src/gromacs/gmxlib/nonbonded/nb_kernel_c/nb_kernel_ElecEwSw_VdwBhamSw_GeomW3P1_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_VdwBhamSw_GeomW3P1_VF_c
  51  * Electrostatics interaction: Ewald
  52  * VdW interaction:            Buckingham
  53  * Geometry:                   Water3-Particle
  54  * Calculate force/pot:        PotentialAndForce
  55  */
  56 void
  57 nb_kernel_ElecEwSw_VdwBhamSw_GeomW3P1_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              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             velec,felec,velecsum,facel,crf,krf,krf2;
  84     real             *charge;
  85     int              nvdwtype;
  86     real             rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,br,vvdwexp,sh_vdw_invrcut6;
  87     int              *vdwtype;
  88     real             *vdwparam;
  89     int              ewitab;
  90     real             ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace;
  91     real             *ewtab;
  92     real             rswitch,swV3,swV4,swV5,swF2,swF3,swF4,d,d2,sw,dsw;
  93
  94     x                = xx[0];
  95     f                = ff[0];
  96
  97     nri              = nlist->nri;
  98     iinr             = nlist->iinr;
  99     jindex           = nlist->jindex;
 100     jjnr             = nlist->jjnr;
 101     shiftidx         = nlist->shift;
 102     gid              = nlist->gid;
 103     shiftvec         = fr->shift_vec[0];
 104     fshift           = fr->fshift[0];
 105     facel            = fr->epsfac;
 106     charge           = mdatoms->chargeA;
 107     nvdwtype         = fr->ntype;
 108     vdwparam         = fr->nbfp;
 109     vdwtype          = mdatoms->typeA;
 110
 111     sh_ewald         = fr->ic->sh_ewald;
 112     ewtab            = fr->ic->tabq_coul_FDV0;
 113     ewtabscale       = fr->ic->tabq_scale;
 114     ewtabhalfspace   = 0.5/ewtabscale;
 115
 116     /* Setup water-specific parameters */
 117     inr              = nlist->iinr[0];
 118     iq0              = facel*charge[inr+0];
 119     iq1              = facel*charge[inr+1];
 120     iq2              = facel*charge[inr+2];
 121     vdwioffset0      = 3*nvdwtype*vdwtype[inr+0];
 122
 123     /* When we use explicit cutoffs the value must be identical for elec and VdW, so use elec as an arbitrary choice */
 124     rcutoff          = fr->rcoulomb;
 125     rcutoff2         = rcutoff*rcutoff;
 126
 127     rswitch          = fr->rcoulomb_switch;
 128     /* Setup switch parameters */
 129     d                = rcutoff-rswitch;
 130     swV3             = -10.0/(d*d*d);
 131     swV4             =  15.0/(d*d*d*d);
 132     swV5             =  -6.0/(d*d*d*d*d);
 133     swF2             = -30.0/(d*d*d);
 134     swF3             =  60.0/(d*d*d*d);
 135     swF4             = -30.0/(d*d*d*d*d);
 136
 137     outeriter        = 0;
 138     inneriter        = 0;
 139
 140     /* Start outer loop over neighborlists */
 141     for(iidx=0; iidx<nri; iidx++)
 142     {
 143         /* Load shift vector for this list */
 144         i_shift_offset   = DIM*shiftidx[iidx];
 145         shX              = shiftvec[i_shift_offset+XX];
 146         shY              = shiftvec[i_shift_offset+YY];
 147         shZ              = shiftvec[i_shift_offset+ZZ];
 148
 149         /* Load limits for loop over neighbors */
 150         j_index_start    = jindex[iidx];
 151         j_index_end      = jindex[iidx+1];
 152
 153         /* Get outer coordinate index */
 154         inr              = iinr[iidx];
 155         i_coord_offset   = DIM*inr;
 156
 157         /* Load i particle coords and add shift vector */
 158         ix0              = shX + x[i_coord_offset+DIM*0+XX];
 159         iy0              = shY + x[i_coord_offset+DIM*0+YY];
 160         iz0              = shZ + x[i_coord_offset+DIM*0+ZZ];
 161         ix1              = shX + x[i_coord_offset+DIM*1+XX];
 162         iy1              = shY + x[i_coord_offset+DIM*1+YY];
 163         iz1              = shZ + x[i_coord_offset+DIM*1+ZZ];
 164         ix2              = shX + x[i_coord_offset+DIM*2+XX];
 165         iy2              = shY + x[i_coord_offset+DIM*2+YY];
 166         iz2              = shZ + x[i_coord_offset+DIM*2+ZZ];
 167
 168         fix0             = 0.0;
 169         fiy0             = 0.0;
 170         fiz0             = 0.0;
 171         fix1             = 0.0;
 172         fiy1             = 0.0;
 173         fiz1             = 0.0;
 174         fix2             = 0.0;
 175         fiy2             = 0.0;
 176         fiz2             = 0.0;
 177
 178         /* Reset potential sums */
 179         velecsum         = 0.0;
 180         vvdwsum          = 0.0;
 181
 182         /* Start inner kernel loop */
 183         for(jidx=j_index_start; jidx<j_index_end; jidx++)
 184         {
 185             /* Get j neighbor index, and coordinate index */
 186             jnr              = jjnr[jidx];
 187             j_coord_offset   = DIM*jnr;
 188
 189             /* load j atom coordinates */
 190             jx0              = x[j_coord_offset+DIM*0+XX];
 191             jy0              = x[j_coord_offset+DIM*0+YY];
 192             jz0              = x[j_coord_offset+DIM*0+ZZ];
 193
 194             /* Calculate displacement vector */
 195             dx00             = ix0 - jx0;
 196             dy00             = iy0 - jy0;
 197             dz00             = iz0 - jz0;
 198             dx10             = ix1 - jx0;
 199             dy10             = iy1 - jy0;
 200             dz10             = iz1 - jz0;
 201             dx20             = ix2 - jx0;
 202             dy20             = iy2 - jy0;
 203             dz20             = iz2 - jz0;
 204
 205             /* Calculate squared distance and things based on it */
 206             rsq00            = dx00*dx00+dy00*dy00+dz00*dz00;
 207             rsq10            = dx10*dx10+dy10*dy10+dz10*dz10;
 208             rsq20            = dx20*dx20+dy20*dy20+dz20*dz20;
 209
 210             rinv00           = gmx_invsqrt(rsq00);
 211             rinv10           = gmx_invsqrt(rsq10);
 212             rinv20           = gmx_invsqrt(rsq20);
 213
 214             rinvsq00         = rinv00*rinv00;
 215             rinvsq10         = rinv10*rinv10;
 216             rinvsq20         = rinv20*rinv20;
 217
 218             /* Load parameters for j particles */
 219             jq0              = charge[jnr+0];
 220             vdwjidx0         = 3*vdwtype[jnr+0];
 221
 222             /**************************
 223              * CALCULATE INTERACTIONS *
 224              **************************/
 225
 226             if (rsq00<rcutoff2)
 227             {
 228
 229             r00              = rsq00*rinv00;
 230
 231             qq00             = iq0*jq0;
 232             c6_00            = vdwparam[vdwioffset0+vdwjidx0];
 233             cexp1_00         = vdwparam[vdwioffset0+vdwjidx0+1];
 234             cexp2_00         = vdwparam[vdwioffset0+vdwjidx0+2];
 235
 236             /* EWALD ELECTROSTATICS */
 237
 238             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 239             ewrt             = r00*ewtabscale;
 240             ewitab           = ewrt;
 241             eweps            = ewrt-ewitab;
 242             ewitab           = 4*ewitab;
 243             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 244             velec            = qq00*(rinv00-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 245             felec            = qq00*rinv00*(rinvsq00-felec);
 246
 247             /* BUCKINGHAM DISPERSION/REPULSION */
 248             rinvsix          = rinvsq00*rinvsq00*rinvsq00;
 249             vvdw6            = c6_00*rinvsix;
 250             br               = cexp2_00*r00;
 251             vvdwexp          = cexp1_00*exp(-br);
 252             vvdw             = vvdwexp - vvdw6*(1.0/6.0);
 253             fvdw             = (br*vvdwexp-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             felec            = felec*sw - rinv00*velec*dsw;
 265             fvdw             = fvdw*sw - rinv00*vvdw*dsw;
 266             velec           *= sw;
 267             vvdw            *= sw;
 268
 269             /* Update potential sums from outer loop */
 270             velecsum        += velec;
 271             vvdwsum         += vvdw;
 272
 273             fscal            = felec+fvdw;
 274
 275             /* Calculate temporary vectorial force */
 276             tx               = fscal*dx00;
 277             ty               = fscal*dy00;
 278             tz               = fscal*dz00;
 279
 280             /* Update vectorial force */
 281             fix0            += tx;
 282             fiy0            += ty;
 283             fiz0            += tz;
 284             f[j_coord_offset+DIM*0+XX] -= tx;
 285             f[j_coord_offset+DIM*0+YY] -= ty;
 286             f[j_coord_offset+DIM*0+ZZ] -= tz;
 287
 288             }
 289
 290             /**************************
 291              * CALCULATE INTERACTIONS *
 292              **************************/
 293
 294             if (rsq10<rcutoff2)
 295             {
 296
 297             r10              = rsq10*rinv10;
 298
 299             qq10             = iq1*jq0;
 300
 301             /* EWALD ELECTROSTATICS */
 302
 303             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 304             ewrt             = r10*ewtabscale;
 305             ewitab           = ewrt;
 306             eweps            = ewrt-ewitab;
 307             ewitab           = 4*ewitab;
 308             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 309             velec            = qq10*(rinv10-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 310             felec            = qq10*rinv10*(rinvsq10-felec);
 311
 312             d                = r10-rswitch;
 313             d                = (d>0.0) ? d : 0.0;
 314             d2               = d*d;
 315             sw               = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
 316
 317             dsw              = d2*(swF2+d*(swF3+d*swF4));
 318
 319             /* Evaluate switch function */
 320             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 321             felec            = felec*sw - rinv10*velec*dsw;
 322             velec           *= sw;
 323
 324             /* Update potential sums from outer loop */
 325             velecsum        += velec;
 326
 327             fscal            = felec;
 328
 329             /* Calculate temporary vectorial force */
 330             tx               = fscal*dx10;
 331             ty               = fscal*dy10;
 332             tz               = fscal*dz10;
 333
 334             /* Update vectorial force */
 335             fix1            += tx;
 336             fiy1            += ty;
 337             fiz1            += tz;
 338             f[j_coord_offset+DIM*0+XX] -= tx;
 339             f[j_coord_offset+DIM*0+YY] -= ty;
 340             f[j_coord_offset+DIM*0+ZZ] -= tz;
 341
 342             }
 343
 344             /**************************
 345              * CALCULATE INTERACTIONS *
 346              **************************/
 347
 348             if (rsq20<rcutoff2)
 349             {
 350
 351             r20              = rsq20*rinv20;
 352
 353             qq20             = iq2*jq0;
 354
 355             /* EWALD ELECTROSTATICS */
 356
 357             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 358             ewrt             = r20*ewtabscale;
 359             ewitab           = ewrt;
 360             eweps            = ewrt-ewitab;
 361             ewitab           = 4*ewitab;
 362             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 363             velec            = qq20*(rinv20-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 364             felec            = qq20*rinv20*(rinvsq20-felec);
 365
 366             d                = r20-rswitch;
 367             d                = (d>0.0) ? d : 0.0;
 368             d2               = d*d;
 369             sw               = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
 370
 371             dsw              = d2*(swF2+d*(swF3+d*swF4));
 372
 373             /* Evaluate switch function */
 374             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 375             felec            = felec*sw - rinv20*velec*dsw;
 376             velec           *= sw;
 377
 378             /* Update potential sums from outer loop */
 379             velecsum        += velec;
 380
 381             fscal            = felec;
 382
 383             /* Calculate temporary vectorial force */
 384             tx               = fscal*dx20;
 385             ty               = fscal*dy20;
 386             tz               = fscal*dz20;
 387
 388             /* Update vectorial force */
 389             fix2            += tx;
 390             fiy2            += ty;
 391             fiz2            += tz;
 392             f[j_coord_offset+DIM*0+XX] -= tx;
 393             f[j_coord_offset+DIM*0+YY] -= ty;
 394             f[j_coord_offset+DIM*0+ZZ] -= tz;
 395
 396             }
 397
 398             /* Inner loop uses 219 flops */
 399         }
 400         /* End of innermost loop */
 401
 402         tx = ty = tz = 0;
 403         f[i_coord_offset+DIM*0+XX] += fix0;
 404         f[i_coord_offset+DIM*0+YY] += fiy0;
 405         f[i_coord_offset+DIM*0+ZZ] += fiz0;
 406         tx                         += fix0;
 407         ty                         += fiy0;
 408         tz                         += fiz0;
 409         f[i_coord_offset+DIM*1+XX] += fix1;
 410         f[i_coord_offset+DIM*1+YY] += fiy1;
 411         f[i_coord_offset+DIM*1+ZZ] += fiz1;
 412         tx                         += fix1;
 413         ty                         += fiy1;
 414         tz                         += fiz1;
 415         f[i_coord_offset+DIM*2+XX] += fix2;
 416         f[i_coord_offset+DIM*2+YY] += fiy2;
 417         f[i_coord_offset+DIM*2+ZZ] += fiz2;
 418         tx                         += fix2;
 419         ty                         += fiy2;
 420         tz                         += fiz2;
 421         fshift[i_shift_offset+XX]  += tx;
 422         fshift[i_shift_offset+YY]  += ty;
 423         fshift[i_shift_offset+ZZ]  += tz;
 424
 425         ggid                        = gid[iidx];
 426         /* Update potential energies */
 427         kernel_data->energygrp_elec[ggid] += velecsum;
 428         kernel_data->energygrp_vdw[ggid] += vvdwsum;
 429
 430         /* Increment number of inner iterations */
 431         inneriter                  += j_index_end - j_index_start;
 432
 433         /* Outer loop uses 32 flops */
 434     }
 435
 436     /* Increment number of outer iterations */
 437     outeriter        += nri;
 438
 439     /* Update outer/inner flops */
 440
 441     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W3_VF,outeriter*32 + inneriter*219);
 442 }
 443 /*
 444  * Gromacs nonbonded kernel:   nb_kernel_ElecEwSw_VdwBhamSw_GeomW3P1_F_c
 445  * Electrostatics interaction: Ewald
 446  * VdW interaction:            Buckingham
 447  * Geometry:                   Water3-Particle
 448  * Calculate force/pot:        Force
 449  */
 450 void
 451 nb_kernel_ElecEwSw_VdwBhamSw_GeomW3P1_F_c
 452                     (t_nblist                    * gmx_restrict       nlist,
 453                      rvec                        * gmx_restrict          xx,
 454                      rvec                        * gmx_restrict          ff,
 455                      t_forcerec                  * gmx_restrict          fr,
 456                      t_mdatoms                   * gmx_restrict     mdatoms,
 457                      nb_kernel_data_t gmx_unused * gmx_restrict kernel_data,
 458                      t_nrnb                      * gmx_restrict        nrnb)
 459 {
 460     int              i_shift_offset,i_coord_offset,j_coord_offset;
 461     int              j_index_start,j_index_end;
 462     int              nri,inr,ggid,iidx,jidx,jnr,outeriter,inneriter;
 463     real             shX,shY,shZ,tx,ty,tz,fscal,rcutoff,rcutoff2;
 464     int              *iinr,*jindex,*jjnr,*shiftidx,*gid;
 465     real             *shiftvec,*fshift,*x,*f;
 466     int              vdwioffset0;
 467     real             ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
 468     int              vdwioffset1;
 469     real             ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
 470     int              vdwioffset2;
 471     real             ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
 472     int              vdwjidx0;
 473     real             jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
 474     real             dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00,cexp1_00,cexp2_00;
 475     real             dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10,cexp1_10,cexp2_10;
 476     real             dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20,cexp1_20,cexp2_20;
 477     real             velec,felec,velecsum,facel,crf,krf,krf2;
 478     real             *charge;
 479     int              nvdwtype;
 480     real             rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,br,vvdwexp,sh_vdw_invrcut6;
 481     int              *vdwtype;
 482     real             *vdwparam;
 483     int              ewitab;
 484     real             ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace;
 485     real             *ewtab;
 486     real             rswitch,swV3,swV4,swV5,swF2,swF3,swF4,d,d2,sw,dsw;
 487
 488     x                = xx[0];
 489     f                = ff[0];
 490
 491     nri              = nlist->nri;
 492     iinr             = nlist->iinr;
 493     jindex           = nlist->jindex;
 494     jjnr             = nlist->jjnr;
 495     shiftidx         = nlist->shift;
 496     gid              = nlist->gid;
 497     shiftvec         = fr->shift_vec[0];
 498     fshift           = fr->fshift[0];
 499     facel            = fr->epsfac;
 500     charge           = mdatoms->chargeA;
 501     nvdwtype         = fr->ntype;
 502     vdwparam         = fr->nbfp;
 503     vdwtype          = mdatoms->typeA;
 504
 505     sh_ewald         = fr->ic->sh_ewald;
 506     ewtab            = fr->ic->tabq_coul_FDV0;
 507     ewtabscale       = fr->ic->tabq_scale;
 508     ewtabhalfspace   = 0.5/ewtabscale;
 509
 510     /* Setup water-specific parameters */
 511     inr              = nlist->iinr[0];
 512     iq0              = facel*charge[inr+0];
 513     iq1              = facel*charge[inr+1];
 514     iq2              = facel*charge[inr+2];
 515     vdwioffset0      = 3*nvdwtype*vdwtype[inr+0];
 516
 517     /* When we use explicit cutoffs the value must be identical for elec and VdW, so use elec as an arbitrary choice */
 518     rcutoff          = fr->rcoulomb;
 519     rcutoff2         = rcutoff*rcutoff;
 520
 521     rswitch          = fr->rcoulomb_switch;
 522     /* Setup switch parameters */
 523     d                = rcutoff-rswitch;
 524     swV3             = -10.0/(d*d*d);
 525     swV4             =  15.0/(d*d*d*d);
 526     swV5             =  -6.0/(d*d*d*d*d);
 527     swF2             = -30.0/(d*d*d);
 528     swF3             =  60.0/(d*d*d*d);
 529     swF4             = -30.0/(d*d*d*d*d);
 530
 531     outeriter        = 0;
 532     inneriter        = 0;
 533
 534     /* Start outer loop over neighborlists */
 535     for(iidx=0; iidx<nri; iidx++)
 536     {
 537         /* Load shift vector for this list */
 538         i_shift_offset   = DIM*shiftidx[iidx];
 539         shX              = shiftvec[i_shift_offset+XX];
 540         shY              = shiftvec[i_shift_offset+YY];
 541         shZ              = shiftvec[i_shift_offset+ZZ];
 542
 543         /* Load limits for loop over neighbors */
 544         j_index_start    = jindex[iidx];
 545         j_index_end      = jindex[iidx+1];
 546
 547         /* Get outer coordinate index */
 548         inr              = iinr[iidx];
 549         i_coord_offset   = DIM*inr;
 550
 551         /* Load i particle coords and add shift vector */
 552         ix0              = shX + x[i_coord_offset+DIM*0+XX];
 553         iy0              = shY + x[i_coord_offset+DIM*0+YY];
 554         iz0              = shZ + x[i_coord_offset+DIM*0+ZZ];
 555         ix1              = shX + x[i_coord_offset+DIM*1+XX];
 556         iy1              = shY + x[i_coord_offset+DIM*1+YY];
 557         iz1              = shZ + x[i_coord_offset+DIM*1+ZZ];
 558         ix2              = shX + x[i_coord_offset+DIM*2+XX];
 559         iy2              = shY + x[i_coord_offset+DIM*2+YY];
 560         iz2              = shZ + x[i_coord_offset+DIM*2+ZZ];
 561
 562         fix0             = 0.0;
 563         fiy0             = 0.0;
 564         fiz0             = 0.0;
 565         fix1             = 0.0;
 566         fiy1             = 0.0;
 567         fiz1             = 0.0;
 568         fix2             = 0.0;
 569         fiy2             = 0.0;
 570         fiz2             = 0.0;
 571
 572         /* Start inner kernel loop */
 573         for(jidx=j_index_start; jidx<j_index_end; jidx++)
 574         {
 575             /* Get j neighbor index, and coordinate index */
 576             jnr              = jjnr[jidx];
 577             j_coord_offset   = DIM*jnr;
 578
 579             /* load j atom coordinates */
 580             jx0              = x[j_coord_offset+DIM*0+XX];
 581             jy0              = x[j_coord_offset+DIM*0+YY];
 582             jz0              = x[j_coord_offset+DIM*0+ZZ];
 583
 584             /* Calculate displacement vector */
 585             dx00             = ix0 - jx0;
 586             dy00             = iy0 - jy0;
 587             dz00             = iz0 - jz0;
 588             dx10             = ix1 - jx0;
 589             dy10             = iy1 - jy0;
 590             dz10             = iz1 - jz0;
 591             dx20             = ix2 - jx0;
 592             dy20             = iy2 - jy0;
 593             dz20             = iz2 - jz0;
 594
 595             /* Calculate squared distance and things based on it */
 596             rsq00            = dx00*dx00+dy00*dy00+dz00*dz00;
 597             rsq10            = dx10*dx10+dy10*dy10+dz10*dz10;
 598             rsq20            = dx20*dx20+dy20*dy20+dz20*dz20;
 599
 600             rinv00           = gmx_invsqrt(rsq00);
 601             rinv10           = gmx_invsqrt(rsq10);
 602             rinv20           = gmx_invsqrt(rsq20);
 603
 604             rinvsq00         = rinv00*rinv00;
 605             rinvsq10         = rinv10*rinv10;
 606             rinvsq20         = rinv20*rinv20;
 607
 608             /* Load parameters for j particles */
 609             jq0              = charge[jnr+0];
 610             vdwjidx0         = 3*vdwtype[jnr+0];
 611
 612             /**************************
 613              * CALCULATE INTERACTIONS *
 614              **************************/
 615
 616             if (rsq00<rcutoff2)
 617             {
 618
 619             r00              = rsq00*rinv00;
 620
 621             qq00             = iq0*jq0;
 622             c6_00            = vdwparam[vdwioffset0+vdwjidx0];
 623             cexp1_00         = vdwparam[vdwioffset0+vdwjidx0+1];
 624             cexp2_00         = vdwparam[vdwioffset0+vdwjidx0+2];
 625
 626             /* EWALD ELECTROSTATICS */
 627
 628             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 629             ewrt             = r00*ewtabscale;
 630             ewitab           = ewrt;
 631             eweps            = ewrt-ewitab;
 632             ewitab           = 4*ewitab;
 633             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 634             velec            = qq00*(rinv00-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 635             felec            = qq00*rinv00*(rinvsq00-felec);
 636
 637             /* BUCKINGHAM DISPERSION/REPULSION */
 638             rinvsix          = rinvsq00*rinvsq00*rinvsq00;
 639             vvdw6            = c6_00*rinvsix;
 640             br               = cexp2_00*r00;
 641             vvdwexp          = cexp1_00*exp(-br);
 642             vvdw             = vvdwexp - vvdw6*(1.0/6.0);
 643             fvdw             = (br*vvdwexp-vvdw6)*rinvsq00;
 644
 645             d                = r00-rswitch;
 646             d                = (d>0.0) ? d : 0.0;
 647             d2               = d*d;
 648             sw               = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
 649
 650             dsw              = d2*(swF2+d*(swF3+d*swF4));
 651
 652             /* Evaluate switch function */
 653             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 654             felec            = felec*sw - rinv00*velec*dsw;
 655             fvdw             = fvdw*sw - rinv00*vvdw*dsw;
 656
 657             fscal            = felec+fvdw;
 658
 659             /* Calculate temporary vectorial force */
 660             tx               = fscal*dx00;
 661             ty               = fscal*dy00;
 662             tz               = fscal*dz00;
 663
 664             /* Update vectorial force */
 665             fix0            += tx;
 666             fiy0            += ty;
 667             fiz0            += tz;
 668             f[j_coord_offset+DIM*0+XX] -= tx;
 669             f[j_coord_offset+DIM*0+YY] -= ty;
 670             f[j_coord_offset+DIM*0+ZZ] -= tz;
 671
 672             }
 673
 674             /**************************
 675              * CALCULATE INTERACTIONS *
 676              **************************/
 677
 678             if (rsq10<rcutoff2)
 679             {
 680
 681             r10              = rsq10*rinv10;
 682
 683             qq10             = iq1*jq0;
 684
 685             /* EWALD ELECTROSTATICS */
 686
 687             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 688             ewrt             = r10*ewtabscale;
 689             ewitab           = ewrt;
 690             eweps            = ewrt-ewitab;
 691             ewitab           = 4*ewitab;
 692             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 693             velec            = qq10*(rinv10-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 694             felec            = qq10*rinv10*(rinvsq10-felec);
 695
 696             d                = r10-rswitch;
 697             d                = (d>0.0) ? d : 0.0;
 698             d2               = d*d;
 699             sw               = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
 700
 701             dsw              = d2*(swF2+d*(swF3+d*swF4));
 702
 703             /* Evaluate switch function */
 704             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 705             felec            = felec*sw - rinv10*velec*dsw;
 706
 707             fscal            = felec;
 708
 709             /* Calculate temporary vectorial force */
 710             tx               = fscal*dx10;
 711             ty               = fscal*dy10;
 712             tz               = fscal*dz10;
 713
 714             /* Update vectorial force */
 715             fix1            += tx;
 716             fiy1            += ty;
 717             fiz1            += tz;
 718             f[j_coord_offset+DIM*0+XX] -= tx;
 719             f[j_coord_offset+DIM*0+YY] -= ty;
 720             f[j_coord_offset+DIM*0+ZZ] -= tz;
 721
 722             }
 723
 724             /**************************
 725              * CALCULATE INTERACTIONS *
 726              **************************/
 727
 728             if (rsq20<rcutoff2)
 729             {
 730
 731             r20              = rsq20*rinv20;
 732
 733             qq20             = iq2*jq0;
 734
 735             /* EWALD ELECTROSTATICS */
 736
 737             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 738             ewrt             = r20*ewtabscale;
 739             ewitab           = ewrt;
 740             eweps            = ewrt-ewitab;
 741             ewitab           = 4*ewitab;
 742             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 743             velec            = qq20*(rinv20-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 744             felec            = qq20*rinv20*(rinvsq20-felec);
 745
 746             d                = r20-rswitch;
 747             d                = (d>0.0) ? d : 0.0;
 748             d2               = d*d;
 749             sw               = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
 750
 751             dsw              = d2*(swF2+d*(swF3+d*swF4));
 752
 753             /* Evaluate switch function */
 754             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 755             felec            = felec*sw - rinv20*velec*dsw;
 756
 757             fscal            = felec;
 758
 759             /* Calculate temporary vectorial force */
 760             tx               = fscal*dx20;
 761             ty               = fscal*dy20;
 762             tz               = fscal*dz20;
 763
 764             /* Update vectorial force */
 765             fix2            += tx;
 766             fiy2            += ty;
 767             fiz2            += tz;
 768             f[j_coord_offset+DIM*0+XX] -= tx;
 769             f[j_coord_offset+DIM*0+YY] -= ty;
 770             f[j_coord_offset+DIM*0+ZZ] -= tz;
 771
 772             }
 773
 774             /* Inner loop uses 211 flops */
 775         }
 776         /* End of innermost loop */
 777
 778         tx = ty = tz = 0;
 779         f[i_coord_offset+DIM*0+XX] += fix0;
 780         f[i_coord_offset+DIM*0+YY] += fiy0;
 781         f[i_coord_offset+DIM*0+ZZ] += fiz0;
 782         tx                         += fix0;
 783         ty                         += fiy0;
 784         tz                         += fiz0;
 785         f[i_coord_offset+DIM*1+XX] += fix1;
 786         f[i_coord_offset+DIM*1+YY] += fiy1;
 787         f[i_coord_offset+DIM*1+ZZ] += fiz1;
 788         tx                         += fix1;
 789         ty                         += fiy1;
 790         tz                         += fiz1;
 791         f[i_coord_offset+DIM*2+XX] += fix2;
 792         f[i_coord_offset+DIM*2+YY] += fiy2;
 793         f[i_coord_offset+DIM*2+ZZ] += fiz2;
 794         tx                         += fix2;
 795         ty                         += fiy2;
 796         tz                         += fiz2;
 797         fshift[i_shift_offset+XX]  += tx;
 798         fshift[i_shift_offset+YY]  += ty;
 799         fshift[i_shift_offset+ZZ]  += tz;
 800
 801         /* Increment number of inner iterations */
 802         inneriter                  += j_index_end - j_index_start;
 803
 804         /* Outer loop uses 30 flops */
 805     }
 806
 807     /* Increment number of outer iterations */
 808     outeriter        += nri;
 809
 810     /* Update outer/inner flops */
 811
 812     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W3_F,outeriter*30 + inneriter*211);
 813 }