src/gromacs/gmxlib/nonbonded/nb_kernel_sparc64_hpc_ace_double/nb_kernel_ElecRF_VdwCSTab_GeomW4W4_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_GeomW4W4_VF_sparc64_hpc_ace_double
  53  * Electrostatics interaction: ReactionField
  54  * VdW interaction:            CubicSplineTable
  55  * Geometry:                   Water4-Water4
  56  * Calculate force/pot:        PotentialAndForce
  57  */
  58 void
  59 nb_kernel_ElecRF_VdwCSTab_GeomW4W4_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              vdwioffset3;
  88     _fjsp_v2r8       ix3,iy3,iz3,fix3,fiy3,fiz3,iq3,isai3;
  89     int              vdwjidx0A,vdwjidx0B;
  90     _fjsp_v2r8       jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
  91     int              vdwjidx1A,vdwjidx1B;
  92     _fjsp_v2r8       jx1,jy1,jz1,fjx1,fjy1,fjz1,jq1,isaj1;
  93     int              vdwjidx2A,vdwjidx2B;
  94     _fjsp_v2r8       jx2,jy2,jz2,fjx2,fjy2,fjz2,jq2,isaj2;
  95     int              vdwjidx3A,vdwjidx3B;
  96     _fjsp_v2r8       jx3,jy3,jz3,fjx3,fjy3,fjz3,jq3,isaj3;
  97     _fjsp_v2r8       dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00;
  98     _fjsp_v2r8       dx11,dy11,dz11,rsq11,rinv11,rinvsq11,r11,qq11,c6_11,c12_11;
  99     _fjsp_v2r8       dx12,dy12,dz12,rsq12,rinv12,rinvsq12,r12,qq12,c6_12,c12_12;
 100     _fjsp_v2r8       dx13,dy13,dz13,rsq13,rinv13,rinvsq13,r13,qq13,c6_13,c12_13;
 101     _fjsp_v2r8       dx21,dy21,dz21,rsq21,rinv21,rinvsq21,r21,qq21,c6_21,c12_21;
 102     _fjsp_v2r8       dx22,dy22,dz22,rsq22,rinv22,rinvsq22,r22,qq22,c6_22,c12_22;
 103     _fjsp_v2r8       dx23,dy23,dz23,rsq23,rinv23,rinvsq23,r23,qq23,c6_23,c12_23;
 104     _fjsp_v2r8       dx31,dy31,dz31,rsq31,rinv31,rinvsq31,r31,qq31,c6_31,c12_31;
 105     _fjsp_v2r8       dx32,dy32,dz32,rsq32,rinv32,rinvsq32,r32,qq32,c6_32,c12_32;
 106     _fjsp_v2r8       dx33,dy33,dz33,rsq33,rinv33,rinvsq33,r33,qq33,c6_33,c12_33;
 107     _fjsp_v2r8       velec,felec,velecsum,facel,crf,krf,krf2;
 108     real             *charge;
 109     int              nvdwtype;
 110     _fjsp_v2r8       rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,sh_vdw_invrcut6;
 111     int              *vdwtype;
 112     real             *vdwparam;
 113     _fjsp_v2r8       one_sixth   = gmx_fjsp_set1_v2r8(1.0/6.0);
 114     _fjsp_v2r8       one_twelfth = gmx_fjsp_set1_v2r8(1.0/12.0);
 115     _fjsp_v2r8       rt,vfeps,vftabscale,Y,F,G,H,Heps,Fp,VV,FF,twovfeps;
 116     real             *vftab;
 117     _fjsp_v2r8       itab_tmp;
 118     _fjsp_v2r8       dummy_mask,cutoff_mask;
 119     _fjsp_v2r8       one     = gmx_fjsp_set1_v2r8(1.0);
 120     _fjsp_v2r8       two     = gmx_fjsp_set1_v2r8(2.0);
 121     union { _fjsp_v2r8 simd; long long int i[2]; } vfconv,gbconv,ewconv;
 122
 123     x                = xx[0];
 124     f                = ff[0];
 125
 126     nri              = nlist->nri;
 127     iinr             = nlist->iinr;
 128     jindex           = nlist->jindex;
 129     jjnr             = nlist->jjnr;
 130     shiftidx         = nlist->shift;
 131     gid              = nlist->gid;
 132     shiftvec         = fr->shift_vec[0];
 133     fshift           = fr->fshift[0];
 134     facel            = gmx_fjsp_set1_v2r8(fr->epsfac);
 135     charge           = mdatoms->chargeA;
 136     krf              = gmx_fjsp_set1_v2r8(fr->ic->k_rf);
 137     krf2             = gmx_fjsp_set1_v2r8(fr->ic->k_rf*2.0);
 138     crf              = gmx_fjsp_set1_v2r8(fr->ic->c_rf);
 139     nvdwtype         = fr->ntype;
 140     vdwparam         = fr->nbfp;
 141     vdwtype          = mdatoms->typeA;
 142
 143     vftab            = kernel_data->table_vdw->data;
 144     vftabscale       = gmx_fjsp_set1_v2r8(kernel_data->table_vdw->scale);
 145
 146     /* Setup water-specific parameters */
 147     inr              = nlist->iinr[0];
 148     iq1              = _fjsp_mul_v2r8(facel,gmx_fjsp_set1_v2r8(charge[inr+1]));
 149     iq2              = _fjsp_mul_v2r8(facel,gmx_fjsp_set1_v2r8(charge[inr+2]));
 150     iq3              = _fjsp_mul_v2r8(facel,gmx_fjsp_set1_v2r8(charge[inr+3]));
 151     vdwioffset0      = 2*nvdwtype*vdwtype[inr+0];
 152
 153     jq1              = gmx_fjsp_set1_v2r8(charge[inr+1]);
 154     jq2              = gmx_fjsp_set1_v2r8(charge[inr+2]);
 155     jq3              = gmx_fjsp_set1_v2r8(charge[inr+3]);
 156     vdwjidx0A        = 2*vdwtype[inr+0];
 157     c6_00            = gmx_fjsp_set1_v2r8(vdwparam[vdwioffset0+vdwjidx0A]);
 158     c12_00           = gmx_fjsp_set1_v2r8(vdwparam[vdwioffset0+vdwjidx0A+1]);
 159     qq11             = _fjsp_mul_v2r8(iq1,jq1);
 160     qq12             = _fjsp_mul_v2r8(iq1,jq2);
 161     qq13             = _fjsp_mul_v2r8(iq1,jq3);
 162     qq21             = _fjsp_mul_v2r8(iq2,jq1);
 163     qq22             = _fjsp_mul_v2r8(iq2,jq2);
 164     qq23             = _fjsp_mul_v2r8(iq2,jq3);
 165     qq31             = _fjsp_mul_v2r8(iq3,jq1);
 166     qq32             = _fjsp_mul_v2r8(iq3,jq2);
 167     qq33             = _fjsp_mul_v2r8(iq3,jq3);
 168
 169     /* Avoid stupid compiler warnings */
 170     jnrA = jnrB = 0;
 171     j_coord_offsetA = 0;
 172     j_coord_offsetB = 0;
 173
 174     outeriter        = 0;
 175     inneriter        = 0;
 176
 177     /* Start outer loop over neighborlists */
 178     for(iidx=0; iidx<nri; iidx++)
 179     {
 180         /* Load shift vector for this list */
 181         i_shift_offset   = DIM*shiftidx[iidx];
 182
 183         /* Load limits for loop over neighbors */
 184         j_index_start    = jindex[iidx];
 185         j_index_end      = jindex[iidx+1];
 186
 187         /* Get outer coordinate index */
 188         inr              = iinr[iidx];
 189         i_coord_offset   = DIM*inr;
 190
 191         /* Load i particle coords and add shift vector */
 192         gmx_fjsp_load_shift_and_4rvec_broadcast_v2r8(shiftvec+i_shift_offset,x+i_coord_offset,
 193                                                  &ix0,&iy0,&iz0,&ix1,&iy1,&iz1,&ix2,&iy2,&iz2,&ix3,&iy3,&iz3);
 194
 195         fix0             = _fjsp_setzero_v2r8();
 196         fiy0             = _fjsp_setzero_v2r8();
 197         fiz0             = _fjsp_setzero_v2r8();
 198         fix1             = _fjsp_setzero_v2r8();
 199         fiy1             = _fjsp_setzero_v2r8();
 200         fiz1             = _fjsp_setzero_v2r8();
 201         fix2             = _fjsp_setzero_v2r8();
 202         fiy2             = _fjsp_setzero_v2r8();
 203         fiz2             = _fjsp_setzero_v2r8();
 204         fix3             = _fjsp_setzero_v2r8();
 205         fiy3             = _fjsp_setzero_v2r8();
 206         fiz3             = _fjsp_setzero_v2r8();
 207
 208         /* Reset potential sums */
 209         velecsum         = _fjsp_setzero_v2r8();
 210         vvdwsum          = _fjsp_setzero_v2r8();
 211
 212         /* Start inner kernel loop */
 213         for(jidx=j_index_start; jidx<j_index_end-1; jidx+=2)
 214         {
 215
 216             /* Get j neighbor index, and coordinate index */
 217             jnrA             = jjnr[jidx];
 218             jnrB             = jjnr[jidx+1];
 219             j_coord_offsetA  = DIM*jnrA;
 220             j_coord_offsetB  = DIM*jnrB;
 221
 222             /* load j atom coordinates */
 223             gmx_fjsp_load_4rvec_2ptr_swizzle_v2r8(x+j_coord_offsetA,x+j_coord_offsetB,
 224                                               &jx0,&jy0,&jz0,&jx1,&jy1,&jz1,&jx2,
 225                                               &jy2,&jz2,&jx3,&jy3,&jz3);
 226
 227             /* Calculate displacement vector */
 228             dx00             = _fjsp_sub_v2r8(ix0,jx0);
 229             dy00             = _fjsp_sub_v2r8(iy0,jy0);
 230             dz00             = _fjsp_sub_v2r8(iz0,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             dx13             = _fjsp_sub_v2r8(ix1,jx3);
 238             dy13             = _fjsp_sub_v2r8(iy1,jy3);
 239             dz13             = _fjsp_sub_v2r8(iz1,jz3);
 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             dx23             = _fjsp_sub_v2r8(ix2,jx3);
 247             dy23             = _fjsp_sub_v2r8(iy2,jy3);
 248             dz23             = _fjsp_sub_v2r8(iz2,jz3);
 249             dx31             = _fjsp_sub_v2r8(ix3,jx1);
 250             dy31             = _fjsp_sub_v2r8(iy3,jy1);
 251             dz31             = _fjsp_sub_v2r8(iz3,jz1);
 252             dx32             = _fjsp_sub_v2r8(ix3,jx2);
 253             dy32             = _fjsp_sub_v2r8(iy3,jy2);
 254             dz32             = _fjsp_sub_v2r8(iz3,jz2);
 255             dx33             = _fjsp_sub_v2r8(ix3,jx3);
 256             dy33             = _fjsp_sub_v2r8(iy3,jy3);
 257             dz33             = _fjsp_sub_v2r8(iz3,jz3);
 258
 259             /* Calculate squared distance and things based on it */
 260             rsq00            = gmx_fjsp_calc_rsq_v2r8(dx00,dy00,dz00);
 261             rsq11            = gmx_fjsp_calc_rsq_v2r8(dx11,dy11,dz11);
 262             rsq12            = gmx_fjsp_calc_rsq_v2r8(dx12,dy12,dz12);
 263             rsq13            = gmx_fjsp_calc_rsq_v2r8(dx13,dy13,dz13);
 264             rsq21            = gmx_fjsp_calc_rsq_v2r8(dx21,dy21,dz21);
 265             rsq22            = gmx_fjsp_calc_rsq_v2r8(dx22,dy22,dz22);
 266             rsq23            = gmx_fjsp_calc_rsq_v2r8(dx23,dy23,dz23);
 267             rsq31            = gmx_fjsp_calc_rsq_v2r8(dx31,dy31,dz31);
 268             rsq32            = gmx_fjsp_calc_rsq_v2r8(dx32,dy32,dz32);
 269             rsq33            = gmx_fjsp_calc_rsq_v2r8(dx33,dy33,dz33);
 270
 271             rinv00           = gmx_fjsp_invsqrt_v2r8(rsq00);
 272             rinv11           = gmx_fjsp_invsqrt_v2r8(rsq11);
 273             rinv12           = gmx_fjsp_invsqrt_v2r8(rsq12);
 274             rinv13           = gmx_fjsp_invsqrt_v2r8(rsq13);
 275             rinv21           = gmx_fjsp_invsqrt_v2r8(rsq21);
 276             rinv22           = gmx_fjsp_invsqrt_v2r8(rsq22);
 277             rinv23           = gmx_fjsp_invsqrt_v2r8(rsq23);
 278             rinv31           = gmx_fjsp_invsqrt_v2r8(rsq31);
 279             rinv32           = gmx_fjsp_invsqrt_v2r8(rsq32);
 280             rinv33           = gmx_fjsp_invsqrt_v2r8(rsq33);
 281
 282             rinvsq11         = _fjsp_mul_v2r8(rinv11,rinv11);
 283             rinvsq12         = _fjsp_mul_v2r8(rinv12,rinv12);
 284             rinvsq13         = _fjsp_mul_v2r8(rinv13,rinv13);
 285             rinvsq21         = _fjsp_mul_v2r8(rinv21,rinv21);
 286             rinvsq22         = _fjsp_mul_v2r8(rinv22,rinv22);
 287             rinvsq23         = _fjsp_mul_v2r8(rinv23,rinv23);
 288             rinvsq31         = _fjsp_mul_v2r8(rinv31,rinv31);
 289             rinvsq32         = _fjsp_mul_v2r8(rinv32,rinv32);
 290             rinvsq33         = _fjsp_mul_v2r8(rinv33,rinv33);
 291
 292             fjx0             = _fjsp_setzero_v2r8();
 293             fjy0             = _fjsp_setzero_v2r8();
 294             fjz0             = _fjsp_setzero_v2r8();
 295             fjx1             = _fjsp_setzero_v2r8();
 296             fjy1             = _fjsp_setzero_v2r8();
 297             fjz1             = _fjsp_setzero_v2r8();
 298             fjx2             = _fjsp_setzero_v2r8();
 299             fjy2             = _fjsp_setzero_v2r8();
 300             fjz2             = _fjsp_setzero_v2r8();
 301             fjx3             = _fjsp_setzero_v2r8();
 302             fjy3             = _fjsp_setzero_v2r8();
 303             fjz3             = _fjsp_setzero_v2r8();
 304
 305             /**************************
 306              * CALCULATE INTERACTIONS *
 307              **************************/
 308
 309             r00              = _fjsp_mul_v2r8(rsq00,rinv00);
 310
 311             /* Calculate table index by multiplying r with table scale and truncate to integer */
 312             rt               = _fjsp_mul_v2r8(r00,vftabscale);
 313             itab_tmp         = _fjsp_dtox_v2r8(rt);
 314             vfeps            = _fjsp_sub_v2r8(rt, _fjsp_xtod_v2r8(itab_tmp));
 315             twovfeps         = _fjsp_add_v2r8(vfeps,vfeps);
 316             _fjsp_store_v2r8(&vfconv.simd,itab_tmp);
 317
 318             vfconv.i[0]     *= 8;
 319             vfconv.i[1]     *= 8;
 320
 321             /* CUBIC SPLINE TABLE DISPERSION */
 322             Y                = _fjsp_load_v2r8( vftab + vfconv.i[0] );
 323             F                = _fjsp_load_v2r8( vftab + vfconv.i[1] );
 324             GMX_FJSP_TRANSPOSE2_V2R8(Y,F);
 325             G                = _fjsp_load_v2r8( vftab + vfconv.i[0] + 2 );
 326             H                = _fjsp_load_v2r8( vftab + vfconv.i[1] + 2 );
 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             vvdw6            = _fjsp_mul_v2r8(c6_00,VV);
 331             FF               = _fjsp_madd_v2r8(vfeps,_fjsp_madd_v2r8(twovfeps,H,G),Fp);
 332             fvdw6            = _fjsp_mul_v2r8(c6_00,FF);
 333
 334             /* CUBIC SPLINE TABLE REPULSION */
 335             Y                = _fjsp_load_v2r8( vftab + vfconv.i[0] + 4 );
 336             F                = _fjsp_load_v2r8( vftab + vfconv.i[1] + 4 );
 337             GMX_FJSP_TRANSPOSE2_V2R8(Y,F);
 338             G                = _fjsp_load_v2r8( vftab + vfconv.i[0] + 6 );
 339             H                = _fjsp_load_v2r8( vftab + vfconv.i[1] + 6 );
 340             GMX_FJSP_TRANSPOSE2_V2R8(G,H);
 341             Fp               = _fjsp_madd_v2r8(vfeps,_fjsp_madd_v2r8(H,vfeps,G),F);
 342             VV               = _fjsp_madd_v2r8(vfeps,Fp,Y);
 343             vvdw12           = _fjsp_mul_v2r8(c12_00,VV);
 344             FF               = _fjsp_madd_v2r8(vfeps,_fjsp_madd_v2r8(twovfeps,H,G),Fp);
 345             fvdw12           = _fjsp_mul_v2r8(c12_00,FF);
 346             vvdw             = _fjsp_add_v2r8(vvdw12,vvdw6);
 347             fvdw             = _fjsp_neg_v2r8(_fjsp_mul_v2r8(_fjsp_add_v2r8(fvdw6,fvdw12),_fjsp_mul_v2r8(vftabscale,rinv00)));
 348
 349             /* Update potential sum for this i atom from the interaction with this j atom. */
 350             vvdwsum          = _fjsp_add_v2r8(vvdwsum,vvdw);
 351
 352             fscal            = fvdw;
 353
 354             /* Update vectorial force */
 355             fix0             = _fjsp_madd_v2r8(dx00,fscal,fix0);
 356             fiy0             = _fjsp_madd_v2r8(dy00,fscal,fiy0);
 357             fiz0             = _fjsp_madd_v2r8(dz00,fscal,fiz0);
 358
 359             fjx0             = _fjsp_madd_v2r8(dx00,fscal,fjx0);
 360             fjy0             = _fjsp_madd_v2r8(dy00,fscal,fjy0);
 361             fjz0             = _fjsp_madd_v2r8(dz00,fscal,fjz0);
 362
 363             /**************************
 364              * CALCULATE INTERACTIONS *
 365              **************************/
 366
 367             /* REACTION-FIELD ELECTROSTATICS */
 368             velec            = _fjsp_mul_v2r8(qq11,_fjsp_sub_v2r8(_fjsp_madd_v2r8(krf,rsq11,rinv11),crf));
 369             felec            = _fjsp_mul_v2r8(qq11,_fjsp_msub_v2r8(rinv11,rinvsq11,krf2));
 370
 371             /* Update potential sum for this i atom from the interaction with this j atom. */
 372             velecsum         = _fjsp_add_v2r8(velecsum,velec);
 373
 374             fscal            = felec;
 375
 376             /* Update vectorial force */
 377             fix1             = _fjsp_madd_v2r8(dx11,fscal,fix1);
 378             fiy1             = _fjsp_madd_v2r8(dy11,fscal,fiy1);
 379             fiz1             = _fjsp_madd_v2r8(dz11,fscal,fiz1);
 380
 381             fjx1             = _fjsp_madd_v2r8(dx11,fscal,fjx1);
 382             fjy1             = _fjsp_madd_v2r8(dy11,fscal,fjy1);
 383             fjz1             = _fjsp_madd_v2r8(dz11,fscal,fjz1);
 384
 385             /**************************
 386              * CALCULATE INTERACTIONS *
 387              **************************/
 388
 389             /* REACTION-FIELD ELECTROSTATICS */
 390             velec            = _fjsp_mul_v2r8(qq12,_fjsp_sub_v2r8(_fjsp_madd_v2r8(krf,rsq12,rinv12),crf));
 391             felec            = _fjsp_mul_v2r8(qq12,_fjsp_msub_v2r8(rinv12,rinvsq12,krf2));
 392
 393             /* Update potential sum for this i atom from the interaction with this j atom. */
 394             velecsum         = _fjsp_add_v2r8(velecsum,velec);
 395
 396             fscal            = felec;
 397
 398             /* Update vectorial force */
 399             fix1             = _fjsp_madd_v2r8(dx12,fscal,fix1);
 400             fiy1             = _fjsp_madd_v2r8(dy12,fscal,fiy1);
 401             fiz1             = _fjsp_madd_v2r8(dz12,fscal,fiz1);
 402
 403             fjx2             = _fjsp_madd_v2r8(dx12,fscal,fjx2);
 404             fjy2             = _fjsp_madd_v2r8(dy12,fscal,fjy2);
 405             fjz2             = _fjsp_madd_v2r8(dz12,fscal,fjz2);
 406
 407             /**************************
 408              * CALCULATE INTERACTIONS *
 409              **************************/
 410
 411             /* REACTION-FIELD ELECTROSTATICS */
 412             velec            = _fjsp_mul_v2r8(qq13,_fjsp_sub_v2r8(_fjsp_madd_v2r8(krf,rsq13,rinv13),crf));
 413             felec            = _fjsp_mul_v2r8(qq13,_fjsp_msub_v2r8(rinv13,rinvsq13,krf2));
 414
 415             /* Update potential sum for this i atom from the interaction with this j atom. */
 416             velecsum         = _fjsp_add_v2r8(velecsum,velec);
 417
 418             fscal            = felec;
 419
 420             /* Update vectorial force */
 421             fix1             = _fjsp_madd_v2r8(dx13,fscal,fix1);
 422             fiy1             = _fjsp_madd_v2r8(dy13,fscal,fiy1);
 423             fiz1             = _fjsp_madd_v2r8(dz13,fscal,fiz1);
 424
 425             fjx3             = _fjsp_madd_v2r8(dx13,fscal,fjx3);
 426             fjy3             = _fjsp_madd_v2r8(dy13,fscal,fjy3);
 427             fjz3             = _fjsp_madd_v2r8(dz13,fscal,fjz3);
 428
 429             /**************************
 430              * CALCULATE INTERACTIONS *
 431              **************************/
 432
 433             /* REACTION-FIELD ELECTROSTATICS */
 434             velec            = _fjsp_mul_v2r8(qq21,_fjsp_sub_v2r8(_fjsp_madd_v2r8(krf,rsq21,rinv21),crf));
 435             felec            = _fjsp_mul_v2r8(qq21,_fjsp_msub_v2r8(rinv21,rinvsq21,krf2));
 436
 437             /* Update potential sum for this i atom from the interaction with this j atom. */
 438             velecsum         = _fjsp_add_v2r8(velecsum,velec);
 439
 440             fscal            = felec;
 441
 442             /* Update vectorial force */
 443             fix2             = _fjsp_madd_v2r8(dx21,fscal,fix2);
 444             fiy2             = _fjsp_madd_v2r8(dy21,fscal,fiy2);
 445             fiz2             = _fjsp_madd_v2r8(dz21,fscal,fiz2);
 446
 447             fjx1             = _fjsp_madd_v2r8(dx21,fscal,fjx1);
 448             fjy1             = _fjsp_madd_v2r8(dy21,fscal,fjy1);
 449             fjz1             = _fjsp_madd_v2r8(dz21,fscal,fjz1);
 450
 451             /**************************
 452              * CALCULATE INTERACTIONS *
 453              **************************/
 454
 455             /* REACTION-FIELD ELECTROSTATICS */
 456             velec            = _fjsp_mul_v2r8(qq22,_fjsp_sub_v2r8(_fjsp_madd_v2r8(krf,rsq22,rinv22),crf));
 457             felec            = _fjsp_mul_v2r8(qq22,_fjsp_msub_v2r8(rinv22,rinvsq22,krf2));
 458
 459             /* Update potential sum for this i atom from the interaction with this j atom. */
 460             velecsum         = _fjsp_add_v2r8(velecsum,velec);
 461
 462             fscal            = felec;
 463
 464             /* Update vectorial force */
 465             fix2             = _fjsp_madd_v2r8(dx22,fscal,fix2);
 466             fiy2             = _fjsp_madd_v2r8(dy22,fscal,fiy2);
 467             fiz2             = _fjsp_madd_v2r8(dz22,fscal,fiz2);
 468
 469             fjx2             = _fjsp_madd_v2r8(dx22,fscal,fjx2);
 470             fjy2             = _fjsp_madd_v2r8(dy22,fscal,fjy2);
 471             fjz2             = _fjsp_madd_v2r8(dz22,fscal,fjz2);
 472
 473             /**************************
 474              * CALCULATE INTERACTIONS *
 475              **************************/
 476
 477             /* REACTION-FIELD ELECTROSTATICS */
 478             velec            = _fjsp_mul_v2r8(qq23,_fjsp_sub_v2r8(_fjsp_madd_v2r8(krf,rsq23,rinv23),crf));
 479             felec            = _fjsp_mul_v2r8(qq23,_fjsp_msub_v2r8(rinv23,rinvsq23,krf2));
 480
 481             /* Update potential sum for this i atom from the interaction with this j atom. */
 482             velecsum         = _fjsp_add_v2r8(velecsum,velec);
 483
 484             fscal            = felec;
 485
 486             /* Update vectorial force */
 487             fix2             = _fjsp_madd_v2r8(dx23,fscal,fix2);
 488             fiy2             = _fjsp_madd_v2r8(dy23,fscal,fiy2);
 489             fiz2             = _fjsp_madd_v2r8(dz23,fscal,fiz2);
 490
 491             fjx3             = _fjsp_madd_v2r8(dx23,fscal,fjx3);
 492             fjy3             = _fjsp_madd_v2r8(dy23,fscal,fjy3);
 493             fjz3             = _fjsp_madd_v2r8(dz23,fscal,fjz3);
 494
 495             /**************************
 496              * CALCULATE INTERACTIONS *
 497              **************************/
 498
 499             /* REACTION-FIELD ELECTROSTATICS */
 500             velec            = _fjsp_mul_v2r8(qq31,_fjsp_sub_v2r8(_fjsp_madd_v2r8(krf,rsq31,rinv31),crf));
 501             felec            = _fjsp_mul_v2r8(qq31,_fjsp_msub_v2r8(rinv31,rinvsq31,krf2));
 502
 503             /* Update potential sum for this i atom from the interaction with this j atom. */
 504             velecsum         = _fjsp_add_v2r8(velecsum,velec);
 505
 506             fscal            = felec;
 507
 508             /* Update vectorial force */
 509             fix3             = _fjsp_madd_v2r8(dx31,fscal,fix3);
 510             fiy3             = _fjsp_madd_v2r8(dy31,fscal,fiy3);
 511             fiz3             = _fjsp_madd_v2r8(dz31,fscal,fiz3);
 512
 513             fjx1             = _fjsp_madd_v2r8(dx31,fscal,fjx1);
 514             fjy1             = _fjsp_madd_v2r8(dy31,fscal,fjy1);
 515             fjz1             = _fjsp_madd_v2r8(dz31,fscal,fjz1);
 516
 517             /**************************
 518              * CALCULATE INTERACTIONS *
 519              **************************/
 520
 521             /* REACTION-FIELD ELECTROSTATICS */
 522             velec            = _fjsp_mul_v2r8(qq32,_fjsp_sub_v2r8(_fjsp_madd_v2r8(krf,rsq32,rinv32),crf));
 523             felec            = _fjsp_mul_v2r8(qq32,_fjsp_msub_v2r8(rinv32,rinvsq32,krf2));
 524
 525             /* Update potential sum for this i atom from the interaction with this j atom. */
 526             velecsum         = _fjsp_add_v2r8(velecsum,velec);
 527
 528             fscal            = felec;
 529
 530             /* Update vectorial force */
 531             fix3             = _fjsp_madd_v2r8(dx32,fscal,fix3);
 532             fiy3             = _fjsp_madd_v2r8(dy32,fscal,fiy3);
 533             fiz3             = _fjsp_madd_v2r8(dz32,fscal,fiz3);
 534
 535             fjx2             = _fjsp_madd_v2r8(dx32,fscal,fjx2);
 536             fjy2             = _fjsp_madd_v2r8(dy32,fscal,fjy2);
 537             fjz2             = _fjsp_madd_v2r8(dz32,fscal,fjz2);
 538
 539             /**************************
 540              * CALCULATE INTERACTIONS *
 541              **************************/
 542
 543             /* REACTION-FIELD ELECTROSTATICS */
 544             velec            = _fjsp_mul_v2r8(qq33,_fjsp_sub_v2r8(_fjsp_madd_v2r8(krf,rsq33,rinv33),crf));
 545             felec            = _fjsp_mul_v2r8(qq33,_fjsp_msub_v2r8(rinv33,rinvsq33,krf2));
 546
 547             /* Update potential sum for this i atom from the interaction with this j atom. */
 548             velecsum         = _fjsp_add_v2r8(velecsum,velec);
 549
 550             fscal            = felec;
 551
 552             /* Update vectorial force */
 553             fix3             = _fjsp_madd_v2r8(dx33,fscal,fix3);
 554             fiy3             = _fjsp_madd_v2r8(dy33,fscal,fiy3);
 555             fiz3             = _fjsp_madd_v2r8(dz33,fscal,fiz3);
 556
 557             fjx3             = _fjsp_madd_v2r8(dx33,fscal,fjx3);
 558             fjy3             = _fjsp_madd_v2r8(dy33,fscal,fjy3);
 559             fjz3             = _fjsp_madd_v2r8(dz33,fscal,fjz3);
 560
 561             gmx_fjsp_decrement_4rvec_2ptr_swizzle_v2r8(f+j_coord_offsetA,f+j_coord_offsetB,fjx0,fjy0,fjz0,fjx1,fjy1,fjz1,fjx2,fjy2,fjz2,fjx3,fjy3,fjz3);
 562
 563             /* Inner loop uses 377 flops */
 564         }
 565
 566         if(jidx<j_index_end)
 567         {
 568
 569             jnrA             = jjnr[jidx];
 570             j_coord_offsetA  = DIM*jnrA;
 571
 572             /* load j atom coordinates */
 573             gmx_fjsp_load_4rvec_1ptr_swizzle_v2r8(x+j_coord_offsetA,
 574                                               &jx0,&jy0,&jz0,&jx1,&jy1,&jz1,&jx2,
 575                                               &jy2,&jz2,&jx3,&jy3,&jz3);
 576
 577             /* Calculate displacement vector */
 578             dx00             = _fjsp_sub_v2r8(ix0,jx0);
 579             dy00             = _fjsp_sub_v2r8(iy0,jy0);
 580             dz00             = _fjsp_sub_v2r8(iz0,jz0);
 581             dx11             = _fjsp_sub_v2r8(ix1,jx1);
 582             dy11             = _fjsp_sub_v2r8(iy1,jy1);
 583             dz11             = _fjsp_sub_v2r8(iz1,jz1);
 584             dx12             = _fjsp_sub_v2r8(ix1,jx2);
 585             dy12             = _fjsp_sub_v2r8(iy1,jy2);
 586             dz12             = _fjsp_sub_v2r8(iz1,jz2);
 587             dx13             = _fjsp_sub_v2r8(ix1,jx3);
 588             dy13             = _fjsp_sub_v2r8(iy1,jy3);
 589             dz13             = _fjsp_sub_v2r8(iz1,jz3);
 590             dx21             = _fjsp_sub_v2r8(ix2,jx1);
 591             dy21             = _fjsp_sub_v2r8(iy2,jy1);
 592             dz21             = _fjsp_sub_v2r8(iz2,jz1);
 593             dx22             = _fjsp_sub_v2r8(ix2,jx2);
 594             dy22             = _fjsp_sub_v2r8(iy2,jy2);
 595             dz22             = _fjsp_sub_v2r8(iz2,jz2);
 596             dx23             = _fjsp_sub_v2r8(ix2,jx3);
 597             dy23             = _fjsp_sub_v2r8(iy2,jy3);
 598             dz23             = _fjsp_sub_v2r8(iz2,jz3);
 599             dx31             = _fjsp_sub_v2r8(ix3,jx1);
 600             dy31             = _fjsp_sub_v2r8(iy3,jy1);
 601             dz31             = _fjsp_sub_v2r8(iz3,jz1);
 602             dx32             = _fjsp_sub_v2r8(ix3,jx2);
 603             dy32             = _fjsp_sub_v2r8(iy3,jy2);
 604             dz32             = _fjsp_sub_v2r8(iz3,jz2);
 605             dx33             = _fjsp_sub_v2r8(ix3,jx3);
 606             dy33             = _fjsp_sub_v2r8(iy3,jy3);
 607             dz33             = _fjsp_sub_v2r8(iz3,jz3);
 608
 609             /* Calculate squared distance and things based on it */
 610             rsq00            = gmx_fjsp_calc_rsq_v2r8(dx00,dy00,dz00);
 611             rsq11            = gmx_fjsp_calc_rsq_v2r8(dx11,dy11,dz11);
 612             rsq12            = gmx_fjsp_calc_rsq_v2r8(dx12,dy12,dz12);
 613             rsq13            = gmx_fjsp_calc_rsq_v2r8(dx13,dy13,dz13);
 614             rsq21            = gmx_fjsp_calc_rsq_v2r8(dx21,dy21,dz21);
 615             rsq22            = gmx_fjsp_calc_rsq_v2r8(dx22,dy22,dz22);
 616             rsq23            = gmx_fjsp_calc_rsq_v2r8(dx23,dy23,dz23);
 617             rsq31            = gmx_fjsp_calc_rsq_v2r8(dx31,dy31,dz31);
 618             rsq32            = gmx_fjsp_calc_rsq_v2r8(dx32,dy32,dz32);
 619             rsq33            = gmx_fjsp_calc_rsq_v2r8(dx33,dy33,dz33);
 620
 621             rinv00           = gmx_fjsp_invsqrt_v2r8(rsq00);
 622             rinv11           = gmx_fjsp_invsqrt_v2r8(rsq11);
 623             rinv12           = gmx_fjsp_invsqrt_v2r8(rsq12);
 624             rinv13           = gmx_fjsp_invsqrt_v2r8(rsq13);
 625             rinv21           = gmx_fjsp_invsqrt_v2r8(rsq21);
 626             rinv22           = gmx_fjsp_invsqrt_v2r8(rsq22);
 627             rinv23           = gmx_fjsp_invsqrt_v2r8(rsq23);
 628             rinv31           = gmx_fjsp_invsqrt_v2r8(rsq31);
 629             rinv32           = gmx_fjsp_invsqrt_v2r8(rsq32);
 630             rinv33           = gmx_fjsp_invsqrt_v2r8(rsq33);
 631
 632             rinvsq11         = _fjsp_mul_v2r8(rinv11,rinv11);
 633             rinvsq12         = _fjsp_mul_v2r8(rinv12,rinv12);
 634             rinvsq13         = _fjsp_mul_v2r8(rinv13,rinv13);
 635             rinvsq21         = _fjsp_mul_v2r8(rinv21,rinv21);
 636             rinvsq22         = _fjsp_mul_v2r8(rinv22,rinv22);
 637             rinvsq23         = _fjsp_mul_v2r8(rinv23,rinv23);
 638             rinvsq31         = _fjsp_mul_v2r8(rinv31,rinv31);
 639             rinvsq32         = _fjsp_mul_v2r8(rinv32,rinv32);
 640             rinvsq33         = _fjsp_mul_v2r8(rinv33,rinv33);
 641
 642             fjx0             = _fjsp_setzero_v2r8();
 643             fjy0             = _fjsp_setzero_v2r8();
 644             fjz0             = _fjsp_setzero_v2r8();
 645             fjx1             = _fjsp_setzero_v2r8();
 646             fjy1             = _fjsp_setzero_v2r8();
 647             fjz1             = _fjsp_setzero_v2r8();
 648             fjx2             = _fjsp_setzero_v2r8();
 649             fjy2             = _fjsp_setzero_v2r8();
 650             fjz2             = _fjsp_setzero_v2r8();
 651             fjx3             = _fjsp_setzero_v2r8();
 652             fjy3             = _fjsp_setzero_v2r8();
 653             fjz3             = _fjsp_setzero_v2r8();
 654
 655             /**************************
 656              * CALCULATE INTERACTIONS *
 657              **************************/
 658
 659             r00              = _fjsp_mul_v2r8(rsq00,rinv00);
 660
 661             /* Calculate table index by multiplying r with table scale and truncate to integer */
 662             rt               = _fjsp_mul_v2r8(r00,vftabscale);
 663             itab_tmp         = _fjsp_dtox_v2r8(rt);
 664             vfeps            = _fjsp_sub_v2r8(rt, _fjsp_xtod_v2r8(itab_tmp));
 665             twovfeps         = _fjsp_add_v2r8(vfeps,vfeps);
 666             _fjsp_store_v2r8(&vfconv.simd,itab_tmp);
 667
 668             vfconv.i[0]     *= 8;
 669             vfconv.i[1]     *= 8;
 670
 671             /* CUBIC SPLINE TABLE DISPERSION */
 672             Y                = _fjsp_load_v2r8( vftab + vfconv.i[0] );
 673             F                = _fjsp_setzero_v2r8();
 674             GMX_FJSP_TRANSPOSE2_V2R8(Y,F);
 675             G                = _fjsp_load_v2r8( vftab + vfconv.i[0] + 2 );
 676             H                = _fjsp_setzero_v2r8();
 677             GMX_FJSP_TRANSPOSE2_V2R8(G,H);
 678             Fp               = _fjsp_madd_v2r8(vfeps,_fjsp_madd_v2r8(H,vfeps,G),F);
 679             VV               = _fjsp_madd_v2r8(vfeps,Fp,Y);
 680             vvdw6            = _fjsp_mul_v2r8(c6_00,VV);
 681             FF               = _fjsp_madd_v2r8(vfeps,_fjsp_madd_v2r8(twovfeps,H,G),Fp);
 682             fvdw6            = _fjsp_mul_v2r8(c6_00,FF);
 683
 684             /* CUBIC SPLINE TABLE REPULSION */
 685             Y                = _fjsp_load_v2r8( vftab + vfconv.i[0] + 4 );
 686             F                = _fjsp_setzero_v2r8();
 687             GMX_FJSP_TRANSPOSE2_V2R8(Y,F);
 688             G                = _fjsp_load_v2r8( vftab + vfconv.i[0] + 6 );
 689             H                = _fjsp_setzero_v2r8();
 690             GMX_FJSP_TRANSPOSE2_V2R8(G,H);
 691             Fp               = _fjsp_madd_v2r8(vfeps,_fjsp_madd_v2r8(H,vfeps,G),F);
 692             VV               = _fjsp_madd_v2r8(vfeps,Fp,Y);
 693             vvdw12           = _fjsp_mul_v2r8(c12_00,VV);
 694             FF               = _fjsp_madd_v2r8(vfeps,_fjsp_madd_v2r8(twovfeps,H,G),Fp);
 695             fvdw12           = _fjsp_mul_v2r8(c12_00,FF);
 696             vvdw             = _fjsp_add_v2r8(vvdw12,vvdw6);
 697             fvdw             = _fjsp_neg_v2r8(_fjsp_mul_v2r8(_fjsp_add_v2r8(fvdw6,fvdw12),_fjsp_mul_v2r8(vftabscale,rinv00)));
 698
 699             /* Update potential sum for this i atom from the interaction with this j atom. */
 700             vvdw             = _fjsp_unpacklo_v2r8(vvdw,_fjsp_setzero_v2r8());
 701             vvdwsum          = _fjsp_add_v2r8(vvdwsum,vvdw);
 702
 703             fscal            = fvdw;
 704
 705             fscal            = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
 706
 707             /* Update vectorial force */
 708             fix0             = _fjsp_madd_v2r8(dx00,fscal,fix0);
 709             fiy0             = _fjsp_madd_v2r8(dy00,fscal,fiy0);
 710             fiz0             = _fjsp_madd_v2r8(dz00,fscal,fiz0);
 711
 712             fjx0             = _fjsp_madd_v2r8(dx00,fscal,fjx0);
 713             fjy0             = _fjsp_madd_v2r8(dy00,fscal,fjy0);
 714             fjz0             = _fjsp_madd_v2r8(dz00,fscal,fjz0);
 715
 716             /**************************
 717              * CALCULATE INTERACTIONS *
 718              **************************/
 719
 720             /* REACTION-FIELD ELECTROSTATICS */
 721             velec            = _fjsp_mul_v2r8(qq11,_fjsp_sub_v2r8(_fjsp_madd_v2r8(krf,rsq11,rinv11),crf));
 722             felec            = _fjsp_mul_v2r8(qq11,_fjsp_msub_v2r8(rinv11,rinvsq11,krf2));
 723
 724             /* Update potential sum for this i atom from the interaction with this j atom. */
 725             velec            = _fjsp_unpacklo_v2r8(velec,_fjsp_setzero_v2r8());
 726             velecsum         = _fjsp_add_v2r8(velecsum,velec);
 727
 728             fscal            = felec;
 729
 730             fscal            = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
 731
 732             /* Update vectorial force */
 733             fix1             = _fjsp_madd_v2r8(dx11,fscal,fix1);
 734             fiy1             = _fjsp_madd_v2r8(dy11,fscal,fiy1);
 735             fiz1             = _fjsp_madd_v2r8(dz11,fscal,fiz1);
 736
 737             fjx1             = _fjsp_madd_v2r8(dx11,fscal,fjx1);
 738             fjy1             = _fjsp_madd_v2r8(dy11,fscal,fjy1);
 739             fjz1             = _fjsp_madd_v2r8(dz11,fscal,fjz1);
 740
 741             /**************************
 742              * CALCULATE INTERACTIONS *
 743              **************************/
 744
 745             /* REACTION-FIELD ELECTROSTATICS */
 746             velec            = _fjsp_mul_v2r8(qq12,_fjsp_sub_v2r8(_fjsp_madd_v2r8(krf,rsq12,rinv12),crf));
 747             felec            = _fjsp_mul_v2r8(qq12,_fjsp_msub_v2r8(rinv12,rinvsq12,krf2));
 748
 749             /* Update potential sum for this i atom from the interaction with this j atom. */
 750             velec            = _fjsp_unpacklo_v2r8(velec,_fjsp_setzero_v2r8());
 751             velecsum         = _fjsp_add_v2r8(velecsum,velec);
 752
 753             fscal            = felec;
 754
 755             fscal            = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
 756
 757             /* Update vectorial force */
 758             fix1             = _fjsp_madd_v2r8(dx12,fscal,fix1);
 759             fiy1             = _fjsp_madd_v2r8(dy12,fscal,fiy1);
 760             fiz1             = _fjsp_madd_v2r8(dz12,fscal,fiz1);
 761
 762             fjx2             = _fjsp_madd_v2r8(dx12,fscal,fjx2);
 763             fjy2             = _fjsp_madd_v2r8(dy12,fscal,fjy2);
 764             fjz2             = _fjsp_madd_v2r8(dz12,fscal,fjz2);
 765
 766             /**************************
 767              * CALCULATE INTERACTIONS *
 768              **************************/
 769
 770             /* REACTION-FIELD ELECTROSTATICS */
 771             velec            = _fjsp_mul_v2r8(qq13,_fjsp_sub_v2r8(_fjsp_madd_v2r8(krf,rsq13,rinv13),crf));
 772             felec            = _fjsp_mul_v2r8(qq13,_fjsp_msub_v2r8(rinv13,rinvsq13,krf2));
 773
 774             /* Update potential sum for this i atom from the interaction with this j atom. */
 775             velec            = _fjsp_unpacklo_v2r8(velec,_fjsp_setzero_v2r8());
 776             velecsum         = _fjsp_add_v2r8(velecsum,velec);
 777
 778             fscal            = felec;
 779
 780             fscal            = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
 781
 782             /* Update vectorial force */
 783             fix1             = _fjsp_madd_v2r8(dx13,fscal,fix1);
 784             fiy1             = _fjsp_madd_v2r8(dy13,fscal,fiy1);
 785             fiz1             = _fjsp_madd_v2r8(dz13,fscal,fiz1);
 786
 787             fjx3             = _fjsp_madd_v2r8(dx13,fscal,fjx3);
 788             fjy3             = _fjsp_madd_v2r8(dy13,fscal,fjy3);
 789             fjz3             = _fjsp_madd_v2r8(dz13,fscal,fjz3);
 790
 791             /**************************
 792              * CALCULATE INTERACTIONS *
 793              **************************/
 794
 795             /* REACTION-FIELD ELECTROSTATICS */
 796             velec            = _fjsp_mul_v2r8(qq21,_fjsp_sub_v2r8(_fjsp_madd_v2r8(krf,rsq21,rinv21),crf));
 797             felec            = _fjsp_mul_v2r8(qq21,_fjsp_msub_v2r8(rinv21,rinvsq21,krf2));
 798
 799             /* Update potential sum for this i atom from the interaction with this j atom. */
 800             velec            = _fjsp_unpacklo_v2r8(velec,_fjsp_setzero_v2r8());
 801             velecsum         = _fjsp_add_v2r8(velecsum,velec);
 802
 803             fscal            = felec;
 804
 805             fscal            = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
 806
 807             /* Update vectorial force */
 808             fix2             = _fjsp_madd_v2r8(dx21,fscal,fix2);
 809             fiy2             = _fjsp_madd_v2r8(dy21,fscal,fiy2);
 810             fiz2             = _fjsp_madd_v2r8(dz21,fscal,fiz2);
 811
 812             fjx1             = _fjsp_madd_v2r8(dx21,fscal,fjx1);
 813             fjy1             = _fjsp_madd_v2r8(dy21,fscal,fjy1);
 814             fjz1             = _fjsp_madd_v2r8(dz21,fscal,fjz1);
 815
 816             /**************************
 817              * CALCULATE INTERACTIONS *
 818              **************************/
 819
 820             /* REACTION-FIELD ELECTROSTATICS */
 821             velec            = _fjsp_mul_v2r8(qq22,_fjsp_sub_v2r8(_fjsp_madd_v2r8(krf,rsq22,rinv22),crf));
 822             felec            = _fjsp_mul_v2r8(qq22,_fjsp_msub_v2r8(rinv22,rinvsq22,krf2));
 823
 824             /* Update potential sum for this i atom from the interaction with this j atom. */
 825             velec            = _fjsp_unpacklo_v2r8(velec,_fjsp_setzero_v2r8());
 826             velecsum         = _fjsp_add_v2r8(velecsum,velec);
 827
 828             fscal            = felec;
 829
 830             fscal            = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
 831
 832             /* Update vectorial force */
 833             fix2             = _fjsp_madd_v2r8(dx22,fscal,fix2);
 834             fiy2             = _fjsp_madd_v2r8(dy22,fscal,fiy2);
 835             fiz2             = _fjsp_madd_v2r8(dz22,fscal,fiz2);
 836
 837             fjx2             = _fjsp_madd_v2r8(dx22,fscal,fjx2);
 838             fjy2             = _fjsp_madd_v2r8(dy22,fscal,fjy2);
 839             fjz2             = _fjsp_madd_v2r8(dz22,fscal,fjz2);
 840
 841             /**************************
 842              * CALCULATE INTERACTIONS *
 843              **************************/
 844
 845             /* REACTION-FIELD ELECTROSTATICS */
 846             velec            = _fjsp_mul_v2r8(qq23,_fjsp_sub_v2r8(_fjsp_madd_v2r8(krf,rsq23,rinv23),crf));
 847             felec            = _fjsp_mul_v2r8(qq23,_fjsp_msub_v2r8(rinv23,rinvsq23,krf2));
 848
 849             /* Update potential sum for this i atom from the interaction with this j atom. */
 850             velec            = _fjsp_unpacklo_v2r8(velec,_fjsp_setzero_v2r8());
 851             velecsum         = _fjsp_add_v2r8(velecsum,velec);
 852
 853             fscal            = felec;
 854
 855             fscal            = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
 856
 857             /* Update vectorial force */
 858             fix2             = _fjsp_madd_v2r8(dx23,fscal,fix2);
 859             fiy2             = _fjsp_madd_v2r8(dy23,fscal,fiy2);
 860             fiz2             = _fjsp_madd_v2r8(dz23,fscal,fiz2);
 861
 862             fjx3             = _fjsp_madd_v2r8(dx23,fscal,fjx3);
 863             fjy3             = _fjsp_madd_v2r8(dy23,fscal,fjy3);
 864             fjz3             = _fjsp_madd_v2r8(dz23,fscal,fjz3);
 865
 866             /**************************
 867              * CALCULATE INTERACTIONS *
 868              **************************/
 869
 870             /* REACTION-FIELD ELECTROSTATICS */
 871             velec            = _fjsp_mul_v2r8(qq31,_fjsp_sub_v2r8(_fjsp_madd_v2r8(krf,rsq31,rinv31),crf));
 872             felec            = _fjsp_mul_v2r8(qq31,_fjsp_msub_v2r8(rinv31,rinvsq31,krf2));
 873
 874             /* Update potential sum for this i atom from the interaction with this j atom. */
 875             velec            = _fjsp_unpacklo_v2r8(velec,_fjsp_setzero_v2r8());
 876             velecsum         = _fjsp_add_v2r8(velecsum,velec);
 877
 878             fscal            = felec;
 879
 880             fscal            = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
 881
 882             /* Update vectorial force */
 883             fix3             = _fjsp_madd_v2r8(dx31,fscal,fix3);
 884             fiy3             = _fjsp_madd_v2r8(dy31,fscal,fiy3);
 885             fiz3             = _fjsp_madd_v2r8(dz31,fscal,fiz3);
 886
 887             fjx1             = _fjsp_madd_v2r8(dx31,fscal,fjx1);
 888             fjy1             = _fjsp_madd_v2r8(dy31,fscal,fjy1);
 889             fjz1             = _fjsp_madd_v2r8(dz31,fscal,fjz1);
 890
 891             /**************************
 892              * CALCULATE INTERACTIONS *
 893              **************************/
 894
 895             /* REACTION-FIELD ELECTROSTATICS */
 896             velec            = _fjsp_mul_v2r8(qq32,_fjsp_sub_v2r8(_fjsp_madd_v2r8(krf,rsq32,rinv32),crf));
 897             felec            = _fjsp_mul_v2r8(qq32,_fjsp_msub_v2r8(rinv32,rinvsq32,krf2));
 898
 899             /* Update potential sum for this i atom from the interaction with this j atom. */
 900             velec            = _fjsp_unpacklo_v2r8(velec,_fjsp_setzero_v2r8());
 901             velecsum         = _fjsp_add_v2r8(velecsum,velec);
 902
 903             fscal            = felec;
 904
 905             fscal            = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
 906
 907             /* Update vectorial force */
 908             fix3             = _fjsp_madd_v2r8(dx32,fscal,fix3);
 909             fiy3             = _fjsp_madd_v2r8(dy32,fscal,fiy3);
 910             fiz3             = _fjsp_madd_v2r8(dz32,fscal,fiz3);
 911
 912             fjx2             = _fjsp_madd_v2r8(dx32,fscal,fjx2);
 913             fjy2             = _fjsp_madd_v2r8(dy32,fscal,fjy2);
 914             fjz2             = _fjsp_madd_v2r8(dz32,fscal,fjz2);
 915
 916             /**************************
 917              * CALCULATE INTERACTIONS *
 918              **************************/
 919
 920             /* REACTION-FIELD ELECTROSTATICS */
 921             velec            = _fjsp_mul_v2r8(qq33,_fjsp_sub_v2r8(_fjsp_madd_v2r8(krf,rsq33,rinv33),crf));
 922             felec            = _fjsp_mul_v2r8(qq33,_fjsp_msub_v2r8(rinv33,rinvsq33,krf2));
 923
 924             /* Update potential sum for this i atom from the interaction with this j atom. */
 925             velec            = _fjsp_unpacklo_v2r8(velec,_fjsp_setzero_v2r8());
 926             velecsum         = _fjsp_add_v2r8(velecsum,velec);
 927
 928             fscal            = felec;
 929
 930             fscal            = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
 931
 932             /* Update vectorial force */
 933             fix3             = _fjsp_madd_v2r8(dx33,fscal,fix3);
 934             fiy3             = _fjsp_madd_v2r8(dy33,fscal,fiy3);
 935             fiz3             = _fjsp_madd_v2r8(dz33,fscal,fiz3);
 936
 937             fjx3             = _fjsp_madd_v2r8(dx33,fscal,fjx3);
 938             fjy3             = _fjsp_madd_v2r8(dy33,fscal,fjy3);
 939             fjz3             = _fjsp_madd_v2r8(dz33,fscal,fjz3);
 940
 941             gmx_fjsp_decrement_4rvec_1ptr_swizzle_v2r8(f+j_coord_offsetA,fjx0,fjy0,fjz0,fjx1,fjy1,fjz1,fjx2,fjy2,fjz2,fjx3,fjy3,fjz3);
 942
 943             /* Inner loop uses 377 flops */
 944         }
 945
 946         /* End of innermost loop */
 947
 948         gmx_fjsp_update_iforce_4atom_swizzle_v2r8(fix0,fiy0,fiz0,fix1,fiy1,fiz1,fix2,fiy2,fiz2,fix3,fiy3,fiz3,
 949                                               f+i_coord_offset,fshift+i_shift_offset);
 950
 951         ggid                        = gid[iidx];
 952         /* Update potential energies */
 953         gmx_fjsp_update_1pot_v2r8(velecsum,kernel_data->energygrp_elec+ggid);
 954         gmx_fjsp_update_1pot_v2r8(vvdwsum,kernel_data->energygrp_vdw+ggid);
 955
 956         /* Increment number of inner iterations */
 957         inneriter                  += j_index_end - j_index_start;
 958
 959         /* Outer loop uses 26 flops */
 960     }
 961
 962     /* Increment number of outer iterations */
 963     outeriter        += nri;
 964
 965     /* Update outer/inner flops */
 966
 967     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W4W4_VF,outeriter*26 + inneriter*377);
 968 }
 969 /*
 970  * Gromacs nonbonded kernel:   nb_kernel_ElecRF_VdwCSTab_GeomW4W4_F_sparc64_hpc_ace_double
 971  * Electrostatics interaction: ReactionField
 972  * VdW interaction:            CubicSplineTable
 973  * Geometry:                   Water4-Water4
 974  * Calculate force/pot:        Force
 975  */
 976 void
 977 nb_kernel_ElecRF_VdwCSTab_GeomW4W4_F_sparc64_hpc_ace_double
 978                     (t_nblist                    * gmx_restrict       nlist,
 979                      rvec                        * gmx_restrict          xx,
 980                      rvec                        * gmx_restrict          ff,
 981                      t_forcerec                  * gmx_restrict          fr,
 982                      t_mdatoms                   * gmx_restrict     mdatoms,
 983                      nb_kernel_data_t gmx_unused * gmx_restrict kernel_data,
 984                      t_nrnb                      * gmx_restrict        nrnb)
 985 {
 986     /* Suffixes 0,1,2,3 refer to particle indices for waters in the inner or outer loop, or
 987      * just 0 for non-waters.
 988      * Suffixes A,B refer to j loop unrolling done with double precision SIMD, e.g. for the two different
 989      * jnr indices corresponding to data put in the four positions in the SIMD register.
 990      */
 991     int              i_shift_offset,i_coord_offset,outeriter,inneriter;
 992     int              j_index_start,j_index_end,jidx,nri,inr,ggid,iidx;
 993     int              jnrA,jnrB;
 994     int              j_coord_offsetA,j_coord_offsetB;
 995     int              *iinr,*jindex,*jjnr,*shiftidx,*gid;
 996     real             rcutoff_scalar;
 997     real             *shiftvec,*fshift,*x,*f;
 998     _fjsp_v2r8       tx,ty,tz,fscal,rcutoff,rcutoff2,jidxall;
 999     int              vdwioffset0;
1000     _fjsp_v2r8       ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
1001     int              vdwioffset1;
1002     _fjsp_v2r8       ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
1003     int              vdwioffset2;
1004     _fjsp_v2r8       ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
1005     int              vdwioffset3;
1006     _fjsp_v2r8       ix3,iy3,iz3,fix3,fiy3,fiz3,iq3,isai3;
1007     int              vdwjidx0A,vdwjidx0B;
1008     _fjsp_v2r8       jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
1009     int              vdwjidx1A,vdwjidx1B;
1010     _fjsp_v2r8       jx1,jy1,jz1,fjx1,fjy1,fjz1,jq1,isaj1;
1011     int              vdwjidx2A,vdwjidx2B;
1012     _fjsp_v2r8       jx2,jy2,jz2,fjx2,fjy2,fjz2,jq2,isaj2;
1013     int              vdwjidx3A,vdwjidx3B;
1014     _fjsp_v2r8       jx3,jy3,jz3,fjx3,fjy3,fjz3,jq3,isaj3;
1015     _fjsp_v2r8       dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00;
1016     _fjsp_v2r8       dx11,dy11,dz11,rsq11,rinv11,rinvsq11,r11,qq11,c6_11,c12_11;
1017     _fjsp_v2r8       dx12,dy12,dz12,rsq12,rinv12,rinvsq12,r12,qq12,c6_12,c12_12;
1018     _fjsp_v2r8       dx13,dy13,dz13,rsq13,rinv13,rinvsq13,r13,qq13,c6_13,c12_13;
1019     _fjsp_v2r8       dx21,dy21,dz21,rsq21,rinv21,rinvsq21,r21,qq21,c6_21,c12_21;
1020     _fjsp_v2r8       dx22,dy22,dz22,rsq22,rinv22,rinvsq22,r22,qq22,c6_22,c12_22;
1021     _fjsp_v2r8       dx23,dy23,dz23,rsq23,rinv23,rinvsq23,r23,qq23,c6_23,c12_23;
1022     _fjsp_v2r8       dx31,dy31,dz31,rsq31,rinv31,rinvsq31,r31,qq31,c6_31,c12_31;
1023     _fjsp_v2r8       dx32,dy32,dz32,rsq32,rinv32,rinvsq32,r32,qq32,c6_32,c12_32;
1024     _fjsp_v2r8       dx33,dy33,dz33,rsq33,rinv33,rinvsq33,r33,qq33,c6_33,c12_33;
1025     _fjsp_v2r8       velec,felec,velecsum,facel,crf,krf,krf2;
1026     real             *charge;
1027     int              nvdwtype;
1028     _fjsp_v2r8       rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,sh_vdw_invrcut6;
1029     int              *vdwtype;
1030     real             *vdwparam;
1031     _fjsp_v2r8       one_sixth   = gmx_fjsp_set1_v2r8(1.0/6.0);
1032     _fjsp_v2r8       one_twelfth = gmx_fjsp_set1_v2r8(1.0/12.0);
1033     _fjsp_v2r8       rt,vfeps,vftabscale,Y,F,G,H,Heps,Fp,VV,FF,twovfeps;
1034     real             *vftab;
1035     _fjsp_v2r8       itab_tmp;
1036     _fjsp_v2r8       dummy_mask,cutoff_mask;
1037     _fjsp_v2r8       one     = gmx_fjsp_set1_v2r8(1.0);
1038     _fjsp_v2r8       two     = gmx_fjsp_set1_v2r8(2.0);
1039     union { _fjsp_v2r8 simd; long long int i[2]; } vfconv,gbconv,ewconv;
1040
1041     x                = xx[0];
1042     f                = ff[0];
1043
1044     nri              = nlist->nri;
1045     iinr             = nlist->iinr;
1046     jindex           = nlist->jindex;
1047     jjnr             = nlist->jjnr;
1048     shiftidx         = nlist->shift;
1049     gid              = nlist->gid;
1050     shiftvec         = fr->shift_vec[0];
1051     fshift           = fr->fshift[0];
1052     facel            = gmx_fjsp_set1_v2r8(fr->epsfac);
1053     charge           = mdatoms->chargeA;
1054     krf              = gmx_fjsp_set1_v2r8(fr->ic->k_rf);
1055     krf2             = gmx_fjsp_set1_v2r8(fr->ic->k_rf*2.0);
1056     crf              = gmx_fjsp_set1_v2r8(fr->ic->c_rf);
1057     nvdwtype         = fr->ntype;
1058     vdwparam         = fr->nbfp;
1059     vdwtype          = mdatoms->typeA;
1060
1061     vftab            = kernel_data->table_vdw->data;
1062     vftabscale       = gmx_fjsp_set1_v2r8(kernel_data->table_vdw->scale);
1063
1064     /* Setup water-specific parameters */
1065     inr              = nlist->iinr[0];
1066     iq1              = _fjsp_mul_v2r8(facel,gmx_fjsp_set1_v2r8(charge[inr+1]));
1067     iq2              = _fjsp_mul_v2r8(facel,gmx_fjsp_set1_v2r8(charge[inr+2]));
1068     iq3              = _fjsp_mul_v2r8(facel,gmx_fjsp_set1_v2r8(charge[inr+3]));
1069     vdwioffset0      = 2*nvdwtype*vdwtype[inr+0];
1070
1071     jq1              = gmx_fjsp_set1_v2r8(charge[inr+1]);
1072     jq2              = gmx_fjsp_set1_v2r8(charge[inr+2]);
1073     jq3              = gmx_fjsp_set1_v2r8(charge[inr+3]);
1074     vdwjidx0A        = 2*vdwtype[inr+0];
1075     c6_00            = gmx_fjsp_set1_v2r8(vdwparam[vdwioffset0+vdwjidx0A]);
1076     c12_00           = gmx_fjsp_set1_v2r8(vdwparam[vdwioffset0+vdwjidx0A+1]);
1077     qq11             = _fjsp_mul_v2r8(iq1,jq1);
1078     qq12             = _fjsp_mul_v2r8(iq1,jq2);
1079     qq13             = _fjsp_mul_v2r8(iq1,jq3);
1080     qq21             = _fjsp_mul_v2r8(iq2,jq1);
1081     qq22             = _fjsp_mul_v2r8(iq2,jq2);
1082     qq23             = _fjsp_mul_v2r8(iq2,jq3);
1083     qq31             = _fjsp_mul_v2r8(iq3,jq1);
1084     qq32             = _fjsp_mul_v2r8(iq3,jq2);
1085     qq33             = _fjsp_mul_v2r8(iq3,jq3);
1086
1087     /* Avoid stupid compiler warnings */
1088     jnrA = jnrB = 0;
1089     j_coord_offsetA = 0;
1090     j_coord_offsetB = 0;
1091
1092     outeriter        = 0;
1093     inneriter        = 0;
1094
1095     /* Start outer loop over neighborlists */
1096     for(iidx=0; iidx<nri; iidx++)
1097     {
1098         /* Load shift vector for this list */
1099         i_shift_offset   = DIM*shiftidx[iidx];
1100
1101         /* Load limits for loop over neighbors */
1102         j_index_start    = jindex[iidx];
1103         j_index_end      = jindex[iidx+1];
1104
1105         /* Get outer coordinate index */
1106         inr              = iinr[iidx];
1107         i_coord_offset   = DIM*inr;
1108
1109         /* Load i particle coords and add shift vector */
1110         gmx_fjsp_load_shift_and_4rvec_broadcast_v2r8(shiftvec+i_shift_offset,x+i_coord_offset,
1111                                                  &ix0,&iy0,&iz0,&ix1,&iy1,&iz1,&ix2,&iy2,&iz2,&ix3,&iy3,&iz3);
1112
1113         fix0             = _fjsp_setzero_v2r8();
1114         fiy0             = _fjsp_setzero_v2r8();
1115         fiz0             = _fjsp_setzero_v2r8();
1116         fix1             = _fjsp_setzero_v2r8();
1117         fiy1             = _fjsp_setzero_v2r8();
1118         fiz1             = _fjsp_setzero_v2r8();
1119         fix2             = _fjsp_setzero_v2r8();
1120         fiy2             = _fjsp_setzero_v2r8();
1121         fiz2             = _fjsp_setzero_v2r8();
1122         fix3             = _fjsp_setzero_v2r8();
1123         fiy3             = _fjsp_setzero_v2r8();
1124         fiz3             = _fjsp_setzero_v2r8();
1125
1126         /* Start inner kernel loop */
1127         for(jidx=j_index_start; jidx<j_index_end-1; jidx+=2)
1128         {
1129
1130             /* Get j neighbor index, and coordinate index */
1131             jnrA             = jjnr[jidx];
1132             jnrB             = jjnr[jidx+1];
1133             j_coord_offsetA  = DIM*jnrA;
1134             j_coord_offsetB  = DIM*jnrB;
1135
1136             /* load j atom coordinates */
1137             gmx_fjsp_load_4rvec_2ptr_swizzle_v2r8(x+j_coord_offsetA,x+j_coord_offsetB,
1138                                               &jx0,&jy0,&jz0,&jx1,&jy1,&jz1,&jx2,
1139                                               &jy2,&jz2,&jx3,&jy3,&jz3);
1140
1141             /* Calculate displacement vector */
1142             dx00             = _fjsp_sub_v2r8(ix0,jx0);
1143             dy00             = _fjsp_sub_v2r8(iy0,jy0);
1144             dz00             = _fjsp_sub_v2r8(iz0,jz0);
1145             dx11             = _fjsp_sub_v2r8(ix1,jx1);
1146             dy11             = _fjsp_sub_v2r8(iy1,jy1);
1147             dz11             = _fjsp_sub_v2r8(iz1,jz1);
1148             dx12             = _fjsp_sub_v2r8(ix1,jx2);
1149             dy12             = _fjsp_sub_v2r8(iy1,jy2);
1150             dz12             = _fjsp_sub_v2r8(iz1,jz2);
1151             dx13             = _fjsp_sub_v2r8(ix1,jx3);
1152             dy13             = _fjsp_sub_v2r8(iy1,jy3);
1153             dz13             = _fjsp_sub_v2r8(iz1,jz3);
1154             dx21             = _fjsp_sub_v2r8(ix2,jx1);
1155             dy21             = _fjsp_sub_v2r8(iy2,jy1);
1156             dz21             = _fjsp_sub_v2r8(iz2,jz1);
1157             dx22             = _fjsp_sub_v2r8(ix2,jx2);
1158             dy22             = _fjsp_sub_v2r8(iy2,jy2);
1159             dz22             = _fjsp_sub_v2r8(iz2,jz2);
1160             dx23             = _fjsp_sub_v2r8(ix2,jx3);
1161             dy23             = _fjsp_sub_v2r8(iy2,jy3);
1162             dz23             = _fjsp_sub_v2r8(iz2,jz3);
1163             dx31             = _fjsp_sub_v2r8(ix3,jx1);
1164             dy31             = _fjsp_sub_v2r8(iy3,jy1);
1165             dz31             = _fjsp_sub_v2r8(iz3,jz1);
1166             dx32             = _fjsp_sub_v2r8(ix3,jx2);
1167             dy32             = _fjsp_sub_v2r8(iy3,jy2);
1168             dz32             = _fjsp_sub_v2r8(iz3,jz2);
1169             dx33             = _fjsp_sub_v2r8(ix3,jx3);
1170             dy33             = _fjsp_sub_v2r8(iy3,jy3);
1171             dz33             = _fjsp_sub_v2r8(iz3,jz3);
1172
1173             /* Calculate squared distance and things based on it */
1174             rsq00            = gmx_fjsp_calc_rsq_v2r8(dx00,dy00,dz00);
1175             rsq11            = gmx_fjsp_calc_rsq_v2r8(dx11,dy11,dz11);
1176             rsq12            = gmx_fjsp_calc_rsq_v2r8(dx12,dy12,dz12);
1177             rsq13            = gmx_fjsp_calc_rsq_v2r8(dx13,dy13,dz13);
1178             rsq21            = gmx_fjsp_calc_rsq_v2r8(dx21,dy21,dz21);
1179             rsq22            = gmx_fjsp_calc_rsq_v2r8(dx22,dy22,dz22);
1180             rsq23            = gmx_fjsp_calc_rsq_v2r8(dx23,dy23,dz23);
1181             rsq31            = gmx_fjsp_calc_rsq_v2r8(dx31,dy31,dz31);
1182             rsq32            = gmx_fjsp_calc_rsq_v2r8(dx32,dy32,dz32);
1183             rsq33            = gmx_fjsp_calc_rsq_v2r8(dx33,dy33,dz33);
1184
1185             rinv00           = gmx_fjsp_invsqrt_v2r8(rsq00);
1186             rinv11           = gmx_fjsp_invsqrt_v2r8(rsq11);
1187             rinv12           = gmx_fjsp_invsqrt_v2r8(rsq12);
1188             rinv13           = gmx_fjsp_invsqrt_v2r8(rsq13);
1189             rinv21           = gmx_fjsp_invsqrt_v2r8(rsq21);
1190             rinv22           = gmx_fjsp_invsqrt_v2r8(rsq22);
1191             rinv23           = gmx_fjsp_invsqrt_v2r8(rsq23);
1192             rinv31           = gmx_fjsp_invsqrt_v2r8(rsq31);
1193             rinv32           = gmx_fjsp_invsqrt_v2r8(rsq32);
1194             rinv33           = gmx_fjsp_invsqrt_v2r8(rsq33);
1195
1196             rinvsq11         = _fjsp_mul_v2r8(rinv11,rinv11);
1197             rinvsq12         = _fjsp_mul_v2r8(rinv12,rinv12);
1198             rinvsq13         = _fjsp_mul_v2r8(rinv13,rinv13);
1199             rinvsq21         = _fjsp_mul_v2r8(rinv21,rinv21);
1200             rinvsq22         = _fjsp_mul_v2r8(rinv22,rinv22);
1201             rinvsq23         = _fjsp_mul_v2r8(rinv23,rinv23);
1202             rinvsq31         = _fjsp_mul_v2r8(rinv31,rinv31);
1203             rinvsq32         = _fjsp_mul_v2r8(rinv32,rinv32);
1204             rinvsq33         = _fjsp_mul_v2r8(rinv33,rinv33);
1205
1206             fjx0             = _fjsp_setzero_v2r8();
1207             fjy0             = _fjsp_setzero_v2r8();
1208             fjz0             = _fjsp_setzero_v2r8();
1209             fjx1             = _fjsp_setzero_v2r8();
1210             fjy1             = _fjsp_setzero_v2r8();
1211             fjz1             = _fjsp_setzero_v2r8();
1212             fjx2             = _fjsp_setzero_v2r8();
1213             fjy2             = _fjsp_setzero_v2r8();
1214             fjz2             = _fjsp_setzero_v2r8();
1215             fjx3             = _fjsp_setzero_v2r8();
1216             fjy3             = _fjsp_setzero_v2r8();
1217             fjz3             = _fjsp_setzero_v2r8();
1218
1219             /**************************
1220              * CALCULATE INTERACTIONS *
1221              **************************/
1222
1223             r00              = _fjsp_mul_v2r8(rsq00,rinv00);
1224
1225             /* Calculate table index by multiplying r with table scale and truncate to integer */
1226             rt               = _fjsp_mul_v2r8(r00,vftabscale);
1227             itab_tmp         = _fjsp_dtox_v2r8(rt);
1228             vfeps            = _fjsp_sub_v2r8(rt, _fjsp_xtod_v2r8(itab_tmp));
1229             twovfeps         = _fjsp_add_v2r8(vfeps,vfeps);
1230             _fjsp_store_v2r8(&vfconv.simd,itab_tmp);
1231
1232             vfconv.i[0]     *= 8;
1233             vfconv.i[1]     *= 8;
1234
1235             /* CUBIC SPLINE TABLE DISPERSION */
1236             Y                = _fjsp_load_v2r8( vftab + vfconv.i[0] );
1237             F                = _fjsp_load_v2r8( vftab + vfconv.i[1] );
1238             GMX_FJSP_TRANSPOSE2_V2R8(Y,F);
1239             G                = _fjsp_load_v2r8( vftab + vfconv.i[0] + 2 );
1240             H                = _fjsp_load_v2r8( vftab + vfconv.i[1] + 2 );
1241             GMX_FJSP_TRANSPOSE2_V2R8(G,H);
1242             Fp               = _fjsp_madd_v2r8(vfeps,_fjsp_madd_v2r8(H,vfeps,G),F);
1243             FF               = _fjsp_madd_v2r8(vfeps,_fjsp_madd_v2r8(twovfeps,H,G),Fp);
1244             fvdw6            = _fjsp_mul_v2r8(c6_00,FF);
1245
1246             /* CUBIC SPLINE TABLE REPULSION */
1247             Y                = _fjsp_load_v2r8( vftab + vfconv.i[0] + 4 );
1248             F                = _fjsp_load_v2r8( vftab + vfconv.i[1] + 4 );
1249             GMX_FJSP_TRANSPOSE2_V2R8(Y,F);
1250             G                = _fjsp_load_v2r8( vftab + vfconv.i[0] + 6 );
1251             H                = _fjsp_load_v2r8( vftab + vfconv.i[1] + 6 );
1252             GMX_FJSP_TRANSPOSE2_V2R8(G,H);
1253             Fp               = _fjsp_madd_v2r8(vfeps,_fjsp_madd_v2r8(H,vfeps,G),F);
1254             FF               = _fjsp_madd_v2r8(vfeps,_fjsp_madd_v2r8(twovfeps,H,G),Fp);
1255             fvdw12           = _fjsp_mul_v2r8(c12_00,FF);
1256             fvdw             = _fjsp_neg_v2r8(_fjsp_mul_v2r8(_fjsp_add_v2r8(fvdw6,fvdw12),_fjsp_mul_v2r8(vftabscale,rinv00)));
1257
1258             fscal            = fvdw;
1259
1260             /* Update vectorial force */
1261             fix0             = _fjsp_madd_v2r8(dx00,fscal,fix0);
1262             fiy0             = _fjsp_madd_v2r8(dy00,fscal,fiy0);
1263             fiz0             = _fjsp_madd_v2r8(dz00,fscal,fiz0);
1264
1265             fjx0             = _fjsp_madd_v2r8(dx00,fscal,fjx0);
1266             fjy0             = _fjsp_madd_v2r8(dy00,fscal,fjy0);
1267             fjz0             = _fjsp_madd_v2r8(dz00,fscal,fjz0);
1268
1269             /**************************
1270              * CALCULATE INTERACTIONS *
1271              **************************/
1272
1273             /* REACTION-FIELD ELECTROSTATICS */
1274             felec            = _fjsp_mul_v2r8(qq11,_fjsp_msub_v2r8(rinv11,rinvsq11,krf2));
1275
1276             fscal            = felec;
1277
1278             /* Update vectorial force */
1279             fix1             = _fjsp_madd_v2r8(dx11,fscal,fix1);
1280             fiy1             = _fjsp_madd_v2r8(dy11,fscal,fiy1);
1281             fiz1             = _fjsp_madd_v2r8(dz11,fscal,fiz1);
1282
1283             fjx1             = _fjsp_madd_v2r8(dx11,fscal,fjx1);
1284             fjy1             = _fjsp_madd_v2r8(dy11,fscal,fjy1);
1285             fjz1             = _fjsp_madd_v2r8(dz11,fscal,fjz1);
1286
1287             /**************************
1288              * CALCULATE INTERACTIONS *
1289              **************************/
1290
1291             /* REACTION-FIELD ELECTROSTATICS */
1292             felec            = _fjsp_mul_v2r8(qq12,_fjsp_msub_v2r8(rinv12,rinvsq12,krf2));
1293
1294             fscal            = felec;
1295
1296             /* Update vectorial force */
1297             fix1             = _fjsp_madd_v2r8(dx12,fscal,fix1);
1298             fiy1             = _fjsp_madd_v2r8(dy12,fscal,fiy1);
1299             fiz1             = _fjsp_madd_v2r8(dz12,fscal,fiz1);
1300
1301             fjx2             = _fjsp_madd_v2r8(dx12,fscal,fjx2);
1302             fjy2             = _fjsp_madd_v2r8(dy12,fscal,fjy2);
1303             fjz2             = _fjsp_madd_v2r8(dz12,fscal,fjz2);
1304
1305             /**************************
1306              * CALCULATE INTERACTIONS *
1307              **************************/
1308
1309             /* REACTION-FIELD ELECTROSTATICS */
1310             felec            = _fjsp_mul_v2r8(qq13,_fjsp_msub_v2r8(rinv13,rinvsq13,krf2));
1311
1312             fscal            = felec;
1313
1314             /* Update vectorial force */
1315             fix1             = _fjsp_madd_v2r8(dx13,fscal,fix1);
1316             fiy1             = _fjsp_madd_v2r8(dy13,fscal,fiy1);
1317             fiz1             = _fjsp_madd_v2r8(dz13,fscal,fiz1);
1318
1319             fjx3             = _fjsp_madd_v2r8(dx13,fscal,fjx3);
1320             fjy3             = _fjsp_madd_v2r8(dy13,fscal,fjy3);
1321             fjz3             = _fjsp_madd_v2r8(dz13,fscal,fjz3);
1322
1323             /**************************
1324              * CALCULATE INTERACTIONS *
1325              **************************/
1326
1327             /* REACTION-FIELD ELECTROSTATICS */
1328             felec            = _fjsp_mul_v2r8(qq21,_fjsp_msub_v2r8(rinv21,rinvsq21,krf2));
1329
1330             fscal            = felec;
1331
1332             /* Update vectorial force */
1333             fix2             = _fjsp_madd_v2r8(dx21,fscal,fix2);
1334             fiy2             = _fjsp_madd_v2r8(dy21,fscal,fiy2);
1335             fiz2             = _fjsp_madd_v2r8(dz21,fscal,fiz2);
1336
1337             fjx1             = _fjsp_madd_v2r8(dx21,fscal,fjx1);
1338             fjy1             = _fjsp_madd_v2r8(dy21,fscal,fjy1);
1339             fjz1             = _fjsp_madd_v2r8(dz21,fscal,fjz1);
1340
1341             /**************************
1342              * CALCULATE INTERACTIONS *
1343              **************************/
1344
1345             /* REACTION-FIELD ELECTROSTATICS */
1346             felec            = _fjsp_mul_v2r8(qq22,_fjsp_msub_v2r8(rinv22,rinvsq22,krf2));
1347
1348             fscal            = felec;
1349
1350             /* Update vectorial force */
1351             fix2             = _fjsp_madd_v2r8(dx22,fscal,fix2);
1352             fiy2             = _fjsp_madd_v2r8(dy22,fscal,fiy2);
1353             fiz2             = _fjsp_madd_v2r8(dz22,fscal,fiz2);
1354
1355             fjx2             = _fjsp_madd_v2r8(dx22,fscal,fjx2);
1356             fjy2             = _fjsp_madd_v2r8(dy22,fscal,fjy2);
1357             fjz2             = _fjsp_madd_v2r8(dz22,fscal,fjz2);
1358
1359             /**************************
1360              * CALCULATE INTERACTIONS *
1361              **************************/
1362
1363             /* REACTION-FIELD ELECTROSTATICS */
1364             felec            = _fjsp_mul_v2r8(qq23,_fjsp_msub_v2r8(rinv23,rinvsq23,krf2));
1365
1366             fscal            = felec;
1367
1368             /* Update vectorial force */
1369             fix2             = _fjsp_madd_v2r8(dx23,fscal,fix2);
1370             fiy2             = _fjsp_madd_v2r8(dy23,fscal,fiy2);
1371             fiz2             = _fjsp_madd_v2r8(dz23,fscal,fiz2);
1372
1373             fjx3             = _fjsp_madd_v2r8(dx23,fscal,fjx3);
1374             fjy3             = _fjsp_madd_v2r8(dy23,fscal,fjy3);
1375             fjz3             = _fjsp_madd_v2r8(dz23,fscal,fjz3);
1376
1377             /**************************
1378              * CALCULATE INTERACTIONS *
1379              **************************/
1380
1381             /* REACTION-FIELD ELECTROSTATICS */
1382             felec            = _fjsp_mul_v2r8(qq31,_fjsp_msub_v2r8(rinv31,rinvsq31,krf2));
1383
1384             fscal            = felec;
1385
1386             /* Update vectorial force */
1387             fix3             = _fjsp_madd_v2r8(dx31,fscal,fix3);
1388             fiy3             = _fjsp_madd_v2r8(dy31,fscal,fiy3);
1389             fiz3             = _fjsp_madd_v2r8(dz31,fscal,fiz3);
1390
1391             fjx1             = _fjsp_madd_v2r8(dx31,fscal,fjx1);
1392             fjy1             = _fjsp_madd_v2r8(dy31,fscal,fjy1);
1393             fjz1             = _fjsp_madd_v2r8(dz31,fscal,fjz1);
1394
1395             /**************************
1396              * CALCULATE INTERACTIONS *
1397              **************************/
1398
1399             /* REACTION-FIELD ELECTROSTATICS */
1400             felec            = _fjsp_mul_v2r8(qq32,_fjsp_msub_v2r8(rinv32,rinvsq32,krf2));
1401
1402             fscal            = felec;
1403
1404             /* Update vectorial force */
1405             fix3             = _fjsp_madd_v2r8(dx32,fscal,fix3);
1406             fiy3             = _fjsp_madd_v2r8(dy32,fscal,fiy3);
1407             fiz3             = _fjsp_madd_v2r8(dz32,fscal,fiz3);
1408
1409             fjx2             = _fjsp_madd_v2r8(dx32,fscal,fjx2);
1410             fjy2             = _fjsp_madd_v2r8(dy32,fscal,fjy2);
1411             fjz2             = _fjsp_madd_v2r8(dz32,fscal,fjz2);
1412
1413             /**************************
1414              * CALCULATE INTERACTIONS *
1415              **************************/
1416
1417             /* REACTION-FIELD ELECTROSTATICS */
1418             felec            = _fjsp_mul_v2r8(qq33,_fjsp_msub_v2r8(rinv33,rinvsq33,krf2));
1419
1420             fscal            = felec;
1421
1422             /* Update vectorial force */
1423             fix3             = _fjsp_madd_v2r8(dx33,fscal,fix3);
1424             fiy3             = _fjsp_madd_v2r8(dy33,fscal,fiy3);
1425             fiz3             = _fjsp_madd_v2r8(dz33,fscal,fiz3);
1426
1427             fjx3             = _fjsp_madd_v2r8(dx33,fscal,fjx3);
1428             fjy3             = _fjsp_madd_v2r8(dy33,fscal,fjy3);
1429             fjz3             = _fjsp_madd_v2r8(dz33,fscal,fjz3);
1430
1431             gmx_fjsp_decrement_4rvec_2ptr_swizzle_v2r8(f+j_coord_offsetA,f+j_coord_offsetB,fjx0,fjy0,fjz0,fjx1,fjy1,fjz1,fjx2,fjy2,fjz2,fjx3,fjy3,fjz3);
1432
1433             /* Inner loop uses 324 flops */
1434         }
1435
1436         if(jidx<j_index_end)
1437         {
1438
1439             jnrA             = jjnr[jidx];
1440             j_coord_offsetA  = DIM*jnrA;
1441
1442             /* load j atom coordinates */
1443             gmx_fjsp_load_4rvec_1ptr_swizzle_v2r8(x+j_coord_offsetA,
1444                                               &jx0,&jy0,&jz0,&jx1,&jy1,&jz1,&jx2,
1445                                               &jy2,&jz2,&jx3,&jy3,&jz3);
1446
1447             /* Calculate displacement vector */
1448             dx00             = _fjsp_sub_v2r8(ix0,jx0);
1449             dy00             = _fjsp_sub_v2r8(iy0,jy0);
1450             dz00             = _fjsp_sub_v2r8(iz0,jz0);
1451             dx11             = _fjsp_sub_v2r8(ix1,jx1);
1452             dy11             = _fjsp_sub_v2r8(iy1,jy1);
1453             dz11             = _fjsp_sub_v2r8(iz1,jz1);
1454             dx12             = _fjsp_sub_v2r8(ix1,jx2);
1455             dy12             = _fjsp_sub_v2r8(iy1,jy2);
1456             dz12             = _fjsp_sub_v2r8(iz1,jz2);
1457             dx13             = _fjsp_sub_v2r8(ix1,jx3);
1458             dy13             = _fjsp_sub_v2r8(iy1,jy3);
1459             dz13             = _fjsp_sub_v2r8(iz1,jz3);
1460             dx21             = _fjsp_sub_v2r8(ix2,jx1);
1461             dy21             = _fjsp_sub_v2r8(iy2,jy1);
1462             dz21             = _fjsp_sub_v2r8(iz2,jz1);
1463             dx22             = _fjsp_sub_v2r8(ix2,jx2);
1464             dy22             = _fjsp_sub_v2r8(iy2,jy2);
1465             dz22             = _fjsp_sub_v2r8(iz2,jz2);
1466             dx23             = _fjsp_sub_v2r8(ix2,jx3);
1467             dy23             = _fjsp_sub_v2r8(iy2,jy3);
1468             dz23             = _fjsp_sub_v2r8(iz2,jz3);
1469             dx31             = _fjsp_sub_v2r8(ix3,jx1);
1470             dy31             = _fjsp_sub_v2r8(iy3,jy1);
1471             dz31             = _fjsp_sub_v2r8(iz3,jz1);
1472             dx32             = _fjsp_sub_v2r8(ix3,jx2);
1473             dy32             = _fjsp_sub_v2r8(iy3,jy2);
1474             dz32             = _fjsp_sub_v2r8(iz3,jz2);
1475             dx33             = _fjsp_sub_v2r8(ix3,jx3);
1476             dy33             = _fjsp_sub_v2r8(iy3,jy3);
1477             dz33             = _fjsp_sub_v2r8(iz3,jz3);
1478
1479             /* Calculate squared distance and things based on it */
1480             rsq00            = gmx_fjsp_calc_rsq_v2r8(dx00,dy00,dz00);
1481             rsq11            = gmx_fjsp_calc_rsq_v2r8(dx11,dy11,dz11);
1482             rsq12            = gmx_fjsp_calc_rsq_v2r8(dx12,dy12,dz12);
1483             rsq13            = gmx_fjsp_calc_rsq_v2r8(dx13,dy13,dz13);
1484             rsq21            = gmx_fjsp_calc_rsq_v2r8(dx21,dy21,dz21);
1485             rsq22            = gmx_fjsp_calc_rsq_v2r8(dx22,dy22,dz22);
1486             rsq23            = gmx_fjsp_calc_rsq_v2r8(dx23,dy23,dz23);
1487             rsq31            = gmx_fjsp_calc_rsq_v2r8(dx31,dy31,dz31);
1488             rsq32            = gmx_fjsp_calc_rsq_v2r8(dx32,dy32,dz32);
1489             rsq33            = gmx_fjsp_calc_rsq_v2r8(dx33,dy33,dz33);
1490
1491             rinv00           = gmx_fjsp_invsqrt_v2r8(rsq00);
1492             rinv11           = gmx_fjsp_invsqrt_v2r8(rsq11);
1493             rinv12           = gmx_fjsp_invsqrt_v2r8(rsq12);
1494             rinv13           = gmx_fjsp_invsqrt_v2r8(rsq13);
1495             rinv21           = gmx_fjsp_invsqrt_v2r8(rsq21);
1496             rinv22           = gmx_fjsp_invsqrt_v2r8(rsq22);
1497             rinv23           = gmx_fjsp_invsqrt_v2r8(rsq23);
1498             rinv31           = gmx_fjsp_invsqrt_v2r8(rsq31);
1499             rinv32           = gmx_fjsp_invsqrt_v2r8(rsq32);
1500             rinv33           = gmx_fjsp_invsqrt_v2r8(rsq33);
1501
1502             rinvsq11         = _fjsp_mul_v2r8(rinv11,rinv11);
1503             rinvsq12         = _fjsp_mul_v2r8(rinv12,rinv12);
1504             rinvsq13         = _fjsp_mul_v2r8(rinv13,rinv13);
1505             rinvsq21         = _fjsp_mul_v2r8(rinv21,rinv21);
1506             rinvsq22         = _fjsp_mul_v2r8(rinv22,rinv22);
1507             rinvsq23         = _fjsp_mul_v2r8(rinv23,rinv23);
1508             rinvsq31         = _fjsp_mul_v2r8(rinv31,rinv31);
1509             rinvsq32         = _fjsp_mul_v2r8(rinv32,rinv32);
1510             rinvsq33         = _fjsp_mul_v2r8(rinv33,rinv33);
1511
1512             fjx0             = _fjsp_setzero_v2r8();
1513             fjy0             = _fjsp_setzero_v2r8();
1514             fjz0             = _fjsp_setzero_v2r8();
1515             fjx1             = _fjsp_setzero_v2r8();
1516             fjy1             = _fjsp_setzero_v2r8();
1517             fjz1             = _fjsp_setzero_v2r8();
1518             fjx2             = _fjsp_setzero_v2r8();
1519             fjy2             = _fjsp_setzero_v2r8();
1520             fjz2             = _fjsp_setzero_v2r8();
1521             fjx3             = _fjsp_setzero_v2r8();
1522             fjy3             = _fjsp_setzero_v2r8();
1523             fjz3             = _fjsp_setzero_v2r8();
1524
1525             /**************************
1526              * CALCULATE INTERACTIONS *
1527              **************************/
1528
1529             r00              = _fjsp_mul_v2r8(rsq00,rinv00);
1530
1531             /* Calculate table index by multiplying r with table scale and truncate to integer */
1532             rt               = _fjsp_mul_v2r8(r00,vftabscale);
1533             itab_tmp         = _fjsp_dtox_v2r8(rt);
1534             vfeps            = _fjsp_sub_v2r8(rt, _fjsp_xtod_v2r8(itab_tmp));
1535             twovfeps         = _fjsp_add_v2r8(vfeps,vfeps);
1536             _fjsp_store_v2r8(&vfconv.simd,itab_tmp);
1537
1538             vfconv.i[0]     *= 8;
1539             vfconv.i[1]     *= 8;
1540
1541             /* CUBIC SPLINE TABLE DISPERSION */
1542             Y                = _fjsp_load_v2r8( vftab + vfconv.i[0] );
1543             F                = _fjsp_setzero_v2r8();
1544             GMX_FJSP_TRANSPOSE2_V2R8(Y,F);
1545             G                = _fjsp_load_v2r8( vftab + vfconv.i[0] + 2 );
1546             H                = _fjsp_setzero_v2r8();
1547             GMX_FJSP_TRANSPOSE2_V2R8(G,H);
1548             Fp               = _fjsp_madd_v2r8(vfeps,_fjsp_madd_v2r8(H,vfeps,G),F);
1549             FF               = _fjsp_madd_v2r8(vfeps,_fjsp_madd_v2r8(twovfeps,H,G),Fp);
1550             fvdw6            = _fjsp_mul_v2r8(c6_00,FF);
1551
1552             /* CUBIC SPLINE TABLE REPULSION */
1553             Y                = _fjsp_load_v2r8( vftab + vfconv.i[0] + 4 );
1554             F                = _fjsp_setzero_v2r8();
1555             GMX_FJSP_TRANSPOSE2_V2R8(Y,F);
1556             G                = _fjsp_load_v2r8( vftab + vfconv.i[0] + 6 );
1557             H                = _fjsp_setzero_v2r8();
1558             GMX_FJSP_TRANSPOSE2_V2R8(G,H);
1559             Fp               = _fjsp_madd_v2r8(vfeps,_fjsp_madd_v2r8(H,vfeps,G),F);
1560             FF               = _fjsp_madd_v2r8(vfeps,_fjsp_madd_v2r8(twovfeps,H,G),Fp);
1561             fvdw12           = _fjsp_mul_v2r8(c12_00,FF);
1562             fvdw             = _fjsp_neg_v2r8(_fjsp_mul_v2r8(_fjsp_add_v2r8(fvdw6,fvdw12),_fjsp_mul_v2r8(vftabscale,rinv00)));
1563
1564             fscal            = fvdw;
1565
1566             fscal            = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
1567
1568             /* Update vectorial force */
1569             fix0             = _fjsp_madd_v2r8(dx00,fscal,fix0);
1570             fiy0             = _fjsp_madd_v2r8(dy00,fscal,fiy0);
1571             fiz0             = _fjsp_madd_v2r8(dz00,fscal,fiz0);
1572
1573             fjx0             = _fjsp_madd_v2r8(dx00,fscal,fjx0);
1574             fjy0             = _fjsp_madd_v2r8(dy00,fscal,fjy0);
1575             fjz0             = _fjsp_madd_v2r8(dz00,fscal,fjz0);
1576
1577             /**************************
1578              * CALCULATE INTERACTIONS *
1579              **************************/
1580
1581             /* REACTION-FIELD ELECTROSTATICS */
1582             felec            = _fjsp_mul_v2r8(qq11,_fjsp_msub_v2r8(rinv11,rinvsq11,krf2));
1583
1584             fscal            = felec;
1585
1586             fscal            = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
1587
1588             /* Update vectorial force */
1589             fix1             = _fjsp_madd_v2r8(dx11,fscal,fix1);
1590             fiy1             = _fjsp_madd_v2r8(dy11,fscal,fiy1);
1591             fiz1             = _fjsp_madd_v2r8(dz11,fscal,fiz1);
1592
1593             fjx1             = _fjsp_madd_v2r8(dx11,fscal,fjx1);
1594             fjy1             = _fjsp_madd_v2r8(dy11,fscal,fjy1);
1595             fjz1             = _fjsp_madd_v2r8(dz11,fscal,fjz1);
1596
1597             /**************************
1598              * CALCULATE INTERACTIONS *
1599              **************************/
1600
1601             /* REACTION-FIELD ELECTROSTATICS */
1602             felec            = _fjsp_mul_v2r8(qq12,_fjsp_msub_v2r8(rinv12,rinvsq12,krf2));
1603
1604             fscal            = felec;
1605
1606             fscal            = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
1607
1608             /* Update vectorial force */
1609             fix1             = _fjsp_madd_v2r8(dx12,fscal,fix1);
1610             fiy1             = _fjsp_madd_v2r8(dy12,fscal,fiy1);
1611             fiz1             = _fjsp_madd_v2r8(dz12,fscal,fiz1);
1612
1613             fjx2             = _fjsp_madd_v2r8(dx12,fscal,fjx2);
1614             fjy2             = _fjsp_madd_v2r8(dy12,fscal,fjy2);
1615             fjz2             = _fjsp_madd_v2r8(dz12,fscal,fjz2);
1616
1617             /**************************
1618              * CALCULATE INTERACTIONS *
1619              **************************/
1620
1621             /* REACTION-FIELD ELECTROSTATICS */
1622             felec            = _fjsp_mul_v2r8(qq13,_fjsp_msub_v2r8(rinv13,rinvsq13,krf2));
1623
1624             fscal            = felec;
1625
1626             fscal            = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
1627
1628             /* Update vectorial force */
1629             fix1             = _fjsp_madd_v2r8(dx13,fscal,fix1);
1630             fiy1             = _fjsp_madd_v2r8(dy13,fscal,fiy1);
1631             fiz1             = _fjsp_madd_v2r8(dz13,fscal,fiz1);
1632
1633             fjx3             = _fjsp_madd_v2r8(dx13,fscal,fjx3);
1634             fjy3             = _fjsp_madd_v2r8(dy13,fscal,fjy3);
1635             fjz3             = _fjsp_madd_v2r8(dz13,fscal,fjz3);
1636
1637             /**************************
1638              * CALCULATE INTERACTIONS *
1639              **************************/
1640
1641             /* REACTION-FIELD ELECTROSTATICS */
1642             felec            = _fjsp_mul_v2r8(qq21,_fjsp_msub_v2r8(rinv21,rinvsq21,krf2));
1643
1644             fscal            = felec;
1645
1646             fscal            = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
1647
1648             /* Update vectorial force */
1649             fix2             = _fjsp_madd_v2r8(dx21,fscal,fix2);
1650             fiy2             = _fjsp_madd_v2r8(dy21,fscal,fiy2);
1651             fiz2             = _fjsp_madd_v2r8(dz21,fscal,fiz2);
1652
1653             fjx1             = _fjsp_madd_v2r8(dx21,fscal,fjx1);
1654             fjy1             = _fjsp_madd_v2r8(dy21,fscal,fjy1);
1655             fjz1             = _fjsp_madd_v2r8(dz21,fscal,fjz1);
1656
1657             /**************************
1658              * CALCULATE INTERACTIONS *
1659              **************************/
1660
1661             /* REACTION-FIELD ELECTROSTATICS */
1662             felec            = _fjsp_mul_v2r8(qq22,_fjsp_msub_v2r8(rinv22,rinvsq22,krf2));
1663
1664             fscal            = felec;
1665
1666             fscal            = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
1667
1668             /* Update vectorial force */
1669             fix2             = _fjsp_madd_v2r8(dx22,fscal,fix2);
1670             fiy2             = _fjsp_madd_v2r8(dy22,fscal,fiy2);
1671             fiz2             = _fjsp_madd_v2r8(dz22,fscal,fiz2);
1672
1673             fjx2             = _fjsp_madd_v2r8(dx22,fscal,fjx2);
1674             fjy2             = _fjsp_madd_v2r8(dy22,fscal,fjy2);
1675             fjz2             = _fjsp_madd_v2r8(dz22,fscal,fjz2);
1676
1677             /**************************
1678              * CALCULATE INTERACTIONS *
1679              **************************/
1680
1681             /* REACTION-FIELD ELECTROSTATICS */
1682             felec            = _fjsp_mul_v2r8(qq23,_fjsp_msub_v2r8(rinv23,rinvsq23,krf2));
1683
1684             fscal            = felec;
1685
1686             fscal            = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
1687
1688             /* Update vectorial force */
1689             fix2             = _fjsp_madd_v2r8(dx23,fscal,fix2);
1690             fiy2             = _fjsp_madd_v2r8(dy23,fscal,fiy2);
1691             fiz2             = _fjsp_madd_v2r8(dz23,fscal,fiz2);
1692
1693             fjx3             = _fjsp_madd_v2r8(dx23,fscal,fjx3);
1694             fjy3             = _fjsp_madd_v2r8(dy23,fscal,fjy3);
1695             fjz3             = _fjsp_madd_v2r8(dz23,fscal,fjz3);
1696
1697             /**************************
1698              * CALCULATE INTERACTIONS *
1699              **************************/
1700
1701             /* REACTION-FIELD ELECTROSTATICS */
1702             felec            = _fjsp_mul_v2r8(qq31,_fjsp_msub_v2r8(rinv31,rinvsq31,krf2));
1703
1704             fscal            = felec;
1705
1706             fscal            = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
1707
1708             /* Update vectorial force */
1709             fix3             = _fjsp_madd_v2r8(dx31,fscal,fix3);
1710             fiy3             = _fjsp_madd_v2r8(dy31,fscal,fiy3);
1711             fiz3             = _fjsp_madd_v2r8(dz31,fscal,fiz3);
1712
1713             fjx1             = _fjsp_madd_v2r8(dx31,fscal,fjx1);
1714             fjy1             = _fjsp_madd_v2r8(dy31,fscal,fjy1);
1715             fjz1             = _fjsp_madd_v2r8(dz31,fscal,fjz1);
1716
1717             /**************************
1718              * CALCULATE INTERACTIONS *
1719              **************************/
1720
1721             /* REACTION-FIELD ELECTROSTATICS */
1722             felec            = _fjsp_mul_v2r8(qq32,_fjsp_msub_v2r8(rinv32,rinvsq32,krf2));
1723
1724             fscal            = felec;
1725
1726             fscal            = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
1727
1728             /* Update vectorial force */
1729             fix3             = _fjsp_madd_v2r8(dx32,fscal,fix3);
1730             fiy3             = _fjsp_madd_v2r8(dy32,fscal,fiy3);
1731             fiz3             = _fjsp_madd_v2r8(dz32,fscal,fiz3);
1732
1733             fjx2             = _fjsp_madd_v2r8(dx32,fscal,fjx2);
1734             fjy2             = _fjsp_madd_v2r8(dy32,fscal,fjy2);
1735             fjz2             = _fjsp_madd_v2r8(dz32,fscal,fjz2);
1736
1737             /**************************
1738              * CALCULATE INTERACTIONS *
1739              **************************/
1740
1741             /* REACTION-FIELD ELECTROSTATICS */
1742             felec            = _fjsp_mul_v2r8(qq33,_fjsp_msub_v2r8(rinv33,rinvsq33,krf2));
1743
1744             fscal            = felec;
1745
1746             fscal            = _fjsp_unpacklo_v2r8(fscal,_fjsp_setzero_v2r8());
1747
1748             /* Update vectorial force */
1749             fix3             = _fjsp_madd_v2r8(dx33,fscal,fix3);
1750             fiy3             = _fjsp_madd_v2r8(dy33,fscal,fiy3);
1751             fiz3             = _fjsp_madd_v2r8(dz33,fscal,fiz3);
1752
1753             fjx3             = _fjsp_madd_v2r8(dx33,fscal,fjx3);
1754             fjy3             = _fjsp_madd_v2r8(dy33,fscal,fjy3);
1755             fjz3             = _fjsp_madd_v2r8(dz33,fscal,fjz3);
1756
1757             gmx_fjsp_decrement_4rvec_1ptr_swizzle_v2r8(f+j_coord_offsetA,fjx0,fjy0,fjz0,fjx1,fjy1,fjz1,fjx2,fjy2,fjz2,fjx3,fjy3,fjz3);
1758
1759             /* Inner loop uses 324 flops */
1760         }
1761
1762         /* End of innermost loop */
1763
1764         gmx_fjsp_update_iforce_4atom_swizzle_v2r8(fix0,fiy0,fiz0,fix1,fiy1,fiz1,fix2,fiy2,fiz2,fix3,fiy3,fiz3,
1765                                               f+i_coord_offset,fshift+i_shift_offset);
1766
1767         /* Increment number of inner iterations */
1768         inneriter                  += j_index_end - j_index_start;
1769
1770         /* Outer loop uses 24 flops */
1771     }
1772
1773     /* Increment number of outer iterations */
1774     outeriter        += nri;
1775
1776     /* Update outer/inner flops */
1777
1778     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W4W4_F,outeriter*24 + inneriter*324);
1779 }