src/gromacs/gmxlib/nonbonded/nb_kernel_avx_256_single/nb_kernel_ElecEw_VdwLJEw_GeomW3W3_avx_256_single.c

   1 /*
   2  * This file is part of the GROMACS molecular simulation package.
   3  *
   4  * Copyright (c) 2012,2013,2014, by the GROMACS development team, led by
   5  * Mark Abraham, David van der Spoel, Berk Hess, and Erik Lindahl,
   6  * and including many others, as listed in the AUTHORS file in the
   7  * top-level source directory and at http://www.gromacs.org.
   8  *
   9  * GROMACS is free software; you can redistribute it and/or
  10  * modify it under the terms of the GNU Lesser General Public License
  11  * as published by the Free Software Foundation; either version 2.1
  12  * of the License, or (at your option) any later version.
  13  *
  14  * GROMACS is distributed in the hope that it will be useful,
  15  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  16  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  17  * Lesser General Public License for more details.
  18  *
  19  * You should have received a copy of the GNU Lesser General Public
  20  * License along with GROMACS; if not, see
  21  * http://www.gnu.org/licenses, or write to the Free Software Foundation,
  22  * Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02110-1301  USA.
  23  *
  24  * If you want to redistribute modifications to GROMACS, please
  25  * consider that scientific software is very special. Version
  26  * control is crucial - bugs must be traceable. We will be happy to
  27  * consider code for inclusion in the official distribution, but
  28  * derived work must not be called official GROMACS. Details are found
  29  * in the README & COPYING files - if they are missing, get the
  30  * official version at http://www.gromacs.org.
  31  *
  32  * To help us fund GROMACS development, we humbly ask that you cite
  33  * the research papers on the package. Check out http://www.gromacs.org.
  34  */
  35 /*
  36  * Note: this file was generated by the GROMACS avx_256_single 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/legacyheaders/types/simple.h"
  46 #include "gromacs/math/vec.h"
  47 #include "gromacs/legacyheaders/nrnb.h"
  48
  49 #include "gromacs/simd/math_x86_avx_256_single.h"
  50 #include "kernelutil_x86_avx_256_single.h"
  51
  52 /*
  53  * Gromacs nonbonded kernel:   nb_kernel_ElecEw_VdwLJEw_GeomW3W3_VF_avx_256_single
  54  * Electrostatics interaction: Ewald
  55  * VdW interaction:            LJEwald
  56  * Geometry:                   Water3-Water3
  57  * Calculate force/pot:        PotentialAndForce
  58  */
  59 void
  60 nb_kernel_ElecEw_VdwLJEw_GeomW3W3_VF_avx_256_single
  61                     (t_nblist                    * gmx_restrict       nlist,
  62                      rvec                        * gmx_restrict          xx,
  63                      rvec                        * gmx_restrict          ff,
  64                      t_forcerec                  * gmx_restrict          fr,
  65                      t_mdatoms                   * gmx_restrict     mdatoms,
  66                      nb_kernel_data_t gmx_unused * gmx_restrict kernel_data,
  67                      t_nrnb                      * gmx_restrict        nrnb)
  68 {
  69     /* Suffixes 0,1,2,3 refer to particle indices for waters in the inner or outer loop, or
  70      * just 0 for non-waters.
  71      * Suffixes A,B,C,D,E,F,G,H refer to j loop unrolling done with AVX, e.g. for the eight different
  72      * jnr indices corresponding to data put in the four positions in the SIMD register.
  73      */
  74     int              i_shift_offset,i_coord_offset,outeriter,inneriter;
  75     int              j_index_start,j_index_end,jidx,nri,inr,ggid,iidx;
  76     int              jnrA,jnrB,jnrC,jnrD;
  77     int              jnrE,jnrF,jnrG,jnrH;
  78     int              jnrlistA,jnrlistB,jnrlistC,jnrlistD;
  79     int              jnrlistE,jnrlistF,jnrlistG,jnrlistH;
  80     int              j_coord_offsetA,j_coord_offsetB,j_coord_offsetC,j_coord_offsetD;
  81     int              j_coord_offsetE,j_coord_offsetF,j_coord_offsetG,j_coord_offsetH;
  82     int              *iinr,*jindex,*jjnr,*shiftidx,*gid;
  83     real             rcutoff_scalar;
  84     real             *shiftvec,*fshift,*x,*f;
  85     real             *fjptrA,*fjptrB,*fjptrC,*fjptrD,*fjptrE,*fjptrF,*fjptrG,*fjptrH;
  86     real             scratch[4*DIM];
  87     __m256           tx,ty,tz,fscal,rcutoff,rcutoff2,jidxall;
  88     real *           vdwioffsetptr0;
  89     real *           vdwgridioffsetptr0;
  90     __m256           ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
  91     real *           vdwioffsetptr1;
  92     real *           vdwgridioffsetptr1;
  93     __m256           ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
  94     real *           vdwioffsetptr2;
  95     real *           vdwgridioffsetptr2;
  96     __m256           ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
  97     int              vdwjidx0A,vdwjidx0B,vdwjidx0C,vdwjidx0D,vdwjidx0E,vdwjidx0F,vdwjidx0G,vdwjidx0H;
  98     __m256           jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
  99     int              vdwjidx1A,vdwjidx1B,vdwjidx1C,vdwjidx1D,vdwjidx1E,vdwjidx1F,vdwjidx1G,vdwjidx1H;
 100     __m256           jx1,jy1,jz1,fjx1,fjy1,fjz1,jq1,isaj1;
 101     int              vdwjidx2A,vdwjidx2B,vdwjidx2C,vdwjidx2D,vdwjidx2E,vdwjidx2F,vdwjidx2G,vdwjidx2H;
 102     __m256           jx2,jy2,jz2,fjx2,fjy2,fjz2,jq2,isaj2;
 103     __m256           dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00;
 104     __m256           dx01,dy01,dz01,rsq01,rinv01,rinvsq01,r01,qq01,c6_01,c12_01;
 105     __m256           dx02,dy02,dz02,rsq02,rinv02,rinvsq02,r02,qq02,c6_02,c12_02;
 106     __m256           dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10;
 107     __m256           dx11,dy11,dz11,rsq11,rinv11,rinvsq11,r11,qq11,c6_11,c12_11;
 108     __m256           dx12,dy12,dz12,rsq12,rinv12,rinvsq12,r12,qq12,c6_12,c12_12;
 109     __m256           dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20;
 110     __m256           dx21,dy21,dz21,rsq21,rinv21,rinvsq21,r21,qq21,c6_21,c12_21;
 111     __m256           dx22,dy22,dz22,rsq22,rinv22,rinvsq22,r22,qq22,c6_22,c12_22;
 112     __m256           velec,felec,velecsum,facel,crf,krf,krf2;
 113     real             *charge;
 114     int              nvdwtype;
 115     __m256           rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,sh_vdw_invrcut6;
 116     int              *vdwtype;
 117     real             *vdwparam;
 118     __m256           one_sixth   = _mm256_set1_ps(1.0/6.0);
 119     __m256           one_twelfth = _mm256_set1_ps(1.0/12.0);
 120     __m256           c6grid_00;
 121     __m256           c6grid_01;
 122     __m256           c6grid_02;
 123     __m256           c6grid_10;
 124     __m256           c6grid_11;
 125     __m256           c6grid_12;
 126     __m256           c6grid_20;
 127     __m256           c6grid_21;
 128     __m256           c6grid_22;
 129     real             *vdwgridparam;
 130     __m256           ewclj,ewclj2,ewclj6,ewcljrsq,poly,exponent,f6A,f6B,sh_lj_ewald;
 131     __m256           one_half  = _mm256_set1_ps(0.5);
 132     __m256           minus_one = _mm256_set1_ps(-1.0);
 133     __m256i          ewitab;
 134     __m128i          ewitab_lo,ewitab_hi;
 135     __m256           ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace,ewtabF,ewtabFn,ewtabD,ewtabV;
 136     __m256           beta,beta2,beta3,zeta2,pmecorrF,pmecorrV,rinv3;
 137     real             *ewtab;
 138     __m256           dummy_mask,cutoff_mask;
 139     __m256           signbit = _mm256_castsi256_ps( _mm256_set1_epi32(0x80000000) );
 140     __m256           one     = _mm256_set1_ps(1.0);
 141     __m256           two     = _mm256_set1_ps(2.0);
 142     x                = xx[0];
 143     f                = ff[0];
 144
 145     nri              = nlist->nri;
 146     iinr             = nlist->iinr;
 147     jindex           = nlist->jindex;
 148     jjnr             = nlist->jjnr;
 149     shiftidx         = nlist->shift;
 150     gid              = nlist->gid;
 151     shiftvec         = fr->shift_vec[0];
 152     fshift           = fr->fshift[0];
 153     facel            = _mm256_set1_ps(fr->epsfac);
 154     charge           = mdatoms->chargeA;
 155     nvdwtype         = fr->ntype;
 156     vdwparam         = fr->nbfp;
 157     vdwtype          = mdatoms->typeA;
 158     vdwgridparam     = fr->ljpme_c6grid;
 159     sh_lj_ewald      = _mm256_set1_ps(fr->ic->sh_lj_ewald);
 160     ewclj            = _mm256_set1_ps(fr->ewaldcoeff_lj);
 161     ewclj2           = _mm256_mul_ps(minus_one,_mm256_mul_ps(ewclj,ewclj));
 162
 163     sh_ewald         = _mm256_set1_ps(fr->ic->sh_ewald);
 164     beta             = _mm256_set1_ps(fr->ic->ewaldcoeff_q);
 165     beta2            = _mm256_mul_ps(beta,beta);
 166     beta3            = _mm256_mul_ps(beta,beta2);
 167
 168     ewtab            = fr->ic->tabq_coul_FDV0;
 169     ewtabscale       = _mm256_set1_ps(fr->ic->tabq_scale);
 170     ewtabhalfspace   = _mm256_set1_ps(0.5/fr->ic->tabq_scale);
 171
 172     /* Setup water-specific parameters */
 173     inr              = nlist->iinr[0];
 174     iq0              = _mm256_mul_ps(facel,_mm256_set1_ps(charge[inr+0]));
 175     iq1              = _mm256_mul_ps(facel,_mm256_set1_ps(charge[inr+1]));
 176     iq2              = _mm256_mul_ps(facel,_mm256_set1_ps(charge[inr+2]));
 177     vdwioffsetptr0   = vdwparam+2*nvdwtype*vdwtype[inr+0];
 178     vdwgridioffsetptr0 = vdwgridparam+2*nvdwtype*vdwtype[inr+0];
 179
 180     jq0              = _mm256_set1_ps(charge[inr+0]);
 181     jq1              = _mm256_set1_ps(charge[inr+1]);
 182     jq2              = _mm256_set1_ps(charge[inr+2]);
 183     vdwjidx0A        = 2*vdwtype[inr+0];
 184     qq00             = _mm256_mul_ps(iq0,jq0);
 185     c6_00            = _mm256_set1_ps(vdwioffsetptr0[vdwjidx0A]);
 186     c12_00           = _mm256_set1_ps(vdwioffsetptr0[vdwjidx0A+1]);
 187     c6grid_00        = _mm256_set1_ps(vdwgridioffsetptr0[vdwjidx0A]);
 188     qq01             = _mm256_mul_ps(iq0,jq1);
 189     qq02             = _mm256_mul_ps(iq0,jq2);
 190     qq10             = _mm256_mul_ps(iq1,jq0);
 191     qq11             = _mm256_mul_ps(iq1,jq1);
 192     qq12             = _mm256_mul_ps(iq1,jq2);
 193     qq20             = _mm256_mul_ps(iq2,jq0);
 194     qq21             = _mm256_mul_ps(iq2,jq1);
 195     qq22             = _mm256_mul_ps(iq2,jq2);
 196
 197     /* Avoid stupid compiler warnings */
 198     jnrA = jnrB = jnrC = jnrD = jnrE = jnrF = jnrG = jnrH = 0;
 199     j_coord_offsetA = 0;
 200     j_coord_offsetB = 0;
 201     j_coord_offsetC = 0;
 202     j_coord_offsetD = 0;
 203     j_coord_offsetE = 0;
 204     j_coord_offsetF = 0;
 205     j_coord_offsetG = 0;
 206     j_coord_offsetH = 0;
 207
 208     outeriter        = 0;
 209     inneriter        = 0;
 210
 211     for(iidx=0;iidx<4*DIM;iidx++)
 212     {
 213         scratch[iidx] = 0.0;
 214     }
 215
 216     /* Start outer loop over neighborlists */
 217     for(iidx=0; iidx<nri; iidx++)
 218     {
 219         /* Load shift vector for this list */
 220         i_shift_offset   = DIM*shiftidx[iidx];
 221
 222         /* Load limits for loop over neighbors */
 223         j_index_start    = jindex[iidx];
 224         j_index_end      = jindex[iidx+1];
 225
 226         /* Get outer coordinate index */
 227         inr              = iinr[iidx];
 228         i_coord_offset   = DIM*inr;
 229
 230         /* Load i particle coords and add shift vector */
 231         gmx_mm256_load_shift_and_3rvec_broadcast_ps(shiftvec+i_shift_offset,x+i_coord_offset,
 232                                                     &ix0,&iy0,&iz0,&ix1,&iy1,&iz1,&ix2,&iy2,&iz2);
 233
 234         fix0             = _mm256_setzero_ps();
 235         fiy0             = _mm256_setzero_ps();
 236         fiz0             = _mm256_setzero_ps();
 237         fix1             = _mm256_setzero_ps();
 238         fiy1             = _mm256_setzero_ps();
 239         fiz1             = _mm256_setzero_ps();
 240         fix2             = _mm256_setzero_ps();
 241         fiy2             = _mm256_setzero_ps();
 242         fiz2             = _mm256_setzero_ps();
 243
 244         /* Reset potential sums */
 245         velecsum         = _mm256_setzero_ps();
 246         vvdwsum          = _mm256_setzero_ps();
 247
 248         /* Start inner kernel loop */
 249         for(jidx=j_index_start; jidx<j_index_end && jjnr[jidx+7]>=0; jidx+=8)
 250         {
 251
 252             /* Get j neighbor index, and coordinate index */
 253             jnrA             = jjnr[jidx];
 254             jnrB             = jjnr[jidx+1];
 255             jnrC             = jjnr[jidx+2];
 256             jnrD             = jjnr[jidx+3];
 257             jnrE             = jjnr[jidx+4];
 258             jnrF             = jjnr[jidx+5];
 259             jnrG             = jjnr[jidx+6];
 260             jnrH             = jjnr[jidx+7];
 261             j_coord_offsetA  = DIM*jnrA;
 262             j_coord_offsetB  = DIM*jnrB;
 263             j_coord_offsetC  = DIM*jnrC;
 264             j_coord_offsetD  = DIM*jnrD;
 265             j_coord_offsetE  = DIM*jnrE;
 266             j_coord_offsetF  = DIM*jnrF;
 267             j_coord_offsetG  = DIM*jnrG;
 268             j_coord_offsetH  = DIM*jnrH;
 269
 270             /* load j atom coordinates */
 271             gmx_mm256_load_3rvec_8ptr_swizzle_ps(x+j_coord_offsetA,x+j_coord_offsetB,
 272                                                  x+j_coord_offsetC,x+j_coord_offsetD,
 273                                                  x+j_coord_offsetE,x+j_coord_offsetF,
 274                                                  x+j_coord_offsetG,x+j_coord_offsetH,
 275                                               &jx0,&jy0,&jz0,&jx1,&jy1,&jz1,&jx2,&jy2,&jz2);
 276
 277             /* Calculate displacement vector */
 278             dx00             = _mm256_sub_ps(ix0,jx0);
 279             dy00             = _mm256_sub_ps(iy0,jy0);
 280             dz00             = _mm256_sub_ps(iz0,jz0);
 281             dx01             = _mm256_sub_ps(ix0,jx1);
 282             dy01             = _mm256_sub_ps(iy0,jy1);
 283             dz01             = _mm256_sub_ps(iz0,jz1);
 284             dx02             = _mm256_sub_ps(ix0,jx2);
 285             dy02             = _mm256_sub_ps(iy0,jy2);
 286             dz02             = _mm256_sub_ps(iz0,jz2);
 287             dx10             = _mm256_sub_ps(ix1,jx0);
 288             dy10             = _mm256_sub_ps(iy1,jy0);
 289             dz10             = _mm256_sub_ps(iz1,jz0);
 290             dx11             = _mm256_sub_ps(ix1,jx1);
 291             dy11             = _mm256_sub_ps(iy1,jy1);
 292             dz11             = _mm256_sub_ps(iz1,jz1);
 293             dx12             = _mm256_sub_ps(ix1,jx2);
 294             dy12             = _mm256_sub_ps(iy1,jy2);
 295             dz12             = _mm256_sub_ps(iz1,jz2);
 296             dx20             = _mm256_sub_ps(ix2,jx0);
 297             dy20             = _mm256_sub_ps(iy2,jy0);
 298             dz20             = _mm256_sub_ps(iz2,jz0);
 299             dx21             = _mm256_sub_ps(ix2,jx1);
 300             dy21             = _mm256_sub_ps(iy2,jy1);
 301             dz21             = _mm256_sub_ps(iz2,jz1);
 302             dx22             = _mm256_sub_ps(ix2,jx2);
 303             dy22             = _mm256_sub_ps(iy2,jy2);
 304             dz22             = _mm256_sub_ps(iz2,jz2);
 305
 306             /* Calculate squared distance and things based on it */
 307             rsq00            = gmx_mm256_calc_rsq_ps(dx00,dy00,dz00);
 308             rsq01            = gmx_mm256_calc_rsq_ps(dx01,dy01,dz01);
 309             rsq02            = gmx_mm256_calc_rsq_ps(dx02,dy02,dz02);
 310             rsq10            = gmx_mm256_calc_rsq_ps(dx10,dy10,dz10);
 311             rsq11            = gmx_mm256_calc_rsq_ps(dx11,dy11,dz11);
 312             rsq12            = gmx_mm256_calc_rsq_ps(dx12,dy12,dz12);
 313             rsq20            = gmx_mm256_calc_rsq_ps(dx20,dy20,dz20);
 314             rsq21            = gmx_mm256_calc_rsq_ps(dx21,dy21,dz21);
 315             rsq22            = gmx_mm256_calc_rsq_ps(dx22,dy22,dz22);
 316
 317             rinv00           = gmx_mm256_invsqrt_ps(rsq00);
 318             rinv01           = gmx_mm256_invsqrt_ps(rsq01);
 319             rinv02           = gmx_mm256_invsqrt_ps(rsq02);
 320             rinv10           = gmx_mm256_invsqrt_ps(rsq10);
 321             rinv11           = gmx_mm256_invsqrt_ps(rsq11);
 322             rinv12           = gmx_mm256_invsqrt_ps(rsq12);
 323             rinv20           = gmx_mm256_invsqrt_ps(rsq20);
 324             rinv21           = gmx_mm256_invsqrt_ps(rsq21);
 325             rinv22           = gmx_mm256_invsqrt_ps(rsq22);
 326
 327             rinvsq00         = _mm256_mul_ps(rinv00,rinv00);
 328             rinvsq01         = _mm256_mul_ps(rinv01,rinv01);
 329             rinvsq02         = _mm256_mul_ps(rinv02,rinv02);
 330             rinvsq10         = _mm256_mul_ps(rinv10,rinv10);
 331             rinvsq11         = _mm256_mul_ps(rinv11,rinv11);
 332             rinvsq12         = _mm256_mul_ps(rinv12,rinv12);
 333             rinvsq20         = _mm256_mul_ps(rinv20,rinv20);
 334             rinvsq21         = _mm256_mul_ps(rinv21,rinv21);
 335             rinvsq22         = _mm256_mul_ps(rinv22,rinv22);
 336
 337             fjx0             = _mm256_setzero_ps();
 338             fjy0             = _mm256_setzero_ps();
 339             fjz0             = _mm256_setzero_ps();
 340             fjx1             = _mm256_setzero_ps();
 341             fjy1             = _mm256_setzero_ps();
 342             fjz1             = _mm256_setzero_ps();
 343             fjx2             = _mm256_setzero_ps();
 344             fjy2             = _mm256_setzero_ps();
 345             fjz2             = _mm256_setzero_ps();
 346
 347             /**************************
 348              * CALCULATE INTERACTIONS *
 349              **************************/
 350
 351             r00              = _mm256_mul_ps(rsq00,rinv00);
 352
 353             /* EWALD ELECTROSTATICS */
 354
 355             /* Analytical PME correction */
 356             zeta2            = _mm256_mul_ps(beta2,rsq00);
 357             rinv3            = _mm256_mul_ps(rinvsq00,rinv00);
 358             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
 359             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
 360             felec            = _mm256_mul_ps(qq00,felec);
 361             pmecorrV         = gmx_mm256_pmecorrV_ps(zeta2);
 362             pmecorrV         = _mm256_mul_ps(pmecorrV,beta);
 363             velec            = _mm256_sub_ps(rinv00,pmecorrV);
 364             velec            = _mm256_mul_ps(qq00,velec);
 365
 366             /* Analytical LJ-PME */
 367             rinvsix          = _mm256_mul_ps(_mm256_mul_ps(rinvsq00,rinvsq00),rinvsq00);
 368             ewcljrsq         = _mm256_mul_ps(ewclj2,rsq00);
 369             ewclj6           = _mm256_mul_ps(ewclj2,_mm256_mul_ps(ewclj2,ewclj2));
 370             exponent         = gmx_simd_exp_r(ewcljrsq);
 371             /* poly = exp(-(beta*r)^2) * (1 + (beta*r)^2 + (beta*r)^4 /2) */
 372             poly             = _mm256_mul_ps(exponent,_mm256_add_ps(_mm256_sub_ps(one,ewcljrsq),_mm256_mul_ps(_mm256_mul_ps(ewcljrsq,ewcljrsq),one_half)));
 373             /* vvdw6 = [C6 - C6grid * (1-poly)]/r6 */
 374             vvdw6            = _mm256_mul_ps(_mm256_sub_ps(c6_00,_mm256_mul_ps(c6grid_00,_mm256_sub_ps(one,poly))),rinvsix);
 375             vvdw12           = _mm256_mul_ps(c12_00,_mm256_mul_ps(rinvsix,rinvsix));
 376             vvdw             = _mm256_sub_ps(_mm256_mul_ps(vvdw12,one_twelfth),_mm256_mul_ps(vvdw6,one_sixth));
 377             /* fvdw = vvdw12/r - (vvdw6/r + (C6grid * exponent * beta^6)/r) */
 378             fvdw             = _mm256_mul_ps(_mm256_sub_ps(vvdw12,_mm256_sub_ps(vvdw6,_mm256_mul_ps(_mm256_mul_ps(c6grid_00,one_sixth),_mm256_mul_ps(exponent,ewclj6)))),rinvsq00);
 379
 380             /* Update potential sum for this i atom from the interaction with this j atom. */
 381             velecsum         = _mm256_add_ps(velecsum,velec);
 382             vvdwsum          = _mm256_add_ps(vvdwsum,vvdw);
 383
 384             fscal            = _mm256_add_ps(felec,fvdw);
 385
 386             /* Calculate temporary vectorial force */
 387             tx               = _mm256_mul_ps(fscal,dx00);
 388             ty               = _mm256_mul_ps(fscal,dy00);
 389             tz               = _mm256_mul_ps(fscal,dz00);
 390
 391             /* Update vectorial force */
 392             fix0             = _mm256_add_ps(fix0,tx);
 393             fiy0             = _mm256_add_ps(fiy0,ty);
 394             fiz0             = _mm256_add_ps(fiz0,tz);
 395
 396             fjx0             = _mm256_add_ps(fjx0,tx);
 397             fjy0             = _mm256_add_ps(fjy0,ty);
 398             fjz0             = _mm256_add_ps(fjz0,tz);
 399
 400             /**************************
 401              * CALCULATE INTERACTIONS *
 402              **************************/
 403
 404             r01              = _mm256_mul_ps(rsq01,rinv01);
 405
 406             /* EWALD ELECTROSTATICS */
 407
 408             /* Analytical PME correction */
 409             zeta2            = _mm256_mul_ps(beta2,rsq01);
 410             rinv3            = _mm256_mul_ps(rinvsq01,rinv01);
 411             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
 412             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
 413             felec            = _mm256_mul_ps(qq01,felec);
 414             pmecorrV         = gmx_mm256_pmecorrV_ps(zeta2);
 415             pmecorrV         = _mm256_mul_ps(pmecorrV,beta);
 416             velec            = _mm256_sub_ps(rinv01,pmecorrV);
 417             velec            = _mm256_mul_ps(qq01,velec);
 418
 419             /* Update potential sum for this i atom from the interaction with this j atom. */
 420             velecsum         = _mm256_add_ps(velecsum,velec);
 421
 422             fscal            = felec;
 423
 424             /* Calculate temporary vectorial force */
 425             tx               = _mm256_mul_ps(fscal,dx01);
 426             ty               = _mm256_mul_ps(fscal,dy01);
 427             tz               = _mm256_mul_ps(fscal,dz01);
 428
 429             /* Update vectorial force */
 430             fix0             = _mm256_add_ps(fix0,tx);
 431             fiy0             = _mm256_add_ps(fiy0,ty);
 432             fiz0             = _mm256_add_ps(fiz0,tz);
 433
 434             fjx1             = _mm256_add_ps(fjx1,tx);
 435             fjy1             = _mm256_add_ps(fjy1,ty);
 436             fjz1             = _mm256_add_ps(fjz1,tz);
 437
 438             /**************************
 439              * CALCULATE INTERACTIONS *
 440              **************************/
 441
 442             r02              = _mm256_mul_ps(rsq02,rinv02);
 443
 444             /* EWALD ELECTROSTATICS */
 445
 446             /* Analytical PME correction */
 447             zeta2            = _mm256_mul_ps(beta2,rsq02);
 448             rinv3            = _mm256_mul_ps(rinvsq02,rinv02);
 449             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
 450             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
 451             felec            = _mm256_mul_ps(qq02,felec);
 452             pmecorrV         = gmx_mm256_pmecorrV_ps(zeta2);
 453             pmecorrV         = _mm256_mul_ps(pmecorrV,beta);
 454             velec            = _mm256_sub_ps(rinv02,pmecorrV);
 455             velec            = _mm256_mul_ps(qq02,velec);
 456
 457             /* Update potential sum for this i atom from the interaction with this j atom. */
 458             velecsum         = _mm256_add_ps(velecsum,velec);
 459
 460             fscal            = felec;
 461
 462             /* Calculate temporary vectorial force */
 463             tx               = _mm256_mul_ps(fscal,dx02);
 464             ty               = _mm256_mul_ps(fscal,dy02);
 465             tz               = _mm256_mul_ps(fscal,dz02);
 466
 467             /* Update vectorial force */
 468             fix0             = _mm256_add_ps(fix0,tx);
 469             fiy0             = _mm256_add_ps(fiy0,ty);
 470             fiz0             = _mm256_add_ps(fiz0,tz);
 471
 472             fjx2             = _mm256_add_ps(fjx2,tx);
 473             fjy2             = _mm256_add_ps(fjy2,ty);
 474             fjz2             = _mm256_add_ps(fjz2,tz);
 475
 476             /**************************
 477              * CALCULATE INTERACTIONS *
 478              **************************/
 479
 480             r10              = _mm256_mul_ps(rsq10,rinv10);
 481
 482             /* EWALD ELECTROSTATICS */
 483
 484             /* Analytical PME correction */
 485             zeta2            = _mm256_mul_ps(beta2,rsq10);
 486             rinv3            = _mm256_mul_ps(rinvsq10,rinv10);
 487             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
 488             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
 489             felec            = _mm256_mul_ps(qq10,felec);
 490             pmecorrV         = gmx_mm256_pmecorrV_ps(zeta2);
 491             pmecorrV         = _mm256_mul_ps(pmecorrV,beta);
 492             velec            = _mm256_sub_ps(rinv10,pmecorrV);
 493             velec            = _mm256_mul_ps(qq10,velec);
 494
 495             /* Update potential sum for this i atom from the interaction with this j atom. */
 496             velecsum         = _mm256_add_ps(velecsum,velec);
 497
 498             fscal            = felec;
 499
 500             /* Calculate temporary vectorial force */
 501             tx               = _mm256_mul_ps(fscal,dx10);
 502             ty               = _mm256_mul_ps(fscal,dy10);
 503             tz               = _mm256_mul_ps(fscal,dz10);
 504
 505             /* Update vectorial force */
 506             fix1             = _mm256_add_ps(fix1,tx);
 507             fiy1             = _mm256_add_ps(fiy1,ty);
 508             fiz1             = _mm256_add_ps(fiz1,tz);
 509
 510             fjx0             = _mm256_add_ps(fjx0,tx);
 511             fjy0             = _mm256_add_ps(fjy0,ty);
 512             fjz0             = _mm256_add_ps(fjz0,tz);
 513
 514             /**************************
 515              * CALCULATE INTERACTIONS *
 516              **************************/
 517
 518             r11              = _mm256_mul_ps(rsq11,rinv11);
 519
 520             /* EWALD ELECTROSTATICS */
 521
 522             /* Analytical PME correction */
 523             zeta2            = _mm256_mul_ps(beta2,rsq11);
 524             rinv3            = _mm256_mul_ps(rinvsq11,rinv11);
 525             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
 526             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
 527             felec            = _mm256_mul_ps(qq11,felec);
 528             pmecorrV         = gmx_mm256_pmecorrV_ps(zeta2);
 529             pmecorrV         = _mm256_mul_ps(pmecorrV,beta);
 530             velec            = _mm256_sub_ps(rinv11,pmecorrV);
 531             velec            = _mm256_mul_ps(qq11,velec);
 532
 533             /* Update potential sum for this i atom from the interaction with this j atom. */
 534             velecsum         = _mm256_add_ps(velecsum,velec);
 535
 536             fscal            = felec;
 537
 538             /* Calculate temporary vectorial force */
 539             tx               = _mm256_mul_ps(fscal,dx11);
 540             ty               = _mm256_mul_ps(fscal,dy11);
 541             tz               = _mm256_mul_ps(fscal,dz11);
 542
 543             /* Update vectorial force */
 544             fix1             = _mm256_add_ps(fix1,tx);
 545             fiy1             = _mm256_add_ps(fiy1,ty);
 546             fiz1             = _mm256_add_ps(fiz1,tz);
 547
 548             fjx1             = _mm256_add_ps(fjx1,tx);
 549             fjy1             = _mm256_add_ps(fjy1,ty);
 550             fjz1             = _mm256_add_ps(fjz1,tz);
 551
 552             /**************************
 553              * CALCULATE INTERACTIONS *
 554              **************************/
 555
 556             r12              = _mm256_mul_ps(rsq12,rinv12);
 557
 558             /* EWALD ELECTROSTATICS */
 559
 560             /* Analytical PME correction */
 561             zeta2            = _mm256_mul_ps(beta2,rsq12);
 562             rinv3            = _mm256_mul_ps(rinvsq12,rinv12);
 563             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
 564             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
 565             felec            = _mm256_mul_ps(qq12,felec);
 566             pmecorrV         = gmx_mm256_pmecorrV_ps(zeta2);
 567             pmecorrV         = _mm256_mul_ps(pmecorrV,beta);
 568             velec            = _mm256_sub_ps(rinv12,pmecorrV);
 569             velec            = _mm256_mul_ps(qq12,velec);
 570
 571             /* Update potential sum for this i atom from the interaction with this j atom. */
 572             velecsum         = _mm256_add_ps(velecsum,velec);
 573
 574             fscal            = felec;
 575
 576             /* Calculate temporary vectorial force */
 577             tx               = _mm256_mul_ps(fscal,dx12);
 578             ty               = _mm256_mul_ps(fscal,dy12);
 579             tz               = _mm256_mul_ps(fscal,dz12);
 580
 581             /* Update vectorial force */
 582             fix1             = _mm256_add_ps(fix1,tx);
 583             fiy1             = _mm256_add_ps(fiy1,ty);
 584             fiz1             = _mm256_add_ps(fiz1,tz);
 585
 586             fjx2             = _mm256_add_ps(fjx2,tx);
 587             fjy2             = _mm256_add_ps(fjy2,ty);
 588             fjz2             = _mm256_add_ps(fjz2,tz);
 589
 590             /**************************
 591              * CALCULATE INTERACTIONS *
 592              **************************/
 593
 594             r20              = _mm256_mul_ps(rsq20,rinv20);
 595
 596             /* EWALD ELECTROSTATICS */
 597
 598             /* Analytical PME correction */
 599             zeta2            = _mm256_mul_ps(beta2,rsq20);
 600             rinv3            = _mm256_mul_ps(rinvsq20,rinv20);
 601             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
 602             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
 603             felec            = _mm256_mul_ps(qq20,felec);
 604             pmecorrV         = gmx_mm256_pmecorrV_ps(zeta2);
 605             pmecorrV         = _mm256_mul_ps(pmecorrV,beta);
 606             velec            = _mm256_sub_ps(rinv20,pmecorrV);
 607             velec            = _mm256_mul_ps(qq20,velec);
 608
 609             /* Update potential sum for this i atom from the interaction with this j atom. */
 610             velecsum         = _mm256_add_ps(velecsum,velec);
 611
 612             fscal            = felec;
 613
 614             /* Calculate temporary vectorial force */
 615             tx               = _mm256_mul_ps(fscal,dx20);
 616             ty               = _mm256_mul_ps(fscal,dy20);
 617             tz               = _mm256_mul_ps(fscal,dz20);
 618
 619             /* Update vectorial force */
 620             fix2             = _mm256_add_ps(fix2,tx);
 621             fiy2             = _mm256_add_ps(fiy2,ty);
 622             fiz2             = _mm256_add_ps(fiz2,tz);
 623
 624             fjx0             = _mm256_add_ps(fjx0,tx);
 625             fjy0             = _mm256_add_ps(fjy0,ty);
 626             fjz0             = _mm256_add_ps(fjz0,tz);
 627
 628             /**************************
 629              * CALCULATE INTERACTIONS *
 630              **************************/
 631
 632             r21              = _mm256_mul_ps(rsq21,rinv21);
 633
 634             /* EWALD ELECTROSTATICS */
 635
 636             /* Analytical PME correction */
 637             zeta2            = _mm256_mul_ps(beta2,rsq21);
 638             rinv3            = _mm256_mul_ps(rinvsq21,rinv21);
 639             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
 640             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
 641             felec            = _mm256_mul_ps(qq21,felec);
 642             pmecorrV         = gmx_mm256_pmecorrV_ps(zeta2);
 643             pmecorrV         = _mm256_mul_ps(pmecorrV,beta);
 644             velec            = _mm256_sub_ps(rinv21,pmecorrV);
 645             velec            = _mm256_mul_ps(qq21,velec);
 646
 647             /* Update potential sum for this i atom from the interaction with this j atom. */
 648             velecsum         = _mm256_add_ps(velecsum,velec);
 649
 650             fscal            = felec;
 651
 652             /* Calculate temporary vectorial force */
 653             tx               = _mm256_mul_ps(fscal,dx21);
 654             ty               = _mm256_mul_ps(fscal,dy21);
 655             tz               = _mm256_mul_ps(fscal,dz21);
 656
 657             /* Update vectorial force */
 658             fix2             = _mm256_add_ps(fix2,tx);
 659             fiy2             = _mm256_add_ps(fiy2,ty);
 660             fiz2             = _mm256_add_ps(fiz2,tz);
 661
 662             fjx1             = _mm256_add_ps(fjx1,tx);
 663             fjy1             = _mm256_add_ps(fjy1,ty);
 664             fjz1             = _mm256_add_ps(fjz1,tz);
 665
 666             /**************************
 667              * CALCULATE INTERACTIONS *
 668              **************************/
 669
 670             r22              = _mm256_mul_ps(rsq22,rinv22);
 671
 672             /* EWALD ELECTROSTATICS */
 673
 674             /* Analytical PME correction */
 675             zeta2            = _mm256_mul_ps(beta2,rsq22);
 676             rinv3            = _mm256_mul_ps(rinvsq22,rinv22);
 677             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
 678             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
 679             felec            = _mm256_mul_ps(qq22,felec);
 680             pmecorrV         = gmx_mm256_pmecorrV_ps(zeta2);
 681             pmecorrV         = _mm256_mul_ps(pmecorrV,beta);
 682             velec            = _mm256_sub_ps(rinv22,pmecorrV);
 683             velec            = _mm256_mul_ps(qq22,velec);
 684
 685             /* Update potential sum for this i atom from the interaction with this j atom. */
 686             velecsum         = _mm256_add_ps(velecsum,velec);
 687
 688             fscal            = felec;
 689
 690             /* Calculate temporary vectorial force */
 691             tx               = _mm256_mul_ps(fscal,dx22);
 692             ty               = _mm256_mul_ps(fscal,dy22);
 693             tz               = _mm256_mul_ps(fscal,dz22);
 694
 695             /* Update vectorial force */
 696             fix2             = _mm256_add_ps(fix2,tx);
 697             fiy2             = _mm256_add_ps(fiy2,ty);
 698             fiz2             = _mm256_add_ps(fiz2,tz);
 699
 700             fjx2             = _mm256_add_ps(fjx2,tx);
 701             fjy2             = _mm256_add_ps(fjy2,ty);
 702             fjz2             = _mm256_add_ps(fjz2,tz);
 703
 704             fjptrA             = f+j_coord_offsetA;
 705             fjptrB             = f+j_coord_offsetB;
 706             fjptrC             = f+j_coord_offsetC;
 707             fjptrD             = f+j_coord_offsetD;
 708             fjptrE             = f+j_coord_offsetE;
 709             fjptrF             = f+j_coord_offsetF;
 710             fjptrG             = f+j_coord_offsetG;
 711             fjptrH             = f+j_coord_offsetH;
 712
 713             gmx_mm256_decrement_3rvec_8ptr_swizzle_ps(fjptrA,fjptrB,fjptrC,fjptrD,fjptrE,fjptrF,fjptrG,fjptrH,
 714                                                       fjx0,fjy0,fjz0,fjx1,fjy1,fjz1,fjx2,fjy2,fjz2);
 715
 716             /* Inner loop uses 784 flops */
 717         }
 718
 719         if(jidx<j_index_end)
 720         {
 721
 722             /* Get j neighbor index, and coordinate index */
 723             jnrlistA         = jjnr[jidx];
 724             jnrlistB         = jjnr[jidx+1];
 725             jnrlistC         = jjnr[jidx+2];
 726             jnrlistD         = jjnr[jidx+3];
 727             jnrlistE         = jjnr[jidx+4];
 728             jnrlistF         = jjnr[jidx+5];
 729             jnrlistG         = jjnr[jidx+6];
 730             jnrlistH         = jjnr[jidx+7];
 731             /* Sign of each element will be negative for non-real atoms.
 732              * This mask will be 0xFFFFFFFF for dummy entries and 0x0 for real ones,
 733              * so use it as val = _mm_andnot_ps(mask,val) to clear dummy entries.
 734              */
 735             dummy_mask = gmx_mm256_set_m128(gmx_mm_castsi128_ps(_mm_cmplt_epi32(_mm_loadu_si128((const __m128i *)(jjnr+jidx+4)),_mm_setzero_si128())),
 736                                             gmx_mm_castsi128_ps(_mm_cmplt_epi32(_mm_loadu_si128((const __m128i *)(jjnr+jidx)),_mm_setzero_si128())));
 737
 738             jnrA       = (jnrlistA>=0) ? jnrlistA : 0;
 739             jnrB       = (jnrlistB>=0) ? jnrlistB : 0;
 740             jnrC       = (jnrlistC>=0) ? jnrlistC : 0;
 741             jnrD       = (jnrlistD>=0) ? jnrlistD : 0;
 742             jnrE       = (jnrlistE>=0) ? jnrlistE : 0;
 743             jnrF       = (jnrlistF>=0) ? jnrlistF : 0;
 744             jnrG       = (jnrlistG>=0) ? jnrlistG : 0;
 745             jnrH       = (jnrlistH>=0) ? jnrlistH : 0;
 746             j_coord_offsetA  = DIM*jnrA;
 747             j_coord_offsetB  = DIM*jnrB;
 748             j_coord_offsetC  = DIM*jnrC;
 749             j_coord_offsetD  = DIM*jnrD;
 750             j_coord_offsetE  = DIM*jnrE;
 751             j_coord_offsetF  = DIM*jnrF;
 752             j_coord_offsetG  = DIM*jnrG;
 753             j_coord_offsetH  = DIM*jnrH;
 754
 755             /* load j atom coordinates */
 756             gmx_mm256_load_3rvec_8ptr_swizzle_ps(x+j_coord_offsetA,x+j_coord_offsetB,
 757                                                  x+j_coord_offsetC,x+j_coord_offsetD,
 758                                                  x+j_coord_offsetE,x+j_coord_offsetF,
 759                                                  x+j_coord_offsetG,x+j_coord_offsetH,
 760                                               &jx0,&jy0,&jz0,&jx1,&jy1,&jz1,&jx2,&jy2,&jz2);
 761
 762             /* Calculate displacement vector */
 763             dx00             = _mm256_sub_ps(ix0,jx0);
 764             dy00             = _mm256_sub_ps(iy0,jy0);
 765             dz00             = _mm256_sub_ps(iz0,jz0);
 766             dx01             = _mm256_sub_ps(ix0,jx1);
 767             dy01             = _mm256_sub_ps(iy0,jy1);
 768             dz01             = _mm256_sub_ps(iz0,jz1);
 769             dx02             = _mm256_sub_ps(ix0,jx2);
 770             dy02             = _mm256_sub_ps(iy0,jy2);
 771             dz02             = _mm256_sub_ps(iz0,jz2);
 772             dx10             = _mm256_sub_ps(ix1,jx0);
 773             dy10             = _mm256_sub_ps(iy1,jy0);
 774             dz10             = _mm256_sub_ps(iz1,jz0);
 775             dx11             = _mm256_sub_ps(ix1,jx1);
 776             dy11             = _mm256_sub_ps(iy1,jy1);
 777             dz11             = _mm256_sub_ps(iz1,jz1);
 778             dx12             = _mm256_sub_ps(ix1,jx2);
 779             dy12             = _mm256_sub_ps(iy1,jy2);
 780             dz12             = _mm256_sub_ps(iz1,jz2);
 781             dx20             = _mm256_sub_ps(ix2,jx0);
 782             dy20             = _mm256_sub_ps(iy2,jy0);
 783             dz20             = _mm256_sub_ps(iz2,jz0);
 784             dx21             = _mm256_sub_ps(ix2,jx1);
 785             dy21             = _mm256_sub_ps(iy2,jy1);
 786             dz21             = _mm256_sub_ps(iz2,jz1);
 787             dx22             = _mm256_sub_ps(ix2,jx2);
 788             dy22             = _mm256_sub_ps(iy2,jy2);
 789             dz22             = _mm256_sub_ps(iz2,jz2);
 790
 791             /* Calculate squared distance and things based on it */
 792             rsq00            = gmx_mm256_calc_rsq_ps(dx00,dy00,dz00);
 793             rsq01            = gmx_mm256_calc_rsq_ps(dx01,dy01,dz01);
 794             rsq02            = gmx_mm256_calc_rsq_ps(dx02,dy02,dz02);
 795             rsq10            = gmx_mm256_calc_rsq_ps(dx10,dy10,dz10);
 796             rsq11            = gmx_mm256_calc_rsq_ps(dx11,dy11,dz11);
 797             rsq12            = gmx_mm256_calc_rsq_ps(dx12,dy12,dz12);
 798             rsq20            = gmx_mm256_calc_rsq_ps(dx20,dy20,dz20);
 799             rsq21            = gmx_mm256_calc_rsq_ps(dx21,dy21,dz21);
 800             rsq22            = gmx_mm256_calc_rsq_ps(dx22,dy22,dz22);
 801
 802             rinv00           = gmx_mm256_invsqrt_ps(rsq00);
 803             rinv01           = gmx_mm256_invsqrt_ps(rsq01);
 804             rinv02           = gmx_mm256_invsqrt_ps(rsq02);
 805             rinv10           = gmx_mm256_invsqrt_ps(rsq10);
 806             rinv11           = gmx_mm256_invsqrt_ps(rsq11);
 807             rinv12           = gmx_mm256_invsqrt_ps(rsq12);
 808             rinv20           = gmx_mm256_invsqrt_ps(rsq20);
 809             rinv21           = gmx_mm256_invsqrt_ps(rsq21);
 810             rinv22           = gmx_mm256_invsqrt_ps(rsq22);
 811
 812             rinvsq00         = _mm256_mul_ps(rinv00,rinv00);
 813             rinvsq01         = _mm256_mul_ps(rinv01,rinv01);
 814             rinvsq02         = _mm256_mul_ps(rinv02,rinv02);
 815             rinvsq10         = _mm256_mul_ps(rinv10,rinv10);
 816             rinvsq11         = _mm256_mul_ps(rinv11,rinv11);
 817             rinvsq12         = _mm256_mul_ps(rinv12,rinv12);
 818             rinvsq20         = _mm256_mul_ps(rinv20,rinv20);
 819             rinvsq21         = _mm256_mul_ps(rinv21,rinv21);
 820             rinvsq22         = _mm256_mul_ps(rinv22,rinv22);
 821
 822             fjx0             = _mm256_setzero_ps();
 823             fjy0             = _mm256_setzero_ps();
 824             fjz0             = _mm256_setzero_ps();
 825             fjx1             = _mm256_setzero_ps();
 826             fjy1             = _mm256_setzero_ps();
 827             fjz1             = _mm256_setzero_ps();
 828             fjx2             = _mm256_setzero_ps();
 829             fjy2             = _mm256_setzero_ps();
 830             fjz2             = _mm256_setzero_ps();
 831
 832             /**************************
 833              * CALCULATE INTERACTIONS *
 834              **************************/
 835
 836             r00              = _mm256_mul_ps(rsq00,rinv00);
 837             r00              = _mm256_andnot_ps(dummy_mask,r00);
 838
 839             /* EWALD ELECTROSTATICS */
 840
 841             /* Analytical PME correction */
 842             zeta2            = _mm256_mul_ps(beta2,rsq00);
 843             rinv3            = _mm256_mul_ps(rinvsq00,rinv00);
 844             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
 845             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
 846             felec            = _mm256_mul_ps(qq00,felec);
 847             pmecorrV         = gmx_mm256_pmecorrV_ps(zeta2);
 848             pmecorrV         = _mm256_mul_ps(pmecorrV,beta);
 849             velec            = _mm256_sub_ps(rinv00,pmecorrV);
 850             velec            = _mm256_mul_ps(qq00,velec);
 851
 852             /* Analytical LJ-PME */
 853             rinvsix          = _mm256_mul_ps(_mm256_mul_ps(rinvsq00,rinvsq00),rinvsq00);
 854             ewcljrsq         = _mm256_mul_ps(ewclj2,rsq00);
 855             ewclj6           = _mm256_mul_ps(ewclj2,_mm256_mul_ps(ewclj2,ewclj2));
 856             exponent         = gmx_simd_exp_r(ewcljrsq);
 857             /* poly = exp(-(beta*r)^2) * (1 + (beta*r)^2 + (beta*r)^4 /2) */
 858             poly             = _mm256_mul_ps(exponent,_mm256_add_ps(_mm256_sub_ps(one,ewcljrsq),_mm256_mul_ps(_mm256_mul_ps(ewcljrsq,ewcljrsq),one_half)));
 859             /* vvdw6 = [C6 - C6grid * (1-poly)]/r6 */
 860             vvdw6            = _mm256_mul_ps(_mm256_sub_ps(c6_00,_mm256_mul_ps(c6grid_00,_mm256_sub_ps(one,poly))),rinvsix);
 861             vvdw12           = _mm256_mul_ps(c12_00,_mm256_mul_ps(rinvsix,rinvsix));
 862             vvdw             = _mm256_sub_ps(_mm256_mul_ps(vvdw12,one_twelfth),_mm256_mul_ps(vvdw6,one_sixth));
 863             /* fvdw = vvdw12/r - (vvdw6/r + (C6grid * exponent * beta^6)/r) */
 864             fvdw             = _mm256_mul_ps(_mm256_sub_ps(vvdw12,_mm256_sub_ps(vvdw6,_mm256_mul_ps(_mm256_mul_ps(c6grid_00,one_sixth),_mm256_mul_ps(exponent,ewclj6)))),rinvsq00);
 865
 866             /* Update potential sum for this i atom from the interaction with this j atom. */
 867             velec            = _mm256_andnot_ps(dummy_mask,velec);
 868             velecsum         = _mm256_add_ps(velecsum,velec);
 869             vvdw             = _mm256_andnot_ps(dummy_mask,vvdw);
 870             vvdwsum          = _mm256_add_ps(vvdwsum,vvdw);
 871
 872             fscal            = _mm256_add_ps(felec,fvdw);
 873
 874             fscal            = _mm256_andnot_ps(dummy_mask,fscal);
 875
 876             /* Calculate temporary vectorial force */
 877             tx               = _mm256_mul_ps(fscal,dx00);
 878             ty               = _mm256_mul_ps(fscal,dy00);
 879             tz               = _mm256_mul_ps(fscal,dz00);
 880
 881             /* Update vectorial force */
 882             fix0             = _mm256_add_ps(fix0,tx);
 883             fiy0             = _mm256_add_ps(fiy0,ty);
 884             fiz0             = _mm256_add_ps(fiz0,tz);
 885
 886             fjx0             = _mm256_add_ps(fjx0,tx);
 887             fjy0             = _mm256_add_ps(fjy0,ty);
 888             fjz0             = _mm256_add_ps(fjz0,tz);
 889
 890             /**************************
 891              * CALCULATE INTERACTIONS *
 892              **************************/
 893
 894             r01              = _mm256_mul_ps(rsq01,rinv01);
 895             r01              = _mm256_andnot_ps(dummy_mask,r01);
 896
 897             /* EWALD ELECTROSTATICS */
 898
 899             /* Analytical PME correction */
 900             zeta2            = _mm256_mul_ps(beta2,rsq01);
 901             rinv3            = _mm256_mul_ps(rinvsq01,rinv01);
 902             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
 903             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
 904             felec            = _mm256_mul_ps(qq01,felec);
 905             pmecorrV         = gmx_mm256_pmecorrV_ps(zeta2);
 906             pmecorrV         = _mm256_mul_ps(pmecorrV,beta);
 907             velec            = _mm256_sub_ps(rinv01,pmecorrV);
 908             velec            = _mm256_mul_ps(qq01,velec);
 909
 910             /* Update potential sum for this i atom from the interaction with this j atom. */
 911             velec            = _mm256_andnot_ps(dummy_mask,velec);
 912             velecsum         = _mm256_add_ps(velecsum,velec);
 913
 914             fscal            = felec;
 915
 916             fscal            = _mm256_andnot_ps(dummy_mask,fscal);
 917
 918             /* Calculate temporary vectorial force */
 919             tx               = _mm256_mul_ps(fscal,dx01);
 920             ty               = _mm256_mul_ps(fscal,dy01);
 921             tz               = _mm256_mul_ps(fscal,dz01);
 922
 923             /* Update vectorial force */
 924             fix0             = _mm256_add_ps(fix0,tx);
 925             fiy0             = _mm256_add_ps(fiy0,ty);
 926             fiz0             = _mm256_add_ps(fiz0,tz);
 927
 928             fjx1             = _mm256_add_ps(fjx1,tx);
 929             fjy1             = _mm256_add_ps(fjy1,ty);
 930             fjz1             = _mm256_add_ps(fjz1,tz);
 931
 932             /**************************
 933              * CALCULATE INTERACTIONS *
 934              **************************/
 935
 936             r02              = _mm256_mul_ps(rsq02,rinv02);
 937             r02              = _mm256_andnot_ps(dummy_mask,r02);
 938
 939             /* EWALD ELECTROSTATICS */
 940
 941             /* Analytical PME correction */
 942             zeta2            = _mm256_mul_ps(beta2,rsq02);
 943             rinv3            = _mm256_mul_ps(rinvsq02,rinv02);
 944             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
 945             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
 946             felec            = _mm256_mul_ps(qq02,felec);
 947             pmecorrV         = gmx_mm256_pmecorrV_ps(zeta2);
 948             pmecorrV         = _mm256_mul_ps(pmecorrV,beta);
 949             velec            = _mm256_sub_ps(rinv02,pmecorrV);
 950             velec            = _mm256_mul_ps(qq02,velec);
 951
 952             /* Update potential sum for this i atom from the interaction with this j atom. */
 953             velec            = _mm256_andnot_ps(dummy_mask,velec);
 954             velecsum         = _mm256_add_ps(velecsum,velec);
 955
 956             fscal            = felec;
 957
 958             fscal            = _mm256_andnot_ps(dummy_mask,fscal);
 959
 960             /* Calculate temporary vectorial force */
 961             tx               = _mm256_mul_ps(fscal,dx02);
 962             ty               = _mm256_mul_ps(fscal,dy02);
 963             tz               = _mm256_mul_ps(fscal,dz02);
 964
 965             /* Update vectorial force */
 966             fix0             = _mm256_add_ps(fix0,tx);
 967             fiy0             = _mm256_add_ps(fiy0,ty);
 968             fiz0             = _mm256_add_ps(fiz0,tz);
 969
 970             fjx2             = _mm256_add_ps(fjx2,tx);
 971             fjy2             = _mm256_add_ps(fjy2,ty);
 972             fjz2             = _mm256_add_ps(fjz2,tz);
 973
 974             /**************************
 975              * CALCULATE INTERACTIONS *
 976              **************************/
 977
 978             r10              = _mm256_mul_ps(rsq10,rinv10);
 979             r10              = _mm256_andnot_ps(dummy_mask,r10);
 980
 981             /* EWALD ELECTROSTATICS */
 982
 983             /* Analytical PME correction */
 984             zeta2            = _mm256_mul_ps(beta2,rsq10);
 985             rinv3            = _mm256_mul_ps(rinvsq10,rinv10);
 986             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
 987             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
 988             felec            = _mm256_mul_ps(qq10,felec);
 989             pmecorrV         = gmx_mm256_pmecorrV_ps(zeta2);
 990             pmecorrV         = _mm256_mul_ps(pmecorrV,beta);
 991             velec            = _mm256_sub_ps(rinv10,pmecorrV);
 992             velec            = _mm256_mul_ps(qq10,velec);
 993
 994             /* Update potential sum for this i atom from the interaction with this j atom. */
 995             velec            = _mm256_andnot_ps(dummy_mask,velec);
 996             velecsum         = _mm256_add_ps(velecsum,velec);
 997
 998             fscal            = felec;
 999
1000             fscal            = _mm256_andnot_ps(dummy_mask,fscal);
1001
1002             /* Calculate temporary vectorial force */
1003             tx               = _mm256_mul_ps(fscal,dx10);
1004             ty               = _mm256_mul_ps(fscal,dy10);
1005             tz               = _mm256_mul_ps(fscal,dz10);
1006
1007             /* Update vectorial force */
1008             fix1             = _mm256_add_ps(fix1,tx);
1009             fiy1             = _mm256_add_ps(fiy1,ty);
1010             fiz1             = _mm256_add_ps(fiz1,tz);
1011
1012             fjx0             = _mm256_add_ps(fjx0,tx);
1013             fjy0             = _mm256_add_ps(fjy0,ty);
1014             fjz0             = _mm256_add_ps(fjz0,tz);
1015
1016             /**************************
1017              * CALCULATE INTERACTIONS *
1018              **************************/
1019
1020             r11              = _mm256_mul_ps(rsq11,rinv11);
1021             r11              = _mm256_andnot_ps(dummy_mask,r11);
1022
1023             /* EWALD ELECTROSTATICS */
1024
1025             /* Analytical PME correction */
1026             zeta2            = _mm256_mul_ps(beta2,rsq11);
1027             rinv3            = _mm256_mul_ps(rinvsq11,rinv11);
1028             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
1029             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
1030             felec            = _mm256_mul_ps(qq11,felec);
1031             pmecorrV         = gmx_mm256_pmecorrV_ps(zeta2);
1032             pmecorrV         = _mm256_mul_ps(pmecorrV,beta);
1033             velec            = _mm256_sub_ps(rinv11,pmecorrV);
1034             velec            = _mm256_mul_ps(qq11,velec);
1035
1036             /* Update potential sum for this i atom from the interaction with this j atom. */
1037             velec            = _mm256_andnot_ps(dummy_mask,velec);
1038             velecsum         = _mm256_add_ps(velecsum,velec);
1039
1040             fscal            = felec;
1041
1042             fscal            = _mm256_andnot_ps(dummy_mask,fscal);
1043
1044             /* Calculate temporary vectorial force */
1045             tx               = _mm256_mul_ps(fscal,dx11);
1046             ty               = _mm256_mul_ps(fscal,dy11);
1047             tz               = _mm256_mul_ps(fscal,dz11);
1048
1049             /* Update vectorial force */
1050             fix1             = _mm256_add_ps(fix1,tx);
1051             fiy1             = _mm256_add_ps(fiy1,ty);
1052             fiz1             = _mm256_add_ps(fiz1,tz);
1053
1054             fjx1             = _mm256_add_ps(fjx1,tx);
1055             fjy1             = _mm256_add_ps(fjy1,ty);
1056             fjz1             = _mm256_add_ps(fjz1,tz);
1057
1058             /**************************
1059              * CALCULATE INTERACTIONS *
1060              **************************/
1061
1062             r12              = _mm256_mul_ps(rsq12,rinv12);
1063             r12              = _mm256_andnot_ps(dummy_mask,r12);
1064
1065             /* EWALD ELECTROSTATICS */
1066
1067             /* Analytical PME correction */
1068             zeta2            = _mm256_mul_ps(beta2,rsq12);
1069             rinv3            = _mm256_mul_ps(rinvsq12,rinv12);
1070             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
1071             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
1072             felec            = _mm256_mul_ps(qq12,felec);
1073             pmecorrV         = gmx_mm256_pmecorrV_ps(zeta2);
1074             pmecorrV         = _mm256_mul_ps(pmecorrV,beta);
1075             velec            = _mm256_sub_ps(rinv12,pmecorrV);
1076             velec            = _mm256_mul_ps(qq12,velec);
1077
1078             /* Update potential sum for this i atom from the interaction with this j atom. */
1079             velec            = _mm256_andnot_ps(dummy_mask,velec);
1080             velecsum         = _mm256_add_ps(velecsum,velec);
1081
1082             fscal            = felec;
1083
1084             fscal            = _mm256_andnot_ps(dummy_mask,fscal);
1085
1086             /* Calculate temporary vectorial force */
1087             tx               = _mm256_mul_ps(fscal,dx12);
1088             ty               = _mm256_mul_ps(fscal,dy12);
1089             tz               = _mm256_mul_ps(fscal,dz12);
1090
1091             /* Update vectorial force */
1092             fix1             = _mm256_add_ps(fix1,tx);
1093             fiy1             = _mm256_add_ps(fiy1,ty);
1094             fiz1             = _mm256_add_ps(fiz1,tz);
1095
1096             fjx2             = _mm256_add_ps(fjx2,tx);
1097             fjy2             = _mm256_add_ps(fjy2,ty);
1098             fjz2             = _mm256_add_ps(fjz2,tz);
1099
1100             /**************************
1101              * CALCULATE INTERACTIONS *
1102              **************************/
1103
1104             r20              = _mm256_mul_ps(rsq20,rinv20);
1105             r20              = _mm256_andnot_ps(dummy_mask,r20);
1106
1107             /* EWALD ELECTROSTATICS */
1108
1109             /* Analytical PME correction */
1110             zeta2            = _mm256_mul_ps(beta2,rsq20);
1111             rinv3            = _mm256_mul_ps(rinvsq20,rinv20);
1112             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
1113             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
1114             felec            = _mm256_mul_ps(qq20,felec);
1115             pmecorrV         = gmx_mm256_pmecorrV_ps(zeta2);
1116             pmecorrV         = _mm256_mul_ps(pmecorrV,beta);
1117             velec            = _mm256_sub_ps(rinv20,pmecorrV);
1118             velec            = _mm256_mul_ps(qq20,velec);
1119
1120             /* Update potential sum for this i atom from the interaction with this j atom. */
1121             velec            = _mm256_andnot_ps(dummy_mask,velec);
1122             velecsum         = _mm256_add_ps(velecsum,velec);
1123
1124             fscal            = felec;
1125
1126             fscal            = _mm256_andnot_ps(dummy_mask,fscal);
1127
1128             /* Calculate temporary vectorial force */
1129             tx               = _mm256_mul_ps(fscal,dx20);
1130             ty               = _mm256_mul_ps(fscal,dy20);
1131             tz               = _mm256_mul_ps(fscal,dz20);
1132
1133             /* Update vectorial force */
1134             fix2             = _mm256_add_ps(fix2,tx);
1135             fiy2             = _mm256_add_ps(fiy2,ty);
1136             fiz2             = _mm256_add_ps(fiz2,tz);
1137
1138             fjx0             = _mm256_add_ps(fjx0,tx);
1139             fjy0             = _mm256_add_ps(fjy0,ty);
1140             fjz0             = _mm256_add_ps(fjz0,tz);
1141
1142             /**************************
1143              * CALCULATE INTERACTIONS *
1144              **************************/
1145
1146             r21              = _mm256_mul_ps(rsq21,rinv21);
1147             r21              = _mm256_andnot_ps(dummy_mask,r21);
1148
1149             /* EWALD ELECTROSTATICS */
1150
1151             /* Analytical PME correction */
1152             zeta2            = _mm256_mul_ps(beta2,rsq21);
1153             rinv3            = _mm256_mul_ps(rinvsq21,rinv21);
1154             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
1155             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
1156             felec            = _mm256_mul_ps(qq21,felec);
1157             pmecorrV         = gmx_mm256_pmecorrV_ps(zeta2);
1158             pmecorrV         = _mm256_mul_ps(pmecorrV,beta);
1159             velec            = _mm256_sub_ps(rinv21,pmecorrV);
1160             velec            = _mm256_mul_ps(qq21,velec);
1161
1162             /* Update potential sum for this i atom from the interaction with this j atom. */
1163             velec            = _mm256_andnot_ps(dummy_mask,velec);
1164             velecsum         = _mm256_add_ps(velecsum,velec);
1165
1166             fscal            = felec;
1167
1168             fscal            = _mm256_andnot_ps(dummy_mask,fscal);
1169
1170             /* Calculate temporary vectorial force */
1171             tx               = _mm256_mul_ps(fscal,dx21);
1172             ty               = _mm256_mul_ps(fscal,dy21);
1173             tz               = _mm256_mul_ps(fscal,dz21);
1174
1175             /* Update vectorial force */
1176             fix2             = _mm256_add_ps(fix2,tx);
1177             fiy2             = _mm256_add_ps(fiy2,ty);
1178             fiz2             = _mm256_add_ps(fiz2,tz);
1179
1180             fjx1             = _mm256_add_ps(fjx1,tx);
1181             fjy1             = _mm256_add_ps(fjy1,ty);
1182             fjz1             = _mm256_add_ps(fjz1,tz);
1183
1184             /**************************
1185              * CALCULATE INTERACTIONS *
1186              **************************/
1187
1188             r22              = _mm256_mul_ps(rsq22,rinv22);
1189             r22              = _mm256_andnot_ps(dummy_mask,r22);
1190
1191             /* EWALD ELECTROSTATICS */
1192
1193             /* Analytical PME correction */
1194             zeta2            = _mm256_mul_ps(beta2,rsq22);
1195             rinv3            = _mm256_mul_ps(rinvsq22,rinv22);
1196             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
1197             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
1198             felec            = _mm256_mul_ps(qq22,felec);
1199             pmecorrV         = gmx_mm256_pmecorrV_ps(zeta2);
1200             pmecorrV         = _mm256_mul_ps(pmecorrV,beta);
1201             velec            = _mm256_sub_ps(rinv22,pmecorrV);
1202             velec            = _mm256_mul_ps(qq22,velec);
1203
1204             /* Update potential sum for this i atom from the interaction with this j atom. */
1205             velec            = _mm256_andnot_ps(dummy_mask,velec);
1206             velecsum         = _mm256_add_ps(velecsum,velec);
1207
1208             fscal            = felec;
1209
1210             fscal            = _mm256_andnot_ps(dummy_mask,fscal);
1211
1212             /* Calculate temporary vectorial force */
1213             tx               = _mm256_mul_ps(fscal,dx22);
1214             ty               = _mm256_mul_ps(fscal,dy22);
1215             tz               = _mm256_mul_ps(fscal,dz22);
1216
1217             /* Update vectorial force */
1218             fix2             = _mm256_add_ps(fix2,tx);
1219             fiy2             = _mm256_add_ps(fiy2,ty);
1220             fiz2             = _mm256_add_ps(fiz2,tz);
1221
1222             fjx2             = _mm256_add_ps(fjx2,tx);
1223             fjy2             = _mm256_add_ps(fjy2,ty);
1224             fjz2             = _mm256_add_ps(fjz2,tz);
1225
1226             fjptrA             = (jnrlistA>=0) ? f+j_coord_offsetA : scratch;
1227             fjptrB             = (jnrlistB>=0) ? f+j_coord_offsetB : scratch;
1228             fjptrC             = (jnrlistC>=0) ? f+j_coord_offsetC : scratch;
1229             fjptrD             = (jnrlistD>=0) ? f+j_coord_offsetD : scratch;
1230             fjptrE             = (jnrlistE>=0) ? f+j_coord_offsetE : scratch;
1231             fjptrF             = (jnrlistF>=0) ? f+j_coord_offsetF : scratch;
1232             fjptrG             = (jnrlistG>=0) ? f+j_coord_offsetG : scratch;
1233             fjptrH             = (jnrlistH>=0) ? f+j_coord_offsetH : scratch;
1234
1235             gmx_mm256_decrement_3rvec_8ptr_swizzle_ps(fjptrA,fjptrB,fjptrC,fjptrD,fjptrE,fjptrF,fjptrG,fjptrH,
1236                                                       fjx0,fjy0,fjz0,fjx1,fjy1,fjz1,fjx2,fjy2,fjz2);
1237
1238             /* Inner loop uses 793 flops */
1239         }
1240
1241         /* End of innermost loop */
1242
1243         gmx_mm256_update_iforce_3atom_swizzle_ps(fix0,fiy0,fiz0,fix1,fiy1,fiz1,fix2,fiy2,fiz2,
1244                                                  f+i_coord_offset,fshift+i_shift_offset);
1245
1246         ggid                        = gid[iidx];
1247         /* Update potential energies */
1248         gmx_mm256_update_1pot_ps(velecsum,kernel_data->energygrp_elec+ggid);
1249         gmx_mm256_update_1pot_ps(vvdwsum,kernel_data->energygrp_vdw+ggid);
1250
1251         /* Increment number of inner iterations */
1252         inneriter                  += j_index_end - j_index_start;
1253
1254         /* Outer loop uses 20 flops */
1255     }
1256
1257     /* Increment number of outer iterations */
1258     outeriter        += nri;
1259
1260     /* Update outer/inner flops */
1261
1262     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W3W3_VF,outeriter*20 + inneriter*793);
1263 }
1264 /*
1265  * Gromacs nonbonded kernel:   nb_kernel_ElecEw_VdwLJEw_GeomW3W3_F_avx_256_single
1266  * Electrostatics interaction: Ewald
1267  * VdW interaction:            LJEwald
1268  * Geometry:                   Water3-Water3
1269  * Calculate force/pot:        Force
1270  */
1271 void
1272 nb_kernel_ElecEw_VdwLJEw_GeomW3W3_F_avx_256_single
1273                     (t_nblist                    * gmx_restrict       nlist,
1274                      rvec                        * gmx_restrict          xx,
1275                      rvec                        * gmx_restrict          ff,
1276                      t_forcerec                  * gmx_restrict          fr,
1277                      t_mdatoms                   * gmx_restrict     mdatoms,
1278                      nb_kernel_data_t gmx_unused * gmx_restrict kernel_data,
1279                      t_nrnb                      * gmx_restrict        nrnb)
1280 {
1281     /* Suffixes 0,1,2,3 refer to particle indices for waters in the inner or outer loop, or
1282      * just 0 for non-waters.
1283      * Suffixes A,B,C,D,E,F,G,H refer to j loop unrolling done with AVX, e.g. for the eight different
1284      * jnr indices corresponding to data put in the four positions in the SIMD register.
1285      */
1286     int              i_shift_offset,i_coord_offset,outeriter,inneriter;
1287     int              j_index_start,j_index_end,jidx,nri,inr,ggid,iidx;
1288     int              jnrA,jnrB,jnrC,jnrD;
1289     int              jnrE,jnrF,jnrG,jnrH;
1290     int              jnrlistA,jnrlistB,jnrlistC,jnrlistD;
1291     int              jnrlistE,jnrlistF,jnrlistG,jnrlistH;
1292     int              j_coord_offsetA,j_coord_offsetB,j_coord_offsetC,j_coord_offsetD;
1293     int              j_coord_offsetE,j_coord_offsetF,j_coord_offsetG,j_coord_offsetH;
1294     int              *iinr,*jindex,*jjnr,*shiftidx,*gid;
1295     real             rcutoff_scalar;
1296     real             *shiftvec,*fshift,*x,*f;
1297     real             *fjptrA,*fjptrB,*fjptrC,*fjptrD,*fjptrE,*fjptrF,*fjptrG,*fjptrH;
1298     real             scratch[4*DIM];
1299     __m256           tx,ty,tz,fscal,rcutoff,rcutoff2,jidxall;
1300     real *           vdwioffsetptr0;
1301     real *           vdwgridioffsetptr0;
1302     __m256           ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
1303     real *           vdwioffsetptr1;
1304     real *           vdwgridioffsetptr1;
1305     __m256           ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
1306     real *           vdwioffsetptr2;
1307     real *           vdwgridioffsetptr2;
1308     __m256           ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
1309     int              vdwjidx0A,vdwjidx0B,vdwjidx0C,vdwjidx0D,vdwjidx0E,vdwjidx0F,vdwjidx0G,vdwjidx0H;
1310     __m256           jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
1311     int              vdwjidx1A,vdwjidx1B,vdwjidx1C,vdwjidx1D,vdwjidx1E,vdwjidx1F,vdwjidx1G,vdwjidx1H;
1312     __m256           jx1,jy1,jz1,fjx1,fjy1,fjz1,jq1,isaj1;
1313     int              vdwjidx2A,vdwjidx2B,vdwjidx2C,vdwjidx2D,vdwjidx2E,vdwjidx2F,vdwjidx2G,vdwjidx2H;
1314     __m256           jx2,jy2,jz2,fjx2,fjy2,fjz2,jq2,isaj2;
1315     __m256           dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00;
1316     __m256           dx01,dy01,dz01,rsq01,rinv01,rinvsq01,r01,qq01,c6_01,c12_01;
1317     __m256           dx02,dy02,dz02,rsq02,rinv02,rinvsq02,r02,qq02,c6_02,c12_02;
1318     __m256           dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10;
1319     __m256           dx11,dy11,dz11,rsq11,rinv11,rinvsq11,r11,qq11,c6_11,c12_11;
1320     __m256           dx12,dy12,dz12,rsq12,rinv12,rinvsq12,r12,qq12,c6_12,c12_12;
1321     __m256           dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20;
1322     __m256           dx21,dy21,dz21,rsq21,rinv21,rinvsq21,r21,qq21,c6_21,c12_21;
1323     __m256           dx22,dy22,dz22,rsq22,rinv22,rinvsq22,r22,qq22,c6_22,c12_22;
1324     __m256           velec,felec,velecsum,facel,crf,krf,krf2;
1325     real             *charge;
1326     int              nvdwtype;
1327     __m256           rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,sh_vdw_invrcut6;
1328     int              *vdwtype;
1329     real             *vdwparam;
1330     __m256           one_sixth   = _mm256_set1_ps(1.0/6.0);
1331     __m256           one_twelfth = _mm256_set1_ps(1.0/12.0);
1332     __m256           c6grid_00;
1333     __m256           c6grid_01;
1334     __m256           c6grid_02;
1335     __m256           c6grid_10;
1336     __m256           c6grid_11;
1337     __m256           c6grid_12;
1338     __m256           c6grid_20;
1339     __m256           c6grid_21;
1340     __m256           c6grid_22;
1341     real             *vdwgridparam;
1342     __m256           ewclj,ewclj2,ewclj6,ewcljrsq,poly,exponent,f6A,f6B,sh_lj_ewald;
1343     __m256           one_half  = _mm256_set1_ps(0.5);
1344     __m256           minus_one = _mm256_set1_ps(-1.0);
1345     __m256i          ewitab;
1346     __m128i          ewitab_lo,ewitab_hi;
1347     __m256           ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace,ewtabF,ewtabFn,ewtabD,ewtabV;
1348     __m256           beta,beta2,beta3,zeta2,pmecorrF,pmecorrV,rinv3;
1349     real             *ewtab;
1350     __m256           dummy_mask,cutoff_mask;
1351     __m256           signbit = _mm256_castsi256_ps( _mm256_set1_epi32(0x80000000) );
1352     __m256           one     = _mm256_set1_ps(1.0);
1353     __m256           two     = _mm256_set1_ps(2.0);
1354     x                = xx[0];
1355     f                = ff[0];
1356
1357     nri              = nlist->nri;
1358     iinr             = nlist->iinr;
1359     jindex           = nlist->jindex;
1360     jjnr             = nlist->jjnr;
1361     shiftidx         = nlist->shift;
1362     gid              = nlist->gid;
1363     shiftvec         = fr->shift_vec[0];
1364     fshift           = fr->fshift[0];
1365     facel            = _mm256_set1_ps(fr->epsfac);
1366     charge           = mdatoms->chargeA;
1367     nvdwtype         = fr->ntype;
1368     vdwparam         = fr->nbfp;
1369     vdwtype          = mdatoms->typeA;
1370     vdwgridparam     = fr->ljpme_c6grid;
1371     sh_lj_ewald      = _mm256_set1_ps(fr->ic->sh_lj_ewald);
1372     ewclj            = _mm256_set1_ps(fr->ewaldcoeff_lj);
1373     ewclj2           = _mm256_mul_ps(minus_one,_mm256_mul_ps(ewclj,ewclj));
1374
1375     sh_ewald         = _mm256_set1_ps(fr->ic->sh_ewald);
1376     beta             = _mm256_set1_ps(fr->ic->ewaldcoeff_q);
1377     beta2            = _mm256_mul_ps(beta,beta);
1378     beta3            = _mm256_mul_ps(beta,beta2);
1379
1380     ewtab            = fr->ic->tabq_coul_F;
1381     ewtabscale       = _mm256_set1_ps(fr->ic->tabq_scale);
1382     ewtabhalfspace   = _mm256_set1_ps(0.5/fr->ic->tabq_scale);
1383
1384     /* Setup water-specific parameters */
1385     inr              = nlist->iinr[0];
1386     iq0              = _mm256_mul_ps(facel,_mm256_set1_ps(charge[inr+0]));
1387     iq1              = _mm256_mul_ps(facel,_mm256_set1_ps(charge[inr+1]));
1388     iq2              = _mm256_mul_ps(facel,_mm256_set1_ps(charge[inr+2]));
1389     vdwioffsetptr0   = vdwparam+2*nvdwtype*vdwtype[inr+0];
1390     vdwgridioffsetptr0 = vdwgridparam+2*nvdwtype*vdwtype[inr+0];
1391
1392     jq0              = _mm256_set1_ps(charge[inr+0]);
1393     jq1              = _mm256_set1_ps(charge[inr+1]);
1394     jq2              = _mm256_set1_ps(charge[inr+2]);
1395     vdwjidx0A        = 2*vdwtype[inr+0];
1396     qq00             = _mm256_mul_ps(iq0,jq0);
1397     c6_00            = _mm256_set1_ps(vdwioffsetptr0[vdwjidx0A]);
1398     c12_00           = _mm256_set1_ps(vdwioffsetptr0[vdwjidx0A+1]);
1399     c6grid_00        = _mm256_set1_ps(vdwgridioffsetptr0[vdwjidx0A]);
1400     qq01             = _mm256_mul_ps(iq0,jq1);
1401     qq02             = _mm256_mul_ps(iq0,jq2);
1402     qq10             = _mm256_mul_ps(iq1,jq0);
1403     qq11             = _mm256_mul_ps(iq1,jq1);
1404     qq12             = _mm256_mul_ps(iq1,jq2);
1405     qq20             = _mm256_mul_ps(iq2,jq0);
1406     qq21             = _mm256_mul_ps(iq2,jq1);
1407     qq22             = _mm256_mul_ps(iq2,jq2);
1408
1409     /* Avoid stupid compiler warnings */
1410     jnrA = jnrB = jnrC = jnrD = jnrE = jnrF = jnrG = jnrH = 0;
1411     j_coord_offsetA = 0;
1412     j_coord_offsetB = 0;
1413     j_coord_offsetC = 0;
1414     j_coord_offsetD = 0;
1415     j_coord_offsetE = 0;
1416     j_coord_offsetF = 0;
1417     j_coord_offsetG = 0;
1418     j_coord_offsetH = 0;
1419
1420     outeriter        = 0;
1421     inneriter        = 0;
1422
1423     for(iidx=0;iidx<4*DIM;iidx++)
1424     {
1425         scratch[iidx] = 0.0;
1426     }
1427
1428     /* Start outer loop over neighborlists */
1429     for(iidx=0; iidx<nri; iidx++)
1430     {
1431         /* Load shift vector for this list */
1432         i_shift_offset   = DIM*shiftidx[iidx];
1433
1434         /* Load limits for loop over neighbors */
1435         j_index_start    = jindex[iidx];
1436         j_index_end      = jindex[iidx+1];
1437
1438         /* Get outer coordinate index */
1439         inr              = iinr[iidx];
1440         i_coord_offset   = DIM*inr;
1441
1442         /* Load i particle coords and add shift vector */
1443         gmx_mm256_load_shift_and_3rvec_broadcast_ps(shiftvec+i_shift_offset,x+i_coord_offset,
1444                                                     &ix0,&iy0,&iz0,&ix1,&iy1,&iz1,&ix2,&iy2,&iz2);
1445
1446         fix0             = _mm256_setzero_ps();
1447         fiy0             = _mm256_setzero_ps();
1448         fiz0             = _mm256_setzero_ps();
1449         fix1             = _mm256_setzero_ps();
1450         fiy1             = _mm256_setzero_ps();
1451         fiz1             = _mm256_setzero_ps();
1452         fix2             = _mm256_setzero_ps();
1453         fiy2             = _mm256_setzero_ps();
1454         fiz2             = _mm256_setzero_ps();
1455
1456         /* Start inner kernel loop */
1457         for(jidx=j_index_start; jidx<j_index_end && jjnr[jidx+7]>=0; jidx+=8)
1458         {
1459
1460             /* Get j neighbor index, and coordinate index */
1461             jnrA             = jjnr[jidx];
1462             jnrB             = jjnr[jidx+1];
1463             jnrC             = jjnr[jidx+2];
1464             jnrD             = jjnr[jidx+3];
1465             jnrE             = jjnr[jidx+4];
1466             jnrF             = jjnr[jidx+5];
1467             jnrG             = jjnr[jidx+6];
1468             jnrH             = jjnr[jidx+7];
1469             j_coord_offsetA  = DIM*jnrA;
1470             j_coord_offsetB  = DIM*jnrB;
1471             j_coord_offsetC  = DIM*jnrC;
1472             j_coord_offsetD  = DIM*jnrD;
1473             j_coord_offsetE  = DIM*jnrE;
1474             j_coord_offsetF  = DIM*jnrF;
1475             j_coord_offsetG  = DIM*jnrG;
1476             j_coord_offsetH  = DIM*jnrH;
1477
1478             /* load j atom coordinates */
1479             gmx_mm256_load_3rvec_8ptr_swizzle_ps(x+j_coord_offsetA,x+j_coord_offsetB,
1480                                                  x+j_coord_offsetC,x+j_coord_offsetD,
1481                                                  x+j_coord_offsetE,x+j_coord_offsetF,
1482                                                  x+j_coord_offsetG,x+j_coord_offsetH,
1483                                               &jx0,&jy0,&jz0,&jx1,&jy1,&jz1,&jx2,&jy2,&jz2);
1484
1485             /* Calculate displacement vector */
1486             dx00             = _mm256_sub_ps(ix0,jx0);
1487             dy00             = _mm256_sub_ps(iy0,jy0);
1488             dz00             = _mm256_sub_ps(iz0,jz0);
1489             dx01             = _mm256_sub_ps(ix0,jx1);
1490             dy01             = _mm256_sub_ps(iy0,jy1);
1491             dz01             = _mm256_sub_ps(iz0,jz1);
1492             dx02             = _mm256_sub_ps(ix0,jx2);
1493             dy02             = _mm256_sub_ps(iy0,jy2);
1494             dz02             = _mm256_sub_ps(iz0,jz2);
1495             dx10             = _mm256_sub_ps(ix1,jx0);
1496             dy10             = _mm256_sub_ps(iy1,jy0);
1497             dz10             = _mm256_sub_ps(iz1,jz0);
1498             dx11             = _mm256_sub_ps(ix1,jx1);
1499             dy11             = _mm256_sub_ps(iy1,jy1);
1500             dz11             = _mm256_sub_ps(iz1,jz1);
1501             dx12             = _mm256_sub_ps(ix1,jx2);
1502             dy12             = _mm256_sub_ps(iy1,jy2);
1503             dz12             = _mm256_sub_ps(iz1,jz2);
1504             dx20             = _mm256_sub_ps(ix2,jx0);
1505             dy20             = _mm256_sub_ps(iy2,jy0);
1506             dz20             = _mm256_sub_ps(iz2,jz0);
1507             dx21             = _mm256_sub_ps(ix2,jx1);
1508             dy21             = _mm256_sub_ps(iy2,jy1);
1509             dz21             = _mm256_sub_ps(iz2,jz1);
1510             dx22             = _mm256_sub_ps(ix2,jx2);
1511             dy22             = _mm256_sub_ps(iy2,jy2);
1512             dz22             = _mm256_sub_ps(iz2,jz2);
1513
1514             /* Calculate squared distance and things based on it */
1515             rsq00            = gmx_mm256_calc_rsq_ps(dx00,dy00,dz00);
1516             rsq01            = gmx_mm256_calc_rsq_ps(dx01,dy01,dz01);
1517             rsq02            = gmx_mm256_calc_rsq_ps(dx02,dy02,dz02);
1518             rsq10            = gmx_mm256_calc_rsq_ps(dx10,dy10,dz10);
1519             rsq11            = gmx_mm256_calc_rsq_ps(dx11,dy11,dz11);
1520             rsq12            = gmx_mm256_calc_rsq_ps(dx12,dy12,dz12);
1521             rsq20            = gmx_mm256_calc_rsq_ps(dx20,dy20,dz20);
1522             rsq21            = gmx_mm256_calc_rsq_ps(dx21,dy21,dz21);
1523             rsq22            = gmx_mm256_calc_rsq_ps(dx22,dy22,dz22);
1524
1525             rinv00           = gmx_mm256_invsqrt_ps(rsq00);
1526             rinv01           = gmx_mm256_invsqrt_ps(rsq01);
1527             rinv02           = gmx_mm256_invsqrt_ps(rsq02);
1528             rinv10           = gmx_mm256_invsqrt_ps(rsq10);
1529             rinv11           = gmx_mm256_invsqrt_ps(rsq11);
1530             rinv12           = gmx_mm256_invsqrt_ps(rsq12);
1531             rinv20           = gmx_mm256_invsqrt_ps(rsq20);
1532             rinv21           = gmx_mm256_invsqrt_ps(rsq21);
1533             rinv22           = gmx_mm256_invsqrt_ps(rsq22);
1534
1535             rinvsq00         = _mm256_mul_ps(rinv00,rinv00);
1536             rinvsq01         = _mm256_mul_ps(rinv01,rinv01);
1537             rinvsq02         = _mm256_mul_ps(rinv02,rinv02);
1538             rinvsq10         = _mm256_mul_ps(rinv10,rinv10);
1539             rinvsq11         = _mm256_mul_ps(rinv11,rinv11);
1540             rinvsq12         = _mm256_mul_ps(rinv12,rinv12);
1541             rinvsq20         = _mm256_mul_ps(rinv20,rinv20);
1542             rinvsq21         = _mm256_mul_ps(rinv21,rinv21);
1543             rinvsq22         = _mm256_mul_ps(rinv22,rinv22);
1544
1545             fjx0             = _mm256_setzero_ps();
1546             fjy0             = _mm256_setzero_ps();
1547             fjz0             = _mm256_setzero_ps();
1548             fjx1             = _mm256_setzero_ps();
1549             fjy1             = _mm256_setzero_ps();
1550             fjz1             = _mm256_setzero_ps();
1551             fjx2             = _mm256_setzero_ps();
1552             fjy2             = _mm256_setzero_ps();
1553             fjz2             = _mm256_setzero_ps();
1554
1555             /**************************
1556              * CALCULATE INTERACTIONS *
1557              **************************/
1558
1559             r00              = _mm256_mul_ps(rsq00,rinv00);
1560
1561             /* EWALD ELECTROSTATICS */
1562
1563             /* Analytical PME correction */
1564             zeta2            = _mm256_mul_ps(beta2,rsq00);
1565             rinv3            = _mm256_mul_ps(rinvsq00,rinv00);
1566             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
1567             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
1568             felec            = _mm256_mul_ps(qq00,felec);
1569
1570             /* Analytical LJ-PME */
1571             rinvsix          = _mm256_mul_ps(_mm256_mul_ps(rinvsq00,rinvsq00),rinvsq00);
1572             ewcljrsq         = _mm256_mul_ps(ewclj2,rsq00);
1573             ewclj6           = _mm256_mul_ps(ewclj2,_mm256_mul_ps(ewclj2,ewclj2));
1574             exponent         = gmx_simd_exp_r(ewcljrsq);
1575             /* poly = exp(-(beta*r)^2) * (1 + (beta*r)^2 + (beta*r)^4 /2) */
1576             poly             = _mm256_mul_ps(exponent,_mm256_add_ps(_mm256_sub_ps(one,ewcljrsq),_mm256_mul_ps(_mm256_mul_ps(ewcljrsq,ewcljrsq),one_half)));
1577             /* f6A = 6 * C6grid * (1 - poly) */
1578             f6A              = _mm256_mul_ps(c6grid_00,_mm256_sub_ps(one,poly));
1579             /* f6B = C6grid * exponent * beta^6 */
1580             f6B              = _mm256_mul_ps(_mm256_mul_ps(c6grid_00,one_sixth),_mm256_mul_ps(exponent,ewclj6));
1581             /* fvdw = 12*C12/r13 - ((6*C6 - f6A)/r6 + f6B)/r */
1582             fvdw              = _mm256_mul_ps(_mm256_add_ps(_mm256_mul_ps(_mm256_sub_ps(_mm256_mul_ps(c12_00,rinvsix),_mm256_sub_ps(c6_00,f6A)),rinvsix),f6B),rinvsq00);
1583
1584             fscal            = _mm256_add_ps(felec,fvdw);
1585
1586             /* Calculate temporary vectorial force */
1587             tx               = _mm256_mul_ps(fscal,dx00);
1588             ty               = _mm256_mul_ps(fscal,dy00);
1589             tz               = _mm256_mul_ps(fscal,dz00);
1590
1591             /* Update vectorial force */
1592             fix0             = _mm256_add_ps(fix0,tx);
1593             fiy0             = _mm256_add_ps(fiy0,ty);
1594             fiz0             = _mm256_add_ps(fiz0,tz);
1595
1596             fjx0             = _mm256_add_ps(fjx0,tx);
1597             fjy0             = _mm256_add_ps(fjy0,ty);
1598             fjz0             = _mm256_add_ps(fjz0,tz);
1599
1600             /**************************
1601              * CALCULATE INTERACTIONS *
1602              **************************/
1603
1604             r01              = _mm256_mul_ps(rsq01,rinv01);
1605
1606             /* EWALD ELECTROSTATICS */
1607
1608             /* Analytical PME correction */
1609             zeta2            = _mm256_mul_ps(beta2,rsq01);
1610             rinv3            = _mm256_mul_ps(rinvsq01,rinv01);
1611             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
1612             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
1613             felec            = _mm256_mul_ps(qq01,felec);
1614
1615             fscal            = felec;
1616
1617             /* Calculate temporary vectorial force */
1618             tx               = _mm256_mul_ps(fscal,dx01);
1619             ty               = _mm256_mul_ps(fscal,dy01);
1620             tz               = _mm256_mul_ps(fscal,dz01);
1621
1622             /* Update vectorial force */
1623             fix0             = _mm256_add_ps(fix0,tx);
1624             fiy0             = _mm256_add_ps(fiy0,ty);
1625             fiz0             = _mm256_add_ps(fiz0,tz);
1626
1627             fjx1             = _mm256_add_ps(fjx1,tx);
1628             fjy1             = _mm256_add_ps(fjy1,ty);
1629             fjz1             = _mm256_add_ps(fjz1,tz);
1630
1631             /**************************
1632              * CALCULATE INTERACTIONS *
1633              **************************/
1634
1635             r02              = _mm256_mul_ps(rsq02,rinv02);
1636
1637             /* EWALD ELECTROSTATICS */
1638
1639             /* Analytical PME correction */
1640             zeta2            = _mm256_mul_ps(beta2,rsq02);
1641             rinv3            = _mm256_mul_ps(rinvsq02,rinv02);
1642             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
1643             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
1644             felec            = _mm256_mul_ps(qq02,felec);
1645
1646             fscal            = felec;
1647
1648             /* Calculate temporary vectorial force */
1649             tx               = _mm256_mul_ps(fscal,dx02);
1650             ty               = _mm256_mul_ps(fscal,dy02);
1651             tz               = _mm256_mul_ps(fscal,dz02);
1652
1653             /* Update vectorial force */
1654             fix0             = _mm256_add_ps(fix0,tx);
1655             fiy0             = _mm256_add_ps(fiy0,ty);
1656             fiz0             = _mm256_add_ps(fiz0,tz);
1657
1658             fjx2             = _mm256_add_ps(fjx2,tx);
1659             fjy2             = _mm256_add_ps(fjy2,ty);
1660             fjz2             = _mm256_add_ps(fjz2,tz);
1661
1662             /**************************
1663              * CALCULATE INTERACTIONS *
1664              **************************/
1665
1666             r10              = _mm256_mul_ps(rsq10,rinv10);
1667
1668             /* EWALD ELECTROSTATICS */
1669
1670             /* Analytical PME correction */
1671             zeta2            = _mm256_mul_ps(beta2,rsq10);
1672             rinv3            = _mm256_mul_ps(rinvsq10,rinv10);
1673             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
1674             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
1675             felec            = _mm256_mul_ps(qq10,felec);
1676
1677             fscal            = felec;
1678
1679             /* Calculate temporary vectorial force */
1680             tx               = _mm256_mul_ps(fscal,dx10);
1681             ty               = _mm256_mul_ps(fscal,dy10);
1682             tz               = _mm256_mul_ps(fscal,dz10);
1683
1684             /* Update vectorial force */
1685             fix1             = _mm256_add_ps(fix1,tx);
1686             fiy1             = _mm256_add_ps(fiy1,ty);
1687             fiz1             = _mm256_add_ps(fiz1,tz);
1688
1689             fjx0             = _mm256_add_ps(fjx0,tx);
1690             fjy0             = _mm256_add_ps(fjy0,ty);
1691             fjz0             = _mm256_add_ps(fjz0,tz);
1692
1693             /**************************
1694              * CALCULATE INTERACTIONS *
1695              **************************/
1696
1697             r11              = _mm256_mul_ps(rsq11,rinv11);
1698
1699             /* EWALD ELECTROSTATICS */
1700
1701             /* Analytical PME correction */
1702             zeta2            = _mm256_mul_ps(beta2,rsq11);
1703             rinv3            = _mm256_mul_ps(rinvsq11,rinv11);
1704             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
1705             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
1706             felec            = _mm256_mul_ps(qq11,felec);
1707
1708             fscal            = felec;
1709
1710             /* Calculate temporary vectorial force */
1711             tx               = _mm256_mul_ps(fscal,dx11);
1712             ty               = _mm256_mul_ps(fscal,dy11);
1713             tz               = _mm256_mul_ps(fscal,dz11);
1714
1715             /* Update vectorial force */
1716             fix1             = _mm256_add_ps(fix1,tx);
1717             fiy1             = _mm256_add_ps(fiy1,ty);
1718             fiz1             = _mm256_add_ps(fiz1,tz);
1719
1720             fjx1             = _mm256_add_ps(fjx1,tx);
1721             fjy1             = _mm256_add_ps(fjy1,ty);
1722             fjz1             = _mm256_add_ps(fjz1,tz);
1723
1724             /**************************
1725              * CALCULATE INTERACTIONS *
1726              **************************/
1727
1728             r12              = _mm256_mul_ps(rsq12,rinv12);
1729
1730             /* EWALD ELECTROSTATICS */
1731
1732             /* Analytical PME correction */
1733             zeta2            = _mm256_mul_ps(beta2,rsq12);
1734             rinv3            = _mm256_mul_ps(rinvsq12,rinv12);
1735             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
1736             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
1737             felec            = _mm256_mul_ps(qq12,felec);
1738
1739             fscal            = felec;
1740
1741             /* Calculate temporary vectorial force */
1742             tx               = _mm256_mul_ps(fscal,dx12);
1743             ty               = _mm256_mul_ps(fscal,dy12);
1744             tz               = _mm256_mul_ps(fscal,dz12);
1745
1746             /* Update vectorial force */
1747             fix1             = _mm256_add_ps(fix1,tx);
1748             fiy1             = _mm256_add_ps(fiy1,ty);
1749             fiz1             = _mm256_add_ps(fiz1,tz);
1750
1751             fjx2             = _mm256_add_ps(fjx2,tx);
1752             fjy2             = _mm256_add_ps(fjy2,ty);
1753             fjz2             = _mm256_add_ps(fjz2,tz);
1754
1755             /**************************
1756              * CALCULATE INTERACTIONS *
1757              **************************/
1758
1759             r20              = _mm256_mul_ps(rsq20,rinv20);
1760
1761             /* EWALD ELECTROSTATICS */
1762
1763             /* Analytical PME correction */
1764             zeta2            = _mm256_mul_ps(beta2,rsq20);
1765             rinv3            = _mm256_mul_ps(rinvsq20,rinv20);
1766             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
1767             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
1768             felec            = _mm256_mul_ps(qq20,felec);
1769
1770             fscal            = felec;
1771
1772             /* Calculate temporary vectorial force */
1773             tx               = _mm256_mul_ps(fscal,dx20);
1774             ty               = _mm256_mul_ps(fscal,dy20);
1775             tz               = _mm256_mul_ps(fscal,dz20);
1776
1777             /* Update vectorial force */
1778             fix2             = _mm256_add_ps(fix2,tx);
1779             fiy2             = _mm256_add_ps(fiy2,ty);
1780             fiz2             = _mm256_add_ps(fiz2,tz);
1781
1782             fjx0             = _mm256_add_ps(fjx0,tx);
1783             fjy0             = _mm256_add_ps(fjy0,ty);
1784             fjz0             = _mm256_add_ps(fjz0,tz);
1785
1786             /**************************
1787              * CALCULATE INTERACTIONS *
1788              **************************/
1789
1790             r21              = _mm256_mul_ps(rsq21,rinv21);
1791
1792             /* EWALD ELECTROSTATICS */
1793
1794             /* Analytical PME correction */
1795             zeta2            = _mm256_mul_ps(beta2,rsq21);
1796             rinv3            = _mm256_mul_ps(rinvsq21,rinv21);
1797             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
1798             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
1799             felec            = _mm256_mul_ps(qq21,felec);
1800
1801             fscal            = felec;
1802
1803             /* Calculate temporary vectorial force */
1804             tx               = _mm256_mul_ps(fscal,dx21);
1805             ty               = _mm256_mul_ps(fscal,dy21);
1806             tz               = _mm256_mul_ps(fscal,dz21);
1807
1808             /* Update vectorial force */
1809             fix2             = _mm256_add_ps(fix2,tx);
1810             fiy2             = _mm256_add_ps(fiy2,ty);
1811             fiz2             = _mm256_add_ps(fiz2,tz);
1812
1813             fjx1             = _mm256_add_ps(fjx1,tx);
1814             fjy1             = _mm256_add_ps(fjy1,ty);
1815             fjz1             = _mm256_add_ps(fjz1,tz);
1816
1817             /**************************
1818              * CALCULATE INTERACTIONS *
1819              **************************/
1820
1821             r22              = _mm256_mul_ps(rsq22,rinv22);
1822
1823             /* EWALD ELECTROSTATICS */
1824
1825             /* Analytical PME correction */
1826             zeta2            = _mm256_mul_ps(beta2,rsq22);
1827             rinv3            = _mm256_mul_ps(rinvsq22,rinv22);
1828             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
1829             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
1830             felec            = _mm256_mul_ps(qq22,felec);
1831
1832             fscal            = felec;
1833
1834             /* Calculate temporary vectorial force */
1835             tx               = _mm256_mul_ps(fscal,dx22);
1836             ty               = _mm256_mul_ps(fscal,dy22);
1837             tz               = _mm256_mul_ps(fscal,dz22);
1838
1839             /* Update vectorial force */
1840             fix2             = _mm256_add_ps(fix2,tx);
1841             fiy2             = _mm256_add_ps(fiy2,ty);
1842             fiz2             = _mm256_add_ps(fiz2,tz);
1843
1844             fjx2             = _mm256_add_ps(fjx2,tx);
1845             fjy2             = _mm256_add_ps(fjy2,ty);
1846             fjz2             = _mm256_add_ps(fjz2,tz);
1847
1848             fjptrA             = f+j_coord_offsetA;
1849             fjptrB             = f+j_coord_offsetB;
1850             fjptrC             = f+j_coord_offsetC;
1851             fjptrD             = f+j_coord_offsetD;
1852             fjptrE             = f+j_coord_offsetE;
1853             fjptrF             = f+j_coord_offsetF;
1854             fjptrG             = f+j_coord_offsetG;
1855             fjptrH             = f+j_coord_offsetH;
1856
1857             gmx_mm256_decrement_3rvec_8ptr_swizzle_ps(fjptrA,fjptrB,fjptrC,fjptrD,fjptrE,fjptrF,fjptrG,fjptrH,
1858                                                       fjx0,fjy0,fjz0,fjx1,fjy1,fjz1,fjx2,fjy2,fjz2);
1859
1860             /* Inner loop uses 527 flops */
1861         }
1862
1863         if(jidx<j_index_end)
1864         {
1865
1866             /* Get j neighbor index, and coordinate index */
1867             jnrlistA         = jjnr[jidx];
1868             jnrlistB         = jjnr[jidx+1];
1869             jnrlistC         = jjnr[jidx+2];
1870             jnrlistD         = jjnr[jidx+3];
1871             jnrlistE         = jjnr[jidx+4];
1872             jnrlistF         = jjnr[jidx+5];
1873             jnrlistG         = jjnr[jidx+6];
1874             jnrlistH         = jjnr[jidx+7];
1875             /* Sign of each element will be negative for non-real atoms.
1876              * This mask will be 0xFFFFFFFF for dummy entries and 0x0 for real ones,
1877              * so use it as val = _mm_andnot_ps(mask,val) to clear dummy entries.
1878              */
1879             dummy_mask = gmx_mm256_set_m128(gmx_mm_castsi128_ps(_mm_cmplt_epi32(_mm_loadu_si128((const __m128i *)(jjnr+jidx+4)),_mm_setzero_si128())),
1880                                             gmx_mm_castsi128_ps(_mm_cmplt_epi32(_mm_loadu_si128((const __m128i *)(jjnr+jidx)),_mm_setzero_si128())));
1881
1882             jnrA       = (jnrlistA>=0) ? jnrlistA : 0;
1883             jnrB       = (jnrlistB>=0) ? jnrlistB : 0;
1884             jnrC       = (jnrlistC>=0) ? jnrlistC : 0;
1885             jnrD       = (jnrlistD>=0) ? jnrlistD : 0;
1886             jnrE       = (jnrlistE>=0) ? jnrlistE : 0;
1887             jnrF       = (jnrlistF>=0) ? jnrlistF : 0;
1888             jnrG       = (jnrlistG>=0) ? jnrlistG : 0;
1889             jnrH       = (jnrlistH>=0) ? jnrlistH : 0;
1890             j_coord_offsetA  = DIM*jnrA;
1891             j_coord_offsetB  = DIM*jnrB;
1892             j_coord_offsetC  = DIM*jnrC;
1893             j_coord_offsetD  = DIM*jnrD;
1894             j_coord_offsetE  = DIM*jnrE;
1895             j_coord_offsetF  = DIM*jnrF;
1896             j_coord_offsetG  = DIM*jnrG;
1897             j_coord_offsetH  = DIM*jnrH;
1898
1899             /* load j atom coordinates */
1900             gmx_mm256_load_3rvec_8ptr_swizzle_ps(x+j_coord_offsetA,x+j_coord_offsetB,
1901                                                  x+j_coord_offsetC,x+j_coord_offsetD,
1902                                                  x+j_coord_offsetE,x+j_coord_offsetF,
1903                                                  x+j_coord_offsetG,x+j_coord_offsetH,
1904                                               &jx0,&jy0,&jz0,&jx1,&jy1,&jz1,&jx2,&jy2,&jz2);
1905
1906             /* Calculate displacement vector */
1907             dx00             = _mm256_sub_ps(ix0,jx0);
1908             dy00             = _mm256_sub_ps(iy0,jy0);
1909             dz00             = _mm256_sub_ps(iz0,jz0);
1910             dx01             = _mm256_sub_ps(ix0,jx1);
1911             dy01             = _mm256_sub_ps(iy0,jy1);
1912             dz01             = _mm256_sub_ps(iz0,jz1);
1913             dx02             = _mm256_sub_ps(ix0,jx2);
1914             dy02             = _mm256_sub_ps(iy0,jy2);
1915             dz02             = _mm256_sub_ps(iz0,jz2);
1916             dx10             = _mm256_sub_ps(ix1,jx0);
1917             dy10             = _mm256_sub_ps(iy1,jy0);
1918             dz10             = _mm256_sub_ps(iz1,jz0);
1919             dx11             = _mm256_sub_ps(ix1,jx1);
1920             dy11             = _mm256_sub_ps(iy1,jy1);
1921             dz11             = _mm256_sub_ps(iz1,jz1);
1922             dx12             = _mm256_sub_ps(ix1,jx2);
1923             dy12             = _mm256_sub_ps(iy1,jy2);
1924             dz12             = _mm256_sub_ps(iz1,jz2);
1925             dx20             = _mm256_sub_ps(ix2,jx0);
1926             dy20             = _mm256_sub_ps(iy2,jy0);
1927             dz20             = _mm256_sub_ps(iz2,jz0);
1928             dx21             = _mm256_sub_ps(ix2,jx1);
1929             dy21             = _mm256_sub_ps(iy2,jy1);
1930             dz21             = _mm256_sub_ps(iz2,jz1);
1931             dx22             = _mm256_sub_ps(ix2,jx2);
1932             dy22             = _mm256_sub_ps(iy2,jy2);
1933             dz22             = _mm256_sub_ps(iz2,jz2);
1934
1935             /* Calculate squared distance and things based on it */
1936             rsq00            = gmx_mm256_calc_rsq_ps(dx00,dy00,dz00);
1937             rsq01            = gmx_mm256_calc_rsq_ps(dx01,dy01,dz01);
1938             rsq02            = gmx_mm256_calc_rsq_ps(dx02,dy02,dz02);
1939             rsq10            = gmx_mm256_calc_rsq_ps(dx10,dy10,dz10);
1940             rsq11            = gmx_mm256_calc_rsq_ps(dx11,dy11,dz11);
1941             rsq12            = gmx_mm256_calc_rsq_ps(dx12,dy12,dz12);
1942             rsq20            = gmx_mm256_calc_rsq_ps(dx20,dy20,dz20);
1943             rsq21            = gmx_mm256_calc_rsq_ps(dx21,dy21,dz21);
1944             rsq22            = gmx_mm256_calc_rsq_ps(dx22,dy22,dz22);
1945
1946             rinv00           = gmx_mm256_invsqrt_ps(rsq00);
1947             rinv01           = gmx_mm256_invsqrt_ps(rsq01);
1948             rinv02           = gmx_mm256_invsqrt_ps(rsq02);
1949             rinv10           = gmx_mm256_invsqrt_ps(rsq10);
1950             rinv11           = gmx_mm256_invsqrt_ps(rsq11);
1951             rinv12           = gmx_mm256_invsqrt_ps(rsq12);
1952             rinv20           = gmx_mm256_invsqrt_ps(rsq20);
1953             rinv21           = gmx_mm256_invsqrt_ps(rsq21);
1954             rinv22           = gmx_mm256_invsqrt_ps(rsq22);
1955
1956             rinvsq00         = _mm256_mul_ps(rinv00,rinv00);
1957             rinvsq01         = _mm256_mul_ps(rinv01,rinv01);
1958             rinvsq02         = _mm256_mul_ps(rinv02,rinv02);
1959             rinvsq10         = _mm256_mul_ps(rinv10,rinv10);
1960             rinvsq11         = _mm256_mul_ps(rinv11,rinv11);
1961             rinvsq12         = _mm256_mul_ps(rinv12,rinv12);
1962             rinvsq20         = _mm256_mul_ps(rinv20,rinv20);
1963             rinvsq21         = _mm256_mul_ps(rinv21,rinv21);
1964             rinvsq22         = _mm256_mul_ps(rinv22,rinv22);
1965
1966             fjx0             = _mm256_setzero_ps();
1967             fjy0             = _mm256_setzero_ps();
1968             fjz0             = _mm256_setzero_ps();
1969             fjx1             = _mm256_setzero_ps();
1970             fjy1             = _mm256_setzero_ps();
1971             fjz1             = _mm256_setzero_ps();
1972             fjx2             = _mm256_setzero_ps();
1973             fjy2             = _mm256_setzero_ps();
1974             fjz2             = _mm256_setzero_ps();
1975
1976             /**************************
1977              * CALCULATE INTERACTIONS *
1978              **************************/
1979
1980             r00              = _mm256_mul_ps(rsq00,rinv00);
1981             r00              = _mm256_andnot_ps(dummy_mask,r00);
1982
1983             /* EWALD ELECTROSTATICS */
1984
1985             /* Analytical PME correction */
1986             zeta2            = _mm256_mul_ps(beta2,rsq00);
1987             rinv3            = _mm256_mul_ps(rinvsq00,rinv00);
1988             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
1989             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
1990             felec            = _mm256_mul_ps(qq00,felec);
1991
1992             /* Analytical LJ-PME */
1993             rinvsix          = _mm256_mul_ps(_mm256_mul_ps(rinvsq00,rinvsq00),rinvsq00);
1994             ewcljrsq         = _mm256_mul_ps(ewclj2,rsq00);
1995             ewclj6           = _mm256_mul_ps(ewclj2,_mm256_mul_ps(ewclj2,ewclj2));
1996             exponent         = gmx_simd_exp_r(ewcljrsq);
1997             /* poly = exp(-(beta*r)^2) * (1 + (beta*r)^2 + (beta*r)^4 /2) */
1998             poly             = _mm256_mul_ps(exponent,_mm256_add_ps(_mm256_sub_ps(one,ewcljrsq),_mm256_mul_ps(_mm256_mul_ps(ewcljrsq,ewcljrsq),one_half)));
1999             /* f6A = 6 * C6grid * (1 - poly) */
2000             f6A              = _mm256_mul_ps(c6grid_00,_mm256_sub_ps(one,poly));
2001             /* f6B = C6grid * exponent * beta^6 */
2002             f6B              = _mm256_mul_ps(_mm256_mul_ps(c6grid_00,one_sixth),_mm256_mul_ps(exponent,ewclj6));
2003             /* fvdw = 12*C12/r13 - ((6*C6 - f6A)/r6 + f6B)/r */
2004             fvdw              = _mm256_mul_ps(_mm256_add_ps(_mm256_mul_ps(_mm256_sub_ps(_mm256_mul_ps(c12_00,rinvsix),_mm256_sub_ps(c6_00,f6A)),rinvsix),f6B),rinvsq00);
2005
2006             fscal            = _mm256_add_ps(felec,fvdw);
2007
2008             fscal            = _mm256_andnot_ps(dummy_mask,fscal);
2009
2010             /* Calculate temporary vectorial force */
2011             tx               = _mm256_mul_ps(fscal,dx00);
2012             ty               = _mm256_mul_ps(fscal,dy00);
2013             tz               = _mm256_mul_ps(fscal,dz00);
2014
2015             /* Update vectorial force */
2016             fix0             = _mm256_add_ps(fix0,tx);
2017             fiy0             = _mm256_add_ps(fiy0,ty);
2018             fiz0             = _mm256_add_ps(fiz0,tz);
2019
2020             fjx0             = _mm256_add_ps(fjx0,tx);
2021             fjy0             = _mm256_add_ps(fjy0,ty);
2022             fjz0             = _mm256_add_ps(fjz0,tz);
2023
2024             /**************************
2025              * CALCULATE INTERACTIONS *
2026              **************************/
2027
2028             r01              = _mm256_mul_ps(rsq01,rinv01);
2029             r01              = _mm256_andnot_ps(dummy_mask,r01);
2030
2031             /* EWALD ELECTROSTATICS */
2032
2033             /* Analytical PME correction */
2034             zeta2            = _mm256_mul_ps(beta2,rsq01);
2035             rinv3            = _mm256_mul_ps(rinvsq01,rinv01);
2036             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
2037             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
2038             felec            = _mm256_mul_ps(qq01,felec);
2039
2040             fscal            = felec;
2041
2042             fscal            = _mm256_andnot_ps(dummy_mask,fscal);
2043
2044             /* Calculate temporary vectorial force */
2045             tx               = _mm256_mul_ps(fscal,dx01);
2046             ty               = _mm256_mul_ps(fscal,dy01);
2047             tz               = _mm256_mul_ps(fscal,dz01);
2048
2049             /* Update vectorial force */
2050             fix0             = _mm256_add_ps(fix0,tx);
2051             fiy0             = _mm256_add_ps(fiy0,ty);
2052             fiz0             = _mm256_add_ps(fiz0,tz);
2053
2054             fjx1             = _mm256_add_ps(fjx1,tx);
2055             fjy1             = _mm256_add_ps(fjy1,ty);
2056             fjz1             = _mm256_add_ps(fjz1,tz);
2057
2058             /**************************
2059              * CALCULATE INTERACTIONS *
2060              **************************/
2061
2062             r02              = _mm256_mul_ps(rsq02,rinv02);
2063             r02              = _mm256_andnot_ps(dummy_mask,r02);
2064
2065             /* EWALD ELECTROSTATICS */
2066
2067             /* Analytical PME correction */
2068             zeta2            = _mm256_mul_ps(beta2,rsq02);
2069             rinv3            = _mm256_mul_ps(rinvsq02,rinv02);
2070             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
2071             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
2072             felec            = _mm256_mul_ps(qq02,felec);
2073
2074             fscal            = felec;
2075
2076             fscal            = _mm256_andnot_ps(dummy_mask,fscal);
2077
2078             /* Calculate temporary vectorial force */
2079             tx               = _mm256_mul_ps(fscal,dx02);
2080             ty               = _mm256_mul_ps(fscal,dy02);
2081             tz               = _mm256_mul_ps(fscal,dz02);
2082
2083             /* Update vectorial force */
2084             fix0             = _mm256_add_ps(fix0,tx);
2085             fiy0             = _mm256_add_ps(fiy0,ty);
2086             fiz0             = _mm256_add_ps(fiz0,tz);
2087
2088             fjx2             = _mm256_add_ps(fjx2,tx);
2089             fjy2             = _mm256_add_ps(fjy2,ty);
2090             fjz2             = _mm256_add_ps(fjz2,tz);
2091
2092             /**************************
2093              * CALCULATE INTERACTIONS *
2094              **************************/
2095
2096             r10              = _mm256_mul_ps(rsq10,rinv10);
2097             r10              = _mm256_andnot_ps(dummy_mask,r10);
2098
2099             /* EWALD ELECTROSTATICS */
2100
2101             /* Analytical PME correction */
2102             zeta2            = _mm256_mul_ps(beta2,rsq10);
2103             rinv3            = _mm256_mul_ps(rinvsq10,rinv10);
2104             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
2105             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
2106             felec            = _mm256_mul_ps(qq10,felec);
2107
2108             fscal            = felec;
2109
2110             fscal            = _mm256_andnot_ps(dummy_mask,fscal);
2111
2112             /* Calculate temporary vectorial force */
2113             tx               = _mm256_mul_ps(fscal,dx10);
2114             ty               = _mm256_mul_ps(fscal,dy10);
2115             tz               = _mm256_mul_ps(fscal,dz10);
2116
2117             /* Update vectorial force */
2118             fix1             = _mm256_add_ps(fix1,tx);
2119             fiy1             = _mm256_add_ps(fiy1,ty);
2120             fiz1             = _mm256_add_ps(fiz1,tz);
2121
2122             fjx0             = _mm256_add_ps(fjx0,tx);
2123             fjy0             = _mm256_add_ps(fjy0,ty);
2124             fjz0             = _mm256_add_ps(fjz0,tz);
2125
2126             /**************************
2127              * CALCULATE INTERACTIONS *
2128              **************************/
2129
2130             r11              = _mm256_mul_ps(rsq11,rinv11);
2131             r11              = _mm256_andnot_ps(dummy_mask,r11);
2132
2133             /* EWALD ELECTROSTATICS */
2134
2135             /* Analytical PME correction */
2136             zeta2            = _mm256_mul_ps(beta2,rsq11);
2137             rinv3            = _mm256_mul_ps(rinvsq11,rinv11);
2138             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
2139             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
2140             felec            = _mm256_mul_ps(qq11,felec);
2141
2142             fscal            = felec;
2143
2144             fscal            = _mm256_andnot_ps(dummy_mask,fscal);
2145
2146             /* Calculate temporary vectorial force */
2147             tx               = _mm256_mul_ps(fscal,dx11);
2148             ty               = _mm256_mul_ps(fscal,dy11);
2149             tz               = _mm256_mul_ps(fscal,dz11);
2150
2151             /* Update vectorial force */
2152             fix1             = _mm256_add_ps(fix1,tx);
2153             fiy1             = _mm256_add_ps(fiy1,ty);
2154             fiz1             = _mm256_add_ps(fiz1,tz);
2155
2156             fjx1             = _mm256_add_ps(fjx1,tx);
2157             fjy1             = _mm256_add_ps(fjy1,ty);
2158             fjz1             = _mm256_add_ps(fjz1,tz);
2159
2160             /**************************
2161              * CALCULATE INTERACTIONS *
2162              **************************/
2163
2164             r12              = _mm256_mul_ps(rsq12,rinv12);
2165             r12              = _mm256_andnot_ps(dummy_mask,r12);
2166
2167             /* EWALD ELECTROSTATICS */
2168
2169             /* Analytical PME correction */
2170             zeta2            = _mm256_mul_ps(beta2,rsq12);
2171             rinv3            = _mm256_mul_ps(rinvsq12,rinv12);
2172             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
2173             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
2174             felec            = _mm256_mul_ps(qq12,felec);
2175
2176             fscal            = felec;
2177
2178             fscal            = _mm256_andnot_ps(dummy_mask,fscal);
2179
2180             /* Calculate temporary vectorial force */
2181             tx               = _mm256_mul_ps(fscal,dx12);
2182             ty               = _mm256_mul_ps(fscal,dy12);
2183             tz               = _mm256_mul_ps(fscal,dz12);
2184
2185             /* Update vectorial force */
2186             fix1             = _mm256_add_ps(fix1,tx);
2187             fiy1             = _mm256_add_ps(fiy1,ty);
2188             fiz1             = _mm256_add_ps(fiz1,tz);
2189
2190             fjx2             = _mm256_add_ps(fjx2,tx);
2191             fjy2             = _mm256_add_ps(fjy2,ty);
2192             fjz2             = _mm256_add_ps(fjz2,tz);
2193
2194             /**************************
2195              * CALCULATE INTERACTIONS *
2196              **************************/
2197
2198             r20              = _mm256_mul_ps(rsq20,rinv20);
2199             r20              = _mm256_andnot_ps(dummy_mask,r20);
2200
2201             /* EWALD ELECTROSTATICS */
2202
2203             /* Analytical PME correction */
2204             zeta2            = _mm256_mul_ps(beta2,rsq20);
2205             rinv3            = _mm256_mul_ps(rinvsq20,rinv20);
2206             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
2207             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
2208             felec            = _mm256_mul_ps(qq20,felec);
2209
2210             fscal            = felec;
2211
2212             fscal            = _mm256_andnot_ps(dummy_mask,fscal);
2213
2214             /* Calculate temporary vectorial force */
2215             tx               = _mm256_mul_ps(fscal,dx20);
2216             ty               = _mm256_mul_ps(fscal,dy20);
2217             tz               = _mm256_mul_ps(fscal,dz20);
2218
2219             /* Update vectorial force */
2220             fix2             = _mm256_add_ps(fix2,tx);
2221             fiy2             = _mm256_add_ps(fiy2,ty);
2222             fiz2             = _mm256_add_ps(fiz2,tz);
2223
2224             fjx0             = _mm256_add_ps(fjx0,tx);
2225             fjy0             = _mm256_add_ps(fjy0,ty);
2226             fjz0             = _mm256_add_ps(fjz0,tz);
2227
2228             /**************************
2229              * CALCULATE INTERACTIONS *
2230              **************************/
2231
2232             r21              = _mm256_mul_ps(rsq21,rinv21);
2233             r21              = _mm256_andnot_ps(dummy_mask,r21);
2234
2235             /* EWALD ELECTROSTATICS */
2236
2237             /* Analytical PME correction */
2238             zeta2            = _mm256_mul_ps(beta2,rsq21);
2239             rinv3            = _mm256_mul_ps(rinvsq21,rinv21);
2240             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
2241             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
2242             felec            = _mm256_mul_ps(qq21,felec);
2243
2244             fscal            = felec;
2245
2246             fscal            = _mm256_andnot_ps(dummy_mask,fscal);
2247
2248             /* Calculate temporary vectorial force */
2249             tx               = _mm256_mul_ps(fscal,dx21);
2250             ty               = _mm256_mul_ps(fscal,dy21);
2251             tz               = _mm256_mul_ps(fscal,dz21);
2252
2253             /* Update vectorial force */
2254             fix2             = _mm256_add_ps(fix2,tx);
2255             fiy2             = _mm256_add_ps(fiy2,ty);
2256             fiz2             = _mm256_add_ps(fiz2,tz);
2257
2258             fjx1             = _mm256_add_ps(fjx1,tx);
2259             fjy1             = _mm256_add_ps(fjy1,ty);
2260             fjz1             = _mm256_add_ps(fjz1,tz);
2261
2262             /**************************
2263              * CALCULATE INTERACTIONS *
2264              **************************/
2265
2266             r22              = _mm256_mul_ps(rsq22,rinv22);
2267             r22              = _mm256_andnot_ps(dummy_mask,r22);
2268
2269             /* EWALD ELECTROSTATICS */
2270
2271             /* Analytical PME correction */
2272             zeta2            = _mm256_mul_ps(beta2,rsq22);
2273             rinv3            = _mm256_mul_ps(rinvsq22,rinv22);
2274             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
2275             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
2276             felec            = _mm256_mul_ps(qq22,felec);
2277
2278             fscal            = felec;
2279
2280             fscal            = _mm256_andnot_ps(dummy_mask,fscal);
2281
2282             /* Calculate temporary vectorial force */
2283             tx               = _mm256_mul_ps(fscal,dx22);
2284             ty               = _mm256_mul_ps(fscal,dy22);
2285             tz               = _mm256_mul_ps(fscal,dz22);
2286
2287             /* Update vectorial force */
2288             fix2             = _mm256_add_ps(fix2,tx);
2289             fiy2             = _mm256_add_ps(fiy2,ty);
2290             fiz2             = _mm256_add_ps(fiz2,tz);
2291
2292             fjx2             = _mm256_add_ps(fjx2,tx);
2293             fjy2             = _mm256_add_ps(fjy2,ty);
2294             fjz2             = _mm256_add_ps(fjz2,tz);
2295
2296             fjptrA             = (jnrlistA>=0) ? f+j_coord_offsetA : scratch;
2297             fjptrB             = (jnrlistB>=0) ? f+j_coord_offsetB : scratch;
2298             fjptrC             = (jnrlistC>=0) ? f+j_coord_offsetC : scratch;
2299             fjptrD             = (jnrlistD>=0) ? f+j_coord_offsetD : scratch;
2300             fjptrE             = (jnrlistE>=0) ? f+j_coord_offsetE : scratch;
2301             fjptrF             = (jnrlistF>=0) ? f+j_coord_offsetF : scratch;
2302             fjptrG             = (jnrlistG>=0) ? f+j_coord_offsetG : scratch;
2303             fjptrH             = (jnrlistH>=0) ? f+j_coord_offsetH : scratch;
2304
2305             gmx_mm256_decrement_3rvec_8ptr_swizzle_ps(fjptrA,fjptrB,fjptrC,fjptrD,fjptrE,fjptrF,fjptrG,fjptrH,
2306                                                       fjx0,fjy0,fjz0,fjx1,fjy1,fjz1,fjx2,fjy2,fjz2);
2307
2308             /* Inner loop uses 536 flops */
2309         }
2310
2311         /* End of innermost loop */
2312
2313         gmx_mm256_update_iforce_3atom_swizzle_ps(fix0,fiy0,fiz0,fix1,fiy1,fiz1,fix2,fiy2,fiz2,
2314                                                  f+i_coord_offset,fshift+i_shift_offset);
2315
2316         /* Increment number of inner iterations */
2317         inneriter                  += j_index_end - j_index_start;
2318
2319         /* Outer loop uses 18 flops */
2320     }
2321
2322     /* Increment number of outer iterations */
2323     outeriter        += nri;
2324
2325     /* Update outer/inner flops */
2326
2327     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W3W3_F,outeriter*18 + inneriter*536);
2328 }