src/gromacs/gmxlib/nonbonded/nb_kernel_sparc64_hpc_ace_double/nb_kernel_ElecRF_VdwCSTab_GeomW3W3_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_ElecRF_VdwCSTab_GeomW3W3_VF_sparc64_hpc_ace_double
  53  * Electrostatics interaction: ReactionField
  54  * VdW interaction:            CubicSplineTable
  55  * Geometry:                   Water3-Water3
  56  * Calculate force/pot:        PotentialAndForce
  57  */
  58 void
  59 nb_kernel_ElecRF_VdwCSTab_GeomW3W3_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              vdwioffset1;
  84     _fjsp_v2r8       ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
  85     int              vdwioffset2;
  86     _fjsp_v2r8       ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
  87     int              vdwjidx0A,vdwjidx0B;
  88     _fjsp_v2r8       jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
  89     int              vdwjidx1A,vdwjidx1B;
  90     _fjsp_v2r8       jx1,jy1,jz1,fjx1,fjy1,fjz1,jq1,isaj1;
  91     int              vdwjidx2A,vdwjidx2B;
  92     _fjsp_v2r8       jx2,jy2,jz2,fjx2,fjy2,fjz2,jq2,isaj2;
  93     _fjsp_v2r8       dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00;
  94     _fjsp_v2r8       dx01,dy01,dz01,rsq01,rinv01,rinvsq01,r01,qq01,c6_01,c12_01;
  95     _fjsp_v2r8       dx02,dy02,dz02,rsq02,rinv02,rinvsq02,r02,qq02,c6_02,c12_02;
  96     _fjsp_v2r8       dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10;
  97     _fjsp_v2r8       dx11,dy11,dz11,rsq11,rinv11,rinvsq11,r11,qq11,c6_11,c12_11;
  98     _fjsp_v2r8       dx12,dy12,dz12,rsq12,rinv12,rinvsq12,r12,qq12,c6_12,c12_12;
  99     _fjsp_v2r8       dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20;
 100     _fjsp_v2r8       dx21,dy21,dz21,rsq21,rinv21,rinvsq21,r21,qq21,c6_21,c12_21;
 101     _fjsp_v2r8       dx22,dy22,dz22,rsq22,rinv22,rinvsq22,r22,qq22,c6_22,c12_22;
 102     _fjsp_v2r8       velec,felec,velecsum,facel,crf,krf,krf2;
 103     real             *charge;
 104     int              nvdwtype;
 105     _fjsp_v2r8       rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,sh_vdw_invrcut6;
 106     int              *vdwtype;
 107     real             *vdwparam;
 108     _fjsp_v2r8       one_sixth   = gmx_fjsp_set1_v2r8(1.0/6.0);
 109     _fjsp_v2r8       one_twelfth = gmx_fjsp_set1_v2r8(1.0/12.0);
 110     _fjsp_v2r8       rt,vfeps,vftabscale,Y,F,G,H,Heps,Fp,VV,FF,twovfeps;
 111     real             *vftab;
 112     _fjsp_v2r8       itab_tmp;
 113     _fjsp_v2r8       dummy_mask,cutoff_mask;
 114     _fjsp_v2r8       one     = gmx_fjsp_set1_v2r8(1.0);
 115     _fjsp_v2r8       two     = gmx_fjsp_set1_v2r8(2.0);
 116     union { _fjsp_v2r8 simd; long long int i[2]; } vfconv,gbconv,ewconv;
 117
 118     x                = xx[0];
 119     f                = ff[0];
 120
 121     nri              = nlist->nri;
 122     iinr             = nlist->iinr;
 123     jindex           = nlist->jindex;
 124     jjnr             = nlist->jjnr;
 125     shiftidx         = nlist->shift;
 126     gid              = nlist->gid;
 127     shiftvec         = fr->shift_vec[0];
 128     fshift           = fr->fshift[0];
 129     facel            = gmx_fjsp_set1_v2r8(fr->epsfac);
 130     charge           = mdatoms->chargeA;
 131     krf              = gmx_fjsp_set1_v2r8(fr->ic->k_rf);
 132     krf2             = gmx_fjsp_set1_v2r8(fr->ic->k_rf*2.0);
 133     crf              = gmx_fjsp_set1_v2r8(fr->ic->c_rf);
 134     nvdwtype         = fr->ntype;
 135     vdwparam         = fr->nbfp;
 136     vdwtype          = mdatoms->typeA;
 137
 138     vftab            = kernel_data->table_vdw->data;
 139     vftabscale       = gmx_fjsp_set1_v2r8(kernel_data->table_vdw->scale);
 140
 141     /* Setup water-specific parameters */
 142     inr              = nlist->iinr[0];
 143     iq0              = _fjsp_mul_v2r8(facel,gmx_fjsp_set1_v2r8(charge[inr+0]));
 144     iq1              = _fjsp_mul_v2r8(facel,gmx_fjsp_set1_v2r8(charge[inr+1]));
 145     iq2              = _fjsp_mul_v2r8(facel,gmx_fjsp_set1_v2r8(charge[inr+2]));
 146     vdwioffset0      = 2*nvdwtype*vdwtype[inr+0];
 147
 148     jq0              = gmx_fjsp_set1_v2r8(charge[inr+0]);
 149     jq1              = gmx_fjsp_set1_v2r8(charge[inr+1]);
 150     jq2              = gmx_fjsp_set1_v2r8(charge[inr+2]);
 151     vdwjidx0A        = 2*vdwtype[inr+0];
 152     qq00             = _fjsp_mul_v2r8(iq0,jq0);
 153     c6_00            = gmx_fjsp_set1_v2r8(vdwparam[vdwioffset0+vdwjidx0A]);
 154     c12_00           = gmx_fjsp_set1_v2r8(vdwparam[vdwioffset0+vdwjidx0A+1]);
 155     qq01             = _fjsp_mul_v2r8(iq0,jq1);
 156     qq02             = _fjsp_mul_v2r8(iq0,jq2);
 157     qq10             = _fjsp_mul_v2r8(iq1,jq0);
 158     qq11             = _fjsp_mul_v2r8(iq1,jq1);
 159     qq12             = _fjsp_mul_v2r8(iq1,jq2);
 160     qq20             = _fjsp_mul_v2r8(iq2,jq0);
 161     qq21             = _fjsp_mul_v2r8(iq2,jq1);
 162     qq22             = _fjsp_mul_v2r8(iq2,jq2);
 163
 164     /* Avoid stupid compiler warnings */
 165     jnrA = jnrB = 0;
 166     j_coord_offsetA = 0;
 167     j_coord_offsetB = 0;
 168
 169     outeriter        = 0;
 170     inneriter        = 0;
 171
 172     /* Start outer loop over neighborlists */
 173     for(iidx=0; iidx<nri; iidx++)
 174     {
 175         /* Load shift vector for this list */
 176         i_shift_offset   = DIM*shiftidx[iidx];
 177
 178         /* Load limits for loop over neighbors */
 179         j_index_start    = jindex[iidx];
 180         j_index_end      = jindex[iidx+1];
 181
 182         /* Get outer coordinate index */
 183         inr              = iinr[iidx];
 184         i_coord_offset   = DIM*inr;
 185
 186         /* Load i particle coords and add shift vector */
 187         gmx_fjsp_load_shift_and_3rvec_broadcast_v2r8(shiftvec+i_shift_offset,x+i_coord_offset,
 188                                                  &ix0,&iy0,&iz0,&ix1,&iy1,&iz1,&ix2,&iy2,&iz2);
 189
 190         fix0             = _fjsp_setzero_v2r8();
 191         fiy0             = _fjsp_setzero_v2r8();
 192         fiz0             = _fjsp_setzero_v2r8();
 193         fix1             = _fjsp_setzero_v2r8();
 194         fiy1             = _fjsp_setzero_v2r8();
 195         fiz1             = _fjsp_setzero_v2r8();
 196         fix2             = _fjsp_setzero_v2r8();
 197         fiy2             = _fjsp_setzero_v2r8();
 198         fiz2             = _fjsp_setzero_v2r8();
 199
 200         /* Reset potential sums */
 201         velecsum         = _fjsp_setzero_v2r8();
 202         vvdwsum          = _fjsp_setzero_v2r8();
 203
 204         /* Start inner kernel loop */
 205         for(jidx=j_index_start; jidx<j_index_end-1; jidx+=2)
 206         {
 207
 208             /* Get j neighbor index, and coordinate index */
 209             jnrA             = jjnr[jidx];
 210             jnrB             = jjnr[jidx+1];
 211             j_coord_offsetA  = DIM*jnrA;
 212             j_coord_offsetB  = DIM*jnrB;
 213
 214             /* load j atom coordinates */
 215             gmx_fjsp_load_3rvec_2ptr_swizzle_v2r8(x+j_coord_offsetA,x+j_coord_offsetB,
 216                                               &jx0,&jy0,&jz0,&jx1,&jy1,&jz1,&jx2,&jy2,&jz2);
 217
 218             /* Calculate displacement vector */
 219             dx00             = _fjsp_sub_v2r8(ix0,jx0);
 220             dy00             = _fjsp_sub_v2r8(iy0,jy0);
 221             dz00             = _fjsp_sub_v2r8(iz0,jz0);
 222             dx01             = _fjsp_sub_v2r8(ix0,jx1);
 223             dy01             = _fjsp_sub_v2r8(iy0,jy1);
 224             dz01             = _fjsp_sub_v2r8(iz0,jz1);
 225             dx02             = _fjsp_sub_v2r8(ix0,jx2);
 226             dy02             = _fjsp_sub_v2r8(iy0,jy2);
 227             dz02             = _fjsp_sub_v2r8(iz0,jz2);
 228             dx10             = _fjsp_sub_v2r8(ix1,jx0);
 229             dy10             = _fjsp_sub_v2r8(iy1,jy0);
 230             dz10             = _fjsp_sub_v2r8(iz1,jz0);
 231             dx11             = _fjsp_sub_v2r8(ix1,jx1);
 232             dy11             = _fjsp_sub_v2r8(iy1,jy1);
 233             dz11             = _fjsp_sub_v2r8(iz1,jz1);
 234             dx12             = _fjsp_sub_v2r8(ix1,jx2);
 235             dy12             = _fjsp_sub_v2r8(iy1,jy2);
 236             dz12             = _fjsp_sub_v2r8(iz1,jz2);
 237             dx20             = _fjsp_sub_v2r8(ix2,jx0);
 238             dy20             = _fjsp_sub_v2r8(iy2,jy0);
 239             dz20             = _fjsp_sub_v2r8(iz2,jz0);
 240             dx21             = _fjsp_sub_v2r8(ix2,jx1);
 241             dy21             = _fjsp_sub_v2r8(iy2,jy1);
 242             dz21             = _fjsp_sub_v2r8(iz2,jz1);
 243             dx22             = _fjsp_sub_v2r8(ix2,jx2);
 244             dy22             = _fjsp_sub_v2r8(iy2,jy2);
 245             dz22             = _fjsp_sub_v2r8(iz2,jz2);
 246
 247             /* Calculate squared distance and things based on it */
 248             rsq00            = gmx_fjsp_calc_rsq_v2r8(dx00,dy00,dz00);
 249             rsq01            = gmx_fjsp_calc_rsq_v2r8(dx01,dy01,dz01);
 250             rsq02            = gmx_fjsp_calc_rsq_v2r8(dx02,dy02,dz02);
 251             rsq10            = gmx_fjsp_calc_rsq_v2r8(dx10,dy10,dz10);
 252             rsq11            = gmx_fjsp_calc_rsq_v2r8(dx11,dy11,dz11);
 253             rsq12            = gmx_fjsp_calc_rsq_v2r8(dx12,dy12,dz12);
 254             rsq20            = gmx_fjsp_calc_rsq_v2r8(dx20,dy20,dz20);
 255             rsq21            = gmx_fjsp_calc_rsq_v2r8(dx21,dy21,dz21);
 256             rsq22            = gmx_fjsp_calc_rsq_v2r8(dx22,dy22,dz22);
 257
 258             rinv00           = gmx_fjsp_invsqrt_v2r8(rsq00);
 259             rinv01           = gmx_fjsp_invsqrt_v2r8(rsq01);
 260             rinv02           = gmx_fjsp_invsqrt_v2r8(rsq02);
 261             rinv10           = gmx_fjsp_invsqrt_v2r8(rsq10);
 262             rinv11           = gmx_fjsp_invsqrt_v2r8(rsq11);
 263             rinv12           = gmx_fjsp_invsqrt_v2r8(rsq12);
 264             rinv20           = gmx_fjsp_invsqrt_v2r8(rsq20);
 265             rinv21           = gmx_fjsp_invsqrt_v2r8(rsq21);
 266             rinv22           = gmx_fjsp_invsqrt_v2r8(rsq22);
 267
 268             rinvsq00         = _fjsp_mul_v2r8(rinv00,rinv00);
 269             rinvsq01         = _fjsp_mul_v2r8(rinv01,rinv01);
 270             rinvsq02         = _fjsp_mul_v2r8(rinv02,rinv02);
 271             rinvsq10         = _fjsp_mul_v2r8(rinv10,rinv10);
 272             rinvsq11         = _fjsp_mul_v2r8(rinv11,rinv11);
 273             rinvsq12         = _fjsp_mul_v2r8(rinv12,rinv12);
 274             rinvsq20         = _fjsp_mul_v2r8(rinv20,rinv20);
 275             rinvsq21         = _fjsp_mul_v2r8(rinv21,rinv21);
 276             rinvsq22         = _fjsp_mul_v2r8(rinv22,rinv22);
 277
 278             fjx0             = _fjsp_setzero_v2r8();
 279             fjy0             = _fjsp_setzero_v2r8();
 280             fjz0             = _fjsp_setzero_v2r8();
 281             fjx1             = _fjsp_setzero_v2r8();
 282             fjy1             = _fjsp_setzero_v2r8();
 283             fjz1             = _fjsp_setzero_v2r8();
 284             fjx2             = _fjsp_setzero_v2r8();
 285             fjy2             = _fjsp_setzero_v2r8();
 286             fjz2             = _fjsp_setzero_v2r8();
 287
 288             /**************************
 289              * CALCULATE INTERACTIONS *
 290              **************************/
 291
 292             r00              = _fjsp_mul_v2r8(rsq00,rinv00);
 293
 294             /* Calculate table index by multiplying r with table scale and truncate to integer */
 295             rt               = _fjsp_mul_v2r8(r00,vftabscale);
 296             itab_tmp         = _fjsp_dtox_v2r8(rt);
 297             vfeps            = _fjsp_sub_v2r8(rt, _fjsp_xtod_v2r8(itab_tmp));
 298             twovfeps         = _fjsp_add_v2r8(vfeps,vfeps);
 299             _fjsp_store_v2r8(&vfconv.simd,itab_tmp);
 300
 301             vfconv.i[0]     *= 8;
 302             vfconv.i[1]     *= 8;
 303
 304             /* REACTION-FIELD ELECTROSTATICS */
 305             velec            = _fjsp_mul_v2r8(qq00,_fjsp_sub_v2r8(_fjsp_madd_v2r8(krf,rsq00,rinv00),crf));
 306             felec            = _fjsp_mul_v2r8(qq00,_fjsp_msub_v2r8(rinv00,rinvsq00,krf2));
 307
 308             /* CUBIC SPLINE TABLE DISPERSION */
 309             Y                = _fjsp_load_v2r8( vftab + vfconv.i[0] );
 310             F                = _fjsp_load_v2r8( vftab + vfconv.i[1] );
 311             GMX_FJSP_TRANSPOSE2_V2R8(Y,F);
 312             G                = _fjsp_load_v2r8( vftab + vfconv.i[0] + 2 );
 313             H                = _fjsp_load_v2r8( vftab + vfconv.i[1] + 2 );
 314             GMX_FJSP_TRANSPOSE2_V2R8(G,H);
 315             Fp               = _fjsp_madd_v2r8(vfeps,_fjsp_madd_v2r8(H,vfeps,G),F);
 316             VV               = _fjsp_madd_v2r8(vfeps,Fp,Y);
 317             vvdw6            = _fjsp_mul_v2r8(c6_00,VV);
 318             FF               = _fjsp_madd_v2r8(vfeps,_fjsp_madd_v2r8(twovfeps,H,G),Fp);
 319             fvdw6            = _fjsp_mul_v2r8(c6_00,FF);
 320
 321             /* CUBIC SPLINE TABLE REPULSION */
 322             Y                = _fjsp_load_v2r8( vftab + vfconv.i[0] + 4 );
 323             F                = _fjsp_load_v2r8( vftab + vfconv.i[1] + 4 );
 324             GMX_FJSP_TRANSPOSE2_V2R8(Y,F);
 325             G                = _fjsp_load_v2r8( vftab + vfconv.i[0] + 6 );
 326             H                = _fjsp_load_v2r8( vftab + vfconv.i[1] + 6 );
 327             GMX_FJSP_TRANSPOSE2_V2R8(G,H);
 328             Fp               = _fjsp_madd_v2r8(vfeps,_fjsp_madd_v2r8(H,vfeps,G),F);
 329             VV               = _fjsp_madd_v2r8(vfeps,Fp,Y);
 330             vvdw12           = _fjsp_mul_v2r8(c12_00,VV);
 331             FF               = _fjsp_madd_v2r8(vfeps,_fjsp_madd_v2r8(twovfeps,H,G),Fp);
 332             fvdw12           = _fjsp_mul_v2r8(c12_00,FF);
 333             vvdw             = _fjsp_add_v2r8(vvdw12,vvdw6);
 334             fvdw             = _fjsp_neg_v2r8(_fjsp_mul_v2r8(_fjsp_add_v2r8(fvdw6,fvdw12),_fjsp_mul_v2r8(vftabscale,rinv00)));
 335
 336             /* Update potential sum for this i atom from the interaction with this j atom. */
 337             velecsum         = _fjsp_add_v2r8(velecsum,velec);
 338             vvdwsum          = _fjsp_add_v2r8(vvdwsum,vvdw);
 339
 340             fscal            = _fjsp_add_v2r8(felec,fvdw);
 341
 342             /* Update vectorial force */
 343             fix0             = _fjsp_madd_v2r8(dx00,fscal,fix0);
 344             fiy0             = _fjsp_madd_v2r8(dy00,fscal,fiy0);
 345             fiz0             = _fjsp_madd_v2r8(dz00,fscal,fiz0);
 346
 347             fjx0             = _fjsp_madd_v2r8(dx00,fscal,fjx0);
 348             fjy0             = _fjsp_madd_v2r8(dy00,fscal,fjy0);
 349             fjz0             = _fjsp_madd_v2r8(dz00,fscal,fjz0);
 350
 351             /**************************
 352              * CALCULATE INTERACTIONS *
 353              **************************/
 354
 355             /* REACTION-FIELD ELECTROSTATICS */
 356             velec            = _fjsp_mul_v2r8(qq01,_fjsp_sub_v2r8(_fjsp_madd_v2r8(krf,rsq01,rinv01),crf));
 357             felec            = _fjsp_mul_v2r8(qq01,_fjsp_msub_v2r8(rinv01,rinvsq01,krf2));
 358
 359             /* Update potential sum for this i atom from the interaction with this j atom. */
 360             velecsum         = _fjsp_add_v2r8(velecsum,velec);
 361
 362             fscal            = felec;
 363
 364             /* Update vectorial force */
 365             fix0             = _fjsp_madd_v2r8(dx01,fscal,fix0);
 366             fiy0             = _fjsp_madd_v2r8(dy01,fscal,fiy0);
 367             fiz0             = _fjsp_madd_v2r8(dz01,fscal,fiz0);
 368
 369             fjx1             = _fjsp_madd_v2r8(dx01,fscal,fjx1);
 370             fjy1             = _fjsp_madd_v2r8(dy01,fscal,fjy1);
 371             fjz1             = _fjsp_madd_v2r8(dz01,fscal,fjz1);
 372
 373             /**************************
 374              * CALCULATE INTERACTIONS *
 375              **************************/
 376
 377             /* REACTION-FIELD ELECTROSTATICS */
 378             velec            = _fjsp_mul_v2r8(qq02,_fjsp_sub_v2r8(_fjsp_madd_v2r8(krf,rsq02,rinv02),crf));
 379             felec            = _fjsp_mul_v2r8(qq02,_fjsp_msub_v2r8(rinv02,rinvsq02,krf2));
 380
 381             /* Update potential sum for this i atom from the interaction with this j atom. */
 382             velecsum         = _fjsp_add_v2r8(velecsum,velec);
 383
 384             fscal            = felec;
 385
 386             /* Update vectorial force */
 387             fix0             = _fjsp_madd_v2r8(dx02,fscal,fix0);
 388             fiy0             = _fjsp_madd_v2r8(dy02,fscal,fiy0);
 389             fiz0             = _fjsp_madd_v2r8(dz02,fscal,fiz0);
 390
 391             fjx2             = _fjsp_madd_v2r8(dx02,fscal,fjx2);
 392             fjy2             = _fjsp_madd_v2r8(dy02,fscal,fjy2);
 393             fjz2             = _fjsp_madd_v2r8(dz02,fscal,fjz2);
 394
 395             /**************************
 396              * CALCULATE INTERACTIONS *
 397              **************************/
 398
 399             /* REACTION-FIELD ELECTROSTATICS */
 400             velec            = _fjsp_mul_v2r8(qq10,_fjsp_sub_v2r8(_fjsp_madd_v2r8(krf,rsq10,rinv10),crf));
 401             felec            = _fjsp_mul_v2r8(qq10,_fjsp_msub_v2r8(rinv10,rinvsq10,krf2));
 402
 403             /* Update potential sum for this i atom from the interaction with this j atom. */
 404             velecsum         = _fjsp_add_v2r8(velecsum,velec);
 405
 406             fscal            = felec;
 407
 408             /* Update vectorial force */
 409             fix1             = _fjsp_madd_v2r8(dx10,fscal,fix1);
 410             fiy1             = _fjsp_madd_v2r8(dy10,fscal,fiy1);
 411             fiz1             = _fjsp_madd_v2r8(dz10,fscal,fiz1);
 412
 413             fjx0             = _fjsp_madd_v2r8(dx10,fscal,fjx0);
 414             fjy0             = _fjsp_madd_v2r8(dy10,fscal,fjy0);
 415             fjz0             = _fjsp_madd_v2r8(dz10,fscal,fjz0);
 416
 417             /**************************
 418              * CALCULATE INTERACTIONS *
 419              **************************/
 420
 421             /* REACTION-FIELD ELECTROSTATICS */
 422             velec            = _fjsp_mul_v2r8(qq11,_fjsp_sub_v2r8(_fjsp_madd_v2r8(krf,rsq11,rinv11),crf));
 423             felec            = _fjsp_mul_v2r8(qq11,_fjsp_msub_v2r8(rinv11,rinvsq11,krf2));
 424
 425             /* Update potential sum for this i atom from the interaction with this j atom. */
 426             velecsum         = _fjsp_add_v2r8(velecsum,velec);
 427
 428             fscal            = felec;
 429
 430             /* Update vectorial force */
 431             fix1             = _fjsp_madd_v2r8(dx11,fscal,fix1);
 432             fiy1             = _fjsp_madd_v2r8(dy11,fscal,fiy1);
 433             fiz1             = _fjsp_madd_v2r8(dz11,fscal,fiz1);
 434
 435             fjx1             = _fjsp_madd_v2r8(dx11,fscal,fjx1);
 436             fjy1             = _fjsp_madd_v2r8(dy11,fscal,fjy1);
 437             fjz1             = _fjsp_madd_v2r8(dz11,fscal,fjz1);
 438
 439             /**************************
 440              * CALCULATE INTERACTIONS *
 441              **************************/
 442
 443             /* REACTION-FIELD ELECTROSTATICS */
 444             velec            = _fjsp_mul_v2r8(qq12,_fjsp_sub_v2r8(_fjsp_madd_v2r8(krf,rsq12,rinv12),crf));
 445             felec            = _fjsp_mul_v2r8(qq12,_fjsp_msub_v2r8(rinv12,rinvsq12,krf2));
 446
 447             /* Update potential sum for this i atom from the interaction with this j atom. */
 448             velecsum         = _fjsp_add_v2r8(velecsum,velec);
 449
 450             fscal            = felec;
 451
 452             /* Update vectorial force */
 453             fix1             = _fjsp_madd_v2r8(dx12,fscal,fix1);
 454             fiy1             = _fjsp_madd_v2r8(dy12,fscal,fiy1);
 455             fiz1             = _fjsp_madd_v2r8(dz12,fscal,fiz1);
 456
 457             fjx2             = _fjsp_madd_v2r8(dx12,fscal,fjx2);
 458             fjy2             = _fjsp_madd_v2r8(dy12,fscal,fjy2);
 459             fjz2             = _fjsp_madd_v2r8(dz12,fscal,fjz2);
 460
 461             /**************************
 462              * CALCULATE INTERACTIONS *
 463              **************************/
 464
 465             /* REACTION-FIELD ELECTROSTATICS */
 466             velec            = _fjsp_mul_v2r8(qq20,_fjsp_sub_v2r8(_fjsp_madd_v2r8(krf,rsq20,rinv20),crf));
 467             felec            = _fjsp_mul_v2r8(qq20,_fjsp_msub_v2r8(rinv20,rinvsq20,krf2));
 468
 469             /* Update potential sum for this i atom from the interaction with this j atom. */
 470             velecsum         = _fjsp_add_v2r8(velecsum,velec);
 471
 472             fscal            = felec;
 473
 474             /* Update vectorial force */
 475             fix2             = _fjsp_madd_v2r8(dx20,fscal,fix2);
 476             fiy2             = _fjsp_madd_v2r8(dy20,fscal,fiy2);
 477             fiz2             = _fjsp_madd_v2r8(dz20,fscal,fiz2);
 478
 479             fjx0             = _fjsp_madd_v2r8(dx20,fscal,fjx0);
 480             fjy0             = _fjsp_madd_v2r8(dy20,fscal,fjy0);
 481             fjz0             = _fjsp_madd_v2r8(dz20,fscal,fjz0);
 482
 483             /**************************
 484              * CALCULATE INTERACTIONS *
 485              **************************/
 486
 487             /* REACTION-FIELD ELECTROSTATICS */
 488             velec            = _fjsp_mul_v2r8(qq21,_fjsp_sub_v2r8(_fjsp_madd_v2r8(krf,rsq21,rinv21),crf));
 489             felec            = _fjsp_mul_v2r8(qq21,_fjsp_msub_v2r8(rinv21,rinvsq21,krf2));
 490
 491             /* Update potential sum for this i atom from the interaction with this j atom. */
 492             velecsum         = _fjsp_add_v2r8(velecsum,velec);
 493
 494             fscal            = felec;
 495
 496             /* Update vectorial force */
 497             fix2             = _fjsp_madd_v2r8(dx21,fscal,fix2);
 498             fiy2             = _fjsp_madd_v2r8(dy21,fscal,fiy2);
 499             fiz2             = _fjsp_madd_v2r8(dz21,fscal,fiz2);
 500
 501             fjx1             = _fjsp_madd_v2r8(dx21,fscal,fjx1);
 502             fjy1             = _fjsp_madd_v2r8(dy21,fscal,fjy1);
 503             fjz1             = _fjsp_madd_v2r8(dz21,fscal,fjz1);
 504
 505             /**************************
 506              * CALCULATE INTERACTIONS *
 507              **************************/
 508
 509             /* REACTION-FIELD ELECTROSTATICS */
 510             velec            = _fjsp_mul_v2r8(qq22,_fjsp_sub_v2r8(_fjsp_madd_v2r8(krf,rsq22,rinv22),crf));
 511             felec            = _fjsp_mul_v2r8(qq22,_fjsp_msub_v2r8(rinv22,rinvsq22,krf2));
 512
 513             /* Update potential sum for this i atom from the interaction with this j atom. */
 514             velecsum         = _fjsp_add_v2r8(velecsum,velec);
 515
 516             fscal            = felec;
 517
 518             /* Update vectorial force */
 519             fix2             = _fjsp_madd_v2r8(dx22,fscal,fix2);
 520             fiy2             = _fjsp_madd_v2r8(dy22,fscal,fiy2);
 521             fiz2             = _fjsp_madd_v2r8(dz22,fscal,fiz2);
 522
 523             fjx2             = _fjsp_madd_v2r8(dx22,fscal,fjx2);
 524             fjy2             = _fjsp_madd_v2r8(dy22,fscal,fjy2);
 525             fjz2             = _fjsp_madd_v2r8(dz22,fscal,fjz2);
 526
 527             gmx_fjsp_decrement_3rvec_2ptr_swizzle_v2r8(f+j_coord_offsetA,f+j_coord_offsetB,fjx0,fjy0,fjz0,fjx1,fjy1,fjz1,fjx2,fjy2,fjz2);
 528
 529             /* Inner loop uses 350 flops */
 530         }
 531
 532         if(jidx<j_index_end)
 533         {
 534
 535             jnrA             = jjnr[jidx];
 536             j_coord_offsetA  = DIM*jnrA;
 537
 538             /* load j atom coordinates */
 539             gmx_fjsp_load_3rvec_1ptr_swizzle_v2r8(x+j_coord_offsetA,
 540                                               &jx0,&jy0,&jz0,&jx1,&jy1,&jz1,&jx2,&jy2,&jz2);
 541
 542             /* Calculate displacement vector */
 543             dx00             = _fjsp_sub_v2r8(ix0,jx0);
 544             dy00             = _fjsp_sub_v2r8(iy0,jy0);
 545             dz00             = _fjsp_sub_v2r8(iz0,jz0);
 546             dx01             = _fjsp_sub_v2r8(ix0,jx1);
 547             dy01             = _fjsp_sub_v2r8(iy0,jy1);
 548             dz01             = _fjsp_sub_v2r8(iz0,jz1);
 549             dx02             = _fjsp_sub_v2r8(ix0,jx2);
 550             dy02             = _fjsp_sub_v2r8(iy0,jy2);
 551             dz02             = _fjsp_sub_v2r8(iz0,jz2);
 552             dx10             = _fjsp_sub_v2r8(ix1,jx0);
 553             dy10             = _fjsp_sub_v2r8(iy1,jy0);
 554             dz10             = _fjsp_sub_v2r8(iz1,jz0);
 555             dx11             = _fjsp_sub_v2r8(ix1,jx1);
 556             dy11             = _fjsp_sub_v2r8(iy1,jy1);
 557             dz11             = _fjsp_sub_v2r8(iz1,jz1);
 558             dx12             = _fjsp_sub_v2r8(ix1,jx2);
 559             dy12             = _fjsp_sub_v2r8(iy1,jy2);
 560             dz12             = _fjsp_sub_v2r8(iz1,jz2);
 561             dx20             = _fjsp_sub_v2r8(ix2,jx0);
 562             dy20             = _fjsp_sub_v2r8(iy2,jy0);
 563             dz20             = _fjsp_sub_v2r8(iz2,jz0);
 564             dx21             = _fjsp_sub_v2r8(ix2,jx1);
 565             dy21             = _fjsp_sub_v2r8(iy2,jy1);
 566             dz21             = _fjsp_sub_v2r8(iz2,jz1);
 567             dx22             = _fjsp_sub_v2r8(ix2,jx2);
 568             dy22             = _fjsp_sub_v2r8(iy2,jy2);
 569             dz22             = _fjsp_sub_v2r8(iz2,jz2);
 570
 571             /* Calculate squared distance and things based on it */
 572             rsq00            = gmx_fjsp_calc_rsq_v2r8(dx00,dy00,dz00);
 573             rsq01            = gmx_fjsp_calc_rsq_v2r8(dx01,dy01,dz01);
 574             rsq02            = gmx_fjsp_calc_rsq_v2r8(dx02,dy02,dz02);
 575             rsq10            = gmx_fjsp_calc_rsq_v2r8(dx10,dy10,dz10);
 576             rsq11            = gmx_fjsp_calc_rsq_v2r8(dx11,dy11,dz11);
 577             rsq12            = gmx_fjsp_calc_rsq_v2r8(dx12,dy12,dz12);
 578             rsq20            = gmx_fjsp_calc_rsq_v2r8(dx20,dy20,dz20);
 579             rsq21            = gmx_fjsp_calc_rsq_v2r8(dx21,dy21,dz21);
 580             rsq22            = gmx_fjsp_calc_rsq_v2r8(dx22,dy22,dz22);
 581
 582             rinv00           = gmx_fjsp_invsqrt_v2r8(rsq00);
 583             rinv01           = gmx_fjsp_invsqrt_v2r8(rsq01);
 584             rinv02           = gmx_fjsp_invsqrt_v2r8(rsq02);
 585             rinv10           = gmx_fjsp_invsqrt_v2r8(rsq10);
 586             rinv11           = gmx_fjsp_invsqrt_v2r8(rsq11);
 587             rinv12           = gmx_fjsp_invsqrt_v2r8(rsq12);
 588             rinv20           = gmx_fjsp_invsqrt_v2r8(rsq20);
 589             rinv21           = gmx_fjsp_invsqrt_v2r8(rsq21);
 590             rinv22           = gmx_fjsp_invsqrt_v2r8(rsq22);
 591
 592             rinvsq00         = _fjsp_mul_v2r8(rinv00,rinv00);
 593             rinvsq01         = _fjsp_mul_v2r8(rinv01,rinv01);
 594             rinvsq02         = _fjsp_mul_v2r8(rinv02,rinv02);
 595             rinvsq10         = _fjsp_mul_v2r8(rinv10,rinv10);
 596             rinvsq11         = _fjsp_mul_v2r8(rinv11,rinv11);
 597             rinvsq12         = _fjsp_mul_v2r8(rinv12,rinv12);
 598             rinvsq20         = _fjsp_mul_v2r8(rinv20,rinv20);
 599             rinvsq21         = _fjsp_mul_v2r8(rinv21,rinv21);
 600             rinvsq22         = _fjsp_mul_v2r8(rinv22,rinv22);
 601
 602             fjx0             = _fjsp_setzero_v2r8();
 603             fjy0             = _fjsp_setzero_v2r8();
 604             fjz0             = _fjsp_setzero_v2r8();
 605             fjx1             = _fjsp_setzero_v2r8();
 606             fjy1             = _fjsp_setzero_v2r8();
 607             fjz1             = _fjsp_setzero_v2r8();
 608             fjx2             = _fjsp_setzero_v2r8();
 609             fjy2             = _fjsp_setzero_v2r8();
 610             fjz2             = _fjsp_setzero_v2r8();
 611
 612             /**************************
 613              * CALCULATE INTERACTIONS *
 614              **************************/
 615
 616             r00              = _fjsp_mul_v2r8(rsq00,rinv00);
 617
 618             /* Calculate table index by multiplying r with table scale and truncate to integer */
 619             rt               = _fjsp_mul_v2r8(r00,vftabscale);
 620             itab_tmp         = _fjsp_dtox_v2r8(rt);
 621             vfeps            = _fjsp_sub_v2r8(rt, _fjsp_xtod_v2r8(itab_tmp));
 622             twovfeps         = _fjsp_add_v2r8(vfeps,vfeps);
 623             _fjsp_store_v2r8(&vfconv.simd,itab_tmp);
 624
 625             vfconv.i[0]     *= 8;
 626             vfconv.i[1]     *= 8;
 627
 628             /* REACTION-FIELD ELECTROSTATICS */
 629             velec            = _fjsp_mul_v2r8(qq00,_fjsp_sub_v2r8(_fjsp_madd_v2r8(krf,rsq00,rinv00),crf));
 630             felec            = _fjsp_mul_v2r8(qq00,_fjsp_msub_v2r8(rinv00,rinvsq00,krf2));
 631
 632             /* CUBIC SPLINE TABLE DISPERSION */
 633             Y                = _fjsp_load_v2r8( vftab + vfconv.i[0] );
 634             F                = _fjsp_setzero_v2r8();
 635             GMX_FJSP_TRANSPOSE2_V2R8(Y,F);
 636             G                = _fjsp_load_v2r8( vftab + vfconv.i[0] + 2 );
 637             H                = _fjsp_setzero_v2r8();
 638             GMX_FJSP_TRANSPOSE2_V2R8(G,H);
 639             Fp               = _fjsp_madd_v2r8(vfeps,_fjsp_madd_v2r8(H,vfeps,G),F);
 640             VV               = _fjsp_madd_v2r8(vfeps,Fp,Y);
 641             vvdw6            = _fjsp_mul_v2r8(c6_00,VV);
 642             FF               = _fjsp_madd_v2r8(vfeps,_fjsp_madd_v2r8(twovfeps,H,G),Fp);
 643             fvdw6            = _fjsp_mul_v2r8(c6_00,FF);
 644
 645             /* CUBIC SPLINE TABLE REPULSION */
 646             Y                = _fjsp_load_v2r8( vftab + vfconv.i[0] + 4 );
 647             F                = _fjsp_setzero_v2r8();
 648             GMX_FJSP_TRANSPOSE2_V2R8(Y,F);
 649             G                = _fjsp_load_v2r8( vftab + vfconv.i[0] + 6 );
 650             H                = _fjsp_setzero_v2r8();
 651             GMX_FJSP_TRANSPOSE2_V2R8(G,H);
 652             Fp               = _fjsp_madd_v2r8(vfeps,_fjsp_madd_v2r8(H,vfeps,G),F);
 653             VV               = _fjsp_madd_v2r8(vfeps,Fp,Y);
 654             vvdw12           = _fjsp_mul_v2r8(c12_00,VV);
 655             FF               = _fjsp_madd_v2r8(vfeps,_fjsp_madd_v2r8(twovfeps,H,G),Fp);
 656             fvdw12           = _fjsp_mul_v2r8(c12_00,FF);
 657             vvdw             = _fjsp_add_v2r8(vvdw12,vvdw6);
 658             fvdw             = _fjsp_neg_v2r8(_fjsp_mul_v2r8(_fjsp_add_v2r8(fvdw6,fvdw12),_fjsp_mul_v2r8(vftabscale,rinv00)));
 659
 660             /* Update potential sum for this i atom from the interaction with this j atom. */
 661             velec            = _fjsp_unpacklo_v2r8(velec,_fjsp_setzero_v2r8());
 662             velecsum         = _fjsp_add_v2r8(velecsum,velec);
 663             vvdw             = _fjsp_unpacklo_v2r8(vvdw,_fjsp_setzero_v2r8());
 664             vvdwsum          = _fjsp_add_v2r8(vvdwsum,vvdw);
 665
 666             fscal            = _fjsp_add_v2r8(felec,fvdw);
 667
 668             fscal            = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
 669
 670             /* Update vectorial force */
 671             fix0             = _fjsp_madd_v2r8(dx00,fscal,fix0);
 672             fiy0             = _fjsp_madd_v2r8(dy00,fscal,fiy0);
 673             fiz0             = _fjsp_madd_v2r8(dz00,fscal,fiz0);
 674
 675             fjx0             = _fjsp_madd_v2r8(dx00,fscal,fjx0);
 676             fjy0             = _fjsp_madd_v2r8(dy00,fscal,fjy0);
 677             fjz0             = _fjsp_madd_v2r8(dz00,fscal,fjz0);
 678
 679             /**************************
 680              * CALCULATE INTERACTIONS *
 681              **************************/
 682
 683             /* REACTION-FIELD ELECTROSTATICS */
 684             velec            = _fjsp_mul_v2r8(qq01,_fjsp_sub_v2r8(_fjsp_madd_v2r8(krf,rsq01,rinv01),crf));
 685             felec            = _fjsp_mul_v2r8(qq01,_fjsp_msub_v2r8(rinv01,rinvsq01,krf2));
 686
 687             /* Update potential sum for this i atom from the interaction with this j atom. */
 688             velec            = _fjsp_unpacklo_v2r8(velec,_fjsp_setzero_v2r8());
 689             velecsum         = _fjsp_add_v2r8(velecsum,velec);
 690
 691             fscal            = felec;
 692
 693             fscal            = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
 694
 695             /* Update vectorial force */
 696             fix0             = _fjsp_madd_v2r8(dx01,fscal,fix0);
 697             fiy0             = _fjsp_madd_v2r8(dy01,fscal,fiy0);
 698             fiz0             = _fjsp_madd_v2r8(dz01,fscal,fiz0);
 699
 700             fjx1             = _fjsp_madd_v2r8(dx01,fscal,fjx1);
 701             fjy1             = _fjsp_madd_v2r8(dy01,fscal,fjy1);
 702             fjz1             = _fjsp_madd_v2r8(dz01,fscal,fjz1);
 703
 704             /**************************
 705              * CALCULATE INTERACTIONS *
 706              **************************/
 707
 708             /* REACTION-FIELD ELECTROSTATICS */
 709             velec            = _fjsp_mul_v2r8(qq02,_fjsp_sub_v2r8(_fjsp_madd_v2r8(krf,rsq02,rinv02),crf));
 710             felec            = _fjsp_mul_v2r8(qq02,_fjsp_msub_v2r8(rinv02,rinvsq02,krf2));
 711
 712             /* Update potential sum for this i atom from the interaction with this j atom. */
 713             velec            = _fjsp_unpacklo_v2r8(velec,_fjsp_setzero_v2r8());
 714             velecsum         = _fjsp_add_v2r8(velecsum,velec);
 715
 716             fscal            = felec;
 717
 718             fscal            = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
 719
 720             /* Update vectorial force */
 721             fix0             = _fjsp_madd_v2r8(dx02,fscal,fix0);
 722             fiy0             = _fjsp_madd_v2r8(dy02,fscal,fiy0);
 723             fiz0             = _fjsp_madd_v2r8(dz02,fscal,fiz0);
 724
 725             fjx2             = _fjsp_madd_v2r8(dx02,fscal,fjx2);
 726             fjy2             = _fjsp_madd_v2r8(dy02,fscal,fjy2);
 727             fjz2             = _fjsp_madd_v2r8(dz02,fscal,fjz2);
 728
 729             /**************************
 730              * CALCULATE INTERACTIONS *
 731              **************************/
 732
 733             /* REACTION-FIELD ELECTROSTATICS */
 734             velec            = _fjsp_mul_v2r8(qq10,_fjsp_sub_v2r8(_fjsp_madd_v2r8(krf,rsq10,rinv10),crf));
 735             felec            = _fjsp_mul_v2r8(qq10,_fjsp_msub_v2r8(rinv10,rinvsq10,krf2));
 736
 737             /* Update potential sum for this i atom from the interaction with this j atom. */
 738             velec            = _fjsp_unpacklo_v2r8(velec,_fjsp_setzero_v2r8());
 739             velecsum         = _fjsp_add_v2r8(velecsum,velec);
 740
 741             fscal            = felec;
 742
 743             fscal            = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
 744
 745             /* Update vectorial force */
 746             fix1             = _fjsp_madd_v2r8(dx10,fscal,fix1);
 747             fiy1             = _fjsp_madd_v2r8(dy10,fscal,fiy1);
 748             fiz1             = _fjsp_madd_v2r8(dz10,fscal,fiz1);
 749
 750             fjx0             = _fjsp_madd_v2r8(dx10,fscal,fjx0);
 751             fjy0             = _fjsp_madd_v2r8(dy10,fscal,fjy0);
 752             fjz0             = _fjsp_madd_v2r8(dz10,fscal,fjz0);
 753
 754             /**************************
 755              * CALCULATE INTERACTIONS *
 756              **************************/
 757
 758             /* REACTION-FIELD ELECTROSTATICS */
 759             velec            = _fjsp_mul_v2r8(qq11,_fjsp_sub_v2r8(_fjsp_madd_v2r8(krf,rsq11,rinv11),crf));
 760             felec            = _fjsp_mul_v2r8(qq11,_fjsp_msub_v2r8(rinv11,rinvsq11,krf2));
 761
 762             /* Update potential sum for this i atom from the interaction with this j atom. */
 763             velec            = _fjsp_unpacklo_v2r8(velec,_fjsp_setzero_v2r8());
 764             velecsum         = _fjsp_add_v2r8(velecsum,velec);
 765
 766             fscal            = felec;
 767
 768             fscal            = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
 769
 770             /* Update vectorial force */
 771             fix1             = _fjsp_madd_v2r8(dx11,fscal,fix1);
 772             fiy1             = _fjsp_madd_v2r8(dy11,fscal,fiy1);
 773             fiz1             = _fjsp_madd_v2r8(dz11,fscal,fiz1);
 774
 775             fjx1             = _fjsp_madd_v2r8(dx11,fscal,fjx1);
 776             fjy1             = _fjsp_madd_v2r8(dy11,fscal,fjy1);
 777             fjz1             = _fjsp_madd_v2r8(dz11,fscal,fjz1);
 778
 779             /**************************
 780              * CALCULATE INTERACTIONS *
 781              **************************/
 782
 783             /* REACTION-FIELD ELECTROSTATICS */
 784             velec            = _fjsp_mul_v2r8(qq12,_fjsp_sub_v2r8(_fjsp_madd_v2r8(krf,rsq12,rinv12),crf));
 785             felec            = _fjsp_mul_v2r8(qq12,_fjsp_msub_v2r8(rinv12,rinvsq12,krf2));
 786
 787             /* Update potential sum for this i atom from the interaction with this j atom. */
 788             velec            = _fjsp_unpacklo_v2r8(velec,_fjsp_setzero_v2r8());
 789             velecsum         = _fjsp_add_v2r8(velecsum,velec);
 790
 791             fscal            = felec;
 792
 793             fscal            = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
 794
 795             /* Update vectorial force */
 796             fix1             = _fjsp_madd_v2r8(dx12,fscal,fix1);
 797             fiy1             = _fjsp_madd_v2r8(dy12,fscal,fiy1);
 798             fiz1             = _fjsp_madd_v2r8(dz12,fscal,fiz1);
 799
 800             fjx2             = _fjsp_madd_v2r8(dx12,fscal,fjx2);
 801             fjy2             = _fjsp_madd_v2r8(dy12,fscal,fjy2);
 802             fjz2             = _fjsp_madd_v2r8(dz12,fscal,fjz2);
 803
 804             /**************************
 805              * CALCULATE INTERACTIONS *
 806              **************************/
 807
 808             /* REACTION-FIELD ELECTROSTATICS */
 809             velec            = _fjsp_mul_v2r8(qq20,_fjsp_sub_v2r8(_fjsp_madd_v2r8(krf,rsq20,rinv20),crf));
 810             felec            = _fjsp_mul_v2r8(qq20,_fjsp_msub_v2r8(rinv20,rinvsq20,krf2));
 811
 812             /* Update potential sum for this i atom from the interaction with this j atom. */
 813             velec            = _fjsp_unpacklo_v2r8(velec,_fjsp_setzero_v2r8());
 814             velecsum         = _fjsp_add_v2r8(velecsum,velec);
 815
 816             fscal            = felec;
 817
 818             fscal            = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
 819
 820             /* Update vectorial force */
 821             fix2             = _fjsp_madd_v2r8(dx20,fscal,fix2);
 822             fiy2             = _fjsp_madd_v2r8(dy20,fscal,fiy2);
 823             fiz2             = _fjsp_madd_v2r8(dz20,fscal,fiz2);
 824
 825             fjx0             = _fjsp_madd_v2r8(dx20,fscal,fjx0);
 826             fjy0             = _fjsp_madd_v2r8(dy20,fscal,fjy0);
 827             fjz0             = _fjsp_madd_v2r8(dz20,fscal,fjz0);
 828
 829             /**************************
 830              * CALCULATE INTERACTIONS *
 831              **************************/
 832
 833             /* REACTION-FIELD ELECTROSTATICS */
 834             velec            = _fjsp_mul_v2r8(qq21,_fjsp_sub_v2r8(_fjsp_madd_v2r8(krf,rsq21,rinv21),crf));
 835             felec            = _fjsp_mul_v2r8(qq21,_fjsp_msub_v2r8(rinv21,rinvsq21,krf2));
 836
 837             /* Update potential sum for this i atom from the interaction with this j atom. */
 838             velec            = _fjsp_unpacklo_v2r8(velec,_fjsp_setzero_v2r8());
 839             velecsum         = _fjsp_add_v2r8(velecsum,velec);
 840
 841             fscal            = felec;
 842
 843             fscal            = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
 844
 845             /* Update vectorial force */
 846             fix2             = _fjsp_madd_v2r8(dx21,fscal,fix2);
 847             fiy2             = _fjsp_madd_v2r8(dy21,fscal,fiy2);
 848             fiz2             = _fjsp_madd_v2r8(dz21,fscal,fiz2);
 849
 850             fjx1             = _fjsp_madd_v2r8(dx21,fscal,fjx1);
 851             fjy1             = _fjsp_madd_v2r8(dy21,fscal,fjy1);
 852             fjz1             = _fjsp_madd_v2r8(dz21,fscal,fjz1);
 853
 854             /**************************
 855              * CALCULATE INTERACTIONS *
 856              **************************/
 857
 858             /* REACTION-FIELD ELECTROSTATICS */
 859             velec            = _fjsp_mul_v2r8(qq22,_fjsp_sub_v2r8(_fjsp_madd_v2r8(krf,rsq22,rinv22),crf));
 860             felec            = _fjsp_mul_v2r8(qq22,_fjsp_msub_v2r8(rinv22,rinvsq22,krf2));
 861
 862             /* Update potential sum for this i atom from the interaction with this j atom. */
 863             velec            = _fjsp_unpacklo_v2r8(velec,_fjsp_setzero_v2r8());
 864             velecsum         = _fjsp_add_v2r8(velecsum,velec);
 865
 866             fscal            = felec;
 867
 868             fscal            = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
 869
 870             /* Update vectorial force */
 871             fix2             = _fjsp_madd_v2r8(dx22,fscal,fix2);
 872             fiy2             = _fjsp_madd_v2r8(dy22,fscal,fiy2);
 873             fiz2             = _fjsp_madd_v2r8(dz22,fscal,fiz2);
 874
 875             fjx2             = _fjsp_madd_v2r8(dx22,fscal,fjx2);
 876             fjy2             = _fjsp_madd_v2r8(dy22,fscal,fjy2);
 877             fjz2             = _fjsp_madd_v2r8(dz22,fscal,fjz2);
 878
 879             gmx_fjsp_decrement_3rvec_1ptr_swizzle_v2r8(f+j_coord_offsetA,fjx0,fjy0,fjz0,fjx1,fjy1,fjz1,fjx2,fjy2,fjz2);
 880
 881             /* Inner loop uses 350 flops */
 882         }
 883
 884         /* End of innermost loop */
 885
 886         gmx_fjsp_update_iforce_3atom_swizzle_v2r8(fix0,fiy0,fiz0,fix1,fiy1,fiz1,fix2,fiy2,fiz2,
 887                                               f+i_coord_offset,fshift+i_shift_offset);
 888
 889         ggid                        = gid[iidx];
 890         /* Update potential energies */
 891         gmx_fjsp_update_1pot_v2r8(velecsum,kernel_data->energygrp_elec+ggid);
 892         gmx_fjsp_update_1pot_v2r8(vvdwsum,kernel_data->energygrp_vdw+ggid);
 893
 894         /* Increment number of inner iterations */
 895         inneriter                  += j_index_end - j_index_start;
 896
 897         /* Outer loop uses 20 flops */
 898     }
 899
 900     /* Increment number of outer iterations */
 901     outeriter        += nri;
 902
 903     /* Update outer/inner flops */
 904
 905     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W3W3_VF,outeriter*20 + inneriter*350);
 906 }
 907 /*
 908  * Gromacs nonbonded kernel:   nb_kernel_ElecRF_VdwCSTab_GeomW3W3_F_sparc64_hpc_ace_double
 909  * Electrostatics interaction: ReactionField
 910  * VdW interaction:            CubicSplineTable
 911  * Geometry:                   Water3-Water3
 912  * Calculate force/pot:        Force
 913  */
 914 void
 915 nb_kernel_ElecRF_VdwCSTab_GeomW3W3_F_sparc64_hpc_ace_double
 916                     (t_nblist                    * gmx_restrict       nlist,
 917                      rvec                        * gmx_restrict          xx,
 918                      rvec                        * gmx_restrict          ff,
 919                      t_forcerec                  * gmx_restrict          fr,
 920                      t_mdatoms                   * gmx_restrict     mdatoms,
 921                      nb_kernel_data_t gmx_unused * gmx_restrict kernel_data,
 922                      t_nrnb                      * gmx_restrict        nrnb)
 923 {
 924     /* Suffixes 0,1,2,3 refer to particle indices for waters in the inner or outer loop, or
 925      * just 0 for non-waters.
 926      * Suffixes A,B refer to j loop unrolling done with double precision SIMD, e.g. for the two different
 927      * jnr indices corresponding to data put in the four positions in the SIMD register.
 928      */
 929     int              i_shift_offset,i_coord_offset,outeriter,inneriter;
 930     int              j_index_start,j_index_end,jidx,nri,inr,ggid,iidx;
 931     int              jnrA,jnrB;
 932     int              j_coord_offsetA,j_coord_offsetB;
 933     int              *iinr,*jindex,*jjnr,*shiftidx,*gid;
 934     real             rcutoff_scalar;
 935     real             *shiftvec,*fshift,*x,*f;
 936     _fjsp_v2r8       tx,ty,tz,fscal,rcutoff,rcutoff2,jidxall;
 937     int              vdwioffset0;
 938     _fjsp_v2r8       ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
 939     int              vdwioffset1;
 940     _fjsp_v2r8       ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
 941     int              vdwioffset2;
 942     _fjsp_v2r8       ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
 943     int              vdwjidx0A,vdwjidx0B;
 944     _fjsp_v2r8       jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
 945     int              vdwjidx1A,vdwjidx1B;
 946     _fjsp_v2r8       jx1,jy1,jz1,fjx1,fjy1,fjz1,jq1,isaj1;
 947     int              vdwjidx2A,vdwjidx2B;
 948     _fjsp_v2r8       jx2,jy2,jz2,fjx2,fjy2,fjz2,jq2,isaj2;
 949     _fjsp_v2r8       dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00;
 950     _fjsp_v2r8       dx01,dy01,dz01,rsq01,rinv01,rinvsq01,r01,qq01,c6_01,c12_01;
 951     _fjsp_v2r8       dx02,dy02,dz02,rsq02,rinv02,rinvsq02,r02,qq02,c6_02,c12_02;
 952     _fjsp_v2r8       dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10;
 953     _fjsp_v2r8       dx11,dy11,dz11,rsq11,rinv11,rinvsq11,r11,qq11,c6_11,c12_11;
 954     _fjsp_v2r8       dx12,dy12,dz12,rsq12,rinv12,rinvsq12,r12,qq12,c6_12,c12_12;
 955     _fjsp_v2r8       dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20;
 956     _fjsp_v2r8       dx21,dy21,dz21,rsq21,rinv21,rinvsq21,r21,qq21,c6_21,c12_21;
 957     _fjsp_v2r8       dx22,dy22,dz22,rsq22,rinv22,rinvsq22,r22,qq22,c6_22,c12_22;
 958     _fjsp_v2r8       velec,felec,velecsum,facel,crf,krf,krf2;
 959     real             *charge;
 960     int              nvdwtype;
 961     _fjsp_v2r8       rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,sh_vdw_invrcut6;
 962     int              *vdwtype;
 963     real             *vdwparam;
 964     _fjsp_v2r8       one_sixth   = gmx_fjsp_set1_v2r8(1.0/6.0);
 965     _fjsp_v2r8       one_twelfth = gmx_fjsp_set1_v2r8(1.0/12.0);
 966     _fjsp_v2r8       rt,vfeps,vftabscale,Y,F,G,H,Heps,Fp,VV,FF,twovfeps;
 967     real             *vftab;
 968     _fjsp_v2r8       itab_tmp;
 969     _fjsp_v2r8       dummy_mask,cutoff_mask;
 970     _fjsp_v2r8       one     = gmx_fjsp_set1_v2r8(1.0);
 971     _fjsp_v2r8       two     = gmx_fjsp_set1_v2r8(2.0);
 972     union { _fjsp_v2r8 simd; long long int i[2]; } vfconv,gbconv,ewconv;
 973
 974     x                = xx[0];
 975     f                = ff[0];
 976
 977     nri              = nlist->nri;
 978     iinr             = nlist->iinr;
 979     jindex           = nlist->jindex;
 980     jjnr             = nlist->jjnr;
 981     shiftidx         = nlist->shift;
 982     gid              = nlist->gid;
 983     shiftvec         = fr->shift_vec[0];
 984     fshift           = fr->fshift[0];
 985     facel            = gmx_fjsp_set1_v2r8(fr->epsfac);
 986     charge           = mdatoms->chargeA;
 987     krf              = gmx_fjsp_set1_v2r8(fr->ic->k_rf);
 988     krf2             = gmx_fjsp_set1_v2r8(fr->ic->k_rf*2.0);
 989     crf              = gmx_fjsp_set1_v2r8(fr->ic->c_rf);
 990     nvdwtype         = fr->ntype;
 991     vdwparam         = fr->nbfp;
 992     vdwtype          = mdatoms->typeA;
 993
 994     vftab            = kernel_data->table_vdw->data;
 995     vftabscale       = gmx_fjsp_set1_v2r8(kernel_data->table_vdw->scale);
 996
 997     /* Setup water-specific parameters */
 998     inr              = nlist->iinr[0];
 999     iq0              = _fjsp_mul_v2r8(facel,gmx_fjsp_set1_v2r8(charge[inr+0]));
1000     iq1              = _fjsp_mul_v2r8(facel,gmx_fjsp_set1_v2r8(charge[inr+1]));
1001     iq2              = _fjsp_mul_v2r8(facel,gmx_fjsp_set1_v2r8(charge[inr+2]));
1002     vdwioffset0      = 2*nvdwtype*vdwtype[inr+0];
1003
1004     jq0              = gmx_fjsp_set1_v2r8(charge[inr+0]);
1005     jq1              = gmx_fjsp_set1_v2r8(charge[inr+1]);
1006     jq2              = gmx_fjsp_set1_v2r8(charge[inr+2]);
1007     vdwjidx0A        = 2*vdwtype[inr+0];
1008     qq00             = _fjsp_mul_v2r8(iq0,jq0);
1009     c6_00            = gmx_fjsp_set1_v2r8(vdwparam[vdwioffset0+vdwjidx0A]);
1010     c12_00           = gmx_fjsp_set1_v2r8(vdwparam[vdwioffset0+vdwjidx0A+1]);
1011     qq01             = _fjsp_mul_v2r8(iq0,jq1);
1012     qq02             = _fjsp_mul_v2r8(iq0,jq2);
1013     qq10             = _fjsp_mul_v2r8(iq1,jq0);
1014     qq11             = _fjsp_mul_v2r8(iq1,jq1);
1015     qq12             = _fjsp_mul_v2r8(iq1,jq2);
1016     qq20             = _fjsp_mul_v2r8(iq2,jq0);
1017     qq21             = _fjsp_mul_v2r8(iq2,jq1);
1018     qq22             = _fjsp_mul_v2r8(iq2,jq2);
1019
1020     /* Avoid stupid compiler warnings */
1021     jnrA = jnrB = 0;
1022     j_coord_offsetA = 0;
1023     j_coord_offsetB = 0;
1024
1025     outeriter        = 0;
1026     inneriter        = 0;
1027
1028     /* Start outer loop over neighborlists */
1029     for(iidx=0; iidx<nri; iidx++)
1030     {
1031         /* Load shift vector for this list */
1032         i_shift_offset   = DIM*shiftidx[iidx];
1033
1034         /* Load limits for loop over neighbors */
1035         j_index_start    = jindex[iidx];
1036         j_index_end      = jindex[iidx+1];
1037
1038         /* Get outer coordinate index */
1039         inr              = iinr[iidx];
1040         i_coord_offset   = DIM*inr;
1041
1042         /* Load i particle coords and add shift vector */
1043         gmx_fjsp_load_shift_and_3rvec_broadcast_v2r8(shiftvec+i_shift_offset,x+i_coord_offset,
1044                                                  &ix0,&iy0,&iz0,&ix1,&iy1,&iz1,&ix2,&iy2,&iz2);
1045
1046         fix0             = _fjsp_setzero_v2r8();
1047         fiy0             = _fjsp_setzero_v2r8();
1048         fiz0             = _fjsp_setzero_v2r8();
1049         fix1             = _fjsp_setzero_v2r8();
1050         fiy1             = _fjsp_setzero_v2r8();
1051         fiz1             = _fjsp_setzero_v2r8();
1052         fix2             = _fjsp_setzero_v2r8();
1053         fiy2             = _fjsp_setzero_v2r8();
1054         fiz2             = _fjsp_setzero_v2r8();
1055
1056         /* Start inner kernel loop */
1057         for(jidx=j_index_start; jidx<j_index_end-1; jidx+=2)
1058         {
1059
1060             /* Get j neighbor index, and coordinate index */
1061             jnrA             = jjnr[jidx];
1062             jnrB             = jjnr[jidx+1];
1063             j_coord_offsetA  = DIM*jnrA;
1064             j_coord_offsetB  = DIM*jnrB;
1065
1066             /* load j atom coordinates */
1067             gmx_fjsp_load_3rvec_2ptr_swizzle_v2r8(x+j_coord_offsetA,x+j_coord_offsetB,
1068                                               &jx0,&jy0,&jz0,&jx1,&jy1,&jz1,&jx2,&jy2,&jz2);
1069
1070             /* Calculate displacement vector */
1071             dx00             = _fjsp_sub_v2r8(ix0,jx0);
1072             dy00             = _fjsp_sub_v2r8(iy0,jy0);
1073             dz00             = _fjsp_sub_v2r8(iz0,jz0);
1074             dx01             = _fjsp_sub_v2r8(ix0,jx1);
1075             dy01             = _fjsp_sub_v2r8(iy0,jy1);
1076             dz01             = _fjsp_sub_v2r8(iz0,jz1);
1077             dx02             = _fjsp_sub_v2r8(ix0,jx2);
1078             dy02             = _fjsp_sub_v2r8(iy0,jy2);
1079             dz02             = _fjsp_sub_v2r8(iz0,jz2);
1080             dx10             = _fjsp_sub_v2r8(ix1,jx0);
1081             dy10             = _fjsp_sub_v2r8(iy1,jy0);
1082             dz10             = _fjsp_sub_v2r8(iz1,jz0);
1083             dx11             = _fjsp_sub_v2r8(ix1,jx1);
1084             dy11             = _fjsp_sub_v2r8(iy1,jy1);
1085             dz11             = _fjsp_sub_v2r8(iz1,jz1);
1086             dx12             = _fjsp_sub_v2r8(ix1,jx2);
1087             dy12             = _fjsp_sub_v2r8(iy1,jy2);
1088             dz12             = _fjsp_sub_v2r8(iz1,jz2);
1089             dx20             = _fjsp_sub_v2r8(ix2,jx0);
1090             dy20             = _fjsp_sub_v2r8(iy2,jy0);
1091             dz20             = _fjsp_sub_v2r8(iz2,jz0);
1092             dx21             = _fjsp_sub_v2r8(ix2,jx1);
1093             dy21             = _fjsp_sub_v2r8(iy2,jy1);
1094             dz21             = _fjsp_sub_v2r8(iz2,jz1);
1095             dx22             = _fjsp_sub_v2r8(ix2,jx2);
1096             dy22             = _fjsp_sub_v2r8(iy2,jy2);
1097             dz22             = _fjsp_sub_v2r8(iz2,jz2);
1098
1099             /* Calculate squared distance and things based on it */
1100             rsq00            = gmx_fjsp_calc_rsq_v2r8(dx00,dy00,dz00);
1101             rsq01            = gmx_fjsp_calc_rsq_v2r8(dx01,dy01,dz01);
1102             rsq02            = gmx_fjsp_calc_rsq_v2r8(dx02,dy02,dz02);
1103             rsq10            = gmx_fjsp_calc_rsq_v2r8(dx10,dy10,dz10);
1104             rsq11            = gmx_fjsp_calc_rsq_v2r8(dx11,dy11,dz11);
1105             rsq12            = gmx_fjsp_calc_rsq_v2r8(dx12,dy12,dz12);
1106             rsq20            = gmx_fjsp_calc_rsq_v2r8(dx20,dy20,dz20);
1107             rsq21            = gmx_fjsp_calc_rsq_v2r8(dx21,dy21,dz21);
1108             rsq22            = gmx_fjsp_calc_rsq_v2r8(dx22,dy22,dz22);
1109
1110             rinv00           = gmx_fjsp_invsqrt_v2r8(rsq00);
1111             rinv01           = gmx_fjsp_invsqrt_v2r8(rsq01);
1112             rinv02           = gmx_fjsp_invsqrt_v2r8(rsq02);
1113             rinv10           = gmx_fjsp_invsqrt_v2r8(rsq10);
1114             rinv11           = gmx_fjsp_invsqrt_v2r8(rsq11);
1115             rinv12           = gmx_fjsp_invsqrt_v2r8(rsq12);
1116             rinv20           = gmx_fjsp_invsqrt_v2r8(rsq20);
1117             rinv21           = gmx_fjsp_invsqrt_v2r8(rsq21);
1118             rinv22           = gmx_fjsp_invsqrt_v2r8(rsq22);
1119
1120             rinvsq00         = _fjsp_mul_v2r8(rinv00,rinv00);
1121             rinvsq01         = _fjsp_mul_v2r8(rinv01,rinv01);
1122             rinvsq02         = _fjsp_mul_v2r8(rinv02,rinv02);
1123             rinvsq10         = _fjsp_mul_v2r8(rinv10,rinv10);
1124             rinvsq11         = _fjsp_mul_v2r8(rinv11,rinv11);
1125             rinvsq12         = _fjsp_mul_v2r8(rinv12,rinv12);
1126             rinvsq20         = _fjsp_mul_v2r8(rinv20,rinv20);
1127             rinvsq21         = _fjsp_mul_v2r8(rinv21,rinv21);
1128             rinvsq22         = _fjsp_mul_v2r8(rinv22,rinv22);
1129
1130             fjx0             = _fjsp_setzero_v2r8();
1131             fjy0             = _fjsp_setzero_v2r8();
1132             fjz0             = _fjsp_setzero_v2r8();
1133             fjx1             = _fjsp_setzero_v2r8();
1134             fjy1             = _fjsp_setzero_v2r8();
1135             fjz1             = _fjsp_setzero_v2r8();
1136             fjx2             = _fjsp_setzero_v2r8();
1137             fjy2             = _fjsp_setzero_v2r8();
1138             fjz2             = _fjsp_setzero_v2r8();
1139
1140             /**************************
1141              * CALCULATE INTERACTIONS *
1142              **************************/
1143
1144             r00              = _fjsp_mul_v2r8(rsq00,rinv00);
1145
1146             /* Calculate table index by multiplying r with table scale and truncate to integer */
1147             rt               = _fjsp_mul_v2r8(r00,vftabscale);
1148             itab_tmp         = _fjsp_dtox_v2r8(rt);
1149             vfeps            = _fjsp_sub_v2r8(rt, _fjsp_xtod_v2r8(itab_tmp));
1150             twovfeps         = _fjsp_add_v2r8(vfeps,vfeps);
1151             _fjsp_store_v2r8(&vfconv.simd,itab_tmp);
1152
1153             vfconv.i[0]     *= 8;
1154             vfconv.i[1]     *= 8;
1155
1156             /* REACTION-FIELD ELECTROSTATICS */
1157             felec            = _fjsp_mul_v2r8(qq00,_fjsp_msub_v2r8(rinv00,rinvsq00,krf2));
1158
1159             /* CUBIC SPLINE TABLE DISPERSION */
1160             Y                = _fjsp_load_v2r8( vftab + vfconv.i[0] );
1161             F                = _fjsp_load_v2r8( vftab + vfconv.i[1] );
1162             GMX_FJSP_TRANSPOSE2_V2R8(Y,F);
1163             G                = _fjsp_load_v2r8( vftab + vfconv.i[0] + 2 );
1164             H                = _fjsp_load_v2r8( vftab + vfconv.i[1] + 2 );
1165             GMX_FJSP_TRANSPOSE2_V2R8(G,H);
1166             Fp               = _fjsp_madd_v2r8(vfeps,_fjsp_madd_v2r8(H,vfeps,G),F);
1167             FF               = _fjsp_madd_v2r8(vfeps,_fjsp_madd_v2r8(twovfeps,H,G),Fp);
1168             fvdw6            = _fjsp_mul_v2r8(c6_00,FF);
1169
1170             /* CUBIC SPLINE TABLE REPULSION */
1171             Y                = _fjsp_load_v2r8( vftab + vfconv.i[0] + 4 );
1172             F                = _fjsp_load_v2r8( vftab + vfconv.i[1] + 4 );
1173             GMX_FJSP_TRANSPOSE2_V2R8(Y,F);
1174             G                = _fjsp_load_v2r8( vftab + vfconv.i[0] + 6 );
1175             H                = _fjsp_load_v2r8( vftab + vfconv.i[1] + 6 );
1176             GMX_FJSP_TRANSPOSE2_V2R8(G,H);
1177             Fp               = _fjsp_madd_v2r8(vfeps,_fjsp_madd_v2r8(H,vfeps,G),F);
1178             FF               = _fjsp_madd_v2r8(vfeps,_fjsp_madd_v2r8(twovfeps,H,G),Fp);
1179             fvdw12           = _fjsp_mul_v2r8(c12_00,FF);
1180             fvdw             = _fjsp_neg_v2r8(_fjsp_mul_v2r8(_fjsp_add_v2r8(fvdw6,fvdw12),_fjsp_mul_v2r8(vftabscale,rinv00)));
1181
1182             fscal            = _fjsp_add_v2r8(felec,fvdw);
1183
1184             /* Update vectorial force */
1185             fix0             = _fjsp_madd_v2r8(dx00,fscal,fix0);
1186             fiy0             = _fjsp_madd_v2r8(dy00,fscal,fiy0);
1187             fiz0             = _fjsp_madd_v2r8(dz00,fscal,fiz0);
1188
1189             fjx0             = _fjsp_madd_v2r8(dx00,fscal,fjx0);
1190             fjy0             = _fjsp_madd_v2r8(dy00,fscal,fjy0);
1191             fjz0             = _fjsp_madd_v2r8(dz00,fscal,fjz0);
1192
1193             /**************************
1194              * CALCULATE INTERACTIONS *
1195              **************************/
1196
1197             /* REACTION-FIELD ELECTROSTATICS */
1198             felec            = _fjsp_mul_v2r8(qq01,_fjsp_msub_v2r8(rinv01,rinvsq01,krf2));
1199
1200             fscal            = felec;
1201
1202             /* Update vectorial force */
1203             fix0             = _fjsp_madd_v2r8(dx01,fscal,fix0);
1204             fiy0             = _fjsp_madd_v2r8(dy01,fscal,fiy0);
1205             fiz0             = _fjsp_madd_v2r8(dz01,fscal,fiz0);
1206
1207             fjx1             = _fjsp_madd_v2r8(dx01,fscal,fjx1);
1208             fjy1             = _fjsp_madd_v2r8(dy01,fscal,fjy1);
1209             fjz1             = _fjsp_madd_v2r8(dz01,fscal,fjz1);
1210
1211             /**************************
1212              * CALCULATE INTERACTIONS *
1213              **************************/
1214
1215             /* REACTION-FIELD ELECTROSTATICS */
1216             felec            = _fjsp_mul_v2r8(qq02,_fjsp_msub_v2r8(rinv02,rinvsq02,krf2));
1217
1218             fscal            = felec;
1219
1220             /* Update vectorial force */
1221             fix0             = _fjsp_madd_v2r8(dx02,fscal,fix0);
1222             fiy0             = _fjsp_madd_v2r8(dy02,fscal,fiy0);
1223             fiz0             = _fjsp_madd_v2r8(dz02,fscal,fiz0);
1224
1225             fjx2             = _fjsp_madd_v2r8(dx02,fscal,fjx2);
1226             fjy2             = _fjsp_madd_v2r8(dy02,fscal,fjy2);
1227             fjz2             = _fjsp_madd_v2r8(dz02,fscal,fjz2);
1228
1229             /**************************
1230              * CALCULATE INTERACTIONS *
1231              **************************/
1232
1233             /* REACTION-FIELD ELECTROSTATICS */
1234             felec            = _fjsp_mul_v2r8(qq10,_fjsp_msub_v2r8(rinv10,rinvsq10,krf2));
1235
1236             fscal            = felec;
1237
1238             /* Update vectorial force */
1239             fix1             = _fjsp_madd_v2r8(dx10,fscal,fix1);
1240             fiy1             = _fjsp_madd_v2r8(dy10,fscal,fiy1);
1241             fiz1             = _fjsp_madd_v2r8(dz10,fscal,fiz1);
1242
1243             fjx0             = _fjsp_madd_v2r8(dx10,fscal,fjx0);
1244             fjy0             = _fjsp_madd_v2r8(dy10,fscal,fjy0);
1245             fjz0             = _fjsp_madd_v2r8(dz10,fscal,fjz0);
1246
1247             /**************************
1248              * CALCULATE INTERACTIONS *
1249              **************************/
1250
1251             /* REACTION-FIELD ELECTROSTATICS */
1252             felec            = _fjsp_mul_v2r8(qq11,_fjsp_msub_v2r8(rinv11,rinvsq11,krf2));
1253
1254             fscal            = felec;
1255
1256             /* Update vectorial force */
1257             fix1             = _fjsp_madd_v2r8(dx11,fscal,fix1);
1258             fiy1             = _fjsp_madd_v2r8(dy11,fscal,fiy1);
1259             fiz1             = _fjsp_madd_v2r8(dz11,fscal,fiz1);
1260
1261             fjx1             = _fjsp_madd_v2r8(dx11,fscal,fjx1);
1262             fjy1             = _fjsp_madd_v2r8(dy11,fscal,fjy1);
1263             fjz1             = _fjsp_madd_v2r8(dz11,fscal,fjz1);
1264
1265             /**************************
1266              * CALCULATE INTERACTIONS *
1267              **************************/
1268
1269             /* REACTION-FIELD ELECTROSTATICS */
1270             felec            = _fjsp_mul_v2r8(qq12,_fjsp_msub_v2r8(rinv12,rinvsq12,krf2));
1271
1272             fscal            = felec;
1273
1274             /* Update vectorial force */
1275             fix1             = _fjsp_madd_v2r8(dx12,fscal,fix1);
1276             fiy1             = _fjsp_madd_v2r8(dy12,fscal,fiy1);
1277             fiz1             = _fjsp_madd_v2r8(dz12,fscal,fiz1);
1278
1279             fjx2             = _fjsp_madd_v2r8(dx12,fscal,fjx2);
1280             fjy2             = _fjsp_madd_v2r8(dy12,fscal,fjy2);
1281             fjz2             = _fjsp_madd_v2r8(dz12,fscal,fjz2);
1282
1283             /**************************
1284              * CALCULATE INTERACTIONS *
1285              **************************/
1286
1287             /* REACTION-FIELD ELECTROSTATICS */
1288             felec            = _fjsp_mul_v2r8(qq20,_fjsp_msub_v2r8(rinv20,rinvsq20,krf2));
1289
1290             fscal            = felec;
1291
1292             /* Update vectorial force */
1293             fix2             = _fjsp_madd_v2r8(dx20,fscal,fix2);
1294             fiy2             = _fjsp_madd_v2r8(dy20,fscal,fiy2);
1295             fiz2             = _fjsp_madd_v2r8(dz20,fscal,fiz2);
1296
1297             fjx0             = _fjsp_madd_v2r8(dx20,fscal,fjx0);
1298             fjy0             = _fjsp_madd_v2r8(dy20,fscal,fjy0);
1299             fjz0             = _fjsp_madd_v2r8(dz20,fscal,fjz0);
1300
1301             /**************************
1302              * CALCULATE INTERACTIONS *
1303              **************************/
1304
1305             /* REACTION-FIELD ELECTROSTATICS */
1306             felec            = _fjsp_mul_v2r8(qq21,_fjsp_msub_v2r8(rinv21,rinvsq21,krf2));
1307
1308             fscal            = felec;
1309
1310             /* Update vectorial force */
1311             fix2             = _fjsp_madd_v2r8(dx21,fscal,fix2);
1312             fiy2             = _fjsp_madd_v2r8(dy21,fscal,fiy2);
1313             fiz2             = _fjsp_madd_v2r8(dz21,fscal,fiz2);
1314
1315             fjx1             = _fjsp_madd_v2r8(dx21,fscal,fjx1);
1316             fjy1             = _fjsp_madd_v2r8(dy21,fscal,fjy1);
1317             fjz1             = _fjsp_madd_v2r8(dz21,fscal,fjz1);
1318
1319             /**************************
1320              * CALCULATE INTERACTIONS *
1321              **************************/
1322
1323             /* REACTION-FIELD ELECTROSTATICS */
1324             felec            = _fjsp_mul_v2r8(qq22,_fjsp_msub_v2r8(rinv22,rinvsq22,krf2));
1325
1326             fscal            = felec;
1327
1328             /* Update vectorial force */
1329             fix2             = _fjsp_madd_v2r8(dx22,fscal,fix2);
1330             fiy2             = _fjsp_madd_v2r8(dy22,fscal,fiy2);
1331             fiz2             = _fjsp_madd_v2r8(dz22,fscal,fiz2);
1332
1333             fjx2             = _fjsp_madd_v2r8(dx22,fscal,fjx2);
1334             fjy2             = _fjsp_madd_v2r8(dy22,fscal,fjy2);
1335             fjz2             = _fjsp_madd_v2r8(dz22,fscal,fjz2);
1336
1337             gmx_fjsp_decrement_3rvec_2ptr_swizzle_v2r8(f+j_coord_offsetA,f+j_coord_offsetB,fjx0,fjy0,fjz0,fjx1,fjy1,fjz1,fjx2,fjy2,fjz2);
1338
1339             /* Inner loop uses 297 flops */
1340         }
1341
1342         if(jidx<j_index_end)
1343         {
1344
1345             jnrA             = jjnr[jidx];
1346             j_coord_offsetA  = DIM*jnrA;
1347
1348             /* load j atom coordinates */
1349             gmx_fjsp_load_3rvec_1ptr_swizzle_v2r8(x+j_coord_offsetA,
1350                                               &jx0,&jy0,&jz0,&jx1,&jy1,&jz1,&jx2,&jy2,&jz2);
1351
1352             /* Calculate displacement vector */
1353             dx00             = _fjsp_sub_v2r8(ix0,jx0);
1354             dy00             = _fjsp_sub_v2r8(iy0,jy0);
1355             dz00             = _fjsp_sub_v2r8(iz0,jz0);
1356             dx01             = _fjsp_sub_v2r8(ix0,jx1);
1357             dy01             = _fjsp_sub_v2r8(iy0,jy1);
1358             dz01             = _fjsp_sub_v2r8(iz0,jz1);
1359             dx02             = _fjsp_sub_v2r8(ix0,jx2);
1360             dy02             = _fjsp_sub_v2r8(iy0,jy2);
1361             dz02             = _fjsp_sub_v2r8(iz0,jz2);
1362             dx10             = _fjsp_sub_v2r8(ix1,jx0);
1363             dy10             = _fjsp_sub_v2r8(iy1,jy0);
1364             dz10             = _fjsp_sub_v2r8(iz1,jz0);
1365             dx11             = _fjsp_sub_v2r8(ix1,jx1);
1366             dy11             = _fjsp_sub_v2r8(iy1,jy1);
1367             dz11             = _fjsp_sub_v2r8(iz1,jz1);
1368             dx12             = _fjsp_sub_v2r8(ix1,jx2);
1369             dy12             = _fjsp_sub_v2r8(iy1,jy2);
1370             dz12             = _fjsp_sub_v2r8(iz1,jz2);
1371             dx20             = _fjsp_sub_v2r8(ix2,jx0);
1372             dy20             = _fjsp_sub_v2r8(iy2,jy0);
1373             dz20             = _fjsp_sub_v2r8(iz2,jz0);
1374             dx21             = _fjsp_sub_v2r8(ix2,jx1);
1375             dy21             = _fjsp_sub_v2r8(iy2,jy1);
1376             dz21             = _fjsp_sub_v2r8(iz2,jz1);
1377             dx22             = _fjsp_sub_v2r8(ix2,jx2);
1378             dy22             = _fjsp_sub_v2r8(iy2,jy2);
1379             dz22             = _fjsp_sub_v2r8(iz2,jz2);
1380
1381             /* Calculate squared distance and things based on it */
1382             rsq00            = gmx_fjsp_calc_rsq_v2r8(dx00,dy00,dz00);
1383             rsq01            = gmx_fjsp_calc_rsq_v2r8(dx01,dy01,dz01);
1384             rsq02            = gmx_fjsp_calc_rsq_v2r8(dx02,dy02,dz02);
1385             rsq10            = gmx_fjsp_calc_rsq_v2r8(dx10,dy10,dz10);
1386             rsq11            = gmx_fjsp_calc_rsq_v2r8(dx11,dy11,dz11);
1387             rsq12            = gmx_fjsp_calc_rsq_v2r8(dx12,dy12,dz12);
1388             rsq20            = gmx_fjsp_calc_rsq_v2r8(dx20,dy20,dz20);
1389             rsq21            = gmx_fjsp_calc_rsq_v2r8(dx21,dy21,dz21);
1390             rsq22            = gmx_fjsp_calc_rsq_v2r8(dx22,dy22,dz22);
1391
1392             rinv00           = gmx_fjsp_invsqrt_v2r8(rsq00);
1393             rinv01           = gmx_fjsp_invsqrt_v2r8(rsq01);
1394             rinv02           = gmx_fjsp_invsqrt_v2r8(rsq02);
1395             rinv10           = gmx_fjsp_invsqrt_v2r8(rsq10);
1396             rinv11           = gmx_fjsp_invsqrt_v2r8(rsq11);
1397             rinv12           = gmx_fjsp_invsqrt_v2r8(rsq12);
1398             rinv20           = gmx_fjsp_invsqrt_v2r8(rsq20);
1399             rinv21           = gmx_fjsp_invsqrt_v2r8(rsq21);
1400             rinv22           = gmx_fjsp_invsqrt_v2r8(rsq22);
1401
1402             rinvsq00         = _fjsp_mul_v2r8(rinv00,rinv00);
1403             rinvsq01         = _fjsp_mul_v2r8(rinv01,rinv01);
1404             rinvsq02         = _fjsp_mul_v2r8(rinv02,rinv02);
1405             rinvsq10         = _fjsp_mul_v2r8(rinv10,rinv10);
1406             rinvsq11         = _fjsp_mul_v2r8(rinv11,rinv11);
1407             rinvsq12         = _fjsp_mul_v2r8(rinv12,rinv12);
1408             rinvsq20         = _fjsp_mul_v2r8(rinv20,rinv20);
1409             rinvsq21         = _fjsp_mul_v2r8(rinv21,rinv21);
1410             rinvsq22         = _fjsp_mul_v2r8(rinv22,rinv22);
1411
1412             fjx0             = _fjsp_setzero_v2r8();
1413             fjy0             = _fjsp_setzero_v2r8();
1414             fjz0             = _fjsp_setzero_v2r8();
1415             fjx1             = _fjsp_setzero_v2r8();
1416             fjy1             = _fjsp_setzero_v2r8();
1417             fjz1             = _fjsp_setzero_v2r8();
1418             fjx2             = _fjsp_setzero_v2r8();
1419             fjy2             = _fjsp_setzero_v2r8();
1420             fjz2             = _fjsp_setzero_v2r8();
1421
1422             /**************************
1423              * CALCULATE INTERACTIONS *
1424              **************************/
1425
1426             r00              = _fjsp_mul_v2r8(rsq00,rinv00);
1427
1428             /* Calculate table index by multiplying r with table scale and truncate to integer */
1429             rt               = _fjsp_mul_v2r8(r00,vftabscale);
1430             itab_tmp         = _fjsp_dtox_v2r8(rt);
1431             vfeps            = _fjsp_sub_v2r8(rt, _fjsp_xtod_v2r8(itab_tmp));
1432             twovfeps         = _fjsp_add_v2r8(vfeps,vfeps);
1433             _fjsp_store_v2r8(&vfconv.simd,itab_tmp);
1434
1435             vfconv.i[0]     *= 8;
1436             vfconv.i[1]     *= 8;
1437
1438             /* REACTION-FIELD ELECTROSTATICS */
1439             felec            = _fjsp_mul_v2r8(qq00,_fjsp_msub_v2r8(rinv00,rinvsq00,krf2));
1440
1441             /* CUBIC SPLINE TABLE DISPERSION */
1442             Y                = _fjsp_load_v2r8( vftab + vfconv.i[0] );
1443             F                = _fjsp_setzero_v2r8();
1444             GMX_FJSP_TRANSPOSE2_V2R8(Y,F);
1445             G                = _fjsp_load_v2r8( vftab + vfconv.i[0] + 2 );
1446             H                = _fjsp_setzero_v2r8();
1447             GMX_FJSP_TRANSPOSE2_V2R8(G,H);
1448             Fp               = _fjsp_madd_v2r8(vfeps,_fjsp_madd_v2r8(H,vfeps,G),F);
1449             FF               = _fjsp_madd_v2r8(vfeps,_fjsp_madd_v2r8(twovfeps,H,G),Fp);
1450             fvdw6            = _fjsp_mul_v2r8(c6_00,FF);
1451
1452             /* CUBIC SPLINE TABLE REPULSION */
1453             Y                = _fjsp_load_v2r8( vftab + vfconv.i[0] + 4 );
1454             F                = _fjsp_setzero_v2r8();
1455             GMX_FJSP_TRANSPOSE2_V2R8(Y,F);
1456             G                = _fjsp_load_v2r8( vftab + vfconv.i[0] + 6 );
1457             H                = _fjsp_setzero_v2r8();
1458             GMX_FJSP_TRANSPOSE2_V2R8(G,H);
1459             Fp               = _fjsp_madd_v2r8(vfeps,_fjsp_madd_v2r8(H,vfeps,G),F);
1460             FF               = _fjsp_madd_v2r8(vfeps,_fjsp_madd_v2r8(twovfeps,H,G),Fp);
1461             fvdw12           = _fjsp_mul_v2r8(c12_00,FF);
1462             fvdw             = _fjsp_neg_v2r8(_fjsp_mul_v2r8(_fjsp_add_v2r8(fvdw6,fvdw12),_fjsp_mul_v2r8(vftabscale,rinv00)));
1463
1464             fscal            = _fjsp_add_v2r8(felec,fvdw);
1465
1466             fscal            = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
1467
1468             /* Update vectorial force */
1469             fix0             = _fjsp_madd_v2r8(dx00,fscal,fix0);
1470             fiy0             = _fjsp_madd_v2r8(dy00,fscal,fiy0);
1471             fiz0             = _fjsp_madd_v2r8(dz00,fscal,fiz0);
1472
1473             fjx0             = _fjsp_madd_v2r8(dx00,fscal,fjx0);
1474             fjy0             = _fjsp_madd_v2r8(dy00,fscal,fjy0);
1475             fjz0             = _fjsp_madd_v2r8(dz00,fscal,fjz0);
1476
1477             /**************************
1478              * CALCULATE INTERACTIONS *
1479              **************************/
1480
1481             /* REACTION-FIELD ELECTROSTATICS */
1482             felec            = _fjsp_mul_v2r8(qq01,_fjsp_msub_v2r8(rinv01,rinvsq01,krf2));
1483
1484             fscal            = felec;
1485
1486             fscal            = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
1487
1488             /* Update vectorial force */
1489             fix0             = _fjsp_madd_v2r8(dx01,fscal,fix0);
1490             fiy0             = _fjsp_madd_v2r8(dy01,fscal,fiy0);
1491             fiz0             = _fjsp_madd_v2r8(dz01,fscal,fiz0);
1492
1493             fjx1             = _fjsp_madd_v2r8(dx01,fscal,fjx1);
1494             fjy1             = _fjsp_madd_v2r8(dy01,fscal,fjy1);
1495             fjz1             = _fjsp_madd_v2r8(dz01,fscal,fjz1);
1496
1497             /**************************
1498              * CALCULATE INTERACTIONS *
1499              **************************/
1500
1501             /* REACTION-FIELD ELECTROSTATICS */
1502             felec            = _fjsp_mul_v2r8(qq02,_fjsp_msub_v2r8(rinv02,rinvsq02,krf2));
1503
1504             fscal            = felec;
1505
1506             fscal            = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
1507
1508             /* Update vectorial force */
1509             fix0             = _fjsp_madd_v2r8(dx02,fscal,fix0);
1510             fiy0             = _fjsp_madd_v2r8(dy02,fscal,fiy0);
1511             fiz0             = _fjsp_madd_v2r8(dz02,fscal,fiz0);
1512
1513             fjx2             = _fjsp_madd_v2r8(dx02,fscal,fjx2);
1514             fjy2             = _fjsp_madd_v2r8(dy02,fscal,fjy2);
1515             fjz2             = _fjsp_madd_v2r8(dz02,fscal,fjz2);
1516
1517             /**************************
1518              * CALCULATE INTERACTIONS *
1519              **************************/
1520
1521             /* REACTION-FIELD ELECTROSTATICS */
1522             felec            = _fjsp_mul_v2r8(qq10,_fjsp_msub_v2r8(rinv10,rinvsq10,krf2));
1523
1524             fscal            = felec;
1525
1526             fscal            = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
1527
1528             /* Update vectorial force */
1529             fix1             = _fjsp_madd_v2r8(dx10,fscal,fix1);
1530             fiy1             = _fjsp_madd_v2r8(dy10,fscal,fiy1);
1531             fiz1             = _fjsp_madd_v2r8(dz10,fscal,fiz1);
1532
1533             fjx0             = _fjsp_madd_v2r8(dx10,fscal,fjx0);
1534             fjy0             = _fjsp_madd_v2r8(dy10,fscal,fjy0);
1535             fjz0             = _fjsp_madd_v2r8(dz10,fscal,fjz0);
1536
1537             /**************************
1538              * CALCULATE INTERACTIONS *
1539              **************************/
1540
1541             /* REACTION-FIELD ELECTROSTATICS */
1542             felec            = _fjsp_mul_v2r8(qq11,_fjsp_msub_v2r8(rinv11,rinvsq11,krf2));
1543
1544             fscal            = felec;
1545
1546             fscal            = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
1547
1548             /* Update vectorial force */
1549             fix1             = _fjsp_madd_v2r8(dx11,fscal,fix1);
1550             fiy1             = _fjsp_madd_v2r8(dy11,fscal,fiy1);
1551             fiz1             = _fjsp_madd_v2r8(dz11,fscal,fiz1);
1552
1553             fjx1             = _fjsp_madd_v2r8(dx11,fscal,fjx1);
1554             fjy1             = _fjsp_madd_v2r8(dy11,fscal,fjy1);
1555             fjz1             = _fjsp_madd_v2r8(dz11,fscal,fjz1);
1556
1557             /**************************
1558              * CALCULATE INTERACTIONS *
1559              **************************/
1560
1561             /* REACTION-FIELD ELECTROSTATICS */
1562             felec            = _fjsp_mul_v2r8(qq12,_fjsp_msub_v2r8(rinv12,rinvsq12,krf2));
1563
1564             fscal            = felec;
1565
1566             fscal            = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
1567
1568             /* Update vectorial force */
1569             fix1             = _fjsp_madd_v2r8(dx12,fscal,fix1);
1570             fiy1             = _fjsp_madd_v2r8(dy12,fscal,fiy1);
1571             fiz1             = _fjsp_madd_v2r8(dz12,fscal,fiz1);
1572
1573             fjx2             = _fjsp_madd_v2r8(dx12,fscal,fjx2);
1574             fjy2             = _fjsp_madd_v2r8(dy12,fscal,fjy2);
1575             fjz2             = _fjsp_madd_v2r8(dz12,fscal,fjz2);
1576
1577             /**************************
1578              * CALCULATE INTERACTIONS *
1579              **************************/
1580
1581             /* REACTION-FIELD ELECTROSTATICS */
1582             felec            = _fjsp_mul_v2r8(qq20,_fjsp_msub_v2r8(rinv20,rinvsq20,krf2));
1583
1584             fscal            = felec;
1585
1586             fscal            = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
1587
1588             /* Update vectorial force */
1589             fix2             = _fjsp_madd_v2r8(dx20,fscal,fix2);
1590             fiy2             = _fjsp_madd_v2r8(dy20,fscal,fiy2);
1591             fiz2             = _fjsp_madd_v2r8(dz20,fscal,fiz2);
1592
1593             fjx0             = _fjsp_madd_v2r8(dx20,fscal,fjx0);
1594             fjy0             = _fjsp_madd_v2r8(dy20,fscal,fjy0);
1595             fjz0             = _fjsp_madd_v2r8(dz20,fscal,fjz0);
1596
1597             /**************************
1598              * CALCULATE INTERACTIONS *
1599              **************************/
1600
1601             /* REACTION-FIELD ELECTROSTATICS */
1602             felec            = _fjsp_mul_v2r8(qq21,_fjsp_msub_v2r8(rinv21,rinvsq21,krf2));
1603
1604             fscal            = felec;
1605
1606             fscal            = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
1607
1608             /* Update vectorial force */
1609             fix2             = _fjsp_madd_v2r8(dx21,fscal,fix2);
1610             fiy2             = _fjsp_madd_v2r8(dy21,fscal,fiy2);
1611             fiz2             = _fjsp_madd_v2r8(dz21,fscal,fiz2);
1612
1613             fjx1             = _fjsp_madd_v2r8(dx21,fscal,fjx1);
1614             fjy1             = _fjsp_madd_v2r8(dy21,fscal,fjy1);
1615             fjz1             = _fjsp_madd_v2r8(dz21,fscal,fjz1);
1616
1617             /**************************
1618              * CALCULATE INTERACTIONS *
1619              **************************/
1620
1621             /* REACTION-FIELD ELECTROSTATICS */
1622             felec            = _fjsp_mul_v2r8(qq22,_fjsp_msub_v2r8(rinv22,rinvsq22,krf2));
1623
1624             fscal            = felec;
1625
1626             fscal            = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
1627
1628             /* Update vectorial force */
1629             fix2             = _fjsp_madd_v2r8(dx22,fscal,fix2);
1630             fiy2             = _fjsp_madd_v2r8(dy22,fscal,fiy2);
1631             fiz2             = _fjsp_madd_v2r8(dz22,fscal,fiz2);
1632
1633             fjx2             = _fjsp_madd_v2r8(dx22,fscal,fjx2);
1634             fjy2             = _fjsp_madd_v2r8(dy22,fscal,fjy2);
1635             fjz2             = _fjsp_madd_v2r8(dz22,fscal,fjz2);
1636
1637             gmx_fjsp_decrement_3rvec_1ptr_swizzle_v2r8(f+j_coord_offsetA,fjx0,fjy0,fjz0,fjx1,fjy1,fjz1,fjx2,fjy2,fjz2);
1638
1639             /* Inner loop uses 297 flops */
1640         }
1641
1642         /* End of innermost loop */
1643
1644         gmx_fjsp_update_iforce_3atom_swizzle_v2r8(fix0,fiy0,fiz0,fix1,fiy1,fiz1,fix2,fiy2,fiz2,
1645                                               f+i_coord_offset,fshift+i_shift_offset);
1646
1647         /* Increment number of inner iterations */
1648         inneriter                  += j_index_end - j_index_start;
1649
1650         /* Outer loop uses 18 flops */
1651     }
1652
1653     /* Increment number of outer iterations */
1654     outeriter        += nri;
1655
1656     /* Update outer/inner flops */
1657
1658     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W3W3_F,outeriter*18 + inneriter*297);
1659 }