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, 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 "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 "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                      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_1pair_swizzle_v2r8(vdwparam+vdwioffset0+vdwjidx0A,&c6_00,&c12_00);
 234
 235             /* LENNARD-JONES DISPERSION/REPULSION */
 236
 237             rinvsix          = _fjsp_mul_v2r8(_fjsp_mul_v2r8(rinvsq00,rinvsq00),rinvsq00);
 238             vvdw6            = _fjsp_mul_v2r8(c6_00,rinvsix);
 239             vvdw12           = _fjsp_mul_v2r8(c12_00,_fjsp_mul_v2r8(rinvsix,rinvsix));
 240             vvdw             = _fjsp_msub_v2r8( vvdw12,one_twelfth, _fjsp_mul_v2r8(vvdw6,one_sixth) );
 241             fvdw             = _fjsp_mul_v2r8(_fjsp_sub_v2r8(vvdw12,vvdw6),rinvsq00);
 242
 243             /* Update potential sum for this i atom from the interaction with this j atom. */
 244             vvdw             = _fjsp_unpacklo_v2r8(vvdw,_fjsp_setzero_v2r8());
 245             vvdwsum          = _fjsp_add_v2r8(vvdwsum,vvdw);
 246
 247             fscal            = fvdw;
 248
 249             fscal            = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
 250
 251             /* Update vectorial force */
 252             fix0             = _fjsp_madd_v2r8(dx00,fscal,fix0);
 253             fiy0             = _fjsp_madd_v2r8(dy00,fscal,fiy0);
 254             fiz0             = _fjsp_madd_v2r8(dz00,fscal,fiz0);
 255
 256             gmx_fjsp_decrement_fma_1rvec_1ptr_swizzle_v2r8(f+j_coord_offsetA,fscal,dx00,dy00,dz00);
 257
 258             /* Inner loop uses 35 flops */
 259         }
 260
 261         /* End of innermost loop */
 262
 263         gmx_fjsp_update_iforce_1atom_swizzle_v2r8(fix0,fiy0,fiz0,
 264                                               f+i_coord_offset,fshift+i_shift_offset);
 265
 266         ggid                        = gid[iidx];
 267         /* Update potential energies */
 268         gmx_fjsp_update_1pot_v2r8(vvdwsum,kernel_data->energygrp_vdw+ggid);
 269
 270         /* Increment number of inner iterations */
 271         inneriter                  += j_index_end - j_index_start;
 272
 273         /* Outer loop uses 7 flops */
 274     }
 275
 276     /* Increment number of outer iterations */
 277     outeriter        += nri;
 278
 279     /* Update outer/inner flops */
 280
 281     inc_nrnb(nrnb,eNR_NBKERNEL_VDW_VF,outeriter*7 + inneriter*35);
 282 }
 283 /*
 284  * Gromacs nonbonded kernel:   nb_kernel_ElecNone_VdwLJ_GeomP1P1_F_sparc64_hpc_ace_double
 285  * Electrostatics interaction: None
 286  * VdW interaction:            LennardJones
 287  * Geometry:                   Particle-Particle
 288  * Calculate force/pot:        Force
 289  */
 290 void
 291 nb_kernel_ElecNone_VdwLJ_GeomP1P1_F_sparc64_hpc_ace_double
 292                     (t_nblist                    * gmx_restrict       nlist,
 293                      rvec                        * gmx_restrict          xx,
 294                      rvec                        * gmx_restrict          ff,
 295                      t_forcerec                  * gmx_restrict          fr,
 296                      t_mdatoms                   * gmx_restrict     mdatoms,
 297                      nb_kernel_data_t gmx_unused * gmx_restrict kernel_data,
 298                      t_nrnb                      * gmx_restrict        nrnb)
 299 {
 300     /* Suffixes 0,1,2,3 refer to particle indices for waters in the inner or outer loop, or
 301      * just 0 for non-waters.
 302      * Suffixes A,B refer to j loop unrolling done with double precision SIMD, e.g. for the two different
 303      * jnr indices corresponding to data put in the four positions in the SIMD register.
 304      */
 305     int              i_shift_offset,i_coord_offset,outeriter,inneriter;
 306     int              j_index_start,j_index_end,jidx,nri,inr,ggid,iidx;
 307     int              jnrA,jnrB;
 308     int              j_coord_offsetA,j_coord_offsetB;
 309     int              *iinr,*jindex,*jjnr,*shiftidx,*gid;
 310     real             rcutoff_scalar;
 311     real             *shiftvec,*fshift,*x,*f;
 312     _fjsp_v2r8       tx,ty,tz,fscal,rcutoff,rcutoff2,jidxall;
 313     int              vdwioffset0;
 314     _fjsp_v2r8       ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
 315     int              vdwjidx0A,vdwjidx0B;
 316     _fjsp_v2r8       jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
 317     _fjsp_v2r8       dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00;
 318     int              nvdwtype;
 319     _fjsp_v2r8       rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,sh_vdw_invrcut6;
 320     int              *vdwtype;
 321     real             *vdwparam;
 322     _fjsp_v2r8       one_sixth   = gmx_fjsp_set1_v2r8(1.0/6.0);
 323     _fjsp_v2r8       one_twelfth = gmx_fjsp_set1_v2r8(1.0/12.0);
 324     _fjsp_v2r8       itab_tmp;
 325     _fjsp_v2r8       dummy_mask,cutoff_mask;
 326     _fjsp_v2r8       one     = gmx_fjsp_set1_v2r8(1.0);
 327     _fjsp_v2r8       two     = gmx_fjsp_set1_v2r8(2.0);
 328     union { _fjsp_v2r8 simd; long long int i[2]; } vfconv,gbconv,ewconv;
 329
 330     x                = xx[0];
 331     f                = ff[0];
 332
 333     nri              = nlist->nri;
 334     iinr             = nlist->iinr;
 335     jindex           = nlist->jindex;
 336     jjnr             = nlist->jjnr;
 337     shiftidx         = nlist->shift;
 338     gid              = nlist->gid;
 339     shiftvec         = fr->shift_vec[0];
 340     fshift           = fr->fshift[0];
 341     nvdwtype         = fr->ntype;
 342     vdwparam         = fr->nbfp;
 343     vdwtype          = mdatoms->typeA;
 344
 345     /* Avoid stupid compiler warnings */
 346     jnrA = jnrB = 0;
 347     j_coord_offsetA = 0;
 348     j_coord_offsetB = 0;
 349
 350     outeriter        = 0;
 351     inneriter        = 0;
 352
 353     /* Start outer loop over neighborlists */
 354     for(iidx=0; iidx<nri; iidx++)
 355     {
 356         /* Load shift vector for this list */
 357         i_shift_offset   = DIM*shiftidx[iidx];
 358
 359         /* Load limits for loop over neighbors */
 360         j_index_start    = jindex[iidx];
 361         j_index_end      = jindex[iidx+1];
 362
 363         /* Get outer coordinate index */
 364         inr              = iinr[iidx];
 365         i_coord_offset   = DIM*inr;
 366
 367         /* Load i particle coords and add shift vector */
 368         gmx_fjsp_load_shift_and_1rvec_broadcast_v2r8(shiftvec+i_shift_offset,x+i_coord_offset,&ix0,&iy0,&iz0);
 369
 370         fix0             = _fjsp_setzero_v2r8();
 371         fiy0             = _fjsp_setzero_v2r8();
 372         fiz0             = _fjsp_setzero_v2r8();
 373
 374         /* Load parameters for i particles */
 375         vdwioffset0      = 2*nvdwtype*vdwtype[inr+0];
 376
 377         /* Start inner kernel loop */
 378         for(jidx=j_index_start; jidx<j_index_end-1; jidx+=2)
 379         {
 380
 381             /* Get j neighbor index, and coordinate index */
 382             jnrA             = jjnr[jidx];
 383             jnrB             = jjnr[jidx+1];
 384             j_coord_offsetA  = DIM*jnrA;
 385             j_coord_offsetB  = DIM*jnrB;
 386
 387             /* load j atom coordinates */
 388             gmx_fjsp_load_1rvec_2ptr_swizzle_v2r8(x+j_coord_offsetA,x+j_coord_offsetB,
 389                                               &jx0,&jy0,&jz0);
 390
 391             /* Calculate displacement vector */
 392             dx00             = _fjsp_sub_v2r8(ix0,jx0);
 393             dy00             = _fjsp_sub_v2r8(iy0,jy0);
 394             dz00             = _fjsp_sub_v2r8(iz0,jz0);
 395
 396             /* Calculate squared distance and things based on it */
 397             rsq00            = gmx_fjsp_calc_rsq_v2r8(dx00,dy00,dz00);
 398
 399             rinvsq00         = gmx_fjsp_inv_v2r8(rsq00);
 400
 401             /* Load parameters for j particles */
 402             vdwjidx0A        = 2*vdwtype[jnrA+0];
 403             vdwjidx0B        = 2*vdwtype[jnrB+0];
 404
 405             /**************************
 406              * CALCULATE INTERACTIONS *
 407              **************************/
 408
 409             /* Compute parameters for interactions between i and j atoms */
 410             gmx_fjsp_load_2pair_swizzle_v2r8(vdwparam+vdwioffset0+vdwjidx0A,
 411                                          vdwparam+vdwioffset0+vdwjidx0B,&c6_00,&c12_00);
 412
 413             /* LENNARD-JONES DISPERSION/REPULSION */
 414
 415             rinvsix          = _fjsp_mul_v2r8(_fjsp_mul_v2r8(rinvsq00,rinvsq00),rinvsq00);
 416             fvdw             = _fjsp_mul_v2r8(_fjsp_msub_v2r8(c12_00,rinvsix,c6_00),_fjsp_mul_v2r8(rinvsix,rinvsq00));
 417
 418             fscal            = fvdw;
 419
 420             /* Update vectorial force */
 421             fix0             = _fjsp_madd_v2r8(dx00,fscal,fix0);
 422             fiy0             = _fjsp_madd_v2r8(dy00,fscal,fiy0);
 423             fiz0             = _fjsp_madd_v2r8(dz00,fscal,fiz0);
 424
 425             gmx_fjsp_decrement_fma_1rvec_2ptr_swizzle_v2r8(f+j_coord_offsetA,f+j_coord_offsetB,fscal,dx00,dy00,dz00);
 426
 427             /* Inner loop uses 30 flops */
 428         }
 429
 430         if(jidx<j_index_end)
 431         {
 432
 433             jnrA             = jjnr[jidx];
 434             j_coord_offsetA  = DIM*jnrA;
 435
 436             /* load j atom coordinates */
 437             gmx_fjsp_load_1rvec_1ptr_swizzle_v2r8(x+j_coord_offsetA,
 438                                               &jx0,&jy0,&jz0);
 439
 440             /* Calculate displacement vector */
 441             dx00             = _fjsp_sub_v2r8(ix0,jx0);
 442             dy00             = _fjsp_sub_v2r8(iy0,jy0);
 443             dz00             = _fjsp_sub_v2r8(iz0,jz0);
 444
 445             /* Calculate squared distance and things based on it */
 446             rsq00            = gmx_fjsp_calc_rsq_v2r8(dx00,dy00,dz00);
 447
 448             rinvsq00         = gmx_fjsp_inv_v2r8(rsq00);
 449
 450             /* Load parameters for j particles */
 451             vdwjidx0A        = 2*vdwtype[jnrA+0];
 452
 453             /**************************
 454              * CALCULATE INTERACTIONS *
 455              **************************/
 456
 457             /* Compute parameters for interactions between i and j atoms */
 458             gmx_fjsp_load_1pair_swizzle_v2r8(vdwparam+vdwioffset0+vdwjidx0A,&c6_00,&c12_00);
 459
 460             /* LENNARD-JONES DISPERSION/REPULSION */
 461
 462             rinvsix          = _fjsp_mul_v2r8(_fjsp_mul_v2r8(rinvsq00,rinvsq00),rinvsq00);
 463             fvdw             = _fjsp_mul_v2r8(_fjsp_msub_v2r8(c12_00,rinvsix,c6_00),_fjsp_mul_v2r8(rinvsix,rinvsq00));
 464
 465             fscal            = fvdw;
 466
 467             fscal            = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
 468
 469             /* Update vectorial force */
 470             fix0             = _fjsp_madd_v2r8(dx00,fscal,fix0);
 471             fiy0             = _fjsp_madd_v2r8(dy00,fscal,fiy0);
 472             fiz0             = _fjsp_madd_v2r8(dz00,fscal,fiz0);
 473
 474             gmx_fjsp_decrement_fma_1rvec_1ptr_swizzle_v2r8(f+j_coord_offsetA,fscal,dx00,dy00,dz00);
 475
 476             /* Inner loop uses 30 flops */
 477         }
 478
 479         /* End of innermost loop */
 480
 481         gmx_fjsp_update_iforce_1atom_swizzle_v2r8(fix0,fiy0,fiz0,
 482                                               f+i_coord_offset,fshift+i_shift_offset);
 483
 484         /* Increment number of inner iterations */
 485         inneriter                  += j_index_end - j_index_start;
 486
 487         /* Outer loop uses 6 flops */
 488     }
 489
 490     /* Increment number of outer iterations */
 491     outeriter        += nri;
 492
 493     /* Update outer/inner flops */
 494
 495     inc_nrnb(nrnb,eNR_NBKERNEL_VDW_F,outeriter*6 + inneriter*30);
 496 }