyright (c) 2012,2013,2014, by the GROMACS development team, led by
   5  * Mark Abraham, David van der Spoel, Berk Hess, and Erik Lindahl,
   6  * and including many others, as listed in the AUTHORS file in the
   7  * top-level source directory and at http://www.gromacs.org.
   8  *
   9  * GROMACS is free software; you can redistribute it and/or
  10  * modify it under the terms of the GNU Lesser General Public License
  11  * as published by the Free Software Foundation; either version 2.1
  12  * of the License, or (at your option) any later version.
  13  *
  14  * GROMACS is distributed in the hope that it will be useful,
  15  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  16  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  17  * Lesser General Public License for more details.
  18  *
  19  * You should have received a copy of the GNU Lesser General Public
  20  * License along with GROMACS; if not, see
  21  * http://www.gnu.org/licenses, or write to the Free Software Foundation,
  22  * Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02110-1301  USA.
  23  *
  24  * If you want to redistribute modifications to GROMACS, please
  25  * consider that scientific software is very special. Version
  26  * control is crucial - bugs must be traceable. We will be happy to
  27  * consider code for inclusion in the official distribution, but
  28  * derived work must not be called official GROMACS. Details are found
  29  * in the README & COPYING files - if they are missing, get the
  30  * official version at http://www.gromacs.org.
  31  *
  32  * To help us fund GROMACS development, we humbly ask that you cite
  33  * the research papers on the package. Check out http://www.gromacs.org.
  34  */
  35 /*
  36  * Note: this file was generated by the GROMACS c kernel generator.
  37  */
  38 #include "gmxpre.h"
  39
  40 #include "config.h"
  41
  42 #include <math.h>
  43
  44 #include "../nb_kernel.h"
  45 #include "gromacs/legacyheaders/types/simple.h"
  46 #include "gromacs/math/vec.h"
  47 #include "gromacs/legacyheaders/nrnb.h"
  48
  49 /*
  50  * Gromacs nonbonded kernel:   nb_kernel_ElecEwSw_VdwLJSw_GeomW4P1_VF_c
  51  * Electrostatics interaction: Ewald
  52  * VdW interaction:            LennardJones
  53  * Geometry:                   Water4-Particle
  54  * Calculate force/pot:        PotentialAndForce
  55  */
  56 void
  57 nb_kernel_ElecEwSw_VdwLJSw_GeomW4P1_VF_c
  58                     (t_nblist                    * gmx_restrict       nlist,
  59                      rvec                        * gmx_restrict          xx,
  60                      rvec                        * gmx_restrict          ff,
  61                      t_forcerec                  * gmx_restrict          fr,
  62                      t_mdatoms                   * gmx_restrict     mdatoms,
  63                      nb_kernel_data_t gmx_unused * gmx_restrict kernel_data,
  64                      t_nrnb                      * gmx_restrict        nrnb)
  65 {
  66     int              i_shift_offset,i_coord_offset,j_coord_offset;
  67     int              j_index_start,j_index_end;
  68     int              nri,inr,ggid,iidx,jidx,jnr,outeriter,inneriter;
  69     real             shX,shY,shZ,tx,ty,tz,fscal,rcutoff,rcutoff2;
  70     int              *iinr,*jindex,*jjnr,*shiftidx,*gid;
  71     real             *shiftvec,*fshift,*x,*f;
  72     int              vdwioffset0;
  73     real             ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
  74     int              vdwioffset1;
  75     real             ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
  76     int              vdwioffset2;
  77     real             ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
  78     int              vdwioffset3;
  79     real             ix3,iy3,iz3,fix3,fiy3,fiz3,iq3,isai3;
  80     int              vdwjidx0;
  81     real             jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
  82     real             dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00,cexp1_00,cexp2_00;
  83     real             dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10,cexp1_10,cexp2_10;
  84     real             dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20,cexp1_20,cexp2_20;
  85     real             dx30,dy30,dz30,rsq30,rinv30,rinvsq30,r30,qq30,c6_30,c12_30,cexp1_30,cexp2_30;
  86     real             velec,felec,velecsum,facel,crf,krf,krf2;
  87     real             *charge;
  88     int              nvdwtype;
  89     real             rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,br,vvdwexp,sh_vdw_invrcut6;
  90     int              *vdwtype;
  91     real             *vdwparam;
  92     int              ewitab;
  93     real             ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace;
  94     real             *ewtab;
  95     real             rswitch,swV3,swV4,swV5,swF2,swF3,swF4,d,d2,sw,dsw;
  96
  97     x                = xx[0];
  98     f                = ff[0];
  99
 100     nri              = nlist->nri;
 101     iinr             = nlist->iinr;
 102     jindex           = nlist->jindex;
 103     jjnr             = nlist->jjnr;
 104     shiftidx         = nlist->shift;
 105     gid              = nlist->gid;
 106     shiftvec         = fr->shift_vec[0];
 107     fshift           = fr->fshift[0];
 108     facel            = fr->epsfac;
 109     charge           = mdatoms->chargeA;
 110     nvdwtype         = fr->ntype;
 111     vdwparam         = fr->nbfp;
 112     vdwtype          = mdatoms->typeA;
 113
 114     sh_ewald         = fr->ic->sh_ewald;
 115     ewtab            = fr->ic->tabq_coul_FDV0;
 116     ewtabscale       = fr->ic->tabq_scale;
 117     ewtabhalfspace   = 0.5/ewtabscale;
 118
 119     /* Setup water-specific parameters */
 120     inr              = nlist->iinr[0];
 121     iq1              = facel*charge[inr+1];
 122     iq2              = facel*charge[inr+2];
 123     iq3              = facel*charge[inr+3];
 124     vdwioffset0      = 2*nvdwtype*vdwtype[inr+0];
 125
 126     /* When we use explicit cutoffs the value must be identical for elec and VdW, so use elec as an arbitrary choice */
 127     rcutoff          = fr->rcoulomb;
 128     rcutoff2         = rcutoff*rcutoff;
 129
 130     rswitch          = fr->rcoulomb_switch;
 131     /* Setup switch parameters */
 132     d                = rcutoff-rswitch;
 133     swV3             = -10.0/(d*d*d);
 134     swV4             =  15.0/(d*d*d*d);
 135     swV5             =  -6.0/(d*d*d*d*d);
 136     swF2             = -30.0/(d*d*d);
 137     swF3             =  60.0/(d*d*d*d);
 138     swF4             = -30.0/(d*d*d*d*d);
 139
 140     outeriter        = 0;
 141     inneriter        = 0;
 142
 143     /* Start outer loop over neighborlists */
 144     for(iidx=0; iidx<nri; iidx++)
 145     {
 146         /* Load shift vector for this list */
 147         i_shift_offset   = DIM*shiftidx[iidx];
 148         shX              = shiftvec[i_shift_offset+XX];
 149         shY              = shiftvec[i_shift_offset+YY];
 150         shZ              = shiftvec[i_shift_offset+ZZ];
 151
 152         /* Load limits for loop over neighbors */
 153         j_index_start    = jindex[iidx];
 154         j_index_end      = jindex[iidx+1];
 155
 156         /* Get outer coordinate index */
 157         inr              = iinr[iidx];
 158         i_coord_offset   = DIM*inr;
 159
 160         /* Load i particle coords and add shift vector */
 161         ix0              = shX + x[i_coord_offset+DIM*0+XX];
 162         iy0              = shY + x[i_coord_offset+DIM*0+YY];
 163         iz0              = shZ + x[i_coord_offset+DIM*0+ZZ];
 164         ix1              = shX + x[i_coord_offset+DIM*1+XX];
 165         iy1              = shY + x[i_coord_offset+DIM*1+YY];
 166         iz1              = shZ + x[i_coord_offset+DIM*1+ZZ];
 167         ix2              = shX + x[i_coord_offset+DIM*2+XX];
 168         iy2              = shY + x[i_coord_offset+DIM*2+YY];
 169         iz2              = shZ + x[i_coord_offset+DIM*2+ZZ];
 170         ix3              = shX + x[i_coord_offset+DIM*3+XX];
 171         iy3              = shY + x[i_coord_offset+DIM*3+YY];
 172         iz3              = shZ + x[i_coord_offset+DIM*3+ZZ];
 173
 174         fix0             = 0.0;
 175         fiy0             = 0.0;
 176         fiz0             = 0.0;
 177         fix1             = 0.0;
 178         fiy1             = 0.0;
 179         fiz1             = 0.0;
 180         fix2             = 0.0;
 181         fiy2             = 0.0;
 182         fiz2             = 0.0;
 183         fix3             = 0.0;
 184         fiy3             = 0.0;
 185         fiz3             = 0.0;
 186
 187         /* Reset potential sums */
 188         velecsum         = 0.0;
 189         vvdwsum          = 0.0;
 190
 191         /* Start inner kernel loop */
 192         for(jidx=j_index_start; jidx<j_index_end; jidx++)
 193         {
 194             /* Get j neighbor index, and coordinate index */
 195             jnr              = jjnr[jidx];
 196             j_coord_offset   = DIM*jnr;
 197
 198             /* load j atom coordinates */
 199             jx0              = x[j_coord_offset+DIM*0+XX];
 200             jy0              = x[j_coord_offset+DIM*0+YY];
 201             jz0              = x[j_coord_offset+DIM*0+ZZ];
 202
 203             /* Calculate displacement vector */
 204             dx00             = ix0 - jx0;
 205             dy00             = iy0 - jy0;
 206             dz00             = iz0 - jz0;
 207             dx10             = ix1 - jx0;
 208             dy10             = iy1 - jy0;
 209             dz10             = iz1 - jz0;
 210             dx20             = ix2 - jx0;
 211             dy20             = iy2 - jy0;
 212             dz20             = iz2 - jz0;
 213             dx30             = ix3 - jx0;
 214             dy30             = iy3 - jy0;
 215             dz30             = iz3 - jz0;
 216
 217             /* Calculate squared distance and things based on it */
 218             rsq00            = dx00*dx00+dy00*dy00+dz00*dz00;
 219             rsq10            = dx10*dx10+dy10*dy10+dz10*dz10;
 220             rsq20            = dx20*dx20+dy20*dy20+dz20*dz20;
 221             rsq30            = dx30*dx30+dy30*dy30+dz30*dz30;
 222
 223             rinv00           = gmx_invsqrt(rsq00);
 224             rinv10           = gmx_invsqrt(rsq10);
 225             rinv20           = gmx_invsqrt(rsq20);
 226             rinv30           = gmx_invsqrt(rsq30);
 227
 228             rinvsq00         = rinv00*rinv00;
 229             rinvsq10         = rinv10*rinv10;
 230             rinvsq20         = rinv20*rinv20;
 231             rinvsq30         = rinv30*rinv30;
 232
 233             /* Load parameters for j particles */
 234             jq0              = charge[jnr+0];
 235             vdwjidx0         = 2*vdwtype[jnr+0];
 236
 237             /**************************
 238              * CALCULATE INTERACTIONS *
 239              **************************/
 240
 241             if (rsq00<rcutoff2)
 242             {
 243
 244             r00              = rsq00*rinv00;
 245
 246             c6_00            = vdwparam[vdwioffset0+vdwjidx0];
 247             c12_00           = vdwparam[vdwioffset0+vdwjidx0+1];
 248
 249             /* LENNARD-JONES DISPERSION/REPULSION */
 250
 251             rinvsix          = rinvsq00*rinvsq00*rinvsq00;
 252             vvdw6            = c6_00*rinvsix;
 253             vvdw12           = c12_00*rinvsix*rinvsix;
 254             vvdw             = vvdw12*(1.0/12.0) - vvdw6*(1.0/6.0);
 255             fvdw             = (vvdw12-vvdw6)*rinvsq00;
 256
 257             d                = r00-rswitch;
 258             d                = (d>0.0) ? d : 0.0;
 259             d2               = d*d;
 260             sw               = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
 261
 262             dsw              = d2*(swF2+d*(swF3+d*swF4));
 263
 264             /* Evaluate switch function */
 265             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 266             fvdw             = fvdw*sw - rinv00*vvdw*dsw;
 267             vvdw            *= sw;
 268
 269             /* Update potential sums from outer loop */
 270             vvdwsum         += vvdw;
 271
 272             fscal            = fvdw;
 273
 274             /* Calculate temporary vectorial force */
 275             tx               = fscal*dx00;
 276             ty               = fscal*dy00;
 277             tz               = fscal*dz00;
 278
 279             /* Update vectorial force */
 280             fix0            += tx;
 281             fiy0            += ty;
 282             fiz0            += tz;
 283             f[j_coord_offset+DIM*0+XX] -= tx;
 284             f[j_coord_offset+DIM*0+YY] -= ty;
 285             f[j_coord_offset+DIM*0+ZZ] -= tz;
 286
 287             }
 288
 289             /**************************
 290              * CALCULATE INTERACTIONS *
 291              **************************/
 292
 293             if (rsq10<rcutoff2)
 294             {
 295
 296             r10              = rsq10*rinv10;
 297
 298             qq10             = iq1*jq0;
 299
 300             /* EWALD ELECTROSTATICS */
 301
 302             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 303             ewrt             = r10*ewtabscale;
 304             ewitab           = ewrt;
 305             eweps            = ewrt-ewitab;
 306             ewitab           = 4*ewitab;
 307             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 308             velec            = qq10*(rinv10-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 309             felec            = qq10*rinv10*(rinvsq10-felec);
 310
 311             d                = r10-rswitch;
 312             d                = (d>0.0) ? d : 0.0;
 313             d2               = d*d;
 314             sw               = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
 315
 316             dsw              = d2*(swF2+d*(swF3+d*swF4));
 317
 318             /* Evaluate switch function */
 319             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 320             felec            = felec*sw - rinv10*velec*dsw;
 321             velec           *= sw;
 322
 323             /* Update potential sums from outer loop */
 324             velecsum        += velec;
 325
 326             fscal            = felec;
 327
 328             /* Calculate temporary vectorial force */
 329             tx               = fscal*dx10;
 330             ty               = fscal*dy10;
 331             tz               = fscal*dz10;
 332
 333             /* Update vectorial force */
 334             fix1            += tx;
 335             fiy1            += ty;
 336             fiz1            += tz;
 337             f[j_coord_offset+DIM*0+XX] -= tx;
 338             f[j_coord_offset+DIM*0+YY] -= ty;
 339             f[j_coord_offset+DIM*0+ZZ] -= tz;
 340
 341             }
 342
 343             /**************************
 344              * CALCULATE INTERACTIONS *
 345              **************************/
 346
 347             if (rsq20<rcutoff2)
 348             {
 349
 350             r20              = rsq20*rinv20;
 351
 352             qq20             = iq2*jq0;
 353
 354             /* EWALD ELECTROSTATICS */
 355
 356             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 357             ewrt             = r20*ewtabscale;
 358             ewitab           = ewrt;
 359             eweps            = ewrt-ewitab;
 360             ewitab           = 4*ewitab;
 361             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 362             velec            = qq20*(rinv20-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 363             felec            = qq20*rinv20*(rinvsq20-felec);
 364
 365             d                = r20-rswitch;
 366             d                = (d>0.0) ? d : 0.0;
 367             d2               = d*d;
 368             sw               = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
 369
 370             dsw              = d2*(swF2+d*(swF3+d*swF4));
 371
 372             /* Evaluate switch function */
 373             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 374             felec            = felec*sw - rinv20*velec*dsw;
 375             velec           *= sw;
 376
 377             /* Update potential sums from outer loop */
 378             velecsum        += velec;
 379
 380             fscal            = felec;
 381
 382             /* Calculate temporary vectorial force */
 383             tx               = fscal*dx20;
 384             ty               = fscal*dy20;
 385             tz               = fscal*dz20;
 386
 387             /* Update vectorial force */
 388             fix2            += tx;
 389             fiy2            += ty;
 390             fiz2            += tz;
 391             f[j_coord_offset+DIM*0+XX] -= tx;
 392             f[j_coord_offset+DIM*0+YY] -= ty;
 393             f[j_coord_offset+DIM*0+ZZ] -= tz;
 394
 395             }
 396
 397             /**************************
 398              * CALCULATE INTERACTIONS *
 399              **************************/
 400
 401             if (rsq30<rcutoff2)
 402             {
 403
 404             r30              = rsq30*rinv30;
 405
 406             qq30             = iq3*jq0;
 407
 408             /* EWALD ELECTROSTATICS */
 409
 410             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 411             ewrt             = r30*ewtabscale;
 412             ewitab           = ewrt;
 413             eweps            = ewrt-ewitab;
 414             ewitab           = 4*ewitab;
 415             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 416             velec            = qq30*(rinv30-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 417             felec            = qq30*rinv30*(rinvsq30-felec);
 418
 419             d                = r30-rswitch;
 420             d                = (d>0.0) ? d : 0.0;
 421             d2               = d*d;
 422             sw               = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
 423
 424             dsw              = d2*(swF2+d*(swF3+d*swF4));
 425
 426             /* Evaluate switch function */
 427             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 428             felec            = felec*sw - rinv30*velec*dsw;
 429             velec           *= sw;
 430
 431             /* Update potential sums from outer loop */
 432             velecsum        += velec;
 433
 434             fscal            = felec;
 435
 436             /* Calculate temporary vectorial force */
 437             tx               = fscal*dx30;
 438             ty               = fscal*dy30;
 439             tz               = fscal*dz30;
 440
 441             /* Update vectorial force */
 442             fix3            += tx;
 443             fiy3            += ty;
 444             fiz3            += tz;
 445             f[j_coord_offset+DIM*0+XX] -= tx;
 446             f[j_coord_offset+DIM*0+YY] -= ty;
 447             f[j_coord_offset+DIM*0+ZZ] -= tz;
 448
 449             }
 450
 451             /* Inner loop uses 230 flops */
 452         }
 453         /* End of innermost loop */
 454
 455         tx = ty = tz = 0;
 456         f[i_coord_offset+DIM*0+XX] += fix0;
 457         f[i_coord_offset+DIM*0+YY] += fiy0;
 458         f[i_coord_offset+DIM*0+ZZ] += fiz0;
 459         tx                         += fix0;
 460         ty                         += fiy0;
 461         tz                         += fiz0;
 462         f[i_coord_offset+DIM*1+XX] += fix1;
 463         f[i_coord_offset+DIM*1+YY] += fiy1;
 464         f[i_coord_offset+DIM*1+ZZ] += fiz1;
 465         tx                         += fix1;
 466         ty                         += fiy1;
 467         tz                         += fiz1;
 468         f[i_coord_offset+DIM*2+XX] += fix2;
 469         f[i_coord_offset+DIM*2+YY] += fiy2;
 470         f[i_coord_offset+DIM*2+ZZ] += fiz2;
 471         tx                         += fix2;
 472         ty                         += fiy2;
 473         tz                         += fiz2;
 474         f[i_coord_offset+DIM*3+XX] += fix3;
 475         f[i_coord_offset+DIM*3+YY] += fiy3;
 476         f[i_coord_offset+DIM*3+ZZ] += fiz3;
 477         tx                         += fix3;
 478         ty                         += fiy3;
 479         tz                         += fiz3;
 480         fshift[i_shift_offset+XX]  += tx;
 481         fshift[i_shift_offset+YY]  += ty;
 482         fshift[i_shift_offset+ZZ]  += tz;
 483
 484         ggid                        = gid[iidx];
 485         /* Update potential energies */
 486         kernel_data->energygrp_elec[ggid] += velecsum;
 487         kernel_data->energygrp_vdw[ggid] += vvdwsum;
 488
 489         /* Increment number of inner iterations */
 490         inneriter                  += j_index_end - j_index_start;
 491
 492         /* Outer loop uses 41 flops */
 493     }
 494
 495     /* Increment number of outer iterations */
 496     outeriter        += nri;
 497
 498     /* Update outer/inner flops */
 499
 500     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W4_VF,outeriter*41 + inneriter*230);
 501 }
 502 /*
 503  * Gromacs nonbonded kernel:   nb_kernel_ElecEwSw_VdwLJSw_GeomW4P1_F_c
 504  * Electrostatics interaction: Ewald
 505  * VdW interaction:            LennardJones
 506  * Geometry:                   Water4-Particle
 507  * Calculate force/pot:        Force
 508  */
 509 void
 510 nb_kernel_ElecEwSw_VdwLJSw_GeomW4P1_F_c
 511                     (t_nblist                    * gmx_restrict       nlist,
 512                      rvec                        * gmx_restrict          xx,
 513                      rvec                        * gmx_restrict          ff,
 514                      t_forcerec                  * gmx_restrict          fr,
 515                      t_mdatoms                   * gmx_restrict     mdatoms,
 516                      nb_kernel_data_t gmx_unused * gmx_restrict kernel_data,
 517                      t_nrnb                      * gmx_restrict        nrnb)
 518 {
 519     int              i_shift_offset,i_coord_offset,j_coord_offset;
 520     int              j_index_start,j_index_end;
 521     int              nri,inr,ggid,iidx,jidx,jnr,outeriter,inneriter;
 522     real             shX,shY,shZ,tx,ty,tz,fscal,rcutoff,rcutoff2;
 523     int              *iinr,*jindex,*jjnr,*shiftidx,*gid;
 524     real             *shiftvec,*fshift,*x,*f;
 525     int              vdwioffset0;
 526     real             ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
 527     int              vdwioffset1;
 528     real             ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
 529     int              vdwioffset2;
 530     real             ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
 531     int              vdwioffset3;
 532     real             ix3,iy3,iz3,fix3,fiy3,fiz3,iq3,isai3;
 533     int              vdwjidx0;
 534     real             jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
 535     real             dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00,cexp1_00,cexp2_00;
 536     real             dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10,cexp1_10,cexp2_10;
 537     real             dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20,cexp1_20,cexp2_20;
 538     real             dx30,dy30,dz30,rsq30,rinv30,rinvsq30,r30,qq30,c6_30,c12_30,cexp1_30,cexp2_30;
 539     real             velec,felec,velecsum,facel,crf,krf,krf2;
 540     real             *charge;
 541     int              nvdwtype;
 542     real             rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,br,vvdwexp,sh_vdw_invrcut6;
 543     int              *vdwtype;
 544     real             *vdwparam;
 545     int              ewitab;
 546     real             ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace;
 547     real             *ewtab;
 548     real             rswitch,swV3,swV4,swV5,swF2,swF3,swF4,d,d2,sw,dsw;
 549
 550     x                = xx[0];
 551     f                = ff[0];
 552
 553     nri              = nlist->nri;
 554     iinr             = nlist->iinr;
 555     jindex           = nlist->jindex;
 556     jjnr             = nlist->jjnr;
 557     shiftidx         = nlist->shift;
 558     gid              = nlist->gid;
 559     shiftvec         = fr->shift_vec[0];
 560     fshift           = fr->fshift[0];
 561     facel            = fr->epsfac;
 562     charge           = mdatoms->chargeA;
 563     nvdwtype         = fr->ntype;
 564     vdwparam         = fr->nbfp;
 565     vdwtype          = mdatoms->typeA;
 566
 567     sh_ewald         = fr->ic->sh_ewald;
 568     ewtab            = fr->ic->tabq_coul_FDV0;
 569     ewtabscale       = fr->ic->tabq_scale;
 570     ewtabhalfspace   = 0.5/ewtabscale;
 571
 572     /* Setup water-specific parameters */
 573     inr              = nlist->iinr[0];
 574     iq1              = facel*charge[inr+1];
 575     iq2              = facel*charge[inr+2];
 576     iq3              = facel*charge[inr+3];
 577     vdwioffset0      = 2*nvdwtype*vdwtype[inr+0];
 578
 579     /* When we use explicit cutoffs the value must be identical for elec and VdW, so use elec as an arbitrary choice */
 580     rcutoff          = fr->rcoulomb;
 581     rcutoff2         = rcutoff*rcutoff;
 582
 583     rswitch          = fr->rcoulomb_switch;
 584     /* Setup switch parameters */
 585     d                = rcutoff-rswitch;
 586     swV3             = -10.0/(d*d*d);
 587     swV4             =  15.0/(d*d*d*d);
 588     swV5             =  -6.0/(d*d*d*d*d);
 589     swF2             = -30.0/(d*d*d);
 590     swF3             =  60.0/(d*d*d*d);
 591     swF4             = -30.0/(d*d*d*d*d);
 592
 593     outeriter        = 0;
 594     inneriter        = 0;
 595
 596     /* Start outer loop over neighborlists */
 597     for(iidx=0; iidx<nri; iidx++)
 598     {
 599         /* Load shift vector for this list */
 600         i_shift_offset   = DIM*shiftidx[iidx];
 601         shX              = shiftvec[i_shift_offset+XX];
 602         shY              = shiftvec[i_shift_offset+YY];
 603         shZ              = shiftvec[i_shift_offset+ZZ];
 604
 605         /* Load limits for loop over neighbors */
 606         j_index_start    = jindex[iidx];
 607         j_index_end      = jindex[iidx+1];
 608
 609         /* Get outer coordinate index */
 610         inr              = iinr[iidx];
 611         i_coord_offset   = DIM*inr;
 612
 613         /* Load i particle coords and add shift vector */
 614         ix0              = shX + x[i_coord_offset+DIM*0+XX];
 615         iy0              = shY + x[i_coord_offset+DIM*0+YY];
 616         iz0              = shZ + x[i_coord_offset+DIM*0+ZZ];
 617         ix1              = shX + x[i_coord_offset+DIM*1+XX];
 618         iy1              = shY + x[i_coord_offset+DIM*1+YY];
 619         iz1              = shZ + x[i_coord_offset+DIM*1+ZZ];
 620         ix2              = shX + x[i_coord_offset+DIM*2+XX];
 621         iy2              = shY + x[i_coord_offset+DIM*2+YY];
 622         iz2              = shZ + x[i_coord_offset+DIM*2+ZZ];
 623         ix3              = shX + x[i_coord_offset+DIM*3+XX];
 624         iy3              = shY + x[i_coord_offset+DIM*3+YY];
 625         iz3              = shZ + x[i_coord_offset+DIM*3+ZZ];
 626
 627         fix0             = 0.0;
 628         fiy0             = 0.0;
 629         fiz0             = 0.0;
 630         fix1             = 0.0;
 631         fiy1             = 0.0;
 632         fiz1             = 0.0;
 633         fix2             = 0.0;
 634         fiy2             = 0.0;
 635         fiz2             = 0.0;
 636         fix3             = 0.0;
 637         fiy3             = 0.0;
 638         fiz3             = 0.0;
 639
 640         /* Start inner kernel loop */
 641         for(jidx=j_index_start; jidx<j_index_end; jidx++)
 642         {
 643             /* Get j neighbor index, and coordinate index */
 644             jnr              = jjnr[jidx];
 645             j_coord_offset   = DIM*jnr;
 646
 647             /* load j atom coordinates */
 648             jx0              = x[j_coord_offset+DIM*0+XX];
 649             jy0              = x[j_coord_offset+DIM*0+YY];
 650             jz0              = x[j_coord_offset+DIM*0+ZZ];
 651
 652             /* Calculate displacement vector */
 653             dx00             = ix0 - jx0;
 654             dy00             = iy0 - jy0;
 655             dz00             = iz0 - jz0;
 656             dx10             = ix1 - jx0;
 657             dy10             = iy1 - jy0;
 658             dz10             = iz1 - jz0;
 659             dx20             = ix2 - jx0;
 660             dy20             = iy2 - jy0;
 661             dz20             = iz2 - jz0;
 662             dx30             = ix3 - jx0;
 663             dy30             = iy3 - jy0;
 664             dz30             = iz3 - jz0;
 665
 666             /* Calculate squared distance and things based on it */
 667             rsq00            = dx00*dx00+dy00*dy00+dz00*dz00;
 668             rsq10            = dx10*dx10+dy10*dy10+dz10*dz10;
 669             rsq20            = dx20*dx20+dy20*dy20+dz20*dz20;
 670             rsq30            = dx30*dx30+dy30*dy30+dz30*dz30;
 671
 672             rinv00           = gmx_invsqrt(rsq00);
 673             rinv10           = gmx_invsqrt(rsq10);
 674             rinv20           = gmx_invsqrt(rsq20);
 675             rinv30           = gmx_invsqrt(rsq30);
 676
 677             rinvsq00         = rinv00*rinv00;
 678             rinvsq10         = rinv10*rinv10;
 679             rinvsq20         = rinv20*rinv20;
 680             rinvsq30         = rinv30*rinv30;
 681
 682             /* Load parameters for j particles */
 683             jq0              = charge[jnr+0];
 684             vdwjidx0         = 2*vdwtype[jnr+0];
 685
 686             /**************************
 687              * CALCULATE INTERACTIONS *
 688              **************************/
 689
 690             if (rsq00<rcutoff2)
 691             {
 692
 693             r00              = rsq00*rinv00;
 694
 695             c6_00            = vdwparam[vdwioffset0+vdwjidx0];
 696             c12_00           = vdwparam[vdwioffset0+vdwjidx0+1];
 697
 698             /* LENNARD-JONES DISPERSION/REPULSION */
 699
 700             rinvsix          = rinvsq00*rinvsq00*rinvsq00;
 701             vvdw6            = c6_00*rinvsix;
 702             vvdw12           = c12_00*rinvsix*rinvsix;
 703             vvdw             = vvdw12*(1.0/12.0) - vvdw6*(1.0/6.0);
 704             fvdw             = (vvdw12-vvdw6)*rinvsq00;
 705
 706             d                = r00-rswitch;
 707             d                = (d>0.0) ? d : 0.0;
 708             d2               = d*d;
 709             sw               = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
 710
 711             dsw              = d2*(swF2+d*(swF3+d*swF4));
 712
 713             /* Evaluate switch function */
 714             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 715             fvdw             = fvdw*sw - rinv00*vvdw*dsw;
 716
 717             fscal            = fvdw;
 718
 719             /* Calculate temporary vectorial force */
 720             tx               = fscal*dx00;
 721             ty               = fscal*dy00;
 722             tz               = fscal*dz00;
 723
 724             /* Update vectorial force */
 725             fix0            += tx;
 726             fiy0            += ty;
 727             fiz0            += tz;
 728             f[j_coord_offset+DIM*0+XX] -= tx;
 729             f[j_coord_offset+DIM*0+YY] -= ty;
 730             f[j_coord_offset+DIM*0+ZZ] -= tz;
 731
 732             }
 733
 734             /**************************
 735              * CALCULATE INTERACTIONS *
 736              **************************/
 737
 738             if (rsq10<rcutoff2)
 739             {
 740
 741             r10              = rsq10*rinv10;
 742
 743             qq10             = iq1*jq0;
 744
 745             /* EWALD ELECTROSTATICS */
 746
 747             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 748             ewrt             = r10*ewtabscale;
 749             ewitab           = ewrt;
 750             eweps            = ewrt-ewitab;
 751             ewitab           = 4*ewitab;
 752             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 753             velec            = qq10*(rinv10-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 754             felec            = qq10*rinv10*(rinvsq10-felec);
 755
 756             d                = r10-rswitch;
 757             d                = (d>0.0) ? d : 0.0;
 758             d2               = d*d;
 759             sw               = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
 760
 761             dsw              = d2*(swF2+d*(swF3+d*swF4));
 762
 763             /* Evaluate switch function */
 764             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 765             felec            = felec*sw - rinv10*velec*dsw;
 766
 767             fscal            = felec;
 768
 769             /* Calculate temporary vectorial force */
 770             tx               = fscal*dx10;
 771             ty               = fscal*dy10;
 772             tz               = fscal*dz10;
 773
 774             /* Update vectorial force */
 775             fix1            += tx;
 776             fiy1            += ty;
 777             fiz1            += tz;
 778             f[j_coord_offset+DIM*0+XX] -= tx;
 779             f[j_coord_offset+DIM*0+YY] -= ty;
 780             f[j_coord_offset+DIM*0+ZZ] -= tz;
 781
 782             }
 783
 784             /**************************
 785              * CALCULATE INTERACTIONS *
 786              **************************/
 787
 788             if (rsq20<rcutoff2)
 789             {
 790
 791             r20              = rsq20*rinv20;
 792
 793             qq20             = iq2*jq0;
 794
 795             /* EWALD ELECTROSTATICS */
 796
 797             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 798             ewrt             = r20*ewtabscale;
 799             ewitab           = ewrt;
 800             eweps            = ewrt-ewitab;
 801             ewitab           = 4*ewitab;
 802             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 803             velec            = qq20*(rinv20-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 804             felec            = qq20*rinv20*(rinvsq20-felec);
 805
 806             d                = r20-rswitch;
 807             d                = (d>0.0) ? d : 0.0;
 808             d2               = d*d;
 809             sw               = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
 810
 811             dsw              = d2*(swF2+d*(swF3+d*swF4));
 812
 813             /* Evaluate switch function */
 814             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 815             felec            = felec*sw - rinv20*velec*dsw;
 816
 817             fscal            = felec;
 818
 819             /* Calculate temporary vectorial force */
 820             tx               = fscal*dx20;
 821             ty               = fscal*dy20;
 822             tz               = fscal*dz20;
 823
 824             /* Update vectorial force */
 825             fix2            += tx;
 826             fiy2            += ty;
 827             fiz2            += tz;
 828             f[j_coord_offset+DIM*0+XX] -= tx;
 829             f[j_coord_offset+DIM*0+YY] -= ty;
 830             f[j_coord_offset+DIM*0+ZZ] -= tz;
 831
 832             }
 833
 834             /**************************
 835              * CALCULATE INTERACTIONS *
 836              **************************/
 837
 838             if (rsq30<rcutoff2)
 839             {
 840
 841             r30              = rsq30*rinv30;
 842
 843             qq30             = iq3*jq0;
 844
 845             /* EWALD ELECTROSTATICS */
 846
 847             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 848             ewrt             = r30*ewtabscale;
 849             ewitab           = ewrt;
 850             eweps            = ewrt-ewitab;
 851             ewitab           = 4*ewitab;
 852             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 853             velec            = qq30*(rinv30-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 854             felec            = qq30*rinv30*(rinvsq30-felec);
 855
 856             d                = r30-rswitch;
 857             d                = (d>0.0) ? d : 0.0;
 858             d2               = d*d;
 859             sw               = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
 860
 861             dsw              = d2*(swF2+d*(swF3+d*swF4));
 862
 863             /* Evaluate switch function */
 864             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 865             felec            = felec*sw - rinv30*velec*dsw;
 866
 867             fscal            = felec;
 868
 869             /* Calculate temporary vectorial force */
 870             tx               = fscal*dx30;
 871             ty               = fscal*dy30;
 872             tz               = fscal*dz30;
 873
 874             /* Update vectorial force */
 875             fix3            += tx;
 876             fiy3            += ty;
 877             fiz3            += tz;
 878             f[j_coord_offset+DIM*0+XX] -= tx;
 879             f[j_coord_offset+DIM*0+YY] -= ty;
 880             f[j_coord_offset+DIM*0+ZZ] -= tz;
 881
 882             }
 883
 884             /* Inner loop uses 222 flops */
 885         }
 886         /* End of innermost loop */
 887
 888         tx = ty = tz = 0;
 889         f[i_coord_offset+DIM*0+XX] += fix0;
 890         f[i_coord_offset+DIM*0+YY] += fiy0;
 891         f[i_coord_offset+DIM*0+ZZ] += fiz0;
 892         tx                         += fix0;
 893         ty                         += fiy0;
 894         tz                         += fiz0;
 895         f[i_coord_offset+DIM*1+XX] += fix1;
 896         f[i_coord_offset+DIM*1+YY] += fiy1;
 897         f[i_coord_offset+DIM*1+ZZ] += fiz1;
 898         tx                         += fix1;
 899         ty                         += fiy1;
 900         tz                         += fiz1;
 901         f[i_coord_offset+DIM*2+XX] += fix2;
 902         f[i_coord_offset+DIM*2+YY] += fiy2;
 903         f[i_coord_offset+DIM*2+ZZ] += fiz2;
 904         tx                         += fix2;
 905         ty                         += fiy2;
 906         tz                         += fiz2;
 907         f[i_coord_offset+DIM*3+XX] += fix3;
 908         f[i_coord_offset+DIM*3+YY] += fiy3;
 909         f[i_coord_offset+DIM*3+ZZ] += fiz3;
 910         tx                         += fix3;
 911         ty                         += fiy3;
 912         tz                         += fiz3;
 913         fshift[i_shift_offset+XX]  += tx;
 914         fshift[i_shift_offset+YY]  += ty;
 915         fshift[i_shift_offset+ZZ]  += tz;
 916
 917         /* Increment number of inner iterations */
 918         inneriter                  += j_index_end - j_index_start;
 919
 920         /* Outer loop uses 39 flops */
 921     }
 922
 923     /* Increment number of outer iterations */
 924     outeriter        += nri;
 925
 926     /* Update outer/inner flops */
 927
 928     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W4_F,outeriter*39 + inneriter*222);
 929 }