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) 2012,2013,2014, 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 "gromacs/legacyheaders/types/simple.h"
  44 #include "gromacs/math/vec.h"
  45 #include "gromacs/legacyheaders/nrnb.h"
  46
  47 #include "kernelutil_sparc64_hpc_ace_double.h"
  48
  49 /*
  50  * Gromacs nonbonded kernel:   nb_kernel_ElecRF_VdwCSTab_GeomW3W3_VF_sparc64_hpc_ace_double
  51  * Electrostatics interaction: ReactionField
  52  * VdW interaction:            CubicSplineTable
  53  * Geometry:                   Water3-Water3
  54  * Calculate force/pot:        PotentialAndForce
  55  */
  56 void
  57 nb_kernel_ElecRF_VdwCSTab_GeomW3W3_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              vdwioffset1;
  82     _fjsp_v2r8       ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
  83     int              vdwioffset2;
  84     _fjsp_v2r8       ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
  85     int              vdwjidx0A,vdwjidx0B;
  86     _fjsp_v2r8       jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
  87     int              vdwjidx1A,vdwjidx1B;
  88     _fjsp_v2r8       jx1,jy1,jz1,fjx1,fjy1,fjz1,jq1,isaj1;
  89     int              vdwjidx2A,vdwjidx2B;
  90     _fjsp_v2r8       jx2,jy2,jz2,fjx2,fjy2,fjz2,jq2,isaj2;
  91     _fjsp_v2r8       dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00;
  92     _fjsp_v2r8       dx01,dy01,dz01,rsq01,rinv01,rinvsq01,r01,qq01,c6_01,c12_01;
  93     _fjsp_v2r8       dx02,dy02,dz02,rsq02,rinv02,rinvsq02,r02,qq02,c6_02,c12_02;
  94     _fjsp_v2r8       dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10;
  95     _fjsp_v2r8       dx11,dy11,dz11,rsq11,rinv11,rinvsq11,r11,qq11,c6_11,c12_11;
  96     _fjsp_v2r8       dx12,dy12,dz12,rsq12,rinv12,rinvsq12,r12,qq12,c6_12,c12_12;
  97     _fjsp_v2r8       dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20;
  98     _fjsp_v2r8       dx21,dy21,dz21,rsq21,rinv21,rinvsq21,r21,qq21,c6_21,c12_21;
  99     _fjsp_v2r8       dx22,dy22,dz22,rsq22,rinv22,rinvsq22,r22,qq22,c6_22,c12_22;
 100     _fjsp_v2r8       velec,felec,velecsum,facel,crf,krf,krf2;
 101     real             *charge;
 102     int              nvdwtype;
 103     _fjsp_v2r8       rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,sh_vdw_invrcut6;
 104     int              *vdwtype;
 105     real             *vdwparam;
 106     _fjsp_v2r8       one_sixth   = gmx_fjsp_set1_v2r8(1.0/6.0);
 107     _fjsp_v2r8       one_twelfth = gmx_fjsp_set1_v2r8(1.0/12.0);
 108     _fjsp_v2r8       rt,vfeps,vftabscale,Y,F,G,H,Heps,Fp,VV,FF,twovfeps;
 109     real             *vftab;
 110     _fjsp_v2r8       itab_tmp;
 111     _fjsp_v2r8       dummy_mask,cutoff_mask;
 112     _fjsp_v2r8       one     = gmx_fjsp_set1_v2r8(1.0);
 113     _fjsp_v2r8       two     = gmx_fjsp_set1_v2r8(2.0);
 114     union { _fjsp_v2r8 simd; long long int i[2]; } vfconv,gbconv,ewconv;
 115
 116     x                = xx[0];
 117     f                = ff[0];
 118
 119     nri              = nlist->nri;
 120     iinr             = nlist->iinr;
 121     jindex           = nlist->jindex;
 122     jjnr             = nlist->jjnr;
 123     shiftidx         = nlist->shift;
 124     gid              = nlist->gid;
 125     shiftvec         = fr->shift_vec[0];
 126     fshift           = fr->fshift[0];
 127     facel            = gmx_fjsp_set1_v2r8(fr->epsfac);
 128     charge           = mdatoms->chargeA;
 129     krf              = gmx_fjsp_set1_v2r8(fr->ic->k_rf);
 130     krf2             = gmx_fjsp_set1_v2r8(fr->ic->k_rf*2.0);
 131     crf              = gmx_fjsp_set1_v2r8(fr->ic->c_rf);
 132     nvdwtype         = fr->ntype;
 133     vdwparam         = fr->nbfp;
 134     vdwtype          = mdatoms->typeA;
 135
 136     vftab            = kernel_data->table_vdw->data;
 137     vftabscale       = gmx_fjsp_set1_v2r8(kernel_data->table_vdw->scale);
 138
 139     /* Setup water-specific parameters */
 140     inr              = nlist->iinr[0];
 141     iq0              = _fjsp_mul_v2r8(facel,gmx_fjsp_set1_v2r8(charge[inr+0]));
 142     iq1              = _fjsp_mul_v2r8(facel,gmx_fjsp_set1_v2r8(charge[inr+1]));
 143     iq2              = _fjsp_mul_v2r8(facel,gmx_fjsp_set1_v2r8(charge[inr+2]));
 144     vdwioffset0      = 2*nvdwtype*vdwtype[inr+0];
 145
 146     jq0              = gmx_fjsp_set1_v2r8(charge[inr+0]);
 147     jq1              = gmx_fjsp_set1_v2r8(charge[inr+1]);
 148     jq2              = gmx_fjsp_set1_v2r8(charge[inr+2]);
 149     vdwjidx0A        = 2*vdwtype[inr+0];
 150     qq00             = _fjsp_mul_v2r8(iq0,jq0);
 151     c6_00            = gmx_fjsp_set1_v2r8(vdwparam[vdwioffset0+vdwjidx0A]);
 152     c12_00           = gmx_fjsp_set1_v2r8(vdwparam[vdwioffset0+vdwjidx0A+1]);
 153     qq01             = _fjsp_mul_v2r8(iq0,jq1);
 154     qq02             = _fjsp_mul_v2r8(iq0,jq2);
 155     qq10             = _fjsp_mul_v2r8(iq1,jq0);
 156     qq11             = _fjsp_mul_v2r8(iq1,jq1);
 157     qq12             = _fjsp_mul_v2r8(iq1,jq2);
 158     qq20             = _fjsp_mul_v2r8(iq2,jq0);
 159     qq21             = _fjsp_mul_v2r8(iq2,jq1);
 160     qq22             = _fjsp_mul_v2r8(iq2,jq2);
 161
 162     /* Avoid stupid compiler warnings */
 163     jnrA = jnrB = 0;
 164     j_coord_offsetA = 0;
 165     j_coord_offsetB = 0;
 166
 167     outeriter        = 0;
 168     inneriter        = 0;
 169
 170     /* Start outer loop over neighborlists */
 171     for(iidx=0; iidx<nri; iidx++)
 172     {
 173         /* Load shift vector for this list */
 174         i_shift_offset   = DIM*shiftidx[iidx];
 175
 176         /* Load limits for loop over neighbors */
 177         j_index_start    = jindex[iidx];
 178         j_index_end      = jindex[iidx+1];
 179
 180         /* Get outer coordinate index */
 181         inr              = iinr[iidx];
 182         i_coord_offset   = DIM*inr;
 183
 184         /* Load i particle coords and add shift vector */
 185         gmx_fjsp_load_shift_and_3rvec_broadcast_v2r8(shiftvec+i_shift_offset,x+i_coord_offset,
 186                                                  &ix0,&iy0,&iz0,&ix1,&iy1,&iz1,&ix2,&iy2,&iz2);
 187
 188         fix0             = _fjsp_setzero_v2r8();
 189         fiy0             = _fjsp_setzero_v2r8();
 190         fiz0             = _fjsp_setzero_v2r8();
 191         fix1             = _fjsp_setzero_v2r8();
 192         fiy1             = _fjsp_setzero_v2r8();
 193         fiz1             = _fjsp_setzero_v2r8();
 194         fix2             = _fjsp_setzero_v2r8();
 195         fiy2             = _fjsp_setzero_v2r8();
 196         fiz2             = _fjsp_setzero_v2r8();
 197
 198         /* Reset potential sums */
 199         velecsum         = _fjsp_setzero_v2r8();
 200         vvdwsum          = _fjsp_setzero_v2r8();
 201
 202         /* Start inner kernel loop */
 203         for(jidx=j_index_start; jidx<j_index_end-1; jidx+=2)
 204         {
 205
 206             /* Get j neighbor index, and coordinate index */
 207             jnrA             = jjnr[jidx];
 208             jnrB             = jjnr[jidx+1];
 209             j_coord_offsetA  = DIM*jnrA;
 210             j_coord_offsetB  = DIM*jnrB;
 211
 212             /* load j atom coordinates */
 213             gmx_fjsp_load_3rvec_2ptr_swizzle_v2r8(x+j_coord_offsetA,x+j_coord_offsetB,
 214                                               &jx0,&jy0,&jz0,&jx1,&jy1,&jz1,&jx2,&jy2,&jz2);
 215
 216             /* Calculate displacement vector */
 217             dx00             = _fjsp_sub_v2r8(ix0,jx0);
 218             dy00             = _fjsp_sub_v2r8(iy0,jy0);
 219             dz00             = _fjsp_sub_v2r8(iz0,jz0);
 220             dx01             = _fjsp_sub_v2r8(ix0,jx1);
 221             dy01             = _fjsp_sub_v2r8(iy0,jy1);
 222             dz01             = _fjsp_sub_v2r8(iz0,jz1);
 223             dx02             = _fjsp_sub_v2r8(ix0,jx2);
 224             dy02             = _fjsp_sub_v2r8(iy0,jy2);
 225             dz02             = _fjsp_sub_v2r8(iz0,jz2);
 226             dx10             = _fjsp_sub_v2r8(ix1,jx0);
 227             dy10             = _fjsp_sub_v2r8(iy1,jy0);
 228             dz10             = _fjsp_sub_v2r8(iz1,jz0);
 229             dx11             = _fjsp_sub_v2r8(ix1,jx1);
 230             dy11             = _fjsp_sub_v2r8(iy1,jy1);
 231             dz11             = _fjsp_sub_v2r8(iz1,jz1);
 232             dx12             = _fjsp_sub_v2r8(ix1,jx2);
 233             dy12             = _fjsp_sub_v2r8(iy1,jy2);
 234             dz12             = _fjsp_sub_v2r8(iz1,jz2);
 235             dx20             = _fjsp_sub_v2r8(ix2,jx0);
 236             dy20             = _fjsp_sub_v2r8(iy2,jy0);
 237             dz20             = _fjsp_sub_v2r8(iz2,jz0);
 238             dx21             = _fjsp_sub_v2r8(ix2,jx1);
 239             dy21             = _fjsp_sub_v2r8(iy2,jy1);
 240             dz21             = _fjsp_sub_v2r8(iz2,jz1);
 241             dx22             = _fjsp_sub_v2r8(ix2,jx2);
 242             dy22             = _fjsp_sub_v2r8(iy2,jy2);
 243             dz22             = _fjsp_sub_v2r8(iz2,jz2);
 244
 245             /* Calculate squared distance and things based on it */
 246             rsq00            = gmx_fjsp_calc_rsq_v2r8(dx00,dy00,dz00);
 247             rsq01            = gmx_fjsp_calc_rsq_v2r8(dx01,dy01,dz01);
 248             rsq02            = gmx_fjsp_calc_rsq_v2r8(dx02,dy02,dz02);
 249             rsq10            = gmx_fjsp_calc_rsq_v2r8(dx10,dy10,dz10);
 250             rsq11            = gmx_fjsp_calc_rsq_v2r8(dx11,dy11,dz11);
 251             rsq12            = gmx_fjsp_calc_rsq_v2r8(dx12,dy12,dz12);
 252             rsq20            = gmx_fjsp_calc_rsq_v2r8(dx20,dy20,dz20);
 253             rsq21            = gmx_fjsp_calc_rsq_v2r8(dx21,dy21,dz21);
 254             rsq22            = gmx_fjsp_calc_rsq_v2r8(dx22,dy22,dz22);
 255
 256             rinv00           = gmx_fjsp_invsqrt_v2r8(rsq00);
 257             rinv01           = gmx_fjsp_invsqrt_v2r8(rsq01);
 258             rinv02           = gmx_fjsp_invsqrt_v2r8(rsq02);
 259             rinv10           = gmx_fjsp_invsqrt_v2r8(rsq10);
 260             rinv11           = gmx_fjsp_invsqrt_v2r8(rsq11);
 261             rinv12           = gmx_fjsp_invsqrt_v2r8(rsq12);
 262             rinv20           = gmx_fjsp_invsqrt_v2r8(rsq20);
 263             rinv21           = gmx_fjsp_invsqrt_v2r8(rsq21);
 264             rinv22           = gmx_fjsp_invsqrt_v2r8(rsq22);
 265
 266             rinvsq00         = _fjsp_mul_v2r8(rinv00,rinv00);
 267             rinvsq01         = _fjsp_mul_v2r8(rinv01,rinv01);
 268             rinvsq02         = _fjsp_mul_v2r8(rinv02,rinv02);
 269             rinvsq10         = _fjsp_mul_v2r8(rinv10,rinv10);
 270             rinvsq11         = _fjsp_mul_v2r8(rinv11,rinv11);
 271             rinvsq12         = _fjsp_mul_v2r8(rinv12,rinv12);
 272             rinvsq20         = _fjsp_mul_v2r8(rinv20,rinv20);
 273             rinvsq21         = _fjsp_mul_v2r8(rinv21,rinv21);
 274             rinvsq22         = _fjsp_mul_v2r8(rinv22,rinv22);
 275
 276             fjx0             = _fjsp_setzero_v2r8();
 277             fjy0             = _fjsp_setzero_v2r8();
 278             fjz0             = _fjsp_setzero_v2r8();
 279             fjx1             = _fjsp_setzero_v2r8();
 280             fjy1             = _fjsp_setzero_v2r8();
 281             fjz1             = _fjsp_setzero_v2r8();
 282             fjx2             = _fjsp_setzero_v2r8();
 283             fjy2             = _fjsp_setzero_v2r8();
 284             fjz2             = _fjsp_setzero_v2r8();
 285
 286             /**************************
 287              * CALCULATE INTERACTIONS *
 288              **************************/
 289
 290             r00              = _fjsp_mul_v2r8(rsq00,rinv00);
 291
 292             /* Calculate table index by multiplying r with table scale and truncate to integer */
 293             rt               = _fjsp_mul_v2r8(r00,vftabscale);
 294             itab_tmp         = _fjsp_dtox_v2r8(rt);
 295             vfeps            = _fjsp_sub_v2r8(rt, _fjsp_xtod_v2r8(itab_tmp));
 296             twovfeps         = _fjsp_add_v2r8(vfeps,vfeps);
 297             _fjsp_store_v2r8(&vfconv.simd,itab_tmp);
 298
 299             vfconv.i[0]     *= 8;
 300             vfconv.i[1]     *= 8;
 301
 302             /* REACTION-FIELD ELECTROSTATICS */
 303             velec            = _fjsp_mul_v2r8(qq00,_fjsp_sub_v2r8(_fjsp_madd_v2r8(krf,rsq00,rinv00),crf));
 304             felec            = _fjsp_mul_v2r8(qq00,_fjsp_msub_v2r8(rinv00,rinvsq00,krf2));
 305
 306             /* CUBIC SPLINE TABLE DISPERSION */
 307             Y                = _fjsp_load_v2r8( vftab + vfconv.i[0] );
 308             F                = _fjsp_load_v2r8( vftab + vfconv.i[1] );
 309             GMX_FJSP_TRANSPOSE2_V2R8(Y,F);
 310             G                = _fjsp_load_v2r8( vftab + vfconv.i[0] + 2 );
 311             H                = _fjsp_load_v2r8( vftab + vfconv.i[1] + 2 );
 312             GMX_FJSP_TRANSPOSE2_V2R8(G,H);
 313             Fp               = _fjsp_madd_v2r8(vfeps,_fjsp_madd_v2r8(H,vfeps,G),F);
 314             VV               = _fjsp_madd_v2r8(vfeps,Fp,Y);
 315             vvdw6            = _fjsp_mul_v2r8(c6_00,VV);
 316             FF               = _fjsp_madd_v2r8(vfeps,_fjsp_madd_v2r8(twovfeps,H,G),Fp);
 317             fvdw6            = _fjsp_mul_v2r8(c6_00,FF);
 318
 319             /* CUBIC SPLINE TABLE REPULSION */
 320             Y                = _fjsp_load_v2r8( vftab + vfconv.i[0] + 4 );
 321             F                = _fjsp_load_v2r8( vftab + vfconv.i[1] + 4 );
 322             GMX_FJSP_TRANSPOSE2_V2R8(Y,F);
 323             G                = _fjsp_load_v2r8( vftab + vfconv.i[0] + 6 );
 324             H                = _fjsp_load_v2r8( vftab + vfconv.i[1] + 6 );
 325             GMX_FJSP_TRANSPOSE2_V2R8(G,H);
 326             Fp               = _fjsp_madd_v2r8(vfeps,_fjsp_madd_v2r8(H,vfeps,G),F);
 327             VV               = _fjsp_madd_v2r8(vfeps,Fp,Y);
 328             vvdw12           = _fjsp_mul_v2r8(c12_00,VV);
 329             FF               = _fjsp_madd_v2r8(vfeps,_fjsp_madd_v2r8(twovfeps,H,G),Fp);
 330             fvdw12           = _fjsp_mul_v2r8(c12_00,FF);
 331             vvdw             = _fjsp_add_v2r8(vvdw12,vvdw6);
 332             fvdw             = _fjsp_neg_v2r8(_fjsp_mul_v2r8(_fjsp_add_v2r8(fvdw6,fvdw12),_fjsp_mul_v2r8(vftabscale,rinv00)));
 333
 334             /* Update potential sum for this i atom from the interaction with this j atom. */
 335             velecsum         = _fjsp_add_v2r8(velecsum,velec);
 336             vvdwsum          = _fjsp_add_v2r8(vvdwsum,vvdw);
 337
 338             fscal            = _fjsp_add_v2r8(felec,fvdw);
 339
 340             /* Update vectorial force */
 341             fix0             = _fjsp_madd_v2r8(dx00,fscal,fix0);
 342             fiy0             = _fjsp_madd_v2r8(dy00,fscal,fiy0);
 343             fiz0             = _fjsp_madd_v2r8(dz00,fscal,fiz0);
 344
 345             fjx0             = _fjsp_madd_v2r8(dx00,fscal,fjx0);
 346             fjy0             = _fjsp_madd_v2r8(dy00,fscal,fjy0);
 347             fjz0             = _fjsp_madd_v2r8(dz00,fscal,fjz0);
 348
 349             /**************************
 350              * CALCULATE INTERACTIONS *
 351              **************************/
 352
 353             /* REACTION-FIELD ELECTROSTATICS */
 354             velec            = _fjsp_mul_v2r8(qq01,_fjsp_sub_v2r8(_fjsp_madd_v2r8(krf,rsq01,rinv01),crf));
 355             felec            = _fjsp_mul_v2r8(qq01,_fjsp_msub_v2r8(rinv01,rinvsq01,krf2));
 356
 357             /* Update potential sum for this i atom from the interaction with this j atom. */
 358             velecsum         = _fjsp_add_v2r8(velecsum,velec);
 359
 360             fscal            = felec;
 361
 362             /* Update vectorial force */
 363             fix0             = _fjsp_madd_v2r8(dx01,fscal,fix0);
 364             fiy0             = _fjsp_madd_v2r8(dy01,fscal,fiy0);
 365             fiz0             = _fjsp_madd_v2r8(dz01,fscal,fiz0);
 366
 367             fjx1             = _fjsp_madd_v2r8(dx01,fscal,fjx1);
 368             fjy1             = _fjsp_madd_v2r8(dy01,fscal,fjy1);
 369             fjz1             = _fjsp_madd_v2r8(dz01,fscal,fjz1);
 370
 371             /**************************
 372              * CALCULATE INTERACTIONS *
 373              **************************/
 374
 375             /* REACTION-FIELD ELECTROSTATICS */
 376             velec            = _fjsp_mul_v2r8(qq02,_fjsp_sub_v2r8(_fjsp_madd_v2r8(krf,rsq02,rinv02),crf));
 377             felec            = _fjsp_mul_v2r8(qq02,_fjsp_msub_v2r8(rinv02,rinvsq02,krf2));
 378
 379             /* Update potential sum for this i atom from the interaction with this j atom. */
 380             velecsum         = _fjsp_add_v2r8(velecsum,velec);
 381
 382             fscal            = felec;
 383
 384             /* Update vectorial force */
 385             fix0             = _fjsp_madd_v2r8(dx02,fscal,fix0);
 386             fiy0             = _fjsp_madd_v2r8(dy02,fscal,fiy0);
 387             fiz0             = _fjsp_madd_v2r8(dz02,fscal,fiz0);
 388
 389             fjx2             = _fjsp_madd_v2r8(dx02,fscal,fjx2);
 390             fjy2             = _fjsp_madd_v2r8(dy02,fscal,fjy2);
 391             fjz2             = _fjsp_madd_v2r8(dz02,fscal,fjz2);
 392
 393             /**************************
 394              * CALCULATE INTERACTIONS *
 395              **************************/
 396
 397             /* REACTION-FIELD ELECTROSTATICS */
 398             velec            = _fjsp_mul_v2r8(qq10,_fjsp_sub_v2r8(_fjsp_madd_v2r8(krf,rsq10,rinv10),crf));
 399             felec            = _fjsp_mul_v2r8(qq10,_fjsp_msub_v2r8(rinv10,rinvsq10,krf2));
 400
 401             /* Update potential sum for this i atom from the interaction with this j atom. */
 402             velecsum         = _fjsp_add_v2r8(velecsum,velec);
 403
 404             fscal            = felec;
 405
 406             /* Update vectorial force */
 407             fix1             = _fjsp_madd_v2r8(dx10,fscal,fix1);
 408             fiy1             = _fjsp_madd_v2r8(dy10,fscal,fiy1);
 409             fiz1             = _fjsp_madd_v2r8(dz10,fscal,fiz1);
 410
 411             fjx0             = _fjsp_madd_v2r8(dx10,fscal,fjx0);
 412             fjy0             = _fjsp_madd_v2r8(dy10,fscal,fjy0);
 413             fjz0             = _fjsp_madd_v2r8(dz10,fscal,fjz0);
 414
 415             /**************************
 416              * CALCULATE INTERACTIONS *
 417              **************************/
 418
 419             /* REACTION-FIELD ELECTROSTATICS */
 420             velec            = _fjsp_mul_v2r8(qq11,_fjsp_sub_v2r8(_fjsp_madd_v2r8(krf,rsq11,rinv11),crf));
 421             felec            = _fjsp_mul_v2r8(qq11,_fjsp_msub_v2r8(rinv11,rinvsq11,krf2));
 422
 423             /* Update potential sum for this i atom from the interaction with this j atom. */
 424             velecsum         = _fjsp_add_v2r8(velecsum,velec);
 425
 426             fscal            = felec;
 427
 428             /* Update vectorial force */
 429             fix1             = _fjsp_madd_v2r8(dx11,fscal,fix1);
 430             fiy1             = _fjsp_madd_v2r8(dy11,fscal,fiy1);
 431             fiz1             = _fjsp_madd_v2r8(dz11,fscal,fiz1);
 432
 433             fjx1             = _fjsp_madd_v2r8(dx11,fscal,fjx1);
 434             fjy1             = _fjsp_madd_v2r8(dy11,fscal,fjy1);
 435             fjz1             = _fjsp_madd_v2r8(dz11,fscal,fjz1);
 436
 437             /**************************
 438              * CALCULATE INTERACTIONS *
 439              **************************/
 440
 441             /* REACTION-FIELD ELECTROSTATICS */
 442             velec            = _fjsp_mul_v2r8(qq12,_fjsp_sub_v2r8(_fjsp_madd_v2r8(krf,rsq12,rinv12),crf));
 443             felec            = _fjsp_mul_v2r8(qq12,_fjsp_msub_v2r8(rinv12,rinvsq12,krf2));
 444
 445             /* Update potential sum for this i atom from the interaction with this j atom. */
 446             velecsum         = _fjsp_add_v2r8(velecsum,velec);
 447
 448             fscal            = felec;
 449
 450             /* Update vectorial force */
 451             fix1             = _fjsp_madd_v2r8(dx12,fscal,fix1);
 452             fiy1             = _fjsp_madd_v2r8(dy12,fscal,fiy1);
 453             fiz1             = _fjsp_madd_v2r8(dz12,fscal,fiz1);
 454
 455             fjx2             = _fjsp_madd_v2r8(dx12,fscal,fjx2);
 456             fjy2             = _fjsp_madd_v2r8(dy12,fscal,fjy2);
 457             fjz2             = _fjsp_madd_v2r8(dz12,fscal,fjz2);
 458
 459             /**************************
 460              * CALCULATE INTERACTIONS *
 461              **************************/
 462
 463             /* REACTION-FIELD ELECTROSTATICS */
 464             velec            = _fjsp_mul_v2r8(qq20,_fjsp_sub_v2r8(_fjsp_madd_v2r8(krf,rsq20,rinv20),crf));
 465             felec            = _fjsp_mul_v2r8(qq20,_fjsp_msub_v2r8(rinv20,rinvsq20,krf2));
 466
 467             /* Update potential sum for this i atom from the interaction with this j atom. */
 468             velecsum         = _fjsp_add_v2r8(velecsum,velec);
 469
 470             fscal            = felec;
 471
 472             /* Update vectorial force */
 473             fix2             = _fjsp_madd_v2r8(dx20,fscal,fix2);
 474             fiy2             = _fjsp_madd_v2r8(dy20,fscal,fiy2);
 475             fiz2             = _fjsp_madd_v2r8(dz20,fscal,fiz2);
 476
 477             fjx0             = _fjsp_madd_v2r8(dx20,fscal,fjx0);
 478             fjy0             = _fjsp_madd_v2r8(dy20,fscal,fjy0);
 479             fjz0             = _fjsp_madd_v2r8(dz20,fscal,fjz0);
 480
 481             /**************************
 482              * CALCULATE INTERACTIONS *
 483              **************************/
 484
 485             /* REACTION-FIELD ELECTROSTATICS */
 486             velec            = _fjsp_mul_v2r8(qq21,_fjsp_sub_v2r8(_fjsp_madd_v2r8(krf,rsq21,rinv21),crf));
 487             felec            = _fjsp_mul_v2r8(qq21,_fjsp_msub_v2r8(rinv21,rinvsq21,krf2));
 488
 489             /* Update potential sum for this i atom from the interaction with this j atom. */
 490             velecsum         = _fjsp_add_v2r8(velecsum,velec);
 491
 492             fscal            = felec;
 493
 494             /* Update vectorial force */
 495             fix2             = _fjsp_madd_v2r8(dx21,fscal,fix2);
 496             fiy2             = _fjsp_madd_v2r8(dy21,fscal,fiy2);
 497             fiz2             = _fjsp_madd_v2r8(dz21,fscal,fiz2);
 498
 499             fjx1             = _fjsp_madd_v2r8(dx21,fscal,fjx1);
 500             fjy1             = _fjsp_madd_v2r8(dy21,fscal,fjy1);
 501             fjz1             = _fjsp_madd_v2r8(dz21,fscal,fjz1);
 502
 503             /**************************
 504              * CALCULATE INTERACTIONS *
 505              **************************/
 506
 507             /* REACTION-FIELD ELECTROSTATICS */
 508             velec            = _fjsp_mul_v2r8(qq22,_fjsp_sub_v2r8(_fjsp_madd_v2r8(krf,rsq22,rinv22),crf));
 509             felec            = _fjsp_mul_v2r8(qq22,_fjsp_msub_v2r8(rinv22,rinvsq22,krf2));
 510
 511             /* Update potential sum for this i atom from the interaction with this j atom. */
 512             velecsum         = _fjsp_add_v2r8(velecsum,velec);
 513
 514             fscal            = felec;
 515
 516             /* Update vectorial force */
 517             fix2             = _fjsp_madd_v2r8(dx22,fscal,fix2);
 518             fiy2             = _fjsp_madd_v2r8(dy22,fscal,fiy2);
 519             fiz2             = _fjsp_madd_v2r8(dz22,fscal,fiz2);
 520
 521             fjx2             = _fjsp_madd_v2r8(dx22,fscal,fjx2);
 522             fjy2             = _fjsp_madd_v2r8(dy22,fscal,fjy2);
 523             fjz2             = _fjsp_madd_v2r8(dz22,fscal,fjz2);
 524
 525             gmx_fjsp_decrement_3rvec_2ptr_swizzle_v2r8(f+j_coord_offsetA,f+j_coord_offsetB,fjx0,fjy0,fjz0,fjx1,fjy1,fjz1,fjx2,fjy2,fjz2);
 526
 527             /* Inner loop uses 350 flops */
 528         }
 529
 530         if(jidx<j_index_end)
 531         {
 532
 533             jnrA             = jjnr[jidx];
 534             j_coord_offsetA  = DIM*jnrA;
 535
 536             /* load j atom coordinates */
 537             gmx_fjsp_load_3rvec_1ptr_swizzle_v2r8(x+j_coord_offsetA,
 538                                               &jx0,&jy0,&jz0,&jx1,&jy1,&jz1,&jx2,&jy2,&jz2);
 539
 540             /* Calculate displacement vector */
 541             dx00             = _fjsp_sub_v2r8(ix0,jx0);
 542             dy00             = _fjsp_sub_v2r8(iy0,jy0);
 543             dz00             = _fjsp_sub_v2r8(iz0,jz0);
 544             dx01             = _fjsp_sub_v2r8(ix0,jx1);
 545             dy01             = _fjsp_sub_v2r8(iy0,jy1);
 546             dz01             = _fjsp_sub_v2r8(iz0,jz1);
 547             dx02             = _fjsp_sub_v2r8(ix0,jx2);
 548             dy02             = _fjsp_sub_v2r8(iy0,jy2);
 549             dz02             = _fjsp_sub_v2r8(iz0,jz2);
 550             dx10             = _fjsp_sub_v2r8(ix1,jx0);
 551             dy10             = _fjsp_sub_v2r8(iy1,jy0);
 552             dz10             = _fjsp_sub_v2r8(iz1,jz0);
 553             dx11             = _fjsp_sub_v2r8(ix1,jx1);
 554             dy11             = _fjsp_sub_v2r8(iy1,jy1);
 555             dz11             = _fjsp_sub_v2r8(iz1,jz1);
 556             dx12             = _fjsp_sub_v2r8(ix1,jx2);
 557             dy12             = _fjsp_sub_v2r8(iy1,jy2);
 558             dz12             = _fjsp_sub_v2r8(iz1,jz2);
 559             dx20             = _fjsp_sub_v2r8(ix2,jx0);
 560             dy20             = _fjsp_sub_v2r8(iy2,jy0);
 561             dz20             = _fjsp_sub_v2r8(iz2,jz0);
 562             dx21             = _fjsp_sub_v2r8(ix2,jx1);
 563             dy21             = _fjsp_sub_v2r8(iy2,jy1);
 564             dz21             = _fjsp_sub_v2r8(iz2,jz1);
 565             dx22             = _fjsp_sub_v2r8(ix2,jx2);
 566             dy22             = _fjsp_sub_v2r8(iy2,jy2);
 567             dz22             = _fjsp_sub_v2r8(iz2,jz2);
 568
 569             /* Calculate squared distance and things based on it */
 570             rsq00            = gmx_fjsp_calc_rsq_v2r8(dx00,dy00,dz00);
 571             rsq01            = gmx_fjsp_calc_rsq_v2r8(dx01,dy01,dz01);
 572             rsq02            = gmx_fjsp_calc_rsq_v2r8(dx02,dy02,dz02);
 573             rsq10            = gmx_fjsp_calc_rsq_v2r8(dx10,dy10,dz10);
 574             rsq11            = gmx_fjsp_calc_rsq_v2r8(dx11,dy11,dz11);
 575             rsq12            = gmx_fjsp_calc_rsq_v2r8(dx12,dy12,dz12);
 576             rsq20            = gmx_fjsp_calc_rsq_v2r8(dx20,dy20,dz20);
 577             rsq21            = gmx_fjsp_calc_rsq_v2r8(dx21,dy21,dz21);
 578             rsq22            = gmx_fjsp_calc_rsq_v2r8(dx22,dy22,dz22);
 579
 580             rinv00           = gmx_fjsp_invsqrt_v2r8(rsq00);
 581             rinv01           = gmx_fjsp_invsqrt_v2r8(rsq01);
 582             rinv02           = gmx_fjsp_invsqrt_v2r8(rsq02);
 583             rinv10           = gmx_fjsp_invsqrt_v2r8(rsq10);
 584             rinv11           = gmx_fjsp_invsqrt_v2r8(rsq11);
 585             rinv12           = gmx_fjsp_invsqrt_v2r8(rsq12);
 586             rinv20           = gmx_fjsp_invsqrt_v2r8(rsq20);
 587             rinv21           = gmx_fjsp_invsqrt_v2r8(rsq21);
 588             rinv22           = gmx_fjsp_invsqrt_v2r8(rsq22);
 589
 590             rinvsq00         = _fjsp_mul_v2r8(rinv00,rinv00);
 591             rinvsq01         = _fjsp_mul_v2r8(rinv01,rinv01);
 592             rinvsq02         = _fjsp_mul_v2r8(rinv02,rinv02);
 593             rinvsq10         = _fjsp_mul_v2r8(rinv10,rinv10);
 594             rinvsq11         = _fjsp_mul_v2r8(rinv11,rinv11);
 595             rinvsq12         = _fjsp_mul_v2r8(rinv12,rinv12);
 596             rinvsq20         = _fjsp_mul_v2r8(rinv20,rinv20);
 597             rinvsq21         = _fjsp_mul_v2r8(rinv21,rinv21);
 598             rinvsq22         = _fjsp_mul_v2r8(rinv22,rinv22);
 599
 600             fjx0             = _fjsp_setzero_v2r8();
 601             fjy0             = _fjsp_setzero_v2r8();
 602             fjz0             = _fjsp_setzero_v2r8();
 603             fjx1             = _fjsp_setzero_v2r8();
 604             fjy1             = _fjsp_setzero_v2r8();
 605             fjz1             = _fjsp_setzero_v2r8();
 606             fjx2             = _fjsp_setzero_v2r8();
 607             fjy2             = _fjsp_setzero_v2r8();
 608             fjz2             = _fjsp_setzero_v2r8();
 609
 610             /**************************
 611              * CALCULATE INTERACTIONS *
 612              **************************/
 613
 614             r00              = _fjsp_mul_v2r8(rsq00,rinv00);
 615
 616             /* Calculate table index by multiplying r with table scale and truncate to integer */
 617             rt               = _fjsp_mul_v2r8(r00,vftabscale);
 618             itab_tmp         = _fjsp_dtox_v2r8(rt);
 619             vfeps            = _fjsp_sub_v2r8(rt, _fjsp_xtod_v2r8(itab_tmp));
 620             twovfeps         = _fjsp_add_v2r8(vfeps,vfeps);
 621             _fjsp_store_v2r8(&vfconv.simd,itab_tmp);
 622
 623             vfconv.i[0]     *= 8;
 624             vfconv.i[1]     *= 8;
 625
 626             /* REACTION-FIELD ELECTROSTATICS */
 627             velec            = _fjsp_mul_v2r8(qq00,_fjsp_sub_v2r8(_fjsp_madd_v2r8(krf,rsq00,rinv00),crf));
 628             felec            = _fjsp_mul_v2r8(qq00,_fjsp_msub_v2r8(rinv00,rinvsq00,krf2));
 629
 630             /* CUBIC SPLINE TABLE DISPERSION */
 631             Y                = _fjsp_load_v2r8( vftab + vfconv.i[0] );
 632             F                = _fjsp_setzero_v2r8();
 633             GMX_FJSP_TRANSPOSE2_V2R8(Y,F);
 634             G                = _fjsp_load_v2r8( vftab + vfconv.i[0] + 2 );
 635             H                = _fjsp_setzero_v2r8();
 636             GMX_FJSP_TRANSPOSE2_V2R8(G,H);
 637             Fp               = _fjsp_madd_v2r8(vfeps,_fjsp_madd_v2r8(H,vfeps,G),F);
 638             VV               = _fjsp_madd_v2r8(vfeps,Fp,Y);
 639             vvdw6            = _fjsp_mul_v2r8(c6_00,VV);
 640             FF               = _fjsp_madd_v2r8(vfeps,_fjsp_madd_v2r8(twovfeps,H,G),Fp);
 641             fvdw6            = _fjsp_mul_v2r8(c6_00,FF);
 642
 643             /* CUBIC SPLINE TABLE REPULSION */
 644             Y                = _fjsp_load_v2r8( vftab + vfconv.i[0] + 4 );
 645             F                = _fjsp_setzero_v2r8();
 646             GMX_FJSP_TRANSPOSE2_V2R8(Y,F);
 647             G                = _fjsp_load_v2r8( vftab + vfconv.i[0] + 6 );
 648             H                = _fjsp_setzero_v2r8();
 649             GMX_FJSP_TRANSPOSE2_V2R8(G,H);
 650             Fp               = _fjsp_madd_v2r8(vfeps,_fjsp_madd_v2r8(H,vfeps,G),F);
 651             VV               = _fjsp_madd_v2r8(vfeps,Fp,Y);
 652             vvdw12           = _fjsp_mul_v2r8(c12_00,VV);
 653             FF               = _fjsp_madd_v2r8(vfeps,_fjsp_madd_v2r8(twovfeps,H,G),Fp);
 654             fvdw12           = _fjsp_mul_v2r8(c12_00,FF);
 655             vvdw             = _fjsp_add_v2r8(vvdw12,vvdw6);
 656             fvdw             = _fjsp_neg_v2r8(_fjsp_mul_v2r8(_fjsp_add_v2r8(fvdw6,fvdw12),_fjsp_mul_v2r8(vftabscale,rinv00)));
 657
 658             /* Update potential sum for this i atom from the interaction with this j atom. */
 659             velec            = _fjsp_unpacklo_v2r8(velec,_fjsp_setzero_v2r8());
 660             velecsum         = _fjsp_add_v2r8(velecsum,velec);
 661             vvdw             = _fjsp_unpacklo_v2r8(vvdw,_fjsp_setzero_v2r8());
 662             vvdwsum          = _fjsp_add_v2r8(vvdwsum,vvdw);
 663
 664             fscal            = _fjsp_add_v2r8(felec,fvdw);
 665
 666             fscal            = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
 667
 668             /* Update vectorial force */
 669             fix0             = _fjsp_madd_v2r8(dx00,fscal,fix0);
 670             fiy0             = _fjsp_madd_v2r8(dy00,fscal,fiy0);
 671             fiz0             = _fjsp_madd_v2r8(dz00,fscal,fiz0);
 672
 673             fjx0             = _fjsp_madd_v2r8(dx00,fscal,fjx0);
 674             fjy0             = _fjsp_madd_v2r8(dy00,fscal,fjy0);
 675             fjz0             = _fjsp_madd_v2r8(dz00,fscal,fjz0);
 676
 677             /**************************
 678              * CALCULATE INTERACTIONS *
 679              **************************/
 680
 681             /* REACTION-FIELD ELECTROSTATICS */
 682             velec            = _fjsp_mul_v2r8(qq01,_fjsp_sub_v2r8(_fjsp_madd_v2r8(krf,rsq01,rinv01),crf));
 683             felec            = _fjsp_mul_v2r8(qq01,_fjsp_msub_v2r8(rinv01,rinvsq01,krf2));
 684
 685             /* Update potential sum for this i atom from the interaction with this j atom. */
 686             velec            = _fjsp_unpacklo_v2r8(velec,_fjsp_setzero_v2r8());
 687             velecsum         = _fjsp_add_v2r8(velecsum,velec);
 688
 689             fscal            = felec;
 690
 691             fscal            = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
 692
 693             /* Update vectorial force */
 694             fix0             = _fjsp_madd_v2r8(dx01,fscal,fix0);
 695             fiy0             = _fjsp_madd_v2r8(dy01,fscal,fiy0);
 696             fiz0             = _fjsp_madd_v2r8(dz01,fscal,fiz0);
 697
 698             fjx1             = _fjsp_madd_v2r8(dx01,fscal,fjx1);
 699             fjy1             = _fjsp_madd_v2r8(dy01,fscal,fjy1);
 700             fjz1             = _fjsp_madd_v2r8(dz01,fscal,fjz1);
 701
 702             /**************************
 703              * CALCULATE INTERACTIONS *
 704              **************************/
 705
 706             /* REACTION-FIELD ELECTROSTATICS */
 707             velec            = _fjsp_mul_v2r8(qq02,_fjsp_sub_v2r8(_fjsp_madd_v2r8(krf,rsq02,rinv02),crf));
 708             felec            = _fjsp_mul_v2r8(qq02,_fjsp_msub_v2r8(rinv02,rinvsq02,krf2));
 709
 710             /* Update potential sum for this i atom from the interaction with this j atom. */
 711             velec            = _fjsp_unpacklo_v2r8(velec,_fjsp_setzero_v2r8());
 712             velecsum         = _fjsp_add_v2r8(velecsum,velec);
 713
 714             fscal            = felec;
 715
 716             fscal            = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
 717
 718             /* Update vectorial force */
 719             fix0             = _fjsp_madd_v2r8(dx02,fscal,fix0);
 720             fiy0             = _fjsp_madd_v2r8(dy02,fscal,fiy0);
 721             fiz0             = _fjsp_madd_v2r8(dz02,fscal,fiz0);
 722
 723             fjx2             = _fjsp_madd_v2r8(dx02,fscal,fjx2);
 724             fjy2             = _fjsp_madd_v2r8(dy02,fscal,fjy2);
 725             fjz2             = _fjsp_madd_v2r8(dz02,fscal,fjz2);
 726
 727             /**************************
 728              * CALCULATE INTERACTIONS *
 729              **************************/
 730
 731             /* REACTION-FIELD ELECTROSTATICS */
 732             velec            = _fjsp_mul_v2r8(qq10,_fjsp_sub_v2r8(_fjsp_madd_v2r8(krf,rsq10,rinv10),crf));
 733             felec            = _fjsp_mul_v2r8(qq10,_fjsp_msub_v2r8(rinv10,rinvsq10,krf2));
 734
 735             /* Update potential sum for this i atom from the interaction with this j atom. */
 736             velec            = _fjsp_unpacklo_v2r8(velec,_fjsp_setzero_v2r8());
 737             velecsum         = _fjsp_add_v2r8(velecsum,velec);
 738
 739             fscal            = felec;
 740
 741             fscal            = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
 742
 743             /* Update vectorial force */
 744             fix1             = _fjsp_madd_v2r8(dx10,fscal,fix1);
 745             fiy1             = _fjsp_madd_v2r8(dy10,fscal,fiy1);
 746             fiz1             = _fjsp_madd_v2r8(dz10,fscal,fiz1);
 747
 748             fjx0             = _fjsp_madd_v2r8(dx10,fscal,fjx0);
 749             fjy0             = _fjsp_madd_v2r8(dy10,fscal,fjy0);
 750             fjz0             = _fjsp_madd_v2r8(dz10,fscal,fjz0);
 751
 752             /**************************
 753              * CALCULATE INTERACTIONS *
 754              **************************/
 755
 756             /* REACTION-FIELD ELECTROSTATICS */
 757             velec            = _fjsp_mul_v2r8(qq11,_fjsp_sub_v2r8(_fjsp_madd_v2r8(krf,rsq11,rinv11),crf));
 758             felec            = _fjsp_mul_v2r8(qq11,_fjsp_msub_v2r8(rinv11,rinvsq11,krf2));
 759
 760             /* Update potential sum for this i atom from the interaction with this j atom. */
 761             velec            = _fjsp_unpacklo_v2r8(velec,_fjsp_setzero_v2r8());
 762             velecsum         = _fjsp_add_v2r8(velecsum,velec);
 763
 764             fscal            = felec;
 765
 766             fscal            = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
 767
 768             /* Update vectorial force */
 769             fix1             = _fjsp_madd_v2r8(dx11,fscal,fix1);
 770             fiy1             = _fjsp_madd_v2r8(dy11,fscal,fiy1);
 771             fiz1             = _fjsp_madd_v2r8(dz11,fscal,fiz1);
 772
 773             fjx1             = _fjsp_madd_v2r8(dx11,fscal,fjx1);
 774             fjy1             = _fjsp_madd_v2r8(dy11,fscal,fjy1);
 775             fjz1             = _fjsp_madd_v2r8(dz11,fscal,fjz1);
 776
 777             /**************************
 778              * CALCULATE INTERACTIONS *
 779              **************************/
 780
 781             /* REACTION-FIELD ELECTROSTATICS */
 782             velec            = _fjsp_mul_v2r8(qq12,_fjsp_sub_v2r8(_fjsp_madd_v2r8(krf,rsq12,rinv12),crf));
 783             felec            = _fjsp_mul_v2r8(qq12,_fjsp_msub_v2r8(rinv12,rinvsq12,krf2));
 784
 785             /* Update potential sum for this i atom from the interaction with this j atom. */
 786             velec            = _fjsp_unpacklo_v2r8(velec,_fjsp_setzero_v2r8());
 787             velecsum         = _fjsp_add_v2r8(velecsum,velec);
 788
 789             fscal            = felec;
 790
 791             fscal            = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
 792
 793             /* Update vectorial force */
 794             fix1             = _fjsp_madd_v2r8(dx12,fscal,fix1);
 795             fiy1             = _fjsp_madd_v2r8(dy12,fscal,fiy1);
 796             fiz1             = _fjsp_madd_v2r8(dz12,fscal,fiz1);
 797
 798             fjx2             = _fjsp_madd_v2r8(dx12,fscal,fjx2);
 799             fjy2             = _fjsp_madd_v2r8(dy12,fscal,fjy2);
 800             fjz2             = _fjsp_madd_v2r8(dz12,fscal,fjz2);
 801
 802             /**************************
 803              * CALCULATE INTERACTIONS *
 804              **************************/
 805
 806             /* REACTION-FIELD ELECTROSTATICS */
 807             velec            = _fjsp_mul_v2r8(qq20,_fjsp_sub_v2r8(_fjsp_madd_v2r8(krf,rsq20,rinv20),crf));
 808             felec            = _fjsp_mul_v2r8(qq20,_fjsp_msub_v2r8(rinv20,rinvsq20,krf2));
 809
 810             /* Update potential sum for this i atom from the interaction with this j atom. */
 811             velec            = _fjsp_unpacklo_v2r8(velec,_fjsp_setzero_v2r8());
 812             velecsum         = _fjsp_add_v2r8(velecsum,velec);
 813
 814             fscal            = felec;
 815
 816             fscal            = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
 817
 818             /* Update vectorial force */
 819             fix2             = _fjsp_madd_v2r8(dx20,fscal,fix2);
 820             fiy2             = _fjsp_madd_v2r8(dy20,fscal,fiy2);
 821             fiz2             = _fjsp_madd_v2r8(dz20,fscal,fiz2);
 822
 823             fjx0             = _fjsp_madd_v2r8(dx20,fscal,fjx0);
 824             fjy0             = _fjsp_madd_v2r8(dy20,fscal,fjy0);
 825             fjz0             = _fjsp_madd_v2r8(dz20,fscal,fjz0);
 826
 827             /**************************
 828              * CALCULATE INTERACTIONS *
 829              **************************/
 830
 831             /* REACTION-FIELD ELECTROSTATICS */
 832             velec            = _fjsp_mul_v2r8(qq21,_fjsp_sub_v2r8(_fjsp_madd_v2r8(krf,rsq21,rinv21),crf));
 833             felec            = _fjsp_mul_v2r8(qq21,_fjsp_msub_v2r8(rinv21,rinvsq21,krf2));
 834
 835             /* Update potential sum for this i atom from the interaction with this j atom. */
 836             velec            = _fjsp_unpacklo_v2r8(velec,_fjsp_setzero_v2r8());
 837             velecsum         = _fjsp_add_v2r8(velecsum,velec);
 838
 839             fscal            = felec;
 840
 841             fscal            = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
 842
 843             /* Update vectorial force */
 844             fix2             = _fjsp_madd_v2r8(dx21,fscal,fix2);
 845             fiy2             = _fjsp_madd_v2r8(dy21,fscal,fiy2);
 846             fiz2             = _fjsp_madd_v2r8(dz21,fscal,fiz2);
 847
 848             fjx1             = _fjsp_madd_v2r8(dx21,fscal,fjx1);
 849             fjy1             = _fjsp_madd_v2r8(dy21,fscal,fjy1);
 850             fjz1             = _fjsp_madd_v2r8(dz21,fscal,fjz1);
 851
 852             /**************************
 853              * CALCULATE INTERACTIONS *
 854              **************************/
 855
 856             /* REACTION-FIELD ELECTROSTATICS */
 857             velec            = _fjsp_mul_v2r8(qq22,_fjsp_sub_v2r8(_fjsp_madd_v2r8(krf,rsq22,rinv22),crf));
 858             felec            = _fjsp_mul_v2r8(qq22,_fjsp_msub_v2r8(rinv22,rinvsq22,krf2));
 859
 860             /* Update potential sum for this i atom from the interaction with this j atom. */
 861             velec            = _fjsp_unpacklo_v2r8(velec,_fjsp_setzero_v2r8());
 862             velecsum         = _fjsp_add_v2r8(velecsum,velec);
 863
 864             fscal            = felec;
 865
 866             fscal            = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
 867
 868             /* Update vectorial force */
 869             fix2             = _fjsp_madd_v2r8(dx22,fscal,fix2);
 870             fiy2             = _fjsp_madd_v2r8(dy22,fscal,fiy2);
 871             fiz2             = _fjsp_madd_v2r8(dz22,fscal,fiz2);
 872
 873             fjx2             = _fjsp_madd_v2r8(dx22,fscal,fjx2);
 874             fjy2             = _fjsp_madd_v2r8(dy22,fscal,fjy2);
 875             fjz2             = _fjsp_madd_v2r8(dz22,fscal,fjz2);
 876
 877             gmx_fjsp_decrement_3rvec_1ptr_swizzle_v2r8(f+j_coord_offsetA,fjx0,fjy0,fjz0,fjx1,fjy1,fjz1,fjx2,fjy2,fjz2);
 878
 879             /* Inner loop uses 350 flops */
 880         }
 881
 882         /* End of innermost loop */
 883
 884         gmx_fjsp_update_iforce_3atom_swizzle_v2r8(fix0,fiy0,fiz0,fix1,fiy1,fiz1,fix2,fiy2,fiz2,
 885                                               f+i_coord_offset,fshift+i_shift_offset);
 886
 887         ggid                        = gid[iidx];
 888         /* Update potential energies */
 889         gmx_fjsp_update_1pot_v2r8(velecsum,kernel_data->energygrp_elec+ggid);
 890         gmx_fjsp_update_1pot_v2r8(vvdwsum,kernel_data->energygrp_vdw+ggid);
 891
 892         /* Increment number of inner iterations */
 893         inneriter                  += j_index_end - j_index_start;
 894
 895         /* Outer loop uses 20 flops */
 896     }
 897
 898     /* Increment number of outer iterations */
 899     outeriter        += nri;
 900
 901     /* Update outer/inner flops */
 902
 903     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W3W3_VF,outeriter*20 + inneriter*350);
 904 }
 905 /*
 906  * Gromacs nonbonded kernel:   nb_kernel_ElecRF_VdwCSTab_GeomW3W3_F_sparc64_hpc_ace_double
 907  * Electrostatics interaction: ReactionField
 908  * VdW interaction:            CubicSplineTable
 909  * Geometry:                   Water3-Water3
 910  * Calculate force/pot:        Force
 911  */
 912 void
 913 nb_kernel_ElecRF_VdwCSTab_GeomW3W3_F_sparc64_hpc_ace_double
 914                     (t_nblist                    * gmx_restrict       nlist,
 915                      rvec                        * gmx_restrict          xx,
 916                      rvec                        * gmx_restrict          ff,
 917                      t_forcerec                  * gmx_restrict          fr,
 918                      t_mdatoms                   * gmx_restrict     mdatoms,
 919                      nb_kernel_data_t gmx_unused * gmx_restrict kernel_data,
 920                      t_nrnb                      * gmx_restrict        nrnb)
 921 {
 922     /* Suffixes 0,1,2,3 refer to particle indices for waters in the inner or outer loop, or
 923      * just 0 for non-waters.
 924      * Suffixes A,B refer to j loop unrolling done with double precision SIMD, e.g. for the two different
 925      * jnr indices corresponding to data put in the four positions in the SIMD register.
 926      */
 927     int              i_shift_offset,i_coord_offset,outeriter,inneriter;
 928     int              j_index_start,j_index_end,jidx,nri,inr,ggid,iidx;
 929     int              jnrA,jnrB;
 930     int              j_coord_offsetA,j_coord_offsetB;
 931     int              *iinr,*jindex,*jjnr,*shiftidx,*gid;
 932     real             rcutoff_scalar;
 933     real             *shiftvec,*fshift,*x,*f;
 934     _fjsp_v2r8       tx,ty,tz,fscal,rcutoff,rcutoff2,jidxall;
 935     int              vdwioffset0;
 936     _fjsp_v2r8       ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
 937     int              vdwioffset1;
 938     _fjsp_v2r8       ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
 939     int              vdwioffset2;
 940     _fjsp_v2r8       ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
 941     int              vdwjidx0A,vdwjidx0B;
 942     _fjsp_v2r8       jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
 943     int              vdwjidx1A,vdwjidx1B;
 944     _fjsp_v2r8       jx1,jy1,jz1,fjx1,fjy1,fjz1,jq1,isaj1;
 945     int              vdwjidx2A,vdwjidx2B;
 946     _fjsp_v2r8       jx2,jy2,jz2,fjx2,fjy2,fjz2,jq2,isaj2;
 947     _fjsp_v2r8       dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00;
 948     _fjsp_v2r8       dx01,dy01,dz01,rsq01,rinv01,rinvsq01,r01,qq01,c6_01,c12_01;
 949     _fjsp_v2r8       dx02,dy02,dz02,rsq02,rinv02,rinvsq02,r02,qq02,c6_02,c12_02;
 950     _fjsp_v2r8       dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10;
 951     _fjsp_v2r8       dx11,dy11,dz11,rsq11,rinv11,rinvsq11,r11,qq11,c6_11,c12_11;
 952     _fjsp_v2r8       dx12,dy12,dz12,rsq12,rinv12,rinvsq12,r12,qq12,c6_12,c12_12;
 953     _fjsp_v2r8       dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20;
 954     _fjsp_v2r8       dx21,dy21,dz21,rsq21,rinv21,rinvsq21,r21,qq21,c6_21,c12_21;
 955     _fjsp_v2r8       dx22,dy22,dz22,rsq22,rinv22,rinvsq22,r22,qq22,c6_22,c12_22;
 956     _fjsp_v2r8       velec,felec,velecsum,facel,crf,krf,krf2;
 957     real             *charge;
 958     int              nvdwtype;
 959     _fjsp_v2r8       rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,sh_vdw_invrcut6;
 960     int              *vdwtype;
 961     real             *vdwparam;
 962     _fjsp_v2r8       one_sixth   = gmx_fjsp_set1_v2r8(1.0/6.0);
 963     _fjsp_v2r8       one_twelfth = gmx_fjsp_set1_v2r8(1.0/12.0);
 964     _fjsp_v2r8       rt,vfeps,vftabscale,Y,F,G,H,Heps,Fp,VV,FF,twovfeps;
 965     real             *vftab;
 966     _fjsp_v2r8       itab_tmp;
 967     _fjsp_v2r8       dummy_mask,cutoff_mask;
 968     _fjsp_v2r8       one     = gmx_fjsp_set1_v2r8(1.0);
 969     _fjsp_v2r8       two     = gmx_fjsp_set1_v2r8(2.0);
 970     union { _fjsp_v2r8 simd; long long int i[2]; } vfconv,gbconv,ewconv;
 971
 972     x                = xx[0];
 973     f                = ff[0];
 974
 975     nri              = nlist->nri;
 976     iinr             = nlist->iinr;
 977     jindex           = nlist->jindex;
 978     jjnr             = nlist->jjnr;
 979     shiftidx         = nlist->shift;
 980     gid              = nlist->gid;
 981     shiftvec         = fr->shift_vec[0];
 982     fshift           = fr->fshift[0];
 983     facel            = gmx_fjsp_set1_v2r8(fr->epsfac);
 984     charge           = mdatoms->chargeA;
 985     krf              = gmx_fjsp_set1_v2r8(fr->ic->k_rf);
 986     krf2             = gmx_fjsp_set1_v2r8(fr->ic->k_rf*2.0);
 987     crf              = gmx_fjsp_set1_v2r8(fr->ic->c_rf);
 988     nvdwtype         = fr->ntype;
 989     vdwparam         = fr->nbfp;
 990     vdwtype          = mdatoms->typeA;
 991
 992     vftab            = kernel_data->table_vdw->data;
 993     vftabscale       = gmx_fjsp_set1_v2r8(kernel_data->table_vdw->scale);
 994
 995     /* Setup water-specific parameters */
 996     inr              = nlist->iinr[0];
 997     iq0              = _fjsp_mul_v2r8(facel,gmx_fjsp_set1_v2r8(charge[inr+0]));
 998     iq1              = _fjsp_mul_v2r8(facel,gmx_fjsp_set1_v2r8(charge[inr+1]));
 999     iq2              = _fjsp_mul_v2r8(facel,gmx_fjsp_set1_v2r8(charge[inr+2]));
1000     vdwioffset0      = 2*nvdwtype*vdwtype[inr+0];
1001
1002     jq0              = gmx_fjsp_set1_v2r8(charge[inr+0]);
1003     jq1              = gmx_fjsp_set1_v2r8(charge[inr+1]);
1004     jq2              = gmx_fjsp_set1_v2r8(charge[inr+2]);
1005     vdwjidx0A        = 2*vdwtype[inr+0];
1006     qq00             = _fjsp_mul_v2r8(iq0,jq0);
1007     c6_00            = gmx_fjsp_set1_v2r8(vdwparam[vdwioffset0+vdwjidx0A]);
1008     c12_00           = gmx_fjsp_set1_v2r8(vdwparam[vdwioffset0+vdwjidx0A+1]);
1009     qq01             = _fjsp_mul_v2r8(iq0,jq1);
1010     qq02             = _fjsp_mul_v2r8(iq0,jq2);
1011     qq10             = _fjsp_mul_v2r8(iq1,jq0);
1012     qq11             = _fjsp_mul_v2r8(iq1,jq1);
1013     qq12             = _fjsp_mul_v2r8(iq1,jq2);
1014     qq20             = _fjsp_mul_v2r8(iq2,jq0);
1015     qq21             = _fjsp_mul_v2r8(iq2,jq1);
1016     qq22             = _fjsp_mul_v2r8(iq2,jq2);
1017
1018     /* Avoid stupid compiler warnings */
1019     jnrA = jnrB = 0;
1020     j_coord_offsetA = 0;
1021     j_coord_offsetB = 0;
1022
1023     outeriter        = 0;
1024     inneriter        = 0;
1025
1026     /* Start outer loop over neighborlists */
1027     for(iidx=0; iidx<nri; iidx++)
1028     {
1029         /* Load shift vector for this list */
1030         i_shift_offset   = DIM*shiftidx[iidx];
1031
1032         /* Load limits for loop over neighbors */
1033         j_index_start    = jindex[iidx];
1034         j_index_end      = jindex[iidx+1];
1035
1036         /* Get outer coordinate index */
1037         inr              = iinr[iidx];
1038         i_coord_offset   = DIM*inr;
1039
1040         /* Load i particle coords and add shift vector */
1041         gmx_fjsp_load_shift_and_3rvec_broadcast_v2r8(shiftvec+i_shift_offset,x+i_coord_offset,
1042                                                  &ix0,&iy0,&iz0,&ix1,&iy1,&iz1,&ix2,&iy2,&iz2);
1043
1044         fix0             = _fjsp_setzero_v2r8();
1045         fiy0             = _fjsp_setzero_v2r8();
1046         fiz0             = _fjsp_setzero_v2r8();
1047         fix1             = _fjsp_setzero_v2r8();
1048         fiy1             = _fjsp_setzero_v2r8();
1049         fiz1             = _fjsp_setzero_v2r8();
1050         fix2             = _fjsp_setzero_v2r8();
1051         fiy2             = _fjsp_setzero_v2r8();
1052         fiz2             = _fjsp_setzero_v2r8();
1053
1054         /* Start inner kernel loop */
1055         for(jidx=j_index_start; jidx<j_index_end-1; jidx+=2)
1056         {
1057
1058             /* Get j neighbor index, and coordinate index */
1059             jnrA             = jjnr[jidx];
1060             jnrB             = jjnr[jidx+1];
1061             j_coord_offsetA  = DIM*jnrA;
1062             j_coord_offsetB  = DIM*jnrB;
1063
1064             /* load j atom coordinates */
1065             gmx_fjsp_load_3rvec_2ptr_swizzle_v2r8(x+j_coord_offsetA,x+j_coord_offsetB,
1066                                               &jx0,&jy0,&jz0,&jx1,&jy1,&jz1,&jx2,&jy2,&jz2);
1067
1068             /* Calculate displacement vector */
1069             dx00             = _fjsp_sub_v2r8(ix0,jx0);
1070             dy00             = _fjsp_sub_v2r8(iy0,jy0);
1071             dz00             = _fjsp_sub_v2r8(iz0,jz0);
1072             dx01             = _fjsp_sub_v2r8(ix0,jx1);
1073             dy01             = _fjsp_sub_v2r8(iy0,jy1);
1074             dz01             = _fjsp_sub_v2r8(iz0,jz1);
1075             dx02             = _fjsp_sub_v2r8(ix0,jx2);
1076             dy02             = _fjsp_sub_v2r8(iy0,jy2);
1077             dz02             = _fjsp_sub_v2r8(iz0,jz2);
1078             dx10             = _fjsp_sub_v2r8(ix1,jx0);
1079             dy10             = _fjsp_sub_v2r8(iy1,jy0);
1080             dz10             = _fjsp_sub_v2r8(iz1,jz0);
1081             dx11             = _fjsp_sub_v2r8(ix1,jx1);
1082             dy11             = _fjsp_sub_v2r8(iy1,jy1);
1083             dz11             = _fjsp_sub_v2r8(iz1,jz1);
1084             dx12             = _fjsp_sub_v2r8(ix1,jx2);
1085             dy12             = _fjsp_sub_v2r8(iy1,jy2);
1086             dz12             = _fjsp_sub_v2r8(iz1,jz2);
1087             dx20             = _fjsp_sub_v2r8(ix2,jx0);
1088             dy20             = _fjsp_sub_v2r8(iy2,jy0);
1089             dz20             = _fjsp_sub_v2r8(iz2,jz0);
1090             dx21             = _fjsp_sub_v2r8(ix2,jx1);
1091             dy21             = _fjsp_sub_v2r8(iy2,jy1);
1092             dz21             = _fjsp_sub_v2r8(iz2,jz1);
1093             dx22             = _fjsp_sub_v2r8(ix2,jx2);
1094             dy22             = _fjsp_sub_v2r8(iy2,jy2);
1095             dz22             = _fjsp_sub_v2r8(iz2,jz2);
1096
1097             /* Calculate squared distance and things based on it */
1098             rsq00            = gmx_fjsp_calc_rsq_v2r8(dx00,dy00,dz00);
1099             rsq01            = gmx_fjsp_calc_rsq_v2r8(dx01,dy01,dz01);
1100             rsq02            = gmx_fjsp_calc_rsq_v2r8(dx02,dy02,dz02);
1101             rsq10            = gmx_fjsp_calc_rsq_v2r8(dx10,dy10,dz10);
1102             rsq11            = gmx_fjsp_calc_rsq_v2r8(dx11,dy11,dz11);
1103             rsq12            = gmx_fjsp_calc_rsq_v2r8(dx12,dy12,dz12);
1104             rsq20            = gmx_fjsp_calc_rsq_v2r8(dx20,dy20,dz20);
1105             rsq21            = gmx_fjsp_calc_rsq_v2r8(dx21,dy21,dz21);
1106             rsq22            = gmx_fjsp_calc_rsq_v2r8(dx22,dy22,dz22);
1107
1108             rinv00           = gmx_fjsp_invsqrt_v2r8(rsq00);
1109             rinv01           = gmx_fjsp_invsqrt_v2r8(rsq01);
1110             rinv02           = gmx_fjsp_invsqrt_v2r8(rsq02);
1111             rinv10           = gmx_fjsp_invsqrt_v2r8(rsq10);
1112             rinv11           = gmx_fjsp_invsqrt_v2r8(rsq11);
1113             rinv12           = gmx_fjsp_invsqrt_v2r8(rsq12);
1114             rinv20           = gmx_fjsp_invsqrt_v2r8(rsq20);
1115             rinv21           = gmx_fjsp_invsqrt_v2r8(rsq21);
1116             rinv22           = gmx_fjsp_invsqrt_v2r8(rsq22);
1117
1118             rinvsq00         = _fjsp_mul_v2r8(rinv00,rinv00);
1119             rinvsq01         = _fjsp_mul_v2r8(rinv01,rinv01);
1120             rinvsq02         = _fjsp_mul_v2r8(rinv02,rinv02);
1121             rinvsq10         = _fjsp_mul_v2r8(rinv10,rinv10);
1122             rinvsq11         = _fjsp_mul_v2r8(rinv11,rinv11);
1123             rinvsq12         = _fjsp_mul_v2r8(rinv12,rinv12);
1124             rinvsq20         = _fjsp_mul_v2r8(rinv20,rinv20);
1125             rinvsq21         = _fjsp_mul_v2r8(rinv21,rinv21);
1126             rinvsq22         = _fjsp_mul_v2r8(rinv22,rinv22);
1127
1128             fjx0             = _fjsp_setzero_v2r8();
1129             fjy0             = _fjsp_setzero_v2r8();
1130             fjz0             = _fjsp_setzero_v2r8();
1131             fjx1             = _fjsp_setzero_v2r8();
1132             fjy1             = _fjsp_setzero_v2r8();
1133             fjz1             = _fjsp_setzero_v2r8();
1134             fjx2             = _fjsp_setzero_v2r8();
1135             fjy2             = _fjsp_setzero_v2r8();
1136             fjz2             = _fjsp_setzero_v2r8();
1137
1138             /**************************
1139              * CALCULATE INTERACTIONS *
1140              **************************/
1141
1142             r00              = _fjsp_mul_v2r8(rsq00,rinv00);
1143
1144             /* Calculate table index by multiplying r with table scale and truncate to integer */
1145             rt               = _fjsp_mul_v2r8(r00,vftabscale);
1146             itab_tmp         = _fjsp_dtox_v2r8(rt);
1147             vfeps            = _fjsp_sub_v2r8(rt, _fjsp_xtod_v2r8(itab_tmp));
1148             twovfeps         = _fjsp_add_v2r8(vfeps,vfeps);
1149             _fjsp_store_v2r8(&vfconv.simd,itab_tmp);
1150
1151             vfconv.i[0]     *= 8;
1152             vfconv.i[1]     *= 8;
1153
1154             /* REACTION-FIELD ELECTROSTATICS */
1155             felec            = _fjsp_mul_v2r8(qq00,_fjsp_msub_v2r8(rinv00,rinvsq00,krf2));
1156
1157             /* CUBIC SPLINE TABLE DISPERSION */
1158             Y                = _fjsp_load_v2r8( vftab + vfconv.i[0] );
1159             F                = _fjsp_load_v2r8( vftab + vfconv.i[1] );
1160             GMX_FJSP_TRANSPOSE2_V2R8(Y,F);
1161             G                = _fjsp_load_v2r8( vftab + vfconv.i[0] + 2 );
1162             H                = _fjsp_load_v2r8( vftab + vfconv.i[1] + 2 );
1163             GMX_FJSP_TRANSPOSE2_V2R8(G,H);
1164             Fp               = _fjsp_madd_v2r8(vfeps,_fjsp_madd_v2r8(H,vfeps,G),F);
1165             FF               = _fjsp_madd_v2r8(vfeps,_fjsp_madd_v2r8(twovfeps,H,G),Fp);
1166             fvdw6            = _fjsp_mul_v2r8(c6_00,FF);
1167
1168             /* CUBIC SPLINE TABLE REPULSION */
1169             Y                = _fjsp_load_v2r8( vftab + vfconv.i[0] + 4 );
1170             F                = _fjsp_load_v2r8( vftab + vfconv.i[1] + 4 );
1171             GMX_FJSP_TRANSPOSE2_V2R8(Y,F);
1172             G                = _fjsp_load_v2r8( vftab + vfconv.i[0] + 6 );
1173             H                = _fjsp_load_v2r8( vftab + vfconv.i[1] + 6 );
1174             GMX_FJSP_TRANSPOSE2_V2R8(G,H);
1175             Fp               = _fjsp_madd_v2r8(vfeps,_fjsp_madd_v2r8(H,vfeps,G),F);
1176             FF               = _fjsp_madd_v2r8(vfeps,_fjsp_madd_v2r8(twovfeps,H,G),Fp);
1177             fvdw12           = _fjsp_mul_v2r8(c12_00,FF);
1178             fvdw             = _fjsp_neg_v2r8(_fjsp_mul_v2r8(_fjsp_add_v2r8(fvdw6,fvdw12),_fjsp_mul_v2r8(vftabscale,rinv00)));
1179
1180             fscal            = _fjsp_add_v2r8(felec,fvdw);
1181
1182             /* Update vectorial force */
1183             fix0             = _fjsp_madd_v2r8(dx00,fscal,fix0);
1184             fiy0             = _fjsp_madd_v2r8(dy00,fscal,fiy0);
1185             fiz0             = _fjsp_madd_v2r8(dz00,fscal,fiz0);
1186
1187             fjx0             = _fjsp_madd_v2r8(dx00,fscal,fjx0);
1188             fjy0             = _fjsp_madd_v2r8(dy00,fscal,fjy0);
1189             fjz0             = _fjsp_madd_v2r8(dz00,fscal,fjz0);
1190
1191             /**************************
1192              * CALCULATE INTERACTIONS *
1193              **************************/
1194
1195             /* REACTION-FIELD ELECTROSTATICS */
1196             felec            = _fjsp_mul_v2r8(qq01,_fjsp_msub_v2r8(rinv01,rinvsq01,krf2));
1197
1198             fscal            = felec;
1199
1200             /* Update vectorial force */
1201             fix0             = _fjsp_madd_v2r8(dx01,fscal,fix0);
1202             fiy0             = _fjsp_madd_v2r8(dy01,fscal,fiy0);
1203             fiz0             = _fjsp_madd_v2r8(dz01,fscal,fiz0);
1204
1205             fjx1             = _fjsp_madd_v2r8(dx01,fscal,fjx1);
1206             fjy1             = _fjsp_madd_v2r8(dy01,fscal,fjy1);
1207             fjz1             = _fjsp_madd_v2r8(dz01,fscal,fjz1);
1208
1209             /**************************
1210              * CALCULATE INTERACTIONS *
1211              **************************/
1212
1213             /* REACTION-FIELD ELECTROSTATICS */
1214             felec            = _fjsp_mul_v2r8(qq02,_fjsp_msub_v2r8(rinv02,rinvsq02,krf2));
1215
1216             fscal            = felec;
1217
1218             /* Update vectorial force */
1219             fix0             = _fjsp_madd_v2r8(dx02,fscal,fix0);
1220             fiy0             = _fjsp_madd_v2r8(dy02,fscal,fiy0);
1221             fiz0             = _fjsp_madd_v2r8(dz02,fscal,fiz0);
1222
1223             fjx2             = _fjsp_madd_v2r8(dx02,fscal,fjx2);
1224             fjy2             = _fjsp_madd_v2r8(dy02,fscal,fjy2);
1225             fjz2             = _fjsp_madd_v2r8(dz02,fscal,fjz2);
1226
1227             /**************************
1228              * CALCULATE INTERACTIONS *
1229              **************************/
1230
1231             /* REACTION-FIELD ELECTROSTATICS */
1232             felec            = _fjsp_mul_v2r8(qq10,_fjsp_msub_v2r8(rinv10,rinvsq10,krf2));
1233
1234             fscal            = felec;
1235
1236             /* Update vectorial force */
1237             fix1             = _fjsp_madd_v2r8(dx10,fscal,fix1);
1238             fiy1             = _fjsp_madd_v2r8(dy10,fscal,fiy1);
1239             fiz1             = _fjsp_madd_v2r8(dz10,fscal,fiz1);
1240
1241             fjx0             = _fjsp_madd_v2r8(dx10,fscal,fjx0);
1242             fjy0             = _fjsp_madd_v2r8(dy10,fscal,fjy0);
1243             fjz0             = _fjsp_madd_v2r8(dz10,fscal,fjz0);
1244
1245             /**************************
1246              * CALCULATE INTERACTIONS *
1247              **************************/
1248
1249             /* REACTION-FIELD ELECTROSTATICS */
1250             felec            = _fjsp_mul_v2r8(qq11,_fjsp_msub_v2r8(rinv11,rinvsq11,krf2));
1251
1252             fscal            = felec;
1253
1254             /* Update vectorial force */
1255             fix1             = _fjsp_madd_v2r8(dx11,fscal,fix1);
1256             fiy1             = _fjsp_madd_v2r8(dy11,fscal,fiy1);
1257             fiz1             = _fjsp_madd_v2r8(dz11,fscal,fiz1);
1258
1259             fjx1             = _fjsp_madd_v2r8(dx11,fscal,fjx1);
1260             fjy1             = _fjsp_madd_v2r8(dy11,fscal,fjy1);
1261             fjz1             = _fjsp_madd_v2r8(dz11,fscal,fjz1);
1262
1263             /**************************
1264              * CALCULATE INTERACTIONS *
1265              **************************/
1266
1267             /* REACTION-FIELD ELECTROSTATICS */
1268             felec            = _fjsp_mul_v2r8(qq12,_fjsp_msub_v2r8(rinv12,rinvsq12,krf2));
1269
1270             fscal            = felec;
1271
1272             /* Update vectorial force */
1273             fix1             = _fjsp_madd_v2r8(dx12,fscal,fix1);
1274             fiy1             = _fjsp_madd_v2r8(dy12,fscal,fiy1);
1275             fiz1             = _fjsp_madd_v2r8(dz12,fscal,fiz1);
1276
1277             fjx2             = _fjsp_madd_v2r8(dx12,fscal,fjx2);
1278             fjy2             = _fjsp_madd_v2r8(dy12,fscal,fjy2);
1279             fjz2             = _fjsp_madd_v2r8(dz12,fscal,fjz2);
1280
1281             /**************************
1282              * CALCULATE INTERACTIONS *
1283              **************************/
1284
1285             /* REACTION-FIELD ELECTROSTATICS */
1286             felec            = _fjsp_mul_v2r8(qq20,_fjsp_msub_v2r8(rinv20,rinvsq20,krf2));
1287
1288             fscal            = felec;
1289
1290             /* Update vectorial force */
1291             fix2             = _fjsp_madd_v2r8(dx20,fscal,fix2);
1292             fiy2             = _fjsp_madd_v2r8(dy20,fscal,fiy2);
1293             fiz2             = _fjsp_madd_v2r8(dz20,fscal,fiz2);
1294
1295             fjx0             = _fjsp_madd_v2r8(dx20,fscal,fjx0);
1296             fjy0             = _fjsp_madd_v2r8(dy20,fscal,fjy0);
1297             fjz0             = _fjsp_madd_v2r8(dz20,fscal,fjz0);
1298
1299             /**************************
1300              * CALCULATE INTERACTIONS *
1301              **************************/
1302
1303             /* REACTION-FIELD ELECTROSTATICS */
1304             felec            = _fjsp_mul_v2r8(qq21,_fjsp_msub_v2r8(rinv21,rinvsq21,krf2));
1305
1306             fscal            = felec;
1307
1308             /* Update vectorial force */
1309             fix2             = _fjsp_madd_v2r8(dx21,fscal,fix2);
1310             fiy2             = _fjsp_madd_v2r8(dy21,fscal,fiy2);
1311             fiz2             = _fjsp_madd_v2r8(dz21,fscal,fiz2);
1312
1313             fjx1             = _fjsp_madd_v2r8(dx21,fscal,fjx1);
1314             fjy1             = _fjsp_madd_v2r8(dy21,fscal,fjy1);
1315             fjz1             = _fjsp_madd_v2r8(dz21,fscal,fjz1);
1316
1317             /**************************
1318              * CALCULATE INTERACTIONS *
1319              **************************/
1320
1321             /* REACTION-FIELD ELECTROSTATICS */
1322             felec            = _fjsp_mul_v2r8(qq22,_fjsp_msub_v2r8(rinv22,rinvsq22,krf2));
1323
1324             fscal            = felec;
1325
1326             /* Update vectorial force */
1327             fix2             = _fjsp_madd_v2r8(dx22,fscal,fix2);
1328             fiy2             = _fjsp_madd_v2r8(dy22,fscal,fiy2);
1329             fiz2             = _fjsp_madd_v2r8(dz22,fscal,fiz2);
1330
1331             fjx2             = _fjsp_madd_v2r8(dx22,fscal,fjx2);
1332             fjy2             = _fjsp_madd_v2r8(dy22,fscal,fjy2);
1333             fjz2             = _fjsp_madd_v2r8(dz22,fscal,fjz2);
1334
1335             gmx_fjsp_decrement_3rvec_2ptr_swizzle_v2r8(f+j_coord_offsetA,f+j_coord_offsetB,fjx0,fjy0,fjz0,fjx1,fjy1,fjz1,fjx2,fjy2,fjz2);
1336
1337             /* Inner loop uses 297 flops */
1338         }
1339
1340         if(jidx<j_index_end)
1341         {
1342
1343             jnrA             = jjnr[jidx];
1344             j_coord_offsetA  = DIM*jnrA;
1345
1346             /* load j atom coordinates */
1347             gmx_fjsp_load_3rvec_1ptr_swizzle_v2r8(x+j_coord_offsetA,
1348                                               &jx0,&jy0,&jz0,&jx1,&jy1,&jz1,&jx2,&jy2,&jz2);
1349
1350             /* Calculate displacement vector */
1351             dx00             = _fjsp_sub_v2r8(ix0,jx0);
1352             dy00             = _fjsp_sub_v2r8(iy0,jy0);
1353             dz00             = _fjsp_sub_v2r8(iz0,jz0);
1354             dx01             = _fjsp_sub_v2r8(ix0,jx1);
1355             dy01             = _fjsp_sub_v2r8(iy0,jy1);
1356             dz01             = _fjsp_sub_v2r8(iz0,jz1);
1357             dx02             = _fjsp_sub_v2r8(ix0,jx2);
1358             dy02             = _fjsp_sub_v2r8(iy0,jy2);
1359             dz02             = _fjsp_sub_v2r8(iz0,jz2);
1360             dx10             = _fjsp_sub_v2r8(ix1,jx0);
1361             dy10             = _fjsp_sub_v2r8(iy1,jy0);
1362             dz10             = _fjsp_sub_v2r8(iz1,jz0);
1363             dx11             = _fjsp_sub_v2r8(ix1,jx1);
1364             dy11             = _fjsp_sub_v2r8(iy1,jy1);
1365             dz11             = _fjsp_sub_v2r8(iz1,jz1);
1366             dx12             = _fjsp_sub_v2r8(ix1,jx2);
1367             dy12             = _fjsp_sub_v2r8(iy1,jy2);
1368             dz12             = _fjsp_sub_v2r8(iz1,jz2);
1369             dx20             = _fjsp_sub_v2r8(ix2,jx0);
1370             dy20             = _fjsp_sub_v2r8(iy2,jy0);
1371             dz20             = _fjsp_sub_v2r8(iz2,jz0);
1372             dx21             = _fjsp_sub_v2r8(ix2,jx1);
1373             dy21             = _fjsp_sub_v2r8(iy2,jy1);
1374             dz21             = _fjsp_sub_v2r8(iz2,jz1);
1375             dx22             = _fjsp_sub_v2r8(ix2,jx2);
1376             dy22             = _fjsp_sub_v2r8(iy2,jy2);
1377             dz22             = _fjsp_sub_v2r8(iz2,jz2);
1378
1379             /* Calculate squared distance and things based on it */
1380             rsq00            = gmx_fjsp_calc_rsq_v2r8(dx00,dy00,dz00);
1381             rsq01            = gmx_fjsp_calc_rsq_v2r8(dx01,dy01,dz01);
1382             rsq02            = gmx_fjsp_calc_rsq_v2r8(dx02,dy02,dz02);
1383             rsq10            = gmx_fjsp_calc_rsq_v2r8(dx10,dy10,dz10);
1384             rsq11            = gmx_fjsp_calc_rsq_v2r8(dx11,dy11,dz11);
1385             rsq12            = gmx_fjsp_calc_rsq_v2r8(dx12,dy12,dz12);
1386             rsq20            = gmx_fjsp_calc_rsq_v2r8(dx20,dy20,dz20);
1387             rsq21            = gmx_fjsp_calc_rsq_v2r8(dx21,dy21,dz21);
1388             rsq22            = gmx_fjsp_calc_rsq_v2r8(dx22,dy22,dz22);
1389
1390             rinv00           = gmx_fjsp_invsqrt_v2r8(rsq00);
1391             rinv01           = gmx_fjsp_invsqrt_v2r8(rsq01);
1392             rinv02           = gmx_fjsp_invsqrt_v2r8(rsq02);
1393             rinv10           = gmx_fjsp_invsqrt_v2r8(rsq10);
1394             rinv11           = gmx_fjsp_invsqrt_v2r8(rsq11);
1395             rinv12           = gmx_fjsp_invsqrt_v2r8(rsq12);
1396             rinv20           = gmx_fjsp_invsqrt_v2r8(rsq20);
1397             rinv21           = gmx_fjsp_invsqrt_v2r8(rsq21);
1398             rinv22           = gmx_fjsp_invsqrt_v2r8(rsq22);
1399
1400             rinvsq00         = _fjsp_mul_v2r8(rinv00,rinv00);
1401             rinvsq01         = _fjsp_mul_v2r8(rinv01,rinv01);
1402             rinvsq02         = _fjsp_mul_v2r8(rinv02,rinv02);
1403             rinvsq10         = _fjsp_mul_v2r8(rinv10,rinv10);
1404             rinvsq11         = _fjsp_mul_v2r8(rinv11,rinv11);
1405             rinvsq12         = _fjsp_mul_v2r8(rinv12,rinv12);
1406             rinvsq20         = _fjsp_mul_v2r8(rinv20,rinv20);
1407             rinvsq21         = _fjsp_mul_v2r8(rinv21,rinv21);
1408             rinvsq22         = _fjsp_mul_v2r8(rinv22,rinv22);
1409
1410             fjx0             = _fjsp_setzero_v2r8();
1411             fjy0             = _fjsp_setzero_v2r8();
1412             fjz0             = _fjsp_setzero_v2r8();
1413             fjx1             = _fjsp_setzero_v2r8();
1414             fjy1             = _fjsp_setzero_v2r8();
1415             fjz1             = _fjsp_setzero_v2r8();
1416             fjx2             = _fjsp_setzero_v2r8();
1417             fjy2             = _fjsp_setzero_v2r8();
1418             fjz2             = _fjsp_setzero_v2r8();
1419
1420             /**************************
1421              * CALCULATE INTERACTIONS *
1422              **************************/
1423
1424             r00              = _fjsp_mul_v2r8(rsq00,rinv00);
1425
1426             /* Calculate table index by multiplying r with table scale and truncate to integer */
1427             rt               = _fjsp_mul_v2r8(r00,vftabscale);
1428             itab_tmp         = _fjsp_dtox_v2r8(rt);
1429             vfeps            = _fjsp_sub_v2r8(rt, _fjsp_xtod_v2r8(itab_tmp));
1430             twovfeps         = _fjsp_add_v2r8(vfeps,vfeps);
1431             _fjsp_store_v2r8(&vfconv.simd,itab_tmp);
1432
1433             vfconv.i[0]     *= 8;
1434             vfconv.i[1]     *= 8;
1435
1436             /* REACTION-FIELD ELECTROSTATICS */
1437             felec            = _fjsp_mul_v2r8(qq00,_fjsp_msub_v2r8(rinv00,rinvsq00,krf2));
1438
1439             /* CUBIC SPLINE TABLE DISPERSION */
1440             Y                = _fjsp_load_v2r8( vftab + vfconv.i[0] );
1441             F                = _fjsp_setzero_v2r8();
1442             GMX_FJSP_TRANSPOSE2_V2R8(Y,F);
1443             G                = _fjsp_load_v2r8( vftab + vfconv.i[0] + 2 );
1444             H                = _fjsp_setzero_v2r8();
1445             GMX_FJSP_TRANSPOSE2_V2R8(G,H);
1446             Fp               = _fjsp_madd_v2r8(vfeps,_fjsp_madd_v2r8(H,vfeps,G),F);
1447             FF               = _fjsp_madd_v2r8(vfeps,_fjsp_madd_v2r8(twovfeps,H,G),Fp);
1448             fvdw6            = _fjsp_mul_v2r8(c6_00,FF);
1449
1450             /* CUBIC SPLINE TABLE REPULSION */
1451             Y                = _fjsp_load_v2r8( vftab + vfconv.i[0] + 4 );
1452             F                = _fjsp_setzero_v2r8();
1453             GMX_FJSP_TRANSPOSE2_V2R8(Y,F);
1454             G                = _fjsp_load_v2r8( vftab + vfconv.i[0] + 6 );
1455             H                = _fjsp_setzero_v2r8();
1456             GMX_FJSP_TRANSPOSE2_V2R8(G,H);
1457             Fp               = _fjsp_madd_v2r8(vfeps,_fjsp_madd_v2r8(H,vfeps,G),F);
1458             FF               = _fjsp_madd_v2r8(vfeps,_fjsp_madd_v2r8(twovfeps,H,G),Fp);
1459             fvdw12           = _fjsp_mul_v2r8(c12_00,FF);
1460             fvdw             = _fjsp_neg_v2r8(_fjsp_mul_v2r8(_fjsp_add_v2r8(fvdw6,fvdw12),_fjsp_mul_v2r8(vftabscale,rinv00)));
1461
1462             fscal            = _fjsp_add_v2r8(felec,fvdw);
1463
1464             fscal            = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
1465
1466             /* Update vectorial force */
1467             fix0             = _fjsp_madd_v2r8(dx00,fscal,fix0);
1468             fiy0             = _fjsp_madd_v2r8(dy00,fscal,fiy0);
1469             fiz0             = _fjsp_madd_v2r8(dz00,fscal,fiz0);
1470
1471             fjx0             = _fjsp_madd_v2r8(dx00,fscal,fjx0);
1472             fjy0             = _fjsp_madd_v2r8(dy00,fscal,fjy0);
1473             fjz0             = _fjsp_madd_v2r8(dz00,fscal,fjz0);
1474
1475             /**************************
1476              * CALCULATE INTERACTIONS *
1477              **************************/
1478
1479             /* REACTION-FIELD ELECTROSTATICS */
1480             felec            = _fjsp_mul_v2r8(qq01,_fjsp_msub_v2r8(rinv01,rinvsq01,krf2));
1481
1482             fscal            = felec;
1483
1484             fscal            = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
1485
1486             /* Update vectorial force */
1487             fix0             = _fjsp_madd_v2r8(dx01,fscal,fix0);
1488             fiy0             = _fjsp_madd_v2r8(dy01,fscal,fiy0);
1489             fiz0             = _fjsp_madd_v2r8(dz01,fscal,fiz0);
1490
1491             fjx1             = _fjsp_madd_v2r8(dx01,fscal,fjx1);
1492             fjy1             = _fjsp_madd_v2r8(dy01,fscal,fjy1);
1493             fjz1             = _fjsp_madd_v2r8(dz01,fscal,fjz1);
1494
1495             /**************************
1496              * CALCULATE INTERACTIONS *
1497              **************************/
1498
1499             /* REACTION-FIELD ELECTROSTATICS */
1500             felec            = _fjsp_mul_v2r8(qq02,_fjsp_msub_v2r8(rinv02,rinvsq02,krf2));
1501
1502             fscal            = felec;
1503
1504             fscal            = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
1505
1506             /* Update vectorial force */
1507             fix0             = _fjsp_madd_v2r8(dx02,fscal,fix0);
1508             fiy0             = _fjsp_madd_v2r8(dy02,fscal,fiy0);
1509             fiz0             = _fjsp_madd_v2r8(dz02,fscal,fiz0);
1510
1511             fjx2             = _fjsp_madd_v2r8(dx02,fscal,fjx2);
1512             fjy2             = _fjsp_madd_v2r8(dy02,fscal,fjy2);
1513             fjz2             = _fjsp_madd_v2r8(dz02,fscal,fjz2);
1514
1515             /**************************
1516              * CALCULATE INTERACTIONS *
1517              **************************/
1518
1519             /* REACTION-FIELD ELECTROSTATICS */
1520             felec            = _fjsp_mul_v2r8(qq10,_fjsp_msub_v2r8(rinv10,rinvsq10,krf2));
1521
1522             fscal            = felec;
1523
1524             fscal            = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
1525
1526             /* Update vectorial force */
1527             fix1             = _fjsp_madd_v2r8(dx10,fscal,fix1);
1528             fiy1             = _fjsp_madd_v2r8(dy10,fscal,fiy1);
1529             fiz1             = _fjsp_madd_v2r8(dz10,fscal,fiz1);
1530
1531             fjx0             = _fjsp_madd_v2r8(dx10,fscal,fjx0);
1532             fjy0             = _fjsp_madd_v2r8(dy10,fscal,fjy0);
1533             fjz0             = _fjsp_madd_v2r8(dz10,fscal,fjz0);
1534
1535             /**************************
1536              * CALCULATE INTERACTIONS *
1537              **************************/
1538
1539             /* REACTION-FIELD ELECTROSTATICS */
1540             felec            = _fjsp_mul_v2r8(qq11,_fjsp_msub_v2r8(rinv11,rinvsq11,krf2));
1541
1542             fscal            = felec;
1543
1544             fscal            = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
1545
1546             /* Update vectorial force */
1547             fix1             = _fjsp_madd_v2r8(dx11,fscal,fix1);
1548             fiy1             = _fjsp_madd_v2r8(dy11,fscal,fiy1);
1549             fiz1             = _fjsp_madd_v2r8(dz11,fscal,fiz1);
1550
1551             fjx1             = _fjsp_madd_v2r8(dx11,fscal,fjx1);
1552             fjy1             = _fjsp_madd_v2r8(dy11,fscal,fjy1);
1553             fjz1             = _fjsp_madd_v2r8(dz11,fscal,fjz1);
1554
1555             /**************************
1556              * CALCULATE INTERACTIONS *
1557              **************************/
1558
1559             /* REACTION-FIELD ELECTROSTATICS */
1560             felec            = _fjsp_mul_v2r8(qq12,_fjsp_msub_v2r8(rinv12,rinvsq12,krf2));
1561
1562             fscal            = felec;
1563
1564             fscal            = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
1565
1566             /* Update vectorial force */
1567             fix1             = _fjsp_madd_v2r8(dx12,fscal,fix1);
1568             fiy1             = _fjsp_madd_v2r8(dy12,fscal,fiy1);
1569             fiz1             = _fjsp_madd_v2r8(dz12,fscal,fiz1);
1570
1571             fjx2             = _fjsp_madd_v2r8(dx12,fscal,fjx2);
1572             fjy2             = _fjsp_madd_v2r8(dy12,fscal,fjy2);
1573             fjz2             = _fjsp_madd_v2r8(dz12,fscal,fjz2);
1574
1575             /**************************
1576              * CALCULATE INTERACTIONS *
1577              **************************/
1578
1579             /* REACTION-FIELD ELECTROSTATICS */
1580             felec            = _fjsp_mul_v2r8(qq20,_fjsp_msub_v2r8(rinv20,rinvsq20,krf2));
1581
1582             fscal            = felec;
1583
1584             fscal            = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
1585
1586             /* Update vectorial force */
1587             fix2             = _fjsp_madd_v2r8(dx20,fscal,fix2);
1588             fiy2             = _fjsp_madd_v2r8(dy20,fscal,fiy2);
1589             fiz2             = _fjsp_madd_v2r8(dz20,fscal,fiz2);
1590
1591             fjx0             = _fjsp_madd_v2r8(dx20,fscal,fjx0);
1592             fjy0             = _fjsp_madd_v2r8(dy20,fscal,fjy0);
1593             fjz0             = _fjsp_madd_v2r8(dz20,fscal,fjz0);
1594
1595             /**************************
1596              * CALCULATE INTERACTIONS *
1597              **************************/
1598
1599             /* REACTION-FIELD ELECTROSTATICS */
1600             felec            = _fjsp_mul_v2r8(qq21,_fjsp_msub_v2r8(rinv21,rinvsq21,krf2));
1601
1602             fscal            = felec;
1603
1604             fscal            = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
1605
1606             /* Update vectorial force */
1607             fix2             = _fjsp_madd_v2r8(dx21,fscal,fix2);
1608             fiy2             = _fjsp_madd_v2r8(dy21,fscal,fiy2);
1609             fiz2             = _fjsp_madd_v2r8(dz21,fscal,fiz2);
1610
1611             fjx1             = _fjsp_madd_v2r8(dx21,fscal,fjx1);
1612             fjy1             = _fjsp_madd_v2r8(dy21,fscal,fjy1);
1613             fjz1             = _fjsp_madd_v2r8(dz21,fscal,fjz1);
1614
1615             /**************************
1616              * CALCULATE INTERACTIONS *
1617              **************************/
1618
1619             /* REACTION-FIELD ELECTROSTATICS */
1620             felec            = _fjsp_mul_v2r8(qq22,_fjsp_msub_v2r8(rinv22,rinvsq22,krf2));
1621
1622             fscal            = felec;
1623
1624             fscal            = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
1625
1626             /* Update vectorial force */
1627             fix2             = _fjsp_madd_v2r8(dx22,fscal,fix2);
1628             fiy2             = _fjsp_madd_v2r8(dy22,fscal,fiy2);
1629             fiz2             = _fjsp_madd_v2r8(dz22,fscal,fiz2);
1630
1631             fjx2             = _fjsp_madd_v2r8(dx22,fscal,fjx2);
1632             fjy2             = _fjsp_madd_v2r8(dy22,fscal,fjy2);
1633             fjz2             = _fjsp_madd_v2r8(dz22,fscal,fjz2);
1634
1635             gmx_fjsp_decrement_3rvec_1ptr_swizzle_v2r8(f+j_coord_offsetA,fjx0,fjy0,fjz0,fjx1,fjy1,fjz1,fjx2,fjy2,fjz2);
1636
1637             /* Inner loop uses 297 flops */
1638         }
1639
1640         /* End of innermost loop */
1641
1642         gmx_fjsp_update_iforce_3atom_swizzle_v2r8(fix0,fiy0,fiz0,fix1,fiy1,fiz1,fix2,fiy2,fiz2,
1643                                               f+i_coord_offset,fshift+i_shift_offset);
1644
1645         /* Increment number of inner iterations */
1646         inneriter                  += j_index_end - j_index_start;
1647
1648         /* Outer loop uses 18 flops */
1649     }
1650
1651     /* Increment number of outer iterations */
1652     outeriter        += nri;
1653
1654     /* Update outer/inner flops */
1655
1656     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W3W3_F,outeriter*18 + inneriter*297);
1657 }