src/gromacs/gmxlib/nonbonded/nb_kernel_sparc64_hpc_ace_double/nb_kernel_ElecNone_VdwLJ_GeomP1P1_sparc64_hpc_ace_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 sparc64_hpc_ace_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 "kernelutil_sparc64_hpc_ace_double.h"
  50
  51 /*
  52  * Gromacs nonbonded kernel:   nb_kernel_ElecNone_VdwLJ_GeomP1P1_VF_sparc64_hpc_ace_double
  53  * Electrostatics interaction: None
  54  * VdW interaction:            LennardJones
  55  * Geometry:                   Particle-Particle
  56  * Calculate force/pot:        PotentialAndForce
  57  */
  58 void
  59 nb_kernel_ElecNone_VdwLJ_GeomP1P1_VF_sparc64_hpc_ace_double
  60                     (t_nblist                    * gmx_restrict       nlist,
  61                      rvec                        * gmx_restrict          xx,
  62                      rvec                        * gmx_restrict          ff,
  63                      t_forcerec                  * gmx_restrict          fr,
  64                      t_mdatoms                   * gmx_restrict     mdatoms,
  65                      nb_kernel_data_t gmx_unused * gmx_restrict kernel_data,
  66                      t_nrnb                      * gmx_restrict        nrnb)
  67 {
  68     /* Suffixes 0,1,2,3 refer to particle indices for waters in the inner or outer loop, or
  69      * just 0 for non-waters.
  70      * Suffixes A,B refer to j loop unrolling done with double precision SIMD, e.g. for the two different
  71      * jnr indices corresponding to data put in the four positions in the SIMD register.
  72      */
  73     int              i_shift_offset,i_coord_offset,outeriter,inneriter;
  74     int              j_index_start,j_index_end,jidx,nri,inr,ggid,iidx;
  75     int              jnrA,jnrB;
  76     int              j_coord_offsetA,j_coord_offsetB;
  77     int              *iinr,*jindex,*jjnr,*shiftidx,*gid;
  78     real             rcutoff_scalar;
  79     real             *shiftvec,*fshift,*x,*f;
  80     _fjsp_v2r8       tx,ty,tz,fscal,rcutoff,rcutoff2,jidxall;
  81     int              vdwioffset0;
  82     _fjsp_v2r8       ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
  83     int              vdwjidx0A,vdwjidx0B;
  84     _fjsp_v2r8       jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
  85     _fjsp_v2r8       dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00;
  86     int              nvdwtype;
  87     _fjsp_v2r8       rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,sh_vdw_invrcut6;
  88     int              *vdwtype;
  89     real             *vdwparam;
  90     _fjsp_v2r8       one_sixth   = gmx_fjsp_set1_v2r8(1.0/6.0);
  91     _fjsp_v2r8       one_twelfth = gmx_fjsp_set1_v2r8(1.0/12.0);
  92     _fjsp_v2r8       itab_tmp;
  93     _fjsp_v2r8       dummy_mask,cutoff_mask;
  94     _fjsp_v2r8       one     = gmx_fjsp_set1_v2r8(1.0);
  95     _fjsp_v2r8       two     = gmx_fjsp_set1_v2r8(2.0);
  96     union { _fjsp_v2r8 simd; long long int i[2]; } vfconv,gbconv,ewconv;
  97
  98     x                = xx[0];
  99     f                = ff[0];
 100
 101     nri              = nlist->nri;
 102     iinr             = nlist->iinr;
 103     jindex           = nlist->jindex;
 104     jjnr             = nlist->jjnr;
 105     shiftidx         = nlist->shift;
 106     gid              = nlist->gid;
 107     shiftvec         = fr->shift_vec[0];
 108     fshift           = fr->fshift[0];
 109     nvdwtype         = fr->ntype;
 110     vdwparam         = fr->nbfp;
 111     vdwtype          = mdatoms->typeA;
 112
 113     /* Avoid stupid compiler warnings */
 114     jnrA = jnrB = 0;
 115     j_coord_offsetA = 0;
 116     j_coord_offsetB = 0;
 117
 118     outeriter        = 0;
 119     inneriter        = 0;
 120
 121     /* Start outer loop over neighborlists */
 122     for(iidx=0; iidx<nri; iidx++)
 123     {
 124         /* Load shift vector for this list */
 125         i_shift_offset   = DIM*shiftidx[iidx];
 126
 127         /* Load limits for loop over neighbors */
 128         j_index_start    = jindex[iidx];
 129         j_index_end      = jindex[iidx+1];
 130
 131         /* Get outer coordinate index */
 132         inr              = iinr[iidx];
 133         i_coord_offset   = DIM*inr;
 134
 135         /* Load i particle coords and add shift vector */
 136         gmx_fjsp_load_shift_and_1rvec_broadcast_v2r8(shiftvec+i_shift_offset,x+i_coord_offset,&ix0,&iy0,&iz0);
 137
 138         fix0             = _fjsp_setzero_v2r8();
 139         fiy0             = _fjsp_setzero_v2r8();
 140         fiz0             = _fjsp_setzero_v2r8();
 141
 142         /* Load parameters for i particles */
 143         vdwioffset0      = 2*nvdwtype*vdwtype[inr+0];
 144
 145         /* Reset potential sums */
 146         vvdwsum          = _fjsp_setzero_v2r8();
 147
 148         /* Start inner kernel loop */
 149         for(jidx=j_index_start; jidx<j_index_end-1; jidx+=2)
 150         {
 151
 152             /* Get j neighbor index, and coordinate index */
 153             jnrA             = jjnr[jidx];
 154             jnrB             = jjnr[jidx+1];
 155             j_coord_offsetA  = DIM*jnrA;
 156             j_coord_offsetB  = DIM*jnrB;
 157
 158             /* load j atom coordinates */
 159             gmx_fjsp_load_1rvec_2ptr_swizzle_v2r8(x+j_coord_offsetA,x+j_coord_offsetB,
 160                                               &jx0,&jy0,&jz0);
 161
 162             /* Calculate displacement vector */
 163             dx00             = _fjsp_sub_v2r8(ix0,jx0);
 164             dy00             = _fjsp_sub_v2r8(iy0,jy0);
 165             dz00             = _fjsp_sub_v2r8(iz0,jz0);
 166
 167             /* Calculate squared distance and things based on it */
 168             rsq00            = gmx_fjsp_calc_rsq_v2r8(dx00,dy00,dz00);
 169
 170             rinvsq00         = gmx_fjsp_inv_v2r8(rsq00);
 171
 172             /* Load parameters for j particles */
 173             vdwjidx0A        = 2*vdwtype[jnrA+0];
 174             vdwjidx0B        = 2*vdwtype[jnrB+0];
 175
 176             /**************************
 177              * CALCULATE INTERACTIONS *
 178              **************************/
 179
 180             /* Compute parameters for interactions between i and j atoms */
 181             gmx_fjsp_load_2pair_swizzle_v2r8(vdwparam+vdwioffset0+vdwjidx0A,
 182                                          vdwparam+vdwioffset0+vdwjidx0B,&c6_00,&c12_00);
 183
 184             /* LENNARD-JONES DISPERSION/REPULSION */
 185
 186             rinvsix          = _fjsp_mul_v2r8(_fjsp_mul_v2r8(rinvsq00,rinvsq00),rinvsq00);
 187             vvdw6            = _fjsp_mul_v2r8(c6_00,rinvsix);
 188             vvdw12           = _fjsp_mul_v2r8(c12_00,_fjsp_mul_v2r8(rinvsix,rinvsix));
 189             vvdw             = _fjsp_msub_v2r8( vvdw12,one_twelfth, _fjsp_mul_v2r8(vvdw6,one_sixth) );
 190             fvdw             = _fjsp_mul_v2r8(_fjsp_sub_v2r8(vvdw12,vvdw6),rinvsq00);
 191
 192             /* Update potential sum for this i atom from the interaction with this j atom. */
 193             vvdwsum          = _fjsp_add_v2r8(vvdwsum,vvdw);
 194
 195             fscal            = fvdw;
 196
 197             /* Update vectorial force */
 198             fix0             = _fjsp_madd_v2r8(dx00,fscal,fix0);
 199             fiy0             = _fjsp_madd_v2r8(dy00,fscal,fiy0);
 200             fiz0             = _fjsp_madd_v2r8(dz00,fscal,fiz0);
 201
 202             gmx_fjsp_decrement_fma_1rvec_2ptr_swizzle_v2r8(f+j_coord_offsetA,f+j_coord_offsetB,fscal,dx00,dy00,dz00);
 203
 204             /* Inner loop uses 35 flops */
 205         }
 206
 207         if(jidx<j_index_end)
 208         {
 209
 210             jnrA             = jjnr[jidx];
 211             j_coord_offsetA  = DIM*jnrA;
 212
 213             /* load j atom coordinates */
 214             gmx_fjsp_load_1rvec_1ptr_swizzle_v2r8(x+j_coord_offsetA,
 215                                               &jx0,&jy0,&jz0);
 216
 217             /* Calculate displacement vector */
 218             dx00             = _fjsp_sub_v2r8(ix0,jx0);
 219             dy00             = _fjsp_sub_v2r8(iy0,jy0);
 220             dz00             = _fjsp_sub_v2r8(iz0,jz0);
 221
 222             /* Calculate squared distance and things based on it */
 223             rsq00            = gmx_fjsp_calc_rsq_v2r8(dx00,dy00,dz00);
 224
 225             rinvsq00         = gmx_fjsp_inv_v2r8(rsq00);
 226
 227             /* Load parameters for j particles */
 228             vdwjidx0A        = 2*vdwtype[jnrA+0];
 229
 230             /**************************
 231              * CALCULATE INTERACTIONS *
 232              **************************/
 233
 234             /* Compute parameters for interactions between i and j atoms */
 235             gmx_fjsp_load_1pair_swizzle_v2r8(vdwparam+vdwioffset0+vdwjidx0A,&c6_00,&c12_00);
 236
 237             /* LENNARD-JONES DISPERSION/REPULSION */
 238
 239             rinvsix          = _fjsp_mul_v2r8(_fjsp_mul_v2r8(rinvsq00,rinvsq00),rinvsq00);
 240             vvdw6            = _fjsp_mul_v2r8(c6_00,rinvsix);
 241             vvdw12           = _fjsp_mul_v2r8(c12_00,_fjsp_mul_v2r8(rinvsix,rinvsix));
 242             vvdw             = _fjsp_msub_v2r8( vvdw12,one_twelfth, _fjsp_mul_v2r8(vvdw6,one_sixth) );
 243             fvdw             = _fjsp_mul_v2r8(_fjsp_sub_v2r8(vvdw12,vvdw6),rinvsq00);
 244
 245             /* Update potential sum for this i atom from the interaction with this j atom. */
 246             vvdw             = _fjsp_unpacklo_v2r8(vvdw,_fjsp_setzero_v2r8());
 247             vvdwsum          = _fjsp_add_v2r8(vvdwsum,vvdw);
 248
 249             fscal            = fvdw;
 250
 251             fscal            = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
 252
 253             /* Update vectorial force */
 254             fix0             = _fjsp_madd_v2r8(dx00,fscal,fix0);
 255             fiy0             = _fjsp_madd_v2r8(dy00,fscal,fiy0);
 256             fiz0             = _fjsp_madd_v2r8(dz00,fscal,fiz0);
 257
 258             gmx_fjsp_decrement_fma_1rvec_1ptr_swizzle_v2r8(f+j_coord_offsetA,fscal,dx00,dy00,dz00);
 259
 260             /* Inner loop uses 35 flops */
 261         }
 262
 263         /* End of innermost loop */
 264
 265         gmx_fjsp_update_iforce_1atom_swizzle_v2r8(fix0,fiy0,fiz0,
 266                                               f+i_coord_offset,fshift+i_shift_offset);
 267
 268         ggid                        = gid[iidx];
 269         /* Update potential energies */
 270         gmx_fjsp_update_1pot_v2r8(vvdwsum,kernel_data->energygrp_vdw+ggid);
 271
 272         /* Increment number of inner iterations */
 273         inneriter                  += j_index_end - j_index_start;
 274
 275         /* Outer loop uses 7 flops */
 276     }
 277
 278     /* Increment number of outer iterations */
 279     outeriter        += nri;
 280
 281     /* Update outer/inner flops */
 282
 283     inc_nrnb(nrnb,eNR_NBKERNEL_VDW_VF,outeriter*7 + inneriter*35);
 284 }
 285 /*
 286  * Gromacs nonbonded kernel:   nb_kernel_ElecNone_VdwLJ_GeomP1P1_F_sparc64_hpc_ace_double
 287  * Electrostatics interaction: None
 288  * VdW interaction:            LennardJones
 289  * Geometry:                   Particle-Particle
 290  * Calculate force/pot:        Force
 291  */
 292 void
 293 nb_kernel_ElecNone_VdwLJ_GeomP1P1_F_sparc64_hpc_ace_double
 294                     (t_nblist                    * gmx_restrict       nlist,
 295                      rvec                        * gmx_restrict          xx,
 296                      rvec                        * gmx_restrict          ff,
 297                      t_forcerec                  * gmx_restrict          fr,
 298                      t_mdatoms                   * gmx_restrict     mdatoms,
 299                      nb_kernel_data_t gmx_unused * gmx_restrict kernel_data,
 300                      t_nrnb                      * gmx_restrict        nrnb)
 301 {
 302     /* Suffixes 0,1,2,3 refer to particle indices for waters in the inner or outer loop, or
 303      * just 0 for non-waters.
 304      * Suffixes A,B refer to j loop unrolling done with double precision SIMD, e.g. for the two different
 305      * jnr indices corresponding to data put in the four positions in the SIMD register.
 306      */
 307     int              i_shift_offset,i_coord_offset,outeriter,inneriter;
 308     int              j_index_start,j_index_end,jidx,nri,inr,ggid,iidx;
 309     int              jnrA,jnrB;
 310     int              j_coord_offsetA,j_coord_offsetB;
 311     int              *iinr,*jindex,*jjnr,*shiftidx,*gid;
 312     real             rcutoff_scalar;
 313     real             *shiftvec,*fshift,*x,*f;
 314     _fjsp_v2r8       tx,ty,tz,fscal,rcutoff,rcutoff2,jidxall;
 315     int              vdwioffset0;
 316     _fjsp_v2r8       ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
 317     int              vdwjidx0A,vdwjidx0B;
 318     _fjsp_v2r8       jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
 319     _fjsp_v2r8       dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00;
 320     int              nvdwtype;
 321     _fjsp_v2r8       rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,sh_vdw_invrcut6;
 322     int              *vdwtype;
 323     real             *vdwparam;
 324     _fjsp_v2r8       one_sixth   = gmx_fjsp_set1_v2r8(1.0/6.0);
 325     _fjsp_v2r8       one_twelfth = gmx_fjsp_set1_v2r8(1.0/12.0);
 326     _fjsp_v2r8       itab_tmp;
 327     _fjsp_v2r8       dummy_mask,cutoff_mask;
 328     _fjsp_v2r8       one     = gmx_fjsp_set1_v2r8(1.0);
 329     _fjsp_v2r8       two     = gmx_fjsp_set1_v2r8(2.0);
 330     union { _fjsp_v2r8 simd; long long int i[2]; } vfconv,gbconv,ewconv;
 331
 332     x                = xx[0];
 333     f                = ff[0];
 334
 335     nri              = nlist->nri;
 336     iinr             = nlist->iinr;
 337     jindex           = nlist->jindex;
 338     jjnr             = nlist->jjnr;
 339     shiftidx         = nlist->shift;
 340     gid              = nlist->gid;
 341     shiftvec         = fr->shift_vec[0];
 342     fshift           = fr->fshift[0];
 343     nvdwtype         = fr->ntype;
 344     vdwparam         = fr->nbfp;
 345     vdwtype          = mdatoms->typeA;
 346
 347     /* Avoid stupid compiler warnings */
 348     jnrA = jnrB = 0;
 349     j_coord_offsetA = 0;
 350     j_coord_offsetB = 0;
 351
 352     outeriter        = 0;
 353     inneriter        = 0;
 354
 355     /* Start outer loop over neighborlists */
 356     for(iidx=0; iidx<nri; iidx++)
 357     {
 358         /* Load shift vector for this list */
 359         i_shift_offset   = DIM*shiftidx[iidx];
 360
 361         /* Load limits for loop over neighbors */
 362         j_index_start    = jindex[iidx];
 363         j_index_end      = jindex[iidx+1];
 364
 365         /* Get outer coordinate index */
 366         inr              = iinr[iidx];
 367         i_coord_offset   = DIM*inr;
 368
 369         /* Load i particle coords and add shift vector */
 370         gmx_fjsp_load_shift_and_1rvec_broadcast_v2r8(shiftvec+i_shift_offset,x+i_coord_offset,&ix0,&iy0,&iz0);
 371
 372         fix0             = _fjsp_setzero_v2r8();
 373         fiy0             = _fjsp_setzero_v2r8();
 374         fiz0             = _fjsp_setzero_v2r8();
 375
 376         /* Load parameters for i particles */
 377         vdwioffset0      = 2*nvdwtype*vdwtype[inr+0];
 378
 379         /* Start inner kernel loop */
 380         for(jidx=j_index_start; jidx<j_index_end-1; jidx+=2)
 381         {
 382
 383             /* Get j neighbor index, and coordinate index */
 384             jnrA             = jjnr[jidx];
 385             jnrB             = jjnr[jidx+1];
 386             j_coord_offsetA  = DIM*jnrA;
 387             j_coord_offsetB  = DIM*jnrB;
 388
 389             /* load j atom coordinates */
 390             gmx_fjsp_load_1rvec_2ptr_swizzle_v2r8(x+j_coord_offsetA,x+j_coord_offsetB,
 391                                               &jx0,&jy0,&jz0);
 392
 393             /* Calculate displacement vector */
 394             dx00             = _fjsp_sub_v2r8(ix0,jx0);
 395             dy00             = _fjsp_sub_v2r8(iy0,jy0);
 396             dz00             = _fjsp_sub_v2r8(iz0,jz0);
 397
 398             /* Calculate squared distance and things based on it */
 399             rsq00            = gmx_fjsp_calc_rsq_v2r8(dx00,dy00,dz00);
 400
 401             rinvsq00         = gmx_fjsp_inv_v2r8(rsq00);
 402
 403             /* Load parameters for j particles */
 404             vdwjidx0A        = 2*vdwtype[jnrA+0];
 405             vdwjidx0B        = 2*vdwtype[jnrB+0];
 406
 407             /**************************
 408              * CALCULATE INTERACTIONS *
 409              **************************/
 410
 411             /* Compute parameters for interactions between i and j atoms */
 412             gmx_fjsp_load_2pair_swizzle_v2r8(vdwparam+vdwioffset0+vdwjidx0A,
 413                                          vdwparam+vdwioffset0+vdwjidx0B,&c6_00,&c12_00);
 414
 415             /* LENNARD-JONES DISPERSION/REPULSION */
 416
 417             rinvsix          = _fjsp_mul_v2r8(_fjsp_mul_v2r8(rinvsq00,rinvsq00),rinvsq00);
 418             fvdw             = _fjsp_mul_v2r8(_fjsp_msub_v2r8(c12_00,rinvsix,c6_00),_fjsp_mul_v2r8(rinvsix,rinvsq00));
 419
 420             fscal            = fvdw;
 421
 422             /* Update vectorial force */
 423             fix0             = _fjsp_madd_v2r8(dx00,fscal,fix0);
 424             fiy0             = _fjsp_madd_v2r8(dy00,fscal,fiy0);
 425             fiz0             = _fjsp_madd_v2r8(dz00,fscal,fiz0);
 426
 427             gmx_fjsp_decrement_fma_1rvec_2ptr_swizzle_v2r8(f+j_coord_offsetA,f+j_coord_offsetB,fscal,dx00,dy00,dz00);
 428
 429             /* Inner loop uses 30 flops */
 430         }
 431
 432         if(jidx<j_index_end)
 433         {
 434
 435             jnrA             = jjnr[jidx];
 436             j_coord_offsetA  = DIM*jnrA;
 437
 438             /* load j atom coordinates */
 439             gmx_fjsp_load_1rvec_1ptr_swizzle_v2r8(x+j_coord_offsetA,
 440                                               &jx0,&jy0,&jz0);
 441
 442             /* Calculate displacement vector */
 443             dx00             = _fjsp_sub_v2r8(ix0,jx0);
 444             dy00             = _fjsp_sub_v2r8(iy0,jy0);
 445             dz00             = _fjsp_sub_v2r8(iz0,jz0);
 446
 447             /* Calculate squared distance and things based on it */
 448             rsq00            = gmx_fjsp_calc_rsq_v2r8(dx00,dy00,dz00);
 449
 450             rinvsq00         = gmx_fjsp_inv_v2r8(rsq00);
 451
 452             /* Load parameters for j particles */
 453             vdwjidx0A        = 2*vdwtype[jnrA+0];
 454
 455             /**************************
 456              * CALCULATE INTERACTIONS *
 457              **************************/
 458
 459             /* Compute parameters for interactions between i and j atoms */
 460             gmx_fjsp_load_1pair_swizzle_v2r8(vdwparam+vdwioffset0+vdwjidx0A,&c6_00,&c12_00);
 461
 462             /* LENNARD-JONES DISPERSION/REPULSION */
 463
 464             rinvsix          = _fjsp_mul_v2r8(_fjsp_mul_v2r8(rinvsq00,rinvsq00),rinvsq00);
 465             fvdw             = _fjsp_mul_v2r8(_fjsp_msub_v2r8(c12_00,rinvsix,c6_00),_fjsp_mul_v2r8(rinvsix,rinvsq00));
 466
 467             fscal            = fvdw;
 468
 469             fscal            = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
 470
 471             /* Update vectorial force */
 472             fix0             = _fjsp_madd_v2r8(dx00,fscal,fix0);
 473             fiy0             = _fjsp_madd_v2r8(dy00,fscal,fiy0);
 474             fiz0             = _fjsp_madd_v2r8(dz00,fscal,fiz0);
 475
 476             gmx_fjsp_decrement_fma_1rvec_1ptr_swizzle_v2r8(f+j_coord_offsetA,fscal,dx00,dy00,dz00);
 477
 478             /* Inner loop uses 30 flops */
 479         }
 480
 481         /* End of innermost loop */
 482
 483         gmx_fjsp_update_iforce_1atom_swizzle_v2r8(fix0,fiy0,fiz0,
 484                                               f+i_coord_offset,fshift+i_shift_offset);
 485
 486         /* Increment number of inner iterations */
 487         inneriter                  += j_index_end - j_index_start;
 488
 489         /* Outer loop uses 6 flops */
 490     }
 491
 492     /* Increment number of outer iterations */
 493     outeriter        += nri;
 494
 495     /* Update outer/inner flops */
 496
 497     inc_nrnb(nrnb,eNR_NBKERNEL_VDW_F,outeriter*6 + inneriter*30);
 498 }