src/gmxlib/nonbonded/nb_kernel_c/nb_kernel_ElecEwSw_VdwBhamSw_GeomW4P1_c.c

   1 /*
   2  * Note: this file was generated by the Gromacs c kernel generator.
   3  *
   4  *                This source code is part of
   5  *
   6  *                 G   R   O   M   A   C   S
   7  *
   8  * Copyright (c) 2001-2012, The GROMACS Development Team
   9  *
  10  * Gromacs is a library for molecular simulation and trajectory analysis,
  11  * written by Erik Lindahl, David van der Spoel, Berk Hess, and others - for
  12  * a full list of developers and information, check out http://www.gromacs.org
  13  *
  14  * This program is free software; you can redistribute it and/or modify it under
  15  * the terms of the GNU Lesser General Public License as published by the Free
  16  * Software Foundation; either version 2 of the License, or (at your option) any
  17  * later version.
  18  *
  19  * To help fund GROMACS development, we humbly ask that you cite
  20  * the papers people have written on it - you can find them on the website.
  21  */
  22 #ifdef HAVE_CONFIG_H
  23 #include <config.h>
  24 #endif
  25
  26 #include <math.h>
  27
  28 #include "../nb_kernel.h"
  29 #include "types/simple.h"
  30 #include "vec.h"
  31 #include "nrnb.h"
  32
  33 /*
  34  * Gromacs nonbonded kernel:   nb_kernel_ElecEwSw_VdwBhamSw_GeomW4P1_VF_c
  35  * Electrostatics interaction: Ewald
  36  * VdW interaction:            Buckingham
  37  * Geometry:                   Water4-Particle
  38  * Calculate force/pot:        PotentialAndForce
  39  */
  40 void
  41 nb_kernel_ElecEwSw_VdwBhamSw_GeomW4P1_VF_c
  42                     (t_nblist * gmx_restrict                nlist,
  43                      rvec * gmx_restrict                    xx,
  44                      rvec * gmx_restrict                    ff,
  45                      t_forcerec * gmx_restrict              fr,
  46                      t_mdatoms * gmx_restrict               mdatoms,
  47                      nb_kernel_data_t * gmx_restrict        kernel_data,
  48                      t_nrnb * gmx_restrict                  nrnb)
  49 {
  50     int              i_shift_offset,i_coord_offset,j_coord_offset;
  51     int              j_index_start,j_index_end;
  52     int              nri,inr,ggid,iidx,jidx,jnr,outeriter,inneriter;
  53     real             shX,shY,shZ,tx,ty,tz,fscal,rcutoff,rcutoff2;
  54     int              *iinr,*jindex,*jjnr,*shiftidx,*gid;
  55     real             *shiftvec,*fshift,*x,*f;
  56     int              vdwioffset0;
  57     real             ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
  58     int              vdwioffset1;
  59     real             ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
  60     int              vdwioffset2;
  61     real             ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
  62     int              vdwioffset3;
  63     real             ix3,iy3,iz3,fix3,fiy3,fiz3,iq3,isai3;
  64     int              vdwjidx0;
  65     real             jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
  66     real             dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00,cexp1_00,cexp2_00;
  67     real             dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10,cexp1_10,cexp2_10;
  68     real             dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20,cexp1_20,cexp2_20;
  69     real             dx30,dy30,dz30,rsq30,rinv30,rinvsq30,r30,qq30,c6_30,c12_30,cexp1_30,cexp2_30;
  70     real             velec,felec,velecsum,facel,crf,krf,krf2;
  71     real             *charge;
  72     int              nvdwtype;
  73     real             rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,br,vvdwexp,sh_vdw_invrcut6;
  74     int              *vdwtype;
  75     real             *vdwparam;
  76     int              ewitab;
  77     real             ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace;
  78     real             *ewtab;
  79     real             rswitch,swV3,swV4,swV5,swF2,swF3,swF4,d,d2,sw,dsw;
  80
  81     x                = xx[0];
  82     f                = ff[0];
  83
  84     nri              = nlist->nri;
  85     iinr             = nlist->iinr;
  86     jindex           = nlist->jindex;
  87     jjnr             = nlist->jjnr;
  88     shiftidx         = nlist->shift;
  89     gid              = nlist->gid;
  90     shiftvec         = fr->shift_vec[0];
  91     fshift           = fr->fshift[0];
  92     facel            = fr->epsfac;
  93     charge           = mdatoms->chargeA;
  94     nvdwtype         = fr->ntype;
  95     vdwparam         = fr->nbfp;
  96     vdwtype          = mdatoms->typeA;
  97
  98     sh_ewald         = fr->ic->sh_ewald;
  99     ewtab            = fr->ic->tabq_coul_FDV0;
 100     ewtabscale       = fr->ic->tabq_scale;
 101     ewtabhalfspace   = 0.5/ewtabscale;
 102
 103     /* Setup water-specific parameters */
 104     inr              = nlist->iinr[0];
 105     iq1              = facel*charge[inr+1];
 106     iq2              = facel*charge[inr+2];
 107     iq3              = facel*charge[inr+3];
 108     vdwioffset0      = 3*nvdwtype*vdwtype[inr+0];
 109
 110     /* When we use explicit cutoffs the value must be identical for elec and VdW, so use elec as an arbitrary choice */
 111     rcutoff          = fr->rcoulomb;
 112     rcutoff2         = rcutoff*rcutoff;
 113
 114     rswitch          = fr->rcoulomb_switch;
 115     /* Setup switch parameters */
 116     d                = rcutoff-rswitch;
 117     swV3             = -10.0/(d*d*d);
 118     swV4             =  15.0/(d*d*d*d);
 119     swV5             =  -6.0/(d*d*d*d*d);
 120     swF2             = -30.0/(d*d*d);
 121     swF3             =  60.0/(d*d*d*d);
 122     swF4             = -30.0/(d*d*d*d*d);
 123
 124     outeriter        = 0;
 125     inneriter        = 0;
 126
 127     /* Start outer loop over neighborlists */
 128     for(iidx=0; iidx<nri; iidx++)
 129     {
 130         /* Load shift vector for this list */
 131         i_shift_offset   = DIM*shiftidx[iidx];
 132         shX              = shiftvec[i_shift_offset+XX];
 133         shY              = shiftvec[i_shift_offset+YY];
 134         shZ              = shiftvec[i_shift_offset+ZZ];
 135
 136         /* Load limits for loop over neighbors */
 137         j_index_start    = jindex[iidx];
 138         j_index_end      = jindex[iidx+1];
 139
 140         /* Get outer coordinate index */
 141         inr              = iinr[iidx];
 142         i_coord_offset   = DIM*inr;
 143
 144         /* Load i particle coords and add shift vector */
 145         ix0              = shX + x[i_coord_offset+DIM*0+XX];
 146         iy0              = shY + x[i_coord_offset+DIM*0+YY];
 147         iz0              = shZ + x[i_coord_offset+DIM*0+ZZ];
 148         ix1              = shX + x[i_coord_offset+DIM*1+XX];
 149         iy1              = shY + x[i_coord_offset+DIM*1+YY];
 150         iz1              = shZ + x[i_coord_offset+DIM*1+ZZ];
 151         ix2              = shX + x[i_coord_offset+DIM*2+XX];
 152         iy2              = shY + x[i_coord_offset+DIM*2+YY];
 153         iz2              = shZ + x[i_coord_offset+DIM*2+ZZ];
 154         ix3              = shX + x[i_coord_offset+DIM*3+XX];
 155         iy3              = shY + x[i_coord_offset+DIM*3+YY];
 156         iz3              = shZ + x[i_coord_offset+DIM*3+ZZ];
 157
 158         fix0             = 0.0;
 159         fiy0             = 0.0;
 160         fiz0             = 0.0;
 161         fix1             = 0.0;
 162         fiy1             = 0.0;
 163         fiz1             = 0.0;
 164         fix2             = 0.0;
 165         fiy2             = 0.0;
 166         fiz2             = 0.0;
 167         fix3             = 0.0;
 168         fiy3             = 0.0;
 169         fiz3             = 0.0;
 170
 171         /* Reset potential sums */
 172         velecsum         = 0.0;
 173         vvdwsum          = 0.0;
 174
 175         /* Start inner kernel loop */
 176         for(jidx=j_index_start; jidx<j_index_end; jidx++)
 177         {
 178             /* Get j neighbor index, and coordinate index */
 179             jnr              = jjnr[jidx];
 180             j_coord_offset   = DIM*jnr;
 181
 182             /* load j atom coordinates */
 183             jx0              = x[j_coord_offset+DIM*0+XX];
 184             jy0              = x[j_coord_offset+DIM*0+YY];
 185             jz0              = x[j_coord_offset+DIM*0+ZZ];
 186
 187             /* Calculate displacement vector */
 188             dx00             = ix0 - jx0;
 189             dy00             = iy0 - jy0;
 190             dz00             = iz0 - jz0;
 191             dx10             = ix1 - jx0;
 192             dy10             = iy1 - jy0;
 193             dz10             = iz1 - jz0;
 194             dx20             = ix2 - jx0;
 195             dy20             = iy2 - jy0;
 196             dz20             = iz2 - jz0;
 197             dx30             = ix3 - jx0;
 198             dy30             = iy3 - jy0;
 199             dz30             = iz3 - jz0;
 200
 201             /* Calculate squared distance and things based on it */
 202             rsq00            = dx00*dx00+dy00*dy00+dz00*dz00;
 203             rsq10            = dx10*dx10+dy10*dy10+dz10*dz10;
 204             rsq20            = dx20*dx20+dy20*dy20+dz20*dz20;
 205             rsq30            = dx30*dx30+dy30*dy30+dz30*dz30;
 206
 207             rinv00           = gmx_invsqrt(rsq00);
 208             rinv10           = gmx_invsqrt(rsq10);
 209             rinv20           = gmx_invsqrt(rsq20);
 210             rinv30           = gmx_invsqrt(rsq30);
 211
 212             rinvsq00         = rinv00*rinv00;
 213             rinvsq10         = rinv10*rinv10;
 214             rinvsq20         = rinv20*rinv20;
 215             rinvsq30         = rinv30*rinv30;
 216
 217             /* Load parameters for j particles */
 218             jq0              = charge[jnr+0];
 219             vdwjidx0         = 3*vdwtype[jnr+0];
 220
 221             /**************************
 222              * CALCULATE INTERACTIONS *
 223              **************************/
 224
 225             if (rsq00<rcutoff2)
 226             {
 227
 228             r00              = rsq00*rinv00;
 229
 230             c6_00            = vdwparam[vdwioffset0+vdwjidx0];
 231             cexp1_00         = vdwparam[vdwioffset0+vdwjidx0+1];
 232             cexp2_00         = vdwparam[vdwioffset0+vdwjidx0+2];
 233
 234             /* BUCKINGHAM DISPERSION/REPULSION */
 235             rinvsix          = rinvsq00*rinvsq00*rinvsq00;
 236             vvdw6            = c6_00*rinvsix;
 237             br               = cexp2_00*r00;
 238             vvdwexp          = cexp1_00*exp(-br);
 239             vvdw             = vvdwexp - vvdw6*(1.0/6.0);
 240             fvdw             = (br*vvdwexp-vvdw6)*rinvsq00;
 241
 242             d                = r00-rswitch;
 243             d                = (d>0.0) ? d : 0.0;
 244             d2               = d*d;
 245             sw               = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
 246
 247             dsw              = d2*(swF2+d*(swF3+d*swF4));
 248
 249             /* Evaluate switch function */
 250             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 251             fvdw             = fvdw*sw - rinv00*vvdw*dsw;
 252             vvdw            *= sw;
 253
 254             /* Update potential sums from outer loop */
 255             vvdwsum         += vvdw;
 256
 257             fscal            = fvdw;
 258
 259             /* Calculate temporary vectorial force */
 260             tx               = fscal*dx00;
 261             ty               = fscal*dy00;
 262             tz               = fscal*dz00;
 263
 264             /* Update vectorial force */
 265             fix0            += tx;
 266             fiy0            += ty;
 267             fiz0            += tz;
 268             f[j_coord_offset+DIM*0+XX] -= tx;
 269             f[j_coord_offset+DIM*0+YY] -= ty;
 270             f[j_coord_offset+DIM*0+ZZ] -= tz;
 271
 272             }
 273
 274             /**************************
 275              * CALCULATE INTERACTIONS *
 276              **************************/
 277
 278             if (rsq10<rcutoff2)
 279             {
 280
 281             r10              = rsq10*rinv10;
 282
 283             qq10             = iq1*jq0;
 284
 285             /* EWALD ELECTROSTATICS */
 286
 287             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 288             ewrt             = r10*ewtabscale;
 289             ewitab           = ewrt;
 290             eweps            = ewrt-ewitab;
 291             ewitab           = 4*ewitab;
 292             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 293             velec            = qq10*(rinv10-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 294             felec            = qq10*rinv10*(rinvsq10-felec);
 295
 296             d                = r10-rswitch;
 297             d                = (d>0.0) ? d : 0.0;
 298             d2               = d*d;
 299             sw               = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
 300
 301             dsw              = d2*(swF2+d*(swF3+d*swF4));
 302
 303             /* Evaluate switch function */
 304             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 305             felec            = felec*sw - rinv10*velec*dsw;
 306             velec           *= sw;
 307
 308             /* Update potential sums from outer loop */
 309             velecsum        += velec;
 310
 311             fscal            = felec;
 312
 313             /* Calculate temporary vectorial force */
 314             tx               = fscal*dx10;
 315             ty               = fscal*dy10;
 316             tz               = fscal*dz10;
 317
 318             /* Update vectorial force */
 319             fix1            += tx;
 320             fiy1            += ty;
 321             fiz1            += tz;
 322             f[j_coord_offset+DIM*0+XX] -= tx;
 323             f[j_coord_offset+DIM*0+YY] -= ty;
 324             f[j_coord_offset+DIM*0+ZZ] -= tz;
 325
 326             }
 327
 328             /**************************
 329              * CALCULATE INTERACTIONS *
 330              **************************/
 331
 332             if (rsq20<rcutoff2)
 333             {
 334
 335             r20              = rsq20*rinv20;
 336
 337             qq20             = iq2*jq0;
 338
 339             /* EWALD ELECTROSTATICS */
 340
 341             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 342             ewrt             = r20*ewtabscale;
 343             ewitab           = ewrt;
 344             eweps            = ewrt-ewitab;
 345             ewitab           = 4*ewitab;
 346             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 347             velec            = qq20*(rinv20-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 348             felec            = qq20*rinv20*(rinvsq20-felec);
 349
 350             d                = r20-rswitch;
 351             d                = (d>0.0) ? d : 0.0;
 352             d2               = d*d;
 353             sw               = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
 354
 355             dsw              = d2*(swF2+d*(swF3+d*swF4));
 356
 357             /* Evaluate switch function */
 358             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 359             felec            = felec*sw - rinv20*velec*dsw;
 360             velec           *= sw;
 361
 362             /* Update potential sums from outer loop */
 363             velecsum        += velec;
 364
 365             fscal            = felec;
 366
 367             /* Calculate temporary vectorial force */
 368             tx               = fscal*dx20;
 369             ty               = fscal*dy20;
 370             tz               = fscal*dz20;
 371
 372             /* Update vectorial force */
 373             fix2            += tx;
 374             fiy2            += ty;
 375             fiz2            += tz;
 376             f[j_coord_offset+DIM*0+XX] -= tx;
 377             f[j_coord_offset+DIM*0+YY] -= ty;
 378             f[j_coord_offset+DIM*0+ZZ] -= tz;
 379
 380             }
 381
 382             /**************************
 383              * CALCULATE INTERACTIONS *
 384              **************************/
 385
 386             if (rsq30<rcutoff2)
 387             {
 388
 389             r30              = rsq30*rinv30;
 390
 391             qq30             = iq3*jq0;
 392
 393             /* EWALD ELECTROSTATICS */
 394
 395             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 396             ewrt             = r30*ewtabscale;
 397             ewitab           = ewrt;
 398             eweps            = ewrt-ewitab;
 399             ewitab           = 4*ewitab;
 400             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 401             velec            = qq30*(rinv30-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 402             felec            = qq30*rinv30*(rinvsq30-felec);
 403
 404             d                = r30-rswitch;
 405             d                = (d>0.0) ? d : 0.0;
 406             d2               = d*d;
 407             sw               = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
 408
 409             dsw              = d2*(swF2+d*(swF3+d*swF4));
 410
 411             /* Evaluate switch function */
 412             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 413             felec            = felec*sw - rinv30*velec*dsw;
 414             velec           *= sw;
 415
 416             /* Update potential sums from outer loop */
 417             velecsum        += velec;
 418
 419             fscal            = felec;
 420
 421             /* Calculate temporary vectorial force */
 422             tx               = fscal*dx30;
 423             ty               = fscal*dy30;
 424             tz               = fscal*dz30;
 425
 426             /* Update vectorial force */
 427             fix3            += tx;
 428             fiy3            += ty;
 429             fiz3            += tz;
 430             f[j_coord_offset+DIM*0+XX] -= tx;
 431             f[j_coord_offset+DIM*0+YY] -= ty;
 432             f[j_coord_offset+DIM*0+ZZ] -= tz;
 433
 434             }
 435
 436             /* Inner loop uses 256 flops */
 437         }
 438         /* End of innermost loop */
 439
 440         tx = ty = tz = 0;
 441         f[i_coord_offset+DIM*0+XX] += fix0;
 442         f[i_coord_offset+DIM*0+YY] += fiy0;
 443         f[i_coord_offset+DIM*0+ZZ] += fiz0;
 444         tx                         += fix0;
 445         ty                         += fiy0;
 446         tz                         += fiz0;
 447         f[i_coord_offset+DIM*1+XX] += fix1;
 448         f[i_coord_offset+DIM*1+YY] += fiy1;
 449         f[i_coord_offset+DIM*1+ZZ] += fiz1;
 450         tx                         += fix1;
 451         ty                         += fiy1;
 452         tz                         += fiz1;
 453         f[i_coord_offset+DIM*2+XX] += fix2;
 454         f[i_coord_offset+DIM*2+YY] += fiy2;
 455         f[i_coord_offset+DIM*2+ZZ] += fiz2;
 456         tx                         += fix2;
 457         ty                         += fiy2;
 458         tz                         += fiz2;
 459         f[i_coord_offset+DIM*3+XX] += fix3;
 460         f[i_coord_offset+DIM*3+YY] += fiy3;
 461         f[i_coord_offset+DIM*3+ZZ] += fiz3;
 462         tx                         += fix3;
 463         ty                         += fiy3;
 464         tz                         += fiz3;
 465         fshift[i_shift_offset+XX]  += tx;
 466         fshift[i_shift_offset+YY]  += ty;
 467         fshift[i_shift_offset+ZZ]  += tz;
 468
 469         ggid                        = gid[iidx];
 470         /* Update potential energies */
 471         kernel_data->energygrp_elec[ggid] += velecsum;
 472         kernel_data->energygrp_vdw[ggid] += vvdwsum;
 473
 474         /* Increment number of inner iterations */
 475         inneriter                  += j_index_end - j_index_start;
 476
 477         /* Outer loop uses 41 flops */
 478     }
 479
 480     /* Increment number of outer iterations */
 481     outeriter        += nri;
 482
 483     /* Update outer/inner flops */
 484
 485     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W4_VF,outeriter*41 + inneriter*256);
 486 }
 487 /*
 488  * Gromacs nonbonded kernel:   nb_kernel_ElecEwSw_VdwBhamSw_GeomW4P1_F_c
 489  * Electrostatics interaction: Ewald
 490  * VdW interaction:            Buckingham
 491  * Geometry:                   Water4-Particle
 492  * Calculate force/pot:        Force
 493  */
 494 void
 495 nb_kernel_ElecEwSw_VdwBhamSw_GeomW4P1_F_c
 496                     (t_nblist * gmx_restrict                nlist,
 497                      rvec * gmx_restrict                    xx,
 498                      rvec * gmx_restrict                    ff,
 499                      t_forcerec * gmx_restrict              fr,
 500                      t_mdatoms * gmx_restrict               mdatoms,
 501                      nb_kernel_data_t * gmx_restrict        kernel_data,
 502                      t_nrnb * gmx_restrict                  nrnb)
 503 {
 504     int              i_shift_offset,i_coord_offset,j_coord_offset;
 505     int              j_index_start,j_index_end;
 506     int              nri,inr,ggid,iidx,jidx,jnr,outeriter,inneriter;
 507     real             shX,shY,shZ,tx,ty,tz,fscal,rcutoff,rcutoff2;
 508     int              *iinr,*jindex,*jjnr,*shiftidx,*gid;
 509     real             *shiftvec,*fshift,*x,*f;
 510     int              vdwioffset0;
 511     real             ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
 512     int              vdwioffset1;
 513     real             ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
 514     int              vdwioffset2;
 515     real             ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
 516     int              vdwioffset3;
 517     real             ix3,iy3,iz3,fix3,fiy3,fiz3,iq3,isai3;
 518     int              vdwjidx0;
 519     real             jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
 520     real             dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00,cexp1_00,cexp2_00;
 521     real             dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10,cexp1_10,cexp2_10;
 522     real             dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20,cexp1_20,cexp2_20;
 523     real             dx30,dy30,dz30,rsq30,rinv30,rinvsq30,r30,qq30,c6_30,c12_30,cexp1_30,cexp2_30;
 524     real             velec,felec,velecsum,facel,crf,krf,krf2;
 525     real             *charge;
 526     int              nvdwtype;
 527     real             rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,br,vvdwexp,sh_vdw_invrcut6;
 528     int              *vdwtype;
 529     real             *vdwparam;
 530     int              ewitab;
 531     real             ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace;
 532     real             *ewtab;
 533     real             rswitch,swV3,swV4,swV5,swF2,swF3,swF4,d,d2,sw,dsw;
 534
 535     x                = xx[0];
 536     f                = ff[0];
 537
 538     nri              = nlist->nri;
 539     iinr             = nlist->iinr;
 540     jindex           = nlist->jindex;
 541     jjnr             = nlist->jjnr;
 542     shiftidx         = nlist->shift;
 543     gid              = nlist->gid;
 544     shiftvec         = fr->shift_vec[0];
 545     fshift           = fr->fshift[0];
 546     facel            = fr->epsfac;
 547     charge           = mdatoms->chargeA;
 548     nvdwtype         = fr->ntype;
 549     vdwparam         = fr->nbfp;
 550     vdwtype          = mdatoms->typeA;
 551
 552     sh_ewald         = fr->ic->sh_ewald;
 553     ewtab            = fr->ic->tabq_coul_FDV0;
 554     ewtabscale       = fr->ic->tabq_scale;
 555     ewtabhalfspace   = 0.5/ewtabscale;
 556
 557     /* Setup water-specific parameters */
 558     inr              = nlist->iinr[0];
 559     iq1              = facel*charge[inr+1];
 560     iq2              = facel*charge[inr+2];
 561     iq3              = facel*charge[inr+3];
 562     vdwioffset0      = 3*nvdwtype*vdwtype[inr+0];
 563
 564     /* When we use explicit cutoffs the value must be identical for elec and VdW, so use elec as an arbitrary choice */
 565     rcutoff          = fr->rcoulomb;
 566     rcutoff2         = rcutoff*rcutoff;
 567
 568     rswitch          = fr->rcoulomb_switch;
 569     /* Setup switch parameters */
 570     d                = rcutoff-rswitch;
 571     swV3             = -10.0/(d*d*d);
 572     swV4             =  15.0/(d*d*d*d);
 573     swV5             =  -6.0/(d*d*d*d*d);
 574     swF2             = -30.0/(d*d*d);
 575     swF3             =  60.0/(d*d*d*d);
 576     swF4             = -30.0/(d*d*d*d*d);
 577
 578     outeriter        = 0;
 579     inneriter        = 0;
 580
 581     /* Start outer loop over neighborlists */
 582     for(iidx=0; iidx<nri; iidx++)
 583     {
 584         /* Load shift vector for this list */
 585         i_shift_offset   = DIM*shiftidx[iidx];
 586         shX              = shiftvec[i_shift_offset+XX];
 587         shY              = shiftvec[i_shift_offset+YY];
 588         shZ              = shiftvec[i_shift_offset+ZZ];
 589
 590         /* Load limits for loop over neighbors */
 591         j_index_start    = jindex[iidx];
 592         j_index_end      = jindex[iidx+1];
 593
 594         /* Get outer coordinate index */
 595         inr              = iinr[iidx];
 596         i_coord_offset   = DIM*inr;
 597
 598         /* Load i particle coords and add shift vector */
 599         ix0              = shX + x[i_coord_offset+DIM*0+XX];
 600         iy0              = shY + x[i_coord_offset+DIM*0+YY];
 601         iz0              = shZ + x[i_coord_offset+DIM*0+ZZ];
 602         ix1              = shX + x[i_coord_offset+DIM*1+XX];
 603         iy1              = shY + x[i_coord_offset+DIM*1+YY];
 604         iz1              = shZ + x[i_coord_offset+DIM*1+ZZ];
 605         ix2              = shX + x[i_coord_offset+DIM*2+XX];
 606         iy2              = shY + x[i_coord_offset+DIM*2+YY];
 607         iz2              = shZ + x[i_coord_offset+DIM*2+ZZ];
 608         ix3              = shX + x[i_coord_offset+DIM*3+XX];
 609         iy3              = shY + x[i_coord_offset+DIM*3+YY];
 610         iz3              = shZ + x[i_coord_offset+DIM*3+ZZ];
 611
 612         fix0             = 0.0;
 613         fiy0             = 0.0;
 614         fiz0             = 0.0;
 615         fix1             = 0.0;
 616         fiy1             = 0.0;
 617         fiz1             = 0.0;
 618         fix2             = 0.0;
 619         fiy2             = 0.0;
 620         fiz2             = 0.0;
 621         fix3             = 0.0;
 622         fiy3             = 0.0;
 623         fiz3             = 0.0;
 624
 625         /* Start inner kernel loop */
 626         for(jidx=j_index_start; jidx<j_index_end; jidx++)
 627         {
 628             /* Get j neighbor index, and coordinate index */
 629             jnr              = jjnr[jidx];
 630             j_coord_offset   = DIM*jnr;
 631
 632             /* load j atom coordinates */
 633             jx0              = x[j_coord_offset+DIM*0+XX];
 634             jy0              = x[j_coord_offset+DIM*0+YY];
 635             jz0              = x[j_coord_offset+DIM*0+ZZ];
 636
 637             /* Calculate displacement vector */
 638             dx00             = ix0 - jx0;
 639             dy00             = iy0 - jy0;
 640             dz00             = iz0 - jz0;
 641             dx10             = ix1 - jx0;
 642             dy10             = iy1 - jy0;
 643             dz10             = iz1 - jz0;
 644             dx20             = ix2 - jx0;
 645             dy20             = iy2 - jy0;
 646             dz20             = iz2 - jz0;
 647             dx30             = ix3 - jx0;
 648             dy30             = iy3 - jy0;
 649             dz30             = iz3 - jz0;
 650
 651             /* Calculate squared distance and things based on it */
 652             rsq00            = dx00*dx00+dy00*dy00+dz00*dz00;
 653             rsq10            = dx10*dx10+dy10*dy10+dz10*dz10;
 654             rsq20            = dx20*dx20+dy20*dy20+dz20*dz20;
 655             rsq30            = dx30*dx30+dy30*dy30+dz30*dz30;
 656
 657             rinv00           = gmx_invsqrt(rsq00);
 658             rinv10           = gmx_invsqrt(rsq10);
 659             rinv20           = gmx_invsqrt(rsq20);
 660             rinv30           = gmx_invsqrt(rsq30);
 661
 662             rinvsq00         = rinv00*rinv00;
 663             rinvsq10         = rinv10*rinv10;
 664             rinvsq20         = rinv20*rinv20;
 665             rinvsq30         = rinv30*rinv30;
 666
 667             /* Load parameters for j particles */
 668             jq0              = charge[jnr+0];
 669             vdwjidx0         = 3*vdwtype[jnr+0];
 670
 671             /**************************
 672              * CALCULATE INTERACTIONS *
 673              **************************/
 674
 675             if (rsq00<rcutoff2)
 676             {
 677
 678             r00              = rsq00*rinv00;
 679
 680             c6_00            = vdwparam[vdwioffset0+vdwjidx0];
 681             cexp1_00         = vdwparam[vdwioffset0+vdwjidx0+1];
 682             cexp2_00         = vdwparam[vdwioffset0+vdwjidx0+2];
 683
 684             /* BUCKINGHAM DISPERSION/REPULSION */
 685             rinvsix          = rinvsq00*rinvsq00*rinvsq00;
 686             vvdw6            = c6_00*rinvsix;
 687             br               = cexp2_00*r00;
 688             vvdwexp          = cexp1_00*exp(-br);
 689             vvdw             = vvdwexp - vvdw6*(1.0/6.0);
 690             fvdw             = (br*vvdwexp-vvdw6)*rinvsq00;
 691
 692             d                = r00-rswitch;
 693             d                = (d>0.0) ? d : 0.0;
 694             d2               = d*d;
 695             sw               = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
 696
 697             dsw              = d2*(swF2+d*(swF3+d*swF4));
 698
 699             /* Evaluate switch function */
 700             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 701             fvdw             = fvdw*sw - rinv00*vvdw*dsw;
 702
 703             fscal            = fvdw;
 704
 705             /* Calculate temporary vectorial force */
 706             tx               = fscal*dx00;
 707             ty               = fscal*dy00;
 708             tz               = fscal*dz00;
 709
 710             /* Update vectorial force */
 711             fix0            += tx;
 712             fiy0            += ty;
 713             fiz0            += tz;
 714             f[j_coord_offset+DIM*0+XX] -= tx;
 715             f[j_coord_offset+DIM*0+YY] -= ty;
 716             f[j_coord_offset+DIM*0+ZZ] -= tz;
 717
 718             }
 719
 720             /**************************
 721              * CALCULATE INTERACTIONS *
 722              **************************/
 723
 724             if (rsq10<rcutoff2)
 725             {
 726
 727             r10              = rsq10*rinv10;
 728
 729             qq10             = iq1*jq0;
 730
 731             /* EWALD ELECTROSTATICS */
 732
 733             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 734             ewrt             = r10*ewtabscale;
 735             ewitab           = ewrt;
 736             eweps            = ewrt-ewitab;
 737             ewitab           = 4*ewitab;
 738             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 739             velec            = qq10*(rinv10-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 740             felec            = qq10*rinv10*(rinvsq10-felec);
 741
 742             d                = r10-rswitch;
 743             d                = (d>0.0) ? d : 0.0;
 744             d2               = d*d;
 745             sw               = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
 746
 747             dsw              = d2*(swF2+d*(swF3+d*swF4));
 748
 749             /* Evaluate switch function */
 750             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 751             felec            = felec*sw - rinv10*velec*dsw;
 752
 753             fscal            = felec;
 754
 755             /* Calculate temporary vectorial force */
 756             tx               = fscal*dx10;
 757             ty               = fscal*dy10;
 758             tz               = fscal*dz10;
 759
 760             /* Update vectorial force */
 761             fix1            += tx;
 762             fiy1            += ty;
 763             fiz1            += tz;
 764             f[j_coord_offset+DIM*0+XX] -= tx;
 765             f[j_coord_offset+DIM*0+YY] -= ty;
 766             f[j_coord_offset+DIM*0+ZZ] -= tz;
 767
 768             }
 769
 770             /**************************
 771              * CALCULATE INTERACTIONS *
 772              **************************/
 773
 774             if (rsq20<rcutoff2)
 775             {
 776
 777             r20              = rsq20*rinv20;
 778
 779             qq20             = iq2*jq0;
 780
 781             /* EWALD ELECTROSTATICS */
 782
 783             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 784             ewrt             = r20*ewtabscale;
 785             ewitab           = ewrt;
 786             eweps            = ewrt-ewitab;
 787             ewitab           = 4*ewitab;
 788             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 789             velec            = qq20*(rinv20-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 790             felec            = qq20*rinv20*(rinvsq20-felec);
 791
 792             d                = r20-rswitch;
 793             d                = (d>0.0) ? d : 0.0;
 794             d2               = d*d;
 795             sw               = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
 796
 797             dsw              = d2*(swF2+d*(swF3+d*swF4));
 798
 799             /* Evaluate switch function */
 800             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 801             felec            = felec*sw - rinv20*velec*dsw;
 802
 803             fscal            = felec;
 804
 805             /* Calculate temporary vectorial force */
 806             tx               = fscal*dx20;
 807             ty               = fscal*dy20;
 808             tz               = fscal*dz20;
 809
 810             /* Update vectorial force */
 811             fix2            += tx;
 812             fiy2            += ty;
 813             fiz2            += tz;
 814             f[j_coord_offset+DIM*0+XX] -= tx;
 815             f[j_coord_offset+DIM*0+YY] -= ty;
 816             f[j_coord_offset+DIM*0+ZZ] -= tz;
 817
 818             }
 819
 820             /**************************
 821              * CALCULATE INTERACTIONS *
 822              **************************/
 823
 824             if (rsq30<rcutoff2)
 825             {
 826
 827             r30              = rsq30*rinv30;
 828
 829             qq30             = iq3*jq0;
 830
 831             /* EWALD ELECTROSTATICS */
 832
 833             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 834             ewrt             = r30*ewtabscale;
 835             ewitab           = ewrt;
 836             eweps            = ewrt-ewitab;
 837             ewitab           = 4*ewitab;
 838             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 839             velec            = qq30*(rinv30-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 840             felec            = qq30*rinv30*(rinvsq30-felec);
 841
 842             d                = r30-rswitch;
 843             d                = (d>0.0) ? d : 0.0;
 844             d2               = d*d;
 845             sw               = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
 846
 847             dsw              = d2*(swF2+d*(swF3+d*swF4));
 848
 849             /* Evaluate switch function */
 850             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 851             felec            = felec*sw - rinv30*velec*dsw;
 852
 853             fscal            = felec;
 854
 855             /* Calculate temporary vectorial force */
 856             tx               = fscal*dx30;
 857             ty               = fscal*dy30;
 858             tz               = fscal*dz30;
 859
 860             /* Update vectorial force */
 861             fix3            += tx;
 862             fiy3            += ty;
 863             fiz3            += tz;
 864             f[j_coord_offset+DIM*0+XX] -= tx;
 865             f[j_coord_offset+DIM*0+YY] -= ty;
 866             f[j_coord_offset+DIM*0+ZZ] -= tz;
 867
 868             }
 869
 870             /* Inner loop uses 248 flops */
 871         }
 872         /* End of innermost loop */
 873
 874         tx = ty = tz = 0;
 875         f[i_coord_offset+DIM*0+XX] += fix0;
 876         f[i_coord_offset+DIM*0+YY] += fiy0;
 877         f[i_coord_offset+DIM*0+ZZ] += fiz0;
 878         tx                         += fix0;
 879         ty                         += fiy0;
 880         tz                         += fiz0;
 881         f[i_coord_offset+DIM*1+XX] += fix1;
 882         f[i_coord_offset+DIM*1+YY] += fiy1;
 883         f[i_coord_offset+DIM*1+ZZ] += fiz1;
 884         tx                         += fix1;
 885         ty                         += fiy1;
 886         tz                         += fiz1;
 887         f[i_coord_offset+DIM*2+XX] += fix2;
 888         f[i_coord_offset+DIM*2+YY] += fiy2;
 889         f[i_coord_offset+DIM*2+ZZ] += fiz2;
 890         tx                         += fix2;
 891         ty                         += fiy2;
 892         tz                         += fiz2;
 893         f[i_coord_offset+DIM*3+XX] += fix3;
 894         f[i_coord_offset+DIM*3+YY] += fiy3;
 895         f[i_coord_offset+DIM*3+ZZ] += fiz3;
 896         tx                         += fix3;
 897         ty                         += fiy3;
 898         tz                         += fiz3;
 899         fshift[i_shift_offset+XX]  += tx;
 900         fshift[i_shift_offset+YY]  += ty;
 901         fshift[i_shift_offset+ZZ]  += tz;
 902
 903         /* Increment number of inner iterations */
 904         inneriter                  += j_index_end - j_index_start;
 905
 906         /* Outer loop uses 39 flops */
 907     }
 908
 909     /* Increment number of outer iterations */
 910     outeriter        += nri;
 911
 912     /* Update outer/inner flops */
 913
 914     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W4_F,outeriter*39 + inneriter*248);
 915 }