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, 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 #ifdef HAVE_CONFIG_H
  39 #include <config.h>
  40 #endif
  41
  42 #include <math.h>
  43
  44 #include "../nb_kernel.h"
  45 #include "types/simple.h"
  46 #include "vec.h"
  47 #include "nrnb.h"
  48
  49 #include "gromacs/simd/math_x86_sse4_1_double.h"
  50 #include "kernelutil_x86_sse4_1_double.h"
  51
  52 /*
  53  * Gromacs nonbonded kernel:   nb_kernel_ElecNone_VdwLJ_GeomP1P1_VF_sse4_1_double
  54  * Electrostatics interaction: None
  55  * VdW interaction:            LennardJones
  56  * Geometry:                   Particle-Particle
  57  * Calculate force/pot:        PotentialAndForce
  58  */
  59 void
  60 nb_kernel_ElecNone_VdwLJ_GeomP1P1_VF_sse4_1_double
  61                     (t_nblist                    * gmx_restrict       nlist,
  62                      rvec                        * gmx_restrict          xx,
  63                      rvec                        * gmx_restrict          ff,
  64                      t_forcerec                  * gmx_restrict          fr,
  65                      t_mdatoms                   * gmx_restrict     mdatoms,
  66                      nb_kernel_data_t gmx_unused * gmx_restrict kernel_data,
  67                      t_nrnb                      * gmx_restrict        nrnb)
  68 {
  69     /* Suffixes 0,1,2,3 refer to particle indices for waters in the inner or outer loop, or
  70      * just 0 for non-waters.
  71      * Suffixes A,B refer to j loop unrolling done with SSE double precision, e.g. for the two different
  72      * jnr indices corresponding to data put in the four positions in the SIMD register.
  73      */
  74     int              i_shift_offset,i_coord_offset,outeriter,inneriter;
  75     int              j_index_start,j_index_end,jidx,nri,inr,ggid,iidx;
  76     int              jnrA,jnrB;
  77     int              j_coord_offsetA,j_coord_offsetB;
  78     int              *iinr,*jindex,*jjnr,*shiftidx,*gid;
  79     real             rcutoff_scalar;
  80     real             *shiftvec,*fshift,*x,*f;
  81     __m128d          tx,ty,tz,fscal,rcutoff,rcutoff2,jidxall;
  82     int              vdwioffset0;
  83     __m128d          ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
  84     int              vdwjidx0A,vdwjidx0B;
  85     __m128d          jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
  86     __m128d          dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00;
  87     int              nvdwtype;
  88     __m128d          rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,sh_vdw_invrcut6;
  89     int              *vdwtype;
  90     real             *vdwparam;
  91     __m128d          one_sixth   = _mm_set1_pd(1.0/6.0);
  92     __m128d          one_twelfth = _mm_set1_pd(1.0/12.0);
  93     __m128d          dummy_mask,cutoff_mask;
  94     __m128d          signbit   = gmx_mm_castsi128_pd( _mm_set_epi32(0x80000000,0x00000000,0x80000000,0x00000000) );
  95     __m128d          one     = _mm_set1_pd(1.0);
  96     __m128d          two     = _mm_set1_pd(2.0);
  97     x                = xx[0];
  98     f                = ff[0];
  99
 100     nri              = nlist->nri;
 101     iinr             = nlist->iinr;
 102     jindex           = nlist->jindex;
 103     jjnr             = nlist->jjnr;
 104     shiftidx         = nlist->shift;
 105     gid              = nlist->gid;
 106     shiftvec         = fr->shift_vec[0];
 107     fshift           = fr->fshift[0];
 108     nvdwtype         = fr->ntype;
 109     vdwparam         = fr->nbfp;
 110     vdwtype          = mdatoms->typeA;
 111
 112     /* Avoid stupid compiler warnings */
 113     jnrA = jnrB = 0;
 114     j_coord_offsetA = 0;
 115     j_coord_offsetB = 0;
 116
 117     outeriter        = 0;
 118     inneriter        = 0;
 119
 120     /* Start outer loop over neighborlists */
 121     for(iidx=0; iidx<nri; iidx++)
 122     {
 123         /* Load shift vector for this list */
 124         i_shift_offset   = DIM*shiftidx[iidx];
 125
 126         /* Load limits for loop over neighbors */
 127         j_index_start    = jindex[iidx];
 128         j_index_end      = jindex[iidx+1];
 129
 130         /* Get outer coordinate index */
 131         inr              = iinr[iidx];
 132         i_coord_offset   = DIM*inr;
 133
 134         /* Load i particle coords and add shift vector */
 135         gmx_mm_load_shift_and_1rvec_broadcast_pd(shiftvec+i_shift_offset,x+i_coord_offset,&ix0,&iy0,&iz0);
 136
 137         fix0             = _mm_setzero_pd();
 138         fiy0             = _mm_setzero_pd();
 139         fiz0             = _mm_setzero_pd();
 140
 141         /* Load parameters for i particles */
 142         vdwioffset0      = 2*nvdwtype*vdwtype[inr+0];
 143
 144         /* Reset potential sums */
 145         vvdwsum          = _mm_setzero_pd();
 146
 147         /* Start inner kernel loop */
 148         for(jidx=j_index_start; jidx<j_index_end-1; jidx+=2)
 149         {
 150
 151             /* Get j neighbor index, and coordinate index */
 152             jnrA             = jjnr[jidx];
 153             jnrB             = jjnr[jidx+1];
 154             j_coord_offsetA  = DIM*jnrA;
 155             j_coord_offsetB  = DIM*jnrB;
 156
 157             /* load j atom coordinates */
 158             gmx_mm_load_1rvec_2ptr_swizzle_pd(x+j_coord_offsetA,x+j_coord_offsetB,
 159                                               &jx0,&jy0,&jz0);
 160
 161             /* Calculate displacement vector */
 162             dx00             = _mm_sub_pd(ix0,jx0);
 163             dy00             = _mm_sub_pd(iy0,jy0);
 164             dz00             = _mm_sub_pd(iz0,jz0);
 165
 166             /* Calculate squared distance and things based on it */
 167             rsq00            = gmx_mm_calc_rsq_pd(dx00,dy00,dz00);
 168
 169             rinvsq00         = gmx_mm_inv_pd(rsq00);
 170
 171             /* Load parameters for j particles */
 172             vdwjidx0A        = 2*vdwtype[jnrA+0];
 173             vdwjidx0B        = 2*vdwtype[jnrB+0];
 174
 175             /**************************
 176              * CALCULATE INTERACTIONS *
 177              **************************/
 178
 179             /* Compute parameters for interactions between i and j atoms */
 180             gmx_mm_load_2pair_swizzle_pd(vdwparam+vdwioffset0+vdwjidx0A,
 181                                          vdwparam+vdwioffset0+vdwjidx0B,&c6_00,&c12_00);
 182
 183             /* LENNARD-JONES DISPERSION/REPULSION */
 184
 185             rinvsix          = _mm_mul_pd(_mm_mul_pd(rinvsq00,rinvsq00),rinvsq00);
 186             vvdw6            = _mm_mul_pd(c6_00,rinvsix);
 187             vvdw12           = _mm_mul_pd(c12_00,_mm_mul_pd(rinvsix,rinvsix));
 188             vvdw             = _mm_sub_pd( _mm_mul_pd(vvdw12,one_twelfth) , _mm_mul_pd(vvdw6,one_sixth) );
 189             fvdw             = _mm_mul_pd(_mm_sub_pd(vvdw12,vvdw6),rinvsq00);
 190
 191             /* Update potential sum for this i atom from the interaction with this j atom. */
 192             vvdwsum          = _mm_add_pd(vvdwsum,vvdw);
 193
 194             fscal            = fvdw;
 195
 196             /* Calculate temporary vectorial force */
 197             tx               = _mm_mul_pd(fscal,dx00);
 198             ty               = _mm_mul_pd(fscal,dy00);
 199             tz               = _mm_mul_pd(fscal,dz00);
 200
 201             /* Update vectorial force */
 202             fix0             = _mm_add_pd(fix0,tx);
 203             fiy0             = _mm_add_pd(fiy0,ty);
 204             fiz0             = _mm_add_pd(fiz0,tz);
 205
 206             gmx_mm_decrement_1rvec_2ptr_swizzle_pd(f+j_coord_offsetA,f+j_coord_offsetB,tx,ty,tz);
 207
 208             /* Inner loop uses 32 flops */
 209         }
 210
 211         if(jidx<j_index_end)
 212         {
 213
 214             jnrA             = jjnr[jidx];
 215             j_coord_offsetA  = DIM*jnrA;
 216
 217             /* load j atom coordinates */
 218             gmx_mm_load_1rvec_1ptr_swizzle_pd(x+j_coord_offsetA,
 219                                               &jx0,&jy0,&jz0);
 220
 221             /* Calculate displacement vector */
 222             dx00             = _mm_sub_pd(ix0,jx0);
 223             dy00             = _mm_sub_pd(iy0,jy0);
 224             dz00             = _mm_sub_pd(iz0,jz0);
 225
 226             /* Calculate squared distance and things based on it */
 227             rsq00            = gmx_mm_calc_rsq_pd(dx00,dy00,dz00);
 228
 229             rinvsq00         = gmx_mm_inv_pd(rsq00);
 230
 231             /* Load parameters for j particles */
 232             vdwjidx0A        = 2*vdwtype[jnrA+0];
 233
 234             /**************************
 235              * CALCULATE INTERACTIONS *
 236              **************************/
 237
 238             /* Compute parameters for interactions between i and j atoms */
 239             gmx_mm_load_1pair_swizzle_pd(vdwparam+vdwioffset0+vdwjidx0A,&c6_00,&c12_00);
 240
 241             /* LENNARD-JONES DISPERSION/REPULSION */
 242
 243             rinvsix          = _mm_mul_pd(_mm_mul_pd(rinvsq00,rinvsq00),rinvsq00);
 244             vvdw6            = _mm_mul_pd(c6_00,rinvsix);
 245             vvdw12           = _mm_mul_pd(c12_00,_mm_mul_pd(rinvsix,rinvsix));
 246             vvdw             = _mm_sub_pd( _mm_mul_pd(vvdw12,one_twelfth) , _mm_mul_pd(vvdw6,one_sixth) );
 247             fvdw             = _mm_mul_pd(_mm_sub_pd(vvdw12,vvdw6),rinvsq00);
 248
 249             /* Update potential sum for this i atom from the interaction with this j atom. */
 250             vvdw             = _mm_unpacklo_pd(vvdw,_mm_setzero_pd());
 251             vvdwsum          = _mm_add_pd(vvdwsum,vvdw);
 252
 253             fscal            = fvdw;
 254
 255             fscal            = _mm_unpacklo_pd(fscal,_mm_setzero_pd());
 256
 257             /* Calculate temporary vectorial force */
 258             tx               = _mm_mul_pd(fscal,dx00);
 259             ty               = _mm_mul_pd(fscal,dy00);
 260             tz               = _mm_mul_pd(fscal,dz00);
 261
 262             /* Update vectorial force */
 263             fix0             = _mm_add_pd(fix0,tx);
 264             fiy0             = _mm_add_pd(fiy0,ty);
 265             fiz0             = _mm_add_pd(fiz0,tz);
 266
 267             gmx_mm_decrement_1rvec_1ptr_swizzle_pd(f+j_coord_offsetA,tx,ty,tz);
 268
 269             /* Inner loop uses 32 flops */
 270         }
 271
 272         /* End of innermost loop */
 273
 274         gmx_mm_update_iforce_1atom_swizzle_pd(fix0,fiy0,fiz0,
 275                                               f+i_coord_offset,fshift+i_shift_offset);
 276
 277         ggid                        = gid[iidx];
 278         /* Update potential energies */
 279         gmx_mm_update_1pot_pd(vvdwsum,kernel_data->energygrp_vdw+ggid);
 280
 281         /* Increment number of inner iterations */
 282         inneriter                  += j_index_end - j_index_start;
 283
 284         /* Outer loop uses 7 flops */
 285     }
 286
 287     /* Increment number of outer iterations */
 288     outeriter        += nri;
 289
 290     /* Update outer/inner flops */
 291
 292     inc_nrnb(nrnb,eNR_NBKERNEL_VDW_VF,outeriter*7 + inneriter*32);
 293 }
 294 /*
 295  * Gromacs nonbonded kernel:   nb_kernel_ElecNone_VdwLJ_GeomP1P1_F_sse4_1_double
 296  * Electrostatics interaction: None
 297  * VdW interaction:            LennardJones
 298  * Geometry:                   Particle-Particle
 299  * Calculate force/pot:        Force
 300  */
 301 void
 302 nb_kernel_ElecNone_VdwLJ_GeomP1P1_F_sse4_1_double
 303                     (t_nblist                    * gmx_restrict       nlist,
 304                      rvec                        * gmx_restrict          xx,
 305                      rvec                        * gmx_restrict          ff,
 306                      t_forcerec                  * gmx_restrict          fr,
 307                      t_mdatoms                   * gmx_restrict     mdatoms,
 308                      nb_kernel_data_t gmx_unused * gmx_restrict kernel_data,
 309                      t_nrnb                      * gmx_restrict        nrnb)
 310 {
 311     /* Suffixes 0,1,2,3 refer to particle indices for waters in the inner or outer loop, or
 312      * just 0 for non-waters.
 313      * Suffixes A,B refer to j loop unrolling done with SSE double precision, e.g. for the two different
 314      * jnr indices corresponding to data put in the four positions in the SIMD register.
 315      */
 316     int              i_shift_offset,i_coord_offset,outeriter,inneriter;
 317     int              j_index_start,j_index_end,jidx,nri,inr,ggid,iidx;
 318     int              jnrA,jnrB;
 319     int              j_coord_offsetA,j_coord_offsetB;
 320     int              *iinr,*jindex,*jjnr,*shiftidx,*gid;
 321     real             rcutoff_scalar;
 322     real             *shiftvec,*fshift,*x,*f;
 323     __m128d          tx,ty,tz,fscal,rcutoff,rcutoff2,jidxall;
 324     int              vdwioffset0;
 325     __m128d          ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
 326     int              vdwjidx0A,vdwjidx0B;
 327     __m128d          jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
 328     __m128d          dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00;
 329     int              nvdwtype;
 330     __m128d          rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,sh_vdw_invrcut6;
 331     int              *vdwtype;
 332     real             *vdwparam;
 333     __m128d          one_sixth   = _mm_set1_pd(1.0/6.0);
 334     __m128d          one_twelfth = _mm_set1_pd(1.0/12.0);
 335     __m128d          dummy_mask,cutoff_mask;
 336     __m128d          signbit   = gmx_mm_castsi128_pd( _mm_set_epi32(0x80000000,0x00000000,0x80000000,0x00000000) );
 337     __m128d          one     = _mm_set1_pd(1.0);
 338     __m128d          two     = _mm_set1_pd(2.0);
 339     x                = xx[0];
 340     f                = ff[0];
 341
 342     nri              = nlist->nri;
 343     iinr             = nlist->iinr;
 344     jindex           = nlist->jindex;
 345     jjnr             = nlist->jjnr;
 346     shiftidx         = nlist->shift;
 347     gid              = nlist->gid;
 348     shiftvec         = fr->shift_vec[0];
 349     fshift           = fr->fshift[0];
 350     nvdwtype         = fr->ntype;
 351     vdwparam         = fr->nbfp;
 352     vdwtype          = mdatoms->typeA;
 353
 354     /* Avoid stupid compiler warnings */
 355     jnrA = jnrB = 0;
 356     j_coord_offsetA = 0;
 357     j_coord_offsetB = 0;
 358
 359     outeriter        = 0;
 360     inneriter        = 0;
 361
 362     /* Start outer loop over neighborlists */
 363     for(iidx=0; iidx<nri; iidx++)
 364     {
 365         /* Load shift vector for this list */
 366         i_shift_offset   = DIM*shiftidx[iidx];
 367
 368         /* Load limits for loop over neighbors */
 369         j_index_start    = jindex[iidx];
 370         j_index_end      = jindex[iidx+1];
 371
 372         /* Get outer coordinate index */
 373         inr              = iinr[iidx];
 374         i_coord_offset   = DIM*inr;
 375
 376         /* Load i particle coords and add shift vector */
 377         gmx_mm_load_shift_and_1rvec_broadcast_pd(shiftvec+i_shift_offset,x+i_coord_offset,&ix0,&iy0,&iz0);
 378
 379         fix0             = _mm_setzero_pd();
 380         fiy0             = _mm_setzero_pd();
 381         fiz0             = _mm_setzero_pd();
 382
 383         /* Load parameters for i particles */
 384         vdwioffset0      = 2*nvdwtype*vdwtype[inr+0];
 385
 386         /* Start inner kernel loop */
 387         for(jidx=j_index_start; jidx<j_index_end-1; jidx+=2)
 388         {
 389
 390             /* Get j neighbor index, and coordinate index */
 391             jnrA             = jjnr[jidx];
 392             jnrB             = jjnr[jidx+1];
 393             j_coord_offsetA  = DIM*jnrA;
 394             j_coord_offsetB  = DIM*jnrB;
 395
 396             /* load j atom coordinates */
 397             gmx_mm_load_1rvec_2ptr_swizzle_pd(x+j_coord_offsetA,x+j_coord_offsetB,
 398                                               &jx0,&jy0,&jz0);
 399
 400             /* Calculate displacement vector */
 401             dx00             = _mm_sub_pd(ix0,jx0);
 402             dy00             = _mm_sub_pd(iy0,jy0);
 403             dz00             = _mm_sub_pd(iz0,jz0);
 404
 405             /* Calculate squared distance and things based on it */
 406             rsq00            = gmx_mm_calc_rsq_pd(dx00,dy00,dz00);
 407
 408             rinvsq00         = gmx_mm_inv_pd(rsq00);
 409
 410             /* Load parameters for j particles */
 411             vdwjidx0A        = 2*vdwtype[jnrA+0];
 412             vdwjidx0B        = 2*vdwtype[jnrB+0];
 413
 414             /**************************
 415              * CALCULATE INTERACTIONS *
 416              **************************/
 417
 418             /* Compute parameters for interactions between i and j atoms */
 419             gmx_mm_load_2pair_swizzle_pd(vdwparam+vdwioffset0+vdwjidx0A,
 420                                          vdwparam+vdwioffset0+vdwjidx0B,&c6_00,&c12_00);
 421
 422             /* LENNARD-JONES DISPERSION/REPULSION */
 423
 424             rinvsix          = _mm_mul_pd(_mm_mul_pd(rinvsq00,rinvsq00),rinvsq00);
 425             fvdw             = _mm_mul_pd(_mm_sub_pd(_mm_mul_pd(c12_00,rinvsix),c6_00),_mm_mul_pd(rinvsix,rinvsq00));
 426
 427             fscal            = fvdw;
 428
 429             /* Calculate temporary vectorial force */
 430             tx               = _mm_mul_pd(fscal,dx00);
 431             ty               = _mm_mul_pd(fscal,dy00);
 432             tz               = _mm_mul_pd(fscal,dz00);
 433
 434             /* Update vectorial force */
 435             fix0             = _mm_add_pd(fix0,tx);
 436             fiy0             = _mm_add_pd(fiy0,ty);
 437             fiz0             = _mm_add_pd(fiz0,tz);
 438
 439             gmx_mm_decrement_1rvec_2ptr_swizzle_pd(f+j_coord_offsetA,f+j_coord_offsetB,tx,ty,tz);
 440
 441             /* Inner loop uses 27 flops */
 442         }
 443
 444         if(jidx<j_index_end)
 445         {
 446
 447             jnrA             = jjnr[jidx];
 448             j_coord_offsetA  = DIM*jnrA;
 449
 450             /* load j atom coordinates */
 451             gmx_mm_load_1rvec_1ptr_swizzle_pd(x+j_coord_offsetA,
 452                                               &jx0,&jy0,&jz0);
 453
 454             /* Calculate displacement vector */
 455             dx00             = _mm_sub_pd(ix0,jx0);
 456             dy00             = _mm_sub_pd(iy0,jy0);
 457             dz00             = _mm_sub_pd(iz0,jz0);
 458
 459             /* Calculate squared distance and things based on it */
 460             rsq00            = gmx_mm_calc_rsq_pd(dx00,dy00,dz00);
 461
 462             rinvsq00         = gmx_mm_inv_pd(rsq00);
 463
 464             /* Load parameters for j particles */
 465             vdwjidx0A        = 2*vdwtype[jnrA+0];
 466
 467             /**************************
 468              * CALCULATE INTERACTIONS *
 469              **************************/
 470
 471             /* Compute parameters for interactions between i and j atoms */
 472             gmx_mm_load_1pair_swizzle_pd(vdwparam+vdwioffset0+vdwjidx0A,&c6_00,&c12_00);
 473
 474             /* LENNARD-JONES DISPERSION/REPULSION */
 475
 476             rinvsix          = _mm_mul_pd(_mm_mul_pd(rinvsq00,rinvsq00),rinvsq00);
 477             fvdw             = _mm_mul_pd(_mm_sub_pd(_mm_mul_pd(c12_00,rinvsix),c6_00),_mm_mul_pd(rinvsix,rinvsq00));
 478
 479             fscal            = fvdw;
 480
 481             fscal            = _mm_unpacklo_pd(fscal,_mm_setzero_pd());
 482
 483             /* Calculate temporary vectorial force */
 484             tx               = _mm_mul_pd(fscal,dx00);
 485             ty               = _mm_mul_pd(fscal,dy00);
 486             tz               = _mm_mul_pd(fscal,dz00);
 487
 488             /* Update vectorial force */
 489             fix0             = _mm_add_pd(fix0,tx);
 490             fiy0             = _mm_add_pd(fiy0,ty);
 491             fiz0             = _mm_add_pd(fiz0,tz);
 492
 493             gmx_mm_decrement_1rvec_1ptr_swizzle_pd(f+j_coord_offsetA,tx,ty,tz);
 494
 495             /* Inner loop uses 27 flops */
 496         }
 497
 498         /* End of innermost loop */
 499
 500         gmx_mm_update_iforce_1atom_swizzle_pd(fix0,fiy0,fiz0,
 501                                               f+i_coord_offset,fshift+i_shift_offset);
 502
 503         /* Increment number of inner iterations */
 504         inneriter                  += j_index_end - j_index_start;
 505
 506         /* Outer loop uses 6 flops */
 507     }
 508
 509     /* Increment number of outer iterations */
 510     outeriter        += nri;
 511
 512     /* Update outer/inner flops */
 513
 514     inc_nrnb(nrnb,eNR_NBKERNEL_VDW_F,outeriter*6 + inneriter*27);
 515 }