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 "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_VdwLJSw_GeomW4P1_VF_c
  49  * Electrostatics interaction: Ewald
  50  * VdW interaction:            LennardJones
  51  * Geometry:                   Water4-Particle
  52  * Calculate force/pot:        PotentialAndForce
  53  */
  54 void
  55 nb_kernel_ElecEwSw_VdwLJSw_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      = 2*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         = 2*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             c12_00           = vdwparam[vdwioffset0+vdwjidx0+1];
 246
 247             /* LENNARD-JONES DISPERSION/REPULSION */
 248
 249             rinvsix          = rinvsq00*rinvsq00*rinvsq00;
 250             vvdw6            = c6_00*rinvsix;
 251             vvdw12           = c12_00*rinvsix*rinvsix;
 252             vvdw             = vvdw12*(1.0/12.0) - vvdw6*(1.0/6.0);
 253             fvdw             = (vvdw12-vvdw6)*rinvsq00;
 254
 255             d                = r00-rswitch;
 256             d                = (d>0.0) ? d : 0.0;
 257             d2               = d*d;
 258             sw               = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
 259
 260             dsw              = d2*(swF2+d*(swF3+d*swF4));
 261
 262             /* Evaluate switch function */
 263             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 264             fvdw             = fvdw*sw - rinv00*vvdw*dsw;
 265             vvdw            *= sw;
 266
 267             /* Update potential sums from outer loop */
 268             vvdwsum         += vvdw;
 269
 270             fscal            = fvdw;
 271
 272             /* Calculate temporary vectorial force */
 273             tx               = fscal*dx00;
 274             ty               = fscal*dy00;
 275             tz               = fscal*dz00;
 276
 277             /* Update vectorial force */
 278             fix0            += tx;
 279             fiy0            += ty;
 280             fiz0            += tz;
 281             f[j_coord_offset+DIM*0+XX] -= tx;
 282             f[j_coord_offset+DIM*0+YY] -= ty;
 283             f[j_coord_offset+DIM*0+ZZ] -= tz;
 284
 285             }
 286
 287             /**************************
 288              * CALCULATE INTERACTIONS *
 289              **************************/
 290
 291             if (rsq10<rcutoff2)
 292             {
 293
 294             r10              = rsq10*rinv10;
 295
 296             qq10             = iq1*jq0;
 297
 298             /* EWALD ELECTROSTATICS */
 299
 300             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 301             ewrt             = r10*ewtabscale;
 302             ewitab           = ewrt;
 303             eweps            = ewrt-ewitab;
 304             ewitab           = 4*ewitab;
 305             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 306             velec            = qq10*(rinv10-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 307             felec            = qq10*rinv10*(rinvsq10-felec);
 308
 309             d                = r10-rswitch;
 310             d                = (d>0.0) ? d : 0.0;
 311             d2               = d*d;
 312             sw               = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
 313
 314             dsw              = d2*(swF2+d*(swF3+d*swF4));
 315
 316             /* Evaluate switch function */
 317             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 318             felec            = felec*sw - rinv10*velec*dsw;
 319             velec           *= sw;
 320
 321             /* Update potential sums from outer loop */
 322             velecsum        += velec;
 323
 324             fscal            = felec;
 325
 326             /* Calculate temporary vectorial force */
 327             tx               = fscal*dx10;
 328             ty               = fscal*dy10;
 329             tz               = fscal*dz10;
 330
 331             /* Update vectorial force */
 332             fix1            += tx;
 333             fiy1            += ty;
 334             fiz1            += tz;
 335             f[j_coord_offset+DIM*0+XX] -= tx;
 336             f[j_coord_offset+DIM*0+YY] -= ty;
 337             f[j_coord_offset+DIM*0+ZZ] -= tz;
 338
 339             }
 340
 341             /**************************
 342              * CALCULATE INTERACTIONS *
 343              **************************/
 344
 345             if (rsq20<rcutoff2)
 346             {
 347
 348             r20              = rsq20*rinv20;
 349
 350             qq20             = iq2*jq0;
 351
 352             /* EWALD ELECTROSTATICS */
 353
 354             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 355             ewrt             = r20*ewtabscale;
 356             ewitab           = ewrt;
 357             eweps            = ewrt-ewitab;
 358             ewitab           = 4*ewitab;
 359             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 360             velec            = qq20*(rinv20-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 361             felec            = qq20*rinv20*(rinvsq20-felec);
 362
 363             d                = r20-rswitch;
 364             d                = (d>0.0) ? d : 0.0;
 365             d2               = d*d;
 366             sw               = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
 367
 368             dsw              = d2*(swF2+d*(swF3+d*swF4));
 369
 370             /* Evaluate switch function */
 371             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 372             felec            = felec*sw - rinv20*velec*dsw;
 373             velec           *= sw;
 374
 375             /* Update potential sums from outer loop */
 376             velecsum        += velec;
 377
 378             fscal            = felec;
 379
 380             /* Calculate temporary vectorial force */
 381             tx               = fscal*dx20;
 382             ty               = fscal*dy20;
 383             tz               = fscal*dz20;
 384
 385             /* Update vectorial force */
 386             fix2            += tx;
 387             fiy2            += ty;
 388             fiz2            += tz;
 389             f[j_coord_offset+DIM*0+XX] -= tx;
 390             f[j_coord_offset+DIM*0+YY] -= ty;
 391             f[j_coord_offset+DIM*0+ZZ] -= tz;
 392
 393             }
 394
 395             /**************************
 396              * CALCULATE INTERACTIONS *
 397              **************************/
 398
 399             if (rsq30<rcutoff2)
 400             {
 401
 402             r30              = rsq30*rinv30;
 403
 404             qq30             = iq3*jq0;
 405
 406             /* EWALD ELECTROSTATICS */
 407
 408             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 409             ewrt             = r30*ewtabscale;
 410             ewitab           = ewrt;
 411             eweps            = ewrt-ewitab;
 412             ewitab           = 4*ewitab;
 413             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 414             velec            = qq30*(rinv30-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 415             felec            = qq30*rinv30*(rinvsq30-felec);
 416
 417             d                = r30-rswitch;
 418             d                = (d>0.0) ? d : 0.0;
 419             d2               = d*d;
 420             sw               = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
 421
 422             dsw              = d2*(swF2+d*(swF3+d*swF4));
 423
 424             /* Evaluate switch function */
 425             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 426             felec            = felec*sw - rinv30*velec*dsw;
 427             velec           *= sw;
 428
 429             /* Update potential sums from outer loop */
 430             velecsum        += velec;
 431
 432             fscal            = felec;
 433
 434             /* Calculate temporary vectorial force */
 435             tx               = fscal*dx30;
 436             ty               = fscal*dy30;
 437             tz               = fscal*dz30;
 438
 439             /* Update vectorial force */
 440             fix3            += tx;
 441             fiy3            += ty;
 442             fiz3            += tz;
 443             f[j_coord_offset+DIM*0+XX] -= tx;
 444             f[j_coord_offset+DIM*0+YY] -= ty;
 445             f[j_coord_offset+DIM*0+ZZ] -= tz;
 446
 447             }
 448
 449             /* Inner loop uses 230 flops */
 450         }
 451         /* End of innermost loop */
 452
 453         tx = ty = tz = 0;
 454         f[i_coord_offset+DIM*0+XX] += fix0;
 455         f[i_coord_offset+DIM*0+YY] += fiy0;
 456         f[i_coord_offset+DIM*0+ZZ] += fiz0;
 457         tx                         += fix0;
 458         ty                         += fiy0;
 459         tz                         += fiz0;
 460         f[i_coord_offset+DIM*1+XX] += fix1;
 461         f[i_coord_offset+DIM*1+YY] += fiy1;
 462         f[i_coord_offset+DIM*1+ZZ] += fiz1;
 463         tx                         += fix1;
 464         ty                         += fiy1;
 465         tz                         += fiz1;
 466         f[i_coord_offset+DIM*2+XX] += fix2;
 467         f[i_coord_offset+DIM*2+YY] += fiy2;
 468         f[i_coord_offset+DIM*2+ZZ] += fiz2;
 469         tx                         += fix2;
 470         ty                         += fiy2;
 471         tz                         += fiz2;
 472         f[i_coord_offset+DIM*3+XX] += fix3;
 473         f[i_coord_offset+DIM*3+YY] += fiy3;
 474         f[i_coord_offset+DIM*3+ZZ] += fiz3;
 475         tx                         += fix3;
 476         ty                         += fiy3;
 477         tz                         += fiz3;
 478         fshift[i_shift_offset+XX]  += tx;
 479         fshift[i_shift_offset+YY]  += ty;
 480         fshift[i_shift_offset+ZZ]  += tz;
 481
 482         ggid                        = gid[iidx];
 483         /* Update potential energies */
 484         kernel_data->energygrp_elec[ggid] += velecsum;
 485         kernel_data->energygrp_vdw[ggid] += vvdwsum;
 486
 487         /* Increment number of inner iterations */
 488         inneriter                  += j_index_end - j_index_start;
 489
 490         /* Outer loop uses 41 flops */
 491     }
 492
 493     /* Increment number of outer iterations */
 494     outeriter        += nri;
 495
 496     /* Update outer/inner flops */
 497
 498     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W4_VF,outeriter*41 + inneriter*230);
 499 }
 500 /*
 501  * Gromacs nonbonded kernel:   nb_kernel_ElecEwSw_VdwLJSw_GeomW4P1_F_c
 502  * Electrostatics interaction: Ewald
 503  * VdW interaction:            LennardJones
 504  * Geometry:                   Water4-Particle
 505  * Calculate force/pot:        Force
 506  */
 507 void
 508 nb_kernel_ElecEwSw_VdwLJSw_GeomW4P1_F_c
 509                     (t_nblist                    * gmx_restrict       nlist,
 510                      rvec                        * gmx_restrict          xx,
 511                      rvec                        * gmx_restrict          ff,
 512                      t_forcerec                  * gmx_restrict          fr,
 513                      t_mdatoms                   * gmx_restrict     mdatoms,
 514                      nb_kernel_data_t gmx_unused * gmx_restrict kernel_data,
 515                      t_nrnb                      * gmx_restrict        nrnb)
 516 {
 517     int              i_shift_offset,i_coord_offset,j_coord_offset;
 518     int              j_index_start,j_index_end;
 519     int              nri,inr,ggid,iidx,jidx,jnr,outeriter,inneriter;
 520     real             shX,shY,shZ,tx,ty,tz,fscal,rcutoff,rcutoff2;
 521     int              *iinr,*jindex,*jjnr,*shiftidx,*gid;
 522     real             *shiftvec,*fshift,*x,*f;
 523     int              vdwioffset0;
 524     real             ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
 525     int              vdwioffset1;
 526     real             ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
 527     int              vdwioffset2;
 528     real             ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
 529     int              vdwioffset3;
 530     real             ix3,iy3,iz3,fix3,fiy3,fiz3,iq3,isai3;
 531     int              vdwjidx0;
 532     real             jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
 533     real             dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00,cexp1_00,cexp2_00;
 534     real             dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10,cexp1_10,cexp2_10;
 535     real             dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20,cexp1_20,cexp2_20;
 536     real             dx30,dy30,dz30,rsq30,rinv30,rinvsq30,r30,qq30,c6_30,c12_30,cexp1_30,cexp2_30;
 537     real             velec,felec,velecsum,facel,crf,krf,krf2;
 538     real             *charge;
 539     int              nvdwtype;
 540     real             rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,br,vvdwexp,sh_vdw_invrcut6;
 541     int              *vdwtype;
 542     real             *vdwparam;
 543     int              ewitab;
 544     real             ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace;
 545     real             *ewtab;
 546     real             rswitch,swV3,swV4,swV5,swF2,swF3,swF4,d,d2,sw,dsw;
 547
 548     x                = xx[0];
 549     f                = ff[0];
 550
 551     nri              = nlist->nri;
 552     iinr             = nlist->iinr;
 553     jindex           = nlist->jindex;
 554     jjnr             = nlist->jjnr;
 555     shiftidx         = nlist->shift;
 556     gid              = nlist->gid;
 557     shiftvec         = fr->shift_vec[0];
 558     fshift           = fr->fshift[0];
 559     facel            = fr->epsfac;
 560     charge           = mdatoms->chargeA;
 561     nvdwtype         = fr->ntype;
 562     vdwparam         = fr->nbfp;
 563     vdwtype          = mdatoms->typeA;
 564
 565     sh_ewald         = fr->ic->sh_ewald;
 566     ewtab            = fr->ic->tabq_coul_FDV0;
 567     ewtabscale       = fr->ic->tabq_scale;
 568     ewtabhalfspace   = 0.5/ewtabscale;
 569
 570     /* Setup water-specific parameters */
 571     inr              = nlist->iinr[0];
 572     iq1              = facel*charge[inr+1];
 573     iq2              = facel*charge[inr+2];
 574     iq3              = facel*charge[inr+3];
 575     vdwioffset0      = 2*nvdwtype*vdwtype[inr+0];
 576
 577     /* When we use explicit cutoffs the value must be identical for elec and VdW, so use elec as an arbitrary choice */
 578     rcutoff          = fr->rcoulomb;
 579     rcutoff2         = rcutoff*rcutoff;
 580
 581     rswitch          = fr->rcoulomb_switch;
 582     /* Setup switch parameters */
 583     d                = rcutoff-rswitch;
 584     swV3             = -10.0/(d*d*d);
 585     swV4             =  15.0/(d*d*d*d);
 586     swV5             =  -6.0/(d*d*d*d*d);
 587     swF2             = -30.0/(d*d*d);
 588     swF3             =  60.0/(d*d*d*d);
 589     swF4             = -30.0/(d*d*d*d*d);
 590
 591     outeriter        = 0;
 592     inneriter        = 0;
 593
 594     /* Start outer loop over neighborlists */
 595     for(iidx=0; iidx<nri; iidx++)
 596     {
 597         /* Load shift vector for this list */
 598         i_shift_offset   = DIM*shiftidx[iidx];
 599         shX              = shiftvec[i_shift_offset+XX];
 600         shY              = shiftvec[i_shift_offset+YY];
 601         shZ              = shiftvec[i_shift_offset+ZZ];
 602
 603         /* Load limits for loop over neighbors */
 604         j_index_start    = jindex[iidx];
 605         j_index_end      = jindex[iidx+1];
 606
 607         /* Get outer coordinate index */
 608         inr              = iinr[iidx];
 609         i_coord_offset   = DIM*inr;
 610
 611         /* Load i particle coords and add shift vector */
 612         ix0              = shX + x[i_coord_offset+DIM*0+XX];
 613         iy0              = shY + x[i_coord_offset+DIM*0+YY];
 614         iz0              = shZ + x[i_coord_offset+DIM*0+ZZ];
 615         ix1              = shX + x[i_coord_offset+DIM*1+XX];
 616         iy1              = shY + x[i_coord_offset+DIM*1+YY];
 617         iz1              = shZ + x[i_coord_offset+DIM*1+ZZ];
 618         ix2              = shX + x[i_coord_offset+DIM*2+XX];
 619         iy2              = shY + x[i_coord_offset+DIM*2+YY];
 620         iz2              = shZ + x[i_coord_offset+DIM*2+ZZ];
 621         ix3              = shX + x[i_coord_offset+DIM*3+XX];
 622         iy3              = shY + x[i_coord_offset+DIM*3+YY];
 623         iz3              = shZ + x[i_coord_offset+DIM*3+ZZ];
 624
 625         fix0             = 0.0;
 626         fiy0             = 0.0;
 627         fiz0             = 0.0;
 628         fix1             = 0.0;
 629         fiy1             = 0.0;
 630         fiz1             = 0.0;
 631         fix2             = 0.0;
 632         fiy2             = 0.0;
 633         fiz2             = 0.0;
 634         fix3             = 0.0;
 635         fiy3             = 0.0;
 636         fiz3             = 0.0;
 637
 638         /* Start inner kernel loop */
 639         for(jidx=j_index_start; jidx<j_index_end; jidx++)
 640         {
 641             /* Get j neighbor index, and coordinate index */
 642             jnr              = jjnr[jidx];
 643             j_coord_offset   = DIM*jnr;
 644
 645             /* load j atom coordinates */
 646             jx0              = x[j_coord_offset+DIM*0+XX];
 647             jy0              = x[j_coord_offset+DIM*0+YY];
 648             jz0              = x[j_coord_offset+DIM*0+ZZ];
 649
 650             /* Calculate displacement vector */
 651             dx00             = ix0 - jx0;
 652             dy00             = iy0 - jy0;
 653             dz00             = iz0 - jz0;
 654             dx10             = ix1 - jx0;
 655             dy10             = iy1 - jy0;
 656             dz10             = iz1 - jz0;
 657             dx20             = ix2 - jx0;
 658             dy20             = iy2 - jy0;
 659             dz20             = iz2 - jz0;
 660             dx30             = ix3 - jx0;
 661             dy30             = iy3 - jy0;
 662             dz30             = iz3 - jz0;
 663
 664             /* Calculate squared distance and things based on it */
 665             rsq00            = dx00*dx00+dy00*dy00+dz00*dz00;
 666             rsq10            = dx10*dx10+dy10*dy10+dz10*dz10;
 667             rsq20            = dx20*dx20+dy20*dy20+dz20*dz20;
 668             rsq30            = dx30*dx30+dy30*dy30+dz30*dz30;
 669
 670             rinv00           = gmx_invsqrt(rsq00);
 671             rinv10           = gmx_invsqrt(rsq10);
 672             rinv20           = gmx_invsqrt(rsq20);
 673             rinv30           = gmx_invsqrt(rsq30);
 674
 675             rinvsq00         = rinv00*rinv00;
 676             rinvsq10         = rinv10*rinv10;
 677             rinvsq20         = rinv20*rinv20;
 678             rinvsq30         = rinv30*rinv30;
 679
 680             /* Load parameters for j particles */
 681             jq0              = charge[jnr+0];
 682             vdwjidx0         = 2*vdwtype[jnr+0];
 683
 684             /**************************
 685              * CALCULATE INTERACTIONS *
 686              **************************/
 687
 688             if (rsq00<rcutoff2)
 689             {
 690
 691             r00              = rsq00*rinv00;
 692
 693             c6_00            = vdwparam[vdwioffset0+vdwjidx0];
 694             c12_00           = vdwparam[vdwioffset0+vdwjidx0+1];
 695
 696             /* LENNARD-JONES DISPERSION/REPULSION */
 697
 698             rinvsix          = rinvsq00*rinvsq00*rinvsq00;
 699             vvdw6            = c6_00*rinvsix;
 700             vvdw12           = c12_00*rinvsix*rinvsix;
 701             vvdw             = vvdw12*(1.0/12.0) - vvdw6*(1.0/6.0);
 702             fvdw             = (vvdw12-vvdw6)*rinvsq00;
 703
 704             d                = r00-rswitch;
 705             d                = (d>0.0) ? d : 0.0;
 706             d2               = d*d;
 707             sw               = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
 708
 709             dsw              = d2*(swF2+d*(swF3+d*swF4));
 710
 711             /* Evaluate switch function */
 712             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 713             fvdw             = fvdw*sw - rinv00*vvdw*dsw;
 714
 715             fscal            = fvdw;
 716
 717             /* Calculate temporary vectorial force */
 718             tx               = fscal*dx00;
 719             ty               = fscal*dy00;
 720             tz               = fscal*dz00;
 721
 722             /* Update vectorial force */
 723             fix0            += tx;
 724             fiy0            += ty;
 725             fiz0            += tz;
 726             f[j_coord_offset+DIM*0+XX] -= tx;
 727             f[j_coord_offset+DIM*0+YY] -= ty;
 728             f[j_coord_offset+DIM*0+ZZ] -= tz;
 729
 730             }
 731
 732             /**************************
 733              * CALCULATE INTERACTIONS *
 734              **************************/
 735
 736             if (rsq10<rcutoff2)
 737             {
 738
 739             r10              = rsq10*rinv10;
 740
 741             qq10             = iq1*jq0;
 742
 743             /* EWALD ELECTROSTATICS */
 744
 745             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 746             ewrt             = r10*ewtabscale;
 747             ewitab           = ewrt;
 748             eweps            = ewrt-ewitab;
 749             ewitab           = 4*ewitab;
 750             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 751             velec            = qq10*(rinv10-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 752             felec            = qq10*rinv10*(rinvsq10-felec);
 753
 754             d                = r10-rswitch;
 755             d                = (d>0.0) ? d : 0.0;
 756             d2               = d*d;
 757             sw               = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
 758
 759             dsw              = d2*(swF2+d*(swF3+d*swF4));
 760
 761             /* Evaluate switch function */
 762             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 763             felec            = felec*sw - rinv10*velec*dsw;
 764
 765             fscal            = felec;
 766
 767             /* Calculate temporary vectorial force */
 768             tx               = fscal*dx10;
 769             ty               = fscal*dy10;
 770             tz               = fscal*dz10;
 771
 772             /* Update vectorial force */
 773             fix1            += tx;
 774             fiy1            += ty;
 775             fiz1            += tz;
 776             f[j_coord_offset+DIM*0+XX] -= tx;
 777             f[j_coord_offset+DIM*0+YY] -= ty;
 778             f[j_coord_offset+DIM*0+ZZ] -= tz;
 779
 780             }
 781
 782             /**************************
 783              * CALCULATE INTERACTIONS *
 784              **************************/
 785
 786             if (rsq20<rcutoff2)
 787             {
 788
 789             r20              = rsq20*rinv20;
 790
 791             qq20             = iq2*jq0;
 792
 793             /* EWALD ELECTROSTATICS */
 794
 795             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 796             ewrt             = r20*ewtabscale;
 797             ewitab           = ewrt;
 798             eweps            = ewrt-ewitab;
 799             ewitab           = 4*ewitab;
 800             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 801             velec            = qq20*(rinv20-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 802             felec            = qq20*rinv20*(rinvsq20-felec);
 803
 804             d                = r20-rswitch;
 805             d                = (d>0.0) ? d : 0.0;
 806             d2               = d*d;
 807             sw               = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
 808
 809             dsw              = d2*(swF2+d*(swF3+d*swF4));
 810
 811             /* Evaluate switch function */
 812             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 813             felec            = felec*sw - rinv20*velec*dsw;
 814
 815             fscal            = felec;
 816
 817             /* Calculate temporary vectorial force */
 818             tx               = fscal*dx20;
 819             ty               = fscal*dy20;
 820             tz               = fscal*dz20;
 821
 822             /* Update vectorial force */
 823             fix2            += tx;
 824             fiy2            += ty;
 825             fiz2            += tz;
 826             f[j_coord_offset+DIM*0+XX] -= tx;
 827             f[j_coord_offset+DIM*0+YY] -= ty;
 828             f[j_coord_offset+DIM*0+ZZ] -= tz;
 829
 830             }
 831
 832             /**************************
 833              * CALCULATE INTERACTIONS *
 834              **************************/
 835
 836             if (rsq30<rcutoff2)
 837             {
 838
 839             r30              = rsq30*rinv30;
 840
 841             qq30             = iq3*jq0;
 842
 843             /* EWALD ELECTROSTATICS */
 844
 845             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 846             ewrt             = r30*ewtabscale;
 847             ewitab           = ewrt;
 848             eweps            = ewrt-ewitab;
 849             ewitab           = 4*ewitab;
 850             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 851             velec            = qq30*(rinv30-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 852             felec            = qq30*rinv30*(rinvsq30-felec);
 853
 854             d                = r30-rswitch;
 855             d                = (d>0.0) ? d : 0.0;
 856             d2               = d*d;
 857             sw               = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
 858
 859             dsw              = d2*(swF2+d*(swF3+d*swF4));
 860
 861             /* Evaluate switch function */
 862             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 863             felec            = felec*sw - rinv30*velec*dsw;
 864
 865             fscal            = felec;
 866
 867             /* Calculate temporary vectorial force */
 868             tx               = fscal*dx30;
 869             ty               = fscal*dy30;
 870             tz               = fscal*dz30;
 871
 872             /* Update vectorial force */
 873             fix3            += tx;
 874             fiy3            += ty;
 875             fiz3            += tz;
 876             f[j_coord_offset+DIM*0+XX] -= tx;
 877             f[j_coord_offset+DIM*0+YY] -= ty;
 878             f[j_coord_offset+DIM*0+ZZ] -= tz;
 879
 880             }
 881
 882             /* Inner loop uses 222 flops */
 883         }
 884         /* End of innermost loop */
 885
 886         tx = ty = tz = 0;
 887         f[i_coord_offset+DIM*0+XX] += fix0;
 888         f[i_coord_offset+DIM*0+YY] += fiy0;
 889         f[i_coord_offset+DIM*0+ZZ] += fiz0;
 890         tx                         += fix0;
 891         ty                         += fiy0;
 892         tz                         += fiz0;
 893         f[i_coord_offset+DIM*1+XX] += fix1;
 894         f[i_coord_offset+DIM*1+YY] += fiy1;
 895         f[i_coord_offset+DIM*1+ZZ] += fiz1;
 896         tx                         += fix1;
 897         ty                         += fiy1;
 898         tz                         += fiz1;
 899         f[i_coord_offset+DIM*2+XX] += fix2;
 900         f[i_coord_offset+DIM*2+YY] += fiy2;
 901         f[i_coord_offset+DIM*2+ZZ] += fiz2;
 902         tx                         += fix2;
 903         ty                         += fiy2;
 904         tz                         += fiz2;
 905         f[i_coord_offset+DIM*3+XX] += fix3;
 906         f[i_coord_offset+DIM*3+YY] += fiy3;
 907         f[i_coord_offset+DIM*3+ZZ] += fiz3;
 908         tx                         += fix3;
 909         ty                         += fiy3;
 910         tz                         += fiz3;
 911         fshift[i_shift_offset+XX]  += tx;
 912         fshift[i_shift_offset+YY]  += ty;
 913         fshift[i_shift_offset+ZZ]  += tz;
 914
 915         /* Increment number of inner iterations */
 916         inneriter                  += j_index_end - j_index_start;
 917
 918         /* Outer loop uses 39 flops */
 919     }
 920
 921     /* Increment number of outer iterations */
 922     outeriter        += nri;
 923
 924     /* Update outer/inner flops */
 925
 926     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W4_F,outeriter*39 + inneriter*222);
 927 }