src/gromacs/gmxlib/nonbonded/nb_kernel_c/nb_kernel_ElecEwSw_VdwBhamSw_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 "config.h"
  39
  40 #include <math.h>
  41
  42 #include "../nb_kernel.h"
  43 #include "gromacs/legacyheaders/types/simple.h"
  44 #include "gromacs/math/vec.h"
  45 #include "gromacs/legacyheaders/nrnb.h"
  46
  47 /*
  48  * Gromacs nonbonded kernel:   nb_kernel_ElecEwSw_VdwBhamSw_GeomW4P1_VF_c
  49  * Electrostatics interaction: Ewald
  50  * VdW interaction:            Buckingham
  51  * Geometry:                   Water4-Particle
  52  * Calculate force/pot:        PotentialAndForce
  53  */
  54 void
  55 nb_kernel_ElecEwSw_VdwBhamSw_GeomW4P1_VF_c
  56                     (t_nblist                    * gmx_restrict       nlist,
  57                      rvec                        * gmx_restrict          xx,
  58                      rvec                        * gmx_restrict          ff,
  59                      t_forcerec                  * gmx_restrict          fr,
  60                      t_mdatoms                   * gmx_restrict     mdatoms,
  61                      nb_kernel_data_t gmx_unused * gmx_restrict kernel_data,
  62                      t_nrnb                      * gmx_restrict        nrnb)
  63 {
  64     int              i_shift_offset,i_coord_offset,j_coord_offset;
  65     int              j_index_start,j_index_end;
  66     int              nri,inr,ggid,iidx,jidx,jnr,outeriter,inneriter;
  67     real             shX,shY,shZ,tx,ty,tz,fscal,rcutoff,rcutoff2;
  68     int              *iinr,*jindex,*jjnr,*shiftidx,*gid;
  69     real             *shiftvec,*fshift,*x,*f;
  70     int              vdwioffset0;
  71     real             ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
  72     int              vdwioffset1;
  73     real             ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
  74     int              vdwioffset2;
  75     real             ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
  76     int              vdwioffset3;
  77     real             ix3,iy3,iz3,fix3,fiy3,fiz3,iq3,isai3;
  78     int              vdwjidx0;
  79     real             jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
  80     real             dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00,cexp1_00,cexp2_00;
  81     real             dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10,cexp1_10,cexp2_10;
  82     real             dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20,cexp1_20,cexp2_20;
  83     real             dx30,dy30,dz30,rsq30,rinv30,rinvsq30,r30,qq30,c6_30,c12_30,cexp1_30,cexp2_30;
  84     real             velec,felec,velecsum,facel,crf,krf,krf2;
  85     real             *charge;
  86     int              nvdwtype;
  87     real             rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,br,vvdwexp,sh_vdw_invrcut6;
  88     int              *vdwtype;
  89     real             *vdwparam;
  90     int              ewitab;
  91     real             ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace;
  92     real             *ewtab;
  93     real             rswitch,swV3,swV4,swV5,swF2,swF3,swF4,d,d2,sw,dsw;
  94
  95     x                = xx[0];
  96     f                = ff[0];
  97
  98     nri              = nlist->nri;
  99     iinr             = nlist->iinr;
 100     jindex           = nlist->jindex;
 101     jjnr             = nlist->jjnr;
 102     shiftidx         = nlist->shift;
 103     gid              = nlist->gid;
 104     shiftvec         = fr->shift_vec[0];
 105     fshift           = fr->fshift[0];
 106     facel            = fr->epsfac;
 107     charge           = mdatoms->chargeA;
 108     nvdwtype         = fr->ntype;
 109     vdwparam         = fr->nbfp;
 110     vdwtype          = mdatoms->typeA;
 111
 112     sh_ewald         = fr->ic->sh_ewald;
 113     ewtab            = fr->ic->tabq_coul_FDV0;
 114     ewtabscale       = fr->ic->tabq_scale;
 115     ewtabhalfspace   = 0.5/ewtabscale;
 116
 117     /* Setup water-specific parameters */
 118     inr              = nlist->iinr[0];
 119     iq1              = facel*charge[inr+1];
 120     iq2              = facel*charge[inr+2];
 121     iq3              = facel*charge[inr+3];
 122     vdwioffset0      = 3*nvdwtype*vdwtype[inr+0];
 123
 124     /* When we use explicit cutoffs the value must be identical for elec and VdW, so use elec as an arbitrary choice */
 125     rcutoff          = fr->rcoulomb;
 126     rcutoff2         = rcutoff*rcutoff;
 127
 128     rswitch          = fr->rcoulomb_switch;
 129     /* Setup switch parameters */
 130     d                = rcutoff-rswitch;
 131     swV3             = -10.0/(d*d*d);
 132     swV4             =  15.0/(d*d*d*d);
 133     swV5             =  -6.0/(d*d*d*d*d);
 134     swF2             = -30.0/(d*d*d);
 135     swF3             =  60.0/(d*d*d*d);
 136     swF4             = -30.0/(d*d*d*d*d);
 137
 138     outeriter        = 0;
 139     inneriter        = 0;
 140
 141     /* Start outer loop over neighborlists */
 142     for(iidx=0; iidx<nri; iidx++)
 143     {
 144         /* Load shift vector for this list */
 145         i_shift_offset   = DIM*shiftidx[iidx];
 146         shX              = shiftvec[i_shift_offset+XX];
 147         shY              = shiftvec[i_shift_offset+YY];
 148         shZ              = shiftvec[i_shift_offset+ZZ];
 149
 150         /* Load limits for loop over neighbors */
 151         j_index_start    = jindex[iidx];
 152         j_index_end      = jindex[iidx+1];
 153
 154         /* Get outer coordinate index */
 155         inr              = iinr[iidx];
 156         i_coord_offset   = DIM*inr;
 157
 158         /* Load i particle coords and add shift vector */
 159         ix0              = shX + x[i_coord_offset+DIM*0+XX];
 160         iy0              = shY + x[i_coord_offset+DIM*0+YY];
 161         iz0              = shZ + x[i_coord_offset+DIM*0+ZZ];
 162         ix1              = shX + x[i_coord_offset+DIM*1+XX];
 163         iy1              = shY + x[i_coord_offset+DIM*1+YY];
 164         iz1              = shZ + x[i_coord_offset+DIM*1+ZZ];
 165         ix2              = shX + x[i_coord_offset+DIM*2+XX];
 166         iy2              = shY + x[i_coord_offset+DIM*2+YY];
 167         iz2              = shZ + x[i_coord_offset+DIM*2+ZZ];
 168         ix3              = shX + x[i_coord_offset+DIM*3+XX];
 169         iy3              = shY + x[i_coord_offset+DIM*3+YY];
 170         iz3              = shZ + x[i_coord_offset+DIM*3+ZZ];
 171
 172         fix0             = 0.0;
 173         fiy0             = 0.0;
 174         fiz0             = 0.0;
 175         fix1             = 0.0;
 176         fiy1             = 0.0;
 177         fiz1             = 0.0;
 178         fix2             = 0.0;
 179         fiy2             = 0.0;
 180         fiz2             = 0.0;
 181         fix3             = 0.0;
 182         fiy3             = 0.0;
 183         fiz3             = 0.0;
 184
 185         /* Reset potential sums */
 186         velecsum         = 0.0;
 187         vvdwsum          = 0.0;
 188
 189         /* Start inner kernel loop */
 190         for(jidx=j_index_start; jidx<j_index_end; jidx++)
 191         {
 192             /* Get j neighbor index, and coordinate index */
 193             jnr              = jjnr[jidx];
 194             j_coord_offset   = DIM*jnr;
 195
 196             /* load j atom coordinates */
 197             jx0              = x[j_coord_offset+DIM*0+XX];
 198             jy0              = x[j_coord_offset+DIM*0+YY];
 199             jz0              = x[j_coord_offset+DIM*0+ZZ];
 200
 201             /* Calculate displacement vector */
 202             dx00             = ix0 - jx0;
 203             dy00             = iy0 - jy0;
 204             dz00             = iz0 - jz0;
 205             dx10             = ix1 - jx0;
 206             dy10             = iy1 - jy0;
 207             dz10             = iz1 - jz0;
 208             dx20             = ix2 - jx0;
 209             dy20             = iy2 - jy0;
 210             dz20             = iz2 - jz0;
 211             dx30             = ix3 - jx0;
 212             dy30             = iy3 - jy0;
 213             dz30             = iz3 - jz0;
 214
 215             /* Calculate squared distance and things based on it */
 216             rsq00            = dx00*dx00+dy00*dy00+dz00*dz00;
 217             rsq10            = dx10*dx10+dy10*dy10+dz10*dz10;
 218             rsq20            = dx20*dx20+dy20*dy20+dz20*dz20;
 219             rsq30            = dx30*dx30+dy30*dy30+dz30*dz30;
 220
 221             rinv00           = gmx_invsqrt(rsq00);
 222             rinv10           = gmx_invsqrt(rsq10);
 223             rinv20           = gmx_invsqrt(rsq20);
 224             rinv30           = gmx_invsqrt(rsq30);
 225
 226             rinvsq00         = rinv00*rinv00;
 227             rinvsq10         = rinv10*rinv10;
 228             rinvsq20         = rinv20*rinv20;
 229             rinvsq30         = rinv30*rinv30;
 230
 231             /* Load parameters for j particles */
 232             jq0              = charge[jnr+0];
 233             vdwjidx0         = 3*vdwtype[jnr+0];
 234
 235             /**************************
 236              * CALCULATE INTERACTIONS *
 237              **************************/
 238
 239             if (rsq00<rcutoff2)
 240             {
 241
 242             r00              = rsq00*rinv00;
 243
 244             c6_00            = vdwparam[vdwioffset0+vdwjidx0];
 245             cexp1_00         = vdwparam[vdwioffset0+vdwjidx0+1];
 246             cexp2_00         = vdwparam[vdwioffset0+vdwjidx0+2];
 247
 248             /* BUCKINGHAM DISPERSION/REPULSION */
 249             rinvsix          = rinvsq00*rinvsq00*rinvsq00;
 250             vvdw6            = c6_00*rinvsix;
 251             br               = cexp2_00*r00;
 252             vvdwexp          = cexp1_00*exp(-br);
 253             vvdw             = vvdwexp - vvdw6*(1.0/6.0);
 254             fvdw             = (br*vvdwexp-vvdw6)*rinvsq00;
 255
 256             d                = r00-rswitch;
 257             d                = (d>0.0) ? d : 0.0;
 258             d2               = d*d;
 259             sw               = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
 260
 261             dsw              = d2*(swF2+d*(swF3+d*swF4));
 262
 263             /* Evaluate switch function */
 264             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 265             fvdw             = fvdw*sw - rinv00*vvdw*dsw;
 266             vvdw            *= sw;
 267
 268             /* Update potential sums from outer loop */
 269             vvdwsum         += vvdw;
 270
 271             fscal            = fvdw;
 272
 273             /* Calculate temporary vectorial force */
 274             tx               = fscal*dx00;
 275             ty               = fscal*dy00;
 276             tz               = fscal*dz00;
 277
 278             /* Update vectorial force */
 279             fix0            += tx;
 280             fiy0            += ty;
 281             fiz0            += tz;
 282             f[j_coord_offset+DIM*0+XX] -= tx;
 283             f[j_coord_offset+DIM*0+YY] -= ty;
 284             f[j_coord_offset+DIM*0+ZZ] -= tz;
 285
 286             }
 287
 288             /**************************
 289              * CALCULATE INTERACTIONS *
 290              **************************/
 291
 292             if (rsq10<rcutoff2)
 293             {
 294
 295             r10              = rsq10*rinv10;
 296
 297             qq10             = iq1*jq0;
 298
 299             /* EWALD ELECTROSTATICS */
 300
 301             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 302             ewrt             = r10*ewtabscale;
 303             ewitab           = ewrt;
 304             eweps            = ewrt-ewitab;
 305             ewitab           = 4*ewitab;
 306             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 307             velec            = qq10*(rinv10-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 308             felec            = qq10*rinv10*(rinvsq10-felec);
 309
 310             d                = r10-rswitch;
 311             d                = (d>0.0) ? d : 0.0;
 312             d2               = d*d;
 313             sw               = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
 314
 315             dsw              = d2*(swF2+d*(swF3+d*swF4));
 316
 317             /* Evaluate switch function */
 318             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 319             felec            = felec*sw - rinv10*velec*dsw;
 320             velec           *= sw;
 321
 322             /* Update potential sums from outer loop */
 323             velecsum        += velec;
 324
 325             fscal            = felec;
 326
 327             /* Calculate temporary vectorial force */
 328             tx               = fscal*dx10;
 329             ty               = fscal*dy10;
 330             tz               = fscal*dz10;
 331
 332             /* Update vectorial force */
 333             fix1            += tx;
 334             fiy1            += ty;
 335             fiz1            += tz;
 336             f[j_coord_offset+DIM*0+XX] -= tx;
 337             f[j_coord_offset+DIM*0+YY] -= ty;
 338             f[j_coord_offset+DIM*0+ZZ] -= tz;
 339
 340             }
 341
 342             /**************************
 343              * CALCULATE INTERACTIONS *
 344              **************************/
 345
 346             if (rsq20<rcutoff2)
 347             {
 348
 349             r20              = rsq20*rinv20;
 350
 351             qq20             = iq2*jq0;
 352
 353             /* EWALD ELECTROSTATICS */
 354
 355             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 356             ewrt             = r20*ewtabscale;
 357             ewitab           = ewrt;
 358             eweps            = ewrt-ewitab;
 359             ewitab           = 4*ewitab;
 360             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 361             velec            = qq20*(rinv20-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 362             felec            = qq20*rinv20*(rinvsq20-felec);
 363
 364             d                = r20-rswitch;
 365             d                = (d>0.0) ? d : 0.0;
 366             d2               = d*d;
 367             sw               = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
 368
 369             dsw              = d2*(swF2+d*(swF3+d*swF4));
 370
 371             /* Evaluate switch function */
 372             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 373             felec            = felec*sw - rinv20*velec*dsw;
 374             velec           *= sw;
 375
 376             /* Update potential sums from outer loop */
 377             velecsum        += velec;
 378
 379             fscal            = felec;
 380
 381             /* Calculate temporary vectorial force */
 382             tx               = fscal*dx20;
 383             ty               = fscal*dy20;
 384             tz               = fscal*dz20;
 385
 386             /* Update vectorial force */
 387             fix2            += tx;
 388             fiy2            += ty;
 389             fiz2            += tz;
 390             f[j_coord_offset+DIM*0+XX] -= tx;
 391             f[j_coord_offset+DIM*0+YY] -= ty;
 392             f[j_coord_offset+DIM*0+ZZ] -= tz;
 393
 394             }
 395
 396             /**************************
 397              * CALCULATE INTERACTIONS *
 398              **************************/
 399
 400             if (rsq30<rcutoff2)
 401             {
 402
 403             r30              = rsq30*rinv30;
 404
 405             qq30             = iq3*jq0;
 406
 407             /* EWALD ELECTROSTATICS */
 408
 409             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 410             ewrt             = r30*ewtabscale;
 411             ewitab           = ewrt;
 412             eweps            = ewrt-ewitab;
 413             ewitab           = 4*ewitab;
 414             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 415             velec            = qq30*(rinv30-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 416             felec            = qq30*rinv30*(rinvsq30-felec);
 417
 418             d                = r30-rswitch;
 419             d                = (d>0.0) ? d : 0.0;
 420             d2               = d*d;
 421             sw               = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
 422
 423             dsw              = d2*(swF2+d*(swF3+d*swF4));
 424
 425             /* Evaluate switch function */
 426             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 427             felec            = felec*sw - rinv30*velec*dsw;
 428             velec           *= sw;
 429
 430             /* Update potential sums from outer loop */
 431             velecsum        += velec;
 432
 433             fscal            = felec;
 434
 435             /* Calculate temporary vectorial force */
 436             tx               = fscal*dx30;
 437             ty               = fscal*dy30;
 438             tz               = fscal*dz30;
 439
 440             /* Update vectorial force */
 441             fix3            += tx;
 442             fiy3            += ty;
 443             fiz3            += tz;
 444             f[j_coord_offset+DIM*0+XX] -= tx;
 445             f[j_coord_offset+DIM*0+YY] -= ty;
 446             f[j_coord_offset+DIM*0+ZZ] -= tz;
 447
 448             }
 449
 450             /* Inner loop uses 256 flops */
 451         }
 452         /* End of innermost loop */
 453
 454         tx = ty = tz = 0;
 455         f[i_coord_offset+DIM*0+XX] += fix0;
 456         f[i_coord_offset+DIM*0+YY] += fiy0;
 457         f[i_coord_offset+DIM*0+ZZ] += fiz0;
 458         tx                         += fix0;
 459         ty                         += fiy0;
 460         tz                         += fiz0;
 461         f[i_coord_offset+DIM*1+XX] += fix1;
 462         f[i_coord_offset+DIM*1+YY] += fiy1;
 463         f[i_coord_offset+DIM*1+ZZ] += fiz1;
 464         tx                         += fix1;
 465         ty                         += fiy1;
 466         tz                         += fiz1;
 467         f[i_coord_offset+DIM*2+XX] += fix2;
 468         f[i_coord_offset+DIM*2+YY] += fiy2;
 469         f[i_coord_offset+DIM*2+ZZ] += fiz2;
 470         tx                         += fix2;
 471         ty                         += fiy2;
 472         tz                         += fiz2;
 473         f[i_coord_offset+DIM*3+XX] += fix3;
 474         f[i_coord_offset+DIM*3+YY] += fiy3;
 475         f[i_coord_offset+DIM*3+ZZ] += fiz3;
 476         tx                         += fix3;
 477         ty                         += fiy3;
 478         tz                         += fiz3;
 479         fshift[i_shift_offset+XX]  += tx;
 480         fshift[i_shift_offset+YY]  += ty;
 481         fshift[i_shift_offset+ZZ]  += tz;
 482
 483         ggid                        = gid[iidx];
 484         /* Update potential energies */
 485         kernel_data->energygrp_elec[ggid] += velecsum;
 486         kernel_data->energygrp_vdw[ggid] += vvdwsum;
 487
 488         /* Increment number of inner iterations */
 489         inneriter                  += j_index_end - j_index_start;
 490
 491         /* Outer loop uses 41 flops */
 492     }
 493
 494     /* Increment number of outer iterations */
 495     outeriter        += nri;
 496
 497     /* Update outer/inner flops */
 498
 499     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W4_VF,outeriter*41 + inneriter*256);
 500 }
 501 /*
 502  * Gromacs nonbonded kernel:   nb_kernel_ElecEwSw_VdwBhamSw_GeomW4P1_F_c
 503  * Electrostatics interaction: Ewald
 504  * VdW interaction:            Buckingham
 505  * Geometry:                   Water4-Particle
 506  * Calculate force/pot:        Force
 507  */
 508 void
 509 nb_kernel_ElecEwSw_VdwBhamSw_GeomW4P1_F_c
 510                     (t_nblist                    * gmx_restrict       nlist,
 511                      rvec                        * gmx_restrict          xx,
 512                      rvec                        * gmx_restrict          ff,
 513                      t_forcerec                  * gmx_restrict          fr,
 514                      t_mdatoms                   * gmx_restrict     mdatoms,
 515                      nb_kernel_data_t gmx_unused * gmx_restrict kernel_data,
 516                      t_nrnb                      * gmx_restrict        nrnb)
 517 {
 518     int              i_shift_offset,i_coord_offset,j_coord_offset;
 519     int              j_index_start,j_index_end;
 520     int              nri,inr,ggid,iidx,jidx,jnr,outeriter,inneriter;
 521     real             shX,shY,shZ,tx,ty,tz,fscal,rcutoff,rcutoff2;
 522     int              *iinr,*jindex,*jjnr,*shiftidx,*gid;
 523     real             *shiftvec,*fshift,*x,*f;
 524     int              vdwioffset0;
 525     real             ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
 526     int              vdwioffset1;
 527     real             ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
 528     int              vdwioffset2;
 529     real             ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
 530     int              vdwioffset3;
 531     real             ix3,iy3,iz3,fix3,fiy3,fiz3,iq3,isai3;
 532     int              vdwjidx0;
 533     real             jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
 534     real             dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00,cexp1_00,cexp2_00;
 535     real             dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10,cexp1_10,cexp2_10;
 536     real             dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20,cexp1_20,cexp2_20;
 537     real             dx30,dy30,dz30,rsq30,rinv30,rinvsq30,r30,qq30,c6_30,c12_30,cexp1_30,cexp2_30;
 538     real             velec,felec,velecsum,facel,crf,krf,krf2;
 539     real             *charge;
 540     int              nvdwtype;
 541     real             rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,br,vvdwexp,sh_vdw_invrcut6;
 542     int              *vdwtype;
 543     real             *vdwparam;
 544     int              ewitab;
 545     real             ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace;
 546     real             *ewtab;
 547     real             rswitch,swV3,swV4,swV5,swF2,swF3,swF4,d,d2,sw,dsw;
 548
 549     x                = xx[0];
 550     f                = ff[0];
 551
 552     nri              = nlist->nri;
 553     iinr             = nlist->iinr;
 554     jindex           = nlist->jindex;
 555     jjnr             = nlist->jjnr;
 556     shiftidx         = nlist->shift;
 557     gid              = nlist->gid;
 558     shiftvec         = fr->shift_vec[0];
 559     fshift           = fr->fshift[0];
 560     facel            = fr->epsfac;
 561     charge           = mdatoms->chargeA;
 562     nvdwtype         = fr->ntype;
 563     vdwparam         = fr->nbfp;
 564     vdwtype          = mdatoms->typeA;
 565
 566     sh_ewald         = fr->ic->sh_ewald;
 567     ewtab            = fr->ic->tabq_coul_FDV0;
 568     ewtabscale       = fr->ic->tabq_scale;
 569     ewtabhalfspace   = 0.5/ewtabscale;
 570
 571     /* Setup water-specific parameters */
 572     inr              = nlist->iinr[0];
 573     iq1              = facel*charge[inr+1];
 574     iq2              = facel*charge[inr+2];
 575     iq3              = facel*charge[inr+3];
 576     vdwioffset0      = 3*nvdwtype*vdwtype[inr+0];
 577
 578     /* When we use explicit cutoffs the value must be identical for elec and VdW, so use elec as an arbitrary choice */
 579     rcutoff          = fr->rcoulomb;
 580     rcutoff2         = rcutoff*rcutoff;
 581
 582     rswitch          = fr->rcoulomb_switch;
 583     /* Setup switch parameters */
 584     d                = rcutoff-rswitch;
 585     swV3             = -10.0/(d*d*d);
 586     swV4             =  15.0/(d*d*d*d);
 587     swV5             =  -6.0/(d*d*d*d*d);
 588     swF2             = -30.0/(d*d*d);
 589     swF3             =  60.0/(d*d*d*d);
 590     swF4             = -30.0/(d*d*d*d*d);
 591
 592     outeriter        = 0;
 593     inneriter        = 0;
 594
 595     /* Start outer loop over neighborlists */
 596     for(iidx=0; iidx<nri; iidx++)
 597     {
 598         /* Load shift vector for this list */
 599         i_shift_offset   = DIM*shiftidx[iidx];
 600         shX              = shiftvec[i_shift_offset+XX];
 601         shY              = shiftvec[i_shift_offset+YY];
 602         shZ              = shiftvec[i_shift_offset+ZZ];
 603
 604         /* Load limits for loop over neighbors */
 605         j_index_start    = jindex[iidx];
 606         j_index_end      = jindex[iidx+1];
 607
 608         /* Get outer coordinate index */
 609         inr              = iinr[iidx];
 610         i_coord_offset   = DIM*inr;
 611
 612         /* Load i particle coords and add shift vector */
 613         ix0              = shX + x[i_coord_offset+DIM*0+XX];
 614         iy0              = shY + x[i_coord_offset+DIM*0+YY];
 615         iz0              = shZ + x[i_coord_offset+DIM*0+ZZ];
 616         ix1              = shX + x[i_coord_offset+DIM*1+XX];
 617         iy1              = shY + x[i_coord_offset+DIM*1+YY];
 618         iz1              = shZ + x[i_coord_offset+DIM*1+ZZ];
 619         ix2              = shX + x[i_coord_offset+DIM*2+XX];
 620         iy2              = shY + x[i_coord_offset+DIM*2+YY];
 621         iz2              = shZ + x[i_coord_offset+DIM*2+ZZ];
 622         ix3              = shX + x[i_coord_offset+DIM*3+XX];
 623         iy3              = shY + x[i_coord_offset+DIM*3+YY];
 624         iz3              = shZ + x[i_coord_offset+DIM*3+ZZ];
 625
 626         fix0             = 0.0;
 627         fiy0             = 0.0;
 628         fiz0             = 0.0;
 629         fix1             = 0.0;
 630         fiy1             = 0.0;
 631         fiz1             = 0.0;
 632         fix2             = 0.0;
 633         fiy2             = 0.0;
 634         fiz2             = 0.0;
 635         fix3             = 0.0;
 636         fiy3             = 0.0;
 637         fiz3             = 0.0;
 638
 639         /* Start inner kernel loop */
 640         for(jidx=j_index_start; jidx<j_index_end; jidx++)
 641         {
 642             /* Get j neighbor index, and coordinate index */
 643             jnr              = jjnr[jidx];
 644             j_coord_offset   = DIM*jnr;
 645
 646             /* load j atom coordinates */
 647             jx0              = x[j_coord_offset+DIM*0+XX];
 648             jy0              = x[j_coord_offset+DIM*0+YY];
 649             jz0              = x[j_coord_offset+DIM*0+ZZ];
 650
 651             /* Calculate displacement vector */
 652             dx00             = ix0 - jx0;
 653             dy00             = iy0 - jy0;
 654             dz00             = iz0 - jz0;
 655             dx10             = ix1 - jx0;
 656             dy10             = iy1 - jy0;
 657             dz10             = iz1 - jz0;
 658             dx20             = ix2 - jx0;
 659             dy20             = iy2 - jy0;
 660             dz20             = iz2 - jz0;
 661             dx30             = ix3 - jx0;
 662             dy30             = iy3 - jy0;
 663             dz30             = iz3 - jz0;
 664
 665             /* Calculate squared distance and things based on it */
 666             rsq00            = dx00*dx00+dy00*dy00+dz00*dz00;
 667             rsq10            = dx10*dx10+dy10*dy10+dz10*dz10;
 668             rsq20            = dx20*dx20+dy20*dy20+dz20*dz20;
 669             rsq30            = dx30*dx30+dy30*dy30+dz30*dz30;
 670
 671             rinv00           = gmx_invsqrt(rsq00);
 672             rinv10           = gmx_invsqrt(rsq10);
 673             rinv20           = gmx_invsqrt(rsq20);
 674             rinv30           = gmx_invsqrt(rsq30);
 675
 676             rinvsq00         = rinv00*rinv00;
 677             rinvsq10         = rinv10*rinv10;
 678             rinvsq20         = rinv20*rinv20;
 679             rinvsq30         = rinv30*rinv30;
 680
 681             /* Load parameters for j particles */
 682             jq0              = charge[jnr+0];
 683             vdwjidx0         = 3*vdwtype[jnr+0];
 684
 685             /**************************
 686              * CALCULATE INTERACTIONS *
 687              **************************/
 688
 689             if (rsq00<rcutoff2)
 690             {
 691
 692             r00              = rsq00*rinv00;
 693
 694             c6_00            = vdwparam[vdwioffset0+vdwjidx0];
 695             cexp1_00         = vdwparam[vdwioffset0+vdwjidx0+1];
 696             cexp2_00         = vdwparam[vdwioffset0+vdwjidx0+2];
 697
 698             /* BUCKINGHAM DISPERSION/REPULSION */
 699             rinvsix          = rinvsq00*rinvsq00*rinvsq00;
 700             vvdw6            = c6_00*rinvsix;
 701             br               = cexp2_00*r00;
 702             vvdwexp          = cexp1_00*exp(-br);
 703             vvdw             = vvdwexp - vvdw6*(1.0/6.0);
 704             fvdw             = (br*vvdwexp-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 248 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*248);
 929 }