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) 2012,2013,2014,2015,2017, 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 #include "gmxpre.h"
  39
  40 #include "config.h"
  41
  42 #include <math.h>
  43
  44 #include "../nb_kernel.h"
  45 #include "gromacs/gmxlib/nrnb.h"
  46
  47 #include "kernelutil_sparc64_hpc_ace_double.h"
  48
  49 /*
  50  * Gromacs nonbonded kernel:   nb_kernel_ElecNone_VdwLJ_GeomP1P1_VF_sparc64_hpc_ace_double
  51  * Electrostatics interaction: None
  52  * VdW interaction:            LennardJones
  53  * Geometry:                   Particle-Particle
  54  * Calculate force/pot:        PotentialAndForce
  55  */
  56 void
  57 nb_kernel_ElecNone_VdwLJ_GeomP1P1_VF_sparc64_hpc_ace_double
  58                     (t_nblist                    * gmx_restrict       nlist,
  59                      rvec                        * gmx_restrict          xx,
  60                      rvec                        * gmx_restrict          ff,
  61                      struct t_forcerec           * gmx_restrict          fr,
  62                      t_mdatoms                   * gmx_restrict     mdatoms,
  63                      nb_kernel_data_t gmx_unused * gmx_restrict kernel_data,
  64                      t_nrnb                      * gmx_restrict        nrnb)
  65 {
  66     /* Suffixes 0,1,2,3 refer to particle indices for waters in the inner or outer loop, or
  67      * just 0 for non-waters.
  68      * Suffixes A,B refer to j loop unrolling done with double precision SIMD, e.g. for the two different
  69      * jnr indices corresponding to data put in the four positions in the SIMD register.
  70      */
  71     int              i_shift_offset,i_coord_offset,outeriter,inneriter;
  72     int              j_index_start,j_index_end,jidx,nri,inr,ggid,iidx;
  73     int              jnrA,jnrB;
  74     int              j_coord_offsetA,j_coord_offsetB;
  75     int              *iinr,*jindex,*jjnr,*shiftidx,*gid;
  76     real             rcutoff_scalar;
  77     real             *shiftvec,*fshift,*x,*f;
  78     _fjsp_v2r8       tx,ty,tz,fscal,rcutoff,rcutoff2,jidxall;
  79     int              vdwioffset0;
  80     _fjsp_v2r8       ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
  81     int              vdwjidx0A,vdwjidx0B;
  82     _fjsp_v2r8       jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
  83     _fjsp_v2r8       dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00;
  84     int              nvdwtype;
  85     _fjsp_v2r8       rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,sh_vdw_invrcut6;
  86     int              *vdwtype;
  87     real             *vdwparam;
  88     _fjsp_v2r8       one_sixth   = gmx_fjsp_set1_v2r8(1.0/6.0);
  89     _fjsp_v2r8       one_twelfth = gmx_fjsp_set1_v2r8(1.0/12.0);
  90     _fjsp_v2r8       itab_tmp;
  91     _fjsp_v2r8       dummy_mask,cutoff_mask;
  92     _fjsp_v2r8       one     = gmx_fjsp_set1_v2r8(1.0);
  93     _fjsp_v2r8       two     = gmx_fjsp_set1_v2r8(2.0);
  94     union { _fjsp_v2r8 simd; long long int i[2]; } vfconv,gbconv,ewconv;
  95
  96     x                = xx[0];
  97     f                = ff[0];
  98
  99     nri              = nlist->nri;
 100     iinr             = nlist->iinr;
 101     jindex           = nlist->jindex;
 102     jjnr             = nlist->jjnr;
 103     shiftidx         = nlist->shift;
 104     gid              = nlist->gid;
 105     shiftvec         = fr->shift_vec[0];
 106     fshift           = fr->fshift[0];
 107     nvdwtype         = fr->ntype;
 108     vdwparam         = fr->nbfp;
 109     vdwtype          = mdatoms->typeA;
 110
 111     /* Avoid stupid compiler warnings */
 112     jnrA = jnrB = 0;
 113     j_coord_offsetA = 0;
 114     j_coord_offsetB = 0;
 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
 125         /* Load limits for loop over neighbors */
 126         j_index_start    = jindex[iidx];
 127         j_index_end      = jindex[iidx+1];
 128
 129         /* Get outer coordinate index */
 130         inr              = iinr[iidx];
 131         i_coord_offset   = DIM*inr;
 132
 133         /* Load i particle coords and add shift vector */
 134         gmx_fjsp_load_shift_and_1rvec_broadcast_v2r8(shiftvec+i_shift_offset,x+i_coord_offset,&ix0,&iy0,&iz0);
 135
 136         fix0             = _fjsp_setzero_v2r8();
 137         fiy0             = _fjsp_setzero_v2r8();
 138         fiz0             = _fjsp_setzero_v2r8();
 139
 140         /* Load parameters for i particles */
 141         vdwioffset0      = 2*nvdwtype*vdwtype[inr+0];
 142
 143         /* Reset potential sums */
 144         vvdwsum          = _fjsp_setzero_v2r8();
 145
 146         /* Start inner kernel loop */
 147         for(jidx=j_index_start; jidx<j_index_end-1; jidx+=2)
 148         {
 149
 150             /* Get j neighbor index, and coordinate index */
 151             jnrA             = jjnr[jidx];
 152             jnrB             = jjnr[jidx+1];
 153             j_coord_offsetA  = DIM*jnrA;
 154             j_coord_offsetB  = DIM*jnrB;
 155
 156             /* load j atom coordinates */
 157             gmx_fjsp_load_1rvec_2ptr_swizzle_v2r8(x+j_coord_offsetA,x+j_coord_offsetB,
 158                                               &jx0,&jy0,&jz0);
 159
 160             /* Calculate displacement vector */
 161             dx00             = _fjsp_sub_v2r8(ix0,jx0);
 162             dy00             = _fjsp_sub_v2r8(iy0,jy0);
 163             dz00             = _fjsp_sub_v2r8(iz0,jz0);
 164
 165             /* Calculate squared distance and things based on it */
 166             rsq00            = gmx_fjsp_calc_rsq_v2r8(dx00,dy00,dz00);
 167
 168             rinvsq00         = gmx_fjsp_inv_v2r8(rsq00);
 169
 170             /* Load parameters for j particles */
 171             vdwjidx0A        = 2*vdwtype[jnrA+0];
 172             vdwjidx0B        = 2*vdwtype[jnrB+0];
 173
 174             /**************************
 175              * CALCULATE INTERACTIONS *
 176              **************************/
 177
 178             /* Compute parameters for interactions between i and j atoms */
 179             gmx_fjsp_load_2pair_swizzle_v2r8(vdwparam+vdwioffset0+vdwjidx0A,
 180                                          vdwparam+vdwioffset0+vdwjidx0B,&c6_00,&c12_00);
 181
 182             /* LENNARD-JONES DISPERSION/REPULSION */
 183
 184             rinvsix          = _fjsp_mul_v2r8(_fjsp_mul_v2r8(rinvsq00,rinvsq00),rinvsq00);
 185             vvdw6            = _fjsp_mul_v2r8(c6_00,rinvsix);
 186             vvdw12           = _fjsp_mul_v2r8(c12_00,_fjsp_mul_v2r8(rinvsix,rinvsix));
 187             vvdw             = _fjsp_msub_v2r8( vvdw12,one_twelfth, _fjsp_mul_v2r8(vvdw6,one_sixth) );
 188             fvdw             = _fjsp_mul_v2r8(_fjsp_sub_v2r8(vvdw12,vvdw6),rinvsq00);
 189
 190             /* Update potential sum for this i atom from the interaction with this j atom. */
 191             vvdwsum          = _fjsp_add_v2r8(vvdwsum,vvdw);
 192
 193             fscal            = fvdw;
 194
 195             /* Update vectorial force */
 196             fix0             = _fjsp_madd_v2r8(dx00,fscal,fix0);
 197             fiy0             = _fjsp_madd_v2r8(dy00,fscal,fiy0);
 198             fiz0             = _fjsp_madd_v2r8(dz00,fscal,fiz0);
 199
 200             gmx_fjsp_decrement_fma_1rvec_2ptr_swizzle_v2r8(f+j_coord_offsetA,f+j_coord_offsetB,fscal,dx00,dy00,dz00);
 201
 202             /* Inner loop uses 35 flops */
 203         }
 204
 205         if(jidx<j_index_end)
 206         {
 207
 208             jnrA             = jjnr[jidx];
 209             j_coord_offsetA  = DIM*jnrA;
 210
 211             /* load j atom coordinates */
 212             gmx_fjsp_load_1rvec_1ptr_swizzle_v2r8(x+j_coord_offsetA,
 213                                               &jx0,&jy0,&jz0);
 214
 215             /* Calculate displacement vector */
 216             dx00             = _fjsp_sub_v2r8(ix0,jx0);
 217             dy00             = _fjsp_sub_v2r8(iy0,jy0);
 218             dz00             = _fjsp_sub_v2r8(iz0,jz0);
 219
 220             /* Calculate squared distance and things based on it */
 221             rsq00            = gmx_fjsp_calc_rsq_v2r8(dx00,dy00,dz00);
 222
 223             rinvsq00         = gmx_fjsp_inv_v2r8(rsq00);
 224
 225             /* Load parameters for j particles */
 226             vdwjidx0A        = 2*vdwtype[jnrA+0];
 227
 228             /**************************
 229              * CALCULATE INTERACTIONS *
 230              **************************/
 231
 232             /* Compute parameters for interactions between i and j atoms */
 233             gmx_fjsp_load_2pair_swizzle_v2r8(vdwparam+vdwioffset0+vdwjidx0A,
 234                                          vdwparam+vdwioffset0+vdwjidx0B,&c6_00,&c12_00);
 235
 236             /* LENNARD-JONES DISPERSION/REPULSION */
 237
 238             rinvsix          = _fjsp_mul_v2r8(_fjsp_mul_v2r8(rinvsq00,rinvsq00),rinvsq00);
 239             vvdw6            = _fjsp_mul_v2r8(c6_00,rinvsix);
 240             vvdw12           = _fjsp_mul_v2r8(c12_00,_fjsp_mul_v2r8(rinvsix,rinvsix));
 241             vvdw             = _fjsp_msub_v2r8( vvdw12,one_twelfth, _fjsp_mul_v2r8(vvdw6,one_sixth) );
 242             fvdw             = _fjsp_mul_v2r8(_fjsp_sub_v2r8(vvdw12,vvdw6),rinvsq00);
 243
 244             /* Update potential sum for this i atom from the interaction with this j atom. */
 245             vvdw             = _fjsp_unpacklo_v2r8(vvdw,_fjsp_setzero_v2r8());
 246             vvdwsum          = _fjsp_add_v2r8(vvdwsum,vvdw);
 247
 248             fscal            = fvdw;
 249
 250             fscal            = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
 251
 252             /* Update vectorial force */
 253             fix0             = _fjsp_madd_v2r8(dx00,fscal,fix0);
 254             fiy0             = _fjsp_madd_v2r8(dy00,fscal,fiy0);
 255             fiz0             = _fjsp_madd_v2r8(dz00,fscal,fiz0);
 256
 257             gmx_fjsp_decrement_fma_1rvec_1ptr_swizzle_v2r8(f+j_coord_offsetA,fscal,dx00,dy00,dz00);
 258
 259             /* Inner loop uses 35 flops */
 260         }
 261
 262         /* End of innermost loop */
 263
 264         gmx_fjsp_update_iforce_1atom_swizzle_v2r8(fix0,fiy0,fiz0,
 265                                               f+i_coord_offset,fshift+i_shift_offset);
 266
 267         ggid                        = gid[iidx];
 268         /* Update potential energies */
 269         gmx_fjsp_update_1pot_v2r8(vvdwsum,kernel_data->energygrp_vdw+ggid);
 270
 271         /* Increment number of inner iterations */
 272         inneriter                  += j_index_end - j_index_start;
 273
 274         /* Outer loop uses 7 flops */
 275     }
 276
 277     /* Increment number of outer iterations */
 278     outeriter        += nri;
 279
 280     /* Update outer/inner flops */
 281
 282     inc_nrnb(nrnb,eNR_NBKERNEL_VDW_VF,outeriter*7 + inneriter*35);
 283 }
 284 /*
 285  * Gromacs nonbonded kernel:   nb_kernel_ElecNone_VdwLJ_GeomP1P1_F_sparc64_hpc_ace_double
 286  * Electrostatics interaction: None
 287  * VdW interaction:            LennardJones
 288  * Geometry:                   Particle-Particle
 289  * Calculate force/pot:        Force
 290  */
 291 void
 292 nb_kernel_ElecNone_VdwLJ_GeomP1P1_F_sparc64_hpc_ace_double
 293                     (t_nblist                    * gmx_restrict       nlist,
 294                      rvec                        * gmx_restrict          xx,
 295                      rvec                        * gmx_restrict          ff,
 296                      struct t_forcerec           * gmx_restrict          fr,
 297                      t_mdatoms                   * gmx_restrict     mdatoms,
 298                      nb_kernel_data_t gmx_unused * gmx_restrict kernel_data,
 299                      t_nrnb                      * gmx_restrict        nrnb)
 300 {
 301     /* Suffixes 0,1,2,3 refer to particle indices for waters in the inner or outer loop, or
 302      * just 0 for non-waters.
 303      * Suffixes A,B refer to j loop unrolling done with double precision SIMD, e.g. for the two different
 304      * jnr indices corresponding to data put in the four positions in the SIMD register.
 305      */
 306     int              i_shift_offset,i_coord_offset,outeriter,inneriter;
 307     int              j_index_start,j_index_end,jidx,nri,inr,ggid,iidx;
 308     int              jnrA,jnrB;
 309     int              j_coord_offsetA,j_coord_offsetB;
 310     int              *iinr,*jindex,*jjnr,*shiftidx,*gid;
 311     real             rcutoff_scalar;
 312     real             *shiftvec,*fshift,*x,*f;
 313     _fjsp_v2r8       tx,ty,tz,fscal,rcutoff,rcutoff2,jidxall;
 314     int              vdwioffset0;
 315     _fjsp_v2r8       ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
 316     int              vdwjidx0A,vdwjidx0B;
 317     _fjsp_v2r8       jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
 318     _fjsp_v2r8       dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00;
 319     int              nvdwtype;
 320     _fjsp_v2r8       rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,sh_vdw_invrcut6;
 321     int              *vdwtype;
 322     real             *vdwparam;
 323     _fjsp_v2r8       one_sixth   = gmx_fjsp_set1_v2r8(1.0/6.0);
 324     _fjsp_v2r8       one_twelfth = gmx_fjsp_set1_v2r8(1.0/12.0);
 325     _fjsp_v2r8       itab_tmp;
 326     _fjsp_v2r8       dummy_mask,cutoff_mask;
 327     _fjsp_v2r8       one     = gmx_fjsp_set1_v2r8(1.0);
 328     _fjsp_v2r8       two     = gmx_fjsp_set1_v2r8(2.0);
 329     union { _fjsp_v2r8 simd; long long int i[2]; } vfconv,gbconv,ewconv;
 330
 331     x                = xx[0];
 332     f                = ff[0];
 333
 334     nri              = nlist->nri;
 335     iinr             = nlist->iinr;
 336     jindex           = nlist->jindex;
 337     jjnr             = nlist->jjnr;
 338     shiftidx         = nlist->shift;
 339     gid              = nlist->gid;
 340     shiftvec         = fr->shift_vec[0];
 341     fshift           = fr->fshift[0];
 342     nvdwtype         = fr->ntype;
 343     vdwparam         = fr->nbfp;
 344     vdwtype          = mdatoms->typeA;
 345
 346     /* Avoid stupid compiler warnings */
 347     jnrA = jnrB = 0;
 348     j_coord_offsetA = 0;
 349     j_coord_offsetB = 0;
 350
 351     outeriter        = 0;
 352     inneriter        = 0;
 353
 354     /* Start outer loop over neighborlists */
 355     for(iidx=0; iidx<nri; iidx++)
 356     {
 357         /* Load shift vector for this list */
 358         i_shift_offset   = DIM*shiftidx[iidx];
 359
 360         /* Load limits for loop over neighbors */
 361         j_index_start    = jindex[iidx];
 362         j_index_end      = jindex[iidx+1];
 363
 364         /* Get outer coordinate index */
 365         inr              = iinr[iidx];
 366         i_coord_offset   = DIM*inr;
 367
 368         /* Load i particle coords and add shift vector */
 369         gmx_fjsp_load_shift_and_1rvec_broadcast_v2r8(shiftvec+i_shift_offset,x+i_coord_offset,&ix0,&iy0,&iz0);
 370
 371         fix0             = _fjsp_setzero_v2r8();
 372         fiy0             = _fjsp_setzero_v2r8();
 373         fiz0             = _fjsp_setzero_v2r8();
 374
 375         /* Load parameters for i particles */
 376         vdwioffset0      = 2*nvdwtype*vdwtype[inr+0];
 377
 378         /* Start inner kernel loop */
 379         for(jidx=j_index_start; jidx<j_index_end-1; jidx+=2)
 380         {
 381
 382             /* Get j neighbor index, and coordinate index */
 383             jnrA             = jjnr[jidx];
 384             jnrB             = jjnr[jidx+1];
 385             j_coord_offsetA  = DIM*jnrA;
 386             j_coord_offsetB  = DIM*jnrB;
 387
 388             /* load j atom coordinates */
 389             gmx_fjsp_load_1rvec_2ptr_swizzle_v2r8(x+j_coord_offsetA,x+j_coord_offsetB,
 390                                               &jx0,&jy0,&jz0);
 391
 392             /* Calculate displacement vector */
 393             dx00             = _fjsp_sub_v2r8(ix0,jx0);
 394             dy00             = _fjsp_sub_v2r8(iy0,jy0);
 395             dz00             = _fjsp_sub_v2r8(iz0,jz0);
 396
 397             /* Calculate squared distance and things based on it */
 398             rsq00            = gmx_fjsp_calc_rsq_v2r8(dx00,dy00,dz00);
 399
 400             rinvsq00         = gmx_fjsp_inv_v2r8(rsq00);
 401
 402             /* Load parameters for j particles */
 403             vdwjidx0A        = 2*vdwtype[jnrA+0];
 404             vdwjidx0B        = 2*vdwtype[jnrB+0];
 405
 406             /**************************
 407              * CALCULATE INTERACTIONS *
 408              **************************/
 409
 410             /* Compute parameters for interactions between i and j atoms */
 411             gmx_fjsp_load_2pair_swizzle_v2r8(vdwparam+vdwioffset0+vdwjidx0A,
 412                                          vdwparam+vdwioffset0+vdwjidx0B,&c6_00,&c12_00);
 413
 414             /* LENNARD-JONES DISPERSION/REPULSION */
 415
 416             rinvsix          = _fjsp_mul_v2r8(_fjsp_mul_v2r8(rinvsq00,rinvsq00),rinvsq00);
 417             fvdw             = _fjsp_mul_v2r8(_fjsp_msub_v2r8(c12_00,rinvsix,c6_00),_fjsp_mul_v2r8(rinvsix,rinvsq00));
 418
 419             fscal            = fvdw;
 420
 421             /* Update vectorial force */
 422             fix0             = _fjsp_madd_v2r8(dx00,fscal,fix0);
 423             fiy0             = _fjsp_madd_v2r8(dy00,fscal,fiy0);
 424             fiz0             = _fjsp_madd_v2r8(dz00,fscal,fiz0);
 425
 426             gmx_fjsp_decrement_fma_1rvec_2ptr_swizzle_v2r8(f+j_coord_offsetA,f+j_coord_offsetB,fscal,dx00,dy00,dz00);
 427
 428             /* Inner loop uses 30 flops */
 429         }
 430
 431         if(jidx<j_index_end)
 432         {
 433
 434             jnrA             = jjnr[jidx];
 435             j_coord_offsetA  = DIM*jnrA;
 436
 437             /* load j atom coordinates */
 438             gmx_fjsp_load_1rvec_1ptr_swizzle_v2r8(x+j_coord_offsetA,
 439                                               &jx0,&jy0,&jz0);
 440
 441             /* Calculate displacement vector */
 442             dx00             = _fjsp_sub_v2r8(ix0,jx0);
 443             dy00             = _fjsp_sub_v2r8(iy0,jy0);
 444             dz00             = _fjsp_sub_v2r8(iz0,jz0);
 445
 446             /* Calculate squared distance and things based on it */
 447             rsq00            = gmx_fjsp_calc_rsq_v2r8(dx00,dy00,dz00);
 448
 449             rinvsq00         = gmx_fjsp_inv_v2r8(rsq00);
 450
 451             /* Load parameters for j particles */
 452             vdwjidx0A        = 2*vdwtype[jnrA+0];
 453
 454             /**************************
 455              * CALCULATE INTERACTIONS *
 456              **************************/
 457
 458             /* Compute parameters for interactions between i and j atoms */
 459             gmx_fjsp_load_2pair_swizzle_v2r8(vdwparam+vdwioffset0+vdwjidx0A,
 460                                          vdwparam+vdwioffset0+vdwjidx0B,&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 }