src/gromacs/gmxlib/nonbonded/nb_kernel_c/nb_kernel_ElecEwSh_VdwNone_GeomW4P1_c.c

   1 /*
   2  * This file is part of the GROMACS molecular simulation package.
   3  *
   4  * Copyright (c) 2012,2013,2014, by the GROMACS development team, led by
   5  * Mark Abraham, David van der Spoel, Berk Hess, and Erik Lindahl,
   6  * and including many others, as listed in the AUTHORS file in the
   7  * top-level source directory and at http://www.gromacs.org.
   8  *
   9  * GROMACS is free software; you can redistribute it and/or
  10  * modify it under the terms of the GNU Lesser General Public License
  11  * as published by the Free Software Foundation; either version 2.1
  12  * of the License, or (at your option) any later version.
  13  *
  14  * GROMACS is distributed in the hope that it will be useful,
  15  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  16  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  17  * Lesser General Public License for more details.
  18  *
  19  * You should have received a copy of the GNU Lesser General Public
  20  * License along with GROMACS; if not, see
  21  * http://www.gnu.org/licenses, or write to the Free Software Foundation,
  22  * Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02110-1301  USA.
  23  *
  24  * If you want to redistribute modifications to GROMACS, please
  25  * consider that scientific software is very special. Version
  26  * control is crucial - bugs must be traceable. We will be happy to
  27  * consider code for inclusion in the official distribution, but
  28  * derived work must not be called official GROMACS. Details are found
  29  * in the README & COPYING files - if they are missing, get the
  30  * official version at http://www.gromacs.org.
  31  *
  32  * To help us fund GROMACS development, we humbly ask that you cite
  33  * the research papers on the package. Check out http://www.gromacs.org.
  34  */
  35 /*
  36  * Note: this file was generated by the GROMACS c kernel generator.
  37  */
  38 #include "config.h"
  39
  40 #include <math.h>
  41
  42 #include "../nb_kernel.h"
  43 #include "types/simple.h"
  44 #include "gromacs/math/vec.h"
  45 #include "nrnb.h"
  46
  47 /*
  48  * Gromacs nonbonded kernel:   nb_kernel_ElecEwSh_VdwNone_GeomW4P1_VF_c
  49  * Electrostatics interaction: Ewald
  50  * VdW interaction:            None
  51  * Geometry:                   Water4-Particle
  52  * Calculate force/pot:        PotentialAndForce
  53  */
  54 void
  55 nb_kernel_ElecEwSh_VdwNone_GeomW4P1_VF_c
  56                     (t_nblist                    * gmx_restrict       nlist,
  57                      rvec                        * gmx_restrict          xx,
  58                      rvec                        * gmx_restrict          ff,
  59                      t_forcerec                  * gmx_restrict          fr,
  60                      t_mdatoms                   * gmx_restrict     mdatoms,
  61                      nb_kernel_data_t gmx_unused * gmx_restrict kernel_data,
  62                      t_nrnb                      * gmx_restrict        nrnb)
  63 {
  64     int              i_shift_offset,i_coord_offset,j_coord_offset;
  65     int              j_index_start,j_index_end;
  66     int              nri,inr,ggid,iidx,jidx,jnr,outeriter,inneriter;
  67     real             shX,shY,shZ,tx,ty,tz,fscal,rcutoff,rcutoff2;
  68     int              *iinr,*jindex,*jjnr,*shiftidx,*gid;
  69     real             *shiftvec,*fshift,*x,*f;
  70     int              vdwioffset1;
  71     real             ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
  72     int              vdwioffset2;
  73     real             ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
  74     int              vdwioffset3;
  75     real             ix3,iy3,iz3,fix3,fiy3,fiz3,iq3,isai3;
  76     int              vdwjidx0;
  77     real             jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
  78     real             dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10,cexp1_10,cexp2_10;
  79     real             dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20,cexp1_20,cexp2_20;
  80     real             dx30,dy30,dz30,rsq30,rinv30,rinvsq30,r30,qq30,c6_30,c12_30,cexp1_30,cexp2_30;
  81     real             velec,felec,velecsum,facel,crf,krf,krf2;
  82     real             *charge;
  83     int              ewitab;
  84     real             ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace;
  85     real             *ewtab;
  86
  87     x                = xx[0];
  88     f                = ff[0];
  89
  90     nri              = nlist->nri;
  91     iinr             = nlist->iinr;
  92     jindex           = nlist->jindex;
  93     jjnr             = nlist->jjnr;
  94     shiftidx         = nlist->shift;
  95     gid              = nlist->gid;
  96     shiftvec         = fr->shift_vec[0];
  97     fshift           = fr->fshift[0];
  98     facel            = fr->epsfac;
  99     charge           = mdatoms->chargeA;
 100
 101     sh_ewald         = fr->ic->sh_ewald;
 102     ewtab            = fr->ic->tabq_coul_FDV0;
 103     ewtabscale       = fr->ic->tabq_scale;
 104     ewtabhalfspace   = 0.5/ewtabscale;
 105
 106     /* Setup water-specific parameters */
 107     inr              = nlist->iinr[0];
 108     iq1              = facel*charge[inr+1];
 109     iq2              = facel*charge[inr+2];
 110     iq3              = facel*charge[inr+3];
 111
 112     /* When we use explicit cutoffs the value must be identical for elec and VdW, so use elec as an arbitrary choice */
 113     rcutoff          = fr->rcoulomb;
 114     rcutoff2         = rcutoff*rcutoff;
 115
 116     outeriter        = 0;
 117     inneriter        = 0;
 118
 119     /* Start outer loop over neighborlists */
 120     for(iidx=0; iidx<nri; iidx++)
 121     {
 122         /* Load shift vector for this list */
 123         i_shift_offset   = DIM*shiftidx[iidx];
 124         shX              = shiftvec[i_shift_offset+XX];
 125         shY              = shiftvec[i_shift_offset+YY];
 126         shZ              = shiftvec[i_shift_offset+ZZ];
 127
 128         /* Load limits for loop over neighbors */
 129         j_index_start    = jindex[iidx];
 130         j_index_end      = jindex[iidx+1];
 131
 132         /* Get outer coordinate index */
 133         inr              = iinr[iidx];
 134         i_coord_offset   = DIM*inr;
 135
 136         /* Load i particle coords and add shift vector */
 137         ix1              = shX + x[i_coord_offset+DIM*1+XX];
 138         iy1              = shY + x[i_coord_offset+DIM*1+YY];
 139         iz1              = shZ + x[i_coord_offset+DIM*1+ZZ];
 140         ix2              = shX + x[i_coord_offset+DIM*2+XX];
 141         iy2              = shY + x[i_coord_offset+DIM*2+YY];
 142         iz2              = shZ + x[i_coord_offset+DIM*2+ZZ];
 143         ix3              = shX + x[i_coord_offset+DIM*3+XX];
 144         iy3              = shY + x[i_coord_offset+DIM*3+YY];
 145         iz3              = shZ + x[i_coord_offset+DIM*3+ZZ];
 146
 147         fix1             = 0.0;
 148         fiy1             = 0.0;
 149         fiz1             = 0.0;
 150         fix2             = 0.0;
 151         fiy2             = 0.0;
 152         fiz2             = 0.0;
 153         fix3             = 0.0;
 154         fiy3             = 0.0;
 155         fiz3             = 0.0;
 156
 157         /* Reset potential sums */
 158         velecsum         = 0.0;
 159
 160         /* Start inner kernel loop */
 161         for(jidx=j_index_start; jidx<j_index_end; jidx++)
 162         {
 163             /* Get j neighbor index, and coordinate index */
 164             jnr              = jjnr[jidx];
 165             j_coord_offset   = DIM*jnr;
 166
 167             /* load j atom coordinates */
 168             jx0              = x[j_coord_offset+DIM*0+XX];
 169             jy0              = x[j_coord_offset+DIM*0+YY];
 170             jz0              = x[j_coord_offset+DIM*0+ZZ];
 171
 172             /* Calculate displacement vector */
 173             dx10             = ix1 - jx0;
 174             dy10             = iy1 - jy0;
 175             dz10             = iz1 - jz0;
 176             dx20             = ix2 - jx0;
 177             dy20             = iy2 - jy0;
 178             dz20             = iz2 - jz0;
 179             dx30             = ix3 - jx0;
 180             dy30             = iy3 - jy0;
 181             dz30             = iz3 - jz0;
 182
 183             /* Calculate squared distance and things based on it */
 184             rsq10            = dx10*dx10+dy10*dy10+dz10*dz10;
 185             rsq20            = dx20*dx20+dy20*dy20+dz20*dz20;
 186             rsq30            = dx30*dx30+dy30*dy30+dz30*dz30;
 187
 188             rinv10           = gmx_invsqrt(rsq10);
 189             rinv20           = gmx_invsqrt(rsq20);
 190             rinv30           = gmx_invsqrt(rsq30);
 191
 192             rinvsq10         = rinv10*rinv10;
 193             rinvsq20         = rinv20*rinv20;
 194             rinvsq30         = rinv30*rinv30;
 195
 196             /* Load parameters for j particles */
 197             jq0              = charge[jnr+0];
 198
 199             /**************************
 200              * CALCULATE INTERACTIONS *
 201              **************************/
 202
 203             if (rsq10<rcutoff2)
 204             {
 205
 206             r10              = rsq10*rinv10;
 207
 208             qq10             = iq1*jq0;
 209
 210             /* EWALD ELECTROSTATICS */
 211
 212             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 213             ewrt             = r10*ewtabscale;
 214             ewitab           = ewrt;
 215             eweps            = ewrt-ewitab;
 216             ewitab           = 4*ewitab;
 217             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 218             velec            = qq10*((rinv10-sh_ewald)-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 219             felec            = qq10*rinv10*(rinvsq10-felec);
 220
 221             /* Update potential sums from outer loop */
 222             velecsum        += velec;
 223
 224             fscal            = felec;
 225
 226             /* Calculate temporary vectorial force */
 227             tx               = fscal*dx10;
 228             ty               = fscal*dy10;
 229             tz               = fscal*dz10;
 230
 231             /* Update vectorial force */
 232             fix1            += tx;
 233             fiy1            += ty;
 234             fiz1            += tz;
 235             f[j_coord_offset+DIM*0+XX] -= tx;
 236             f[j_coord_offset+DIM*0+YY] -= ty;
 237             f[j_coord_offset+DIM*0+ZZ] -= tz;
 238
 239             }
 240
 241             /**************************
 242              * CALCULATE INTERACTIONS *
 243              **************************/
 244
 245             if (rsq20<rcutoff2)
 246             {
 247
 248             r20              = rsq20*rinv20;
 249
 250             qq20             = iq2*jq0;
 251
 252             /* EWALD ELECTROSTATICS */
 253
 254             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 255             ewrt             = r20*ewtabscale;
 256             ewitab           = ewrt;
 257             eweps            = ewrt-ewitab;
 258             ewitab           = 4*ewitab;
 259             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 260             velec            = qq20*((rinv20-sh_ewald)-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 261             felec            = qq20*rinv20*(rinvsq20-felec);
 262
 263             /* Update potential sums from outer loop */
 264             velecsum        += velec;
 265
 266             fscal            = felec;
 267
 268             /* Calculate temporary vectorial force */
 269             tx               = fscal*dx20;
 270             ty               = fscal*dy20;
 271             tz               = fscal*dz20;
 272
 273             /* Update vectorial force */
 274             fix2            += tx;
 275             fiy2            += ty;
 276             fiz2            += tz;
 277             f[j_coord_offset+DIM*0+XX] -= tx;
 278             f[j_coord_offset+DIM*0+YY] -= ty;
 279             f[j_coord_offset+DIM*0+ZZ] -= tz;
 280
 281             }
 282
 283             /**************************
 284              * CALCULATE INTERACTIONS *
 285              **************************/
 286
 287             if (rsq30<rcutoff2)
 288             {
 289
 290             r30              = rsq30*rinv30;
 291
 292             qq30             = iq3*jq0;
 293
 294             /* EWALD ELECTROSTATICS */
 295
 296             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 297             ewrt             = r30*ewtabscale;
 298             ewitab           = ewrt;
 299             eweps            = ewrt-ewitab;
 300             ewitab           = 4*ewitab;
 301             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 302             velec            = qq30*((rinv30-sh_ewald)-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 303             felec            = qq30*rinv30*(rinvsq30-felec);
 304
 305             /* Update potential sums from outer loop */
 306             velecsum        += velec;
 307
 308             fscal            = felec;
 309
 310             /* Calculate temporary vectorial force */
 311             tx               = fscal*dx30;
 312             ty               = fscal*dy30;
 313             tz               = fscal*dz30;
 314
 315             /* Update vectorial force */
 316             fix3            += tx;
 317             fiy3            += ty;
 318             fiz3            += tz;
 319             f[j_coord_offset+DIM*0+XX] -= tx;
 320             f[j_coord_offset+DIM*0+YY] -= ty;
 321             f[j_coord_offset+DIM*0+ZZ] -= tz;
 322
 323             }
 324
 325             /* Inner loop uses 126 flops */
 326         }
 327         /* End of innermost loop */
 328
 329         tx = ty = tz = 0;
 330         f[i_coord_offset+DIM*1+XX] += fix1;
 331         f[i_coord_offset+DIM*1+YY] += fiy1;
 332         f[i_coord_offset+DIM*1+ZZ] += fiz1;
 333         tx                         += fix1;
 334         ty                         += fiy1;
 335         tz                         += fiz1;
 336         f[i_coord_offset+DIM*2+XX] += fix2;
 337         f[i_coord_offset+DIM*2+YY] += fiy2;
 338         f[i_coord_offset+DIM*2+ZZ] += fiz2;
 339         tx                         += fix2;
 340         ty                         += fiy2;
 341         tz                         += fiz2;
 342         f[i_coord_offset+DIM*3+XX] += fix3;
 343         f[i_coord_offset+DIM*3+YY] += fiy3;
 344         f[i_coord_offset+DIM*3+ZZ] += fiz3;
 345         tx                         += fix3;
 346         ty                         += fiy3;
 347         tz                         += fiz3;
 348         fshift[i_shift_offset+XX]  += tx;
 349         fshift[i_shift_offset+YY]  += ty;
 350         fshift[i_shift_offset+ZZ]  += tz;
 351
 352         ggid                        = gid[iidx];
 353         /* Update potential energies */
 354         kernel_data->energygrp_elec[ggid] += velecsum;
 355
 356         /* Increment number of inner iterations */
 357         inneriter                  += j_index_end - j_index_start;
 358
 359         /* Outer loop uses 31 flops */
 360     }
 361
 362     /* Increment number of outer iterations */
 363     outeriter        += nri;
 364
 365     /* Update outer/inner flops */
 366
 367     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_W4_VF,outeriter*31 + inneriter*126);
 368 }
 369 /*
 370  * Gromacs nonbonded kernel:   nb_kernel_ElecEwSh_VdwNone_GeomW4P1_F_c
 371  * Electrostatics interaction: Ewald
 372  * VdW interaction:            None
 373  * Geometry:                   Water4-Particle
 374  * Calculate force/pot:        Force
 375  */
 376 void
 377 nb_kernel_ElecEwSh_VdwNone_GeomW4P1_F_c
 378                     (t_nblist                    * gmx_restrict       nlist,
 379                      rvec                        * gmx_restrict          xx,
 380                      rvec                        * gmx_restrict          ff,
 381                      t_forcerec                  * gmx_restrict          fr,
 382                      t_mdatoms                   * gmx_restrict     mdatoms,
 383                      nb_kernel_data_t gmx_unused * gmx_restrict kernel_data,
 384                      t_nrnb                      * gmx_restrict        nrnb)
 385 {
 386     int              i_shift_offset,i_coord_offset,j_coord_offset;
 387     int              j_index_start,j_index_end;
 388     int              nri,inr,ggid,iidx,jidx,jnr,outeriter,inneriter;
 389     real             shX,shY,shZ,tx,ty,tz,fscal,rcutoff,rcutoff2;
 390     int              *iinr,*jindex,*jjnr,*shiftidx,*gid;
 391     real             *shiftvec,*fshift,*x,*f;
 392     int              vdwioffset1;
 393     real             ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
 394     int              vdwioffset2;
 395     real             ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
 396     int              vdwioffset3;
 397     real             ix3,iy3,iz3,fix3,fiy3,fiz3,iq3,isai3;
 398     int              vdwjidx0;
 399     real             jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
 400     real             dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10,cexp1_10,cexp2_10;
 401     real             dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20,cexp1_20,cexp2_20;
 402     real             dx30,dy30,dz30,rsq30,rinv30,rinvsq30,r30,qq30,c6_30,c12_30,cexp1_30,cexp2_30;
 403     real             velec,felec,velecsum,facel,crf,krf,krf2;
 404     real             *charge;
 405     int              ewitab;
 406     real             ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace;
 407     real             *ewtab;
 408
 409     x                = xx[0];
 410     f                = ff[0];
 411
 412     nri              = nlist->nri;
 413     iinr             = nlist->iinr;
 414     jindex           = nlist->jindex;
 415     jjnr             = nlist->jjnr;
 416     shiftidx         = nlist->shift;
 417     gid              = nlist->gid;
 418     shiftvec         = fr->shift_vec[0];
 419     fshift           = fr->fshift[0];
 420     facel            = fr->epsfac;
 421     charge           = mdatoms->chargeA;
 422
 423     sh_ewald         = fr->ic->sh_ewald;
 424     ewtab            = fr->ic->tabq_coul_F;
 425     ewtabscale       = fr->ic->tabq_scale;
 426     ewtabhalfspace   = 0.5/ewtabscale;
 427
 428     /* Setup water-specific parameters */
 429     inr              = nlist->iinr[0];
 430     iq1              = facel*charge[inr+1];
 431     iq2              = facel*charge[inr+2];
 432     iq3              = facel*charge[inr+3];
 433
 434     /* When we use explicit cutoffs the value must be identical for elec and VdW, so use elec as an arbitrary choice */
 435     rcutoff          = fr->rcoulomb;
 436     rcutoff2         = rcutoff*rcutoff;
 437
 438     outeriter        = 0;
 439     inneriter        = 0;
 440
 441     /* Start outer loop over neighborlists */
 442     for(iidx=0; iidx<nri; iidx++)
 443     {
 444         /* Load shift vector for this list */
 445         i_shift_offset   = DIM*shiftidx[iidx];
 446         shX              = shiftvec[i_shift_offset+XX];
 447         shY              = shiftvec[i_shift_offset+YY];
 448         shZ              = shiftvec[i_shift_offset+ZZ];
 449
 450         /* Load limits for loop over neighbors */
 451         j_index_start    = jindex[iidx];
 452         j_index_end      = jindex[iidx+1];
 453
 454         /* Get outer coordinate index */
 455         inr              = iinr[iidx];
 456         i_coord_offset   = DIM*inr;
 457
 458         /* Load i particle coords and add shift vector */
 459         ix1              = shX + x[i_coord_offset+DIM*1+XX];
 460         iy1              = shY + x[i_coord_offset+DIM*1+YY];
 461         iz1              = shZ + x[i_coord_offset+DIM*1+ZZ];
 462         ix2              = shX + x[i_coord_offset+DIM*2+XX];
 463         iy2              = shY + x[i_coord_offset+DIM*2+YY];
 464         iz2              = shZ + x[i_coord_offset+DIM*2+ZZ];
 465         ix3              = shX + x[i_coord_offset+DIM*3+XX];
 466         iy3              = shY + x[i_coord_offset+DIM*3+YY];
 467         iz3              = shZ + x[i_coord_offset+DIM*3+ZZ];
 468
 469         fix1             = 0.0;
 470         fiy1             = 0.0;
 471         fiz1             = 0.0;
 472         fix2             = 0.0;
 473         fiy2             = 0.0;
 474         fiz2             = 0.0;
 475         fix3             = 0.0;
 476         fiy3             = 0.0;
 477         fiz3             = 0.0;
 478
 479         /* Start inner kernel loop */
 480         for(jidx=j_index_start; jidx<j_index_end; jidx++)
 481         {
 482             /* Get j neighbor index, and coordinate index */
 483             jnr              = jjnr[jidx];
 484             j_coord_offset   = DIM*jnr;
 485
 486             /* load j atom coordinates */
 487             jx0              = x[j_coord_offset+DIM*0+XX];
 488             jy0              = x[j_coord_offset+DIM*0+YY];
 489             jz0              = x[j_coord_offset+DIM*0+ZZ];
 490
 491             /* Calculate displacement vector */
 492             dx10             = ix1 - jx0;
 493             dy10             = iy1 - jy0;
 494             dz10             = iz1 - jz0;
 495             dx20             = ix2 - jx0;
 496             dy20             = iy2 - jy0;
 497             dz20             = iz2 - jz0;
 498             dx30             = ix3 - jx0;
 499             dy30             = iy3 - jy0;
 500             dz30             = iz3 - jz0;
 501
 502             /* Calculate squared distance and things based on it */
 503             rsq10            = dx10*dx10+dy10*dy10+dz10*dz10;
 504             rsq20            = dx20*dx20+dy20*dy20+dz20*dz20;
 505             rsq30            = dx30*dx30+dy30*dy30+dz30*dz30;
 506
 507             rinv10           = gmx_invsqrt(rsq10);
 508             rinv20           = gmx_invsqrt(rsq20);
 509             rinv30           = gmx_invsqrt(rsq30);
 510
 511             rinvsq10         = rinv10*rinv10;
 512             rinvsq20         = rinv20*rinv20;
 513             rinvsq30         = rinv30*rinv30;
 514
 515             /* Load parameters for j particles */
 516             jq0              = charge[jnr+0];
 517
 518             /**************************
 519              * CALCULATE INTERACTIONS *
 520              **************************/
 521
 522             if (rsq10<rcutoff2)
 523             {
 524
 525             r10              = rsq10*rinv10;
 526
 527             qq10             = iq1*jq0;
 528
 529             /* EWALD ELECTROSTATICS */
 530
 531             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 532             ewrt             = r10*ewtabscale;
 533             ewitab           = ewrt;
 534             eweps            = ewrt-ewitab;
 535             felec            = (1.0-eweps)*ewtab[ewitab]+eweps*ewtab[ewitab+1];
 536             felec            = qq10*rinv10*(rinvsq10-felec);
 537
 538             fscal            = felec;
 539
 540             /* Calculate temporary vectorial force */
 541             tx               = fscal*dx10;
 542             ty               = fscal*dy10;
 543             tz               = fscal*dz10;
 544
 545             /* Update vectorial force */
 546             fix1            += tx;
 547             fiy1            += ty;
 548             fiz1            += tz;
 549             f[j_coord_offset+DIM*0+XX] -= tx;
 550             f[j_coord_offset+DIM*0+YY] -= ty;
 551             f[j_coord_offset+DIM*0+ZZ] -= tz;
 552
 553             }
 554
 555             /**************************
 556              * CALCULATE INTERACTIONS *
 557              **************************/
 558
 559             if (rsq20<rcutoff2)
 560             {
 561
 562             r20              = rsq20*rinv20;
 563
 564             qq20             = iq2*jq0;
 565
 566             /* EWALD ELECTROSTATICS */
 567
 568             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 569             ewrt             = r20*ewtabscale;
 570             ewitab           = ewrt;
 571             eweps            = ewrt-ewitab;
 572             felec            = (1.0-eweps)*ewtab[ewitab]+eweps*ewtab[ewitab+1];
 573             felec            = qq20*rinv20*(rinvsq20-felec);
 574
 575             fscal            = felec;
 576
 577             /* Calculate temporary vectorial force */
 578             tx               = fscal*dx20;
 579             ty               = fscal*dy20;
 580             tz               = fscal*dz20;
 581
 582             /* Update vectorial force */
 583             fix2            += tx;
 584             fiy2            += ty;
 585             fiz2            += tz;
 586             f[j_coord_offset+DIM*0+XX] -= tx;
 587             f[j_coord_offset+DIM*0+YY] -= ty;
 588             f[j_coord_offset+DIM*0+ZZ] -= tz;
 589
 590             }
 591
 592             /**************************
 593              * CALCULATE INTERACTIONS *
 594              **************************/
 595
 596             if (rsq30<rcutoff2)
 597             {
 598
 599             r30              = rsq30*rinv30;
 600
 601             qq30             = iq3*jq0;
 602
 603             /* EWALD ELECTROSTATICS */
 604
 605             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 606             ewrt             = r30*ewtabscale;
 607             ewitab           = ewrt;
 608             eweps            = ewrt-ewitab;
 609             felec            = (1.0-eweps)*ewtab[ewitab]+eweps*ewtab[ewitab+1];
 610             felec            = qq30*rinv30*(rinvsq30-felec);
 611
 612             fscal            = felec;
 613
 614             /* Calculate temporary vectorial force */
 615             tx               = fscal*dx30;
 616             ty               = fscal*dy30;
 617             tz               = fscal*dz30;
 618
 619             /* Update vectorial force */
 620             fix3            += tx;
 621             fiy3            += ty;
 622             fiz3            += tz;
 623             f[j_coord_offset+DIM*0+XX] -= tx;
 624             f[j_coord_offset+DIM*0+YY] -= ty;
 625             f[j_coord_offset+DIM*0+ZZ] -= tz;
 626
 627             }
 628
 629             /* Inner loop uses 102 flops */
 630         }
 631         /* End of innermost loop */
 632
 633         tx = ty = tz = 0;
 634         f[i_coord_offset+DIM*1+XX] += fix1;
 635         f[i_coord_offset+DIM*1+YY] += fiy1;
 636         f[i_coord_offset+DIM*1+ZZ] += fiz1;
 637         tx                         += fix1;
 638         ty                         += fiy1;
 639         tz                         += fiz1;
 640         f[i_coord_offset+DIM*2+XX] += fix2;
 641         f[i_coord_offset+DIM*2+YY] += fiy2;
 642         f[i_coord_offset+DIM*2+ZZ] += fiz2;
 643         tx                         += fix2;
 644         ty                         += fiy2;
 645         tz                         += fiz2;
 646         f[i_coord_offset+DIM*3+XX] += fix3;
 647         f[i_coord_offset+DIM*3+YY] += fiy3;
 648         f[i_coord_offset+DIM*3+ZZ] += fiz3;
 649         tx                         += fix3;
 650         ty                         += fiy3;
 651         tz                         += fiz3;
 652         fshift[i_shift_offset+XX]  += tx;
 653         fshift[i_shift_offset+YY]  += ty;
 654         fshift[i_shift_offset+ZZ]  += tz;
 655
 656         /* Increment number of inner iterations */
 657         inneriter                  += j_index_end - j_index_start;
 658
 659         /* Outer loop uses 30 flops */
 660     }
 661
 662     /* Increment number of outer iterations */
 663     outeriter        += nri;
 664
 665     /* Update outer/inner flops */
 666
 667     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_W4_F,outeriter*30 + inneriter*102);
 668 }