src/gromacs/gmxlib/nonbonded/nb_kernel_sse4_1_double/nb_kernel_ElecNone_VdwLJ_GeomP1P1_sse4_1_double.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 sse4_1_double 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 #include "gromacs/simd/math_x86_sse4_1_double.h"
  48 #include "kernelutil_x86_sse4_1_double.h"
  49
  50 /*
  51  * Gromacs nonbonded kernel:   nb_kernel_ElecNone_VdwLJ_GeomP1P1_VF_sse4_1_double
  52  * Electrostatics interaction: None
  53  * VdW interaction:            LennardJones
  54  * Geometry:                   Particle-Particle
  55  * Calculate force/pot:        PotentialAndForce
  56  */
  57 void
  58 nb_kernel_ElecNone_VdwLJ_GeomP1P1_VF_sse4_1_double
  59                     (t_nblist                    * gmx_restrict       nlist,
  60                      rvec                        * gmx_restrict          xx,
  61                      rvec                        * gmx_restrict          ff,
  62                      t_forcerec                  * gmx_restrict          fr,
  63                      t_mdatoms                   * gmx_restrict     mdatoms,
  64                      nb_kernel_data_t gmx_unused * gmx_restrict kernel_data,
  65                      t_nrnb                      * gmx_restrict        nrnb)
  66 {
  67     /* Suffixes 0,1,2,3 refer to particle indices for waters in the inner or outer loop, or
  68      * just 0 for non-waters.
  69      * Suffixes A,B refer to j loop unrolling done with SSE double precision, e.g. for the two different
  70      * jnr indices corresponding to data put in the four positions in the SIMD register.
  71      */
  72     int              i_shift_offset,i_coord_offset,outeriter,inneriter;
  73     int              j_index_start,j_index_end,jidx,nri,inr,ggid,iidx;
  74     int              jnrA,jnrB;
  75     int              j_coord_offsetA,j_coord_offsetB;
  76     int              *iinr,*jindex,*jjnr,*shiftidx,*gid;
  77     real             rcutoff_scalar;
  78     real             *shiftvec,*fshift,*x,*f;
  79     __m128d          tx,ty,tz,fscal,rcutoff,rcutoff2,jidxall;
  80     int              vdwioffset0;
  81     __m128d          ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
  82     int              vdwjidx0A,vdwjidx0B;
  83     __m128d          jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
  84     __m128d          dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00;
  85     int              nvdwtype;
  86     __m128d          rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,sh_vdw_invrcut6;
  87     int              *vdwtype;
  88     real             *vdwparam;
  89     __m128d          one_sixth   = _mm_set1_pd(1.0/6.0);
  90     __m128d          one_twelfth = _mm_set1_pd(1.0/12.0);
  91     __m128d          dummy_mask,cutoff_mask;
  92     __m128d          signbit   = gmx_mm_castsi128_pd( _mm_set_epi32(0x80000000,0x00000000,0x80000000,0x00000000) );
  93     __m128d          one     = _mm_set1_pd(1.0);
  94     __m128d          two     = _mm_set1_pd(2.0);
  95     x                = xx[0];
  96     f                = ff[0];
  97
  98     nri              = nlist->nri;
  99     iinr             = nlist->iinr;
 100     jindex           = nlist->jindex;
 101     jjnr             = nlist->jjnr;
 102     shiftidx         = nlist->shift;
 103     gid              = nlist->gid;
 104     shiftvec         = fr->shift_vec[0];
 105     fshift           = fr->fshift[0];
 106     nvdwtype         = fr->ntype;
 107     vdwparam         = fr->nbfp;
 108     vdwtype          = mdatoms->typeA;
 109
 110     /* Avoid stupid compiler warnings */
 111     jnrA = jnrB = 0;
 112     j_coord_offsetA = 0;
 113     j_coord_offsetB = 0;
 114
 115     outeriter        = 0;
 116     inneriter        = 0;
 117
 118     /* Start outer loop over neighborlists */
 119     for(iidx=0; iidx<nri; iidx++)
 120     {
 121         /* Load shift vector for this list */
 122         i_shift_offset   = DIM*shiftidx[iidx];
 123
 124         /* Load limits for loop over neighbors */
 125         j_index_start    = jindex[iidx];
 126         j_index_end      = jindex[iidx+1];
 127
 128         /* Get outer coordinate index */
 129         inr              = iinr[iidx];
 130         i_coord_offset   = DIM*inr;
 131
 132         /* Load i particle coords and add shift vector */
 133         gmx_mm_load_shift_and_1rvec_broadcast_pd(shiftvec+i_shift_offset,x+i_coord_offset,&ix0,&iy0,&iz0);
 134
 135         fix0             = _mm_setzero_pd();
 136         fiy0             = _mm_setzero_pd();
 137         fiz0             = _mm_setzero_pd();
 138
 139         /* Load parameters for i particles */
 140         vdwioffset0      = 2*nvdwtype*vdwtype[inr+0];
 141
 142         /* Reset potential sums */
 143         vvdwsum          = _mm_setzero_pd();
 144
 145         /* Start inner kernel loop */
 146         for(jidx=j_index_start; jidx<j_index_end-1; jidx+=2)
 147         {
 148
 149             /* Get j neighbor index, and coordinate index */
 150             jnrA             = jjnr[jidx];
 151             jnrB             = jjnr[jidx+1];
 152             j_coord_offsetA  = DIM*jnrA;
 153             j_coord_offsetB  = DIM*jnrB;
 154
 155             /* load j atom coordinates */
 156             gmx_mm_load_1rvec_2ptr_swizzle_pd(x+j_coord_offsetA,x+j_coord_offsetB,
 157                                               &jx0,&jy0,&jz0);
 158
 159             /* Calculate displacement vector */
 160             dx00             = _mm_sub_pd(ix0,jx0);
 161             dy00             = _mm_sub_pd(iy0,jy0);
 162             dz00             = _mm_sub_pd(iz0,jz0);
 163
 164             /* Calculate squared distance and things based on it */
 165             rsq00            = gmx_mm_calc_rsq_pd(dx00,dy00,dz00);
 166
 167             rinvsq00         = gmx_mm_inv_pd(rsq00);
 168
 169             /* Load parameters for j particles */
 170             vdwjidx0A        = 2*vdwtype[jnrA+0];
 171             vdwjidx0B        = 2*vdwtype[jnrB+0];
 172
 173             /**************************
 174              * CALCULATE INTERACTIONS *
 175              **************************/
 176
 177             /* Compute parameters for interactions between i and j atoms */
 178             gmx_mm_load_2pair_swizzle_pd(vdwparam+vdwioffset0+vdwjidx0A,
 179                                          vdwparam+vdwioffset0+vdwjidx0B,&c6_00,&c12_00);
 180
 181             /* LENNARD-JONES DISPERSION/REPULSION */
 182
 183             rinvsix          = _mm_mul_pd(_mm_mul_pd(rinvsq00,rinvsq00),rinvsq00);
 184             vvdw6            = _mm_mul_pd(c6_00,rinvsix);
 185             vvdw12           = _mm_mul_pd(c12_00,_mm_mul_pd(rinvsix,rinvsix));
 186             vvdw             = _mm_sub_pd( _mm_mul_pd(vvdw12,one_twelfth) , _mm_mul_pd(vvdw6,one_sixth) );
 187             fvdw             = _mm_mul_pd(_mm_sub_pd(vvdw12,vvdw6),rinvsq00);
 188
 189             /* Update potential sum for this i atom from the interaction with this j atom. */
 190             vvdwsum          = _mm_add_pd(vvdwsum,vvdw);
 191
 192             fscal            = fvdw;
 193
 194             /* Calculate temporary vectorial force */
 195             tx               = _mm_mul_pd(fscal,dx00);
 196             ty               = _mm_mul_pd(fscal,dy00);
 197             tz               = _mm_mul_pd(fscal,dz00);
 198
 199             /* Update vectorial force */
 200             fix0             = _mm_add_pd(fix0,tx);
 201             fiy0             = _mm_add_pd(fiy0,ty);
 202             fiz0             = _mm_add_pd(fiz0,tz);
 203
 204             gmx_mm_decrement_1rvec_2ptr_swizzle_pd(f+j_coord_offsetA,f+j_coord_offsetB,tx,ty,tz);
 205
 206             /* Inner loop uses 32 flops */
 207         }
 208
 209         if(jidx<j_index_end)
 210         {
 211
 212             jnrA             = jjnr[jidx];
 213             j_coord_offsetA  = DIM*jnrA;
 214
 215             /* load j atom coordinates */
 216             gmx_mm_load_1rvec_1ptr_swizzle_pd(x+j_coord_offsetA,
 217                                               &jx0,&jy0,&jz0);
 218
 219             /* Calculate displacement vector */
 220             dx00             = _mm_sub_pd(ix0,jx0);
 221             dy00             = _mm_sub_pd(iy0,jy0);
 222             dz00             = _mm_sub_pd(iz0,jz0);
 223
 224             /* Calculate squared distance and things based on it */
 225             rsq00            = gmx_mm_calc_rsq_pd(dx00,dy00,dz00);
 226
 227             rinvsq00         = gmx_mm_inv_pd(rsq00);
 228
 229             /* Load parameters for j particles */
 230             vdwjidx0A        = 2*vdwtype[jnrA+0];
 231
 232             /**************************
 233              * CALCULATE INTERACTIONS *
 234              **************************/
 235
 236             /* Compute parameters for interactions between i and j atoms */
 237             gmx_mm_load_1pair_swizzle_pd(vdwparam+vdwioffset0+vdwjidx0A,&c6_00,&c12_00);
 238
 239             /* LENNARD-JONES DISPERSION/REPULSION */
 240
 241             rinvsix          = _mm_mul_pd(_mm_mul_pd(rinvsq00,rinvsq00),rinvsq00);
 242             vvdw6            = _mm_mul_pd(c6_00,rinvsix);
 243             vvdw12           = _mm_mul_pd(c12_00,_mm_mul_pd(rinvsix,rinvsix));
 244             vvdw             = _mm_sub_pd( _mm_mul_pd(vvdw12,one_twelfth) , _mm_mul_pd(vvdw6,one_sixth) );
 245             fvdw             = _mm_mul_pd(_mm_sub_pd(vvdw12,vvdw6),rinvsq00);
 246
 247             /* Update potential sum for this i atom from the interaction with this j atom. */
 248             vvdw             = _mm_unpacklo_pd(vvdw,_mm_setzero_pd());
 249             vvdwsum          = _mm_add_pd(vvdwsum,vvdw);
 250
 251             fscal            = fvdw;
 252
 253             fscal            = _mm_unpacklo_pd(fscal,_mm_setzero_pd());
 254
 255             /* Calculate temporary vectorial force */
 256             tx               = _mm_mul_pd(fscal,dx00);
 257             ty               = _mm_mul_pd(fscal,dy00);
 258             tz               = _mm_mul_pd(fscal,dz00);
 259
 260             /* Update vectorial force */
 261             fix0             = _mm_add_pd(fix0,tx);
 262             fiy0             = _mm_add_pd(fiy0,ty);
 263             fiz0             = _mm_add_pd(fiz0,tz);
 264
 265             gmx_mm_decrement_1rvec_1ptr_swizzle_pd(f+j_coord_offsetA,tx,ty,tz);
 266
 267             /* Inner loop uses 32 flops */
 268         }
 269
 270         /* End of innermost loop */
 271
 272         gmx_mm_update_iforce_1atom_swizzle_pd(fix0,fiy0,fiz0,
 273                                               f+i_coord_offset,fshift+i_shift_offset);
 274
 275         ggid                        = gid[iidx];
 276         /* Update potential energies */
 277         gmx_mm_update_1pot_pd(vvdwsum,kernel_data->energygrp_vdw+ggid);
 278
 279         /* Increment number of inner iterations */
 280         inneriter                  += j_index_end - j_index_start;
 281
 282         /* Outer loop uses 7 flops */
 283     }
 284
 285     /* Increment number of outer iterations */
 286     outeriter        += nri;
 287
 288     /* Update outer/inner flops */
 289
 290     inc_nrnb(nrnb,eNR_NBKERNEL_VDW_VF,outeriter*7 + inneriter*32);
 291 }
 292 /*
 293  * Gromacs nonbonded kernel:   nb_kernel_ElecNone_VdwLJ_GeomP1P1_F_sse4_1_double
 294  * Electrostatics interaction: None
 295  * VdW interaction:            LennardJones
 296  * Geometry:                   Particle-Particle
 297  * Calculate force/pot:        Force
 298  */
 299 void
 300 nb_kernel_ElecNone_VdwLJ_GeomP1P1_F_sse4_1_double
 301                     (t_nblist                    * gmx_restrict       nlist,
 302                      rvec                        * gmx_restrict          xx,
 303                      rvec                        * gmx_restrict          ff,
 304                      t_forcerec                  * gmx_restrict          fr,
 305                      t_mdatoms                   * gmx_restrict     mdatoms,
 306                      nb_kernel_data_t gmx_unused * gmx_restrict kernel_data,
 307                      t_nrnb                      * gmx_restrict        nrnb)
 308 {
 309     /* Suffixes 0,1,2,3 refer to particle indices for waters in the inner or outer loop, or
 310      * just 0 for non-waters.
 311      * Suffixes A,B refer to j loop unrolling done with SSE double precision, e.g. for the two different
 312      * jnr indices corresponding to data put in the four positions in the SIMD register.
 313      */
 314     int              i_shift_offset,i_coord_offset,outeriter,inneriter;
 315     int              j_index_start,j_index_end,jidx,nri,inr,ggid,iidx;
 316     int              jnrA,jnrB;
 317     int              j_coord_offsetA,j_coord_offsetB;
 318     int              *iinr,*jindex,*jjnr,*shiftidx,*gid;
 319     real             rcutoff_scalar;
 320     real             *shiftvec,*fshift,*x,*f;
 321     __m128d          tx,ty,tz,fscal,rcutoff,rcutoff2,jidxall;
 322     int              vdwioffset0;
 323     __m128d          ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
 324     int              vdwjidx0A,vdwjidx0B;
 325     __m128d          jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
 326     __m128d          dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00;
 327     int              nvdwtype;
 328     __m128d          rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,sh_vdw_invrcut6;
 329     int              *vdwtype;
 330     real             *vdwparam;
 331     __m128d          one_sixth   = _mm_set1_pd(1.0/6.0);
 332     __m128d          one_twelfth = _mm_set1_pd(1.0/12.0);
 333     __m128d          dummy_mask,cutoff_mask;
 334     __m128d          signbit   = gmx_mm_castsi128_pd( _mm_set_epi32(0x80000000,0x00000000,0x80000000,0x00000000) );
 335     __m128d          one     = _mm_set1_pd(1.0);
 336     __m128d          two     = _mm_set1_pd(2.0);
 337     x                = xx[0];
 338     f                = ff[0];
 339
 340     nri              = nlist->nri;
 341     iinr             = nlist->iinr;
 342     jindex           = nlist->jindex;
 343     jjnr             = nlist->jjnr;
 344     shiftidx         = nlist->shift;
 345     gid              = nlist->gid;
 346     shiftvec         = fr->shift_vec[0];
 347     fshift           = fr->fshift[0];
 348     nvdwtype         = fr->ntype;
 349     vdwparam         = fr->nbfp;
 350     vdwtype          = mdatoms->typeA;
 351
 352     /* Avoid stupid compiler warnings */
 353     jnrA = jnrB = 0;
 354     j_coord_offsetA = 0;
 355     j_coord_offsetB = 0;
 356
 357     outeriter        = 0;
 358     inneriter        = 0;
 359
 360     /* Start outer loop over neighborlists */
 361     for(iidx=0; iidx<nri; iidx++)
 362     {
 363         /* Load shift vector for this list */
 364         i_shift_offset   = DIM*shiftidx[iidx];
 365
 366         /* Load limits for loop over neighbors */
 367         j_index_start    = jindex[iidx];
 368         j_index_end      = jindex[iidx+1];
 369
 370         /* Get outer coordinate index */
 371         inr              = iinr[iidx];
 372         i_coord_offset   = DIM*inr;
 373
 374         /* Load i particle coords and add shift vector */
 375         gmx_mm_load_shift_and_1rvec_broadcast_pd(shiftvec+i_shift_offset,x+i_coord_offset,&ix0,&iy0,&iz0);
 376
 377         fix0             = _mm_setzero_pd();
 378         fiy0             = _mm_setzero_pd();
 379         fiz0             = _mm_setzero_pd();
 380
 381         /* Load parameters for i particles */
 382         vdwioffset0      = 2*nvdwtype*vdwtype[inr+0];
 383
 384         /* Start inner kernel loop */
 385         for(jidx=j_index_start; jidx<j_index_end-1; jidx+=2)
 386         {
 387
 388             /* Get j neighbor index, and coordinate index */
 389             jnrA             = jjnr[jidx];
 390             jnrB             = jjnr[jidx+1];
 391             j_coord_offsetA  = DIM*jnrA;
 392             j_coord_offsetB  = DIM*jnrB;
 393
 394             /* load j atom coordinates */
 395             gmx_mm_load_1rvec_2ptr_swizzle_pd(x+j_coord_offsetA,x+j_coord_offsetB,
 396                                               &jx0,&jy0,&jz0);
 397
 398             /* Calculate displacement vector */
 399             dx00             = _mm_sub_pd(ix0,jx0);
 400             dy00             = _mm_sub_pd(iy0,jy0);
 401             dz00             = _mm_sub_pd(iz0,jz0);
 402
 403             /* Calculate squared distance and things based on it */
 404             rsq00            = gmx_mm_calc_rsq_pd(dx00,dy00,dz00);
 405
 406             rinvsq00         = gmx_mm_inv_pd(rsq00);
 407
 408             /* Load parameters for j particles */
 409             vdwjidx0A        = 2*vdwtype[jnrA+0];
 410             vdwjidx0B        = 2*vdwtype[jnrB+0];
 411
 412             /**************************
 413              * CALCULATE INTERACTIONS *
 414              **************************/
 415
 416             /* Compute parameters for interactions between i and j atoms */
 417             gmx_mm_load_2pair_swizzle_pd(vdwparam+vdwioffset0+vdwjidx0A,
 418                                          vdwparam+vdwioffset0+vdwjidx0B,&c6_00,&c12_00);
 419
 420             /* LENNARD-JONES DISPERSION/REPULSION */
 421
 422             rinvsix          = _mm_mul_pd(_mm_mul_pd(rinvsq00,rinvsq00),rinvsq00);
 423             fvdw             = _mm_mul_pd(_mm_sub_pd(_mm_mul_pd(c12_00,rinvsix),c6_00),_mm_mul_pd(rinvsix,rinvsq00));
 424
 425             fscal            = fvdw;
 426
 427             /* Calculate temporary vectorial force */
 428             tx               = _mm_mul_pd(fscal,dx00);
 429             ty               = _mm_mul_pd(fscal,dy00);
 430             tz               = _mm_mul_pd(fscal,dz00);
 431
 432             /* Update vectorial force */
 433             fix0             = _mm_add_pd(fix0,tx);
 434             fiy0             = _mm_add_pd(fiy0,ty);
 435             fiz0             = _mm_add_pd(fiz0,tz);
 436
 437             gmx_mm_decrement_1rvec_2ptr_swizzle_pd(f+j_coord_offsetA,f+j_coord_offsetB,tx,ty,tz);
 438
 439             /* Inner loop uses 27 flops */
 440         }
 441
 442         if(jidx<j_index_end)
 443         {
 444
 445             jnrA             = jjnr[jidx];
 446             j_coord_offsetA  = DIM*jnrA;
 447
 448             /* load j atom coordinates */
 449             gmx_mm_load_1rvec_1ptr_swizzle_pd(x+j_coord_offsetA,
 450                                               &jx0,&jy0,&jz0);
 451
 452             /* Calculate displacement vector */
 453             dx00             = _mm_sub_pd(ix0,jx0);
 454             dy00             = _mm_sub_pd(iy0,jy0);
 455             dz00             = _mm_sub_pd(iz0,jz0);
 456
 457             /* Calculate squared distance and things based on it */
 458             rsq00            = gmx_mm_calc_rsq_pd(dx00,dy00,dz00);
 459
 460             rinvsq00         = gmx_mm_inv_pd(rsq00);
 461
 462             /* Load parameters for j particles */
 463             vdwjidx0A        = 2*vdwtype[jnrA+0];
 464
 465             /**************************
 466              * CALCULATE INTERACTIONS *
 467              **************************/
 468
 469             /* Compute parameters for interactions between i and j atoms */
 470             gmx_mm_load_1pair_swizzle_pd(vdwparam+vdwioffset0+vdwjidx0A,&c6_00,&c12_00);
 471
 472             /* LENNARD-JONES DISPERSION/REPULSION */
 473
 474             rinvsix          = _mm_mul_pd(_mm_mul_pd(rinvsq00,rinvsq00),rinvsq00);
 475             fvdw             = _mm_mul_pd(_mm_sub_pd(_mm_mul_pd(c12_00,rinvsix),c6_00),_mm_mul_pd(rinvsix,rinvsq00));
 476
 477             fscal            = fvdw;
 478
 479             fscal            = _mm_unpacklo_pd(fscal,_mm_setzero_pd());
 480
 481             /* Calculate temporary vectorial force */
 482             tx               = _mm_mul_pd(fscal,dx00);
 483             ty               = _mm_mul_pd(fscal,dy00);
 484             tz               = _mm_mul_pd(fscal,dz00);
 485
 486             /* Update vectorial force */
 487             fix0             = _mm_add_pd(fix0,tx);
 488             fiy0             = _mm_add_pd(fiy0,ty);
 489             fiz0             = _mm_add_pd(fiz0,tz);
 490
 491             gmx_mm_decrement_1rvec_1ptr_swizzle_pd(f+j_coord_offsetA,tx,ty,tz);
 492
 493             /* Inner loop uses 27 flops */
 494         }
 495
 496         /* End of innermost loop */
 497
 498         gmx_mm_update_iforce_1atom_swizzle_pd(fix0,fiy0,fiz0,
 499                                               f+i_coord_offset,fshift+i_shift_offset);
 500
 501         /* Increment number of inner iterations */
 502         inneriter                  += j_index_end - j_index_start;
 503
 504         /* Outer loop uses 6 flops */
 505     }
 506
 507     /* Increment number of outer iterations */
 508     outeriter        += nri;
 509
 510     /* Update outer/inner flops */
 511
 512     inc_nrnb(nrnb,eNR_NBKERNEL_VDW_F,outeriter*6 + inneriter*27);
 513 }