src/gromacs/gmxlib/nonbonded/nb_kernel_avx_256_double/nb_kernel_ElecEw_VdwLJEw_GeomW3W3_avx_256_double.cpp

   1 /*
   2  * This file is part of the GROMACS molecular simulation package.
   3  *
   4  * Copyright (c) 2012,2013,2014,2015,2017,2018, 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 avx_256_double kernel generator.
  37  */
  38 #include "gmxpre.h"
  39
  40 #include "config.h"
  41
  42 #include <math.h>
  43
  44 #include "../nb_kernel.h"
  45 #include "gromacs/gmxlib/nrnb.h"
  46
  47 #include "kernelutil_x86_avx_256_double.h"
  48
  49 /*
  50  * Gromacs nonbonded kernel:   nb_kernel_ElecEw_VdwLJEw_GeomW3W3_VF_avx_256_double
  51  * Electrostatics interaction: Ewald
  52  * VdW interaction:            LJEwald
  53  * Geometry:                   Water3-Water3
  54  * Calculate force/pot:        PotentialAndForce
  55  */
  56 void
  57 nb_kernel_ElecEw_VdwLJEw_GeomW3W3_VF_avx_256_double
  58                     (t_nblist                    * gmx_restrict       nlist,
  59                      rvec                        * gmx_restrict          xx,
  60                      rvec                        * gmx_restrict          ff,
  61                      struct t_forcerec           * gmx_restrict          fr,
  62                      t_mdatoms                   * gmx_restrict     mdatoms,
  63                      nb_kernel_data_t gmx_unused * gmx_restrict kernel_data,
  64                      t_nrnb                      * gmx_restrict        nrnb)
  65 {
  66     /* Suffixes 0,1,2,3 refer to particle indices for waters in the inner or outer loop, or
  67      * just 0 for non-waters.
  68      * Suffixes A,B,C,D refer to j loop unrolling done with AVX, e.g. for the four 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,jnrC,jnrD;
  74     int              jnrlistA,jnrlistB,jnrlistC,jnrlistD;
  75     int              jnrlistE,jnrlistF,jnrlistG,jnrlistH;
  76     int              j_coord_offsetA,j_coord_offsetB,j_coord_offsetC,j_coord_offsetD;
  77     int              *iinr,*jindex,*jjnr,*shiftidx,*gid;
  78     real             rcutoff_scalar;
  79     real             *shiftvec,*fshift,*x,*f;
  80     real             *fjptrA,*fjptrB,*fjptrC,*fjptrD;
  81     real             scratch[4*DIM];
  82     __m256d          tx,ty,tz,fscal,rcutoff,rcutoff2,jidxall;
  83     real *           vdwioffsetptr0;
  84     real *           vdwgridioffsetptr0;
  85     __m256d          ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
  86     real *           vdwioffsetptr1;
  87     real *           vdwgridioffsetptr1;
  88     __m256d          ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
  89     real *           vdwioffsetptr2;
  90     real *           vdwgridioffsetptr2;
  91     __m256d          ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
  92     int              vdwjidx0A,vdwjidx0B,vdwjidx0C,vdwjidx0D;
  93     __m256d          jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
  94     int              vdwjidx1A,vdwjidx1B,vdwjidx1C,vdwjidx1D;
  95     __m256d          jx1,jy1,jz1,fjx1,fjy1,fjz1,jq1,isaj1;
  96     int              vdwjidx2A,vdwjidx2B,vdwjidx2C,vdwjidx2D;
  97     __m256d          jx2,jy2,jz2,fjx2,fjy2,fjz2,jq2,isaj2;
  98     __m256d          dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00;
  99     __m256d          dx01,dy01,dz01,rsq01,rinv01,rinvsq01,r01,qq01,c6_01,c12_01;
 100     __m256d          dx02,dy02,dz02,rsq02,rinv02,rinvsq02,r02,qq02,c6_02,c12_02;
 101     __m256d          dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10;
 102     __m256d          dx11,dy11,dz11,rsq11,rinv11,rinvsq11,r11,qq11,c6_11,c12_11;
 103     __m256d          dx12,dy12,dz12,rsq12,rinv12,rinvsq12,r12,qq12,c6_12,c12_12;
 104     __m256d          dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20;
 105     __m256d          dx21,dy21,dz21,rsq21,rinv21,rinvsq21,r21,qq21,c6_21,c12_21;
 106     __m256d          dx22,dy22,dz22,rsq22,rinv22,rinvsq22,r22,qq22,c6_22,c12_22;
 107     __m256d          velec,felec,velecsum,facel,crf,krf,krf2;
 108     real             *charge;
 109     int              nvdwtype;
 110     __m256d          rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,sh_vdw_invrcut6;
 111     int              *vdwtype;
 112     real             *vdwparam;
 113     __m256d          one_sixth   = _mm256_set1_pd(1.0/6.0);
 114     __m256d          one_twelfth = _mm256_set1_pd(1.0/12.0);
 115     __m256d           c6grid_00;
 116     __m256d           c6grid_01;
 117     __m256d           c6grid_02;
 118     __m256d           c6grid_10;
 119     __m256d           c6grid_11;
 120     __m256d           c6grid_12;
 121     __m256d           c6grid_20;
 122     __m256d           c6grid_21;
 123     __m256d           c6grid_22;
 124     real             *vdwgridparam;
 125     __m256d           ewclj,ewclj2,ewclj6,ewcljrsq,poly,exponent,f6A,f6B,sh_lj_ewald;
 126     __m256d           one_half  = _mm256_set1_pd(0.5);
 127     __m256d           minus_one = _mm256_set1_pd(-1.0);
 128     __m128i          ewitab;
 129     __m256d          ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace,ewtabF,ewtabFn,ewtabD,ewtabV;
 130     __m256d          beta,beta2,beta3,zeta2,pmecorrF,pmecorrV,rinv3;
 131     real             *ewtab;
 132     __m256d          dummy_mask,cutoff_mask;
 133     __m128           tmpmask0,tmpmask1;
 134     __m256d          signbit = _mm256_castsi256_pd( _mm256_set1_epi32(0x80000000) );
 135     __m256d          one     = _mm256_set1_pd(1.0);
 136     __m256d          two     = _mm256_set1_pd(2.0);
 137     x                = xx[0];
 138     f                = ff[0];
 139
 140     nri              = nlist->nri;
 141     iinr             = nlist->iinr;
 142     jindex           = nlist->jindex;
 143     jjnr             = nlist->jjnr;
 144     shiftidx         = nlist->shift;
 145     gid              = nlist->gid;
 146     shiftvec         = fr->shift_vec[0];
 147     fshift           = fr->fshift[0];
 148     facel            = _mm256_set1_pd(fr->ic->epsfac);
 149     charge           = mdatoms->chargeA;
 150     nvdwtype         = fr->ntype;
 151     vdwparam         = fr->nbfp;
 152     vdwtype          = mdatoms->typeA;
 153     vdwgridparam     = fr->ljpme_c6grid;
 154     sh_lj_ewald      = _mm256_set1_pd(fr->ic->sh_lj_ewald);
 155     ewclj            = _mm256_set1_pd(fr->ic->ewaldcoeff_lj);
 156     ewclj2           = _mm256_mul_pd(minus_one,_mm256_mul_pd(ewclj,ewclj));
 157
 158     sh_ewald         = _mm256_set1_pd(fr->ic->sh_ewald);
 159     beta             = _mm256_set1_pd(fr->ic->ewaldcoeff_q);
 160     beta2            = _mm256_mul_pd(beta,beta);
 161     beta3            = _mm256_mul_pd(beta,beta2);
 162
 163     ewtab            = fr->ic->tabq_coul_FDV0;
 164     ewtabscale       = _mm256_set1_pd(fr->ic->tabq_scale);
 165     ewtabhalfspace   = _mm256_set1_pd(0.5/fr->ic->tabq_scale);
 166
 167     /* Setup water-specific parameters */
 168     inr              = nlist->iinr[0];
 169     iq0              = _mm256_mul_pd(facel,_mm256_set1_pd(charge[inr+0]));
 170     iq1              = _mm256_mul_pd(facel,_mm256_set1_pd(charge[inr+1]));
 171     iq2              = _mm256_mul_pd(facel,_mm256_set1_pd(charge[inr+2]));
 172     vdwioffsetptr0   = vdwparam+2*nvdwtype*vdwtype[inr+0];
 173     vdwgridioffsetptr0 = vdwgridparam+2*nvdwtype*vdwtype[inr+0];
 174
 175     jq0              = _mm256_set1_pd(charge[inr+0]);
 176     jq1              = _mm256_set1_pd(charge[inr+1]);
 177     jq2              = _mm256_set1_pd(charge[inr+2]);
 178     vdwjidx0A        = 2*vdwtype[inr+0];
 179     qq00             = _mm256_mul_pd(iq0,jq0);
 180     c6_00            = _mm256_set1_pd(vdwioffsetptr0[vdwjidx0A]);
 181     c12_00           = _mm256_set1_pd(vdwioffsetptr0[vdwjidx0A+1]);
 182     c6grid_00        = _mm256_set1_pd(vdwgridioffsetptr0[vdwjidx0A]);
 183     qq01             = _mm256_mul_pd(iq0,jq1);
 184     qq02             = _mm256_mul_pd(iq0,jq2);
 185     qq10             = _mm256_mul_pd(iq1,jq0);
 186     qq11             = _mm256_mul_pd(iq1,jq1);
 187     qq12             = _mm256_mul_pd(iq1,jq2);
 188     qq20             = _mm256_mul_pd(iq2,jq0);
 189     qq21             = _mm256_mul_pd(iq2,jq1);
 190     qq22             = _mm256_mul_pd(iq2,jq2);
 191
 192     /* Avoid stupid compiler warnings */
 193     jnrA = jnrB = jnrC = jnrD = 0;
 194     j_coord_offsetA = 0;
 195     j_coord_offsetB = 0;
 196     j_coord_offsetC = 0;
 197     j_coord_offsetD = 0;
 198
 199     outeriter        = 0;
 200     inneriter        = 0;
 201
 202     for(iidx=0;iidx<4*DIM;iidx++)
 203     {
 204         scratch[iidx] = 0.0;
 205     }
 206
 207     /* Start outer loop over neighborlists */
 208     for(iidx=0; iidx<nri; iidx++)
 209     {
 210         /* Load shift vector for this list */
 211         i_shift_offset   = DIM*shiftidx[iidx];
 212
 213         /* Load limits for loop over neighbors */
 214         j_index_start    = jindex[iidx];
 215         j_index_end      = jindex[iidx+1];
 216
 217         /* Get outer coordinate index */
 218         inr              = iinr[iidx];
 219         i_coord_offset   = DIM*inr;
 220
 221         /* Load i particle coords and add shift vector */
 222         gmx_mm256_load_shift_and_3rvec_broadcast_pd(shiftvec+i_shift_offset,x+i_coord_offset,
 223                                                     &ix0,&iy0,&iz0,&ix1,&iy1,&iz1,&ix2,&iy2,&iz2);
 224
 225         fix0             = _mm256_setzero_pd();
 226         fiy0             = _mm256_setzero_pd();
 227         fiz0             = _mm256_setzero_pd();
 228         fix1             = _mm256_setzero_pd();
 229         fiy1             = _mm256_setzero_pd();
 230         fiz1             = _mm256_setzero_pd();
 231         fix2             = _mm256_setzero_pd();
 232         fiy2             = _mm256_setzero_pd();
 233         fiz2             = _mm256_setzero_pd();
 234
 235         /* Reset potential sums */
 236         velecsum         = _mm256_setzero_pd();
 237         vvdwsum          = _mm256_setzero_pd();
 238
 239         /* Start inner kernel loop */
 240         for(jidx=j_index_start; jidx<j_index_end && jjnr[jidx+3]>=0; jidx+=4)
 241         {
 242
 243             /* Get j neighbor index, and coordinate index */
 244             jnrA             = jjnr[jidx];
 245             jnrB             = jjnr[jidx+1];
 246             jnrC             = jjnr[jidx+2];
 247             jnrD             = jjnr[jidx+3];
 248             j_coord_offsetA  = DIM*jnrA;
 249             j_coord_offsetB  = DIM*jnrB;
 250             j_coord_offsetC  = DIM*jnrC;
 251             j_coord_offsetD  = DIM*jnrD;
 252
 253             /* load j atom coordinates */
 254             gmx_mm256_load_3rvec_4ptr_swizzle_pd(x+j_coord_offsetA,x+j_coord_offsetB,
 255                                                  x+j_coord_offsetC,x+j_coord_offsetD,
 256                                               &jx0,&jy0,&jz0,&jx1,&jy1,&jz1,&jx2,&jy2,&jz2);
 257
 258             /* Calculate displacement vector */
 259             dx00             = _mm256_sub_pd(ix0,jx0);
 260             dy00             = _mm256_sub_pd(iy0,jy0);
 261             dz00             = _mm256_sub_pd(iz0,jz0);
 262             dx01             = _mm256_sub_pd(ix0,jx1);
 263             dy01             = _mm256_sub_pd(iy0,jy1);
 264             dz01             = _mm256_sub_pd(iz0,jz1);
 265             dx02             = _mm256_sub_pd(ix0,jx2);
 266             dy02             = _mm256_sub_pd(iy0,jy2);
 267             dz02             = _mm256_sub_pd(iz0,jz2);
 268             dx10             = _mm256_sub_pd(ix1,jx0);
 269             dy10             = _mm256_sub_pd(iy1,jy0);
 270             dz10             = _mm256_sub_pd(iz1,jz0);
 271             dx11             = _mm256_sub_pd(ix1,jx1);
 272             dy11             = _mm256_sub_pd(iy1,jy1);
 273             dz11             = _mm256_sub_pd(iz1,jz1);
 274             dx12             = _mm256_sub_pd(ix1,jx2);
 275             dy12             = _mm256_sub_pd(iy1,jy2);
 276             dz12             = _mm256_sub_pd(iz1,jz2);
 277             dx20             = _mm256_sub_pd(ix2,jx0);
 278             dy20             = _mm256_sub_pd(iy2,jy0);
 279             dz20             = _mm256_sub_pd(iz2,jz0);
 280             dx21             = _mm256_sub_pd(ix2,jx1);
 281             dy21             = _mm256_sub_pd(iy2,jy1);
 282             dz21             = _mm256_sub_pd(iz2,jz1);
 283             dx22             = _mm256_sub_pd(ix2,jx2);
 284             dy22             = _mm256_sub_pd(iy2,jy2);
 285             dz22             = _mm256_sub_pd(iz2,jz2);
 286
 287             /* Calculate squared distance and things based on it */
 288             rsq00            = gmx_mm256_calc_rsq_pd(dx00,dy00,dz00);
 289             rsq01            = gmx_mm256_calc_rsq_pd(dx01,dy01,dz01);
 290             rsq02            = gmx_mm256_calc_rsq_pd(dx02,dy02,dz02);
 291             rsq10            = gmx_mm256_calc_rsq_pd(dx10,dy10,dz10);
 292             rsq11            = gmx_mm256_calc_rsq_pd(dx11,dy11,dz11);
 293             rsq12            = gmx_mm256_calc_rsq_pd(dx12,dy12,dz12);
 294             rsq20            = gmx_mm256_calc_rsq_pd(dx20,dy20,dz20);
 295             rsq21            = gmx_mm256_calc_rsq_pd(dx21,dy21,dz21);
 296             rsq22            = gmx_mm256_calc_rsq_pd(dx22,dy22,dz22);
 297
 298             rinv00           = avx256_invsqrt_d(rsq00);
 299             rinv01           = avx256_invsqrt_d(rsq01);
 300             rinv02           = avx256_invsqrt_d(rsq02);
 301             rinv10           = avx256_invsqrt_d(rsq10);
 302             rinv11           = avx256_invsqrt_d(rsq11);
 303             rinv12           = avx256_invsqrt_d(rsq12);
 304             rinv20           = avx256_invsqrt_d(rsq20);
 305             rinv21           = avx256_invsqrt_d(rsq21);
 306             rinv22           = avx256_invsqrt_d(rsq22);
 307
 308             rinvsq00         = _mm256_mul_pd(rinv00,rinv00);
 309             rinvsq01         = _mm256_mul_pd(rinv01,rinv01);
 310             rinvsq02         = _mm256_mul_pd(rinv02,rinv02);
 311             rinvsq10         = _mm256_mul_pd(rinv10,rinv10);
 312             rinvsq11         = _mm256_mul_pd(rinv11,rinv11);
 313             rinvsq12         = _mm256_mul_pd(rinv12,rinv12);
 314             rinvsq20         = _mm256_mul_pd(rinv20,rinv20);
 315             rinvsq21         = _mm256_mul_pd(rinv21,rinv21);
 316             rinvsq22         = _mm256_mul_pd(rinv22,rinv22);
 317
 318             fjx0             = _mm256_setzero_pd();
 319             fjy0             = _mm256_setzero_pd();
 320             fjz0             = _mm256_setzero_pd();
 321             fjx1             = _mm256_setzero_pd();
 322             fjy1             = _mm256_setzero_pd();
 323             fjz1             = _mm256_setzero_pd();
 324             fjx2             = _mm256_setzero_pd();
 325             fjy2             = _mm256_setzero_pd();
 326             fjz2             = _mm256_setzero_pd();
 327
 328             /**************************
 329              * CALCULATE INTERACTIONS *
 330              **************************/
 331
 332             r00              = _mm256_mul_pd(rsq00,rinv00);
 333
 334             /* EWALD ELECTROSTATICS */
 335
 336             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 337             ewrt             = _mm256_mul_pd(r00,ewtabscale);
 338             ewitab           = _mm256_cvttpd_epi32(ewrt);
 339             eweps            = _mm256_sub_pd(ewrt,_mm256_round_pd(ewrt, _MM_FROUND_FLOOR));
 340             ewitab           = _mm_slli_epi32(ewitab,2);
 341             ewtabF           = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,0) );
 342             ewtabD           = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,1) );
 343             ewtabV           = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,2) );
 344             ewtabFn          = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,3) );
 345             GMX_MM256_FULLTRANSPOSE4_PD(ewtabF,ewtabD,ewtabV,ewtabFn);
 346             felec            = _mm256_add_pd(ewtabF,_mm256_mul_pd(eweps,ewtabD));
 347             velec            = _mm256_sub_pd(ewtabV,_mm256_mul_pd(_mm256_mul_pd(ewtabhalfspace,eweps),_mm256_add_pd(ewtabF,felec)));
 348             velec            = _mm256_mul_pd(qq00,_mm256_sub_pd(rinv00,velec));
 349             felec            = _mm256_mul_pd(_mm256_mul_pd(qq00,rinv00),_mm256_sub_pd(rinvsq00,felec));
 350
 351             /* Analytical LJ-PME */
 352             rinvsix          = _mm256_mul_pd(_mm256_mul_pd(rinvsq00,rinvsq00),rinvsq00);
 353             ewcljrsq         = _mm256_mul_pd(ewclj2,rsq00);
 354             ewclj6           = _mm256_mul_pd(ewclj2,_mm256_mul_pd(ewclj2,ewclj2));
 355             exponent         = avx256_exp_d(ewcljrsq);
 356             /* poly = exp(-(beta*r)^2) * (1 + (beta*r)^2 + (beta*r)^4 /2) */
 357             poly             = _mm256_mul_pd(exponent,_mm256_add_pd(_mm256_sub_pd(one,ewcljrsq),_mm256_mul_pd(_mm256_mul_pd(ewcljrsq,ewcljrsq),one_half)));
 358             /* vvdw6 = [C6 - C6grid * (1-poly)]/r6 */
 359             vvdw6            = _mm256_mul_pd(_mm256_sub_pd(c6_00,_mm256_mul_pd(c6grid_00,_mm256_sub_pd(one,poly))),rinvsix);
 360             vvdw12           = _mm256_mul_pd(c12_00,_mm256_mul_pd(rinvsix,rinvsix));
 361             vvdw             = _mm256_sub_pd(_mm256_mul_pd(vvdw12,one_twelfth),_mm256_mul_pd(vvdw6,one_sixth));
 362             /* fvdw = vvdw12/r - (vvdw6/r + (C6grid * exponent * beta^6)/r) */
 363             fvdw             = _mm256_mul_pd(_mm256_sub_pd(vvdw12,_mm256_sub_pd(vvdw6,_mm256_mul_pd(_mm256_mul_pd(c6grid_00,one_sixth),_mm256_mul_pd(exponent,ewclj6)))),rinvsq00);
 364
 365             /* Update potential sum for this i atom from the interaction with this j atom. */
 366             velecsum         = _mm256_add_pd(velecsum,velec);
 367             vvdwsum          = _mm256_add_pd(vvdwsum,vvdw);
 368
 369             fscal            = _mm256_add_pd(felec,fvdw);
 370
 371             /* Calculate temporary vectorial force */
 372             tx               = _mm256_mul_pd(fscal,dx00);
 373             ty               = _mm256_mul_pd(fscal,dy00);
 374             tz               = _mm256_mul_pd(fscal,dz00);
 375
 376             /* Update vectorial force */
 377             fix0             = _mm256_add_pd(fix0,tx);
 378             fiy0             = _mm256_add_pd(fiy0,ty);
 379             fiz0             = _mm256_add_pd(fiz0,tz);
 380
 381             fjx0             = _mm256_add_pd(fjx0,tx);
 382             fjy0             = _mm256_add_pd(fjy0,ty);
 383             fjz0             = _mm256_add_pd(fjz0,tz);
 384
 385             /**************************
 386              * CALCULATE INTERACTIONS *
 387              **************************/
 388
 389             r01              = _mm256_mul_pd(rsq01,rinv01);
 390
 391             /* EWALD ELECTROSTATICS */
 392
 393             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 394             ewrt             = _mm256_mul_pd(r01,ewtabscale);
 395             ewitab           = _mm256_cvttpd_epi32(ewrt);
 396             eweps            = _mm256_sub_pd(ewrt,_mm256_round_pd(ewrt, _MM_FROUND_FLOOR));
 397             ewitab           = _mm_slli_epi32(ewitab,2);
 398             ewtabF           = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,0) );
 399             ewtabD           = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,1) );
 400             ewtabV           = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,2) );
 401             ewtabFn          = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,3) );
 402             GMX_MM256_FULLTRANSPOSE4_PD(ewtabF,ewtabD,ewtabV,ewtabFn);
 403             felec            = _mm256_add_pd(ewtabF,_mm256_mul_pd(eweps,ewtabD));
 404             velec            = _mm256_sub_pd(ewtabV,_mm256_mul_pd(_mm256_mul_pd(ewtabhalfspace,eweps),_mm256_add_pd(ewtabF,felec)));
 405             velec            = _mm256_mul_pd(qq01,_mm256_sub_pd(rinv01,velec));
 406             felec            = _mm256_mul_pd(_mm256_mul_pd(qq01,rinv01),_mm256_sub_pd(rinvsq01,felec));
 407
 408             /* Update potential sum for this i atom from the interaction with this j atom. */
 409             velecsum         = _mm256_add_pd(velecsum,velec);
 410
 411             fscal            = felec;
 412
 413             /* Calculate temporary vectorial force */
 414             tx               = _mm256_mul_pd(fscal,dx01);
 415             ty               = _mm256_mul_pd(fscal,dy01);
 416             tz               = _mm256_mul_pd(fscal,dz01);
 417
 418             /* Update vectorial force */
 419             fix0             = _mm256_add_pd(fix0,tx);
 420             fiy0             = _mm256_add_pd(fiy0,ty);
 421             fiz0             = _mm256_add_pd(fiz0,tz);
 422
 423             fjx1             = _mm256_add_pd(fjx1,tx);
 424             fjy1             = _mm256_add_pd(fjy1,ty);
 425             fjz1             = _mm256_add_pd(fjz1,tz);
 426
 427             /**************************
 428              * CALCULATE INTERACTIONS *
 429              **************************/
 430
 431             r02              = _mm256_mul_pd(rsq02,rinv02);
 432
 433             /* EWALD ELECTROSTATICS */
 434
 435             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 436             ewrt             = _mm256_mul_pd(r02,ewtabscale);
 437             ewitab           = _mm256_cvttpd_epi32(ewrt);
 438             eweps            = _mm256_sub_pd(ewrt,_mm256_round_pd(ewrt, _MM_FROUND_FLOOR));
 439             ewitab           = _mm_slli_epi32(ewitab,2);
 440             ewtabF           = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,0) );
 441             ewtabD           = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,1) );
 442             ewtabV           = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,2) );
 443             ewtabFn          = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,3) );
 444             GMX_MM256_FULLTRANSPOSE4_PD(ewtabF,ewtabD,ewtabV,ewtabFn);
 445             felec            = _mm256_add_pd(ewtabF,_mm256_mul_pd(eweps,ewtabD));
 446             velec            = _mm256_sub_pd(ewtabV,_mm256_mul_pd(_mm256_mul_pd(ewtabhalfspace,eweps),_mm256_add_pd(ewtabF,felec)));
 447             velec            = _mm256_mul_pd(qq02,_mm256_sub_pd(rinv02,velec));
 448             felec            = _mm256_mul_pd(_mm256_mul_pd(qq02,rinv02),_mm256_sub_pd(rinvsq02,felec));
 449
 450             /* Update potential sum for this i atom from the interaction with this j atom. */
 451             velecsum         = _mm256_add_pd(velecsum,velec);
 452
 453             fscal            = felec;
 454
 455             /* Calculate temporary vectorial force */
 456             tx               = _mm256_mul_pd(fscal,dx02);
 457             ty               = _mm256_mul_pd(fscal,dy02);
 458             tz               = _mm256_mul_pd(fscal,dz02);
 459
 460             /* Update vectorial force */
 461             fix0             = _mm256_add_pd(fix0,tx);
 462             fiy0             = _mm256_add_pd(fiy0,ty);
 463             fiz0             = _mm256_add_pd(fiz0,tz);
 464
 465             fjx2             = _mm256_add_pd(fjx2,tx);
 466             fjy2             = _mm256_add_pd(fjy2,ty);
 467             fjz2             = _mm256_add_pd(fjz2,tz);
 468
 469             /**************************
 470              * CALCULATE INTERACTIONS *
 471              **************************/
 472
 473             r10              = _mm256_mul_pd(rsq10,rinv10);
 474
 475             /* EWALD ELECTROSTATICS */
 476
 477             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 478             ewrt             = _mm256_mul_pd(r10,ewtabscale);
 479             ewitab           = _mm256_cvttpd_epi32(ewrt);
 480             eweps            = _mm256_sub_pd(ewrt,_mm256_round_pd(ewrt, _MM_FROUND_FLOOR));
 481             ewitab           = _mm_slli_epi32(ewitab,2);
 482             ewtabF           = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,0) );
 483             ewtabD           = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,1) );
 484             ewtabV           = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,2) );
 485             ewtabFn          = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,3) );
 486             GMX_MM256_FULLTRANSPOSE4_PD(ewtabF,ewtabD,ewtabV,ewtabFn);
 487             felec            = _mm256_add_pd(ewtabF,_mm256_mul_pd(eweps,ewtabD));
 488             velec            = _mm256_sub_pd(ewtabV,_mm256_mul_pd(_mm256_mul_pd(ewtabhalfspace,eweps),_mm256_add_pd(ewtabF,felec)));
 489             velec            = _mm256_mul_pd(qq10,_mm256_sub_pd(rinv10,velec));
 490             felec            = _mm256_mul_pd(_mm256_mul_pd(qq10,rinv10),_mm256_sub_pd(rinvsq10,felec));
 491
 492             /* Update potential sum for this i atom from the interaction with this j atom. */
 493             velecsum         = _mm256_add_pd(velecsum,velec);
 494
 495             fscal            = felec;
 496
 497             /* Calculate temporary vectorial force */
 498             tx               = _mm256_mul_pd(fscal,dx10);
 499             ty               = _mm256_mul_pd(fscal,dy10);
 500             tz               = _mm256_mul_pd(fscal,dz10);
 501
 502             /* Update vectorial force */
 503             fix1             = _mm256_add_pd(fix1,tx);
 504             fiy1             = _mm256_add_pd(fiy1,ty);
 505             fiz1             = _mm256_add_pd(fiz1,tz);
 506
 507             fjx0             = _mm256_add_pd(fjx0,tx);
 508             fjy0             = _mm256_add_pd(fjy0,ty);
 509             fjz0             = _mm256_add_pd(fjz0,tz);
 510
 511             /**************************
 512              * CALCULATE INTERACTIONS *
 513              **************************/
 514
 515             r11              = _mm256_mul_pd(rsq11,rinv11);
 516
 517             /* EWALD ELECTROSTATICS */
 518
 519             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 520             ewrt             = _mm256_mul_pd(r11,ewtabscale);
 521             ewitab           = _mm256_cvttpd_epi32(ewrt);
 522             eweps            = _mm256_sub_pd(ewrt,_mm256_round_pd(ewrt, _MM_FROUND_FLOOR));
 523             ewitab           = _mm_slli_epi32(ewitab,2);
 524             ewtabF           = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,0) );
 525             ewtabD           = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,1) );
 526             ewtabV           = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,2) );
 527             ewtabFn          = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,3) );
 528             GMX_MM256_FULLTRANSPOSE4_PD(ewtabF,ewtabD,ewtabV,ewtabFn);
 529             felec            = _mm256_add_pd(ewtabF,_mm256_mul_pd(eweps,ewtabD));
 530             velec            = _mm256_sub_pd(ewtabV,_mm256_mul_pd(_mm256_mul_pd(ewtabhalfspace,eweps),_mm256_add_pd(ewtabF,felec)));
 531             velec            = _mm256_mul_pd(qq11,_mm256_sub_pd(rinv11,velec));
 532             felec            = _mm256_mul_pd(_mm256_mul_pd(qq11,rinv11),_mm256_sub_pd(rinvsq11,felec));
 533
 534             /* Update potential sum for this i atom from the interaction with this j atom. */
 535             velecsum         = _mm256_add_pd(velecsum,velec);
 536
 537             fscal            = felec;
 538
 539             /* Calculate temporary vectorial force */
 540             tx               = _mm256_mul_pd(fscal,dx11);
 541             ty               = _mm256_mul_pd(fscal,dy11);
 542             tz               = _mm256_mul_pd(fscal,dz11);
 543
 544             /* Update vectorial force */
 545             fix1             = _mm256_add_pd(fix1,tx);
 546             fiy1             = _mm256_add_pd(fiy1,ty);
 547             fiz1             = _mm256_add_pd(fiz1,tz);
 548
 549             fjx1             = _mm256_add_pd(fjx1,tx);
 550             fjy1             = _mm256_add_pd(fjy1,ty);
 551             fjz1             = _mm256_add_pd(fjz1,tz);
 552
 553             /**************************
 554              * CALCULATE INTERACTIONS *
 555              **************************/
 556
 557             r12              = _mm256_mul_pd(rsq12,rinv12);
 558
 559             /* EWALD ELECTROSTATICS */
 560
 561             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 562             ewrt             = _mm256_mul_pd(r12,ewtabscale);
 563             ewitab           = _mm256_cvttpd_epi32(ewrt);
 564             eweps            = _mm256_sub_pd(ewrt,_mm256_round_pd(ewrt, _MM_FROUND_FLOOR));
 565             ewitab           = _mm_slli_epi32(ewitab,2);
 566             ewtabF           = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,0) );
 567             ewtabD           = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,1) );
 568             ewtabV           = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,2) );
 569             ewtabFn          = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,3) );
 570             GMX_MM256_FULLTRANSPOSE4_PD(ewtabF,ewtabD,ewtabV,ewtabFn);
 571             felec            = _mm256_add_pd(ewtabF,_mm256_mul_pd(eweps,ewtabD));
 572             velec            = _mm256_sub_pd(ewtabV,_mm256_mul_pd(_mm256_mul_pd(ewtabhalfspace,eweps),_mm256_add_pd(ewtabF,felec)));
 573             velec            = _mm256_mul_pd(qq12,_mm256_sub_pd(rinv12,velec));
 574             felec            = _mm256_mul_pd(_mm256_mul_pd(qq12,rinv12),_mm256_sub_pd(rinvsq12,felec));
 575
 576             /* Update potential sum for this i atom from the interaction with this j atom. */
 577             velecsum         = _mm256_add_pd(velecsum,velec);
 578
 579             fscal            = felec;
 580
 581             /* Calculate temporary vectorial force */
 582             tx               = _mm256_mul_pd(fscal,dx12);
 583             ty               = _mm256_mul_pd(fscal,dy12);
 584             tz               = _mm256_mul_pd(fscal,dz12);
 585
 586             /* Update vectorial force */
 587             fix1             = _mm256_add_pd(fix1,tx);
 588             fiy1             = _mm256_add_pd(fiy1,ty);
 589             fiz1             = _mm256_add_pd(fiz1,tz);
 590
 591             fjx2             = _mm256_add_pd(fjx2,tx);
 592             fjy2             = _mm256_add_pd(fjy2,ty);
 593             fjz2             = _mm256_add_pd(fjz2,tz);
 594
 595             /**************************
 596              * CALCULATE INTERACTIONS *
 597              **************************/
 598
 599             r20              = _mm256_mul_pd(rsq20,rinv20);
 600
 601             /* EWALD ELECTROSTATICS */
 602
 603             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 604             ewrt             = _mm256_mul_pd(r20,ewtabscale);
 605             ewitab           = _mm256_cvttpd_epi32(ewrt);
 606             eweps            = _mm256_sub_pd(ewrt,_mm256_round_pd(ewrt, _MM_FROUND_FLOOR));
 607             ewitab           = _mm_slli_epi32(ewitab,2);
 608             ewtabF           = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,0) );
 609             ewtabD           = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,1) );
 610             ewtabV           = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,2) );
 611             ewtabFn          = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,3) );
 612             GMX_MM256_FULLTRANSPOSE4_PD(ewtabF,ewtabD,ewtabV,ewtabFn);
 613             felec            = _mm256_add_pd(ewtabF,_mm256_mul_pd(eweps,ewtabD));
 614             velec            = _mm256_sub_pd(ewtabV,_mm256_mul_pd(_mm256_mul_pd(ewtabhalfspace,eweps),_mm256_add_pd(ewtabF,felec)));
 615             velec            = _mm256_mul_pd(qq20,_mm256_sub_pd(rinv20,velec));
 616             felec            = _mm256_mul_pd(_mm256_mul_pd(qq20,rinv20),_mm256_sub_pd(rinvsq20,felec));
 617
 618             /* Update potential sum for this i atom from the interaction with this j atom. */
 619             velecsum         = _mm256_add_pd(velecsum,velec);
 620
 621             fscal            = felec;
 622
 623             /* Calculate temporary vectorial force */
 624             tx               = _mm256_mul_pd(fscal,dx20);
 625             ty               = _mm256_mul_pd(fscal,dy20);
 626             tz               = _mm256_mul_pd(fscal,dz20);
 627
 628             /* Update vectorial force */
 629             fix2             = _mm256_add_pd(fix2,tx);
 630             fiy2             = _mm256_add_pd(fiy2,ty);
 631             fiz2             = _mm256_add_pd(fiz2,tz);
 632
 633             fjx0             = _mm256_add_pd(fjx0,tx);
 634             fjy0             = _mm256_add_pd(fjy0,ty);
 635             fjz0             = _mm256_add_pd(fjz0,tz);
 636
 637             /**************************
 638              * CALCULATE INTERACTIONS *
 639              **************************/
 640
 641             r21              = _mm256_mul_pd(rsq21,rinv21);
 642
 643             /* EWALD ELECTROSTATICS */
 644
 645             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 646             ewrt             = _mm256_mul_pd(r21,ewtabscale);
 647             ewitab           = _mm256_cvttpd_epi32(ewrt);
 648             eweps            = _mm256_sub_pd(ewrt,_mm256_round_pd(ewrt, _MM_FROUND_FLOOR));
 649             ewitab           = _mm_slli_epi32(ewitab,2);
 650             ewtabF           = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,0) );
 651             ewtabD           = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,1) );
 652             ewtabV           = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,2) );
 653             ewtabFn          = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,3) );
 654             GMX_MM256_FULLTRANSPOSE4_PD(ewtabF,ewtabD,ewtabV,ewtabFn);
 655             felec            = _mm256_add_pd(ewtabF,_mm256_mul_pd(eweps,ewtabD));
 656             velec            = _mm256_sub_pd(ewtabV,_mm256_mul_pd(_mm256_mul_pd(ewtabhalfspace,eweps),_mm256_add_pd(ewtabF,felec)));
 657             velec            = _mm256_mul_pd(qq21,_mm256_sub_pd(rinv21,velec));
 658             felec            = _mm256_mul_pd(_mm256_mul_pd(qq21,rinv21),_mm256_sub_pd(rinvsq21,felec));
 659
 660             /* Update potential sum for this i atom from the interaction with this j atom. */
 661             velecsum         = _mm256_add_pd(velecsum,velec);
 662
 663             fscal            = felec;
 664
 665             /* Calculate temporary vectorial force */
 666             tx               = _mm256_mul_pd(fscal,dx21);
 667             ty               = _mm256_mul_pd(fscal,dy21);
 668             tz               = _mm256_mul_pd(fscal,dz21);
 669
 670             /* Update vectorial force */
 671             fix2             = _mm256_add_pd(fix2,tx);
 672             fiy2             = _mm256_add_pd(fiy2,ty);
 673             fiz2             = _mm256_add_pd(fiz2,tz);
 674
 675             fjx1             = _mm256_add_pd(fjx1,tx);
 676             fjy1             = _mm256_add_pd(fjy1,ty);
 677             fjz1             = _mm256_add_pd(fjz1,tz);
 678
 679             /**************************
 680              * CALCULATE INTERACTIONS *
 681              **************************/
 682
 683             r22              = _mm256_mul_pd(rsq22,rinv22);
 684
 685             /* EWALD ELECTROSTATICS */
 686
 687             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 688             ewrt             = _mm256_mul_pd(r22,ewtabscale);
 689             ewitab           = _mm256_cvttpd_epi32(ewrt);
 690             eweps            = _mm256_sub_pd(ewrt,_mm256_round_pd(ewrt, _MM_FROUND_FLOOR));
 691             ewitab           = _mm_slli_epi32(ewitab,2);
 692             ewtabF           = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,0) );
 693             ewtabD           = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,1) );
 694             ewtabV           = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,2) );
 695             ewtabFn          = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,3) );
 696             GMX_MM256_FULLTRANSPOSE4_PD(ewtabF,ewtabD,ewtabV,ewtabFn);
 697             felec            = _mm256_add_pd(ewtabF,_mm256_mul_pd(eweps,ewtabD));
 698             velec            = _mm256_sub_pd(ewtabV,_mm256_mul_pd(_mm256_mul_pd(ewtabhalfspace,eweps),_mm256_add_pd(ewtabF,felec)));
 699             velec            = _mm256_mul_pd(qq22,_mm256_sub_pd(rinv22,velec));
 700             felec            = _mm256_mul_pd(_mm256_mul_pd(qq22,rinv22),_mm256_sub_pd(rinvsq22,felec));
 701
 702             /* Update potential sum for this i atom from the interaction with this j atom. */
 703             velecsum         = _mm256_add_pd(velecsum,velec);
 704
 705             fscal            = felec;
 706
 707             /* Calculate temporary vectorial force */
 708             tx               = _mm256_mul_pd(fscal,dx22);
 709             ty               = _mm256_mul_pd(fscal,dy22);
 710             tz               = _mm256_mul_pd(fscal,dz22);
 711
 712             /* Update vectorial force */
 713             fix2             = _mm256_add_pd(fix2,tx);
 714             fiy2             = _mm256_add_pd(fiy2,ty);
 715             fiz2             = _mm256_add_pd(fiz2,tz);
 716
 717             fjx2             = _mm256_add_pd(fjx2,tx);
 718             fjy2             = _mm256_add_pd(fjy2,ty);
 719             fjz2             = _mm256_add_pd(fjz2,tz);
 720
 721             fjptrA             = f+j_coord_offsetA;
 722             fjptrB             = f+j_coord_offsetB;
 723             fjptrC             = f+j_coord_offsetC;
 724             fjptrD             = f+j_coord_offsetD;
 725
 726             gmx_mm256_decrement_3rvec_4ptr_swizzle_pd(fjptrA,fjptrB,fjptrC,fjptrD,
 727                                                       fjx0,fjy0,fjz0,fjx1,fjy1,fjz1,fjx2,fjy2,fjz2);
 728
 729             /* Inner loop uses 400 flops */
 730         }
 731
 732         if(jidx<j_index_end)
 733         {
 734
 735             /* Get j neighbor index, and coordinate index */
 736             jnrlistA         = jjnr[jidx];
 737             jnrlistB         = jjnr[jidx+1];
 738             jnrlistC         = jjnr[jidx+2];
 739             jnrlistD         = jjnr[jidx+3];
 740             /* Sign of each element will be negative for non-real atoms.
 741              * This mask will be 0xFFFFFFFF for dummy entries and 0x0 for real ones,
 742              * so use it as val = _mm_andnot_pd(mask,val) to clear dummy entries.
 743              */
 744             tmpmask0 = gmx_mm_castsi128_ps(_mm_cmplt_epi32(_mm_loadu_si128((const __m128i *)(jjnr+jidx)),_mm_setzero_si128()));
 745
 746             tmpmask1 = _mm_permute_ps(tmpmask0,_GMX_MM_PERMUTE(3,3,2,2));
 747             tmpmask0 = _mm_permute_ps(tmpmask0,_GMX_MM_PERMUTE(1,1,0,0));
 748             dummy_mask = _mm256_castps_pd(gmx_mm256_set_m128(tmpmask1,tmpmask0));
 749
 750             jnrA       = (jnrlistA>=0) ? jnrlistA : 0;
 751             jnrB       = (jnrlistB>=0) ? jnrlistB : 0;
 752             jnrC       = (jnrlistC>=0) ? jnrlistC : 0;
 753             jnrD       = (jnrlistD>=0) ? jnrlistD : 0;
 754             j_coord_offsetA  = DIM*jnrA;
 755             j_coord_offsetB  = DIM*jnrB;
 756             j_coord_offsetC  = DIM*jnrC;
 757             j_coord_offsetD  = DIM*jnrD;
 758
 759             /* load j atom coordinates */
 760             gmx_mm256_load_3rvec_4ptr_swizzle_pd(x+j_coord_offsetA,x+j_coord_offsetB,
 761                                                  x+j_coord_offsetC,x+j_coord_offsetD,
 762                                               &jx0,&jy0,&jz0,&jx1,&jy1,&jz1,&jx2,&jy2,&jz2);
 763
 764             /* Calculate displacement vector */
 765             dx00             = _mm256_sub_pd(ix0,jx0);
 766             dy00             = _mm256_sub_pd(iy0,jy0);
 767             dz00             = _mm256_sub_pd(iz0,jz0);
 768             dx01             = _mm256_sub_pd(ix0,jx1);
 769             dy01             = _mm256_sub_pd(iy0,jy1);
 770             dz01             = _mm256_sub_pd(iz0,jz1);
 771             dx02             = _mm256_sub_pd(ix0,jx2);
 772             dy02             = _mm256_sub_pd(iy0,jy2);
 773             dz02             = _mm256_sub_pd(iz0,jz2);
 774             dx10             = _mm256_sub_pd(ix1,jx0);
 775             dy10             = _mm256_sub_pd(iy1,jy0);
 776             dz10             = _mm256_sub_pd(iz1,jz0);
 777             dx11             = _mm256_sub_pd(ix1,jx1);
 778             dy11             = _mm256_sub_pd(iy1,jy1);
 779             dz11             = _mm256_sub_pd(iz1,jz1);
 780             dx12             = _mm256_sub_pd(ix1,jx2);
 781             dy12             = _mm256_sub_pd(iy1,jy2);
 782             dz12             = _mm256_sub_pd(iz1,jz2);
 783             dx20             = _mm256_sub_pd(ix2,jx0);
 784             dy20             = _mm256_sub_pd(iy2,jy0);
 785             dz20             = _mm256_sub_pd(iz2,jz0);
 786             dx21             = _mm256_sub_pd(ix2,jx1);
 787             dy21             = _mm256_sub_pd(iy2,jy1);
 788             dz21             = _mm256_sub_pd(iz2,jz1);
 789             dx22             = _mm256_sub_pd(ix2,jx2);
 790             dy22             = _mm256_sub_pd(iy2,jy2);
 791             dz22             = _mm256_sub_pd(iz2,jz2);
 792
 793             /* Calculate squared distance and things based on it */
 794             rsq00            = gmx_mm256_calc_rsq_pd(dx00,dy00,dz00);
 795             rsq01            = gmx_mm256_calc_rsq_pd(dx01,dy01,dz01);
 796             rsq02            = gmx_mm256_calc_rsq_pd(dx02,dy02,dz02);
 797             rsq10            = gmx_mm256_calc_rsq_pd(dx10,dy10,dz10);
 798             rsq11            = gmx_mm256_calc_rsq_pd(dx11,dy11,dz11);
 799             rsq12            = gmx_mm256_calc_rsq_pd(dx12,dy12,dz12);
 800             rsq20            = gmx_mm256_calc_rsq_pd(dx20,dy20,dz20);
 801             rsq21            = gmx_mm256_calc_rsq_pd(dx21,dy21,dz21);
 802             rsq22            = gmx_mm256_calc_rsq_pd(dx22,dy22,dz22);
 803
 804             rinv00           = avx256_invsqrt_d(rsq00);
 805             rinv01           = avx256_invsqrt_d(rsq01);
 806             rinv02           = avx256_invsqrt_d(rsq02);
 807             rinv10           = avx256_invsqrt_d(rsq10);
 808             rinv11           = avx256_invsqrt_d(rsq11);
 809             rinv12           = avx256_invsqrt_d(rsq12);
 810             rinv20           = avx256_invsqrt_d(rsq20);
 811             rinv21           = avx256_invsqrt_d(rsq21);
 812             rinv22           = avx256_invsqrt_d(rsq22);
 813
 814             rinvsq00         = _mm256_mul_pd(rinv00,rinv00);
 815             rinvsq01         = _mm256_mul_pd(rinv01,rinv01);
 816             rinvsq02         = _mm256_mul_pd(rinv02,rinv02);
 817             rinvsq10         = _mm256_mul_pd(rinv10,rinv10);
 818             rinvsq11         = _mm256_mul_pd(rinv11,rinv11);
 819             rinvsq12         = _mm256_mul_pd(rinv12,rinv12);
 820             rinvsq20         = _mm256_mul_pd(rinv20,rinv20);
 821             rinvsq21         = _mm256_mul_pd(rinv21,rinv21);
 822             rinvsq22         = _mm256_mul_pd(rinv22,rinv22);
 823
 824             fjx0             = _mm256_setzero_pd();
 825             fjy0             = _mm256_setzero_pd();
 826             fjz0             = _mm256_setzero_pd();
 827             fjx1             = _mm256_setzero_pd();
 828             fjy1             = _mm256_setzero_pd();
 829             fjz1             = _mm256_setzero_pd();
 830             fjx2             = _mm256_setzero_pd();
 831             fjy2             = _mm256_setzero_pd();
 832             fjz2             = _mm256_setzero_pd();
 833
 834             /**************************
 835              * CALCULATE INTERACTIONS *
 836              **************************/
 837
 838             r00              = _mm256_mul_pd(rsq00,rinv00);
 839             r00              = _mm256_andnot_pd(dummy_mask,r00);
 840
 841             /* EWALD ELECTROSTATICS */
 842
 843             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 844             ewrt             = _mm256_mul_pd(r00,ewtabscale);
 845             ewitab           = _mm256_cvttpd_epi32(ewrt);
 846             eweps            = _mm256_sub_pd(ewrt,_mm256_round_pd(ewrt, _MM_FROUND_FLOOR));
 847             ewitab           = _mm_slli_epi32(ewitab,2);
 848             ewtabF           = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,0) );
 849             ewtabD           = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,1) );
 850             ewtabV           = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,2) );
 851             ewtabFn          = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,3) );
 852             GMX_MM256_FULLTRANSPOSE4_PD(ewtabF,ewtabD,ewtabV,ewtabFn);
 853             felec            = _mm256_add_pd(ewtabF,_mm256_mul_pd(eweps,ewtabD));
 854             velec            = _mm256_sub_pd(ewtabV,_mm256_mul_pd(_mm256_mul_pd(ewtabhalfspace,eweps),_mm256_add_pd(ewtabF,felec)));
 855             velec            = _mm256_mul_pd(qq00,_mm256_sub_pd(rinv00,velec));
 856             felec            = _mm256_mul_pd(_mm256_mul_pd(qq00,rinv00),_mm256_sub_pd(rinvsq00,felec));
 857
 858             /* Analytical LJ-PME */
 859             rinvsix          = _mm256_mul_pd(_mm256_mul_pd(rinvsq00,rinvsq00),rinvsq00);
 860             ewcljrsq         = _mm256_mul_pd(ewclj2,rsq00);
 861             ewclj6           = _mm256_mul_pd(ewclj2,_mm256_mul_pd(ewclj2,ewclj2));
 862             exponent         = avx256_exp_d(ewcljrsq);
 863             /* poly = exp(-(beta*r)^2) * (1 + (beta*r)^2 + (beta*r)^4 /2) */
 864             poly             = _mm256_mul_pd(exponent,_mm256_add_pd(_mm256_sub_pd(one,ewcljrsq),_mm256_mul_pd(_mm256_mul_pd(ewcljrsq,ewcljrsq),one_half)));
 865             /* vvdw6 = [C6 - C6grid * (1-poly)]/r6 */
 866             vvdw6            = _mm256_mul_pd(_mm256_sub_pd(c6_00,_mm256_mul_pd(c6grid_00,_mm256_sub_pd(one,poly))),rinvsix);
 867             vvdw12           = _mm256_mul_pd(c12_00,_mm256_mul_pd(rinvsix,rinvsix));
 868             vvdw             = _mm256_sub_pd(_mm256_mul_pd(vvdw12,one_twelfth),_mm256_mul_pd(vvdw6,one_sixth));
 869             /* fvdw = vvdw12/r - (vvdw6/r + (C6grid * exponent * beta^6)/r) */
 870             fvdw             = _mm256_mul_pd(_mm256_sub_pd(vvdw12,_mm256_sub_pd(vvdw6,_mm256_mul_pd(_mm256_mul_pd(c6grid_00,one_sixth),_mm256_mul_pd(exponent,ewclj6)))),rinvsq00);
 871
 872             /* Update potential sum for this i atom from the interaction with this j atom. */
 873             velec            = _mm256_andnot_pd(dummy_mask,velec);
 874             velecsum         = _mm256_add_pd(velecsum,velec);
 875             vvdw             = _mm256_andnot_pd(dummy_mask,vvdw);
 876             vvdwsum          = _mm256_add_pd(vvdwsum,vvdw);
 877
 878             fscal            = _mm256_add_pd(felec,fvdw);
 879
 880             fscal            = _mm256_andnot_pd(dummy_mask,fscal);
 881
 882             /* Calculate temporary vectorial force */
 883             tx               = _mm256_mul_pd(fscal,dx00);
 884             ty               = _mm256_mul_pd(fscal,dy00);
 885             tz               = _mm256_mul_pd(fscal,dz00);
 886
 887             /* Update vectorial force */
 888             fix0             = _mm256_add_pd(fix0,tx);
 889             fiy0             = _mm256_add_pd(fiy0,ty);
 890             fiz0             = _mm256_add_pd(fiz0,tz);
 891
 892             fjx0             = _mm256_add_pd(fjx0,tx);
 893             fjy0             = _mm256_add_pd(fjy0,ty);
 894             fjz0             = _mm256_add_pd(fjz0,tz);
 895
 896             /**************************
 897              * CALCULATE INTERACTIONS *
 898              **************************/
 899
 900             r01              = _mm256_mul_pd(rsq01,rinv01);
 901             r01              = _mm256_andnot_pd(dummy_mask,r01);
 902
 903             /* EWALD ELECTROSTATICS */
 904
 905             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 906             ewrt             = _mm256_mul_pd(r01,ewtabscale);
 907             ewitab           = _mm256_cvttpd_epi32(ewrt);
 908             eweps            = _mm256_sub_pd(ewrt,_mm256_round_pd(ewrt, _MM_FROUND_FLOOR));
 909             ewitab           = _mm_slli_epi32(ewitab,2);
 910             ewtabF           = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,0) );
 911             ewtabD           = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,1) );
 912             ewtabV           = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,2) );
 913             ewtabFn          = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,3) );
 914             GMX_MM256_FULLTRANSPOSE4_PD(ewtabF,ewtabD,ewtabV,ewtabFn);
 915             felec            = _mm256_add_pd(ewtabF,_mm256_mul_pd(eweps,ewtabD));
 916             velec            = _mm256_sub_pd(ewtabV,_mm256_mul_pd(_mm256_mul_pd(ewtabhalfspace,eweps),_mm256_add_pd(ewtabF,felec)));
 917             velec            = _mm256_mul_pd(qq01,_mm256_sub_pd(rinv01,velec));
 918             felec            = _mm256_mul_pd(_mm256_mul_pd(qq01,rinv01),_mm256_sub_pd(rinvsq01,felec));
 919
 920             /* Update potential sum for this i atom from the interaction with this j atom. */
 921             velec            = _mm256_andnot_pd(dummy_mask,velec);
 922             velecsum         = _mm256_add_pd(velecsum,velec);
 923
 924             fscal            = felec;
 925
 926             fscal            = _mm256_andnot_pd(dummy_mask,fscal);
 927
 928             /* Calculate temporary vectorial force */
 929             tx               = _mm256_mul_pd(fscal,dx01);
 930             ty               = _mm256_mul_pd(fscal,dy01);
 931             tz               = _mm256_mul_pd(fscal,dz01);
 932
 933             /* Update vectorial force */
 934             fix0             = _mm256_add_pd(fix0,tx);
 935             fiy0             = _mm256_add_pd(fiy0,ty);
 936             fiz0             = _mm256_add_pd(fiz0,tz);
 937
 938             fjx1             = _mm256_add_pd(fjx1,tx);
 939             fjy1             = _mm256_add_pd(fjy1,ty);
 940             fjz1             = _mm256_add_pd(fjz1,tz);
 941
 942             /**************************
 943              * CALCULATE INTERACTIONS *
 944              **************************/
 945
 946             r02              = _mm256_mul_pd(rsq02,rinv02);
 947             r02              = _mm256_andnot_pd(dummy_mask,r02);
 948
 949             /* EWALD ELECTROSTATICS */
 950
 951             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 952             ewrt             = _mm256_mul_pd(r02,ewtabscale);
 953             ewitab           = _mm256_cvttpd_epi32(ewrt);
 954             eweps            = _mm256_sub_pd(ewrt,_mm256_round_pd(ewrt, _MM_FROUND_FLOOR));
 955             ewitab           = _mm_slli_epi32(ewitab,2);
 956             ewtabF           = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,0) );
 957             ewtabD           = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,1) );
 958             ewtabV           = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,2) );
 959             ewtabFn          = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,3) );
 960             GMX_MM256_FULLTRANSPOSE4_PD(ewtabF,ewtabD,ewtabV,ewtabFn);
 961             felec            = _mm256_add_pd(ewtabF,_mm256_mul_pd(eweps,ewtabD));
 962             velec            = _mm256_sub_pd(ewtabV,_mm256_mul_pd(_mm256_mul_pd(ewtabhalfspace,eweps),_mm256_add_pd(ewtabF,felec)));
 963             velec            = _mm256_mul_pd(qq02,_mm256_sub_pd(rinv02,velec));
 964             felec            = _mm256_mul_pd(_mm256_mul_pd(qq02,rinv02),_mm256_sub_pd(rinvsq02,felec));
 965
 966             /* Update potential sum for this i atom from the interaction with this j atom. */
 967             velec            = _mm256_andnot_pd(dummy_mask,velec);
 968             velecsum         = _mm256_add_pd(velecsum,velec);
 969
 970             fscal            = felec;
 971
 972             fscal            = _mm256_andnot_pd(dummy_mask,fscal);
 973
 974             /* Calculate temporary vectorial force */
 975             tx               = _mm256_mul_pd(fscal,dx02);
 976             ty               = _mm256_mul_pd(fscal,dy02);
 977             tz               = _mm256_mul_pd(fscal,dz02);
 978
 979             /* Update vectorial force */
 980             fix0             = _mm256_add_pd(fix0,tx);
 981             fiy0             = _mm256_add_pd(fiy0,ty);
 982             fiz0             = _mm256_add_pd(fiz0,tz);
 983
 984             fjx2             = _mm256_add_pd(fjx2,tx);
 985             fjy2             = _mm256_add_pd(fjy2,ty);
 986             fjz2             = _mm256_add_pd(fjz2,tz);
 987
 988             /**************************
 989              * CALCULATE INTERACTIONS *
 990              **************************/
 991
 992             r10              = _mm256_mul_pd(rsq10,rinv10);
 993             r10              = _mm256_andnot_pd(dummy_mask,r10);
 994
 995             /* EWALD ELECTROSTATICS */
 996
 997             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 998             ewrt             = _mm256_mul_pd(r10,ewtabscale);
 999             ewitab           = _mm256_cvttpd_epi32(ewrt);
1000             eweps            = _mm256_sub_pd(ewrt,_mm256_round_pd(ewrt, _MM_FROUND_FLOOR));
1001             ewitab           = _mm_slli_epi32(ewitab,2);
1002             ewtabF           = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,0) );
1003             ewtabD           = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,1) );
1004             ewtabV           = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,2) );
1005             ewtabFn          = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,3) );
1006             GMX_MM256_FULLTRANSPOSE4_PD(ewtabF,ewtabD,ewtabV,ewtabFn);
1007             felec            = _mm256_add_pd(ewtabF,_mm256_mul_pd(eweps,ewtabD));
1008             velec            = _mm256_sub_pd(ewtabV,_mm256_mul_pd(_mm256_mul_pd(ewtabhalfspace,eweps),_mm256_add_pd(ewtabF,felec)));
1009             velec            = _mm256_mul_pd(qq10,_mm256_sub_pd(rinv10,velec));
1010             felec            = _mm256_mul_pd(_mm256_mul_pd(qq10,rinv10),_mm256_sub_pd(rinvsq10,felec));
1011
1012             /* Update potential sum for this i atom from the interaction with this j atom. */
1013             velec            = _mm256_andnot_pd(dummy_mask,velec);
1014             velecsum         = _mm256_add_pd(velecsum,velec);
1015
1016             fscal            = felec;
1017
1018             fscal            = _mm256_andnot_pd(dummy_mask,fscal);
1019
1020             /* Calculate temporary vectorial force */
1021             tx               = _mm256_mul_pd(fscal,dx10);
1022             ty               = _mm256_mul_pd(fscal,dy10);
1023             tz               = _mm256_mul_pd(fscal,dz10);
1024
1025             /* Update vectorial force */
1026             fix1             = _mm256_add_pd(fix1,tx);
1027             fiy1             = _mm256_add_pd(fiy1,ty);
1028             fiz1             = _mm256_add_pd(fiz1,tz);
1029
1030             fjx0             = _mm256_add_pd(fjx0,tx);
1031             fjy0             = _mm256_add_pd(fjy0,ty);
1032             fjz0             = _mm256_add_pd(fjz0,tz);
1033
1034             /**************************
1035              * CALCULATE INTERACTIONS *
1036              **************************/
1037
1038             r11              = _mm256_mul_pd(rsq11,rinv11);
1039             r11              = _mm256_andnot_pd(dummy_mask,r11);
1040
1041             /* EWALD ELECTROSTATICS */
1042
1043             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1044             ewrt             = _mm256_mul_pd(r11,ewtabscale);
1045             ewitab           = _mm256_cvttpd_epi32(ewrt);
1046             eweps            = _mm256_sub_pd(ewrt,_mm256_round_pd(ewrt, _MM_FROUND_FLOOR));
1047             ewitab           = _mm_slli_epi32(ewitab,2);
1048             ewtabF           = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,0) );
1049             ewtabD           = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,1) );
1050             ewtabV           = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,2) );
1051             ewtabFn          = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,3) );
1052             GMX_MM256_FULLTRANSPOSE4_PD(ewtabF,ewtabD,ewtabV,ewtabFn);
1053             felec            = _mm256_add_pd(ewtabF,_mm256_mul_pd(eweps,ewtabD));
1054             velec            = _mm256_sub_pd(ewtabV,_mm256_mul_pd(_mm256_mul_pd(ewtabhalfspace,eweps),_mm256_add_pd(ewtabF,felec)));
1055             velec            = _mm256_mul_pd(qq11,_mm256_sub_pd(rinv11,velec));
1056             felec            = _mm256_mul_pd(_mm256_mul_pd(qq11,rinv11),_mm256_sub_pd(rinvsq11,felec));
1057
1058             /* Update potential sum for this i atom from the interaction with this j atom. */
1059             velec            = _mm256_andnot_pd(dummy_mask,velec);
1060             velecsum         = _mm256_add_pd(velecsum,velec);
1061
1062             fscal            = felec;
1063
1064             fscal            = _mm256_andnot_pd(dummy_mask,fscal);
1065
1066             /* Calculate temporary vectorial force */
1067             tx               = _mm256_mul_pd(fscal,dx11);
1068             ty               = _mm256_mul_pd(fscal,dy11);
1069             tz               = _mm256_mul_pd(fscal,dz11);
1070
1071             /* Update vectorial force */
1072             fix1             = _mm256_add_pd(fix1,tx);
1073             fiy1             = _mm256_add_pd(fiy1,ty);
1074             fiz1             = _mm256_add_pd(fiz1,tz);
1075
1076             fjx1             = _mm256_add_pd(fjx1,tx);
1077             fjy1             = _mm256_add_pd(fjy1,ty);
1078             fjz1             = _mm256_add_pd(fjz1,tz);
1079
1080             /**************************
1081              * CALCULATE INTERACTIONS *
1082              **************************/
1083
1084             r12              = _mm256_mul_pd(rsq12,rinv12);
1085             r12              = _mm256_andnot_pd(dummy_mask,r12);
1086
1087             /* EWALD ELECTROSTATICS */
1088
1089             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1090             ewrt             = _mm256_mul_pd(r12,ewtabscale);
1091             ewitab           = _mm256_cvttpd_epi32(ewrt);
1092             eweps            = _mm256_sub_pd(ewrt,_mm256_round_pd(ewrt, _MM_FROUND_FLOOR));
1093             ewitab           = _mm_slli_epi32(ewitab,2);
1094             ewtabF           = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,0) );
1095             ewtabD           = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,1) );
1096             ewtabV           = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,2) );
1097             ewtabFn          = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,3) );
1098             GMX_MM256_FULLTRANSPOSE4_PD(ewtabF,ewtabD,ewtabV,ewtabFn);
1099             felec            = _mm256_add_pd(ewtabF,_mm256_mul_pd(eweps,ewtabD));
1100             velec            = _mm256_sub_pd(ewtabV,_mm256_mul_pd(_mm256_mul_pd(ewtabhalfspace,eweps),_mm256_add_pd(ewtabF,felec)));
1101             velec            = _mm256_mul_pd(qq12,_mm256_sub_pd(rinv12,velec));
1102             felec            = _mm256_mul_pd(_mm256_mul_pd(qq12,rinv12),_mm256_sub_pd(rinvsq12,felec));
1103
1104             /* Update potential sum for this i atom from the interaction with this j atom. */
1105             velec            = _mm256_andnot_pd(dummy_mask,velec);
1106             velecsum         = _mm256_add_pd(velecsum,velec);
1107
1108             fscal            = felec;
1109
1110             fscal            = _mm256_andnot_pd(dummy_mask,fscal);
1111
1112             /* Calculate temporary vectorial force */
1113             tx               = _mm256_mul_pd(fscal,dx12);
1114             ty               = _mm256_mul_pd(fscal,dy12);
1115             tz               = _mm256_mul_pd(fscal,dz12);
1116
1117             /* Update vectorial force */
1118             fix1             = _mm256_add_pd(fix1,tx);
1119             fiy1             = _mm256_add_pd(fiy1,ty);
1120             fiz1             = _mm256_add_pd(fiz1,tz);
1121
1122             fjx2             = _mm256_add_pd(fjx2,tx);
1123             fjy2             = _mm256_add_pd(fjy2,ty);
1124             fjz2             = _mm256_add_pd(fjz2,tz);
1125
1126             /**************************
1127              * CALCULATE INTERACTIONS *
1128              **************************/
1129
1130             r20              = _mm256_mul_pd(rsq20,rinv20);
1131             r20              = _mm256_andnot_pd(dummy_mask,r20);
1132
1133             /* EWALD ELECTROSTATICS */
1134
1135             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1136             ewrt             = _mm256_mul_pd(r20,ewtabscale);
1137             ewitab           = _mm256_cvttpd_epi32(ewrt);
1138             eweps            = _mm256_sub_pd(ewrt,_mm256_round_pd(ewrt, _MM_FROUND_FLOOR));
1139             ewitab           = _mm_slli_epi32(ewitab,2);
1140             ewtabF           = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,0) );
1141             ewtabD           = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,1) );
1142             ewtabV           = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,2) );
1143             ewtabFn          = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,3) );
1144             GMX_MM256_FULLTRANSPOSE4_PD(ewtabF,ewtabD,ewtabV,ewtabFn);
1145             felec            = _mm256_add_pd(ewtabF,_mm256_mul_pd(eweps,ewtabD));
1146             velec            = _mm256_sub_pd(ewtabV,_mm256_mul_pd(_mm256_mul_pd(ewtabhalfspace,eweps),_mm256_add_pd(ewtabF,felec)));
1147             velec            = _mm256_mul_pd(qq20,_mm256_sub_pd(rinv20,velec));
1148             felec            = _mm256_mul_pd(_mm256_mul_pd(qq20,rinv20),_mm256_sub_pd(rinvsq20,felec));
1149
1150             /* Update potential sum for this i atom from the interaction with this j atom. */
1151             velec            = _mm256_andnot_pd(dummy_mask,velec);
1152             velecsum         = _mm256_add_pd(velecsum,velec);
1153
1154             fscal            = felec;
1155
1156             fscal            = _mm256_andnot_pd(dummy_mask,fscal);
1157
1158             /* Calculate temporary vectorial force */
1159             tx               = _mm256_mul_pd(fscal,dx20);
1160             ty               = _mm256_mul_pd(fscal,dy20);
1161             tz               = _mm256_mul_pd(fscal,dz20);
1162
1163             /* Update vectorial force */
1164             fix2             = _mm256_add_pd(fix2,tx);
1165             fiy2             = _mm256_add_pd(fiy2,ty);
1166             fiz2             = _mm256_add_pd(fiz2,tz);
1167
1168             fjx0             = _mm256_add_pd(fjx0,tx);
1169             fjy0             = _mm256_add_pd(fjy0,ty);
1170             fjz0             = _mm256_add_pd(fjz0,tz);
1171
1172             /**************************
1173              * CALCULATE INTERACTIONS *
1174              **************************/
1175
1176             r21              = _mm256_mul_pd(rsq21,rinv21);
1177             r21              = _mm256_andnot_pd(dummy_mask,r21);
1178
1179             /* EWALD ELECTROSTATICS */
1180
1181             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1182             ewrt             = _mm256_mul_pd(r21,ewtabscale);
1183             ewitab           = _mm256_cvttpd_epi32(ewrt);
1184             eweps            = _mm256_sub_pd(ewrt,_mm256_round_pd(ewrt, _MM_FROUND_FLOOR));
1185             ewitab           = _mm_slli_epi32(ewitab,2);
1186             ewtabF           = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,0) );
1187             ewtabD           = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,1) );
1188             ewtabV           = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,2) );
1189             ewtabFn          = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,3) );
1190             GMX_MM256_FULLTRANSPOSE4_PD(ewtabF,ewtabD,ewtabV,ewtabFn);
1191             felec            = _mm256_add_pd(ewtabF,_mm256_mul_pd(eweps,ewtabD));
1192             velec            = _mm256_sub_pd(ewtabV,_mm256_mul_pd(_mm256_mul_pd(ewtabhalfspace,eweps),_mm256_add_pd(ewtabF,felec)));
1193             velec            = _mm256_mul_pd(qq21,_mm256_sub_pd(rinv21,velec));
1194             felec            = _mm256_mul_pd(_mm256_mul_pd(qq21,rinv21),_mm256_sub_pd(rinvsq21,felec));
1195
1196             /* Update potential sum for this i atom from the interaction with this j atom. */
1197             velec            = _mm256_andnot_pd(dummy_mask,velec);
1198             velecsum         = _mm256_add_pd(velecsum,velec);
1199
1200             fscal            = felec;
1201
1202             fscal            = _mm256_andnot_pd(dummy_mask,fscal);
1203
1204             /* Calculate temporary vectorial force */
1205             tx               = _mm256_mul_pd(fscal,dx21);
1206             ty               = _mm256_mul_pd(fscal,dy21);
1207             tz               = _mm256_mul_pd(fscal,dz21);
1208
1209             /* Update vectorial force */
1210             fix2             = _mm256_add_pd(fix2,tx);
1211             fiy2             = _mm256_add_pd(fiy2,ty);
1212             fiz2             = _mm256_add_pd(fiz2,tz);
1213
1214             fjx1             = _mm256_add_pd(fjx1,tx);
1215             fjy1             = _mm256_add_pd(fjy1,ty);
1216             fjz1             = _mm256_add_pd(fjz1,tz);
1217
1218             /**************************
1219              * CALCULATE INTERACTIONS *
1220              **************************/
1221
1222             r22              = _mm256_mul_pd(rsq22,rinv22);
1223             r22              = _mm256_andnot_pd(dummy_mask,r22);
1224
1225             /* EWALD ELECTROSTATICS */
1226
1227             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1228             ewrt             = _mm256_mul_pd(r22,ewtabscale);
1229             ewitab           = _mm256_cvttpd_epi32(ewrt);
1230             eweps            = _mm256_sub_pd(ewrt,_mm256_round_pd(ewrt, _MM_FROUND_FLOOR));
1231             ewitab           = _mm_slli_epi32(ewitab,2);
1232             ewtabF           = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,0) );
1233             ewtabD           = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,1) );
1234             ewtabV           = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,2) );
1235             ewtabFn          = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,3) );
1236             GMX_MM256_FULLTRANSPOSE4_PD(ewtabF,ewtabD,ewtabV,ewtabFn);
1237             felec            = _mm256_add_pd(ewtabF,_mm256_mul_pd(eweps,ewtabD));
1238             velec            = _mm256_sub_pd(ewtabV,_mm256_mul_pd(_mm256_mul_pd(ewtabhalfspace,eweps),_mm256_add_pd(ewtabF,felec)));
1239             velec            = _mm256_mul_pd(qq22,_mm256_sub_pd(rinv22,velec));
1240             felec            = _mm256_mul_pd(_mm256_mul_pd(qq22,rinv22),_mm256_sub_pd(rinvsq22,felec));
1241
1242             /* Update potential sum for this i atom from the interaction with this j atom. */
1243             velec            = _mm256_andnot_pd(dummy_mask,velec);
1244             velecsum         = _mm256_add_pd(velecsum,velec);
1245
1246             fscal            = felec;
1247
1248             fscal            = _mm256_andnot_pd(dummy_mask,fscal);
1249
1250             /* Calculate temporary vectorial force */
1251             tx               = _mm256_mul_pd(fscal,dx22);
1252             ty               = _mm256_mul_pd(fscal,dy22);
1253             tz               = _mm256_mul_pd(fscal,dz22);
1254
1255             /* Update vectorial force */
1256             fix2             = _mm256_add_pd(fix2,tx);
1257             fiy2             = _mm256_add_pd(fiy2,ty);
1258             fiz2             = _mm256_add_pd(fiz2,tz);
1259
1260             fjx2             = _mm256_add_pd(fjx2,tx);
1261             fjy2             = _mm256_add_pd(fjy2,ty);
1262             fjz2             = _mm256_add_pd(fjz2,tz);
1263
1264             fjptrA             = (jnrlistA>=0) ? f+j_coord_offsetA : scratch;
1265             fjptrB             = (jnrlistB>=0) ? f+j_coord_offsetB : scratch;
1266             fjptrC             = (jnrlistC>=0) ? f+j_coord_offsetC : scratch;
1267             fjptrD             = (jnrlistD>=0) ? f+j_coord_offsetD : scratch;
1268
1269             gmx_mm256_decrement_3rvec_4ptr_swizzle_pd(fjptrA,fjptrB,fjptrC,fjptrD,
1270                                                       fjx0,fjy0,fjz0,fjx1,fjy1,fjz1,fjx2,fjy2,fjz2);
1271
1272             /* Inner loop uses 409 flops */
1273         }
1274
1275         /* End of innermost loop */
1276
1277         gmx_mm256_update_iforce_3atom_swizzle_pd(fix0,fiy0,fiz0,fix1,fiy1,fiz1,fix2,fiy2,fiz2,
1278                                                  f+i_coord_offset,fshift+i_shift_offset);
1279
1280         ggid                        = gid[iidx];
1281         /* Update potential energies */
1282         gmx_mm256_update_1pot_pd(velecsum,kernel_data->energygrp_elec+ggid);
1283         gmx_mm256_update_1pot_pd(vvdwsum,kernel_data->energygrp_vdw+ggid);
1284
1285         /* Increment number of inner iterations */
1286         inneriter                  += j_index_end - j_index_start;
1287
1288         /* Outer loop uses 20 flops */
1289     }
1290
1291     /* Increment number of outer iterations */
1292     outeriter        += nri;
1293
1294     /* Update outer/inner flops */
1295
1296     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W3W3_VF,outeriter*20 + inneriter*409);
1297 }
1298 /*
1299  * Gromacs nonbonded kernel:   nb_kernel_ElecEw_VdwLJEw_GeomW3W3_F_avx_256_double
1300  * Electrostatics interaction: Ewald
1301  * VdW interaction:            LJEwald
1302  * Geometry:                   Water3-Water3
1303  * Calculate force/pot:        Force
1304  */
1305 void
1306 nb_kernel_ElecEw_VdwLJEw_GeomW3W3_F_avx_256_double
1307                     (t_nblist                    * gmx_restrict       nlist,
1308                      rvec                        * gmx_restrict          xx,
1309                      rvec                        * gmx_restrict          ff,
1310                      struct t_forcerec           * gmx_restrict          fr,
1311                      t_mdatoms                   * gmx_restrict     mdatoms,
1312                      nb_kernel_data_t gmx_unused * gmx_restrict kernel_data,
1313                      t_nrnb                      * gmx_restrict        nrnb)
1314 {
1315     /* Suffixes 0,1,2,3 refer to particle indices for waters in the inner or outer loop, or
1316      * just 0 for non-waters.
1317      * Suffixes A,B,C,D refer to j loop unrolling done with AVX, e.g. for the four different
1318      * jnr indices corresponding to data put in the four positions in the SIMD register.
1319      */
1320     int              i_shift_offset,i_coord_offset,outeriter,inneriter;
1321     int              j_index_start,j_index_end,jidx,nri,inr,ggid,iidx;
1322     int              jnrA,jnrB,jnrC,jnrD;
1323     int              jnrlistA,jnrlistB,jnrlistC,jnrlistD;
1324     int              jnrlistE,jnrlistF,jnrlistG,jnrlistH;
1325     int              j_coord_offsetA,j_coord_offsetB,j_coord_offsetC,j_coord_offsetD;
1326     int              *iinr,*jindex,*jjnr,*shiftidx,*gid;
1327     real             rcutoff_scalar;
1328     real             *shiftvec,*fshift,*x,*f;
1329     real             *fjptrA,*fjptrB,*fjptrC,*fjptrD;
1330     real             scratch[4*DIM];
1331     __m256d          tx,ty,tz,fscal,rcutoff,rcutoff2,jidxall;
1332     real *           vdwioffsetptr0;
1333     real *           vdwgridioffsetptr0;
1334     __m256d          ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
1335     real *           vdwioffsetptr1;
1336     real *           vdwgridioffsetptr1;
1337     __m256d          ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
1338     real *           vdwioffsetptr2;
1339     real *           vdwgridioffsetptr2;
1340     __m256d          ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
1341     int              vdwjidx0A,vdwjidx0B,vdwjidx0C,vdwjidx0D;
1342     __m256d          jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
1343     int              vdwjidx1A,vdwjidx1B,vdwjidx1C,vdwjidx1D;
1344     __m256d          jx1,jy1,jz1,fjx1,fjy1,fjz1,jq1,isaj1;
1345     int              vdwjidx2A,vdwjidx2B,vdwjidx2C,vdwjidx2D;
1346     __m256d          jx2,jy2,jz2,fjx2,fjy2,fjz2,jq2,isaj2;
1347     __m256d          dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00;
1348     __m256d          dx01,dy01,dz01,rsq01,rinv01,rinvsq01,r01,qq01,c6_01,c12_01;
1349     __m256d          dx02,dy02,dz02,rsq02,rinv02,rinvsq02,r02,qq02,c6_02,c12_02;
1350     __m256d          dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10;
1351     __m256d          dx11,dy11,dz11,rsq11,rinv11,rinvsq11,r11,qq11,c6_11,c12_11;
1352     __m256d          dx12,dy12,dz12,rsq12,rinv12,rinvsq12,r12,qq12,c6_12,c12_12;
1353     __m256d          dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20;
1354     __m256d          dx21,dy21,dz21,rsq21,rinv21,rinvsq21,r21,qq21,c6_21,c12_21;
1355     __m256d          dx22,dy22,dz22,rsq22,rinv22,rinvsq22,r22,qq22,c6_22,c12_22;
1356     __m256d          velec,felec,velecsum,facel,crf,krf,krf2;
1357     real             *charge;
1358     int              nvdwtype;
1359     __m256d          rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,sh_vdw_invrcut6;
1360     int              *vdwtype;
1361     real             *vdwparam;
1362     __m256d          one_sixth   = _mm256_set1_pd(1.0/6.0);
1363     __m256d          one_twelfth = _mm256_set1_pd(1.0/12.0);
1364     __m256d           c6grid_00;
1365     __m256d           c6grid_01;
1366     __m256d           c6grid_02;
1367     __m256d           c6grid_10;
1368     __m256d           c6grid_11;
1369     __m256d           c6grid_12;
1370     __m256d           c6grid_20;
1371     __m256d           c6grid_21;
1372     __m256d           c6grid_22;
1373     real             *vdwgridparam;
1374     __m256d           ewclj,ewclj2,ewclj6,ewcljrsq,poly,exponent,f6A,f6B,sh_lj_ewald;
1375     __m256d           one_half  = _mm256_set1_pd(0.5);
1376     __m256d           minus_one = _mm256_set1_pd(-1.0);
1377     __m128i          ewitab;
1378     __m256d          ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace,ewtabF,ewtabFn,ewtabD,ewtabV;
1379     __m256d          beta,beta2,beta3,zeta2,pmecorrF,pmecorrV,rinv3;
1380     real             *ewtab;
1381     __m256d          dummy_mask,cutoff_mask;
1382     __m128           tmpmask0,tmpmask1;
1383     __m256d          signbit = _mm256_castsi256_pd( _mm256_set1_epi32(0x80000000) );
1384     __m256d          one     = _mm256_set1_pd(1.0);
1385     __m256d          two     = _mm256_set1_pd(2.0);
1386     x                = xx[0];
1387     f                = ff[0];
1388
1389     nri              = nlist->nri;
1390     iinr             = nlist->iinr;
1391     jindex           = nlist->jindex;
1392     jjnr             = nlist->jjnr;
1393     shiftidx         = nlist->shift;
1394     gid              = nlist->gid;
1395     shiftvec         = fr->shift_vec[0];
1396     fshift           = fr->fshift[0];
1397     facel            = _mm256_set1_pd(fr->ic->epsfac);
1398     charge           = mdatoms->chargeA;
1399     nvdwtype         = fr->ntype;
1400     vdwparam         = fr->nbfp;
1401     vdwtype          = mdatoms->typeA;
1402     vdwgridparam     = fr->ljpme_c6grid;
1403     sh_lj_ewald      = _mm256_set1_pd(fr->ic->sh_lj_ewald);
1404     ewclj            = _mm256_set1_pd(fr->ic->ewaldcoeff_lj);
1405     ewclj2           = _mm256_mul_pd(minus_one,_mm256_mul_pd(ewclj,ewclj));
1406
1407     sh_ewald         = _mm256_set1_pd(fr->ic->sh_ewald);
1408     beta             = _mm256_set1_pd(fr->ic->ewaldcoeff_q);
1409     beta2            = _mm256_mul_pd(beta,beta);
1410     beta3            = _mm256_mul_pd(beta,beta2);
1411
1412     ewtab            = fr->ic->tabq_coul_F;
1413     ewtabscale       = _mm256_set1_pd(fr->ic->tabq_scale);
1414     ewtabhalfspace   = _mm256_set1_pd(0.5/fr->ic->tabq_scale);
1415
1416     /* Setup water-specific parameters */
1417     inr              = nlist->iinr[0];
1418     iq0              = _mm256_mul_pd(facel,_mm256_set1_pd(charge[inr+0]));
1419     iq1              = _mm256_mul_pd(facel,_mm256_set1_pd(charge[inr+1]));
1420     iq2              = _mm256_mul_pd(facel,_mm256_set1_pd(charge[inr+2]));
1421     vdwioffsetptr0   = vdwparam+2*nvdwtype*vdwtype[inr+0];
1422     vdwgridioffsetptr0 = vdwgridparam+2*nvdwtype*vdwtype[inr+0];
1423
1424     jq0              = _mm256_set1_pd(charge[inr+0]);
1425     jq1              = _mm256_set1_pd(charge[inr+1]);
1426     jq2              = _mm256_set1_pd(charge[inr+2]);
1427     vdwjidx0A        = 2*vdwtype[inr+0];
1428     qq00             = _mm256_mul_pd(iq0,jq0);
1429     c6_00            = _mm256_set1_pd(vdwioffsetptr0[vdwjidx0A]);
1430     c12_00           = _mm256_set1_pd(vdwioffsetptr0[vdwjidx0A+1]);
1431     c6grid_00        = _mm256_set1_pd(vdwgridioffsetptr0[vdwjidx0A]);
1432     qq01             = _mm256_mul_pd(iq0,jq1);
1433     qq02             = _mm256_mul_pd(iq0,jq2);
1434     qq10             = _mm256_mul_pd(iq1,jq0);
1435     qq11             = _mm256_mul_pd(iq1,jq1);
1436     qq12             = _mm256_mul_pd(iq1,jq2);
1437     qq20             = _mm256_mul_pd(iq2,jq0);
1438     qq21             = _mm256_mul_pd(iq2,jq1);
1439     qq22             = _mm256_mul_pd(iq2,jq2);
1440
1441     /* Avoid stupid compiler warnings */
1442     jnrA = jnrB = jnrC = jnrD = 0;
1443     j_coord_offsetA = 0;
1444     j_coord_offsetB = 0;
1445     j_coord_offsetC = 0;
1446     j_coord_offsetD = 0;
1447
1448     outeriter        = 0;
1449     inneriter        = 0;
1450
1451     for(iidx=0;iidx<4*DIM;iidx++)
1452     {
1453         scratch[iidx] = 0.0;
1454     }
1455
1456     /* Start outer loop over neighborlists */
1457     for(iidx=0; iidx<nri; iidx++)
1458     {
1459         /* Load shift vector for this list */
1460         i_shift_offset   = DIM*shiftidx[iidx];
1461
1462         /* Load limits for loop over neighbors */
1463         j_index_start    = jindex[iidx];
1464         j_index_end      = jindex[iidx+1];
1465
1466         /* Get outer coordinate index */
1467         inr              = iinr[iidx];
1468         i_coord_offset   = DIM*inr;
1469
1470         /* Load i particle coords and add shift vector */
1471         gmx_mm256_load_shift_and_3rvec_broadcast_pd(shiftvec+i_shift_offset,x+i_coord_offset,
1472                                                     &ix0,&iy0,&iz0,&ix1,&iy1,&iz1,&ix2,&iy2,&iz2);
1473
1474         fix0             = _mm256_setzero_pd();
1475         fiy0             = _mm256_setzero_pd();
1476         fiz0             = _mm256_setzero_pd();
1477         fix1             = _mm256_setzero_pd();
1478         fiy1             = _mm256_setzero_pd();
1479         fiz1             = _mm256_setzero_pd();
1480         fix2             = _mm256_setzero_pd();
1481         fiy2             = _mm256_setzero_pd();
1482         fiz2             = _mm256_setzero_pd();
1483
1484         /* Start inner kernel loop */
1485         for(jidx=j_index_start; jidx<j_index_end && jjnr[jidx+3]>=0; jidx+=4)
1486         {
1487
1488             /* Get j neighbor index, and coordinate index */
1489             jnrA             = jjnr[jidx];
1490             jnrB             = jjnr[jidx+1];
1491             jnrC             = jjnr[jidx+2];
1492             jnrD             = jjnr[jidx+3];
1493             j_coord_offsetA  = DIM*jnrA;
1494             j_coord_offsetB  = DIM*jnrB;
1495             j_coord_offsetC  = DIM*jnrC;
1496             j_coord_offsetD  = DIM*jnrD;
1497
1498             /* load j atom coordinates */
1499             gmx_mm256_load_3rvec_4ptr_swizzle_pd(x+j_coord_offsetA,x+j_coord_offsetB,
1500                                                  x+j_coord_offsetC,x+j_coord_offsetD,
1501                                               &jx0,&jy0,&jz0,&jx1,&jy1,&jz1,&jx2,&jy2,&jz2);
1502
1503             /* Calculate displacement vector */
1504             dx00             = _mm256_sub_pd(ix0,jx0);
1505             dy00             = _mm256_sub_pd(iy0,jy0);
1506             dz00             = _mm256_sub_pd(iz0,jz0);
1507             dx01             = _mm256_sub_pd(ix0,jx1);
1508             dy01             = _mm256_sub_pd(iy0,jy1);
1509             dz01             = _mm256_sub_pd(iz0,jz1);
1510             dx02             = _mm256_sub_pd(ix0,jx2);
1511             dy02             = _mm256_sub_pd(iy0,jy2);
1512             dz02             = _mm256_sub_pd(iz0,jz2);
1513             dx10             = _mm256_sub_pd(ix1,jx0);
1514             dy10             = _mm256_sub_pd(iy1,jy0);
1515             dz10             = _mm256_sub_pd(iz1,jz0);
1516             dx11             = _mm256_sub_pd(ix1,jx1);
1517             dy11             = _mm256_sub_pd(iy1,jy1);
1518             dz11             = _mm256_sub_pd(iz1,jz1);
1519             dx12             = _mm256_sub_pd(ix1,jx2);
1520             dy12             = _mm256_sub_pd(iy1,jy2);
1521             dz12             = _mm256_sub_pd(iz1,jz2);
1522             dx20             = _mm256_sub_pd(ix2,jx0);
1523             dy20             = _mm256_sub_pd(iy2,jy0);
1524             dz20             = _mm256_sub_pd(iz2,jz0);
1525             dx21             = _mm256_sub_pd(ix2,jx1);
1526             dy21             = _mm256_sub_pd(iy2,jy1);
1527             dz21             = _mm256_sub_pd(iz2,jz1);
1528             dx22             = _mm256_sub_pd(ix2,jx2);
1529             dy22             = _mm256_sub_pd(iy2,jy2);
1530             dz22             = _mm256_sub_pd(iz2,jz2);
1531
1532             /* Calculate squared distance and things based on it */
1533             rsq00            = gmx_mm256_calc_rsq_pd(dx00,dy00,dz00);
1534             rsq01            = gmx_mm256_calc_rsq_pd(dx01,dy01,dz01);
1535             rsq02            = gmx_mm256_calc_rsq_pd(dx02,dy02,dz02);
1536             rsq10            = gmx_mm256_calc_rsq_pd(dx10,dy10,dz10);
1537             rsq11            = gmx_mm256_calc_rsq_pd(dx11,dy11,dz11);
1538             rsq12            = gmx_mm256_calc_rsq_pd(dx12,dy12,dz12);
1539             rsq20            = gmx_mm256_calc_rsq_pd(dx20,dy20,dz20);
1540             rsq21            = gmx_mm256_calc_rsq_pd(dx21,dy21,dz21);
1541             rsq22            = gmx_mm256_calc_rsq_pd(dx22,dy22,dz22);
1542
1543             rinv00           = avx256_invsqrt_d(rsq00);
1544             rinv01           = avx256_invsqrt_d(rsq01);
1545             rinv02           = avx256_invsqrt_d(rsq02);
1546             rinv10           = avx256_invsqrt_d(rsq10);
1547             rinv11           = avx256_invsqrt_d(rsq11);
1548             rinv12           = avx256_invsqrt_d(rsq12);
1549             rinv20           = avx256_invsqrt_d(rsq20);
1550             rinv21           = avx256_invsqrt_d(rsq21);
1551             rinv22           = avx256_invsqrt_d(rsq22);
1552
1553             rinvsq00         = _mm256_mul_pd(rinv00,rinv00);
1554             rinvsq01         = _mm256_mul_pd(rinv01,rinv01);
1555             rinvsq02         = _mm256_mul_pd(rinv02,rinv02);
1556             rinvsq10         = _mm256_mul_pd(rinv10,rinv10);
1557             rinvsq11         = _mm256_mul_pd(rinv11,rinv11);
1558             rinvsq12         = _mm256_mul_pd(rinv12,rinv12);
1559             rinvsq20         = _mm256_mul_pd(rinv20,rinv20);
1560             rinvsq21         = _mm256_mul_pd(rinv21,rinv21);
1561             rinvsq22         = _mm256_mul_pd(rinv22,rinv22);
1562
1563             fjx0             = _mm256_setzero_pd();
1564             fjy0             = _mm256_setzero_pd();
1565             fjz0             = _mm256_setzero_pd();
1566             fjx1             = _mm256_setzero_pd();
1567             fjy1             = _mm256_setzero_pd();
1568             fjz1             = _mm256_setzero_pd();
1569             fjx2             = _mm256_setzero_pd();
1570             fjy2             = _mm256_setzero_pd();
1571             fjz2             = _mm256_setzero_pd();
1572
1573             /**************************
1574              * CALCULATE INTERACTIONS *
1575              **************************/
1576
1577             r00              = _mm256_mul_pd(rsq00,rinv00);
1578
1579             /* EWALD ELECTROSTATICS */
1580
1581             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1582             ewrt             = _mm256_mul_pd(r00,ewtabscale);
1583             ewitab           = _mm256_cvttpd_epi32(ewrt);
1584             eweps            = _mm256_sub_pd(ewrt,_mm256_round_pd(ewrt, _MM_FROUND_FLOOR));
1585             gmx_mm256_load_4pair_swizzle_pd(ewtab + _mm_extract_epi32(ewitab,0),ewtab + _mm_extract_epi32(ewitab,1),
1586                                             ewtab + _mm_extract_epi32(ewitab,2),ewtab + _mm_extract_epi32(ewitab,3),
1587                                             &ewtabF,&ewtabFn);
1588             felec            = _mm256_add_pd(_mm256_mul_pd( _mm256_sub_pd(one,eweps),ewtabF),_mm256_mul_pd(eweps,ewtabFn));
1589             felec            = _mm256_mul_pd(_mm256_mul_pd(qq00,rinv00),_mm256_sub_pd(rinvsq00,felec));
1590
1591             /* Analytical LJ-PME */
1592             rinvsix          = _mm256_mul_pd(_mm256_mul_pd(rinvsq00,rinvsq00),rinvsq00);
1593             ewcljrsq         = _mm256_mul_pd(ewclj2,rsq00);
1594             ewclj6           = _mm256_mul_pd(ewclj2,_mm256_mul_pd(ewclj2,ewclj2));
1595             exponent         = avx256_exp_d(ewcljrsq);
1596             /* poly = exp(-(beta*r)^2) * (1 + (beta*r)^2 + (beta*r)^4 /2) */
1597             poly             = _mm256_mul_pd(exponent,_mm256_add_pd(_mm256_sub_pd(one,ewcljrsq),_mm256_mul_pd(_mm256_mul_pd(ewcljrsq,ewcljrsq),one_half)));
1598             /* f6A = 6 * C6grid * (1 - poly) */
1599             f6A              = _mm256_mul_pd(c6grid_00,_mm256_sub_pd(one,poly));
1600             /* f6B = C6grid * exponent * beta^6 */
1601             f6B              = _mm256_mul_pd(_mm256_mul_pd(c6grid_00,one_sixth),_mm256_mul_pd(exponent,ewclj6));
1602             /* fvdw = 12*C12/r13 - ((6*C6 - f6A)/r6 + f6B)/r */
1603             fvdw              = _mm256_mul_pd(_mm256_add_pd(_mm256_mul_pd(_mm256_sub_pd(_mm256_mul_pd(c12_00,rinvsix),_mm256_sub_pd(c6_00,f6A)),rinvsix),f6B),rinvsq00);
1604
1605             fscal            = _mm256_add_pd(felec,fvdw);
1606
1607             /* Calculate temporary vectorial force */
1608             tx               = _mm256_mul_pd(fscal,dx00);
1609             ty               = _mm256_mul_pd(fscal,dy00);
1610             tz               = _mm256_mul_pd(fscal,dz00);
1611
1612             /* Update vectorial force */
1613             fix0             = _mm256_add_pd(fix0,tx);
1614             fiy0             = _mm256_add_pd(fiy0,ty);
1615             fiz0             = _mm256_add_pd(fiz0,tz);
1616
1617             fjx0             = _mm256_add_pd(fjx0,tx);
1618             fjy0             = _mm256_add_pd(fjy0,ty);
1619             fjz0             = _mm256_add_pd(fjz0,tz);
1620
1621             /**************************
1622              * CALCULATE INTERACTIONS *
1623              **************************/
1624
1625             r01              = _mm256_mul_pd(rsq01,rinv01);
1626
1627             /* EWALD ELECTROSTATICS */
1628
1629             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1630             ewrt             = _mm256_mul_pd(r01,ewtabscale);
1631             ewitab           = _mm256_cvttpd_epi32(ewrt);
1632             eweps            = _mm256_sub_pd(ewrt,_mm256_round_pd(ewrt, _MM_FROUND_FLOOR));
1633             gmx_mm256_load_4pair_swizzle_pd(ewtab + _mm_extract_epi32(ewitab,0),ewtab + _mm_extract_epi32(ewitab,1),
1634                                             ewtab + _mm_extract_epi32(ewitab,2),ewtab + _mm_extract_epi32(ewitab,3),
1635                                             &ewtabF,&ewtabFn);
1636             felec            = _mm256_add_pd(_mm256_mul_pd( _mm256_sub_pd(one,eweps),ewtabF),_mm256_mul_pd(eweps,ewtabFn));
1637             felec            = _mm256_mul_pd(_mm256_mul_pd(qq01,rinv01),_mm256_sub_pd(rinvsq01,felec));
1638
1639             fscal            = felec;
1640
1641             /* Calculate temporary vectorial force */
1642             tx               = _mm256_mul_pd(fscal,dx01);
1643             ty               = _mm256_mul_pd(fscal,dy01);
1644             tz               = _mm256_mul_pd(fscal,dz01);
1645
1646             /* Update vectorial force */
1647             fix0             = _mm256_add_pd(fix0,tx);
1648             fiy0             = _mm256_add_pd(fiy0,ty);
1649             fiz0             = _mm256_add_pd(fiz0,tz);
1650
1651             fjx1             = _mm256_add_pd(fjx1,tx);
1652             fjy1             = _mm256_add_pd(fjy1,ty);
1653             fjz1             = _mm256_add_pd(fjz1,tz);
1654
1655             /**************************
1656              * CALCULATE INTERACTIONS *
1657              **************************/
1658
1659             r02              = _mm256_mul_pd(rsq02,rinv02);
1660
1661             /* EWALD ELECTROSTATICS */
1662
1663             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1664             ewrt             = _mm256_mul_pd(r02,ewtabscale);
1665             ewitab           = _mm256_cvttpd_epi32(ewrt);
1666             eweps            = _mm256_sub_pd(ewrt,_mm256_round_pd(ewrt, _MM_FROUND_FLOOR));
1667             gmx_mm256_load_4pair_swizzle_pd(ewtab + _mm_extract_epi32(ewitab,0),ewtab + _mm_extract_epi32(ewitab,1),
1668                                             ewtab + _mm_extract_epi32(ewitab,2),ewtab + _mm_extract_epi32(ewitab,3),
1669                                             &ewtabF,&ewtabFn);
1670             felec            = _mm256_add_pd(_mm256_mul_pd( _mm256_sub_pd(one,eweps),ewtabF),_mm256_mul_pd(eweps,ewtabFn));
1671             felec            = _mm256_mul_pd(_mm256_mul_pd(qq02,rinv02),_mm256_sub_pd(rinvsq02,felec));
1672
1673             fscal            = felec;
1674
1675             /* Calculate temporary vectorial force */
1676             tx               = _mm256_mul_pd(fscal,dx02);
1677             ty               = _mm256_mul_pd(fscal,dy02);
1678             tz               = _mm256_mul_pd(fscal,dz02);
1679
1680             /* Update vectorial force */
1681             fix0             = _mm256_add_pd(fix0,tx);
1682             fiy0             = _mm256_add_pd(fiy0,ty);
1683             fiz0             = _mm256_add_pd(fiz0,tz);
1684
1685             fjx2             = _mm256_add_pd(fjx2,tx);
1686             fjy2             = _mm256_add_pd(fjy2,ty);
1687             fjz2             = _mm256_add_pd(fjz2,tz);
1688
1689             /**************************
1690              * CALCULATE INTERACTIONS *
1691              **************************/
1692
1693             r10              = _mm256_mul_pd(rsq10,rinv10);
1694
1695             /* EWALD ELECTROSTATICS */
1696
1697             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1698             ewrt             = _mm256_mul_pd(r10,ewtabscale);
1699             ewitab           = _mm256_cvttpd_epi32(ewrt);
1700             eweps            = _mm256_sub_pd(ewrt,_mm256_round_pd(ewrt, _MM_FROUND_FLOOR));
1701             gmx_mm256_load_4pair_swizzle_pd(ewtab + _mm_extract_epi32(ewitab,0),ewtab + _mm_extract_epi32(ewitab,1),
1702                                             ewtab + _mm_extract_epi32(ewitab,2),ewtab + _mm_extract_epi32(ewitab,3),
1703                                             &ewtabF,&ewtabFn);
1704             felec            = _mm256_add_pd(_mm256_mul_pd( _mm256_sub_pd(one,eweps),ewtabF),_mm256_mul_pd(eweps,ewtabFn));
1705             felec            = _mm256_mul_pd(_mm256_mul_pd(qq10,rinv10),_mm256_sub_pd(rinvsq10,felec));
1706
1707             fscal            = felec;
1708
1709             /* Calculate temporary vectorial force */
1710             tx               = _mm256_mul_pd(fscal,dx10);
1711             ty               = _mm256_mul_pd(fscal,dy10);
1712             tz               = _mm256_mul_pd(fscal,dz10);
1713
1714             /* Update vectorial force */
1715             fix1             = _mm256_add_pd(fix1,tx);
1716             fiy1             = _mm256_add_pd(fiy1,ty);
1717             fiz1             = _mm256_add_pd(fiz1,tz);
1718
1719             fjx0             = _mm256_add_pd(fjx0,tx);
1720             fjy0             = _mm256_add_pd(fjy0,ty);
1721             fjz0             = _mm256_add_pd(fjz0,tz);
1722
1723             /**************************
1724              * CALCULATE INTERACTIONS *
1725              **************************/
1726
1727             r11              = _mm256_mul_pd(rsq11,rinv11);
1728
1729             /* EWALD ELECTROSTATICS */
1730
1731             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1732             ewrt             = _mm256_mul_pd(r11,ewtabscale);
1733             ewitab           = _mm256_cvttpd_epi32(ewrt);
1734             eweps            = _mm256_sub_pd(ewrt,_mm256_round_pd(ewrt, _MM_FROUND_FLOOR));
1735             gmx_mm256_load_4pair_swizzle_pd(ewtab + _mm_extract_epi32(ewitab,0),ewtab + _mm_extract_epi32(ewitab,1),
1736                                             ewtab + _mm_extract_epi32(ewitab,2),ewtab + _mm_extract_epi32(ewitab,3),
1737                                             &ewtabF,&ewtabFn);
1738             felec            = _mm256_add_pd(_mm256_mul_pd( _mm256_sub_pd(one,eweps),ewtabF),_mm256_mul_pd(eweps,ewtabFn));
1739             felec            = _mm256_mul_pd(_mm256_mul_pd(qq11,rinv11),_mm256_sub_pd(rinvsq11,felec));
1740
1741             fscal            = felec;
1742
1743             /* Calculate temporary vectorial force */
1744             tx               = _mm256_mul_pd(fscal,dx11);
1745             ty               = _mm256_mul_pd(fscal,dy11);
1746             tz               = _mm256_mul_pd(fscal,dz11);
1747
1748             /* Update vectorial force */
1749             fix1             = _mm256_add_pd(fix1,tx);
1750             fiy1             = _mm256_add_pd(fiy1,ty);
1751             fiz1             = _mm256_add_pd(fiz1,tz);
1752
1753             fjx1             = _mm256_add_pd(fjx1,tx);
1754             fjy1             = _mm256_add_pd(fjy1,ty);
1755             fjz1             = _mm256_add_pd(fjz1,tz);
1756
1757             /**************************
1758              * CALCULATE INTERACTIONS *
1759              **************************/
1760
1761             r12              = _mm256_mul_pd(rsq12,rinv12);
1762
1763             /* EWALD ELECTROSTATICS */
1764
1765             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1766             ewrt             = _mm256_mul_pd(r12,ewtabscale);
1767             ewitab           = _mm256_cvttpd_epi32(ewrt);
1768             eweps            = _mm256_sub_pd(ewrt,_mm256_round_pd(ewrt, _MM_FROUND_FLOOR));
1769             gmx_mm256_load_4pair_swizzle_pd(ewtab + _mm_extract_epi32(ewitab,0),ewtab + _mm_extract_epi32(ewitab,1),
1770                                             ewtab + _mm_extract_epi32(ewitab,2),ewtab + _mm_extract_epi32(ewitab,3),
1771                                             &ewtabF,&ewtabFn);
1772             felec            = _mm256_add_pd(_mm256_mul_pd( _mm256_sub_pd(one,eweps),ewtabF),_mm256_mul_pd(eweps,ewtabFn));
1773             felec            = _mm256_mul_pd(_mm256_mul_pd(qq12,rinv12),_mm256_sub_pd(rinvsq12,felec));
1774
1775             fscal            = felec;
1776
1777             /* Calculate temporary vectorial force */
1778             tx               = _mm256_mul_pd(fscal,dx12);
1779             ty               = _mm256_mul_pd(fscal,dy12);
1780             tz               = _mm256_mul_pd(fscal,dz12);
1781
1782             /* Update vectorial force */
1783             fix1             = _mm256_add_pd(fix1,tx);
1784             fiy1             = _mm256_add_pd(fiy1,ty);
1785             fiz1             = _mm256_add_pd(fiz1,tz);
1786
1787             fjx2             = _mm256_add_pd(fjx2,tx);
1788             fjy2             = _mm256_add_pd(fjy2,ty);
1789             fjz2             = _mm256_add_pd(fjz2,tz);
1790
1791             /**************************
1792              * CALCULATE INTERACTIONS *
1793              **************************/
1794
1795             r20              = _mm256_mul_pd(rsq20,rinv20);
1796
1797             /* EWALD ELECTROSTATICS */
1798
1799             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1800             ewrt             = _mm256_mul_pd(r20,ewtabscale);
1801             ewitab           = _mm256_cvttpd_epi32(ewrt);
1802             eweps            = _mm256_sub_pd(ewrt,_mm256_round_pd(ewrt, _MM_FROUND_FLOOR));
1803             gmx_mm256_load_4pair_swizzle_pd(ewtab + _mm_extract_epi32(ewitab,0),ewtab + _mm_extract_epi32(ewitab,1),
1804                                             ewtab + _mm_extract_epi32(ewitab,2),ewtab + _mm_extract_epi32(ewitab,3),
1805                                             &ewtabF,&ewtabFn);
1806             felec            = _mm256_add_pd(_mm256_mul_pd( _mm256_sub_pd(one,eweps),ewtabF),_mm256_mul_pd(eweps,ewtabFn));
1807             felec            = _mm256_mul_pd(_mm256_mul_pd(qq20,rinv20),_mm256_sub_pd(rinvsq20,felec));
1808
1809             fscal            = felec;
1810
1811             /* Calculate temporary vectorial force */
1812             tx               = _mm256_mul_pd(fscal,dx20);
1813             ty               = _mm256_mul_pd(fscal,dy20);
1814             tz               = _mm256_mul_pd(fscal,dz20);
1815
1816             /* Update vectorial force */
1817             fix2             = _mm256_add_pd(fix2,tx);
1818             fiy2             = _mm256_add_pd(fiy2,ty);
1819             fiz2             = _mm256_add_pd(fiz2,tz);
1820
1821             fjx0             = _mm256_add_pd(fjx0,tx);
1822             fjy0             = _mm256_add_pd(fjy0,ty);
1823             fjz0             = _mm256_add_pd(fjz0,tz);
1824
1825             /**************************
1826              * CALCULATE INTERACTIONS *
1827              **************************/
1828
1829             r21              = _mm256_mul_pd(rsq21,rinv21);
1830
1831             /* EWALD ELECTROSTATICS */
1832
1833             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1834             ewrt             = _mm256_mul_pd(r21,ewtabscale);
1835             ewitab           = _mm256_cvttpd_epi32(ewrt);
1836             eweps            = _mm256_sub_pd(ewrt,_mm256_round_pd(ewrt, _MM_FROUND_FLOOR));
1837             gmx_mm256_load_4pair_swizzle_pd(ewtab + _mm_extract_epi32(ewitab,0),ewtab + _mm_extract_epi32(ewitab,1),
1838                                             ewtab + _mm_extract_epi32(ewitab,2),ewtab + _mm_extract_epi32(ewitab,3),
1839                                             &ewtabF,&ewtabFn);
1840             felec            = _mm256_add_pd(_mm256_mul_pd( _mm256_sub_pd(one,eweps),ewtabF),_mm256_mul_pd(eweps,ewtabFn));
1841             felec            = _mm256_mul_pd(_mm256_mul_pd(qq21,rinv21),_mm256_sub_pd(rinvsq21,felec));
1842
1843             fscal            = felec;
1844
1845             /* Calculate temporary vectorial force */
1846             tx               = _mm256_mul_pd(fscal,dx21);
1847             ty               = _mm256_mul_pd(fscal,dy21);
1848             tz               = _mm256_mul_pd(fscal,dz21);
1849
1850             /* Update vectorial force */
1851             fix2             = _mm256_add_pd(fix2,tx);
1852             fiy2             = _mm256_add_pd(fiy2,ty);
1853             fiz2             = _mm256_add_pd(fiz2,tz);
1854
1855             fjx1             = _mm256_add_pd(fjx1,tx);
1856             fjy1             = _mm256_add_pd(fjy1,ty);
1857             fjz1             = _mm256_add_pd(fjz1,tz);
1858
1859             /**************************
1860              * CALCULATE INTERACTIONS *
1861              **************************/
1862
1863             r22              = _mm256_mul_pd(rsq22,rinv22);
1864
1865             /* EWALD ELECTROSTATICS */
1866
1867             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1868             ewrt             = _mm256_mul_pd(r22,ewtabscale);
1869             ewitab           = _mm256_cvttpd_epi32(ewrt);
1870             eweps            = _mm256_sub_pd(ewrt,_mm256_round_pd(ewrt, _MM_FROUND_FLOOR));
1871             gmx_mm256_load_4pair_swizzle_pd(ewtab + _mm_extract_epi32(ewitab,0),ewtab + _mm_extract_epi32(ewitab,1),
1872                                             ewtab + _mm_extract_epi32(ewitab,2),ewtab + _mm_extract_epi32(ewitab,3),
1873                                             &ewtabF,&ewtabFn);
1874             felec            = _mm256_add_pd(_mm256_mul_pd( _mm256_sub_pd(one,eweps),ewtabF),_mm256_mul_pd(eweps,ewtabFn));
1875             felec            = _mm256_mul_pd(_mm256_mul_pd(qq22,rinv22),_mm256_sub_pd(rinvsq22,felec));
1876
1877             fscal            = felec;
1878
1879             /* Calculate temporary vectorial force */
1880             tx               = _mm256_mul_pd(fscal,dx22);
1881             ty               = _mm256_mul_pd(fscal,dy22);
1882             tz               = _mm256_mul_pd(fscal,dz22);
1883
1884             /* Update vectorial force */
1885             fix2             = _mm256_add_pd(fix2,tx);
1886             fiy2             = _mm256_add_pd(fiy2,ty);
1887             fiz2             = _mm256_add_pd(fiz2,tz);
1888
1889             fjx2             = _mm256_add_pd(fjx2,tx);
1890             fjy2             = _mm256_add_pd(fjy2,ty);
1891             fjz2             = _mm256_add_pd(fjz2,tz);
1892
1893             fjptrA             = f+j_coord_offsetA;
1894             fjptrB             = f+j_coord_offsetB;
1895             fjptrC             = f+j_coord_offsetC;
1896             fjptrD             = f+j_coord_offsetD;
1897
1898             gmx_mm256_decrement_3rvec_4ptr_swizzle_pd(fjptrA,fjptrB,fjptrC,fjptrD,
1899                                                       fjx0,fjy0,fjz0,fjx1,fjy1,fjz1,fjx2,fjy2,fjz2);
1900
1901             /* Inner loop uses 347 flops */
1902         }
1903
1904         if(jidx<j_index_end)
1905         {
1906
1907             /* Get j neighbor index, and coordinate index */
1908             jnrlistA         = jjnr[jidx];
1909             jnrlistB         = jjnr[jidx+1];
1910             jnrlistC         = jjnr[jidx+2];
1911             jnrlistD         = jjnr[jidx+3];
1912             /* Sign of each element will be negative for non-real atoms.
1913              * This mask will be 0xFFFFFFFF for dummy entries and 0x0 for real ones,
1914              * so use it as val = _mm_andnot_pd(mask,val) to clear dummy entries.
1915              */
1916             tmpmask0 = gmx_mm_castsi128_ps(_mm_cmplt_epi32(_mm_loadu_si128((const __m128i *)(jjnr+jidx)),_mm_setzero_si128()));
1917
1918             tmpmask1 = _mm_permute_ps(tmpmask0,_GMX_MM_PERMUTE(3,3,2,2));
1919             tmpmask0 = _mm_permute_ps(tmpmask0,_GMX_MM_PERMUTE(1,1,0,0));
1920             dummy_mask = _mm256_castps_pd(gmx_mm256_set_m128(tmpmask1,tmpmask0));
1921
1922             jnrA       = (jnrlistA>=0) ? jnrlistA : 0;
1923             jnrB       = (jnrlistB>=0) ? jnrlistB : 0;
1924             jnrC       = (jnrlistC>=0) ? jnrlistC : 0;
1925             jnrD       = (jnrlistD>=0) ? jnrlistD : 0;
1926             j_coord_offsetA  = DIM*jnrA;
1927             j_coord_offsetB  = DIM*jnrB;
1928             j_coord_offsetC  = DIM*jnrC;
1929             j_coord_offsetD  = DIM*jnrD;
1930
1931             /* load j atom coordinates */
1932             gmx_mm256_load_3rvec_4ptr_swizzle_pd(x+j_coord_offsetA,x+j_coord_offsetB,
1933                                                  x+j_coord_offsetC,x+j_coord_offsetD,
1934                                               &jx0,&jy0,&jz0,&jx1,&jy1,&jz1,&jx2,&jy2,&jz2);
1935
1936             /* Calculate displacement vector */
1937             dx00             = _mm256_sub_pd(ix0,jx0);
1938             dy00             = _mm256_sub_pd(iy0,jy0);
1939             dz00             = _mm256_sub_pd(iz0,jz0);
1940             dx01             = _mm256_sub_pd(ix0,jx1);
1941             dy01             = _mm256_sub_pd(iy0,jy1);
1942             dz01             = _mm256_sub_pd(iz0,jz1);
1943             dx02             = _mm256_sub_pd(ix0,jx2);
1944             dy02             = _mm256_sub_pd(iy0,jy2);
1945             dz02             = _mm256_sub_pd(iz0,jz2);
1946             dx10             = _mm256_sub_pd(ix1,jx0);
1947             dy10             = _mm256_sub_pd(iy1,jy0);
1948             dz10             = _mm256_sub_pd(iz1,jz0);
1949             dx11             = _mm256_sub_pd(ix1,jx1);
1950             dy11             = _mm256_sub_pd(iy1,jy1);
1951             dz11             = _mm256_sub_pd(iz1,jz1);
1952             dx12             = _mm256_sub_pd(ix1,jx2);
1953             dy12             = _mm256_sub_pd(iy1,jy2);
1954             dz12             = _mm256_sub_pd(iz1,jz2);
1955             dx20             = _mm256_sub_pd(ix2,jx0);
1956             dy20             = _mm256_sub_pd(iy2,jy0);
1957             dz20             = _mm256_sub_pd(iz2,jz0);
1958             dx21             = _mm256_sub_pd(ix2,jx1);
1959             dy21             = _mm256_sub_pd(iy2,jy1);
1960             dz21             = _mm256_sub_pd(iz2,jz1);
1961             dx22             = _mm256_sub_pd(ix2,jx2);
1962             dy22             = _mm256_sub_pd(iy2,jy2);
1963             dz22             = _mm256_sub_pd(iz2,jz2);
1964
1965             /* Calculate squared distance and things based on it */
1966             rsq00            = gmx_mm256_calc_rsq_pd(dx00,dy00,dz00);
1967             rsq01            = gmx_mm256_calc_rsq_pd(dx01,dy01,dz01);
1968             rsq02            = gmx_mm256_calc_rsq_pd(dx02,dy02,dz02);
1969             rsq10            = gmx_mm256_calc_rsq_pd(dx10,dy10,dz10);
1970             rsq11            = gmx_mm256_calc_rsq_pd(dx11,dy11,dz11);
1971             rsq12            = gmx_mm256_calc_rsq_pd(dx12,dy12,dz12);
1972             rsq20            = gmx_mm256_calc_rsq_pd(dx20,dy20,dz20);
1973             rsq21            = gmx_mm256_calc_rsq_pd(dx21,dy21,dz21);
1974             rsq22            = gmx_mm256_calc_rsq_pd(dx22,dy22,dz22);
1975
1976             rinv00           = avx256_invsqrt_d(rsq00);
1977             rinv01           = avx256_invsqrt_d(rsq01);
1978             rinv02           = avx256_invsqrt_d(rsq02);
1979             rinv10           = avx256_invsqrt_d(rsq10);
1980             rinv11           = avx256_invsqrt_d(rsq11);
1981             rinv12           = avx256_invsqrt_d(rsq12);
1982             rinv20           = avx256_invsqrt_d(rsq20);
1983             rinv21           = avx256_invsqrt_d(rsq21);
1984             rinv22           = avx256_invsqrt_d(rsq22);
1985
1986             rinvsq00         = _mm256_mul_pd(rinv00,rinv00);
1987             rinvsq01         = _mm256_mul_pd(rinv01,rinv01);
1988             rinvsq02         = _mm256_mul_pd(rinv02,rinv02);
1989             rinvsq10         = _mm256_mul_pd(rinv10,rinv10);
1990             rinvsq11         = _mm256_mul_pd(rinv11,rinv11);
1991             rinvsq12         = _mm256_mul_pd(rinv12,rinv12);
1992             rinvsq20         = _mm256_mul_pd(rinv20,rinv20);
1993             rinvsq21         = _mm256_mul_pd(rinv21,rinv21);
1994             rinvsq22         = _mm256_mul_pd(rinv22,rinv22);
1995
1996             fjx0             = _mm256_setzero_pd();
1997             fjy0             = _mm256_setzero_pd();
1998             fjz0             = _mm256_setzero_pd();
1999             fjx1             = _mm256_setzero_pd();
2000             fjy1             = _mm256_setzero_pd();
2001             fjz1             = _mm256_setzero_pd();
2002             fjx2             = _mm256_setzero_pd();
2003             fjy2             = _mm256_setzero_pd();
2004             fjz2             = _mm256_setzero_pd();
2005
2006             /**************************
2007              * CALCULATE INTERACTIONS *
2008              **************************/
2009
2010             r00              = _mm256_mul_pd(rsq00,rinv00);
2011             r00              = _mm256_andnot_pd(dummy_mask,r00);
2012
2013             /* EWALD ELECTROSTATICS */
2014
2015             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
2016             ewrt             = _mm256_mul_pd(r00,ewtabscale);
2017             ewitab           = _mm256_cvttpd_epi32(ewrt);
2018             eweps            = _mm256_sub_pd(ewrt,_mm256_round_pd(ewrt, _MM_FROUND_FLOOR));
2019             gmx_mm256_load_4pair_swizzle_pd(ewtab + _mm_extract_epi32(ewitab,0),ewtab + _mm_extract_epi32(ewitab,1),
2020                                             ewtab + _mm_extract_epi32(ewitab,2),ewtab + _mm_extract_epi32(ewitab,3),
2021                                             &ewtabF,&ewtabFn);
2022             felec            = _mm256_add_pd(_mm256_mul_pd( _mm256_sub_pd(one,eweps),ewtabF),_mm256_mul_pd(eweps,ewtabFn));
2023             felec            = _mm256_mul_pd(_mm256_mul_pd(qq00,rinv00),_mm256_sub_pd(rinvsq00,felec));
2024
2025             /* Analytical LJ-PME */
2026             rinvsix          = _mm256_mul_pd(_mm256_mul_pd(rinvsq00,rinvsq00),rinvsq00);
2027             ewcljrsq         = _mm256_mul_pd(ewclj2,rsq00);
2028             ewclj6           = _mm256_mul_pd(ewclj2,_mm256_mul_pd(ewclj2,ewclj2));
2029             exponent         = avx256_exp_d(ewcljrsq);
2030             /* poly = exp(-(beta*r)^2) * (1 + (beta*r)^2 + (beta*r)^4 /2) */
2031             poly             = _mm256_mul_pd(exponent,_mm256_add_pd(_mm256_sub_pd(one,ewcljrsq),_mm256_mul_pd(_mm256_mul_pd(ewcljrsq,ewcljrsq),one_half)));
2032             /* f6A = 6 * C6grid * (1 - poly) */
2033             f6A              = _mm256_mul_pd(c6grid_00,_mm256_sub_pd(one,poly));
2034             /* f6B = C6grid * exponent * beta^6 */
2035             f6B              = _mm256_mul_pd(_mm256_mul_pd(c6grid_00,one_sixth),_mm256_mul_pd(exponent,ewclj6));
2036             /* fvdw = 12*C12/r13 - ((6*C6 - f6A)/r6 + f6B)/r */
2037             fvdw              = _mm256_mul_pd(_mm256_add_pd(_mm256_mul_pd(_mm256_sub_pd(_mm256_mul_pd(c12_00,rinvsix),_mm256_sub_pd(c6_00,f6A)),rinvsix),f6B),rinvsq00);
2038
2039             fscal            = _mm256_add_pd(felec,fvdw);
2040
2041             fscal            = _mm256_andnot_pd(dummy_mask,fscal);
2042
2043             /* Calculate temporary vectorial force */
2044             tx               = _mm256_mul_pd(fscal,dx00);
2045             ty               = _mm256_mul_pd(fscal,dy00);
2046             tz               = _mm256_mul_pd(fscal,dz00);
2047
2048             /* Update vectorial force */
2049             fix0             = _mm256_add_pd(fix0,tx);
2050             fiy0             = _mm256_add_pd(fiy0,ty);
2051             fiz0             = _mm256_add_pd(fiz0,tz);
2052
2053             fjx0             = _mm256_add_pd(fjx0,tx);
2054             fjy0             = _mm256_add_pd(fjy0,ty);
2055             fjz0             = _mm256_add_pd(fjz0,tz);
2056
2057             /**************************
2058              * CALCULATE INTERACTIONS *
2059              **************************/
2060
2061             r01              = _mm256_mul_pd(rsq01,rinv01);
2062             r01              = _mm256_andnot_pd(dummy_mask,r01);
2063
2064             /* EWALD ELECTROSTATICS */
2065
2066             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
2067             ewrt             = _mm256_mul_pd(r01,ewtabscale);
2068             ewitab           = _mm256_cvttpd_epi32(ewrt);
2069             eweps            = _mm256_sub_pd(ewrt,_mm256_round_pd(ewrt, _MM_FROUND_FLOOR));
2070             gmx_mm256_load_4pair_swizzle_pd(ewtab + _mm_extract_epi32(ewitab,0),ewtab + _mm_extract_epi32(ewitab,1),
2071                                             ewtab + _mm_extract_epi32(ewitab,2),ewtab + _mm_extract_epi32(ewitab,3),
2072                                             &ewtabF,&ewtabFn);
2073             felec            = _mm256_add_pd(_mm256_mul_pd( _mm256_sub_pd(one,eweps),ewtabF),_mm256_mul_pd(eweps,ewtabFn));
2074             felec            = _mm256_mul_pd(_mm256_mul_pd(qq01,rinv01),_mm256_sub_pd(rinvsq01,felec));
2075
2076             fscal            = felec;
2077
2078             fscal            = _mm256_andnot_pd(dummy_mask,fscal);
2079
2080             /* Calculate temporary vectorial force */
2081             tx               = _mm256_mul_pd(fscal,dx01);
2082             ty               = _mm256_mul_pd(fscal,dy01);
2083             tz               = _mm256_mul_pd(fscal,dz01);
2084
2085             /* Update vectorial force */
2086             fix0             = _mm256_add_pd(fix0,tx);
2087             fiy0             = _mm256_add_pd(fiy0,ty);
2088             fiz0             = _mm256_add_pd(fiz0,tz);
2089
2090             fjx1             = _mm256_add_pd(fjx1,tx);
2091             fjy1             = _mm256_add_pd(fjy1,ty);
2092             fjz1             = _mm256_add_pd(fjz1,tz);
2093
2094             /**************************
2095              * CALCULATE INTERACTIONS *
2096              **************************/
2097
2098             r02              = _mm256_mul_pd(rsq02,rinv02);
2099             r02              = _mm256_andnot_pd(dummy_mask,r02);
2100
2101             /* EWALD ELECTROSTATICS */
2102
2103             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
2104             ewrt             = _mm256_mul_pd(r02,ewtabscale);
2105             ewitab           = _mm256_cvttpd_epi32(ewrt);
2106             eweps            = _mm256_sub_pd(ewrt,_mm256_round_pd(ewrt, _MM_FROUND_FLOOR));
2107             gmx_mm256_load_4pair_swizzle_pd(ewtab + _mm_extract_epi32(ewitab,0),ewtab + _mm_extract_epi32(ewitab,1),
2108                                             ewtab + _mm_extract_epi32(ewitab,2),ewtab + _mm_extract_epi32(ewitab,3),
2109                                             &ewtabF,&ewtabFn);
2110             felec            = _mm256_add_pd(_mm256_mul_pd( _mm256_sub_pd(one,eweps),ewtabF),_mm256_mul_pd(eweps,ewtabFn));
2111             felec            = _mm256_mul_pd(_mm256_mul_pd(qq02,rinv02),_mm256_sub_pd(rinvsq02,felec));
2112
2113             fscal            = felec;
2114
2115             fscal            = _mm256_andnot_pd(dummy_mask,fscal);
2116
2117             /* Calculate temporary vectorial force */
2118             tx               = _mm256_mul_pd(fscal,dx02);
2119             ty               = _mm256_mul_pd(fscal,dy02);
2120             tz               = _mm256_mul_pd(fscal,dz02);
2121
2122             /* Update vectorial force */
2123             fix0             = _mm256_add_pd(fix0,tx);
2124             fiy0             = _mm256_add_pd(fiy0,ty);
2125             fiz0             = _mm256_add_pd(fiz0,tz);
2126
2127             fjx2             = _mm256_add_pd(fjx2,tx);
2128             fjy2             = _mm256_add_pd(fjy2,ty);
2129             fjz2             = _mm256_add_pd(fjz2,tz);
2130
2131             /**************************
2132              * CALCULATE INTERACTIONS *
2133              **************************/
2134
2135             r10              = _mm256_mul_pd(rsq10,rinv10);
2136             r10              = _mm256_andnot_pd(dummy_mask,r10);
2137
2138             /* EWALD ELECTROSTATICS */
2139
2140             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
2141             ewrt             = _mm256_mul_pd(r10,ewtabscale);
2142             ewitab           = _mm256_cvttpd_epi32(ewrt);
2143             eweps            = _mm256_sub_pd(ewrt,_mm256_round_pd(ewrt, _MM_FROUND_FLOOR));
2144             gmx_mm256_load_4pair_swizzle_pd(ewtab + _mm_extract_epi32(ewitab,0),ewtab + _mm_extract_epi32(ewitab,1),
2145                                             ewtab + _mm_extract_epi32(ewitab,2),ewtab + _mm_extract_epi32(ewitab,3),
2146                                             &ewtabF,&ewtabFn);
2147             felec            = _mm256_add_pd(_mm256_mul_pd( _mm256_sub_pd(one,eweps),ewtabF),_mm256_mul_pd(eweps,ewtabFn));
2148             felec            = _mm256_mul_pd(_mm256_mul_pd(qq10,rinv10),_mm256_sub_pd(rinvsq10,felec));
2149
2150             fscal            = felec;
2151
2152             fscal            = _mm256_andnot_pd(dummy_mask,fscal);
2153
2154             /* Calculate temporary vectorial force */
2155             tx               = _mm256_mul_pd(fscal,dx10);
2156             ty               = _mm256_mul_pd(fscal,dy10);
2157             tz               = _mm256_mul_pd(fscal,dz10);
2158
2159             /* Update vectorial force */
2160             fix1             = _mm256_add_pd(fix1,tx);
2161             fiy1             = _mm256_add_pd(fiy1,ty);
2162             fiz1             = _mm256_add_pd(fiz1,tz);
2163
2164             fjx0             = _mm256_add_pd(fjx0,tx);
2165             fjy0             = _mm256_add_pd(fjy0,ty);
2166             fjz0             = _mm256_add_pd(fjz0,tz);
2167
2168             /**************************
2169              * CALCULATE INTERACTIONS *
2170              **************************/
2171
2172             r11              = _mm256_mul_pd(rsq11,rinv11);
2173             r11              = _mm256_andnot_pd(dummy_mask,r11);
2174
2175             /* EWALD ELECTROSTATICS */
2176
2177             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
2178             ewrt             = _mm256_mul_pd(r11,ewtabscale);
2179             ewitab           = _mm256_cvttpd_epi32(ewrt);
2180             eweps            = _mm256_sub_pd(ewrt,_mm256_round_pd(ewrt, _MM_FROUND_FLOOR));
2181             gmx_mm256_load_4pair_swizzle_pd(ewtab + _mm_extract_epi32(ewitab,0),ewtab + _mm_extract_epi32(ewitab,1),
2182                                             ewtab + _mm_extract_epi32(ewitab,2),ewtab + _mm_extract_epi32(ewitab,3),
2183                                             &ewtabF,&ewtabFn);
2184             felec            = _mm256_add_pd(_mm256_mul_pd( _mm256_sub_pd(one,eweps),ewtabF),_mm256_mul_pd(eweps,ewtabFn));
2185             felec            = _mm256_mul_pd(_mm256_mul_pd(qq11,rinv11),_mm256_sub_pd(rinvsq11,felec));
2186
2187             fscal            = felec;
2188
2189             fscal            = _mm256_andnot_pd(dummy_mask,fscal);
2190
2191             /* Calculate temporary vectorial force */
2192             tx               = _mm256_mul_pd(fscal,dx11);
2193             ty               = _mm256_mul_pd(fscal,dy11);
2194             tz               = _mm256_mul_pd(fscal,dz11);
2195
2196             /* Update vectorial force */
2197             fix1             = _mm256_add_pd(fix1,tx);
2198             fiy1             = _mm256_add_pd(fiy1,ty);
2199             fiz1             = _mm256_add_pd(fiz1,tz);
2200
2201             fjx1             = _mm256_add_pd(fjx1,tx);
2202             fjy1             = _mm256_add_pd(fjy1,ty);
2203             fjz1             = _mm256_add_pd(fjz1,tz);
2204
2205             /**************************
2206              * CALCULATE INTERACTIONS *
2207              **************************/
2208
2209             r12              = _mm256_mul_pd(rsq12,rinv12);
2210             r12              = _mm256_andnot_pd(dummy_mask,r12);
2211
2212             /* EWALD ELECTROSTATICS */
2213
2214             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
2215             ewrt             = _mm256_mul_pd(r12,ewtabscale);
2216             ewitab           = _mm256_cvttpd_epi32(ewrt);
2217             eweps            = _mm256_sub_pd(ewrt,_mm256_round_pd(ewrt, _MM_FROUND_FLOOR));
2218             gmx_mm256_load_4pair_swizzle_pd(ewtab + _mm_extract_epi32(ewitab,0),ewtab + _mm_extract_epi32(ewitab,1),
2219                                             ewtab + _mm_extract_epi32(ewitab,2),ewtab + _mm_extract_epi32(ewitab,3),
2220                                             &ewtabF,&ewtabFn);
2221             felec            = _mm256_add_pd(_mm256_mul_pd( _mm256_sub_pd(one,eweps),ewtabF),_mm256_mul_pd(eweps,ewtabFn));
2222             felec            = _mm256_mul_pd(_mm256_mul_pd(qq12,rinv12),_mm256_sub_pd(rinvsq12,felec));
2223
2224             fscal            = felec;
2225
2226             fscal            = _mm256_andnot_pd(dummy_mask,fscal);
2227
2228             /* Calculate temporary vectorial force */
2229             tx               = _mm256_mul_pd(fscal,dx12);
2230             ty               = _mm256_mul_pd(fscal,dy12);
2231             tz               = _mm256_mul_pd(fscal,dz12);
2232
2233             /* Update vectorial force */
2234             fix1             = _mm256_add_pd(fix1,tx);
2235             fiy1             = _mm256_add_pd(fiy1,ty);
2236             fiz1             = _mm256_add_pd(fiz1,tz);
2237
2238             fjx2             = _mm256_add_pd(fjx2,tx);
2239             fjy2             = _mm256_add_pd(fjy2,ty);
2240             fjz2             = _mm256_add_pd(fjz2,tz);
2241
2242             /**************************
2243              * CALCULATE INTERACTIONS *
2244              **************************/
2245
2246             r20              = _mm256_mul_pd(rsq20,rinv20);
2247             r20              = _mm256_andnot_pd(dummy_mask,r20);
2248
2249             /* EWALD ELECTROSTATICS */
2250
2251             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
2252             ewrt             = _mm256_mul_pd(r20,ewtabscale);
2253             ewitab           = _mm256_cvttpd_epi32(ewrt);
2254             eweps            = _mm256_sub_pd(ewrt,_mm256_round_pd(ewrt, _MM_FROUND_FLOOR));
2255             gmx_mm256_load_4pair_swizzle_pd(ewtab + _mm_extract_epi32(ewitab,0),ewtab + _mm_extract_epi32(ewitab,1),
2256                                             ewtab + _mm_extract_epi32(ewitab,2),ewtab + _mm_extract_epi32(ewitab,3),
2257                                             &ewtabF,&ewtabFn);
2258             felec            = _mm256_add_pd(_mm256_mul_pd( _mm256_sub_pd(one,eweps),ewtabF),_mm256_mul_pd(eweps,ewtabFn));
2259             felec            = _mm256_mul_pd(_mm256_mul_pd(qq20,rinv20),_mm256_sub_pd(rinvsq20,felec));
2260
2261             fscal            = felec;
2262
2263             fscal            = _mm256_andnot_pd(dummy_mask,fscal);
2264
2265             /* Calculate temporary vectorial force */
2266             tx               = _mm256_mul_pd(fscal,dx20);
2267             ty               = _mm256_mul_pd(fscal,dy20);
2268             tz               = _mm256_mul_pd(fscal,dz20);
2269
2270             /* Update vectorial force */
2271             fix2             = _mm256_add_pd(fix2,tx);
2272             fiy2             = _mm256_add_pd(fiy2,ty);
2273             fiz2             = _mm256_add_pd(fiz2,tz);
2274
2275             fjx0             = _mm256_add_pd(fjx0,tx);
2276             fjy0             = _mm256_add_pd(fjy0,ty);
2277             fjz0             = _mm256_add_pd(fjz0,tz);
2278
2279             /**************************
2280              * CALCULATE INTERACTIONS *
2281              **************************/
2282
2283             r21              = _mm256_mul_pd(rsq21,rinv21);
2284             r21              = _mm256_andnot_pd(dummy_mask,r21);
2285
2286             /* EWALD ELECTROSTATICS */
2287
2288             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
2289             ewrt             = _mm256_mul_pd(r21,ewtabscale);
2290             ewitab           = _mm256_cvttpd_epi32(ewrt);
2291             eweps            = _mm256_sub_pd(ewrt,_mm256_round_pd(ewrt, _MM_FROUND_FLOOR));
2292             gmx_mm256_load_4pair_swizzle_pd(ewtab + _mm_extract_epi32(ewitab,0),ewtab + _mm_extract_epi32(ewitab,1),
2293                                             ewtab + _mm_extract_epi32(ewitab,2),ewtab + _mm_extract_epi32(ewitab,3),
2294                                             &ewtabF,&ewtabFn);
2295             felec            = _mm256_add_pd(_mm256_mul_pd( _mm256_sub_pd(one,eweps),ewtabF),_mm256_mul_pd(eweps,ewtabFn));
2296             felec            = _mm256_mul_pd(_mm256_mul_pd(qq21,rinv21),_mm256_sub_pd(rinvsq21,felec));
2297
2298             fscal            = felec;
2299
2300             fscal            = _mm256_andnot_pd(dummy_mask,fscal);
2301
2302             /* Calculate temporary vectorial force */
2303             tx               = _mm256_mul_pd(fscal,dx21);
2304             ty               = _mm256_mul_pd(fscal,dy21);
2305             tz               = _mm256_mul_pd(fscal,dz21);
2306
2307             /* Update vectorial force */
2308             fix2             = _mm256_add_pd(fix2,tx);
2309             fiy2             = _mm256_add_pd(fiy2,ty);
2310             fiz2             = _mm256_add_pd(fiz2,tz);
2311
2312             fjx1             = _mm256_add_pd(fjx1,tx);
2313             fjy1             = _mm256_add_pd(fjy1,ty);
2314             fjz1             = _mm256_add_pd(fjz1,tz);
2315
2316             /**************************
2317              * CALCULATE INTERACTIONS *
2318              **************************/
2319
2320             r22              = _mm256_mul_pd(rsq22,rinv22);
2321             r22              = _mm256_andnot_pd(dummy_mask,r22);
2322
2323             /* EWALD ELECTROSTATICS */
2324
2325             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
2326             ewrt             = _mm256_mul_pd(r22,ewtabscale);
2327             ewitab           = _mm256_cvttpd_epi32(ewrt);
2328             eweps            = _mm256_sub_pd(ewrt,_mm256_round_pd(ewrt, _MM_FROUND_FLOOR));
2329             gmx_mm256_load_4pair_swizzle_pd(ewtab + _mm_extract_epi32(ewitab,0),ewtab + _mm_extract_epi32(ewitab,1),
2330                                             ewtab + _mm_extract_epi32(ewitab,2),ewtab + _mm_extract_epi32(ewitab,3),
2331                                             &ewtabF,&ewtabFn);
2332             felec            = _mm256_add_pd(_mm256_mul_pd( _mm256_sub_pd(one,eweps),ewtabF),_mm256_mul_pd(eweps,ewtabFn));
2333             felec            = _mm256_mul_pd(_mm256_mul_pd(qq22,rinv22),_mm256_sub_pd(rinvsq22,felec));
2334
2335             fscal            = felec;
2336
2337             fscal            = _mm256_andnot_pd(dummy_mask,fscal);
2338
2339             /* Calculate temporary vectorial force */
2340             tx               = _mm256_mul_pd(fscal,dx22);
2341             ty               = _mm256_mul_pd(fscal,dy22);
2342             tz               = _mm256_mul_pd(fscal,dz22);
2343
2344             /* Update vectorial force */
2345             fix2             = _mm256_add_pd(fix2,tx);
2346             fiy2             = _mm256_add_pd(fiy2,ty);
2347             fiz2             = _mm256_add_pd(fiz2,tz);
2348
2349             fjx2             = _mm256_add_pd(fjx2,tx);
2350             fjy2             = _mm256_add_pd(fjy2,ty);
2351             fjz2             = _mm256_add_pd(fjz2,tz);
2352
2353             fjptrA             = (jnrlistA>=0) ? f+j_coord_offsetA : scratch;
2354             fjptrB             = (jnrlistB>=0) ? f+j_coord_offsetB : scratch;
2355             fjptrC             = (jnrlistC>=0) ? f+j_coord_offsetC : scratch;
2356             fjptrD             = (jnrlistD>=0) ? f+j_coord_offsetD : scratch;
2357
2358             gmx_mm256_decrement_3rvec_4ptr_swizzle_pd(fjptrA,fjptrB,fjptrC,fjptrD,
2359                                                       fjx0,fjy0,fjz0,fjx1,fjy1,fjz1,fjx2,fjy2,fjz2);
2360
2361             /* Inner loop uses 356 flops */
2362         }
2363
2364         /* End of innermost loop */
2365
2366         gmx_mm256_update_iforce_3atom_swizzle_pd(fix0,fiy0,fiz0,fix1,fiy1,fiz1,fix2,fiy2,fiz2,
2367                                                  f+i_coord_offset,fshift+i_shift_offset);
2368
2369         /* Increment number of inner iterations */
2370         inneriter                  += j_index_end - j_index_start;
2371
2372         /* Outer loop uses 18 flops */
2373     }
2374
2375     /* Increment number of outer iterations */
2376     outeriter        += nri;
2377
2378     /* Update outer/inner flops */
2379
2380     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W3W3_F,outeriter*18 + inneriter*356);
2381 }