src/gromacs/gmxlib/nonbonded/nb_kernel_avx_256_single/nb_kernel_ElecEwSh_VdwNone_GeomW4W4_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_ElecEwSh_VdwNone_GeomW4W4_VF_avx_256_single
  54  * Electrostatics interaction: Ewald
  55  * VdW interaction:            None
  56  * Geometry:                   Water4-Water4
  57  * Calculate force/pot:        PotentialAndForce
  58  */
  59 void
  60 nb_kernel_ElecEwSh_VdwNone_GeomW4W4_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 *           vdwioffsetptr1;
  89     __m256           ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
  90     real *           vdwioffsetptr2;
  91     __m256           ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
  92     real *           vdwioffsetptr3;
  93     __m256           ix3,iy3,iz3,fix3,fiy3,fiz3,iq3,isai3;
  94     int              vdwjidx1A,vdwjidx1B,vdwjidx1C,vdwjidx1D,vdwjidx1E,vdwjidx1F,vdwjidx1G,vdwjidx1H;
  95     __m256           jx1,jy1,jz1,fjx1,fjy1,fjz1,jq1,isaj1;
  96     int              vdwjidx2A,vdwjidx2B,vdwjidx2C,vdwjidx2D,vdwjidx2E,vdwjidx2F,vdwjidx2G,vdwjidx2H;
  97     __m256           jx2,jy2,jz2,fjx2,fjy2,fjz2,jq2,isaj2;
  98     int              vdwjidx3A,vdwjidx3B,vdwjidx3C,vdwjidx3D,vdwjidx3E,vdwjidx3F,vdwjidx3G,vdwjidx3H;
  99     __m256           jx3,jy3,jz3,fjx3,fjy3,fjz3,jq3,isaj3;
 100     __m256           dx11,dy11,dz11,rsq11,rinv11,rinvsq11,r11,qq11,c6_11,c12_11;
 101     __m256           dx12,dy12,dz12,rsq12,rinv12,rinvsq12,r12,qq12,c6_12,c12_12;
 102     __m256           dx13,dy13,dz13,rsq13,rinv13,rinvsq13,r13,qq13,c6_13,c12_13;
 103     __m256           dx21,dy21,dz21,rsq21,rinv21,rinvsq21,r21,qq21,c6_21,c12_21;
 104     __m256           dx22,dy22,dz22,rsq22,rinv22,rinvsq22,r22,qq22,c6_22,c12_22;
 105     __m256           dx23,dy23,dz23,rsq23,rinv23,rinvsq23,r23,qq23,c6_23,c12_23;
 106     __m256           dx31,dy31,dz31,rsq31,rinv31,rinvsq31,r31,qq31,c6_31,c12_31;
 107     __m256           dx32,dy32,dz32,rsq32,rinv32,rinvsq32,r32,qq32,c6_32,c12_32;
 108     __m256           dx33,dy33,dz33,rsq33,rinv33,rinvsq33,r33,qq33,c6_33,c12_33;
 109     __m256           velec,felec,velecsum,facel,crf,krf,krf2;
 110     real             *charge;
 111     __m256i          ewitab;
 112     __m128i          ewitab_lo,ewitab_hi;
 113     __m256           ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace,ewtabF,ewtabFn,ewtabD,ewtabV;
 114     __m256           beta,beta2,beta3,zeta2,pmecorrF,pmecorrV,rinv3;
 115     real             *ewtab;
 116     __m256           dummy_mask,cutoff_mask;
 117     __m256           signbit = _mm256_castsi256_ps( _mm256_set1_epi32(0x80000000) );
 118     __m256           one     = _mm256_set1_ps(1.0);
 119     __m256           two     = _mm256_set1_ps(2.0);
 120     x                = xx[0];
 121     f                = ff[0];
 122
 123     nri              = nlist->nri;
 124     iinr             = nlist->iinr;
 125     jindex           = nlist->jindex;
 126     jjnr             = nlist->jjnr;
 127     shiftidx         = nlist->shift;
 128     gid              = nlist->gid;
 129     shiftvec         = fr->shift_vec[0];
 130     fshift           = fr->fshift[0];
 131     facel            = _mm256_set1_ps(fr->epsfac);
 132     charge           = mdatoms->chargeA;
 133
 134     sh_ewald         = _mm256_set1_ps(fr->ic->sh_ewald);
 135     beta             = _mm256_set1_ps(fr->ic->ewaldcoeff_q);
 136     beta2            = _mm256_mul_ps(beta,beta);
 137     beta3            = _mm256_mul_ps(beta,beta2);
 138
 139     ewtab            = fr->ic->tabq_coul_FDV0;
 140     ewtabscale       = _mm256_set1_ps(fr->ic->tabq_scale);
 141     ewtabhalfspace   = _mm256_set1_ps(0.5/fr->ic->tabq_scale);
 142
 143     /* Setup water-specific parameters */
 144     inr              = nlist->iinr[0];
 145     iq1              = _mm256_mul_ps(facel,_mm256_set1_ps(charge[inr+1]));
 146     iq2              = _mm256_mul_ps(facel,_mm256_set1_ps(charge[inr+2]));
 147     iq3              = _mm256_mul_ps(facel,_mm256_set1_ps(charge[inr+3]));
 148
 149     jq1              = _mm256_set1_ps(charge[inr+1]);
 150     jq2              = _mm256_set1_ps(charge[inr+2]);
 151     jq3              = _mm256_set1_ps(charge[inr+3]);
 152     qq11             = _mm256_mul_ps(iq1,jq1);
 153     qq12             = _mm256_mul_ps(iq1,jq2);
 154     qq13             = _mm256_mul_ps(iq1,jq3);
 155     qq21             = _mm256_mul_ps(iq2,jq1);
 156     qq22             = _mm256_mul_ps(iq2,jq2);
 157     qq23             = _mm256_mul_ps(iq2,jq3);
 158     qq31             = _mm256_mul_ps(iq3,jq1);
 159     qq32             = _mm256_mul_ps(iq3,jq2);
 160     qq33             = _mm256_mul_ps(iq3,jq3);
 161
 162     /* When we use explicit cutoffs the value must be identical for elec and VdW, so use elec as an arbitrary choice */
 163     rcutoff_scalar   = fr->rcoulomb;
 164     rcutoff          = _mm256_set1_ps(rcutoff_scalar);
 165     rcutoff2         = _mm256_mul_ps(rcutoff,rcutoff);
 166
 167     /* Avoid stupid compiler warnings */
 168     jnrA = jnrB = jnrC = jnrD = jnrE = jnrF = jnrG = jnrH = 0;
 169     j_coord_offsetA = 0;
 170     j_coord_offsetB = 0;
 171     j_coord_offsetC = 0;
 172     j_coord_offsetD = 0;
 173     j_coord_offsetE = 0;
 174     j_coord_offsetF = 0;
 175     j_coord_offsetG = 0;
 176     j_coord_offsetH = 0;
 177
 178     outeriter        = 0;
 179     inneriter        = 0;
 180
 181     for(iidx=0;iidx<4*DIM;iidx++)
 182     {
 183         scratch[iidx] = 0.0;
 184     }
 185
 186     /* Start outer loop over neighborlists */
 187     for(iidx=0; iidx<nri; iidx++)
 188     {
 189         /* Load shift vector for this list */
 190         i_shift_offset   = DIM*shiftidx[iidx];
 191
 192         /* Load limits for loop over neighbors */
 193         j_index_start    = jindex[iidx];
 194         j_index_end      = jindex[iidx+1];
 195
 196         /* Get outer coordinate index */
 197         inr              = iinr[iidx];
 198         i_coord_offset   = DIM*inr;
 199
 200         /* Load i particle coords and add shift vector */
 201         gmx_mm256_load_shift_and_3rvec_broadcast_ps(shiftvec+i_shift_offset,x+i_coord_offset+DIM,
 202                                                     &ix1,&iy1,&iz1,&ix2,&iy2,&iz2,&ix3,&iy3,&iz3);
 203
 204         fix1             = _mm256_setzero_ps();
 205         fiy1             = _mm256_setzero_ps();
 206         fiz1             = _mm256_setzero_ps();
 207         fix2             = _mm256_setzero_ps();
 208         fiy2             = _mm256_setzero_ps();
 209         fiz2             = _mm256_setzero_ps();
 210         fix3             = _mm256_setzero_ps();
 211         fiy3             = _mm256_setzero_ps();
 212         fiz3             = _mm256_setzero_ps();
 213
 214         /* Reset potential sums */
 215         velecsum         = _mm256_setzero_ps();
 216
 217         /* Start inner kernel loop */
 218         for(jidx=j_index_start; jidx<j_index_end && jjnr[jidx+7]>=0; jidx+=8)
 219         {
 220
 221             /* Get j neighbor index, and coordinate index */
 222             jnrA             = jjnr[jidx];
 223             jnrB             = jjnr[jidx+1];
 224             jnrC             = jjnr[jidx+2];
 225             jnrD             = jjnr[jidx+3];
 226             jnrE             = jjnr[jidx+4];
 227             jnrF             = jjnr[jidx+5];
 228             jnrG             = jjnr[jidx+6];
 229             jnrH             = jjnr[jidx+7];
 230             j_coord_offsetA  = DIM*jnrA;
 231             j_coord_offsetB  = DIM*jnrB;
 232             j_coord_offsetC  = DIM*jnrC;
 233             j_coord_offsetD  = DIM*jnrD;
 234             j_coord_offsetE  = DIM*jnrE;
 235             j_coord_offsetF  = DIM*jnrF;
 236             j_coord_offsetG  = DIM*jnrG;
 237             j_coord_offsetH  = DIM*jnrH;
 238
 239             /* load j atom coordinates */
 240             gmx_mm256_load_3rvec_8ptr_swizzle_ps(x+j_coord_offsetA+DIM,x+j_coord_offsetB+DIM,
 241                                                  x+j_coord_offsetC+DIM,x+j_coord_offsetD+DIM,
 242                                                  x+j_coord_offsetE+DIM,x+j_coord_offsetF+DIM,
 243                                                  x+j_coord_offsetG+DIM,x+j_coord_offsetH+DIM,
 244                                                  &jx1,&jy1,&jz1,&jx2,&jy2,&jz2,&jx3,&jy3,&jz3);
 245
 246             /* Calculate displacement vector */
 247             dx11             = _mm256_sub_ps(ix1,jx1);
 248             dy11             = _mm256_sub_ps(iy1,jy1);
 249             dz11             = _mm256_sub_ps(iz1,jz1);
 250             dx12             = _mm256_sub_ps(ix1,jx2);
 251             dy12             = _mm256_sub_ps(iy1,jy2);
 252             dz12             = _mm256_sub_ps(iz1,jz2);
 253             dx13             = _mm256_sub_ps(ix1,jx3);
 254             dy13             = _mm256_sub_ps(iy1,jy3);
 255             dz13             = _mm256_sub_ps(iz1,jz3);
 256             dx21             = _mm256_sub_ps(ix2,jx1);
 257             dy21             = _mm256_sub_ps(iy2,jy1);
 258             dz21             = _mm256_sub_ps(iz2,jz1);
 259             dx22             = _mm256_sub_ps(ix2,jx2);
 260             dy22             = _mm256_sub_ps(iy2,jy2);
 261             dz22             = _mm256_sub_ps(iz2,jz2);
 262             dx23             = _mm256_sub_ps(ix2,jx3);
 263             dy23             = _mm256_sub_ps(iy2,jy3);
 264             dz23             = _mm256_sub_ps(iz2,jz3);
 265             dx31             = _mm256_sub_ps(ix3,jx1);
 266             dy31             = _mm256_sub_ps(iy3,jy1);
 267             dz31             = _mm256_sub_ps(iz3,jz1);
 268             dx32             = _mm256_sub_ps(ix3,jx2);
 269             dy32             = _mm256_sub_ps(iy3,jy2);
 270             dz32             = _mm256_sub_ps(iz3,jz2);
 271             dx33             = _mm256_sub_ps(ix3,jx3);
 272             dy33             = _mm256_sub_ps(iy3,jy3);
 273             dz33             = _mm256_sub_ps(iz3,jz3);
 274
 275             /* Calculate squared distance and things based on it */
 276             rsq11            = gmx_mm256_calc_rsq_ps(dx11,dy11,dz11);
 277             rsq12            = gmx_mm256_calc_rsq_ps(dx12,dy12,dz12);
 278             rsq13            = gmx_mm256_calc_rsq_ps(dx13,dy13,dz13);
 279             rsq21            = gmx_mm256_calc_rsq_ps(dx21,dy21,dz21);
 280             rsq22            = gmx_mm256_calc_rsq_ps(dx22,dy22,dz22);
 281             rsq23            = gmx_mm256_calc_rsq_ps(dx23,dy23,dz23);
 282             rsq31            = gmx_mm256_calc_rsq_ps(dx31,dy31,dz31);
 283             rsq32            = gmx_mm256_calc_rsq_ps(dx32,dy32,dz32);
 284             rsq33            = gmx_mm256_calc_rsq_ps(dx33,dy33,dz33);
 285
 286             rinv11           = gmx_mm256_invsqrt_ps(rsq11);
 287             rinv12           = gmx_mm256_invsqrt_ps(rsq12);
 288             rinv13           = gmx_mm256_invsqrt_ps(rsq13);
 289             rinv21           = gmx_mm256_invsqrt_ps(rsq21);
 290             rinv22           = gmx_mm256_invsqrt_ps(rsq22);
 291             rinv23           = gmx_mm256_invsqrt_ps(rsq23);
 292             rinv31           = gmx_mm256_invsqrt_ps(rsq31);
 293             rinv32           = gmx_mm256_invsqrt_ps(rsq32);
 294             rinv33           = gmx_mm256_invsqrt_ps(rsq33);
 295
 296             rinvsq11         = _mm256_mul_ps(rinv11,rinv11);
 297             rinvsq12         = _mm256_mul_ps(rinv12,rinv12);
 298             rinvsq13         = _mm256_mul_ps(rinv13,rinv13);
 299             rinvsq21         = _mm256_mul_ps(rinv21,rinv21);
 300             rinvsq22         = _mm256_mul_ps(rinv22,rinv22);
 301             rinvsq23         = _mm256_mul_ps(rinv23,rinv23);
 302             rinvsq31         = _mm256_mul_ps(rinv31,rinv31);
 303             rinvsq32         = _mm256_mul_ps(rinv32,rinv32);
 304             rinvsq33         = _mm256_mul_ps(rinv33,rinv33);
 305
 306             fjx1             = _mm256_setzero_ps();
 307             fjy1             = _mm256_setzero_ps();
 308             fjz1             = _mm256_setzero_ps();
 309             fjx2             = _mm256_setzero_ps();
 310             fjy2             = _mm256_setzero_ps();
 311             fjz2             = _mm256_setzero_ps();
 312             fjx3             = _mm256_setzero_ps();
 313             fjy3             = _mm256_setzero_ps();
 314             fjz3             = _mm256_setzero_ps();
 315
 316             /**************************
 317              * CALCULATE INTERACTIONS *
 318              **************************/
 319
 320             if (gmx_mm256_any_lt(rsq11,rcutoff2))
 321             {
 322
 323             r11              = _mm256_mul_ps(rsq11,rinv11);
 324
 325             /* EWALD ELECTROSTATICS */
 326
 327             /* Analytical PME correction */
 328             zeta2            = _mm256_mul_ps(beta2,rsq11);
 329             rinv3            = _mm256_mul_ps(rinvsq11,rinv11);
 330             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
 331             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
 332             felec            = _mm256_mul_ps(qq11,felec);
 333             pmecorrV         = gmx_mm256_pmecorrV_ps(zeta2);
 334             pmecorrV         = _mm256_mul_ps(pmecorrV,beta);
 335             velec            = _mm256_sub_ps(_mm256_sub_ps(rinv11,sh_ewald),pmecorrV);
 336             velec            = _mm256_mul_ps(qq11,velec);
 337
 338             cutoff_mask      = _mm256_cmp_ps(rsq11,rcutoff2,_CMP_LT_OQ);
 339
 340             /* Update potential sum for this i atom from the interaction with this j atom. */
 341             velec            = _mm256_and_ps(velec,cutoff_mask);
 342             velecsum         = _mm256_add_ps(velecsum,velec);
 343
 344             fscal            = felec;
 345
 346             fscal            = _mm256_and_ps(fscal,cutoff_mask);
 347
 348             /* Calculate temporary vectorial force */
 349             tx               = _mm256_mul_ps(fscal,dx11);
 350             ty               = _mm256_mul_ps(fscal,dy11);
 351             tz               = _mm256_mul_ps(fscal,dz11);
 352
 353             /* Update vectorial force */
 354             fix1             = _mm256_add_ps(fix1,tx);
 355             fiy1             = _mm256_add_ps(fiy1,ty);
 356             fiz1             = _mm256_add_ps(fiz1,tz);
 357
 358             fjx1             = _mm256_add_ps(fjx1,tx);
 359             fjy1             = _mm256_add_ps(fjy1,ty);
 360             fjz1             = _mm256_add_ps(fjz1,tz);
 361
 362             }
 363
 364             /**************************
 365              * CALCULATE INTERACTIONS *
 366              **************************/
 367
 368             if (gmx_mm256_any_lt(rsq12,rcutoff2))
 369             {
 370
 371             r12              = _mm256_mul_ps(rsq12,rinv12);
 372
 373             /* EWALD ELECTROSTATICS */
 374
 375             /* Analytical PME correction */
 376             zeta2            = _mm256_mul_ps(beta2,rsq12);
 377             rinv3            = _mm256_mul_ps(rinvsq12,rinv12);
 378             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
 379             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
 380             felec            = _mm256_mul_ps(qq12,felec);
 381             pmecorrV         = gmx_mm256_pmecorrV_ps(zeta2);
 382             pmecorrV         = _mm256_mul_ps(pmecorrV,beta);
 383             velec            = _mm256_sub_ps(_mm256_sub_ps(rinv12,sh_ewald),pmecorrV);
 384             velec            = _mm256_mul_ps(qq12,velec);
 385
 386             cutoff_mask      = _mm256_cmp_ps(rsq12,rcutoff2,_CMP_LT_OQ);
 387
 388             /* Update potential sum for this i atom from the interaction with this j atom. */
 389             velec            = _mm256_and_ps(velec,cutoff_mask);
 390             velecsum         = _mm256_add_ps(velecsum,velec);
 391
 392             fscal            = felec;
 393
 394             fscal            = _mm256_and_ps(fscal,cutoff_mask);
 395
 396             /* Calculate temporary vectorial force */
 397             tx               = _mm256_mul_ps(fscal,dx12);
 398             ty               = _mm256_mul_ps(fscal,dy12);
 399             tz               = _mm256_mul_ps(fscal,dz12);
 400
 401             /* Update vectorial force */
 402             fix1             = _mm256_add_ps(fix1,tx);
 403             fiy1             = _mm256_add_ps(fiy1,ty);
 404             fiz1             = _mm256_add_ps(fiz1,tz);
 405
 406             fjx2             = _mm256_add_ps(fjx2,tx);
 407             fjy2             = _mm256_add_ps(fjy2,ty);
 408             fjz2             = _mm256_add_ps(fjz2,tz);
 409
 410             }
 411
 412             /**************************
 413              * CALCULATE INTERACTIONS *
 414              **************************/
 415
 416             if (gmx_mm256_any_lt(rsq13,rcutoff2))
 417             {
 418
 419             r13              = _mm256_mul_ps(rsq13,rinv13);
 420
 421             /* EWALD ELECTROSTATICS */
 422
 423             /* Analytical PME correction */
 424             zeta2            = _mm256_mul_ps(beta2,rsq13);
 425             rinv3            = _mm256_mul_ps(rinvsq13,rinv13);
 426             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
 427             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
 428             felec            = _mm256_mul_ps(qq13,felec);
 429             pmecorrV         = gmx_mm256_pmecorrV_ps(zeta2);
 430             pmecorrV         = _mm256_mul_ps(pmecorrV,beta);
 431             velec            = _mm256_sub_ps(_mm256_sub_ps(rinv13,sh_ewald),pmecorrV);
 432             velec            = _mm256_mul_ps(qq13,velec);
 433
 434             cutoff_mask      = _mm256_cmp_ps(rsq13,rcutoff2,_CMP_LT_OQ);
 435
 436             /* Update potential sum for this i atom from the interaction with this j atom. */
 437             velec            = _mm256_and_ps(velec,cutoff_mask);
 438             velecsum         = _mm256_add_ps(velecsum,velec);
 439
 440             fscal            = felec;
 441
 442             fscal            = _mm256_and_ps(fscal,cutoff_mask);
 443
 444             /* Calculate temporary vectorial force */
 445             tx               = _mm256_mul_ps(fscal,dx13);
 446             ty               = _mm256_mul_ps(fscal,dy13);
 447             tz               = _mm256_mul_ps(fscal,dz13);
 448
 449             /* Update vectorial force */
 450             fix1             = _mm256_add_ps(fix1,tx);
 451             fiy1             = _mm256_add_ps(fiy1,ty);
 452             fiz1             = _mm256_add_ps(fiz1,tz);
 453
 454             fjx3             = _mm256_add_ps(fjx3,tx);
 455             fjy3             = _mm256_add_ps(fjy3,ty);
 456             fjz3             = _mm256_add_ps(fjz3,tz);
 457
 458             }
 459
 460             /**************************
 461              * CALCULATE INTERACTIONS *
 462              **************************/
 463
 464             if (gmx_mm256_any_lt(rsq21,rcutoff2))
 465             {
 466
 467             r21              = _mm256_mul_ps(rsq21,rinv21);
 468
 469             /* EWALD ELECTROSTATICS */
 470
 471             /* Analytical PME correction */
 472             zeta2            = _mm256_mul_ps(beta2,rsq21);
 473             rinv3            = _mm256_mul_ps(rinvsq21,rinv21);
 474             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
 475             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
 476             felec            = _mm256_mul_ps(qq21,felec);
 477             pmecorrV         = gmx_mm256_pmecorrV_ps(zeta2);
 478             pmecorrV         = _mm256_mul_ps(pmecorrV,beta);
 479             velec            = _mm256_sub_ps(_mm256_sub_ps(rinv21,sh_ewald),pmecorrV);
 480             velec            = _mm256_mul_ps(qq21,velec);
 481
 482             cutoff_mask      = _mm256_cmp_ps(rsq21,rcutoff2,_CMP_LT_OQ);
 483
 484             /* Update potential sum for this i atom from the interaction with this j atom. */
 485             velec            = _mm256_and_ps(velec,cutoff_mask);
 486             velecsum         = _mm256_add_ps(velecsum,velec);
 487
 488             fscal            = felec;
 489
 490             fscal            = _mm256_and_ps(fscal,cutoff_mask);
 491
 492             /* Calculate temporary vectorial force */
 493             tx               = _mm256_mul_ps(fscal,dx21);
 494             ty               = _mm256_mul_ps(fscal,dy21);
 495             tz               = _mm256_mul_ps(fscal,dz21);
 496
 497             /* Update vectorial force */
 498             fix2             = _mm256_add_ps(fix2,tx);
 499             fiy2             = _mm256_add_ps(fiy2,ty);
 500             fiz2             = _mm256_add_ps(fiz2,tz);
 501
 502             fjx1             = _mm256_add_ps(fjx1,tx);
 503             fjy1             = _mm256_add_ps(fjy1,ty);
 504             fjz1             = _mm256_add_ps(fjz1,tz);
 505
 506             }
 507
 508             /**************************
 509              * CALCULATE INTERACTIONS *
 510              **************************/
 511
 512             if (gmx_mm256_any_lt(rsq22,rcutoff2))
 513             {
 514
 515             r22              = _mm256_mul_ps(rsq22,rinv22);
 516
 517             /* EWALD ELECTROSTATICS */
 518
 519             /* Analytical PME correction */
 520             zeta2            = _mm256_mul_ps(beta2,rsq22);
 521             rinv3            = _mm256_mul_ps(rinvsq22,rinv22);
 522             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
 523             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
 524             felec            = _mm256_mul_ps(qq22,felec);
 525             pmecorrV         = gmx_mm256_pmecorrV_ps(zeta2);
 526             pmecorrV         = _mm256_mul_ps(pmecorrV,beta);
 527             velec            = _mm256_sub_ps(_mm256_sub_ps(rinv22,sh_ewald),pmecorrV);
 528             velec            = _mm256_mul_ps(qq22,velec);
 529
 530             cutoff_mask      = _mm256_cmp_ps(rsq22,rcutoff2,_CMP_LT_OQ);
 531
 532             /* Update potential sum for this i atom from the interaction with this j atom. */
 533             velec            = _mm256_and_ps(velec,cutoff_mask);
 534             velecsum         = _mm256_add_ps(velecsum,velec);
 535
 536             fscal            = felec;
 537
 538             fscal            = _mm256_and_ps(fscal,cutoff_mask);
 539
 540             /* Calculate temporary vectorial force */
 541             tx               = _mm256_mul_ps(fscal,dx22);
 542             ty               = _mm256_mul_ps(fscal,dy22);
 543             tz               = _mm256_mul_ps(fscal,dz22);
 544
 545             /* Update vectorial force */
 546             fix2             = _mm256_add_ps(fix2,tx);
 547             fiy2             = _mm256_add_ps(fiy2,ty);
 548             fiz2             = _mm256_add_ps(fiz2,tz);
 549
 550             fjx2             = _mm256_add_ps(fjx2,tx);
 551             fjy2             = _mm256_add_ps(fjy2,ty);
 552             fjz2             = _mm256_add_ps(fjz2,tz);
 553
 554             }
 555
 556             /**************************
 557              * CALCULATE INTERACTIONS *
 558              **************************/
 559
 560             if (gmx_mm256_any_lt(rsq23,rcutoff2))
 561             {
 562
 563             r23              = _mm256_mul_ps(rsq23,rinv23);
 564
 565             /* EWALD ELECTROSTATICS */
 566
 567             /* Analytical PME correction */
 568             zeta2            = _mm256_mul_ps(beta2,rsq23);
 569             rinv3            = _mm256_mul_ps(rinvsq23,rinv23);
 570             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
 571             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
 572             felec            = _mm256_mul_ps(qq23,felec);
 573             pmecorrV         = gmx_mm256_pmecorrV_ps(zeta2);
 574             pmecorrV         = _mm256_mul_ps(pmecorrV,beta);
 575             velec            = _mm256_sub_ps(_mm256_sub_ps(rinv23,sh_ewald),pmecorrV);
 576             velec            = _mm256_mul_ps(qq23,velec);
 577
 578             cutoff_mask      = _mm256_cmp_ps(rsq23,rcutoff2,_CMP_LT_OQ);
 579
 580             /* Update potential sum for this i atom from the interaction with this j atom. */
 581             velec            = _mm256_and_ps(velec,cutoff_mask);
 582             velecsum         = _mm256_add_ps(velecsum,velec);
 583
 584             fscal            = felec;
 585
 586             fscal            = _mm256_and_ps(fscal,cutoff_mask);
 587
 588             /* Calculate temporary vectorial force */
 589             tx               = _mm256_mul_ps(fscal,dx23);
 590             ty               = _mm256_mul_ps(fscal,dy23);
 591             tz               = _mm256_mul_ps(fscal,dz23);
 592
 593             /* Update vectorial force */
 594             fix2             = _mm256_add_ps(fix2,tx);
 595             fiy2             = _mm256_add_ps(fiy2,ty);
 596             fiz2             = _mm256_add_ps(fiz2,tz);
 597
 598             fjx3             = _mm256_add_ps(fjx3,tx);
 599             fjy3             = _mm256_add_ps(fjy3,ty);
 600             fjz3             = _mm256_add_ps(fjz3,tz);
 601
 602             }
 603
 604             /**************************
 605              * CALCULATE INTERACTIONS *
 606              **************************/
 607
 608             if (gmx_mm256_any_lt(rsq31,rcutoff2))
 609             {
 610
 611             r31              = _mm256_mul_ps(rsq31,rinv31);
 612
 613             /* EWALD ELECTROSTATICS */
 614
 615             /* Analytical PME correction */
 616             zeta2            = _mm256_mul_ps(beta2,rsq31);
 617             rinv3            = _mm256_mul_ps(rinvsq31,rinv31);
 618             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
 619             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
 620             felec            = _mm256_mul_ps(qq31,felec);
 621             pmecorrV         = gmx_mm256_pmecorrV_ps(zeta2);
 622             pmecorrV         = _mm256_mul_ps(pmecorrV,beta);
 623             velec            = _mm256_sub_ps(_mm256_sub_ps(rinv31,sh_ewald),pmecorrV);
 624             velec            = _mm256_mul_ps(qq31,velec);
 625
 626             cutoff_mask      = _mm256_cmp_ps(rsq31,rcutoff2,_CMP_LT_OQ);
 627
 628             /* Update potential sum for this i atom from the interaction with this j atom. */
 629             velec            = _mm256_and_ps(velec,cutoff_mask);
 630             velecsum         = _mm256_add_ps(velecsum,velec);
 631
 632             fscal            = felec;
 633
 634             fscal            = _mm256_and_ps(fscal,cutoff_mask);
 635
 636             /* Calculate temporary vectorial force */
 637             tx               = _mm256_mul_ps(fscal,dx31);
 638             ty               = _mm256_mul_ps(fscal,dy31);
 639             tz               = _mm256_mul_ps(fscal,dz31);
 640
 641             /* Update vectorial force */
 642             fix3             = _mm256_add_ps(fix3,tx);
 643             fiy3             = _mm256_add_ps(fiy3,ty);
 644             fiz3             = _mm256_add_ps(fiz3,tz);
 645
 646             fjx1             = _mm256_add_ps(fjx1,tx);
 647             fjy1             = _mm256_add_ps(fjy1,ty);
 648             fjz1             = _mm256_add_ps(fjz1,tz);
 649
 650             }
 651
 652             /**************************
 653              * CALCULATE INTERACTIONS *
 654              **************************/
 655
 656             if (gmx_mm256_any_lt(rsq32,rcutoff2))
 657             {
 658
 659             r32              = _mm256_mul_ps(rsq32,rinv32);
 660
 661             /* EWALD ELECTROSTATICS */
 662
 663             /* Analytical PME correction */
 664             zeta2            = _mm256_mul_ps(beta2,rsq32);
 665             rinv3            = _mm256_mul_ps(rinvsq32,rinv32);
 666             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
 667             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
 668             felec            = _mm256_mul_ps(qq32,felec);
 669             pmecorrV         = gmx_mm256_pmecorrV_ps(zeta2);
 670             pmecorrV         = _mm256_mul_ps(pmecorrV,beta);
 671             velec            = _mm256_sub_ps(_mm256_sub_ps(rinv32,sh_ewald),pmecorrV);
 672             velec            = _mm256_mul_ps(qq32,velec);
 673
 674             cutoff_mask      = _mm256_cmp_ps(rsq32,rcutoff2,_CMP_LT_OQ);
 675
 676             /* Update potential sum for this i atom from the interaction with this j atom. */
 677             velec            = _mm256_and_ps(velec,cutoff_mask);
 678             velecsum         = _mm256_add_ps(velecsum,velec);
 679
 680             fscal            = felec;
 681
 682             fscal            = _mm256_and_ps(fscal,cutoff_mask);
 683
 684             /* Calculate temporary vectorial force */
 685             tx               = _mm256_mul_ps(fscal,dx32);
 686             ty               = _mm256_mul_ps(fscal,dy32);
 687             tz               = _mm256_mul_ps(fscal,dz32);
 688
 689             /* Update vectorial force */
 690             fix3             = _mm256_add_ps(fix3,tx);
 691             fiy3             = _mm256_add_ps(fiy3,ty);
 692             fiz3             = _mm256_add_ps(fiz3,tz);
 693
 694             fjx2             = _mm256_add_ps(fjx2,tx);
 695             fjy2             = _mm256_add_ps(fjy2,ty);
 696             fjz2             = _mm256_add_ps(fjz2,tz);
 697
 698             }
 699
 700             /**************************
 701              * CALCULATE INTERACTIONS *
 702              **************************/
 703
 704             if (gmx_mm256_any_lt(rsq33,rcutoff2))
 705             {
 706
 707             r33              = _mm256_mul_ps(rsq33,rinv33);
 708
 709             /* EWALD ELECTROSTATICS */
 710
 711             /* Analytical PME correction */
 712             zeta2            = _mm256_mul_ps(beta2,rsq33);
 713             rinv3            = _mm256_mul_ps(rinvsq33,rinv33);
 714             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
 715             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
 716             felec            = _mm256_mul_ps(qq33,felec);
 717             pmecorrV         = gmx_mm256_pmecorrV_ps(zeta2);
 718             pmecorrV         = _mm256_mul_ps(pmecorrV,beta);
 719             velec            = _mm256_sub_ps(_mm256_sub_ps(rinv33,sh_ewald),pmecorrV);
 720             velec            = _mm256_mul_ps(qq33,velec);
 721
 722             cutoff_mask      = _mm256_cmp_ps(rsq33,rcutoff2,_CMP_LT_OQ);
 723
 724             /* Update potential sum for this i atom from the interaction with this j atom. */
 725             velec            = _mm256_and_ps(velec,cutoff_mask);
 726             velecsum         = _mm256_add_ps(velecsum,velec);
 727
 728             fscal            = felec;
 729
 730             fscal            = _mm256_and_ps(fscal,cutoff_mask);
 731
 732             /* Calculate temporary vectorial force */
 733             tx               = _mm256_mul_ps(fscal,dx33);
 734             ty               = _mm256_mul_ps(fscal,dy33);
 735             tz               = _mm256_mul_ps(fscal,dz33);
 736
 737             /* Update vectorial force */
 738             fix3             = _mm256_add_ps(fix3,tx);
 739             fiy3             = _mm256_add_ps(fiy3,ty);
 740             fiz3             = _mm256_add_ps(fiz3,tz);
 741
 742             fjx3             = _mm256_add_ps(fjx3,tx);
 743             fjy3             = _mm256_add_ps(fjy3,ty);
 744             fjz3             = _mm256_add_ps(fjz3,tz);
 745
 746             }
 747
 748             fjptrA             = f+j_coord_offsetA;
 749             fjptrB             = f+j_coord_offsetB;
 750             fjptrC             = f+j_coord_offsetC;
 751             fjptrD             = f+j_coord_offsetD;
 752             fjptrE             = f+j_coord_offsetE;
 753             fjptrF             = f+j_coord_offsetF;
 754             fjptrG             = f+j_coord_offsetG;
 755             fjptrH             = f+j_coord_offsetH;
 756
 757             gmx_mm256_decrement_3rvec_8ptr_swizzle_ps(fjptrA+DIM,fjptrB+DIM,fjptrC+DIM,fjptrD+DIM,
 758                                                       fjptrE+DIM,fjptrF+DIM,fjptrG+DIM,fjptrH+DIM,
 759                                                       fjx1,fjy1,fjz1,fjx2,fjy2,fjz2,fjx3,fjy3,fjz3);
 760
 761             /* Inner loop uses 981 flops */
 762         }
 763
 764         if(jidx<j_index_end)
 765         {
 766
 767             /* Get j neighbor index, and coordinate index */
 768             jnrlistA         = jjnr[jidx];
 769             jnrlistB         = jjnr[jidx+1];
 770             jnrlistC         = jjnr[jidx+2];
 771             jnrlistD         = jjnr[jidx+3];
 772             jnrlistE         = jjnr[jidx+4];
 773             jnrlistF         = jjnr[jidx+5];
 774             jnrlistG         = jjnr[jidx+6];
 775             jnrlistH         = jjnr[jidx+7];
 776             /* Sign of each element will be negative for non-real atoms.
 777              * This mask will be 0xFFFFFFFF for dummy entries and 0x0 for real ones,
 778              * so use it as val = _mm_andnot_ps(mask,val) to clear dummy entries.
 779              */
 780             dummy_mask = gmx_mm256_set_m128(gmx_mm_castsi128_ps(_mm_cmplt_epi32(_mm_loadu_si128((const __m128i *)(jjnr+jidx+4)),_mm_setzero_si128())),
 781                                             gmx_mm_castsi128_ps(_mm_cmplt_epi32(_mm_loadu_si128((const __m128i *)(jjnr+jidx)),_mm_setzero_si128())));
 782
 783             jnrA       = (jnrlistA>=0) ? jnrlistA : 0;
 784             jnrB       = (jnrlistB>=0) ? jnrlistB : 0;
 785             jnrC       = (jnrlistC>=0) ? jnrlistC : 0;
 786             jnrD       = (jnrlistD>=0) ? jnrlistD : 0;
 787             jnrE       = (jnrlistE>=0) ? jnrlistE : 0;
 788             jnrF       = (jnrlistF>=0) ? jnrlistF : 0;
 789             jnrG       = (jnrlistG>=0) ? jnrlistG : 0;
 790             jnrH       = (jnrlistH>=0) ? jnrlistH : 0;
 791             j_coord_offsetA  = DIM*jnrA;
 792             j_coord_offsetB  = DIM*jnrB;
 793             j_coord_offsetC  = DIM*jnrC;
 794             j_coord_offsetD  = DIM*jnrD;
 795             j_coord_offsetE  = DIM*jnrE;
 796             j_coord_offsetF  = DIM*jnrF;
 797             j_coord_offsetG  = DIM*jnrG;
 798             j_coord_offsetH  = DIM*jnrH;
 799
 800             /* load j atom coordinates */
 801             gmx_mm256_load_3rvec_8ptr_swizzle_ps(x+j_coord_offsetA+DIM,x+j_coord_offsetB+DIM,
 802                                                  x+j_coord_offsetC+DIM,x+j_coord_offsetD+DIM,
 803                                                  x+j_coord_offsetE+DIM,x+j_coord_offsetF+DIM,
 804                                                  x+j_coord_offsetG+DIM,x+j_coord_offsetH+DIM,
 805                                                  &jx1,&jy1,&jz1,&jx2,&jy2,&jz2,&jx3,&jy3,&jz3);
 806
 807             /* Calculate displacement vector */
 808             dx11             = _mm256_sub_ps(ix1,jx1);
 809             dy11             = _mm256_sub_ps(iy1,jy1);
 810             dz11             = _mm256_sub_ps(iz1,jz1);
 811             dx12             = _mm256_sub_ps(ix1,jx2);
 812             dy12             = _mm256_sub_ps(iy1,jy2);
 813             dz12             = _mm256_sub_ps(iz1,jz2);
 814             dx13             = _mm256_sub_ps(ix1,jx3);
 815             dy13             = _mm256_sub_ps(iy1,jy3);
 816             dz13             = _mm256_sub_ps(iz1,jz3);
 817             dx21             = _mm256_sub_ps(ix2,jx1);
 818             dy21             = _mm256_sub_ps(iy2,jy1);
 819             dz21             = _mm256_sub_ps(iz2,jz1);
 820             dx22             = _mm256_sub_ps(ix2,jx2);
 821             dy22             = _mm256_sub_ps(iy2,jy2);
 822             dz22             = _mm256_sub_ps(iz2,jz2);
 823             dx23             = _mm256_sub_ps(ix2,jx3);
 824             dy23             = _mm256_sub_ps(iy2,jy3);
 825             dz23             = _mm256_sub_ps(iz2,jz3);
 826             dx31             = _mm256_sub_ps(ix3,jx1);
 827             dy31             = _mm256_sub_ps(iy3,jy1);
 828             dz31             = _mm256_sub_ps(iz3,jz1);
 829             dx32             = _mm256_sub_ps(ix3,jx2);
 830             dy32             = _mm256_sub_ps(iy3,jy2);
 831             dz32             = _mm256_sub_ps(iz3,jz2);
 832             dx33             = _mm256_sub_ps(ix3,jx3);
 833             dy33             = _mm256_sub_ps(iy3,jy3);
 834             dz33             = _mm256_sub_ps(iz3,jz3);
 835
 836             /* Calculate squared distance and things based on it */
 837             rsq11            = gmx_mm256_calc_rsq_ps(dx11,dy11,dz11);
 838             rsq12            = gmx_mm256_calc_rsq_ps(dx12,dy12,dz12);
 839             rsq13            = gmx_mm256_calc_rsq_ps(dx13,dy13,dz13);
 840             rsq21            = gmx_mm256_calc_rsq_ps(dx21,dy21,dz21);
 841             rsq22            = gmx_mm256_calc_rsq_ps(dx22,dy22,dz22);
 842             rsq23            = gmx_mm256_calc_rsq_ps(dx23,dy23,dz23);
 843             rsq31            = gmx_mm256_calc_rsq_ps(dx31,dy31,dz31);
 844             rsq32            = gmx_mm256_calc_rsq_ps(dx32,dy32,dz32);
 845             rsq33            = gmx_mm256_calc_rsq_ps(dx33,dy33,dz33);
 846
 847             rinv11           = gmx_mm256_invsqrt_ps(rsq11);
 848             rinv12           = gmx_mm256_invsqrt_ps(rsq12);
 849             rinv13           = gmx_mm256_invsqrt_ps(rsq13);
 850             rinv21           = gmx_mm256_invsqrt_ps(rsq21);
 851             rinv22           = gmx_mm256_invsqrt_ps(rsq22);
 852             rinv23           = gmx_mm256_invsqrt_ps(rsq23);
 853             rinv31           = gmx_mm256_invsqrt_ps(rsq31);
 854             rinv32           = gmx_mm256_invsqrt_ps(rsq32);
 855             rinv33           = gmx_mm256_invsqrt_ps(rsq33);
 856
 857             rinvsq11         = _mm256_mul_ps(rinv11,rinv11);
 858             rinvsq12         = _mm256_mul_ps(rinv12,rinv12);
 859             rinvsq13         = _mm256_mul_ps(rinv13,rinv13);
 860             rinvsq21         = _mm256_mul_ps(rinv21,rinv21);
 861             rinvsq22         = _mm256_mul_ps(rinv22,rinv22);
 862             rinvsq23         = _mm256_mul_ps(rinv23,rinv23);
 863             rinvsq31         = _mm256_mul_ps(rinv31,rinv31);
 864             rinvsq32         = _mm256_mul_ps(rinv32,rinv32);
 865             rinvsq33         = _mm256_mul_ps(rinv33,rinv33);
 866
 867             fjx1             = _mm256_setzero_ps();
 868             fjy1             = _mm256_setzero_ps();
 869             fjz1             = _mm256_setzero_ps();
 870             fjx2             = _mm256_setzero_ps();
 871             fjy2             = _mm256_setzero_ps();
 872             fjz2             = _mm256_setzero_ps();
 873             fjx3             = _mm256_setzero_ps();
 874             fjy3             = _mm256_setzero_ps();
 875             fjz3             = _mm256_setzero_ps();
 876
 877             /**************************
 878              * CALCULATE INTERACTIONS *
 879              **************************/
 880
 881             if (gmx_mm256_any_lt(rsq11,rcutoff2))
 882             {
 883
 884             r11              = _mm256_mul_ps(rsq11,rinv11);
 885             r11              = _mm256_andnot_ps(dummy_mask,r11);
 886
 887             /* EWALD ELECTROSTATICS */
 888
 889             /* Analytical PME correction */
 890             zeta2            = _mm256_mul_ps(beta2,rsq11);
 891             rinv3            = _mm256_mul_ps(rinvsq11,rinv11);
 892             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
 893             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
 894             felec            = _mm256_mul_ps(qq11,felec);
 895             pmecorrV         = gmx_mm256_pmecorrV_ps(zeta2);
 896             pmecorrV         = _mm256_mul_ps(pmecorrV,beta);
 897             velec            = _mm256_sub_ps(_mm256_sub_ps(rinv11,sh_ewald),pmecorrV);
 898             velec            = _mm256_mul_ps(qq11,velec);
 899
 900             cutoff_mask      = _mm256_cmp_ps(rsq11,rcutoff2,_CMP_LT_OQ);
 901
 902             /* Update potential sum for this i atom from the interaction with this j atom. */
 903             velec            = _mm256_and_ps(velec,cutoff_mask);
 904             velec            = _mm256_andnot_ps(dummy_mask,velec);
 905             velecsum         = _mm256_add_ps(velecsum,velec);
 906
 907             fscal            = felec;
 908
 909             fscal            = _mm256_and_ps(fscal,cutoff_mask);
 910
 911             fscal            = _mm256_andnot_ps(dummy_mask,fscal);
 912
 913             /* Calculate temporary vectorial force */
 914             tx               = _mm256_mul_ps(fscal,dx11);
 915             ty               = _mm256_mul_ps(fscal,dy11);
 916             tz               = _mm256_mul_ps(fscal,dz11);
 917
 918             /* Update vectorial force */
 919             fix1             = _mm256_add_ps(fix1,tx);
 920             fiy1             = _mm256_add_ps(fiy1,ty);
 921             fiz1             = _mm256_add_ps(fiz1,tz);
 922
 923             fjx1             = _mm256_add_ps(fjx1,tx);
 924             fjy1             = _mm256_add_ps(fjy1,ty);
 925             fjz1             = _mm256_add_ps(fjz1,tz);
 926
 927             }
 928
 929             /**************************
 930              * CALCULATE INTERACTIONS *
 931              **************************/
 932
 933             if (gmx_mm256_any_lt(rsq12,rcutoff2))
 934             {
 935
 936             r12              = _mm256_mul_ps(rsq12,rinv12);
 937             r12              = _mm256_andnot_ps(dummy_mask,r12);
 938
 939             /* EWALD ELECTROSTATICS */
 940
 941             /* Analytical PME correction */
 942             zeta2            = _mm256_mul_ps(beta2,rsq12);
 943             rinv3            = _mm256_mul_ps(rinvsq12,rinv12);
 944             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
 945             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
 946             felec            = _mm256_mul_ps(qq12,felec);
 947             pmecorrV         = gmx_mm256_pmecorrV_ps(zeta2);
 948             pmecorrV         = _mm256_mul_ps(pmecorrV,beta);
 949             velec            = _mm256_sub_ps(_mm256_sub_ps(rinv12,sh_ewald),pmecorrV);
 950             velec            = _mm256_mul_ps(qq12,velec);
 951
 952             cutoff_mask      = _mm256_cmp_ps(rsq12,rcutoff2,_CMP_LT_OQ);
 953
 954             /* Update potential sum for this i atom from the interaction with this j atom. */
 955             velec            = _mm256_and_ps(velec,cutoff_mask);
 956             velec            = _mm256_andnot_ps(dummy_mask,velec);
 957             velecsum         = _mm256_add_ps(velecsum,velec);
 958
 959             fscal            = felec;
 960
 961             fscal            = _mm256_and_ps(fscal,cutoff_mask);
 962
 963             fscal            = _mm256_andnot_ps(dummy_mask,fscal);
 964
 965             /* Calculate temporary vectorial force */
 966             tx               = _mm256_mul_ps(fscal,dx12);
 967             ty               = _mm256_mul_ps(fscal,dy12);
 968             tz               = _mm256_mul_ps(fscal,dz12);
 969
 970             /* Update vectorial force */
 971             fix1             = _mm256_add_ps(fix1,tx);
 972             fiy1             = _mm256_add_ps(fiy1,ty);
 973             fiz1             = _mm256_add_ps(fiz1,tz);
 974
 975             fjx2             = _mm256_add_ps(fjx2,tx);
 976             fjy2             = _mm256_add_ps(fjy2,ty);
 977             fjz2             = _mm256_add_ps(fjz2,tz);
 978
 979             }
 980
 981             /**************************
 982              * CALCULATE INTERACTIONS *
 983              **************************/
 984
 985             if (gmx_mm256_any_lt(rsq13,rcutoff2))
 986             {
 987
 988             r13              = _mm256_mul_ps(rsq13,rinv13);
 989             r13              = _mm256_andnot_ps(dummy_mask,r13);
 990
 991             /* EWALD ELECTROSTATICS */
 992
 993             /* Analytical PME correction */
 994             zeta2            = _mm256_mul_ps(beta2,rsq13);
 995             rinv3            = _mm256_mul_ps(rinvsq13,rinv13);
 996             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
 997             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
 998             felec            = _mm256_mul_ps(qq13,felec);
 999             pmecorrV         = gmx_mm256_pmecorrV_ps(zeta2);
1000             pmecorrV         = _mm256_mul_ps(pmecorrV,beta);
1001             velec            = _mm256_sub_ps(_mm256_sub_ps(rinv13,sh_ewald),pmecorrV);
1002             velec            = _mm256_mul_ps(qq13,velec);
1003
1004             cutoff_mask      = _mm256_cmp_ps(rsq13,rcutoff2,_CMP_LT_OQ);
1005
1006             /* Update potential sum for this i atom from the interaction with this j atom. */
1007             velec            = _mm256_and_ps(velec,cutoff_mask);
1008             velec            = _mm256_andnot_ps(dummy_mask,velec);
1009             velecsum         = _mm256_add_ps(velecsum,velec);
1010
1011             fscal            = felec;
1012
1013             fscal            = _mm256_and_ps(fscal,cutoff_mask);
1014
1015             fscal            = _mm256_andnot_ps(dummy_mask,fscal);
1016
1017             /* Calculate temporary vectorial force */
1018             tx               = _mm256_mul_ps(fscal,dx13);
1019             ty               = _mm256_mul_ps(fscal,dy13);
1020             tz               = _mm256_mul_ps(fscal,dz13);
1021
1022             /* Update vectorial force */
1023             fix1             = _mm256_add_ps(fix1,tx);
1024             fiy1             = _mm256_add_ps(fiy1,ty);
1025             fiz1             = _mm256_add_ps(fiz1,tz);
1026
1027             fjx3             = _mm256_add_ps(fjx3,tx);
1028             fjy3             = _mm256_add_ps(fjy3,ty);
1029             fjz3             = _mm256_add_ps(fjz3,tz);
1030
1031             }
1032
1033             /**************************
1034              * CALCULATE INTERACTIONS *
1035              **************************/
1036
1037             if (gmx_mm256_any_lt(rsq21,rcutoff2))
1038             {
1039
1040             r21              = _mm256_mul_ps(rsq21,rinv21);
1041             r21              = _mm256_andnot_ps(dummy_mask,r21);
1042
1043             /* EWALD ELECTROSTATICS */
1044
1045             /* Analytical PME correction */
1046             zeta2            = _mm256_mul_ps(beta2,rsq21);
1047             rinv3            = _mm256_mul_ps(rinvsq21,rinv21);
1048             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
1049             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
1050             felec            = _mm256_mul_ps(qq21,felec);
1051             pmecorrV         = gmx_mm256_pmecorrV_ps(zeta2);
1052             pmecorrV         = _mm256_mul_ps(pmecorrV,beta);
1053             velec            = _mm256_sub_ps(_mm256_sub_ps(rinv21,sh_ewald),pmecorrV);
1054             velec            = _mm256_mul_ps(qq21,velec);
1055
1056             cutoff_mask      = _mm256_cmp_ps(rsq21,rcutoff2,_CMP_LT_OQ);
1057
1058             /* Update potential sum for this i atom from the interaction with this j atom. */
1059             velec            = _mm256_and_ps(velec,cutoff_mask);
1060             velec            = _mm256_andnot_ps(dummy_mask,velec);
1061             velecsum         = _mm256_add_ps(velecsum,velec);
1062
1063             fscal            = felec;
1064
1065             fscal            = _mm256_and_ps(fscal,cutoff_mask);
1066
1067             fscal            = _mm256_andnot_ps(dummy_mask,fscal);
1068
1069             /* Calculate temporary vectorial force */
1070             tx               = _mm256_mul_ps(fscal,dx21);
1071             ty               = _mm256_mul_ps(fscal,dy21);
1072             tz               = _mm256_mul_ps(fscal,dz21);
1073
1074             /* Update vectorial force */
1075             fix2             = _mm256_add_ps(fix2,tx);
1076             fiy2             = _mm256_add_ps(fiy2,ty);
1077             fiz2             = _mm256_add_ps(fiz2,tz);
1078
1079             fjx1             = _mm256_add_ps(fjx1,tx);
1080             fjy1             = _mm256_add_ps(fjy1,ty);
1081             fjz1             = _mm256_add_ps(fjz1,tz);
1082
1083             }
1084
1085             /**************************
1086              * CALCULATE INTERACTIONS *
1087              **************************/
1088
1089             if (gmx_mm256_any_lt(rsq22,rcutoff2))
1090             {
1091
1092             r22              = _mm256_mul_ps(rsq22,rinv22);
1093             r22              = _mm256_andnot_ps(dummy_mask,r22);
1094
1095             /* EWALD ELECTROSTATICS */
1096
1097             /* Analytical PME correction */
1098             zeta2            = _mm256_mul_ps(beta2,rsq22);
1099             rinv3            = _mm256_mul_ps(rinvsq22,rinv22);
1100             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
1101             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
1102             felec            = _mm256_mul_ps(qq22,felec);
1103             pmecorrV         = gmx_mm256_pmecorrV_ps(zeta2);
1104             pmecorrV         = _mm256_mul_ps(pmecorrV,beta);
1105             velec            = _mm256_sub_ps(_mm256_sub_ps(rinv22,sh_ewald),pmecorrV);
1106             velec            = _mm256_mul_ps(qq22,velec);
1107
1108             cutoff_mask      = _mm256_cmp_ps(rsq22,rcutoff2,_CMP_LT_OQ);
1109
1110             /* Update potential sum for this i atom from the interaction with this j atom. */
1111             velec            = _mm256_and_ps(velec,cutoff_mask);
1112             velec            = _mm256_andnot_ps(dummy_mask,velec);
1113             velecsum         = _mm256_add_ps(velecsum,velec);
1114
1115             fscal            = felec;
1116
1117             fscal            = _mm256_and_ps(fscal,cutoff_mask);
1118
1119             fscal            = _mm256_andnot_ps(dummy_mask,fscal);
1120
1121             /* Calculate temporary vectorial force */
1122             tx               = _mm256_mul_ps(fscal,dx22);
1123             ty               = _mm256_mul_ps(fscal,dy22);
1124             tz               = _mm256_mul_ps(fscal,dz22);
1125
1126             /* Update vectorial force */
1127             fix2             = _mm256_add_ps(fix2,tx);
1128             fiy2             = _mm256_add_ps(fiy2,ty);
1129             fiz2             = _mm256_add_ps(fiz2,tz);
1130
1131             fjx2             = _mm256_add_ps(fjx2,tx);
1132             fjy2             = _mm256_add_ps(fjy2,ty);
1133             fjz2             = _mm256_add_ps(fjz2,tz);
1134
1135             }
1136
1137             /**************************
1138              * CALCULATE INTERACTIONS *
1139              **************************/
1140
1141             if (gmx_mm256_any_lt(rsq23,rcutoff2))
1142             {
1143
1144             r23              = _mm256_mul_ps(rsq23,rinv23);
1145             r23              = _mm256_andnot_ps(dummy_mask,r23);
1146
1147             /* EWALD ELECTROSTATICS */
1148
1149             /* Analytical PME correction */
1150             zeta2            = _mm256_mul_ps(beta2,rsq23);
1151             rinv3            = _mm256_mul_ps(rinvsq23,rinv23);
1152             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
1153             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
1154             felec            = _mm256_mul_ps(qq23,felec);
1155             pmecorrV         = gmx_mm256_pmecorrV_ps(zeta2);
1156             pmecorrV         = _mm256_mul_ps(pmecorrV,beta);
1157             velec            = _mm256_sub_ps(_mm256_sub_ps(rinv23,sh_ewald),pmecorrV);
1158             velec            = _mm256_mul_ps(qq23,velec);
1159
1160             cutoff_mask      = _mm256_cmp_ps(rsq23,rcutoff2,_CMP_LT_OQ);
1161
1162             /* Update potential sum for this i atom from the interaction with this j atom. */
1163             velec            = _mm256_and_ps(velec,cutoff_mask);
1164             velec            = _mm256_andnot_ps(dummy_mask,velec);
1165             velecsum         = _mm256_add_ps(velecsum,velec);
1166
1167             fscal            = felec;
1168
1169             fscal            = _mm256_and_ps(fscal,cutoff_mask);
1170
1171             fscal            = _mm256_andnot_ps(dummy_mask,fscal);
1172
1173             /* Calculate temporary vectorial force */
1174             tx               = _mm256_mul_ps(fscal,dx23);
1175             ty               = _mm256_mul_ps(fscal,dy23);
1176             tz               = _mm256_mul_ps(fscal,dz23);
1177
1178             /* Update vectorial force */
1179             fix2             = _mm256_add_ps(fix2,tx);
1180             fiy2             = _mm256_add_ps(fiy2,ty);
1181             fiz2             = _mm256_add_ps(fiz2,tz);
1182
1183             fjx3             = _mm256_add_ps(fjx3,tx);
1184             fjy3             = _mm256_add_ps(fjy3,ty);
1185             fjz3             = _mm256_add_ps(fjz3,tz);
1186
1187             }
1188
1189             /**************************
1190              * CALCULATE INTERACTIONS *
1191              **************************/
1192
1193             if (gmx_mm256_any_lt(rsq31,rcutoff2))
1194             {
1195
1196             r31              = _mm256_mul_ps(rsq31,rinv31);
1197             r31              = _mm256_andnot_ps(dummy_mask,r31);
1198
1199             /* EWALD ELECTROSTATICS */
1200
1201             /* Analytical PME correction */
1202             zeta2            = _mm256_mul_ps(beta2,rsq31);
1203             rinv3            = _mm256_mul_ps(rinvsq31,rinv31);
1204             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
1205             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
1206             felec            = _mm256_mul_ps(qq31,felec);
1207             pmecorrV         = gmx_mm256_pmecorrV_ps(zeta2);
1208             pmecorrV         = _mm256_mul_ps(pmecorrV,beta);
1209             velec            = _mm256_sub_ps(_mm256_sub_ps(rinv31,sh_ewald),pmecorrV);
1210             velec            = _mm256_mul_ps(qq31,velec);
1211
1212             cutoff_mask      = _mm256_cmp_ps(rsq31,rcutoff2,_CMP_LT_OQ);
1213
1214             /* Update potential sum for this i atom from the interaction with this j atom. */
1215             velec            = _mm256_and_ps(velec,cutoff_mask);
1216             velec            = _mm256_andnot_ps(dummy_mask,velec);
1217             velecsum         = _mm256_add_ps(velecsum,velec);
1218
1219             fscal            = felec;
1220
1221             fscal            = _mm256_and_ps(fscal,cutoff_mask);
1222
1223             fscal            = _mm256_andnot_ps(dummy_mask,fscal);
1224
1225             /* Calculate temporary vectorial force */
1226             tx               = _mm256_mul_ps(fscal,dx31);
1227             ty               = _mm256_mul_ps(fscal,dy31);
1228             tz               = _mm256_mul_ps(fscal,dz31);
1229
1230             /* Update vectorial force */
1231             fix3             = _mm256_add_ps(fix3,tx);
1232             fiy3             = _mm256_add_ps(fiy3,ty);
1233             fiz3             = _mm256_add_ps(fiz3,tz);
1234
1235             fjx1             = _mm256_add_ps(fjx1,tx);
1236             fjy1             = _mm256_add_ps(fjy1,ty);
1237             fjz1             = _mm256_add_ps(fjz1,tz);
1238
1239             }
1240
1241             /**************************
1242              * CALCULATE INTERACTIONS *
1243              **************************/
1244
1245             if (gmx_mm256_any_lt(rsq32,rcutoff2))
1246             {
1247
1248             r32              = _mm256_mul_ps(rsq32,rinv32);
1249             r32              = _mm256_andnot_ps(dummy_mask,r32);
1250
1251             /* EWALD ELECTROSTATICS */
1252
1253             /* Analytical PME correction */
1254             zeta2            = _mm256_mul_ps(beta2,rsq32);
1255             rinv3            = _mm256_mul_ps(rinvsq32,rinv32);
1256             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
1257             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
1258             felec            = _mm256_mul_ps(qq32,felec);
1259             pmecorrV         = gmx_mm256_pmecorrV_ps(zeta2);
1260             pmecorrV         = _mm256_mul_ps(pmecorrV,beta);
1261             velec            = _mm256_sub_ps(_mm256_sub_ps(rinv32,sh_ewald),pmecorrV);
1262             velec            = _mm256_mul_ps(qq32,velec);
1263
1264             cutoff_mask      = _mm256_cmp_ps(rsq32,rcutoff2,_CMP_LT_OQ);
1265
1266             /* Update potential sum for this i atom from the interaction with this j atom. */
1267             velec            = _mm256_and_ps(velec,cutoff_mask);
1268             velec            = _mm256_andnot_ps(dummy_mask,velec);
1269             velecsum         = _mm256_add_ps(velecsum,velec);
1270
1271             fscal            = felec;
1272
1273             fscal            = _mm256_and_ps(fscal,cutoff_mask);
1274
1275             fscal            = _mm256_andnot_ps(dummy_mask,fscal);
1276
1277             /* Calculate temporary vectorial force */
1278             tx               = _mm256_mul_ps(fscal,dx32);
1279             ty               = _mm256_mul_ps(fscal,dy32);
1280             tz               = _mm256_mul_ps(fscal,dz32);
1281
1282             /* Update vectorial force */
1283             fix3             = _mm256_add_ps(fix3,tx);
1284             fiy3             = _mm256_add_ps(fiy3,ty);
1285             fiz3             = _mm256_add_ps(fiz3,tz);
1286
1287             fjx2             = _mm256_add_ps(fjx2,tx);
1288             fjy2             = _mm256_add_ps(fjy2,ty);
1289             fjz2             = _mm256_add_ps(fjz2,tz);
1290
1291             }
1292
1293             /**************************
1294              * CALCULATE INTERACTIONS *
1295              **************************/
1296
1297             if (gmx_mm256_any_lt(rsq33,rcutoff2))
1298             {
1299
1300             r33              = _mm256_mul_ps(rsq33,rinv33);
1301             r33              = _mm256_andnot_ps(dummy_mask,r33);
1302
1303             /* EWALD ELECTROSTATICS */
1304
1305             /* Analytical PME correction */
1306             zeta2            = _mm256_mul_ps(beta2,rsq33);
1307             rinv3            = _mm256_mul_ps(rinvsq33,rinv33);
1308             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
1309             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
1310             felec            = _mm256_mul_ps(qq33,felec);
1311             pmecorrV         = gmx_mm256_pmecorrV_ps(zeta2);
1312             pmecorrV         = _mm256_mul_ps(pmecorrV,beta);
1313             velec            = _mm256_sub_ps(_mm256_sub_ps(rinv33,sh_ewald),pmecorrV);
1314             velec            = _mm256_mul_ps(qq33,velec);
1315
1316             cutoff_mask      = _mm256_cmp_ps(rsq33,rcutoff2,_CMP_LT_OQ);
1317
1318             /* Update potential sum for this i atom from the interaction with this j atom. */
1319             velec            = _mm256_and_ps(velec,cutoff_mask);
1320             velec            = _mm256_andnot_ps(dummy_mask,velec);
1321             velecsum         = _mm256_add_ps(velecsum,velec);
1322
1323             fscal            = felec;
1324
1325             fscal            = _mm256_and_ps(fscal,cutoff_mask);
1326
1327             fscal            = _mm256_andnot_ps(dummy_mask,fscal);
1328
1329             /* Calculate temporary vectorial force */
1330             tx               = _mm256_mul_ps(fscal,dx33);
1331             ty               = _mm256_mul_ps(fscal,dy33);
1332             tz               = _mm256_mul_ps(fscal,dz33);
1333
1334             /* Update vectorial force */
1335             fix3             = _mm256_add_ps(fix3,tx);
1336             fiy3             = _mm256_add_ps(fiy3,ty);
1337             fiz3             = _mm256_add_ps(fiz3,tz);
1338
1339             fjx3             = _mm256_add_ps(fjx3,tx);
1340             fjy3             = _mm256_add_ps(fjy3,ty);
1341             fjz3             = _mm256_add_ps(fjz3,tz);
1342
1343             }
1344
1345             fjptrA             = (jnrlistA>=0) ? f+j_coord_offsetA : scratch;
1346             fjptrB             = (jnrlistB>=0) ? f+j_coord_offsetB : scratch;
1347             fjptrC             = (jnrlistC>=0) ? f+j_coord_offsetC : scratch;
1348             fjptrD             = (jnrlistD>=0) ? f+j_coord_offsetD : scratch;
1349             fjptrE             = (jnrlistE>=0) ? f+j_coord_offsetE : scratch;
1350             fjptrF             = (jnrlistF>=0) ? f+j_coord_offsetF : scratch;
1351             fjptrG             = (jnrlistG>=0) ? f+j_coord_offsetG : scratch;
1352             fjptrH             = (jnrlistH>=0) ? f+j_coord_offsetH : scratch;
1353
1354             gmx_mm256_decrement_3rvec_8ptr_swizzle_ps(fjptrA+DIM,fjptrB+DIM,fjptrC+DIM,fjptrD+DIM,
1355                                                       fjptrE+DIM,fjptrF+DIM,fjptrG+DIM,fjptrH+DIM,
1356                                                       fjx1,fjy1,fjz1,fjx2,fjy2,fjz2,fjx3,fjy3,fjz3);
1357
1358             /* Inner loop uses 990 flops */
1359         }
1360
1361         /* End of innermost loop */
1362
1363         gmx_mm256_update_iforce_3atom_swizzle_ps(fix1,fiy1,fiz1,fix2,fiy2,fiz2,fix3,fiy3,fiz3,
1364                                                  f+i_coord_offset+DIM,fshift+i_shift_offset);
1365
1366         ggid                        = gid[iidx];
1367         /* Update potential energies */
1368         gmx_mm256_update_1pot_ps(velecsum,kernel_data->energygrp_elec+ggid);
1369
1370         /* Increment number of inner iterations */
1371         inneriter                  += j_index_end - j_index_start;
1372
1373         /* Outer loop uses 19 flops */
1374     }
1375
1376     /* Increment number of outer iterations */
1377     outeriter        += nri;
1378
1379     /* Update outer/inner flops */
1380
1381     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_W4W4_VF,outeriter*19 + inneriter*990);
1382 }
1383 /*
1384  * Gromacs nonbonded kernel:   nb_kernel_ElecEwSh_VdwNone_GeomW4W4_F_avx_256_single
1385  * Electrostatics interaction: Ewald
1386  * VdW interaction:            None
1387  * Geometry:                   Water4-Water4
1388  * Calculate force/pot:        Force
1389  */
1390 void
1391 nb_kernel_ElecEwSh_VdwNone_GeomW4W4_F_avx_256_single
1392                     (t_nblist                    * gmx_restrict       nlist,
1393                      rvec                        * gmx_restrict          xx,
1394                      rvec                        * gmx_restrict          ff,
1395                      t_forcerec                  * gmx_restrict          fr,
1396                      t_mdatoms                   * gmx_restrict     mdatoms,
1397                      nb_kernel_data_t gmx_unused * gmx_restrict kernel_data,
1398                      t_nrnb                      * gmx_restrict        nrnb)
1399 {
1400     /* Suffixes 0,1,2,3 refer to particle indices for waters in the inner or outer loop, or
1401      * just 0 for non-waters.
1402      * Suffixes A,B,C,D,E,F,G,H refer to j loop unrolling done with AVX, e.g. for the eight different
1403      * jnr indices corresponding to data put in the four positions in the SIMD register.
1404      */
1405     int              i_shift_offset,i_coord_offset,outeriter,inneriter;
1406     int              j_index_start,j_index_end,jidx,nri,inr,ggid,iidx;
1407     int              jnrA,jnrB,jnrC,jnrD;
1408     int              jnrE,jnrF,jnrG,jnrH;
1409     int              jnrlistA,jnrlistB,jnrlistC,jnrlistD;
1410     int              jnrlistE,jnrlistF,jnrlistG,jnrlistH;
1411     int              j_coord_offsetA,j_coord_offsetB,j_coord_offsetC,j_coord_offsetD;
1412     int              j_coord_offsetE,j_coord_offsetF,j_coord_offsetG,j_coord_offsetH;
1413     int              *iinr,*jindex,*jjnr,*shiftidx,*gid;
1414     real             rcutoff_scalar;
1415     real             *shiftvec,*fshift,*x,*f;
1416     real             *fjptrA,*fjptrB,*fjptrC,*fjptrD,*fjptrE,*fjptrF,*fjptrG,*fjptrH;
1417     real             scratch[4*DIM];
1418     __m256           tx,ty,tz,fscal,rcutoff,rcutoff2,jidxall;
1419     real *           vdwioffsetptr1;
1420     __m256           ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
1421     real *           vdwioffsetptr2;
1422     __m256           ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
1423     real *           vdwioffsetptr3;
1424     __m256           ix3,iy3,iz3,fix3,fiy3,fiz3,iq3,isai3;
1425     int              vdwjidx1A,vdwjidx1B,vdwjidx1C,vdwjidx1D,vdwjidx1E,vdwjidx1F,vdwjidx1G,vdwjidx1H;
1426     __m256           jx1,jy1,jz1,fjx1,fjy1,fjz1,jq1,isaj1;
1427     int              vdwjidx2A,vdwjidx2B,vdwjidx2C,vdwjidx2D,vdwjidx2E,vdwjidx2F,vdwjidx2G,vdwjidx2H;
1428     __m256           jx2,jy2,jz2,fjx2,fjy2,fjz2,jq2,isaj2;
1429     int              vdwjidx3A,vdwjidx3B,vdwjidx3C,vdwjidx3D,vdwjidx3E,vdwjidx3F,vdwjidx3G,vdwjidx3H;
1430     __m256           jx3,jy3,jz3,fjx3,fjy3,fjz3,jq3,isaj3;
1431     __m256           dx11,dy11,dz11,rsq11,rinv11,rinvsq11,r11,qq11,c6_11,c12_11;
1432     __m256           dx12,dy12,dz12,rsq12,rinv12,rinvsq12,r12,qq12,c6_12,c12_12;
1433     __m256           dx13,dy13,dz13,rsq13,rinv13,rinvsq13,r13,qq13,c6_13,c12_13;
1434     __m256           dx21,dy21,dz21,rsq21,rinv21,rinvsq21,r21,qq21,c6_21,c12_21;
1435     __m256           dx22,dy22,dz22,rsq22,rinv22,rinvsq22,r22,qq22,c6_22,c12_22;
1436     __m256           dx23,dy23,dz23,rsq23,rinv23,rinvsq23,r23,qq23,c6_23,c12_23;
1437     __m256           dx31,dy31,dz31,rsq31,rinv31,rinvsq31,r31,qq31,c6_31,c12_31;
1438     __m256           dx32,dy32,dz32,rsq32,rinv32,rinvsq32,r32,qq32,c6_32,c12_32;
1439     __m256           dx33,dy33,dz33,rsq33,rinv33,rinvsq33,r33,qq33,c6_33,c12_33;
1440     __m256           velec,felec,velecsum,facel,crf,krf,krf2;
1441     real             *charge;
1442     __m256i          ewitab;
1443     __m128i          ewitab_lo,ewitab_hi;
1444     __m256           ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace,ewtabF,ewtabFn,ewtabD,ewtabV;
1445     __m256           beta,beta2,beta3,zeta2,pmecorrF,pmecorrV,rinv3;
1446     real             *ewtab;
1447     __m256           dummy_mask,cutoff_mask;
1448     __m256           signbit = _mm256_castsi256_ps( _mm256_set1_epi32(0x80000000) );
1449     __m256           one     = _mm256_set1_ps(1.0);
1450     __m256           two     = _mm256_set1_ps(2.0);
1451     x                = xx[0];
1452     f                = ff[0];
1453
1454     nri              = nlist->nri;
1455     iinr             = nlist->iinr;
1456     jindex           = nlist->jindex;
1457     jjnr             = nlist->jjnr;
1458     shiftidx         = nlist->shift;
1459     gid              = nlist->gid;
1460     shiftvec         = fr->shift_vec[0];
1461     fshift           = fr->fshift[0];
1462     facel            = _mm256_set1_ps(fr->epsfac);
1463     charge           = mdatoms->chargeA;
1464
1465     sh_ewald         = _mm256_set1_ps(fr->ic->sh_ewald);
1466     beta             = _mm256_set1_ps(fr->ic->ewaldcoeff_q);
1467     beta2            = _mm256_mul_ps(beta,beta);
1468     beta3            = _mm256_mul_ps(beta,beta2);
1469
1470     ewtab            = fr->ic->tabq_coul_F;
1471     ewtabscale       = _mm256_set1_ps(fr->ic->tabq_scale);
1472     ewtabhalfspace   = _mm256_set1_ps(0.5/fr->ic->tabq_scale);
1473
1474     /* Setup water-specific parameters */
1475     inr              = nlist->iinr[0];
1476     iq1              = _mm256_mul_ps(facel,_mm256_set1_ps(charge[inr+1]));
1477     iq2              = _mm256_mul_ps(facel,_mm256_set1_ps(charge[inr+2]));
1478     iq3              = _mm256_mul_ps(facel,_mm256_set1_ps(charge[inr+3]));
1479
1480     jq1              = _mm256_set1_ps(charge[inr+1]);
1481     jq2              = _mm256_set1_ps(charge[inr+2]);
1482     jq3              = _mm256_set1_ps(charge[inr+3]);
1483     qq11             = _mm256_mul_ps(iq1,jq1);
1484     qq12             = _mm256_mul_ps(iq1,jq2);
1485     qq13             = _mm256_mul_ps(iq1,jq3);
1486     qq21             = _mm256_mul_ps(iq2,jq1);
1487     qq22             = _mm256_mul_ps(iq2,jq2);
1488     qq23             = _mm256_mul_ps(iq2,jq3);
1489     qq31             = _mm256_mul_ps(iq3,jq1);
1490     qq32             = _mm256_mul_ps(iq3,jq2);
1491     qq33             = _mm256_mul_ps(iq3,jq3);
1492
1493     /* When we use explicit cutoffs the value must be identical for elec and VdW, so use elec as an arbitrary choice */
1494     rcutoff_scalar   = fr->rcoulomb;
1495     rcutoff          = _mm256_set1_ps(rcutoff_scalar);
1496     rcutoff2         = _mm256_mul_ps(rcutoff,rcutoff);
1497
1498     /* Avoid stupid compiler warnings */
1499     jnrA = jnrB = jnrC = jnrD = jnrE = jnrF = jnrG = jnrH = 0;
1500     j_coord_offsetA = 0;
1501     j_coord_offsetB = 0;
1502     j_coord_offsetC = 0;
1503     j_coord_offsetD = 0;
1504     j_coord_offsetE = 0;
1505     j_coord_offsetF = 0;
1506     j_coord_offsetG = 0;
1507     j_coord_offsetH = 0;
1508
1509     outeriter        = 0;
1510     inneriter        = 0;
1511
1512     for(iidx=0;iidx<4*DIM;iidx++)
1513     {
1514         scratch[iidx] = 0.0;
1515     }
1516
1517     /* Start outer loop over neighborlists */
1518     for(iidx=0; iidx<nri; iidx++)
1519     {
1520         /* Load shift vector for this list */
1521         i_shift_offset   = DIM*shiftidx[iidx];
1522
1523         /* Load limits for loop over neighbors */
1524         j_index_start    = jindex[iidx];
1525         j_index_end      = jindex[iidx+1];
1526
1527         /* Get outer coordinate index */
1528         inr              = iinr[iidx];
1529         i_coord_offset   = DIM*inr;
1530
1531         /* Load i particle coords and add shift vector */
1532         gmx_mm256_load_shift_and_3rvec_broadcast_ps(shiftvec+i_shift_offset,x+i_coord_offset+DIM,
1533                                                     &ix1,&iy1,&iz1,&ix2,&iy2,&iz2,&ix3,&iy3,&iz3);
1534
1535         fix1             = _mm256_setzero_ps();
1536         fiy1             = _mm256_setzero_ps();
1537         fiz1             = _mm256_setzero_ps();
1538         fix2             = _mm256_setzero_ps();
1539         fiy2             = _mm256_setzero_ps();
1540         fiz2             = _mm256_setzero_ps();
1541         fix3             = _mm256_setzero_ps();
1542         fiy3             = _mm256_setzero_ps();
1543         fiz3             = _mm256_setzero_ps();
1544
1545         /* Start inner kernel loop */
1546         for(jidx=j_index_start; jidx<j_index_end && jjnr[jidx+7]>=0; jidx+=8)
1547         {
1548
1549             /* Get j neighbor index, and coordinate index */
1550             jnrA             = jjnr[jidx];
1551             jnrB             = jjnr[jidx+1];
1552             jnrC             = jjnr[jidx+2];
1553             jnrD             = jjnr[jidx+3];
1554             jnrE             = jjnr[jidx+4];
1555             jnrF             = jjnr[jidx+5];
1556             jnrG             = jjnr[jidx+6];
1557             jnrH             = jjnr[jidx+7];
1558             j_coord_offsetA  = DIM*jnrA;
1559             j_coord_offsetB  = DIM*jnrB;
1560             j_coord_offsetC  = DIM*jnrC;
1561             j_coord_offsetD  = DIM*jnrD;
1562             j_coord_offsetE  = DIM*jnrE;
1563             j_coord_offsetF  = DIM*jnrF;
1564             j_coord_offsetG  = DIM*jnrG;
1565             j_coord_offsetH  = DIM*jnrH;
1566
1567             /* load j atom coordinates */
1568             gmx_mm256_load_3rvec_8ptr_swizzle_ps(x+j_coord_offsetA+DIM,x+j_coord_offsetB+DIM,
1569                                                  x+j_coord_offsetC+DIM,x+j_coord_offsetD+DIM,
1570                                                  x+j_coord_offsetE+DIM,x+j_coord_offsetF+DIM,
1571                                                  x+j_coord_offsetG+DIM,x+j_coord_offsetH+DIM,
1572                                                  &jx1,&jy1,&jz1,&jx2,&jy2,&jz2,&jx3,&jy3,&jz3);
1573
1574             /* Calculate displacement vector */
1575             dx11             = _mm256_sub_ps(ix1,jx1);
1576             dy11             = _mm256_sub_ps(iy1,jy1);
1577             dz11             = _mm256_sub_ps(iz1,jz1);
1578             dx12             = _mm256_sub_ps(ix1,jx2);
1579             dy12             = _mm256_sub_ps(iy1,jy2);
1580             dz12             = _mm256_sub_ps(iz1,jz2);
1581             dx13             = _mm256_sub_ps(ix1,jx3);
1582             dy13             = _mm256_sub_ps(iy1,jy3);
1583             dz13             = _mm256_sub_ps(iz1,jz3);
1584             dx21             = _mm256_sub_ps(ix2,jx1);
1585             dy21             = _mm256_sub_ps(iy2,jy1);
1586             dz21             = _mm256_sub_ps(iz2,jz1);
1587             dx22             = _mm256_sub_ps(ix2,jx2);
1588             dy22             = _mm256_sub_ps(iy2,jy2);
1589             dz22             = _mm256_sub_ps(iz2,jz2);
1590             dx23             = _mm256_sub_ps(ix2,jx3);
1591             dy23             = _mm256_sub_ps(iy2,jy3);
1592             dz23             = _mm256_sub_ps(iz2,jz3);
1593             dx31             = _mm256_sub_ps(ix3,jx1);
1594             dy31             = _mm256_sub_ps(iy3,jy1);
1595             dz31             = _mm256_sub_ps(iz3,jz1);
1596             dx32             = _mm256_sub_ps(ix3,jx2);
1597             dy32             = _mm256_sub_ps(iy3,jy2);
1598             dz32             = _mm256_sub_ps(iz3,jz2);
1599             dx33             = _mm256_sub_ps(ix3,jx3);
1600             dy33             = _mm256_sub_ps(iy3,jy3);
1601             dz33             = _mm256_sub_ps(iz3,jz3);
1602
1603             /* Calculate squared distance and things based on it */
1604             rsq11            = gmx_mm256_calc_rsq_ps(dx11,dy11,dz11);
1605             rsq12            = gmx_mm256_calc_rsq_ps(dx12,dy12,dz12);
1606             rsq13            = gmx_mm256_calc_rsq_ps(dx13,dy13,dz13);
1607             rsq21            = gmx_mm256_calc_rsq_ps(dx21,dy21,dz21);
1608             rsq22            = gmx_mm256_calc_rsq_ps(dx22,dy22,dz22);
1609             rsq23            = gmx_mm256_calc_rsq_ps(dx23,dy23,dz23);
1610             rsq31            = gmx_mm256_calc_rsq_ps(dx31,dy31,dz31);
1611             rsq32            = gmx_mm256_calc_rsq_ps(dx32,dy32,dz32);
1612             rsq33            = gmx_mm256_calc_rsq_ps(dx33,dy33,dz33);
1613
1614             rinv11           = gmx_mm256_invsqrt_ps(rsq11);
1615             rinv12           = gmx_mm256_invsqrt_ps(rsq12);
1616             rinv13           = gmx_mm256_invsqrt_ps(rsq13);
1617             rinv21           = gmx_mm256_invsqrt_ps(rsq21);
1618             rinv22           = gmx_mm256_invsqrt_ps(rsq22);
1619             rinv23           = gmx_mm256_invsqrt_ps(rsq23);
1620             rinv31           = gmx_mm256_invsqrt_ps(rsq31);
1621             rinv32           = gmx_mm256_invsqrt_ps(rsq32);
1622             rinv33           = gmx_mm256_invsqrt_ps(rsq33);
1623
1624             rinvsq11         = _mm256_mul_ps(rinv11,rinv11);
1625             rinvsq12         = _mm256_mul_ps(rinv12,rinv12);
1626             rinvsq13         = _mm256_mul_ps(rinv13,rinv13);
1627             rinvsq21         = _mm256_mul_ps(rinv21,rinv21);
1628             rinvsq22         = _mm256_mul_ps(rinv22,rinv22);
1629             rinvsq23         = _mm256_mul_ps(rinv23,rinv23);
1630             rinvsq31         = _mm256_mul_ps(rinv31,rinv31);
1631             rinvsq32         = _mm256_mul_ps(rinv32,rinv32);
1632             rinvsq33         = _mm256_mul_ps(rinv33,rinv33);
1633
1634             fjx1             = _mm256_setzero_ps();
1635             fjy1             = _mm256_setzero_ps();
1636             fjz1             = _mm256_setzero_ps();
1637             fjx2             = _mm256_setzero_ps();
1638             fjy2             = _mm256_setzero_ps();
1639             fjz2             = _mm256_setzero_ps();
1640             fjx3             = _mm256_setzero_ps();
1641             fjy3             = _mm256_setzero_ps();
1642             fjz3             = _mm256_setzero_ps();
1643
1644             /**************************
1645              * CALCULATE INTERACTIONS *
1646              **************************/
1647
1648             if (gmx_mm256_any_lt(rsq11,rcutoff2))
1649             {
1650
1651             r11              = _mm256_mul_ps(rsq11,rinv11);
1652
1653             /* EWALD ELECTROSTATICS */
1654
1655             /* Analytical PME correction */
1656             zeta2            = _mm256_mul_ps(beta2,rsq11);
1657             rinv3            = _mm256_mul_ps(rinvsq11,rinv11);
1658             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
1659             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
1660             felec            = _mm256_mul_ps(qq11,felec);
1661
1662             cutoff_mask      = _mm256_cmp_ps(rsq11,rcutoff2,_CMP_LT_OQ);
1663
1664             fscal            = felec;
1665
1666             fscal            = _mm256_and_ps(fscal,cutoff_mask);
1667
1668             /* Calculate temporary vectorial force */
1669             tx               = _mm256_mul_ps(fscal,dx11);
1670             ty               = _mm256_mul_ps(fscal,dy11);
1671             tz               = _mm256_mul_ps(fscal,dz11);
1672
1673             /* Update vectorial force */
1674             fix1             = _mm256_add_ps(fix1,tx);
1675             fiy1             = _mm256_add_ps(fiy1,ty);
1676             fiz1             = _mm256_add_ps(fiz1,tz);
1677
1678             fjx1             = _mm256_add_ps(fjx1,tx);
1679             fjy1             = _mm256_add_ps(fjy1,ty);
1680             fjz1             = _mm256_add_ps(fjz1,tz);
1681
1682             }
1683
1684             /**************************
1685              * CALCULATE INTERACTIONS *
1686              **************************/
1687
1688             if (gmx_mm256_any_lt(rsq12,rcutoff2))
1689             {
1690
1691             r12              = _mm256_mul_ps(rsq12,rinv12);
1692
1693             /* EWALD ELECTROSTATICS */
1694
1695             /* Analytical PME correction */
1696             zeta2            = _mm256_mul_ps(beta2,rsq12);
1697             rinv3            = _mm256_mul_ps(rinvsq12,rinv12);
1698             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
1699             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
1700             felec            = _mm256_mul_ps(qq12,felec);
1701
1702             cutoff_mask      = _mm256_cmp_ps(rsq12,rcutoff2,_CMP_LT_OQ);
1703
1704             fscal            = felec;
1705
1706             fscal            = _mm256_and_ps(fscal,cutoff_mask);
1707
1708             /* Calculate temporary vectorial force */
1709             tx               = _mm256_mul_ps(fscal,dx12);
1710             ty               = _mm256_mul_ps(fscal,dy12);
1711             tz               = _mm256_mul_ps(fscal,dz12);
1712
1713             /* Update vectorial force */
1714             fix1             = _mm256_add_ps(fix1,tx);
1715             fiy1             = _mm256_add_ps(fiy1,ty);
1716             fiz1             = _mm256_add_ps(fiz1,tz);
1717
1718             fjx2             = _mm256_add_ps(fjx2,tx);
1719             fjy2             = _mm256_add_ps(fjy2,ty);
1720             fjz2             = _mm256_add_ps(fjz2,tz);
1721
1722             }
1723
1724             /**************************
1725              * CALCULATE INTERACTIONS *
1726              **************************/
1727
1728             if (gmx_mm256_any_lt(rsq13,rcutoff2))
1729             {
1730
1731             r13              = _mm256_mul_ps(rsq13,rinv13);
1732
1733             /* EWALD ELECTROSTATICS */
1734
1735             /* Analytical PME correction */
1736             zeta2            = _mm256_mul_ps(beta2,rsq13);
1737             rinv3            = _mm256_mul_ps(rinvsq13,rinv13);
1738             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
1739             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
1740             felec            = _mm256_mul_ps(qq13,felec);
1741
1742             cutoff_mask      = _mm256_cmp_ps(rsq13,rcutoff2,_CMP_LT_OQ);
1743
1744             fscal            = felec;
1745
1746             fscal            = _mm256_and_ps(fscal,cutoff_mask);
1747
1748             /* Calculate temporary vectorial force */
1749             tx               = _mm256_mul_ps(fscal,dx13);
1750             ty               = _mm256_mul_ps(fscal,dy13);
1751             tz               = _mm256_mul_ps(fscal,dz13);
1752
1753             /* Update vectorial force */
1754             fix1             = _mm256_add_ps(fix1,tx);
1755             fiy1             = _mm256_add_ps(fiy1,ty);
1756             fiz1             = _mm256_add_ps(fiz1,tz);
1757
1758             fjx3             = _mm256_add_ps(fjx3,tx);
1759             fjy3             = _mm256_add_ps(fjy3,ty);
1760             fjz3             = _mm256_add_ps(fjz3,tz);
1761
1762             }
1763
1764             /**************************
1765              * CALCULATE INTERACTIONS *
1766              **************************/
1767
1768             if (gmx_mm256_any_lt(rsq21,rcutoff2))
1769             {
1770
1771             r21              = _mm256_mul_ps(rsq21,rinv21);
1772
1773             /* EWALD ELECTROSTATICS */
1774
1775             /* Analytical PME correction */
1776             zeta2            = _mm256_mul_ps(beta2,rsq21);
1777             rinv3            = _mm256_mul_ps(rinvsq21,rinv21);
1778             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
1779             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
1780             felec            = _mm256_mul_ps(qq21,felec);
1781
1782             cutoff_mask      = _mm256_cmp_ps(rsq21,rcutoff2,_CMP_LT_OQ);
1783
1784             fscal            = felec;
1785
1786             fscal            = _mm256_and_ps(fscal,cutoff_mask);
1787
1788             /* Calculate temporary vectorial force */
1789             tx               = _mm256_mul_ps(fscal,dx21);
1790             ty               = _mm256_mul_ps(fscal,dy21);
1791             tz               = _mm256_mul_ps(fscal,dz21);
1792
1793             /* Update vectorial force */
1794             fix2             = _mm256_add_ps(fix2,tx);
1795             fiy2             = _mm256_add_ps(fiy2,ty);
1796             fiz2             = _mm256_add_ps(fiz2,tz);
1797
1798             fjx1             = _mm256_add_ps(fjx1,tx);
1799             fjy1             = _mm256_add_ps(fjy1,ty);
1800             fjz1             = _mm256_add_ps(fjz1,tz);
1801
1802             }
1803
1804             /**************************
1805              * CALCULATE INTERACTIONS *
1806              **************************/
1807
1808             if (gmx_mm256_any_lt(rsq22,rcutoff2))
1809             {
1810
1811             r22              = _mm256_mul_ps(rsq22,rinv22);
1812
1813             /* EWALD ELECTROSTATICS */
1814
1815             /* Analytical PME correction */
1816             zeta2            = _mm256_mul_ps(beta2,rsq22);
1817             rinv3            = _mm256_mul_ps(rinvsq22,rinv22);
1818             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
1819             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
1820             felec            = _mm256_mul_ps(qq22,felec);
1821
1822             cutoff_mask      = _mm256_cmp_ps(rsq22,rcutoff2,_CMP_LT_OQ);
1823
1824             fscal            = felec;
1825
1826             fscal            = _mm256_and_ps(fscal,cutoff_mask);
1827
1828             /* Calculate temporary vectorial force */
1829             tx               = _mm256_mul_ps(fscal,dx22);
1830             ty               = _mm256_mul_ps(fscal,dy22);
1831             tz               = _mm256_mul_ps(fscal,dz22);
1832
1833             /* Update vectorial force */
1834             fix2             = _mm256_add_ps(fix2,tx);
1835             fiy2             = _mm256_add_ps(fiy2,ty);
1836             fiz2             = _mm256_add_ps(fiz2,tz);
1837
1838             fjx2             = _mm256_add_ps(fjx2,tx);
1839             fjy2             = _mm256_add_ps(fjy2,ty);
1840             fjz2             = _mm256_add_ps(fjz2,tz);
1841
1842             }
1843
1844             /**************************
1845              * CALCULATE INTERACTIONS *
1846              **************************/
1847
1848             if (gmx_mm256_any_lt(rsq23,rcutoff2))
1849             {
1850
1851             r23              = _mm256_mul_ps(rsq23,rinv23);
1852
1853             /* EWALD ELECTROSTATICS */
1854
1855             /* Analytical PME correction */
1856             zeta2            = _mm256_mul_ps(beta2,rsq23);
1857             rinv3            = _mm256_mul_ps(rinvsq23,rinv23);
1858             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
1859             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
1860             felec            = _mm256_mul_ps(qq23,felec);
1861
1862             cutoff_mask      = _mm256_cmp_ps(rsq23,rcutoff2,_CMP_LT_OQ);
1863
1864             fscal            = felec;
1865
1866             fscal            = _mm256_and_ps(fscal,cutoff_mask);
1867
1868             /* Calculate temporary vectorial force */
1869             tx               = _mm256_mul_ps(fscal,dx23);
1870             ty               = _mm256_mul_ps(fscal,dy23);
1871             tz               = _mm256_mul_ps(fscal,dz23);
1872
1873             /* Update vectorial force */
1874             fix2             = _mm256_add_ps(fix2,tx);
1875             fiy2             = _mm256_add_ps(fiy2,ty);
1876             fiz2             = _mm256_add_ps(fiz2,tz);
1877
1878             fjx3             = _mm256_add_ps(fjx3,tx);
1879             fjy3             = _mm256_add_ps(fjy3,ty);
1880             fjz3             = _mm256_add_ps(fjz3,tz);
1881
1882             }
1883
1884             /**************************
1885              * CALCULATE INTERACTIONS *
1886              **************************/
1887
1888             if (gmx_mm256_any_lt(rsq31,rcutoff2))
1889             {
1890
1891             r31              = _mm256_mul_ps(rsq31,rinv31);
1892
1893             /* EWALD ELECTROSTATICS */
1894
1895             /* Analytical PME correction */
1896             zeta2            = _mm256_mul_ps(beta2,rsq31);
1897             rinv3            = _mm256_mul_ps(rinvsq31,rinv31);
1898             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
1899             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
1900             felec            = _mm256_mul_ps(qq31,felec);
1901
1902             cutoff_mask      = _mm256_cmp_ps(rsq31,rcutoff2,_CMP_LT_OQ);
1903
1904             fscal            = felec;
1905
1906             fscal            = _mm256_and_ps(fscal,cutoff_mask);
1907
1908             /* Calculate temporary vectorial force */
1909             tx               = _mm256_mul_ps(fscal,dx31);
1910             ty               = _mm256_mul_ps(fscal,dy31);
1911             tz               = _mm256_mul_ps(fscal,dz31);
1912
1913             /* Update vectorial force */
1914             fix3             = _mm256_add_ps(fix3,tx);
1915             fiy3             = _mm256_add_ps(fiy3,ty);
1916             fiz3             = _mm256_add_ps(fiz3,tz);
1917
1918             fjx1             = _mm256_add_ps(fjx1,tx);
1919             fjy1             = _mm256_add_ps(fjy1,ty);
1920             fjz1             = _mm256_add_ps(fjz1,tz);
1921
1922             }
1923
1924             /**************************
1925              * CALCULATE INTERACTIONS *
1926              **************************/
1927
1928             if (gmx_mm256_any_lt(rsq32,rcutoff2))
1929             {
1930
1931             r32              = _mm256_mul_ps(rsq32,rinv32);
1932
1933             /* EWALD ELECTROSTATICS */
1934
1935             /* Analytical PME correction */
1936             zeta2            = _mm256_mul_ps(beta2,rsq32);
1937             rinv3            = _mm256_mul_ps(rinvsq32,rinv32);
1938             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
1939             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
1940             felec            = _mm256_mul_ps(qq32,felec);
1941
1942             cutoff_mask      = _mm256_cmp_ps(rsq32,rcutoff2,_CMP_LT_OQ);
1943
1944             fscal            = felec;
1945
1946             fscal            = _mm256_and_ps(fscal,cutoff_mask);
1947
1948             /* Calculate temporary vectorial force */
1949             tx               = _mm256_mul_ps(fscal,dx32);
1950             ty               = _mm256_mul_ps(fscal,dy32);
1951             tz               = _mm256_mul_ps(fscal,dz32);
1952
1953             /* Update vectorial force */
1954             fix3             = _mm256_add_ps(fix3,tx);
1955             fiy3             = _mm256_add_ps(fiy3,ty);
1956             fiz3             = _mm256_add_ps(fiz3,tz);
1957
1958             fjx2             = _mm256_add_ps(fjx2,tx);
1959             fjy2             = _mm256_add_ps(fjy2,ty);
1960             fjz2             = _mm256_add_ps(fjz2,tz);
1961
1962             }
1963
1964             /**************************
1965              * CALCULATE INTERACTIONS *
1966              **************************/
1967
1968             if (gmx_mm256_any_lt(rsq33,rcutoff2))
1969             {
1970
1971             r33              = _mm256_mul_ps(rsq33,rinv33);
1972
1973             /* EWALD ELECTROSTATICS */
1974
1975             /* Analytical PME correction */
1976             zeta2            = _mm256_mul_ps(beta2,rsq33);
1977             rinv3            = _mm256_mul_ps(rinvsq33,rinv33);
1978             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
1979             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
1980             felec            = _mm256_mul_ps(qq33,felec);
1981
1982             cutoff_mask      = _mm256_cmp_ps(rsq33,rcutoff2,_CMP_LT_OQ);
1983
1984             fscal            = felec;
1985
1986             fscal            = _mm256_and_ps(fscal,cutoff_mask);
1987
1988             /* Calculate temporary vectorial force */
1989             tx               = _mm256_mul_ps(fscal,dx33);
1990             ty               = _mm256_mul_ps(fscal,dy33);
1991             tz               = _mm256_mul_ps(fscal,dz33);
1992
1993             /* Update vectorial force */
1994             fix3             = _mm256_add_ps(fix3,tx);
1995             fiy3             = _mm256_add_ps(fiy3,ty);
1996             fiz3             = _mm256_add_ps(fiz3,tz);
1997
1998             fjx3             = _mm256_add_ps(fjx3,tx);
1999             fjy3             = _mm256_add_ps(fjy3,ty);
2000             fjz3             = _mm256_add_ps(fjz3,tz);
2001
2002             }
2003
2004             fjptrA             = f+j_coord_offsetA;
2005             fjptrB             = f+j_coord_offsetB;
2006             fjptrC             = f+j_coord_offsetC;
2007             fjptrD             = f+j_coord_offsetD;
2008             fjptrE             = f+j_coord_offsetE;
2009             fjptrF             = f+j_coord_offsetF;
2010             fjptrG             = f+j_coord_offsetG;
2011             fjptrH             = f+j_coord_offsetH;
2012
2013             gmx_mm256_decrement_3rvec_8ptr_swizzle_ps(fjptrA+DIM,fjptrB+DIM,fjptrC+DIM,fjptrD+DIM,
2014                                                       fjptrE+DIM,fjptrF+DIM,fjptrG+DIM,fjptrH+DIM,
2015                                                       fjx1,fjy1,fjz1,fjx2,fjy2,fjz2,fjx3,fjy3,fjz3);
2016
2017             /* Inner loop uses 531 flops */
2018         }
2019
2020         if(jidx<j_index_end)
2021         {
2022
2023             /* Get j neighbor index, and coordinate index */
2024             jnrlistA         = jjnr[jidx];
2025             jnrlistB         = jjnr[jidx+1];
2026             jnrlistC         = jjnr[jidx+2];
2027             jnrlistD         = jjnr[jidx+3];
2028             jnrlistE         = jjnr[jidx+4];
2029             jnrlistF         = jjnr[jidx+5];
2030             jnrlistG         = jjnr[jidx+6];
2031             jnrlistH         = jjnr[jidx+7];
2032             /* Sign of each element will be negative for non-real atoms.
2033              * This mask will be 0xFFFFFFFF for dummy entries and 0x0 for real ones,
2034              * so use it as val = _mm_andnot_ps(mask,val) to clear dummy entries.
2035              */
2036             dummy_mask = gmx_mm256_set_m128(gmx_mm_castsi128_ps(_mm_cmplt_epi32(_mm_loadu_si128((const __m128i *)(jjnr+jidx+4)),_mm_setzero_si128())),
2037                                             gmx_mm_castsi128_ps(_mm_cmplt_epi32(_mm_loadu_si128((const __m128i *)(jjnr+jidx)),_mm_setzero_si128())));
2038
2039             jnrA       = (jnrlistA>=0) ? jnrlistA : 0;
2040             jnrB       = (jnrlistB>=0) ? jnrlistB : 0;
2041             jnrC       = (jnrlistC>=0) ? jnrlistC : 0;
2042             jnrD       = (jnrlistD>=0) ? jnrlistD : 0;
2043             jnrE       = (jnrlistE>=0) ? jnrlistE : 0;
2044             jnrF       = (jnrlistF>=0) ? jnrlistF : 0;
2045             jnrG       = (jnrlistG>=0) ? jnrlistG : 0;
2046             jnrH       = (jnrlistH>=0) ? jnrlistH : 0;
2047             j_coord_offsetA  = DIM*jnrA;
2048             j_coord_offsetB  = DIM*jnrB;
2049             j_coord_offsetC  = DIM*jnrC;
2050             j_coord_offsetD  = DIM*jnrD;
2051             j_coord_offsetE  = DIM*jnrE;
2052             j_coord_offsetF  = DIM*jnrF;
2053             j_coord_offsetG  = DIM*jnrG;
2054             j_coord_offsetH  = DIM*jnrH;
2055
2056             /* load j atom coordinates */
2057             gmx_mm256_load_3rvec_8ptr_swizzle_ps(x+j_coord_offsetA+DIM,x+j_coord_offsetB+DIM,
2058                                                  x+j_coord_offsetC+DIM,x+j_coord_offsetD+DIM,
2059                                                  x+j_coord_offsetE+DIM,x+j_coord_offsetF+DIM,
2060                                                  x+j_coord_offsetG+DIM,x+j_coord_offsetH+DIM,
2061                                                  &jx1,&jy1,&jz1,&jx2,&jy2,&jz2,&jx3,&jy3,&jz3);
2062
2063             /* Calculate displacement vector */
2064             dx11             = _mm256_sub_ps(ix1,jx1);
2065             dy11             = _mm256_sub_ps(iy1,jy1);
2066             dz11             = _mm256_sub_ps(iz1,jz1);
2067             dx12             = _mm256_sub_ps(ix1,jx2);
2068             dy12             = _mm256_sub_ps(iy1,jy2);
2069             dz12             = _mm256_sub_ps(iz1,jz2);
2070             dx13             = _mm256_sub_ps(ix1,jx3);
2071             dy13             = _mm256_sub_ps(iy1,jy3);
2072             dz13             = _mm256_sub_ps(iz1,jz3);
2073             dx21             = _mm256_sub_ps(ix2,jx1);
2074             dy21             = _mm256_sub_ps(iy2,jy1);
2075             dz21             = _mm256_sub_ps(iz2,jz1);
2076             dx22             = _mm256_sub_ps(ix2,jx2);
2077             dy22             = _mm256_sub_ps(iy2,jy2);
2078             dz22             = _mm256_sub_ps(iz2,jz2);
2079             dx23             = _mm256_sub_ps(ix2,jx3);
2080             dy23             = _mm256_sub_ps(iy2,jy3);
2081             dz23             = _mm256_sub_ps(iz2,jz3);
2082             dx31             = _mm256_sub_ps(ix3,jx1);
2083             dy31             = _mm256_sub_ps(iy3,jy1);
2084             dz31             = _mm256_sub_ps(iz3,jz1);
2085             dx32             = _mm256_sub_ps(ix3,jx2);
2086             dy32             = _mm256_sub_ps(iy3,jy2);
2087             dz32             = _mm256_sub_ps(iz3,jz2);
2088             dx33             = _mm256_sub_ps(ix3,jx3);
2089             dy33             = _mm256_sub_ps(iy3,jy3);
2090             dz33             = _mm256_sub_ps(iz3,jz3);
2091
2092             /* Calculate squared distance and things based on it */
2093             rsq11            = gmx_mm256_calc_rsq_ps(dx11,dy11,dz11);
2094             rsq12            = gmx_mm256_calc_rsq_ps(dx12,dy12,dz12);
2095             rsq13            = gmx_mm256_calc_rsq_ps(dx13,dy13,dz13);
2096             rsq21            = gmx_mm256_calc_rsq_ps(dx21,dy21,dz21);
2097             rsq22            = gmx_mm256_calc_rsq_ps(dx22,dy22,dz22);
2098             rsq23            = gmx_mm256_calc_rsq_ps(dx23,dy23,dz23);
2099             rsq31            = gmx_mm256_calc_rsq_ps(dx31,dy31,dz31);
2100             rsq32            = gmx_mm256_calc_rsq_ps(dx32,dy32,dz32);
2101             rsq33            = gmx_mm256_calc_rsq_ps(dx33,dy33,dz33);
2102
2103             rinv11           = gmx_mm256_invsqrt_ps(rsq11);
2104             rinv12           = gmx_mm256_invsqrt_ps(rsq12);
2105             rinv13           = gmx_mm256_invsqrt_ps(rsq13);
2106             rinv21           = gmx_mm256_invsqrt_ps(rsq21);
2107             rinv22           = gmx_mm256_invsqrt_ps(rsq22);
2108             rinv23           = gmx_mm256_invsqrt_ps(rsq23);
2109             rinv31           = gmx_mm256_invsqrt_ps(rsq31);
2110             rinv32           = gmx_mm256_invsqrt_ps(rsq32);
2111             rinv33           = gmx_mm256_invsqrt_ps(rsq33);
2112
2113             rinvsq11         = _mm256_mul_ps(rinv11,rinv11);
2114             rinvsq12         = _mm256_mul_ps(rinv12,rinv12);
2115             rinvsq13         = _mm256_mul_ps(rinv13,rinv13);
2116             rinvsq21         = _mm256_mul_ps(rinv21,rinv21);
2117             rinvsq22         = _mm256_mul_ps(rinv22,rinv22);
2118             rinvsq23         = _mm256_mul_ps(rinv23,rinv23);
2119             rinvsq31         = _mm256_mul_ps(rinv31,rinv31);
2120             rinvsq32         = _mm256_mul_ps(rinv32,rinv32);
2121             rinvsq33         = _mm256_mul_ps(rinv33,rinv33);
2122
2123             fjx1             = _mm256_setzero_ps();
2124             fjy1             = _mm256_setzero_ps();
2125             fjz1             = _mm256_setzero_ps();
2126             fjx2             = _mm256_setzero_ps();
2127             fjy2             = _mm256_setzero_ps();
2128             fjz2             = _mm256_setzero_ps();
2129             fjx3             = _mm256_setzero_ps();
2130             fjy3             = _mm256_setzero_ps();
2131             fjz3             = _mm256_setzero_ps();
2132
2133             /**************************
2134              * CALCULATE INTERACTIONS *
2135              **************************/
2136
2137             if (gmx_mm256_any_lt(rsq11,rcutoff2))
2138             {
2139
2140             r11              = _mm256_mul_ps(rsq11,rinv11);
2141             r11              = _mm256_andnot_ps(dummy_mask,r11);
2142
2143             /* EWALD ELECTROSTATICS */
2144
2145             /* Analytical PME correction */
2146             zeta2            = _mm256_mul_ps(beta2,rsq11);
2147             rinv3            = _mm256_mul_ps(rinvsq11,rinv11);
2148             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
2149             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
2150             felec            = _mm256_mul_ps(qq11,felec);
2151
2152             cutoff_mask      = _mm256_cmp_ps(rsq11,rcutoff2,_CMP_LT_OQ);
2153
2154             fscal            = felec;
2155
2156             fscal            = _mm256_and_ps(fscal,cutoff_mask);
2157
2158             fscal            = _mm256_andnot_ps(dummy_mask,fscal);
2159
2160             /* Calculate temporary vectorial force */
2161             tx               = _mm256_mul_ps(fscal,dx11);
2162             ty               = _mm256_mul_ps(fscal,dy11);
2163             tz               = _mm256_mul_ps(fscal,dz11);
2164
2165             /* Update vectorial force */
2166             fix1             = _mm256_add_ps(fix1,tx);
2167             fiy1             = _mm256_add_ps(fiy1,ty);
2168             fiz1             = _mm256_add_ps(fiz1,tz);
2169
2170             fjx1             = _mm256_add_ps(fjx1,tx);
2171             fjy1             = _mm256_add_ps(fjy1,ty);
2172             fjz1             = _mm256_add_ps(fjz1,tz);
2173
2174             }
2175
2176             /**************************
2177              * CALCULATE INTERACTIONS *
2178              **************************/
2179
2180             if (gmx_mm256_any_lt(rsq12,rcutoff2))
2181             {
2182
2183             r12              = _mm256_mul_ps(rsq12,rinv12);
2184             r12              = _mm256_andnot_ps(dummy_mask,r12);
2185
2186             /* EWALD ELECTROSTATICS */
2187
2188             /* Analytical PME correction */
2189             zeta2            = _mm256_mul_ps(beta2,rsq12);
2190             rinv3            = _mm256_mul_ps(rinvsq12,rinv12);
2191             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
2192             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
2193             felec            = _mm256_mul_ps(qq12,felec);
2194
2195             cutoff_mask      = _mm256_cmp_ps(rsq12,rcutoff2,_CMP_LT_OQ);
2196
2197             fscal            = felec;
2198
2199             fscal            = _mm256_and_ps(fscal,cutoff_mask);
2200
2201             fscal            = _mm256_andnot_ps(dummy_mask,fscal);
2202
2203             /* Calculate temporary vectorial force */
2204             tx               = _mm256_mul_ps(fscal,dx12);
2205             ty               = _mm256_mul_ps(fscal,dy12);
2206             tz               = _mm256_mul_ps(fscal,dz12);
2207
2208             /* Update vectorial force */
2209             fix1             = _mm256_add_ps(fix1,tx);
2210             fiy1             = _mm256_add_ps(fiy1,ty);
2211             fiz1             = _mm256_add_ps(fiz1,tz);
2212
2213             fjx2             = _mm256_add_ps(fjx2,tx);
2214             fjy2             = _mm256_add_ps(fjy2,ty);
2215             fjz2             = _mm256_add_ps(fjz2,tz);
2216
2217             }
2218
2219             /**************************
2220              * CALCULATE INTERACTIONS *
2221              **************************/
2222
2223             if (gmx_mm256_any_lt(rsq13,rcutoff2))
2224             {
2225
2226             r13              = _mm256_mul_ps(rsq13,rinv13);
2227             r13              = _mm256_andnot_ps(dummy_mask,r13);
2228
2229             /* EWALD ELECTROSTATICS */
2230
2231             /* Analytical PME correction */
2232             zeta2            = _mm256_mul_ps(beta2,rsq13);
2233             rinv3            = _mm256_mul_ps(rinvsq13,rinv13);
2234             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
2235             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
2236             felec            = _mm256_mul_ps(qq13,felec);
2237
2238             cutoff_mask      = _mm256_cmp_ps(rsq13,rcutoff2,_CMP_LT_OQ);
2239
2240             fscal            = felec;
2241
2242             fscal            = _mm256_and_ps(fscal,cutoff_mask);
2243
2244             fscal            = _mm256_andnot_ps(dummy_mask,fscal);
2245
2246             /* Calculate temporary vectorial force */
2247             tx               = _mm256_mul_ps(fscal,dx13);
2248             ty               = _mm256_mul_ps(fscal,dy13);
2249             tz               = _mm256_mul_ps(fscal,dz13);
2250
2251             /* Update vectorial force */
2252             fix1             = _mm256_add_ps(fix1,tx);
2253             fiy1             = _mm256_add_ps(fiy1,ty);
2254             fiz1             = _mm256_add_ps(fiz1,tz);
2255
2256             fjx3             = _mm256_add_ps(fjx3,tx);
2257             fjy3             = _mm256_add_ps(fjy3,ty);
2258             fjz3             = _mm256_add_ps(fjz3,tz);
2259
2260             }
2261
2262             /**************************
2263              * CALCULATE INTERACTIONS *
2264              **************************/
2265
2266             if (gmx_mm256_any_lt(rsq21,rcutoff2))
2267             {
2268
2269             r21              = _mm256_mul_ps(rsq21,rinv21);
2270             r21              = _mm256_andnot_ps(dummy_mask,r21);
2271
2272             /* EWALD ELECTROSTATICS */
2273
2274             /* Analytical PME correction */
2275             zeta2            = _mm256_mul_ps(beta2,rsq21);
2276             rinv3            = _mm256_mul_ps(rinvsq21,rinv21);
2277             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
2278             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
2279             felec            = _mm256_mul_ps(qq21,felec);
2280
2281             cutoff_mask      = _mm256_cmp_ps(rsq21,rcutoff2,_CMP_LT_OQ);
2282
2283             fscal            = felec;
2284
2285             fscal            = _mm256_and_ps(fscal,cutoff_mask);
2286
2287             fscal            = _mm256_andnot_ps(dummy_mask,fscal);
2288
2289             /* Calculate temporary vectorial force */
2290             tx               = _mm256_mul_ps(fscal,dx21);
2291             ty               = _mm256_mul_ps(fscal,dy21);
2292             tz               = _mm256_mul_ps(fscal,dz21);
2293
2294             /* Update vectorial force */
2295             fix2             = _mm256_add_ps(fix2,tx);
2296             fiy2             = _mm256_add_ps(fiy2,ty);
2297             fiz2             = _mm256_add_ps(fiz2,tz);
2298
2299             fjx1             = _mm256_add_ps(fjx1,tx);
2300             fjy1             = _mm256_add_ps(fjy1,ty);
2301             fjz1             = _mm256_add_ps(fjz1,tz);
2302
2303             }
2304
2305             /**************************
2306              * CALCULATE INTERACTIONS *
2307              **************************/
2308
2309             if (gmx_mm256_any_lt(rsq22,rcutoff2))
2310             {
2311
2312             r22              = _mm256_mul_ps(rsq22,rinv22);
2313             r22              = _mm256_andnot_ps(dummy_mask,r22);
2314
2315             /* EWALD ELECTROSTATICS */
2316
2317             /* Analytical PME correction */
2318             zeta2            = _mm256_mul_ps(beta2,rsq22);
2319             rinv3            = _mm256_mul_ps(rinvsq22,rinv22);
2320             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
2321             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
2322             felec            = _mm256_mul_ps(qq22,felec);
2323
2324             cutoff_mask      = _mm256_cmp_ps(rsq22,rcutoff2,_CMP_LT_OQ);
2325
2326             fscal            = felec;
2327
2328             fscal            = _mm256_and_ps(fscal,cutoff_mask);
2329
2330             fscal            = _mm256_andnot_ps(dummy_mask,fscal);
2331
2332             /* Calculate temporary vectorial force */
2333             tx               = _mm256_mul_ps(fscal,dx22);
2334             ty               = _mm256_mul_ps(fscal,dy22);
2335             tz               = _mm256_mul_ps(fscal,dz22);
2336
2337             /* Update vectorial force */
2338             fix2             = _mm256_add_ps(fix2,tx);
2339             fiy2             = _mm256_add_ps(fiy2,ty);
2340             fiz2             = _mm256_add_ps(fiz2,tz);
2341
2342             fjx2             = _mm256_add_ps(fjx2,tx);
2343             fjy2             = _mm256_add_ps(fjy2,ty);
2344             fjz2             = _mm256_add_ps(fjz2,tz);
2345
2346             }
2347
2348             /**************************
2349              * CALCULATE INTERACTIONS *
2350              **************************/
2351
2352             if (gmx_mm256_any_lt(rsq23,rcutoff2))
2353             {
2354
2355             r23              = _mm256_mul_ps(rsq23,rinv23);
2356             r23              = _mm256_andnot_ps(dummy_mask,r23);
2357
2358             /* EWALD ELECTROSTATICS */
2359
2360             /* Analytical PME correction */
2361             zeta2            = _mm256_mul_ps(beta2,rsq23);
2362             rinv3            = _mm256_mul_ps(rinvsq23,rinv23);
2363             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
2364             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
2365             felec            = _mm256_mul_ps(qq23,felec);
2366
2367             cutoff_mask      = _mm256_cmp_ps(rsq23,rcutoff2,_CMP_LT_OQ);
2368
2369             fscal            = felec;
2370
2371             fscal            = _mm256_and_ps(fscal,cutoff_mask);
2372
2373             fscal            = _mm256_andnot_ps(dummy_mask,fscal);
2374
2375             /* Calculate temporary vectorial force */
2376             tx               = _mm256_mul_ps(fscal,dx23);
2377             ty               = _mm256_mul_ps(fscal,dy23);
2378             tz               = _mm256_mul_ps(fscal,dz23);
2379
2380             /* Update vectorial force */
2381             fix2             = _mm256_add_ps(fix2,tx);
2382             fiy2             = _mm256_add_ps(fiy2,ty);
2383             fiz2             = _mm256_add_ps(fiz2,tz);
2384
2385             fjx3             = _mm256_add_ps(fjx3,tx);
2386             fjy3             = _mm256_add_ps(fjy3,ty);
2387             fjz3             = _mm256_add_ps(fjz3,tz);
2388
2389             }
2390
2391             /**************************
2392              * CALCULATE INTERACTIONS *
2393              **************************/
2394
2395             if (gmx_mm256_any_lt(rsq31,rcutoff2))
2396             {
2397
2398             r31              = _mm256_mul_ps(rsq31,rinv31);
2399             r31              = _mm256_andnot_ps(dummy_mask,r31);
2400
2401             /* EWALD ELECTROSTATICS */
2402
2403             /* Analytical PME correction */
2404             zeta2            = _mm256_mul_ps(beta2,rsq31);
2405             rinv3            = _mm256_mul_ps(rinvsq31,rinv31);
2406             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
2407             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
2408             felec            = _mm256_mul_ps(qq31,felec);
2409
2410             cutoff_mask      = _mm256_cmp_ps(rsq31,rcutoff2,_CMP_LT_OQ);
2411
2412             fscal            = felec;
2413
2414             fscal            = _mm256_and_ps(fscal,cutoff_mask);
2415
2416             fscal            = _mm256_andnot_ps(dummy_mask,fscal);
2417
2418             /* Calculate temporary vectorial force */
2419             tx               = _mm256_mul_ps(fscal,dx31);
2420             ty               = _mm256_mul_ps(fscal,dy31);
2421             tz               = _mm256_mul_ps(fscal,dz31);
2422
2423             /* Update vectorial force */
2424             fix3             = _mm256_add_ps(fix3,tx);
2425             fiy3             = _mm256_add_ps(fiy3,ty);
2426             fiz3             = _mm256_add_ps(fiz3,tz);
2427
2428             fjx1             = _mm256_add_ps(fjx1,tx);
2429             fjy1             = _mm256_add_ps(fjy1,ty);
2430             fjz1             = _mm256_add_ps(fjz1,tz);
2431
2432             }
2433
2434             /**************************
2435              * CALCULATE INTERACTIONS *
2436              **************************/
2437
2438             if (gmx_mm256_any_lt(rsq32,rcutoff2))
2439             {
2440
2441             r32              = _mm256_mul_ps(rsq32,rinv32);
2442             r32              = _mm256_andnot_ps(dummy_mask,r32);
2443
2444             /* EWALD ELECTROSTATICS */
2445
2446             /* Analytical PME correction */
2447             zeta2            = _mm256_mul_ps(beta2,rsq32);
2448             rinv3            = _mm256_mul_ps(rinvsq32,rinv32);
2449             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
2450             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
2451             felec            = _mm256_mul_ps(qq32,felec);
2452
2453             cutoff_mask      = _mm256_cmp_ps(rsq32,rcutoff2,_CMP_LT_OQ);
2454
2455             fscal            = felec;
2456
2457             fscal            = _mm256_and_ps(fscal,cutoff_mask);
2458
2459             fscal            = _mm256_andnot_ps(dummy_mask,fscal);
2460
2461             /* Calculate temporary vectorial force */
2462             tx               = _mm256_mul_ps(fscal,dx32);
2463             ty               = _mm256_mul_ps(fscal,dy32);
2464             tz               = _mm256_mul_ps(fscal,dz32);
2465
2466             /* Update vectorial force */
2467             fix3             = _mm256_add_ps(fix3,tx);
2468             fiy3             = _mm256_add_ps(fiy3,ty);
2469             fiz3             = _mm256_add_ps(fiz3,tz);
2470
2471             fjx2             = _mm256_add_ps(fjx2,tx);
2472             fjy2             = _mm256_add_ps(fjy2,ty);
2473             fjz2             = _mm256_add_ps(fjz2,tz);
2474
2475             }
2476
2477             /**************************
2478              * CALCULATE INTERACTIONS *
2479              **************************/
2480
2481             if (gmx_mm256_any_lt(rsq33,rcutoff2))
2482             {
2483
2484             r33              = _mm256_mul_ps(rsq33,rinv33);
2485             r33              = _mm256_andnot_ps(dummy_mask,r33);
2486
2487             /* EWALD ELECTROSTATICS */
2488
2489             /* Analytical PME correction */
2490             zeta2            = _mm256_mul_ps(beta2,rsq33);
2491             rinv3            = _mm256_mul_ps(rinvsq33,rinv33);
2492             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
2493             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
2494             felec            = _mm256_mul_ps(qq33,felec);
2495
2496             cutoff_mask      = _mm256_cmp_ps(rsq33,rcutoff2,_CMP_LT_OQ);
2497
2498             fscal            = felec;
2499
2500             fscal            = _mm256_and_ps(fscal,cutoff_mask);
2501
2502             fscal            = _mm256_andnot_ps(dummy_mask,fscal);
2503
2504             /* Calculate temporary vectorial force */
2505             tx               = _mm256_mul_ps(fscal,dx33);
2506             ty               = _mm256_mul_ps(fscal,dy33);
2507             tz               = _mm256_mul_ps(fscal,dz33);
2508
2509             /* Update vectorial force */
2510             fix3             = _mm256_add_ps(fix3,tx);
2511             fiy3             = _mm256_add_ps(fiy3,ty);
2512             fiz3             = _mm256_add_ps(fiz3,tz);
2513
2514             fjx3             = _mm256_add_ps(fjx3,tx);
2515             fjy3             = _mm256_add_ps(fjy3,ty);
2516             fjz3             = _mm256_add_ps(fjz3,tz);
2517
2518             }
2519
2520             fjptrA             = (jnrlistA>=0) ? f+j_coord_offsetA : scratch;
2521             fjptrB             = (jnrlistB>=0) ? f+j_coord_offsetB : scratch;
2522             fjptrC             = (jnrlistC>=0) ? f+j_coord_offsetC : scratch;
2523             fjptrD             = (jnrlistD>=0) ? f+j_coord_offsetD : scratch;
2524             fjptrE             = (jnrlistE>=0) ? f+j_coord_offsetE : scratch;
2525             fjptrF             = (jnrlistF>=0) ? f+j_coord_offsetF : scratch;
2526             fjptrG             = (jnrlistG>=0) ? f+j_coord_offsetG : scratch;
2527             fjptrH             = (jnrlistH>=0) ? f+j_coord_offsetH : scratch;
2528
2529             gmx_mm256_decrement_3rvec_8ptr_swizzle_ps(fjptrA+DIM,fjptrB+DIM,fjptrC+DIM,fjptrD+DIM,
2530                                                       fjptrE+DIM,fjptrF+DIM,fjptrG+DIM,fjptrH+DIM,
2531                                                       fjx1,fjy1,fjz1,fjx2,fjy2,fjz2,fjx3,fjy3,fjz3);
2532
2533             /* Inner loop uses 540 flops */
2534         }
2535
2536         /* End of innermost loop */
2537
2538         gmx_mm256_update_iforce_3atom_swizzle_ps(fix1,fiy1,fiz1,fix2,fiy2,fiz2,fix3,fiy3,fiz3,
2539                                                  f+i_coord_offset+DIM,fshift+i_shift_offset);
2540
2541         /* Increment number of inner iterations */
2542         inneriter                  += j_index_end - j_index_start;
2543
2544         /* Outer loop uses 18 flops */
2545     }
2546
2547     /* Increment number of outer iterations */
2548     outeriter        += nri;
2549
2550     /* Update outer/inner flops */
2551
2552     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_W4W4_F,outeriter*18 + inneriter*540);
2553 }