src/gmxlib/nonbonded/nb_kernel_c/nb_kernel_ElecEwSw_VdwLJSw_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_VdwLJSw_GeomW4P1_VF_c
  35  * Electrostatics interaction: Ewald
  36  * VdW interaction:            LennardJones
  37  * Geometry:                   Water4-Particle
  38  * Calculate force/pot:        PotentialAndForce
  39  */
  40 void
  41 nb_kernel_ElecEwSw_VdwLJSw_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      = 2*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         = 2*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             c12_00           = vdwparam[vdwioffset0+vdwjidx0+1];
 232
 233             /* LENNARD-JONES DISPERSION/REPULSION */
 234
 235             rinvsix          = rinvsq00*rinvsq00*rinvsq00;
 236             vvdw6            = c6_00*rinvsix;
 237             vvdw12           = c12_00*rinvsix*rinvsix;
 238             vvdw             = vvdw12*(1.0/12.0) - vvdw6*(1.0/6.0);
 239             fvdw             = (vvdw12-vvdw6)*rinvsq00;
 240
 241             d                = r00-rswitch;
 242             d                = (d>0.0) ? d : 0.0;
 243             d2               = d*d;
 244             sw               = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
 245
 246             dsw              = d2*(swF2+d*(swF3+d*swF4));
 247
 248             /* Evaluate switch function */
 249             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 250             fvdw             = fvdw*sw - rinv00*vvdw*dsw;
 251             vvdw            *= sw;
 252
 253             /* Update potential sums from outer loop */
 254             vvdwsum         += vvdw;
 255
 256             fscal            = fvdw;
 257
 258             /* Calculate temporary vectorial force */
 259             tx               = fscal*dx00;
 260             ty               = fscal*dy00;
 261             tz               = fscal*dz00;
 262
 263             /* Update vectorial force */
 264             fix0            += tx;
 265             fiy0            += ty;
 266             fiz0            += tz;
 267             f[j_coord_offset+DIM*0+XX] -= tx;
 268             f[j_coord_offset+DIM*0+YY] -= ty;
 269             f[j_coord_offset+DIM*0+ZZ] -= tz;
 270
 271             }
 272
 273             /**************************
 274              * CALCULATE INTERACTIONS *
 275              **************************/
 276
 277             if (rsq10<rcutoff2)
 278             {
 279
 280             r10              = rsq10*rinv10;
 281
 282             qq10             = iq1*jq0;
 283
 284             /* EWALD ELECTROSTATICS */
 285
 286             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 287             ewrt             = r10*ewtabscale;
 288             ewitab           = ewrt;
 289             eweps            = ewrt-ewitab;
 290             ewitab           = 4*ewitab;
 291             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 292             velec            = qq10*(rinv10-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 293             felec            = qq10*rinv10*(rinvsq10-felec);
 294
 295             d                = r10-rswitch;
 296             d                = (d>0.0) ? d : 0.0;
 297             d2               = d*d;
 298             sw               = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
 299
 300             dsw              = d2*(swF2+d*(swF3+d*swF4));
 301
 302             /* Evaluate switch function */
 303             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 304             felec            = felec*sw - rinv10*velec*dsw;
 305             velec           *= sw;
 306
 307             /* Update potential sums from outer loop */
 308             velecsum        += velec;
 309
 310             fscal            = felec;
 311
 312             /* Calculate temporary vectorial force */
 313             tx               = fscal*dx10;
 314             ty               = fscal*dy10;
 315             tz               = fscal*dz10;
 316
 317             /* Update vectorial force */
 318             fix1            += tx;
 319             fiy1            += ty;
 320             fiz1            += tz;
 321             f[j_coord_offset+DIM*0+XX] -= tx;
 322             f[j_coord_offset+DIM*0+YY] -= ty;
 323             f[j_coord_offset+DIM*0+ZZ] -= tz;
 324
 325             }
 326
 327             /**************************
 328              * CALCULATE INTERACTIONS *
 329              **************************/
 330
 331             if (rsq20<rcutoff2)
 332             {
 333
 334             r20              = rsq20*rinv20;
 335
 336             qq20             = iq2*jq0;
 337
 338             /* EWALD ELECTROSTATICS */
 339
 340             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 341             ewrt             = r20*ewtabscale;
 342             ewitab           = ewrt;
 343             eweps            = ewrt-ewitab;
 344             ewitab           = 4*ewitab;
 345             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 346             velec            = qq20*(rinv20-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 347             felec            = qq20*rinv20*(rinvsq20-felec);
 348
 349             d                = r20-rswitch;
 350             d                = (d>0.0) ? d : 0.0;
 351             d2               = d*d;
 352             sw               = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
 353
 354             dsw              = d2*(swF2+d*(swF3+d*swF4));
 355
 356             /* Evaluate switch function */
 357             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 358             felec            = felec*sw - rinv20*velec*dsw;
 359             velec           *= sw;
 360
 361             /* Update potential sums from outer loop */
 362             velecsum        += velec;
 363
 364             fscal            = felec;
 365
 366             /* Calculate temporary vectorial force */
 367             tx               = fscal*dx20;
 368             ty               = fscal*dy20;
 369             tz               = fscal*dz20;
 370
 371             /* Update vectorial force */
 372             fix2            += tx;
 373             fiy2            += ty;
 374             fiz2            += tz;
 375             f[j_coord_offset+DIM*0+XX] -= tx;
 376             f[j_coord_offset+DIM*0+YY] -= ty;
 377             f[j_coord_offset+DIM*0+ZZ] -= tz;
 378
 379             }
 380
 381             /**************************
 382              * CALCULATE INTERACTIONS *
 383              **************************/
 384
 385             if (rsq30<rcutoff2)
 386             {
 387
 388             r30              = rsq30*rinv30;
 389
 390             qq30             = iq3*jq0;
 391
 392             /* EWALD ELECTROSTATICS */
 393
 394             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 395             ewrt             = r30*ewtabscale;
 396             ewitab           = ewrt;
 397             eweps            = ewrt-ewitab;
 398             ewitab           = 4*ewitab;
 399             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 400             velec            = qq30*(rinv30-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 401             felec            = qq30*rinv30*(rinvsq30-felec);
 402
 403             d                = r30-rswitch;
 404             d                = (d>0.0) ? d : 0.0;
 405             d2               = d*d;
 406             sw               = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
 407
 408             dsw              = d2*(swF2+d*(swF3+d*swF4));
 409
 410             /* Evaluate switch function */
 411             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 412             felec            = felec*sw - rinv30*velec*dsw;
 413             velec           *= sw;
 414
 415             /* Update potential sums from outer loop */
 416             velecsum        += velec;
 417
 418             fscal            = felec;
 419
 420             /* Calculate temporary vectorial force */
 421             tx               = fscal*dx30;
 422             ty               = fscal*dy30;
 423             tz               = fscal*dz30;
 424
 425             /* Update vectorial force */
 426             fix3            += tx;
 427             fiy3            += ty;
 428             fiz3            += tz;
 429             f[j_coord_offset+DIM*0+XX] -= tx;
 430             f[j_coord_offset+DIM*0+YY] -= ty;
 431             f[j_coord_offset+DIM*0+ZZ] -= tz;
 432
 433             }
 434
 435             /* Inner loop uses 230 flops */
 436         }
 437         /* End of innermost loop */
 438
 439         tx = ty = tz = 0;
 440         f[i_coord_offset+DIM*0+XX] += fix0;
 441         f[i_coord_offset+DIM*0+YY] += fiy0;
 442         f[i_coord_offset+DIM*0+ZZ] += fiz0;
 443         tx                         += fix0;
 444         ty                         += fiy0;
 445         tz                         += fiz0;
 446         f[i_coord_offset+DIM*1+XX] += fix1;
 447         f[i_coord_offset+DIM*1+YY] += fiy1;
 448         f[i_coord_offset+DIM*1+ZZ] += fiz1;
 449         tx                         += fix1;
 450         ty                         += fiy1;
 451         tz                         += fiz1;
 452         f[i_coord_offset+DIM*2+XX] += fix2;
 453         f[i_coord_offset+DIM*2+YY] += fiy2;
 454         f[i_coord_offset+DIM*2+ZZ] += fiz2;
 455         tx                         += fix2;
 456         ty                         += fiy2;
 457         tz                         += fiz2;
 458         f[i_coord_offset+DIM*3+XX] += fix3;
 459         f[i_coord_offset+DIM*3+YY] += fiy3;
 460         f[i_coord_offset+DIM*3+ZZ] += fiz3;
 461         tx                         += fix3;
 462         ty                         += fiy3;
 463         tz                         += fiz3;
 464         fshift[i_shift_offset+XX]  += tx;
 465         fshift[i_shift_offset+YY]  += ty;
 466         fshift[i_shift_offset+ZZ]  += tz;
 467
 468         ggid                        = gid[iidx];
 469         /* Update potential energies */
 470         kernel_data->energygrp_elec[ggid] += velecsum;
 471         kernel_data->energygrp_vdw[ggid] += vvdwsum;
 472
 473         /* Increment number of inner iterations */
 474         inneriter                  += j_index_end - j_index_start;
 475
 476         /* Outer loop uses 41 flops */
 477     }
 478
 479     /* Increment number of outer iterations */
 480     outeriter        += nri;
 481
 482     /* Update outer/inner flops */
 483
 484     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W4_VF,outeriter*41 + inneriter*230);
 485 }
 486 /*
 487  * Gromacs nonbonded kernel:   nb_kernel_ElecEwSw_VdwLJSw_GeomW4P1_F_c
 488  * Electrostatics interaction: Ewald
 489  * VdW interaction:            LennardJones
 490  * Geometry:                   Water4-Particle
 491  * Calculate force/pot:        Force
 492  */
 493 void
 494 nb_kernel_ElecEwSw_VdwLJSw_GeomW4P1_F_c
 495                     (t_nblist * gmx_restrict                nlist,
 496                      rvec * gmx_restrict                    xx,
 497                      rvec * gmx_restrict                    ff,
 498                      t_forcerec * gmx_restrict              fr,
 499                      t_mdatoms * gmx_restrict               mdatoms,
 500                      nb_kernel_data_t * gmx_restrict        kernel_data,
 501                      t_nrnb * gmx_restrict                  nrnb)
 502 {
 503     int              i_shift_offset,i_coord_offset,j_coord_offset;
 504     int              j_index_start,j_index_end;
 505     int              nri,inr,ggid,iidx,jidx,jnr,outeriter,inneriter;
 506     real             shX,shY,shZ,tx,ty,tz,fscal,rcutoff,rcutoff2;
 507     int              *iinr,*jindex,*jjnr,*shiftidx,*gid;
 508     real             *shiftvec,*fshift,*x,*f;
 509     int              vdwioffset0;
 510     real             ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
 511     int              vdwioffset1;
 512     real             ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
 513     int              vdwioffset2;
 514     real             ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
 515     int              vdwioffset3;
 516     real             ix3,iy3,iz3,fix3,fiy3,fiz3,iq3,isai3;
 517     int              vdwjidx0;
 518     real             jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
 519     real             dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00,cexp1_00,cexp2_00;
 520     real             dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10,cexp1_10,cexp2_10;
 521     real             dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20,cexp1_20,cexp2_20;
 522     real             dx30,dy30,dz30,rsq30,rinv30,rinvsq30,r30,qq30,c6_30,c12_30,cexp1_30,cexp2_30;
 523     real             velec,felec,velecsum,facel,crf,krf,krf2;
 524     real             *charge;
 525     int              nvdwtype;
 526     real             rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,br,vvdwexp,sh_vdw_invrcut6;
 527     int              *vdwtype;
 528     real             *vdwparam;
 529     int              ewitab;
 530     real             ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace;
 531     real             *ewtab;
 532     real             rswitch,swV3,swV4,swV5,swF2,swF3,swF4,d,d2,sw,dsw;
 533
 534     x                = xx[0];
 535     f                = ff[0];
 536
 537     nri              = nlist->nri;
 538     iinr             = nlist->iinr;
 539     jindex           = nlist->jindex;
 540     jjnr             = nlist->jjnr;
 541     shiftidx         = nlist->shift;
 542     gid              = nlist->gid;
 543     shiftvec         = fr->shift_vec[0];
 544     fshift           = fr->fshift[0];
 545     facel            = fr->epsfac;
 546     charge           = mdatoms->chargeA;
 547     nvdwtype         = fr->ntype;
 548     vdwparam         = fr->nbfp;
 549     vdwtype          = mdatoms->typeA;
 550
 551     sh_ewald         = fr->ic->sh_ewald;
 552     ewtab            = fr->ic->tabq_coul_FDV0;
 553     ewtabscale       = fr->ic->tabq_scale;
 554     ewtabhalfspace   = 0.5/ewtabscale;
 555
 556     /* Setup water-specific parameters */
 557     inr              = nlist->iinr[0];
 558     iq1              = facel*charge[inr+1];
 559     iq2              = facel*charge[inr+2];
 560     iq3              = facel*charge[inr+3];
 561     vdwioffset0      = 2*nvdwtype*vdwtype[inr+0];
 562
 563     /* When we use explicit cutoffs the value must be identical for elec and VdW, so use elec as an arbitrary choice */
 564     rcutoff          = fr->rcoulomb;
 565     rcutoff2         = rcutoff*rcutoff;
 566
 567     rswitch          = fr->rcoulomb_switch;
 568     /* Setup switch parameters */
 569     d                = rcutoff-rswitch;
 570     swV3             = -10.0/(d*d*d);
 571     swV4             =  15.0/(d*d*d*d);
 572     swV5             =  -6.0/(d*d*d*d*d);
 573     swF2             = -30.0/(d*d*d);
 574     swF3             =  60.0/(d*d*d*d);
 575     swF4             = -30.0/(d*d*d*d*d);
 576
 577     outeriter        = 0;
 578     inneriter        = 0;
 579
 580     /* Start outer loop over neighborlists */
 581     for(iidx=0; iidx<nri; iidx++)
 582     {
 583         /* Load shift vector for this list */
 584         i_shift_offset   = DIM*shiftidx[iidx];
 585         shX              = shiftvec[i_shift_offset+XX];
 586         shY              = shiftvec[i_shift_offset+YY];
 587         shZ              = shiftvec[i_shift_offset+ZZ];
 588
 589         /* Load limits for loop over neighbors */
 590         j_index_start    = jindex[iidx];
 591         j_index_end      = jindex[iidx+1];
 592
 593         /* Get outer coordinate index */
 594         inr              = iinr[iidx];
 595         i_coord_offset   = DIM*inr;
 596
 597         /* Load i particle coords and add shift vector */
 598         ix0              = shX + x[i_coord_offset+DIM*0+XX];
 599         iy0              = shY + x[i_coord_offset+DIM*0+YY];
 600         iz0              = shZ + x[i_coord_offset+DIM*0+ZZ];
 601         ix1              = shX + x[i_coord_offset+DIM*1+XX];
 602         iy1              = shY + x[i_coord_offset+DIM*1+YY];
 603         iz1              = shZ + x[i_coord_offset+DIM*1+ZZ];
 604         ix2              = shX + x[i_coord_offset+DIM*2+XX];
 605         iy2              = shY + x[i_coord_offset+DIM*2+YY];
 606         iz2              = shZ + x[i_coord_offset+DIM*2+ZZ];
 607         ix3              = shX + x[i_coord_offset+DIM*3+XX];
 608         iy3              = shY + x[i_coord_offset+DIM*3+YY];
 609         iz3              = shZ + x[i_coord_offset+DIM*3+ZZ];
 610
 611         fix0             = 0.0;
 612         fiy0             = 0.0;
 613         fiz0             = 0.0;
 614         fix1             = 0.0;
 615         fiy1             = 0.0;
 616         fiz1             = 0.0;
 617         fix2             = 0.0;
 618         fiy2             = 0.0;
 619         fiz2             = 0.0;
 620         fix3             = 0.0;
 621         fiy3             = 0.0;
 622         fiz3             = 0.0;
 623
 624         /* Start inner kernel loop */
 625         for(jidx=j_index_start; jidx<j_index_end; jidx++)
 626         {
 627             /* Get j neighbor index, and coordinate index */
 628             jnr              = jjnr[jidx];
 629             j_coord_offset   = DIM*jnr;
 630
 631             /* load j atom coordinates */
 632             jx0              = x[j_coord_offset+DIM*0+XX];
 633             jy0              = x[j_coord_offset+DIM*0+YY];
 634             jz0              = x[j_coord_offset+DIM*0+ZZ];
 635
 636             /* Calculate displacement vector */
 637             dx00             = ix0 - jx0;
 638             dy00             = iy0 - jy0;
 639             dz00             = iz0 - jz0;
 640             dx10             = ix1 - jx0;
 641             dy10             = iy1 - jy0;
 642             dz10             = iz1 - jz0;
 643             dx20             = ix2 - jx0;
 644             dy20             = iy2 - jy0;
 645             dz20             = iz2 - jz0;
 646             dx30             = ix3 - jx0;
 647             dy30             = iy3 - jy0;
 648             dz30             = iz3 - jz0;
 649
 650             /* Calculate squared distance and things based on it */
 651             rsq00            = dx00*dx00+dy00*dy00+dz00*dz00;
 652             rsq10            = dx10*dx10+dy10*dy10+dz10*dz10;
 653             rsq20            = dx20*dx20+dy20*dy20+dz20*dz20;
 654             rsq30            = dx30*dx30+dy30*dy30+dz30*dz30;
 655
 656             rinv00           = gmx_invsqrt(rsq00);
 657             rinv10           = gmx_invsqrt(rsq10);
 658             rinv20           = gmx_invsqrt(rsq20);
 659             rinv30           = gmx_invsqrt(rsq30);
 660
 661             rinvsq00         = rinv00*rinv00;
 662             rinvsq10         = rinv10*rinv10;
 663             rinvsq20         = rinv20*rinv20;
 664             rinvsq30         = rinv30*rinv30;
 665
 666             /* Load parameters for j particles */
 667             jq0              = charge[jnr+0];
 668             vdwjidx0         = 2*vdwtype[jnr+0];
 669
 670             /**************************
 671              * CALCULATE INTERACTIONS *
 672              **************************/
 673
 674             if (rsq00<rcutoff2)
 675             {
 676
 677             r00              = rsq00*rinv00;
 678
 679             c6_00            = vdwparam[vdwioffset0+vdwjidx0];
 680             c12_00           = vdwparam[vdwioffset0+vdwjidx0+1];
 681
 682             /* LENNARD-JONES DISPERSION/REPULSION */
 683
 684             rinvsix          = rinvsq00*rinvsq00*rinvsq00;
 685             vvdw6            = c6_00*rinvsix;
 686             vvdw12           = c12_00*rinvsix*rinvsix;
 687             vvdw             = vvdw12*(1.0/12.0) - vvdw6*(1.0/6.0);
 688             fvdw             = (vvdw12-vvdw6)*rinvsq00;
 689
 690             d                = r00-rswitch;
 691             d                = (d>0.0) ? d : 0.0;
 692             d2               = d*d;
 693             sw               = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
 694
 695             dsw              = d2*(swF2+d*(swF3+d*swF4));
 696
 697             /* Evaluate switch function */
 698             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 699             fvdw             = fvdw*sw - rinv00*vvdw*dsw;
 700
 701             fscal            = fvdw;
 702
 703             /* Calculate temporary vectorial force */
 704             tx               = fscal*dx00;
 705             ty               = fscal*dy00;
 706             tz               = fscal*dz00;
 707
 708             /* Update vectorial force */
 709             fix0            += tx;
 710             fiy0            += ty;
 711             fiz0            += tz;
 712             f[j_coord_offset+DIM*0+XX] -= tx;
 713             f[j_coord_offset+DIM*0+YY] -= ty;
 714             f[j_coord_offset+DIM*0+ZZ] -= tz;
 715
 716             }
 717
 718             /**************************
 719              * CALCULATE INTERACTIONS *
 720              **************************/
 721
 722             if (rsq10<rcutoff2)
 723             {
 724
 725             r10              = rsq10*rinv10;
 726
 727             qq10             = iq1*jq0;
 728
 729             /* EWALD ELECTROSTATICS */
 730
 731             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 732             ewrt             = r10*ewtabscale;
 733             ewitab           = ewrt;
 734             eweps            = ewrt-ewitab;
 735             ewitab           = 4*ewitab;
 736             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 737             velec            = qq10*(rinv10-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 738             felec            = qq10*rinv10*(rinvsq10-felec);
 739
 740             d                = r10-rswitch;
 741             d                = (d>0.0) ? d : 0.0;
 742             d2               = d*d;
 743             sw               = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
 744
 745             dsw              = d2*(swF2+d*(swF3+d*swF4));
 746
 747             /* Evaluate switch function */
 748             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 749             felec            = felec*sw - rinv10*velec*dsw;
 750
 751             fscal            = felec;
 752
 753             /* Calculate temporary vectorial force */
 754             tx               = fscal*dx10;
 755             ty               = fscal*dy10;
 756             tz               = fscal*dz10;
 757
 758             /* Update vectorial force */
 759             fix1            += tx;
 760             fiy1            += ty;
 761             fiz1            += tz;
 762             f[j_coord_offset+DIM*0+XX] -= tx;
 763             f[j_coord_offset+DIM*0+YY] -= ty;
 764             f[j_coord_offset+DIM*0+ZZ] -= tz;
 765
 766             }
 767
 768             /**************************
 769              * CALCULATE INTERACTIONS *
 770              **************************/
 771
 772             if (rsq20<rcutoff2)
 773             {
 774
 775             r20              = rsq20*rinv20;
 776
 777             qq20             = iq2*jq0;
 778
 779             /* EWALD ELECTROSTATICS */
 780
 781             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 782             ewrt             = r20*ewtabscale;
 783             ewitab           = ewrt;
 784             eweps            = ewrt-ewitab;
 785             ewitab           = 4*ewitab;
 786             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 787             velec            = qq20*(rinv20-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 788             felec            = qq20*rinv20*(rinvsq20-felec);
 789
 790             d                = r20-rswitch;
 791             d                = (d>0.0) ? d : 0.0;
 792             d2               = d*d;
 793             sw               = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
 794
 795             dsw              = d2*(swF2+d*(swF3+d*swF4));
 796
 797             /* Evaluate switch function */
 798             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 799             felec            = felec*sw - rinv20*velec*dsw;
 800
 801             fscal            = felec;
 802
 803             /* Calculate temporary vectorial force */
 804             tx               = fscal*dx20;
 805             ty               = fscal*dy20;
 806             tz               = fscal*dz20;
 807
 808             /* Update vectorial force */
 809             fix2            += tx;
 810             fiy2            += ty;
 811             fiz2            += tz;
 812             f[j_coord_offset+DIM*0+XX] -= tx;
 813             f[j_coord_offset+DIM*0+YY] -= ty;
 814             f[j_coord_offset+DIM*0+ZZ] -= tz;
 815
 816             }
 817
 818             /**************************
 819              * CALCULATE INTERACTIONS *
 820              **************************/
 821
 822             if (rsq30<rcutoff2)
 823             {
 824
 825             r30              = rsq30*rinv30;
 826
 827             qq30             = iq3*jq0;
 828
 829             /* EWALD ELECTROSTATICS */
 830
 831             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 832             ewrt             = r30*ewtabscale;
 833             ewitab           = ewrt;
 834             eweps            = ewrt-ewitab;
 835             ewitab           = 4*ewitab;
 836             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 837             velec            = qq30*(rinv30-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 838             felec            = qq30*rinv30*(rinvsq30-felec);
 839
 840             d                = r30-rswitch;
 841             d                = (d>0.0) ? d : 0.0;
 842             d2               = d*d;
 843             sw               = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
 844
 845             dsw              = d2*(swF2+d*(swF3+d*swF4));
 846
 847             /* Evaluate switch function */
 848             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 849             felec            = felec*sw - rinv30*velec*dsw;
 850
 851             fscal            = felec;
 852
 853             /* Calculate temporary vectorial force */
 854             tx               = fscal*dx30;
 855             ty               = fscal*dy30;
 856             tz               = fscal*dz30;
 857
 858             /* Update vectorial force */
 859             fix3            += tx;
 860             fiy3            += ty;
 861             fiz3            += tz;
 862             f[j_coord_offset+DIM*0+XX] -= tx;
 863             f[j_coord_offset+DIM*0+YY] -= ty;
 864             f[j_coord_offset+DIM*0+ZZ] -= tz;
 865
 866             }
 867
 868             /* Inner loop uses 222 flops */
 869         }
 870         /* End of innermost loop */
 871
 872         tx = ty = tz = 0;
 873         f[i_coord_offset+DIM*0+XX] += fix0;
 874         f[i_coord_offset+DIM*0+YY] += fiy0;
 875         f[i_coord_offset+DIM*0+ZZ] += fiz0;
 876         tx                         += fix0;
 877         ty                         += fiy0;
 878         tz                         += fiz0;
 879         f[i_coord_offset+DIM*1+XX] += fix1;
 880         f[i_coord_offset+DIM*1+YY] += fiy1;
 881         f[i_coord_offset+DIM*1+ZZ] += fiz1;
 882         tx                         += fix1;
 883         ty                         += fiy1;
 884         tz                         += fiz1;
 885         f[i_coord_offset+DIM*2+XX] += fix2;
 886         f[i_coord_offset+DIM*2+YY] += fiy2;
 887         f[i_coord_offset+DIM*2+ZZ] += fiz2;
 888         tx                         += fix2;
 889         ty                         += fiy2;
 890         tz                         += fiz2;
 891         f[i_coord_offset+DIM*3+XX] += fix3;
 892         f[i_coord_offset+DIM*3+YY] += fiy3;
 893         f[i_coord_offset+DIM*3+ZZ] += fiz3;
 894         tx                         += fix3;
 895         ty                         += fiy3;
 896         tz                         += fiz3;
 897         fshift[i_shift_offset+XX]  += tx;
 898         fshift[i_shift_offset+YY]  += ty;
 899         fshift[i_shift_offset+ZZ]  += tz;
 900
 901         /* Increment number of inner iterations */
 902         inneriter                  += j_index_end - j_index_start;
 903
 904         /* Outer loop uses 39 flops */
 905     }
 906
 907     /* Increment number of outer iterations */
 908     outeriter        += nri;
 909
 910     /* Update outer/inner flops */
 911
 912     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W4_F,outeriter*39 + inneriter*222);
 913 }