src/gromacs/gmxlib/nonbonded/nb_kernel_avx_256_single/nb_kernel_ElecEwSw_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_ElecEwSw_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_ElecEwSw_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           rswitch,swV3,swV4,swV5,swF2,swF3,swF4,d,d2,sw,dsw;
 117     real             rswitch_scalar,d_scalar;
 118     __m256           dummy_mask,cutoff_mask;
 119     __m256           signbit = _mm256_castsi256_ps( _mm256_set1_epi32(0x80000000) );
 120     __m256           one     = _mm256_set1_ps(1.0);
 121     __m256           two     = _mm256_set1_ps(2.0);
 122     x                = xx[0];
 123     f                = ff[0];
 124
 125     nri              = nlist->nri;
 126     iinr             = nlist->iinr;
 127     jindex           = nlist->jindex;
 128     jjnr             = nlist->jjnr;
 129     shiftidx         = nlist->shift;
 130     gid              = nlist->gid;
 131     shiftvec         = fr->shift_vec[0];
 132     fshift           = fr->fshift[0];
 133     facel            = _mm256_set1_ps(fr->epsfac);
 134     charge           = mdatoms->chargeA;
 135
 136     sh_ewald         = _mm256_set1_ps(fr->ic->sh_ewald);
 137     beta             = _mm256_set1_ps(fr->ic->ewaldcoeff_q);
 138     beta2            = _mm256_mul_ps(beta,beta);
 139     beta3            = _mm256_mul_ps(beta,beta2);
 140
 141     ewtab            = fr->ic->tabq_coul_FDV0;
 142     ewtabscale       = _mm256_set1_ps(fr->ic->tabq_scale);
 143     ewtabhalfspace   = _mm256_set1_ps(0.5/fr->ic->tabq_scale);
 144
 145     /* Setup water-specific parameters */
 146     inr              = nlist->iinr[0];
 147     iq1              = _mm256_mul_ps(facel,_mm256_set1_ps(charge[inr+1]));
 148     iq2              = _mm256_mul_ps(facel,_mm256_set1_ps(charge[inr+2]));
 149     iq3              = _mm256_mul_ps(facel,_mm256_set1_ps(charge[inr+3]));
 150
 151     jq1              = _mm256_set1_ps(charge[inr+1]);
 152     jq2              = _mm256_set1_ps(charge[inr+2]);
 153     jq3              = _mm256_set1_ps(charge[inr+3]);
 154     qq11             = _mm256_mul_ps(iq1,jq1);
 155     qq12             = _mm256_mul_ps(iq1,jq2);
 156     qq13             = _mm256_mul_ps(iq1,jq3);
 157     qq21             = _mm256_mul_ps(iq2,jq1);
 158     qq22             = _mm256_mul_ps(iq2,jq2);
 159     qq23             = _mm256_mul_ps(iq2,jq3);
 160     qq31             = _mm256_mul_ps(iq3,jq1);
 161     qq32             = _mm256_mul_ps(iq3,jq2);
 162     qq33             = _mm256_mul_ps(iq3,jq3);
 163
 164     /* When we use explicit cutoffs the value must be identical for elec and VdW, so use elec as an arbitrary choice */
 165     rcutoff_scalar   = fr->rcoulomb;
 166     rcutoff          = _mm256_set1_ps(rcutoff_scalar);
 167     rcutoff2         = _mm256_mul_ps(rcutoff,rcutoff);
 168
 169     rswitch_scalar   = fr->rcoulomb_switch;
 170     rswitch          = _mm256_set1_ps(rswitch_scalar);
 171     /* Setup switch parameters */
 172     d_scalar         = rcutoff_scalar-rswitch_scalar;
 173     d                = _mm256_set1_ps(d_scalar);
 174     swV3             = _mm256_set1_ps(-10.0/(d_scalar*d_scalar*d_scalar));
 175     swV4             = _mm256_set1_ps( 15.0/(d_scalar*d_scalar*d_scalar*d_scalar));
 176     swV5             = _mm256_set1_ps( -6.0/(d_scalar*d_scalar*d_scalar*d_scalar*d_scalar));
 177     swF2             = _mm256_set1_ps(-30.0/(d_scalar*d_scalar*d_scalar));
 178     swF3             = _mm256_set1_ps( 60.0/(d_scalar*d_scalar*d_scalar*d_scalar));
 179     swF4             = _mm256_set1_ps(-30.0/(d_scalar*d_scalar*d_scalar*d_scalar*d_scalar));
 180
 181     /* Avoid stupid compiler warnings */
 182     jnrA = jnrB = jnrC = jnrD = jnrE = jnrF = jnrG = jnrH = 0;
 183     j_coord_offsetA = 0;
 184     j_coord_offsetB = 0;
 185     j_coord_offsetC = 0;
 186     j_coord_offsetD = 0;
 187     j_coord_offsetE = 0;
 188     j_coord_offsetF = 0;
 189     j_coord_offsetG = 0;
 190     j_coord_offsetH = 0;
 191
 192     outeriter        = 0;
 193     inneriter        = 0;
 194
 195     for(iidx=0;iidx<4*DIM;iidx++)
 196     {
 197         scratch[iidx] = 0.0;
 198     }
 199
 200     /* Start outer loop over neighborlists */
 201     for(iidx=0; iidx<nri; iidx++)
 202     {
 203         /* Load shift vector for this list */
 204         i_shift_offset   = DIM*shiftidx[iidx];
 205
 206         /* Load limits for loop over neighbors */
 207         j_index_start    = jindex[iidx];
 208         j_index_end      = jindex[iidx+1];
 209
 210         /* Get outer coordinate index */
 211         inr              = iinr[iidx];
 212         i_coord_offset   = DIM*inr;
 213
 214         /* Load i particle coords and add shift vector */
 215         gmx_mm256_load_shift_and_3rvec_broadcast_ps(shiftvec+i_shift_offset,x+i_coord_offset+DIM,
 216                                                     &ix1,&iy1,&iz1,&ix2,&iy2,&iz2,&ix3,&iy3,&iz3);
 217
 218         fix1             = _mm256_setzero_ps();
 219         fiy1             = _mm256_setzero_ps();
 220         fiz1             = _mm256_setzero_ps();
 221         fix2             = _mm256_setzero_ps();
 222         fiy2             = _mm256_setzero_ps();
 223         fiz2             = _mm256_setzero_ps();
 224         fix3             = _mm256_setzero_ps();
 225         fiy3             = _mm256_setzero_ps();
 226         fiz3             = _mm256_setzero_ps();
 227
 228         /* Reset potential sums */
 229         velecsum         = _mm256_setzero_ps();
 230
 231         /* Start inner kernel loop */
 232         for(jidx=j_index_start; jidx<j_index_end && jjnr[jidx+7]>=0; jidx+=8)
 233         {
 234
 235             /* Get j neighbor index, and coordinate index */
 236             jnrA             = jjnr[jidx];
 237             jnrB             = jjnr[jidx+1];
 238             jnrC             = jjnr[jidx+2];
 239             jnrD             = jjnr[jidx+3];
 240             jnrE             = jjnr[jidx+4];
 241             jnrF             = jjnr[jidx+5];
 242             jnrG             = jjnr[jidx+6];
 243             jnrH             = jjnr[jidx+7];
 244             j_coord_offsetA  = DIM*jnrA;
 245             j_coord_offsetB  = DIM*jnrB;
 246             j_coord_offsetC  = DIM*jnrC;
 247             j_coord_offsetD  = DIM*jnrD;
 248             j_coord_offsetE  = DIM*jnrE;
 249             j_coord_offsetF  = DIM*jnrF;
 250             j_coord_offsetG  = DIM*jnrG;
 251             j_coord_offsetH  = DIM*jnrH;
 252
 253             /* load j atom coordinates */
 254             gmx_mm256_load_3rvec_8ptr_swizzle_ps(x+j_coord_offsetA+DIM,x+j_coord_offsetB+DIM,
 255                                                  x+j_coord_offsetC+DIM,x+j_coord_offsetD+DIM,
 256                                                  x+j_coord_offsetE+DIM,x+j_coord_offsetF+DIM,
 257                                                  x+j_coord_offsetG+DIM,x+j_coord_offsetH+DIM,
 258                                                  &jx1,&jy1,&jz1,&jx2,&jy2,&jz2,&jx3,&jy3,&jz3);
 259
 260             /* Calculate displacement vector */
 261             dx11             = _mm256_sub_ps(ix1,jx1);
 262             dy11             = _mm256_sub_ps(iy1,jy1);
 263             dz11             = _mm256_sub_ps(iz1,jz1);
 264             dx12             = _mm256_sub_ps(ix1,jx2);
 265             dy12             = _mm256_sub_ps(iy1,jy2);
 266             dz12             = _mm256_sub_ps(iz1,jz2);
 267             dx13             = _mm256_sub_ps(ix1,jx3);
 268             dy13             = _mm256_sub_ps(iy1,jy3);
 269             dz13             = _mm256_sub_ps(iz1,jz3);
 270             dx21             = _mm256_sub_ps(ix2,jx1);
 271             dy21             = _mm256_sub_ps(iy2,jy1);
 272             dz21             = _mm256_sub_ps(iz2,jz1);
 273             dx22             = _mm256_sub_ps(ix2,jx2);
 274             dy22             = _mm256_sub_ps(iy2,jy2);
 275             dz22             = _mm256_sub_ps(iz2,jz2);
 276             dx23             = _mm256_sub_ps(ix2,jx3);
 277             dy23             = _mm256_sub_ps(iy2,jy3);
 278             dz23             = _mm256_sub_ps(iz2,jz3);
 279             dx31             = _mm256_sub_ps(ix3,jx1);
 280             dy31             = _mm256_sub_ps(iy3,jy1);
 281             dz31             = _mm256_sub_ps(iz3,jz1);
 282             dx32             = _mm256_sub_ps(ix3,jx2);
 283             dy32             = _mm256_sub_ps(iy3,jy2);
 284             dz32             = _mm256_sub_ps(iz3,jz2);
 285             dx33             = _mm256_sub_ps(ix3,jx3);
 286             dy33             = _mm256_sub_ps(iy3,jy3);
 287             dz33             = _mm256_sub_ps(iz3,jz3);
 288
 289             /* Calculate squared distance and things based on it */
 290             rsq11            = gmx_mm256_calc_rsq_ps(dx11,dy11,dz11);
 291             rsq12            = gmx_mm256_calc_rsq_ps(dx12,dy12,dz12);
 292             rsq13            = gmx_mm256_calc_rsq_ps(dx13,dy13,dz13);
 293             rsq21            = gmx_mm256_calc_rsq_ps(dx21,dy21,dz21);
 294             rsq22            = gmx_mm256_calc_rsq_ps(dx22,dy22,dz22);
 295             rsq23            = gmx_mm256_calc_rsq_ps(dx23,dy23,dz23);
 296             rsq31            = gmx_mm256_calc_rsq_ps(dx31,dy31,dz31);
 297             rsq32            = gmx_mm256_calc_rsq_ps(dx32,dy32,dz32);
 298             rsq33            = gmx_mm256_calc_rsq_ps(dx33,dy33,dz33);
 299
 300             rinv11           = gmx_mm256_invsqrt_ps(rsq11);
 301             rinv12           = gmx_mm256_invsqrt_ps(rsq12);
 302             rinv13           = gmx_mm256_invsqrt_ps(rsq13);
 303             rinv21           = gmx_mm256_invsqrt_ps(rsq21);
 304             rinv22           = gmx_mm256_invsqrt_ps(rsq22);
 305             rinv23           = gmx_mm256_invsqrt_ps(rsq23);
 306             rinv31           = gmx_mm256_invsqrt_ps(rsq31);
 307             rinv32           = gmx_mm256_invsqrt_ps(rsq32);
 308             rinv33           = gmx_mm256_invsqrt_ps(rsq33);
 309
 310             rinvsq11         = _mm256_mul_ps(rinv11,rinv11);
 311             rinvsq12         = _mm256_mul_ps(rinv12,rinv12);
 312             rinvsq13         = _mm256_mul_ps(rinv13,rinv13);
 313             rinvsq21         = _mm256_mul_ps(rinv21,rinv21);
 314             rinvsq22         = _mm256_mul_ps(rinv22,rinv22);
 315             rinvsq23         = _mm256_mul_ps(rinv23,rinv23);
 316             rinvsq31         = _mm256_mul_ps(rinv31,rinv31);
 317             rinvsq32         = _mm256_mul_ps(rinv32,rinv32);
 318             rinvsq33         = _mm256_mul_ps(rinv33,rinv33);
 319
 320             fjx1             = _mm256_setzero_ps();
 321             fjy1             = _mm256_setzero_ps();
 322             fjz1             = _mm256_setzero_ps();
 323             fjx2             = _mm256_setzero_ps();
 324             fjy2             = _mm256_setzero_ps();
 325             fjz2             = _mm256_setzero_ps();
 326             fjx3             = _mm256_setzero_ps();
 327             fjy3             = _mm256_setzero_ps();
 328             fjz3             = _mm256_setzero_ps();
 329
 330             /**************************
 331              * CALCULATE INTERACTIONS *
 332              **************************/
 333
 334             if (gmx_mm256_any_lt(rsq11,rcutoff2))
 335             {
 336
 337             r11              = _mm256_mul_ps(rsq11,rinv11);
 338
 339             /* EWALD ELECTROSTATICS */
 340
 341             /* Analytical PME correction */
 342             zeta2            = _mm256_mul_ps(beta2,rsq11);
 343             rinv3            = _mm256_mul_ps(rinvsq11,rinv11);
 344             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
 345             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
 346             felec            = _mm256_mul_ps(qq11,felec);
 347             pmecorrV         = gmx_mm256_pmecorrV_ps(zeta2);
 348             pmecorrV         = _mm256_mul_ps(pmecorrV,beta);
 349             velec            = _mm256_sub_ps(rinv11,pmecorrV);
 350             velec            = _mm256_mul_ps(qq11,velec);
 351
 352             d                = _mm256_sub_ps(r11,rswitch);
 353             d                = _mm256_max_ps(d,_mm256_setzero_ps());
 354             d2               = _mm256_mul_ps(d,d);
 355             sw               = _mm256_add_ps(one,_mm256_mul_ps(d2,_mm256_mul_ps(d,_mm256_add_ps(swV3,_mm256_mul_ps(d,_mm256_add_ps(swV4,_mm256_mul_ps(d,swV5)))))));
 356
 357             dsw              = _mm256_mul_ps(d2,_mm256_add_ps(swF2,_mm256_mul_ps(d,_mm256_add_ps(swF3,_mm256_mul_ps(d,swF4)))));
 358
 359             /* Evaluate switch function */
 360             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 361             felec            = _mm256_sub_ps( _mm256_mul_ps(felec,sw) , _mm256_mul_ps(rinv11,_mm256_mul_ps(velec,dsw)) );
 362             velec            = _mm256_mul_ps(velec,sw);
 363             cutoff_mask      = _mm256_cmp_ps(rsq11,rcutoff2,_CMP_LT_OQ);
 364
 365             /* Update potential sum for this i atom from the interaction with this j atom. */
 366             velec            = _mm256_and_ps(velec,cutoff_mask);
 367             velecsum         = _mm256_add_ps(velecsum,velec);
 368
 369             fscal            = felec;
 370
 371             fscal            = _mm256_and_ps(fscal,cutoff_mask);
 372
 373             /* Calculate temporary vectorial force */
 374             tx               = _mm256_mul_ps(fscal,dx11);
 375             ty               = _mm256_mul_ps(fscal,dy11);
 376             tz               = _mm256_mul_ps(fscal,dz11);
 377
 378             /* Update vectorial force */
 379             fix1             = _mm256_add_ps(fix1,tx);
 380             fiy1             = _mm256_add_ps(fiy1,ty);
 381             fiz1             = _mm256_add_ps(fiz1,tz);
 382
 383             fjx1             = _mm256_add_ps(fjx1,tx);
 384             fjy1             = _mm256_add_ps(fjy1,ty);
 385             fjz1             = _mm256_add_ps(fjz1,tz);
 386
 387             }
 388
 389             /**************************
 390              * CALCULATE INTERACTIONS *
 391              **************************/
 392
 393             if (gmx_mm256_any_lt(rsq12,rcutoff2))
 394             {
 395
 396             r12              = _mm256_mul_ps(rsq12,rinv12);
 397
 398             /* EWALD ELECTROSTATICS */
 399
 400             /* Analytical PME correction */
 401             zeta2            = _mm256_mul_ps(beta2,rsq12);
 402             rinv3            = _mm256_mul_ps(rinvsq12,rinv12);
 403             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
 404             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
 405             felec            = _mm256_mul_ps(qq12,felec);
 406             pmecorrV         = gmx_mm256_pmecorrV_ps(zeta2);
 407             pmecorrV         = _mm256_mul_ps(pmecorrV,beta);
 408             velec            = _mm256_sub_ps(rinv12,pmecorrV);
 409             velec            = _mm256_mul_ps(qq12,velec);
 410
 411             d                = _mm256_sub_ps(r12,rswitch);
 412             d                = _mm256_max_ps(d,_mm256_setzero_ps());
 413             d2               = _mm256_mul_ps(d,d);
 414             sw               = _mm256_add_ps(one,_mm256_mul_ps(d2,_mm256_mul_ps(d,_mm256_add_ps(swV3,_mm256_mul_ps(d,_mm256_add_ps(swV4,_mm256_mul_ps(d,swV5)))))));
 415
 416             dsw              = _mm256_mul_ps(d2,_mm256_add_ps(swF2,_mm256_mul_ps(d,_mm256_add_ps(swF3,_mm256_mul_ps(d,swF4)))));
 417
 418             /* Evaluate switch function */
 419             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 420             felec            = _mm256_sub_ps( _mm256_mul_ps(felec,sw) , _mm256_mul_ps(rinv12,_mm256_mul_ps(velec,dsw)) );
 421             velec            = _mm256_mul_ps(velec,sw);
 422             cutoff_mask      = _mm256_cmp_ps(rsq12,rcutoff2,_CMP_LT_OQ);
 423
 424             /* Update potential sum for this i atom from the interaction with this j atom. */
 425             velec            = _mm256_and_ps(velec,cutoff_mask);
 426             velecsum         = _mm256_add_ps(velecsum,velec);
 427
 428             fscal            = felec;
 429
 430             fscal            = _mm256_and_ps(fscal,cutoff_mask);
 431
 432             /* Calculate temporary vectorial force */
 433             tx               = _mm256_mul_ps(fscal,dx12);
 434             ty               = _mm256_mul_ps(fscal,dy12);
 435             tz               = _mm256_mul_ps(fscal,dz12);
 436
 437             /* Update vectorial force */
 438             fix1             = _mm256_add_ps(fix1,tx);
 439             fiy1             = _mm256_add_ps(fiy1,ty);
 440             fiz1             = _mm256_add_ps(fiz1,tz);
 441
 442             fjx2             = _mm256_add_ps(fjx2,tx);
 443             fjy2             = _mm256_add_ps(fjy2,ty);
 444             fjz2             = _mm256_add_ps(fjz2,tz);
 445
 446             }
 447
 448             /**************************
 449              * CALCULATE INTERACTIONS *
 450              **************************/
 451
 452             if (gmx_mm256_any_lt(rsq13,rcutoff2))
 453             {
 454
 455             r13              = _mm256_mul_ps(rsq13,rinv13);
 456
 457             /* EWALD ELECTROSTATICS */
 458
 459             /* Analytical PME correction */
 460             zeta2            = _mm256_mul_ps(beta2,rsq13);
 461             rinv3            = _mm256_mul_ps(rinvsq13,rinv13);
 462             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
 463             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
 464             felec            = _mm256_mul_ps(qq13,felec);
 465             pmecorrV         = gmx_mm256_pmecorrV_ps(zeta2);
 466             pmecorrV         = _mm256_mul_ps(pmecorrV,beta);
 467             velec            = _mm256_sub_ps(rinv13,pmecorrV);
 468             velec            = _mm256_mul_ps(qq13,velec);
 469
 470             d                = _mm256_sub_ps(r13,rswitch);
 471             d                = _mm256_max_ps(d,_mm256_setzero_ps());
 472             d2               = _mm256_mul_ps(d,d);
 473             sw               = _mm256_add_ps(one,_mm256_mul_ps(d2,_mm256_mul_ps(d,_mm256_add_ps(swV3,_mm256_mul_ps(d,_mm256_add_ps(swV4,_mm256_mul_ps(d,swV5)))))));
 474
 475             dsw              = _mm256_mul_ps(d2,_mm256_add_ps(swF2,_mm256_mul_ps(d,_mm256_add_ps(swF3,_mm256_mul_ps(d,swF4)))));
 476
 477             /* Evaluate switch function */
 478             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 479             felec            = _mm256_sub_ps( _mm256_mul_ps(felec,sw) , _mm256_mul_ps(rinv13,_mm256_mul_ps(velec,dsw)) );
 480             velec            = _mm256_mul_ps(velec,sw);
 481             cutoff_mask      = _mm256_cmp_ps(rsq13,rcutoff2,_CMP_LT_OQ);
 482
 483             /* Update potential sum for this i atom from the interaction with this j atom. */
 484             velec            = _mm256_and_ps(velec,cutoff_mask);
 485             velecsum         = _mm256_add_ps(velecsum,velec);
 486
 487             fscal            = felec;
 488
 489             fscal            = _mm256_and_ps(fscal,cutoff_mask);
 490
 491             /* Calculate temporary vectorial force */
 492             tx               = _mm256_mul_ps(fscal,dx13);
 493             ty               = _mm256_mul_ps(fscal,dy13);
 494             tz               = _mm256_mul_ps(fscal,dz13);
 495
 496             /* Update vectorial force */
 497             fix1             = _mm256_add_ps(fix1,tx);
 498             fiy1             = _mm256_add_ps(fiy1,ty);
 499             fiz1             = _mm256_add_ps(fiz1,tz);
 500
 501             fjx3             = _mm256_add_ps(fjx3,tx);
 502             fjy3             = _mm256_add_ps(fjy3,ty);
 503             fjz3             = _mm256_add_ps(fjz3,tz);
 504
 505             }
 506
 507             /**************************
 508              * CALCULATE INTERACTIONS *
 509              **************************/
 510
 511             if (gmx_mm256_any_lt(rsq21,rcutoff2))
 512             {
 513
 514             r21              = _mm256_mul_ps(rsq21,rinv21);
 515
 516             /* EWALD ELECTROSTATICS */
 517
 518             /* Analytical PME correction */
 519             zeta2            = _mm256_mul_ps(beta2,rsq21);
 520             rinv3            = _mm256_mul_ps(rinvsq21,rinv21);
 521             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
 522             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
 523             felec            = _mm256_mul_ps(qq21,felec);
 524             pmecorrV         = gmx_mm256_pmecorrV_ps(zeta2);
 525             pmecorrV         = _mm256_mul_ps(pmecorrV,beta);
 526             velec            = _mm256_sub_ps(rinv21,pmecorrV);
 527             velec            = _mm256_mul_ps(qq21,velec);
 528
 529             d                = _mm256_sub_ps(r21,rswitch);
 530             d                = _mm256_max_ps(d,_mm256_setzero_ps());
 531             d2               = _mm256_mul_ps(d,d);
 532             sw               = _mm256_add_ps(one,_mm256_mul_ps(d2,_mm256_mul_ps(d,_mm256_add_ps(swV3,_mm256_mul_ps(d,_mm256_add_ps(swV4,_mm256_mul_ps(d,swV5)))))));
 533
 534             dsw              = _mm256_mul_ps(d2,_mm256_add_ps(swF2,_mm256_mul_ps(d,_mm256_add_ps(swF3,_mm256_mul_ps(d,swF4)))));
 535
 536             /* Evaluate switch function */
 537             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 538             felec            = _mm256_sub_ps( _mm256_mul_ps(felec,sw) , _mm256_mul_ps(rinv21,_mm256_mul_ps(velec,dsw)) );
 539             velec            = _mm256_mul_ps(velec,sw);
 540             cutoff_mask      = _mm256_cmp_ps(rsq21,rcutoff2,_CMP_LT_OQ);
 541
 542             /* Update potential sum for this i atom from the interaction with this j atom. */
 543             velec            = _mm256_and_ps(velec,cutoff_mask);
 544             velecsum         = _mm256_add_ps(velecsum,velec);
 545
 546             fscal            = felec;
 547
 548             fscal            = _mm256_and_ps(fscal,cutoff_mask);
 549
 550             /* Calculate temporary vectorial force */
 551             tx               = _mm256_mul_ps(fscal,dx21);
 552             ty               = _mm256_mul_ps(fscal,dy21);
 553             tz               = _mm256_mul_ps(fscal,dz21);
 554
 555             /* Update vectorial force */
 556             fix2             = _mm256_add_ps(fix2,tx);
 557             fiy2             = _mm256_add_ps(fiy2,ty);
 558             fiz2             = _mm256_add_ps(fiz2,tz);
 559
 560             fjx1             = _mm256_add_ps(fjx1,tx);
 561             fjy1             = _mm256_add_ps(fjy1,ty);
 562             fjz1             = _mm256_add_ps(fjz1,tz);
 563
 564             }
 565
 566             /**************************
 567              * CALCULATE INTERACTIONS *
 568              **************************/
 569
 570             if (gmx_mm256_any_lt(rsq22,rcutoff2))
 571             {
 572
 573             r22              = _mm256_mul_ps(rsq22,rinv22);
 574
 575             /* EWALD ELECTROSTATICS */
 576
 577             /* Analytical PME correction */
 578             zeta2            = _mm256_mul_ps(beta2,rsq22);
 579             rinv3            = _mm256_mul_ps(rinvsq22,rinv22);
 580             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
 581             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
 582             felec            = _mm256_mul_ps(qq22,felec);
 583             pmecorrV         = gmx_mm256_pmecorrV_ps(zeta2);
 584             pmecorrV         = _mm256_mul_ps(pmecorrV,beta);
 585             velec            = _mm256_sub_ps(rinv22,pmecorrV);
 586             velec            = _mm256_mul_ps(qq22,velec);
 587
 588             d                = _mm256_sub_ps(r22,rswitch);
 589             d                = _mm256_max_ps(d,_mm256_setzero_ps());
 590             d2               = _mm256_mul_ps(d,d);
 591             sw               = _mm256_add_ps(one,_mm256_mul_ps(d2,_mm256_mul_ps(d,_mm256_add_ps(swV3,_mm256_mul_ps(d,_mm256_add_ps(swV4,_mm256_mul_ps(d,swV5)))))));
 592
 593             dsw              = _mm256_mul_ps(d2,_mm256_add_ps(swF2,_mm256_mul_ps(d,_mm256_add_ps(swF3,_mm256_mul_ps(d,swF4)))));
 594
 595             /* Evaluate switch function */
 596             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 597             felec            = _mm256_sub_ps( _mm256_mul_ps(felec,sw) , _mm256_mul_ps(rinv22,_mm256_mul_ps(velec,dsw)) );
 598             velec            = _mm256_mul_ps(velec,sw);
 599             cutoff_mask      = _mm256_cmp_ps(rsq22,rcutoff2,_CMP_LT_OQ);
 600
 601             /* Update potential sum for this i atom from the interaction with this j atom. */
 602             velec            = _mm256_and_ps(velec,cutoff_mask);
 603             velecsum         = _mm256_add_ps(velecsum,velec);
 604
 605             fscal            = felec;
 606
 607             fscal            = _mm256_and_ps(fscal,cutoff_mask);
 608
 609             /* Calculate temporary vectorial force */
 610             tx               = _mm256_mul_ps(fscal,dx22);
 611             ty               = _mm256_mul_ps(fscal,dy22);
 612             tz               = _mm256_mul_ps(fscal,dz22);
 613
 614             /* Update vectorial force */
 615             fix2             = _mm256_add_ps(fix2,tx);
 616             fiy2             = _mm256_add_ps(fiy2,ty);
 617             fiz2             = _mm256_add_ps(fiz2,tz);
 618
 619             fjx2             = _mm256_add_ps(fjx2,tx);
 620             fjy2             = _mm256_add_ps(fjy2,ty);
 621             fjz2             = _mm256_add_ps(fjz2,tz);
 622
 623             }
 624
 625             /**************************
 626              * CALCULATE INTERACTIONS *
 627              **************************/
 628
 629             if (gmx_mm256_any_lt(rsq23,rcutoff2))
 630             {
 631
 632             r23              = _mm256_mul_ps(rsq23,rinv23);
 633
 634             /* EWALD ELECTROSTATICS */
 635
 636             /* Analytical PME correction */
 637             zeta2            = _mm256_mul_ps(beta2,rsq23);
 638             rinv3            = _mm256_mul_ps(rinvsq23,rinv23);
 639             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
 640             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
 641             felec            = _mm256_mul_ps(qq23,felec);
 642             pmecorrV         = gmx_mm256_pmecorrV_ps(zeta2);
 643             pmecorrV         = _mm256_mul_ps(pmecorrV,beta);
 644             velec            = _mm256_sub_ps(rinv23,pmecorrV);
 645             velec            = _mm256_mul_ps(qq23,velec);
 646
 647             d                = _mm256_sub_ps(r23,rswitch);
 648             d                = _mm256_max_ps(d,_mm256_setzero_ps());
 649             d2               = _mm256_mul_ps(d,d);
 650             sw               = _mm256_add_ps(one,_mm256_mul_ps(d2,_mm256_mul_ps(d,_mm256_add_ps(swV3,_mm256_mul_ps(d,_mm256_add_ps(swV4,_mm256_mul_ps(d,swV5)))))));
 651
 652             dsw              = _mm256_mul_ps(d2,_mm256_add_ps(swF2,_mm256_mul_ps(d,_mm256_add_ps(swF3,_mm256_mul_ps(d,swF4)))));
 653
 654             /* Evaluate switch function */
 655             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 656             felec            = _mm256_sub_ps( _mm256_mul_ps(felec,sw) , _mm256_mul_ps(rinv23,_mm256_mul_ps(velec,dsw)) );
 657             velec            = _mm256_mul_ps(velec,sw);
 658             cutoff_mask      = _mm256_cmp_ps(rsq23,rcutoff2,_CMP_LT_OQ);
 659
 660             /* Update potential sum for this i atom from the interaction with this j atom. */
 661             velec            = _mm256_and_ps(velec,cutoff_mask);
 662             velecsum         = _mm256_add_ps(velecsum,velec);
 663
 664             fscal            = felec;
 665
 666             fscal            = _mm256_and_ps(fscal,cutoff_mask);
 667
 668             /* Calculate temporary vectorial force */
 669             tx               = _mm256_mul_ps(fscal,dx23);
 670             ty               = _mm256_mul_ps(fscal,dy23);
 671             tz               = _mm256_mul_ps(fscal,dz23);
 672
 673             /* Update vectorial force */
 674             fix2             = _mm256_add_ps(fix2,tx);
 675             fiy2             = _mm256_add_ps(fiy2,ty);
 676             fiz2             = _mm256_add_ps(fiz2,tz);
 677
 678             fjx3             = _mm256_add_ps(fjx3,tx);
 679             fjy3             = _mm256_add_ps(fjy3,ty);
 680             fjz3             = _mm256_add_ps(fjz3,tz);
 681
 682             }
 683
 684             /**************************
 685              * CALCULATE INTERACTIONS *
 686              **************************/
 687
 688             if (gmx_mm256_any_lt(rsq31,rcutoff2))
 689             {
 690
 691             r31              = _mm256_mul_ps(rsq31,rinv31);
 692
 693             /* EWALD ELECTROSTATICS */
 694
 695             /* Analytical PME correction */
 696             zeta2            = _mm256_mul_ps(beta2,rsq31);
 697             rinv3            = _mm256_mul_ps(rinvsq31,rinv31);
 698             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
 699             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
 700             felec            = _mm256_mul_ps(qq31,felec);
 701             pmecorrV         = gmx_mm256_pmecorrV_ps(zeta2);
 702             pmecorrV         = _mm256_mul_ps(pmecorrV,beta);
 703             velec            = _mm256_sub_ps(rinv31,pmecorrV);
 704             velec            = _mm256_mul_ps(qq31,velec);
 705
 706             d                = _mm256_sub_ps(r31,rswitch);
 707             d                = _mm256_max_ps(d,_mm256_setzero_ps());
 708             d2               = _mm256_mul_ps(d,d);
 709             sw               = _mm256_add_ps(one,_mm256_mul_ps(d2,_mm256_mul_ps(d,_mm256_add_ps(swV3,_mm256_mul_ps(d,_mm256_add_ps(swV4,_mm256_mul_ps(d,swV5)))))));
 710
 711             dsw              = _mm256_mul_ps(d2,_mm256_add_ps(swF2,_mm256_mul_ps(d,_mm256_add_ps(swF3,_mm256_mul_ps(d,swF4)))));
 712
 713             /* Evaluate switch function */
 714             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 715             felec            = _mm256_sub_ps( _mm256_mul_ps(felec,sw) , _mm256_mul_ps(rinv31,_mm256_mul_ps(velec,dsw)) );
 716             velec            = _mm256_mul_ps(velec,sw);
 717             cutoff_mask      = _mm256_cmp_ps(rsq31,rcutoff2,_CMP_LT_OQ);
 718
 719             /* Update potential sum for this i atom from the interaction with this j atom. */
 720             velec            = _mm256_and_ps(velec,cutoff_mask);
 721             velecsum         = _mm256_add_ps(velecsum,velec);
 722
 723             fscal            = felec;
 724
 725             fscal            = _mm256_and_ps(fscal,cutoff_mask);
 726
 727             /* Calculate temporary vectorial force */
 728             tx               = _mm256_mul_ps(fscal,dx31);
 729             ty               = _mm256_mul_ps(fscal,dy31);
 730             tz               = _mm256_mul_ps(fscal,dz31);
 731
 732             /* Update vectorial force */
 733             fix3             = _mm256_add_ps(fix3,tx);
 734             fiy3             = _mm256_add_ps(fiy3,ty);
 735             fiz3             = _mm256_add_ps(fiz3,tz);
 736
 737             fjx1             = _mm256_add_ps(fjx1,tx);
 738             fjy1             = _mm256_add_ps(fjy1,ty);
 739             fjz1             = _mm256_add_ps(fjz1,tz);
 740
 741             }
 742
 743             /**************************
 744              * CALCULATE INTERACTIONS *
 745              **************************/
 746
 747             if (gmx_mm256_any_lt(rsq32,rcutoff2))
 748             {
 749
 750             r32              = _mm256_mul_ps(rsq32,rinv32);
 751
 752             /* EWALD ELECTROSTATICS */
 753
 754             /* Analytical PME correction */
 755             zeta2            = _mm256_mul_ps(beta2,rsq32);
 756             rinv3            = _mm256_mul_ps(rinvsq32,rinv32);
 757             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
 758             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
 759             felec            = _mm256_mul_ps(qq32,felec);
 760             pmecorrV         = gmx_mm256_pmecorrV_ps(zeta2);
 761             pmecorrV         = _mm256_mul_ps(pmecorrV,beta);
 762             velec            = _mm256_sub_ps(rinv32,pmecorrV);
 763             velec            = _mm256_mul_ps(qq32,velec);
 764
 765             d                = _mm256_sub_ps(r32,rswitch);
 766             d                = _mm256_max_ps(d,_mm256_setzero_ps());
 767             d2               = _mm256_mul_ps(d,d);
 768             sw               = _mm256_add_ps(one,_mm256_mul_ps(d2,_mm256_mul_ps(d,_mm256_add_ps(swV3,_mm256_mul_ps(d,_mm256_add_ps(swV4,_mm256_mul_ps(d,swV5)))))));
 769
 770             dsw              = _mm256_mul_ps(d2,_mm256_add_ps(swF2,_mm256_mul_ps(d,_mm256_add_ps(swF3,_mm256_mul_ps(d,swF4)))));
 771
 772             /* Evaluate switch function */
 773             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 774             felec            = _mm256_sub_ps( _mm256_mul_ps(felec,sw) , _mm256_mul_ps(rinv32,_mm256_mul_ps(velec,dsw)) );
 775             velec            = _mm256_mul_ps(velec,sw);
 776             cutoff_mask      = _mm256_cmp_ps(rsq32,rcutoff2,_CMP_LT_OQ);
 777
 778             /* Update potential sum for this i atom from the interaction with this j atom. */
 779             velec            = _mm256_and_ps(velec,cutoff_mask);
 780             velecsum         = _mm256_add_ps(velecsum,velec);
 781
 782             fscal            = felec;
 783
 784             fscal            = _mm256_and_ps(fscal,cutoff_mask);
 785
 786             /* Calculate temporary vectorial force */
 787             tx               = _mm256_mul_ps(fscal,dx32);
 788             ty               = _mm256_mul_ps(fscal,dy32);
 789             tz               = _mm256_mul_ps(fscal,dz32);
 790
 791             /* Update vectorial force */
 792             fix3             = _mm256_add_ps(fix3,tx);
 793             fiy3             = _mm256_add_ps(fiy3,ty);
 794             fiz3             = _mm256_add_ps(fiz3,tz);
 795
 796             fjx2             = _mm256_add_ps(fjx2,tx);
 797             fjy2             = _mm256_add_ps(fjy2,ty);
 798             fjz2             = _mm256_add_ps(fjz2,tz);
 799
 800             }
 801
 802             /**************************
 803              * CALCULATE INTERACTIONS *
 804              **************************/
 805
 806             if (gmx_mm256_any_lt(rsq33,rcutoff2))
 807             {
 808
 809             r33              = _mm256_mul_ps(rsq33,rinv33);
 810
 811             /* EWALD ELECTROSTATICS */
 812
 813             /* Analytical PME correction */
 814             zeta2            = _mm256_mul_ps(beta2,rsq33);
 815             rinv3            = _mm256_mul_ps(rinvsq33,rinv33);
 816             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
 817             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
 818             felec            = _mm256_mul_ps(qq33,felec);
 819             pmecorrV         = gmx_mm256_pmecorrV_ps(zeta2);
 820             pmecorrV         = _mm256_mul_ps(pmecorrV,beta);
 821             velec            = _mm256_sub_ps(rinv33,pmecorrV);
 822             velec            = _mm256_mul_ps(qq33,velec);
 823
 824             d                = _mm256_sub_ps(r33,rswitch);
 825             d                = _mm256_max_ps(d,_mm256_setzero_ps());
 826             d2               = _mm256_mul_ps(d,d);
 827             sw               = _mm256_add_ps(one,_mm256_mul_ps(d2,_mm256_mul_ps(d,_mm256_add_ps(swV3,_mm256_mul_ps(d,_mm256_add_ps(swV4,_mm256_mul_ps(d,swV5)))))));
 828
 829             dsw              = _mm256_mul_ps(d2,_mm256_add_ps(swF2,_mm256_mul_ps(d,_mm256_add_ps(swF3,_mm256_mul_ps(d,swF4)))));
 830
 831             /* Evaluate switch function */
 832             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 833             felec            = _mm256_sub_ps( _mm256_mul_ps(felec,sw) , _mm256_mul_ps(rinv33,_mm256_mul_ps(velec,dsw)) );
 834             velec            = _mm256_mul_ps(velec,sw);
 835             cutoff_mask      = _mm256_cmp_ps(rsq33,rcutoff2,_CMP_LT_OQ);
 836
 837             /* Update potential sum for this i atom from the interaction with this j atom. */
 838             velec            = _mm256_and_ps(velec,cutoff_mask);
 839             velecsum         = _mm256_add_ps(velecsum,velec);
 840
 841             fscal            = felec;
 842
 843             fscal            = _mm256_and_ps(fscal,cutoff_mask);
 844
 845             /* Calculate temporary vectorial force */
 846             tx               = _mm256_mul_ps(fscal,dx33);
 847             ty               = _mm256_mul_ps(fscal,dy33);
 848             tz               = _mm256_mul_ps(fscal,dz33);
 849
 850             /* Update vectorial force */
 851             fix3             = _mm256_add_ps(fix3,tx);
 852             fiy3             = _mm256_add_ps(fiy3,ty);
 853             fiz3             = _mm256_add_ps(fiz3,tz);
 854
 855             fjx3             = _mm256_add_ps(fjx3,tx);
 856             fjy3             = _mm256_add_ps(fjy3,ty);
 857             fjz3             = _mm256_add_ps(fjz3,tz);
 858
 859             }
 860
 861             fjptrA             = f+j_coord_offsetA;
 862             fjptrB             = f+j_coord_offsetB;
 863             fjptrC             = f+j_coord_offsetC;
 864             fjptrD             = f+j_coord_offsetD;
 865             fjptrE             = f+j_coord_offsetE;
 866             fjptrF             = f+j_coord_offsetF;
 867             fjptrG             = f+j_coord_offsetG;
 868             fjptrH             = f+j_coord_offsetH;
 869
 870             gmx_mm256_decrement_3rvec_8ptr_swizzle_ps(fjptrA+DIM,fjptrB+DIM,fjptrC+DIM,fjptrD+DIM,
 871                                                       fjptrE+DIM,fjptrF+DIM,fjptrG+DIM,fjptrH+DIM,
 872                                                       fjx1,fjy1,fjz1,fjx2,fjy2,fjz2,fjx3,fjy3,fjz3);
 873
 874             /* Inner loop uses 972 flops */
 875         }
 876
 877         if(jidx<j_index_end)
 878         {
 879
 880             /* Get j neighbor index, and coordinate index */
 881             jnrlistA         = jjnr[jidx];
 882             jnrlistB         = jjnr[jidx+1];
 883             jnrlistC         = jjnr[jidx+2];
 884             jnrlistD         = jjnr[jidx+3];
 885             jnrlistE         = jjnr[jidx+4];
 886             jnrlistF         = jjnr[jidx+5];
 887             jnrlistG         = jjnr[jidx+6];
 888             jnrlistH         = jjnr[jidx+7];
 889             /* Sign of each element will be negative for non-real atoms.
 890              * This mask will be 0xFFFFFFFF for dummy entries and 0x0 for real ones,
 891              * so use it as val = _mm_andnot_ps(mask,val) to clear dummy entries.
 892              */
 893             dummy_mask = gmx_mm256_set_m128(gmx_mm_castsi128_ps(_mm_cmplt_epi32(_mm_loadu_si128((const __m128i *)(jjnr+jidx+4)),_mm_setzero_si128())),
 894                                             gmx_mm_castsi128_ps(_mm_cmplt_epi32(_mm_loadu_si128((const __m128i *)(jjnr+jidx)),_mm_setzero_si128())));
 895
 896             jnrA       = (jnrlistA>=0) ? jnrlistA : 0;
 897             jnrB       = (jnrlistB>=0) ? jnrlistB : 0;
 898             jnrC       = (jnrlistC>=0) ? jnrlistC : 0;
 899             jnrD       = (jnrlistD>=0) ? jnrlistD : 0;
 900             jnrE       = (jnrlistE>=0) ? jnrlistE : 0;
 901             jnrF       = (jnrlistF>=0) ? jnrlistF : 0;
 902             jnrG       = (jnrlistG>=0) ? jnrlistG : 0;
 903             jnrH       = (jnrlistH>=0) ? jnrlistH : 0;
 904             j_coord_offsetA  = DIM*jnrA;
 905             j_coord_offsetB  = DIM*jnrB;
 906             j_coord_offsetC  = DIM*jnrC;
 907             j_coord_offsetD  = DIM*jnrD;
 908             j_coord_offsetE  = DIM*jnrE;
 909             j_coord_offsetF  = DIM*jnrF;
 910             j_coord_offsetG  = DIM*jnrG;
 911             j_coord_offsetH  = DIM*jnrH;
 912
 913             /* load j atom coordinates */
 914             gmx_mm256_load_3rvec_8ptr_swizzle_ps(x+j_coord_offsetA+DIM,x+j_coord_offsetB+DIM,
 915                                                  x+j_coord_offsetC+DIM,x+j_coord_offsetD+DIM,
 916                                                  x+j_coord_offsetE+DIM,x+j_coord_offsetF+DIM,
 917                                                  x+j_coord_offsetG+DIM,x+j_coord_offsetH+DIM,
 918                                                  &jx1,&jy1,&jz1,&jx2,&jy2,&jz2,&jx3,&jy3,&jz3);
 919
 920             /* Calculate displacement vector */
 921             dx11             = _mm256_sub_ps(ix1,jx1);
 922             dy11             = _mm256_sub_ps(iy1,jy1);
 923             dz11             = _mm256_sub_ps(iz1,jz1);
 924             dx12             = _mm256_sub_ps(ix1,jx2);
 925             dy12             = _mm256_sub_ps(iy1,jy2);
 926             dz12             = _mm256_sub_ps(iz1,jz2);
 927             dx13             = _mm256_sub_ps(ix1,jx3);
 928             dy13             = _mm256_sub_ps(iy1,jy3);
 929             dz13             = _mm256_sub_ps(iz1,jz3);
 930             dx21             = _mm256_sub_ps(ix2,jx1);
 931             dy21             = _mm256_sub_ps(iy2,jy1);
 932             dz21             = _mm256_sub_ps(iz2,jz1);
 933             dx22             = _mm256_sub_ps(ix2,jx2);
 934             dy22             = _mm256_sub_ps(iy2,jy2);
 935             dz22             = _mm256_sub_ps(iz2,jz2);
 936             dx23             = _mm256_sub_ps(ix2,jx3);
 937             dy23             = _mm256_sub_ps(iy2,jy3);
 938             dz23             = _mm256_sub_ps(iz2,jz3);
 939             dx31             = _mm256_sub_ps(ix3,jx1);
 940             dy31             = _mm256_sub_ps(iy3,jy1);
 941             dz31             = _mm256_sub_ps(iz3,jz1);
 942             dx32             = _mm256_sub_ps(ix3,jx2);
 943             dy32             = _mm256_sub_ps(iy3,jy2);
 944             dz32             = _mm256_sub_ps(iz3,jz2);
 945             dx33             = _mm256_sub_ps(ix3,jx3);
 946             dy33             = _mm256_sub_ps(iy3,jy3);
 947             dz33             = _mm256_sub_ps(iz3,jz3);
 948
 949             /* Calculate squared distance and things based on it */
 950             rsq11            = gmx_mm256_calc_rsq_ps(dx11,dy11,dz11);
 951             rsq12            = gmx_mm256_calc_rsq_ps(dx12,dy12,dz12);
 952             rsq13            = gmx_mm256_calc_rsq_ps(dx13,dy13,dz13);
 953             rsq21            = gmx_mm256_calc_rsq_ps(dx21,dy21,dz21);
 954             rsq22            = gmx_mm256_calc_rsq_ps(dx22,dy22,dz22);
 955             rsq23            = gmx_mm256_calc_rsq_ps(dx23,dy23,dz23);
 956             rsq31            = gmx_mm256_calc_rsq_ps(dx31,dy31,dz31);
 957             rsq32            = gmx_mm256_calc_rsq_ps(dx32,dy32,dz32);
 958             rsq33            = gmx_mm256_calc_rsq_ps(dx33,dy33,dz33);
 959
 960             rinv11           = gmx_mm256_invsqrt_ps(rsq11);
 961             rinv12           = gmx_mm256_invsqrt_ps(rsq12);
 962             rinv13           = gmx_mm256_invsqrt_ps(rsq13);
 963             rinv21           = gmx_mm256_invsqrt_ps(rsq21);
 964             rinv22           = gmx_mm256_invsqrt_ps(rsq22);
 965             rinv23           = gmx_mm256_invsqrt_ps(rsq23);
 966             rinv31           = gmx_mm256_invsqrt_ps(rsq31);
 967             rinv32           = gmx_mm256_invsqrt_ps(rsq32);
 968             rinv33           = gmx_mm256_invsqrt_ps(rsq33);
 969
 970             rinvsq11         = _mm256_mul_ps(rinv11,rinv11);
 971             rinvsq12         = _mm256_mul_ps(rinv12,rinv12);
 972             rinvsq13         = _mm256_mul_ps(rinv13,rinv13);
 973             rinvsq21         = _mm256_mul_ps(rinv21,rinv21);
 974             rinvsq22         = _mm256_mul_ps(rinv22,rinv22);
 975             rinvsq23         = _mm256_mul_ps(rinv23,rinv23);
 976             rinvsq31         = _mm256_mul_ps(rinv31,rinv31);
 977             rinvsq32         = _mm256_mul_ps(rinv32,rinv32);
 978             rinvsq33         = _mm256_mul_ps(rinv33,rinv33);
 979
 980             fjx1             = _mm256_setzero_ps();
 981             fjy1             = _mm256_setzero_ps();
 982             fjz1             = _mm256_setzero_ps();
 983             fjx2             = _mm256_setzero_ps();
 984             fjy2             = _mm256_setzero_ps();
 985             fjz2             = _mm256_setzero_ps();
 986             fjx3             = _mm256_setzero_ps();
 987             fjy3             = _mm256_setzero_ps();
 988             fjz3             = _mm256_setzero_ps();
 989
 990             /**************************
 991              * CALCULATE INTERACTIONS *
 992              **************************/
 993
 994             if (gmx_mm256_any_lt(rsq11,rcutoff2))
 995             {
 996
 997             r11              = _mm256_mul_ps(rsq11,rinv11);
 998             r11              = _mm256_andnot_ps(dummy_mask,r11);
 999
1000             /* EWALD ELECTROSTATICS */
1001
1002             /* Analytical PME correction */
1003             zeta2            = _mm256_mul_ps(beta2,rsq11);
1004             rinv3            = _mm256_mul_ps(rinvsq11,rinv11);
1005             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
1006             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
1007             felec            = _mm256_mul_ps(qq11,felec);
1008             pmecorrV         = gmx_mm256_pmecorrV_ps(zeta2);
1009             pmecorrV         = _mm256_mul_ps(pmecorrV,beta);
1010             velec            = _mm256_sub_ps(rinv11,pmecorrV);
1011             velec            = _mm256_mul_ps(qq11,velec);
1012
1013             d                = _mm256_sub_ps(r11,rswitch);
1014             d                = _mm256_max_ps(d,_mm256_setzero_ps());
1015             d2               = _mm256_mul_ps(d,d);
1016             sw               = _mm256_add_ps(one,_mm256_mul_ps(d2,_mm256_mul_ps(d,_mm256_add_ps(swV3,_mm256_mul_ps(d,_mm256_add_ps(swV4,_mm256_mul_ps(d,swV5)))))));
1017
1018             dsw              = _mm256_mul_ps(d2,_mm256_add_ps(swF2,_mm256_mul_ps(d,_mm256_add_ps(swF3,_mm256_mul_ps(d,swF4)))));
1019
1020             /* Evaluate switch function */
1021             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
1022             felec            = _mm256_sub_ps( _mm256_mul_ps(felec,sw) , _mm256_mul_ps(rinv11,_mm256_mul_ps(velec,dsw)) );
1023             velec            = _mm256_mul_ps(velec,sw);
1024             cutoff_mask      = _mm256_cmp_ps(rsq11,rcutoff2,_CMP_LT_OQ);
1025
1026             /* Update potential sum for this i atom from the interaction with this j atom. */
1027             velec            = _mm256_and_ps(velec,cutoff_mask);
1028             velec            = _mm256_andnot_ps(dummy_mask,velec);
1029             velecsum         = _mm256_add_ps(velecsum,velec);
1030
1031             fscal            = felec;
1032
1033             fscal            = _mm256_and_ps(fscal,cutoff_mask);
1034
1035             fscal            = _mm256_andnot_ps(dummy_mask,fscal);
1036
1037             /* Calculate temporary vectorial force */
1038             tx               = _mm256_mul_ps(fscal,dx11);
1039             ty               = _mm256_mul_ps(fscal,dy11);
1040             tz               = _mm256_mul_ps(fscal,dz11);
1041
1042             /* Update vectorial force */
1043             fix1             = _mm256_add_ps(fix1,tx);
1044             fiy1             = _mm256_add_ps(fiy1,ty);
1045             fiz1             = _mm256_add_ps(fiz1,tz);
1046
1047             fjx1             = _mm256_add_ps(fjx1,tx);
1048             fjy1             = _mm256_add_ps(fjy1,ty);
1049             fjz1             = _mm256_add_ps(fjz1,tz);
1050
1051             }
1052
1053             /**************************
1054              * CALCULATE INTERACTIONS *
1055              **************************/
1056
1057             if (gmx_mm256_any_lt(rsq12,rcutoff2))
1058             {
1059
1060             r12              = _mm256_mul_ps(rsq12,rinv12);
1061             r12              = _mm256_andnot_ps(dummy_mask,r12);
1062
1063             /* EWALD ELECTROSTATICS */
1064
1065             /* Analytical PME correction */
1066             zeta2            = _mm256_mul_ps(beta2,rsq12);
1067             rinv3            = _mm256_mul_ps(rinvsq12,rinv12);
1068             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
1069             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
1070             felec            = _mm256_mul_ps(qq12,felec);
1071             pmecorrV         = gmx_mm256_pmecorrV_ps(zeta2);
1072             pmecorrV         = _mm256_mul_ps(pmecorrV,beta);
1073             velec            = _mm256_sub_ps(rinv12,pmecorrV);
1074             velec            = _mm256_mul_ps(qq12,velec);
1075
1076             d                = _mm256_sub_ps(r12,rswitch);
1077             d                = _mm256_max_ps(d,_mm256_setzero_ps());
1078             d2               = _mm256_mul_ps(d,d);
1079             sw               = _mm256_add_ps(one,_mm256_mul_ps(d2,_mm256_mul_ps(d,_mm256_add_ps(swV3,_mm256_mul_ps(d,_mm256_add_ps(swV4,_mm256_mul_ps(d,swV5)))))));
1080
1081             dsw              = _mm256_mul_ps(d2,_mm256_add_ps(swF2,_mm256_mul_ps(d,_mm256_add_ps(swF3,_mm256_mul_ps(d,swF4)))));
1082
1083             /* Evaluate switch function */
1084             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
1085             felec            = _mm256_sub_ps( _mm256_mul_ps(felec,sw) , _mm256_mul_ps(rinv12,_mm256_mul_ps(velec,dsw)) );
1086             velec            = _mm256_mul_ps(velec,sw);
1087             cutoff_mask      = _mm256_cmp_ps(rsq12,rcutoff2,_CMP_LT_OQ);
1088
1089             /* Update potential sum for this i atom from the interaction with this j atom. */
1090             velec            = _mm256_and_ps(velec,cutoff_mask);
1091             velec            = _mm256_andnot_ps(dummy_mask,velec);
1092             velecsum         = _mm256_add_ps(velecsum,velec);
1093
1094             fscal            = felec;
1095
1096             fscal            = _mm256_and_ps(fscal,cutoff_mask);
1097
1098             fscal            = _mm256_andnot_ps(dummy_mask,fscal);
1099
1100             /* Calculate temporary vectorial force */
1101             tx               = _mm256_mul_ps(fscal,dx12);
1102             ty               = _mm256_mul_ps(fscal,dy12);
1103             tz               = _mm256_mul_ps(fscal,dz12);
1104
1105             /* Update vectorial force */
1106             fix1             = _mm256_add_ps(fix1,tx);
1107             fiy1             = _mm256_add_ps(fiy1,ty);
1108             fiz1             = _mm256_add_ps(fiz1,tz);
1109
1110             fjx2             = _mm256_add_ps(fjx2,tx);
1111             fjy2             = _mm256_add_ps(fjy2,ty);
1112             fjz2             = _mm256_add_ps(fjz2,tz);
1113
1114             }
1115
1116             /**************************
1117              * CALCULATE INTERACTIONS *
1118              **************************/
1119
1120             if (gmx_mm256_any_lt(rsq13,rcutoff2))
1121             {
1122
1123             r13              = _mm256_mul_ps(rsq13,rinv13);
1124             r13              = _mm256_andnot_ps(dummy_mask,r13);
1125
1126             /* EWALD ELECTROSTATICS */
1127
1128             /* Analytical PME correction */
1129             zeta2            = _mm256_mul_ps(beta2,rsq13);
1130             rinv3            = _mm256_mul_ps(rinvsq13,rinv13);
1131             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
1132             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
1133             felec            = _mm256_mul_ps(qq13,felec);
1134             pmecorrV         = gmx_mm256_pmecorrV_ps(zeta2);
1135             pmecorrV         = _mm256_mul_ps(pmecorrV,beta);
1136             velec            = _mm256_sub_ps(rinv13,pmecorrV);
1137             velec            = _mm256_mul_ps(qq13,velec);
1138
1139             d                = _mm256_sub_ps(r13,rswitch);
1140             d                = _mm256_max_ps(d,_mm256_setzero_ps());
1141             d2               = _mm256_mul_ps(d,d);
1142             sw               = _mm256_add_ps(one,_mm256_mul_ps(d2,_mm256_mul_ps(d,_mm256_add_ps(swV3,_mm256_mul_ps(d,_mm256_add_ps(swV4,_mm256_mul_ps(d,swV5)))))));
1143
1144             dsw              = _mm256_mul_ps(d2,_mm256_add_ps(swF2,_mm256_mul_ps(d,_mm256_add_ps(swF3,_mm256_mul_ps(d,swF4)))));
1145
1146             /* Evaluate switch function */
1147             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
1148             felec            = _mm256_sub_ps( _mm256_mul_ps(felec,sw) , _mm256_mul_ps(rinv13,_mm256_mul_ps(velec,dsw)) );
1149             velec            = _mm256_mul_ps(velec,sw);
1150             cutoff_mask      = _mm256_cmp_ps(rsq13,rcutoff2,_CMP_LT_OQ);
1151
1152             /* Update potential sum for this i atom from the interaction with this j atom. */
1153             velec            = _mm256_and_ps(velec,cutoff_mask);
1154             velec            = _mm256_andnot_ps(dummy_mask,velec);
1155             velecsum         = _mm256_add_ps(velecsum,velec);
1156
1157             fscal            = felec;
1158
1159             fscal            = _mm256_and_ps(fscal,cutoff_mask);
1160
1161             fscal            = _mm256_andnot_ps(dummy_mask,fscal);
1162
1163             /* Calculate temporary vectorial force */
1164             tx               = _mm256_mul_ps(fscal,dx13);
1165             ty               = _mm256_mul_ps(fscal,dy13);
1166             tz               = _mm256_mul_ps(fscal,dz13);
1167
1168             /* Update vectorial force */
1169             fix1             = _mm256_add_ps(fix1,tx);
1170             fiy1             = _mm256_add_ps(fiy1,ty);
1171             fiz1             = _mm256_add_ps(fiz1,tz);
1172
1173             fjx3             = _mm256_add_ps(fjx3,tx);
1174             fjy3             = _mm256_add_ps(fjy3,ty);
1175             fjz3             = _mm256_add_ps(fjz3,tz);
1176
1177             }
1178
1179             /**************************
1180              * CALCULATE INTERACTIONS *
1181              **************************/
1182
1183             if (gmx_mm256_any_lt(rsq21,rcutoff2))
1184             {
1185
1186             r21              = _mm256_mul_ps(rsq21,rinv21);
1187             r21              = _mm256_andnot_ps(dummy_mask,r21);
1188
1189             /* EWALD ELECTROSTATICS */
1190
1191             /* Analytical PME correction */
1192             zeta2            = _mm256_mul_ps(beta2,rsq21);
1193             rinv3            = _mm256_mul_ps(rinvsq21,rinv21);
1194             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
1195             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
1196             felec            = _mm256_mul_ps(qq21,felec);
1197             pmecorrV         = gmx_mm256_pmecorrV_ps(zeta2);
1198             pmecorrV         = _mm256_mul_ps(pmecorrV,beta);
1199             velec            = _mm256_sub_ps(rinv21,pmecorrV);
1200             velec            = _mm256_mul_ps(qq21,velec);
1201
1202             d                = _mm256_sub_ps(r21,rswitch);
1203             d                = _mm256_max_ps(d,_mm256_setzero_ps());
1204             d2               = _mm256_mul_ps(d,d);
1205             sw               = _mm256_add_ps(one,_mm256_mul_ps(d2,_mm256_mul_ps(d,_mm256_add_ps(swV3,_mm256_mul_ps(d,_mm256_add_ps(swV4,_mm256_mul_ps(d,swV5)))))));
1206
1207             dsw              = _mm256_mul_ps(d2,_mm256_add_ps(swF2,_mm256_mul_ps(d,_mm256_add_ps(swF3,_mm256_mul_ps(d,swF4)))));
1208
1209             /* Evaluate switch function */
1210             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
1211             felec            = _mm256_sub_ps( _mm256_mul_ps(felec,sw) , _mm256_mul_ps(rinv21,_mm256_mul_ps(velec,dsw)) );
1212             velec            = _mm256_mul_ps(velec,sw);
1213             cutoff_mask      = _mm256_cmp_ps(rsq21,rcutoff2,_CMP_LT_OQ);
1214
1215             /* Update potential sum for this i atom from the interaction with this j atom. */
1216             velec            = _mm256_and_ps(velec,cutoff_mask);
1217             velec            = _mm256_andnot_ps(dummy_mask,velec);
1218             velecsum         = _mm256_add_ps(velecsum,velec);
1219
1220             fscal            = felec;
1221
1222             fscal            = _mm256_and_ps(fscal,cutoff_mask);
1223
1224             fscal            = _mm256_andnot_ps(dummy_mask,fscal);
1225
1226             /* Calculate temporary vectorial force */
1227             tx               = _mm256_mul_ps(fscal,dx21);
1228             ty               = _mm256_mul_ps(fscal,dy21);
1229             tz               = _mm256_mul_ps(fscal,dz21);
1230
1231             /* Update vectorial force */
1232             fix2             = _mm256_add_ps(fix2,tx);
1233             fiy2             = _mm256_add_ps(fiy2,ty);
1234             fiz2             = _mm256_add_ps(fiz2,tz);
1235
1236             fjx1             = _mm256_add_ps(fjx1,tx);
1237             fjy1             = _mm256_add_ps(fjy1,ty);
1238             fjz1             = _mm256_add_ps(fjz1,tz);
1239
1240             }
1241
1242             /**************************
1243              * CALCULATE INTERACTIONS *
1244              **************************/
1245
1246             if (gmx_mm256_any_lt(rsq22,rcutoff2))
1247             {
1248
1249             r22              = _mm256_mul_ps(rsq22,rinv22);
1250             r22              = _mm256_andnot_ps(dummy_mask,r22);
1251
1252             /* EWALD ELECTROSTATICS */
1253
1254             /* Analytical PME correction */
1255             zeta2            = _mm256_mul_ps(beta2,rsq22);
1256             rinv3            = _mm256_mul_ps(rinvsq22,rinv22);
1257             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
1258             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
1259             felec            = _mm256_mul_ps(qq22,felec);
1260             pmecorrV         = gmx_mm256_pmecorrV_ps(zeta2);
1261             pmecorrV         = _mm256_mul_ps(pmecorrV,beta);
1262             velec            = _mm256_sub_ps(rinv22,pmecorrV);
1263             velec            = _mm256_mul_ps(qq22,velec);
1264
1265             d                = _mm256_sub_ps(r22,rswitch);
1266             d                = _mm256_max_ps(d,_mm256_setzero_ps());
1267             d2               = _mm256_mul_ps(d,d);
1268             sw               = _mm256_add_ps(one,_mm256_mul_ps(d2,_mm256_mul_ps(d,_mm256_add_ps(swV3,_mm256_mul_ps(d,_mm256_add_ps(swV4,_mm256_mul_ps(d,swV5)))))));
1269
1270             dsw              = _mm256_mul_ps(d2,_mm256_add_ps(swF2,_mm256_mul_ps(d,_mm256_add_ps(swF3,_mm256_mul_ps(d,swF4)))));
1271
1272             /* Evaluate switch function */
1273             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
1274             felec            = _mm256_sub_ps( _mm256_mul_ps(felec,sw) , _mm256_mul_ps(rinv22,_mm256_mul_ps(velec,dsw)) );
1275             velec            = _mm256_mul_ps(velec,sw);
1276             cutoff_mask      = _mm256_cmp_ps(rsq22,rcutoff2,_CMP_LT_OQ);
1277
1278             /* Update potential sum for this i atom from the interaction with this j atom. */
1279             velec            = _mm256_and_ps(velec,cutoff_mask);
1280             velec            = _mm256_andnot_ps(dummy_mask,velec);
1281             velecsum         = _mm256_add_ps(velecsum,velec);
1282
1283             fscal            = felec;
1284
1285             fscal            = _mm256_and_ps(fscal,cutoff_mask);
1286
1287             fscal            = _mm256_andnot_ps(dummy_mask,fscal);
1288
1289             /* Calculate temporary vectorial force */
1290             tx               = _mm256_mul_ps(fscal,dx22);
1291             ty               = _mm256_mul_ps(fscal,dy22);
1292             tz               = _mm256_mul_ps(fscal,dz22);
1293
1294             /* Update vectorial force */
1295             fix2             = _mm256_add_ps(fix2,tx);
1296             fiy2             = _mm256_add_ps(fiy2,ty);
1297             fiz2             = _mm256_add_ps(fiz2,tz);
1298
1299             fjx2             = _mm256_add_ps(fjx2,tx);
1300             fjy2             = _mm256_add_ps(fjy2,ty);
1301             fjz2             = _mm256_add_ps(fjz2,tz);
1302
1303             }
1304
1305             /**************************
1306              * CALCULATE INTERACTIONS *
1307              **************************/
1308
1309             if (gmx_mm256_any_lt(rsq23,rcutoff2))
1310             {
1311
1312             r23              = _mm256_mul_ps(rsq23,rinv23);
1313             r23              = _mm256_andnot_ps(dummy_mask,r23);
1314
1315             /* EWALD ELECTROSTATICS */
1316
1317             /* Analytical PME correction */
1318             zeta2            = _mm256_mul_ps(beta2,rsq23);
1319             rinv3            = _mm256_mul_ps(rinvsq23,rinv23);
1320             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
1321             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
1322             felec            = _mm256_mul_ps(qq23,felec);
1323             pmecorrV         = gmx_mm256_pmecorrV_ps(zeta2);
1324             pmecorrV         = _mm256_mul_ps(pmecorrV,beta);
1325             velec            = _mm256_sub_ps(rinv23,pmecorrV);
1326             velec            = _mm256_mul_ps(qq23,velec);
1327
1328             d                = _mm256_sub_ps(r23,rswitch);
1329             d                = _mm256_max_ps(d,_mm256_setzero_ps());
1330             d2               = _mm256_mul_ps(d,d);
1331             sw               = _mm256_add_ps(one,_mm256_mul_ps(d2,_mm256_mul_ps(d,_mm256_add_ps(swV3,_mm256_mul_ps(d,_mm256_add_ps(swV4,_mm256_mul_ps(d,swV5)))))));
1332
1333             dsw              = _mm256_mul_ps(d2,_mm256_add_ps(swF2,_mm256_mul_ps(d,_mm256_add_ps(swF3,_mm256_mul_ps(d,swF4)))));
1334
1335             /* Evaluate switch function */
1336             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
1337             felec            = _mm256_sub_ps( _mm256_mul_ps(felec,sw) , _mm256_mul_ps(rinv23,_mm256_mul_ps(velec,dsw)) );
1338             velec            = _mm256_mul_ps(velec,sw);
1339             cutoff_mask      = _mm256_cmp_ps(rsq23,rcutoff2,_CMP_LT_OQ);
1340
1341             /* Update potential sum for this i atom from the interaction with this j atom. */
1342             velec            = _mm256_and_ps(velec,cutoff_mask);
1343             velec            = _mm256_andnot_ps(dummy_mask,velec);
1344             velecsum         = _mm256_add_ps(velecsum,velec);
1345
1346             fscal            = felec;
1347
1348             fscal            = _mm256_and_ps(fscal,cutoff_mask);
1349
1350             fscal            = _mm256_andnot_ps(dummy_mask,fscal);
1351
1352             /* Calculate temporary vectorial force */
1353             tx               = _mm256_mul_ps(fscal,dx23);
1354             ty               = _mm256_mul_ps(fscal,dy23);
1355             tz               = _mm256_mul_ps(fscal,dz23);
1356
1357             /* Update vectorial force */
1358             fix2             = _mm256_add_ps(fix2,tx);
1359             fiy2             = _mm256_add_ps(fiy2,ty);
1360             fiz2             = _mm256_add_ps(fiz2,tz);
1361
1362             fjx3             = _mm256_add_ps(fjx3,tx);
1363             fjy3             = _mm256_add_ps(fjy3,ty);
1364             fjz3             = _mm256_add_ps(fjz3,tz);
1365
1366             }
1367
1368             /**************************
1369              * CALCULATE INTERACTIONS *
1370              **************************/
1371
1372             if (gmx_mm256_any_lt(rsq31,rcutoff2))
1373             {
1374
1375             r31              = _mm256_mul_ps(rsq31,rinv31);
1376             r31              = _mm256_andnot_ps(dummy_mask,r31);
1377
1378             /* EWALD ELECTROSTATICS */
1379
1380             /* Analytical PME correction */
1381             zeta2            = _mm256_mul_ps(beta2,rsq31);
1382             rinv3            = _mm256_mul_ps(rinvsq31,rinv31);
1383             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
1384             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
1385             felec            = _mm256_mul_ps(qq31,felec);
1386             pmecorrV         = gmx_mm256_pmecorrV_ps(zeta2);
1387             pmecorrV         = _mm256_mul_ps(pmecorrV,beta);
1388             velec            = _mm256_sub_ps(rinv31,pmecorrV);
1389             velec            = _mm256_mul_ps(qq31,velec);
1390
1391             d                = _mm256_sub_ps(r31,rswitch);
1392             d                = _mm256_max_ps(d,_mm256_setzero_ps());
1393             d2               = _mm256_mul_ps(d,d);
1394             sw               = _mm256_add_ps(one,_mm256_mul_ps(d2,_mm256_mul_ps(d,_mm256_add_ps(swV3,_mm256_mul_ps(d,_mm256_add_ps(swV4,_mm256_mul_ps(d,swV5)))))));
1395
1396             dsw              = _mm256_mul_ps(d2,_mm256_add_ps(swF2,_mm256_mul_ps(d,_mm256_add_ps(swF3,_mm256_mul_ps(d,swF4)))));
1397
1398             /* Evaluate switch function */
1399             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
1400             felec            = _mm256_sub_ps( _mm256_mul_ps(felec,sw) , _mm256_mul_ps(rinv31,_mm256_mul_ps(velec,dsw)) );
1401             velec            = _mm256_mul_ps(velec,sw);
1402             cutoff_mask      = _mm256_cmp_ps(rsq31,rcutoff2,_CMP_LT_OQ);
1403
1404             /* Update potential sum for this i atom from the interaction with this j atom. */
1405             velec            = _mm256_and_ps(velec,cutoff_mask);
1406             velec            = _mm256_andnot_ps(dummy_mask,velec);
1407             velecsum         = _mm256_add_ps(velecsum,velec);
1408
1409             fscal            = felec;
1410
1411             fscal            = _mm256_and_ps(fscal,cutoff_mask);
1412
1413             fscal            = _mm256_andnot_ps(dummy_mask,fscal);
1414
1415             /* Calculate temporary vectorial force */
1416             tx               = _mm256_mul_ps(fscal,dx31);
1417             ty               = _mm256_mul_ps(fscal,dy31);
1418             tz               = _mm256_mul_ps(fscal,dz31);
1419
1420             /* Update vectorial force */
1421             fix3             = _mm256_add_ps(fix3,tx);
1422             fiy3             = _mm256_add_ps(fiy3,ty);
1423             fiz3             = _mm256_add_ps(fiz3,tz);
1424
1425             fjx1             = _mm256_add_ps(fjx1,tx);
1426             fjy1             = _mm256_add_ps(fjy1,ty);
1427             fjz1             = _mm256_add_ps(fjz1,tz);
1428
1429             }
1430
1431             /**************************
1432              * CALCULATE INTERACTIONS *
1433              **************************/
1434
1435             if (gmx_mm256_any_lt(rsq32,rcutoff2))
1436             {
1437
1438             r32              = _mm256_mul_ps(rsq32,rinv32);
1439             r32              = _mm256_andnot_ps(dummy_mask,r32);
1440
1441             /* EWALD ELECTROSTATICS */
1442
1443             /* Analytical PME correction */
1444             zeta2            = _mm256_mul_ps(beta2,rsq32);
1445             rinv3            = _mm256_mul_ps(rinvsq32,rinv32);
1446             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
1447             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
1448             felec            = _mm256_mul_ps(qq32,felec);
1449             pmecorrV         = gmx_mm256_pmecorrV_ps(zeta2);
1450             pmecorrV         = _mm256_mul_ps(pmecorrV,beta);
1451             velec            = _mm256_sub_ps(rinv32,pmecorrV);
1452             velec            = _mm256_mul_ps(qq32,velec);
1453
1454             d                = _mm256_sub_ps(r32,rswitch);
1455             d                = _mm256_max_ps(d,_mm256_setzero_ps());
1456             d2               = _mm256_mul_ps(d,d);
1457             sw               = _mm256_add_ps(one,_mm256_mul_ps(d2,_mm256_mul_ps(d,_mm256_add_ps(swV3,_mm256_mul_ps(d,_mm256_add_ps(swV4,_mm256_mul_ps(d,swV5)))))));
1458
1459             dsw              = _mm256_mul_ps(d2,_mm256_add_ps(swF2,_mm256_mul_ps(d,_mm256_add_ps(swF3,_mm256_mul_ps(d,swF4)))));
1460
1461             /* Evaluate switch function */
1462             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
1463             felec            = _mm256_sub_ps( _mm256_mul_ps(felec,sw) , _mm256_mul_ps(rinv32,_mm256_mul_ps(velec,dsw)) );
1464             velec            = _mm256_mul_ps(velec,sw);
1465             cutoff_mask      = _mm256_cmp_ps(rsq32,rcutoff2,_CMP_LT_OQ);
1466
1467             /* Update potential sum for this i atom from the interaction with this j atom. */
1468             velec            = _mm256_and_ps(velec,cutoff_mask);
1469             velec            = _mm256_andnot_ps(dummy_mask,velec);
1470             velecsum         = _mm256_add_ps(velecsum,velec);
1471
1472             fscal            = felec;
1473
1474             fscal            = _mm256_and_ps(fscal,cutoff_mask);
1475
1476             fscal            = _mm256_andnot_ps(dummy_mask,fscal);
1477
1478             /* Calculate temporary vectorial force */
1479             tx               = _mm256_mul_ps(fscal,dx32);
1480             ty               = _mm256_mul_ps(fscal,dy32);
1481             tz               = _mm256_mul_ps(fscal,dz32);
1482
1483             /* Update vectorial force */
1484             fix3             = _mm256_add_ps(fix3,tx);
1485             fiy3             = _mm256_add_ps(fiy3,ty);
1486             fiz3             = _mm256_add_ps(fiz3,tz);
1487
1488             fjx2             = _mm256_add_ps(fjx2,tx);
1489             fjy2             = _mm256_add_ps(fjy2,ty);
1490             fjz2             = _mm256_add_ps(fjz2,tz);
1491
1492             }
1493
1494             /**************************
1495              * CALCULATE INTERACTIONS *
1496              **************************/
1497
1498             if (gmx_mm256_any_lt(rsq33,rcutoff2))
1499             {
1500
1501             r33              = _mm256_mul_ps(rsq33,rinv33);
1502             r33              = _mm256_andnot_ps(dummy_mask,r33);
1503
1504             /* EWALD ELECTROSTATICS */
1505
1506             /* Analytical PME correction */
1507             zeta2            = _mm256_mul_ps(beta2,rsq33);
1508             rinv3            = _mm256_mul_ps(rinvsq33,rinv33);
1509             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
1510             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
1511             felec            = _mm256_mul_ps(qq33,felec);
1512             pmecorrV         = gmx_mm256_pmecorrV_ps(zeta2);
1513             pmecorrV         = _mm256_mul_ps(pmecorrV,beta);
1514             velec            = _mm256_sub_ps(rinv33,pmecorrV);
1515             velec            = _mm256_mul_ps(qq33,velec);
1516
1517             d                = _mm256_sub_ps(r33,rswitch);
1518             d                = _mm256_max_ps(d,_mm256_setzero_ps());
1519             d2               = _mm256_mul_ps(d,d);
1520             sw               = _mm256_add_ps(one,_mm256_mul_ps(d2,_mm256_mul_ps(d,_mm256_add_ps(swV3,_mm256_mul_ps(d,_mm256_add_ps(swV4,_mm256_mul_ps(d,swV5)))))));
1521
1522             dsw              = _mm256_mul_ps(d2,_mm256_add_ps(swF2,_mm256_mul_ps(d,_mm256_add_ps(swF3,_mm256_mul_ps(d,swF4)))));
1523
1524             /* Evaluate switch function */
1525             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
1526             felec            = _mm256_sub_ps( _mm256_mul_ps(felec,sw) , _mm256_mul_ps(rinv33,_mm256_mul_ps(velec,dsw)) );
1527             velec            = _mm256_mul_ps(velec,sw);
1528             cutoff_mask      = _mm256_cmp_ps(rsq33,rcutoff2,_CMP_LT_OQ);
1529
1530             /* Update potential sum for this i atom from the interaction with this j atom. */
1531             velec            = _mm256_and_ps(velec,cutoff_mask);
1532             velec            = _mm256_andnot_ps(dummy_mask,velec);
1533             velecsum         = _mm256_add_ps(velecsum,velec);
1534
1535             fscal            = felec;
1536
1537             fscal            = _mm256_and_ps(fscal,cutoff_mask);
1538
1539             fscal            = _mm256_andnot_ps(dummy_mask,fscal);
1540
1541             /* Calculate temporary vectorial force */
1542             tx               = _mm256_mul_ps(fscal,dx33);
1543             ty               = _mm256_mul_ps(fscal,dy33);
1544             tz               = _mm256_mul_ps(fscal,dz33);
1545
1546             /* Update vectorial force */
1547             fix3             = _mm256_add_ps(fix3,tx);
1548             fiy3             = _mm256_add_ps(fiy3,ty);
1549             fiz3             = _mm256_add_ps(fiz3,tz);
1550
1551             fjx3             = _mm256_add_ps(fjx3,tx);
1552             fjy3             = _mm256_add_ps(fjy3,ty);
1553             fjz3             = _mm256_add_ps(fjz3,tz);
1554
1555             }
1556
1557             fjptrA             = (jnrlistA>=0) ? f+j_coord_offsetA : scratch;
1558             fjptrB             = (jnrlistB>=0) ? f+j_coord_offsetB : scratch;
1559             fjptrC             = (jnrlistC>=0) ? f+j_coord_offsetC : scratch;
1560             fjptrD             = (jnrlistD>=0) ? f+j_coord_offsetD : scratch;
1561             fjptrE             = (jnrlistE>=0) ? f+j_coord_offsetE : scratch;
1562             fjptrF             = (jnrlistF>=0) ? f+j_coord_offsetF : scratch;
1563             fjptrG             = (jnrlistG>=0) ? f+j_coord_offsetG : scratch;
1564             fjptrH             = (jnrlistH>=0) ? f+j_coord_offsetH : scratch;
1565
1566             gmx_mm256_decrement_3rvec_8ptr_swizzle_ps(fjptrA+DIM,fjptrB+DIM,fjptrC+DIM,fjptrD+DIM,
1567                                                       fjptrE+DIM,fjptrF+DIM,fjptrG+DIM,fjptrH+DIM,
1568                                                       fjx1,fjy1,fjz1,fjx2,fjy2,fjz2,fjx3,fjy3,fjz3);
1569
1570             /* Inner loop uses 981 flops */
1571         }
1572
1573         /* End of innermost loop */
1574
1575         gmx_mm256_update_iforce_3atom_swizzle_ps(fix1,fiy1,fiz1,fix2,fiy2,fiz2,fix3,fiy3,fiz3,
1576                                                  f+i_coord_offset+DIM,fshift+i_shift_offset);
1577
1578         ggid                        = gid[iidx];
1579         /* Update potential energies */
1580         gmx_mm256_update_1pot_ps(velecsum,kernel_data->energygrp_elec+ggid);
1581
1582         /* Increment number of inner iterations */
1583         inneriter                  += j_index_end - j_index_start;
1584
1585         /* Outer loop uses 19 flops */
1586     }
1587
1588     /* Increment number of outer iterations */
1589     outeriter        += nri;
1590
1591     /* Update outer/inner flops */
1592
1593     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_W4W4_VF,outeriter*19 + inneriter*981);
1594 }
1595 /*
1596  * Gromacs nonbonded kernel:   nb_kernel_ElecEwSw_VdwNone_GeomW4W4_F_avx_256_single
1597  * Electrostatics interaction: Ewald
1598  * VdW interaction:            None
1599  * Geometry:                   Water4-Water4
1600  * Calculate force/pot:        Force
1601  */
1602 void
1603 nb_kernel_ElecEwSw_VdwNone_GeomW4W4_F_avx_256_single
1604                     (t_nblist                    * gmx_restrict       nlist,
1605                      rvec                        * gmx_restrict          xx,
1606                      rvec                        * gmx_restrict          ff,
1607                      t_forcerec                  * gmx_restrict          fr,
1608                      t_mdatoms                   * gmx_restrict     mdatoms,
1609                      nb_kernel_data_t gmx_unused * gmx_restrict kernel_data,
1610                      t_nrnb                      * gmx_restrict        nrnb)
1611 {
1612     /* Suffixes 0,1,2,3 refer to particle indices for waters in the inner or outer loop, or
1613      * just 0 for non-waters.
1614      * Suffixes A,B,C,D,E,F,G,H refer to j loop unrolling done with AVX, e.g. for the eight different
1615      * jnr indices corresponding to data put in the four positions in the SIMD register.
1616      */
1617     int              i_shift_offset,i_coord_offset,outeriter,inneriter;
1618     int              j_index_start,j_index_end,jidx,nri,inr,ggid,iidx;
1619     int              jnrA,jnrB,jnrC,jnrD;
1620     int              jnrE,jnrF,jnrG,jnrH;
1621     int              jnrlistA,jnrlistB,jnrlistC,jnrlistD;
1622     int              jnrlistE,jnrlistF,jnrlistG,jnrlistH;
1623     int              j_coord_offsetA,j_coord_offsetB,j_coord_offsetC,j_coord_offsetD;
1624     int              j_coord_offsetE,j_coord_offsetF,j_coord_offsetG,j_coord_offsetH;
1625     int              *iinr,*jindex,*jjnr,*shiftidx,*gid;
1626     real             rcutoff_scalar;
1627     real             *shiftvec,*fshift,*x,*f;
1628     real             *fjptrA,*fjptrB,*fjptrC,*fjptrD,*fjptrE,*fjptrF,*fjptrG,*fjptrH;
1629     real             scratch[4*DIM];
1630     __m256           tx,ty,tz,fscal,rcutoff,rcutoff2,jidxall;
1631     real *           vdwioffsetptr1;
1632     __m256           ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
1633     real *           vdwioffsetptr2;
1634     __m256           ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
1635     real *           vdwioffsetptr3;
1636     __m256           ix3,iy3,iz3,fix3,fiy3,fiz3,iq3,isai3;
1637     int              vdwjidx1A,vdwjidx1B,vdwjidx1C,vdwjidx1D,vdwjidx1E,vdwjidx1F,vdwjidx1G,vdwjidx1H;
1638     __m256           jx1,jy1,jz1,fjx1,fjy1,fjz1,jq1,isaj1;
1639     int              vdwjidx2A,vdwjidx2B,vdwjidx2C,vdwjidx2D,vdwjidx2E,vdwjidx2F,vdwjidx2G,vdwjidx2H;
1640     __m256           jx2,jy2,jz2,fjx2,fjy2,fjz2,jq2,isaj2;
1641     int              vdwjidx3A,vdwjidx3B,vdwjidx3C,vdwjidx3D,vdwjidx3E,vdwjidx3F,vdwjidx3G,vdwjidx3H;
1642     __m256           jx3,jy3,jz3,fjx3,fjy3,fjz3,jq3,isaj3;
1643     __m256           dx11,dy11,dz11,rsq11,rinv11,rinvsq11,r11,qq11,c6_11,c12_11;
1644     __m256           dx12,dy12,dz12,rsq12,rinv12,rinvsq12,r12,qq12,c6_12,c12_12;
1645     __m256           dx13,dy13,dz13,rsq13,rinv13,rinvsq13,r13,qq13,c6_13,c12_13;
1646     __m256           dx21,dy21,dz21,rsq21,rinv21,rinvsq21,r21,qq21,c6_21,c12_21;
1647     __m256           dx22,dy22,dz22,rsq22,rinv22,rinvsq22,r22,qq22,c6_22,c12_22;
1648     __m256           dx23,dy23,dz23,rsq23,rinv23,rinvsq23,r23,qq23,c6_23,c12_23;
1649     __m256           dx31,dy31,dz31,rsq31,rinv31,rinvsq31,r31,qq31,c6_31,c12_31;
1650     __m256           dx32,dy32,dz32,rsq32,rinv32,rinvsq32,r32,qq32,c6_32,c12_32;
1651     __m256           dx33,dy33,dz33,rsq33,rinv33,rinvsq33,r33,qq33,c6_33,c12_33;
1652     __m256           velec,felec,velecsum,facel,crf,krf,krf2;
1653     real             *charge;
1654     __m256i          ewitab;
1655     __m128i          ewitab_lo,ewitab_hi;
1656     __m256           ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace,ewtabF,ewtabFn,ewtabD,ewtabV;
1657     __m256           beta,beta2,beta3,zeta2,pmecorrF,pmecorrV,rinv3;
1658     real             *ewtab;
1659     __m256           rswitch,swV3,swV4,swV5,swF2,swF3,swF4,d,d2,sw,dsw;
1660     real             rswitch_scalar,d_scalar;
1661     __m256           dummy_mask,cutoff_mask;
1662     __m256           signbit = _mm256_castsi256_ps( _mm256_set1_epi32(0x80000000) );
1663     __m256           one     = _mm256_set1_ps(1.0);
1664     __m256           two     = _mm256_set1_ps(2.0);
1665     x                = xx[0];
1666     f                = ff[0];
1667
1668     nri              = nlist->nri;
1669     iinr             = nlist->iinr;
1670     jindex           = nlist->jindex;
1671     jjnr             = nlist->jjnr;
1672     shiftidx         = nlist->shift;
1673     gid              = nlist->gid;
1674     shiftvec         = fr->shift_vec[0];
1675     fshift           = fr->fshift[0];
1676     facel            = _mm256_set1_ps(fr->epsfac);
1677     charge           = mdatoms->chargeA;
1678
1679     sh_ewald         = _mm256_set1_ps(fr->ic->sh_ewald);
1680     beta             = _mm256_set1_ps(fr->ic->ewaldcoeff_q);
1681     beta2            = _mm256_mul_ps(beta,beta);
1682     beta3            = _mm256_mul_ps(beta,beta2);
1683
1684     ewtab            = fr->ic->tabq_coul_FDV0;
1685     ewtabscale       = _mm256_set1_ps(fr->ic->tabq_scale);
1686     ewtabhalfspace   = _mm256_set1_ps(0.5/fr->ic->tabq_scale);
1687
1688     /* Setup water-specific parameters */
1689     inr              = nlist->iinr[0];
1690     iq1              = _mm256_mul_ps(facel,_mm256_set1_ps(charge[inr+1]));
1691     iq2              = _mm256_mul_ps(facel,_mm256_set1_ps(charge[inr+2]));
1692     iq3              = _mm256_mul_ps(facel,_mm256_set1_ps(charge[inr+3]));
1693
1694     jq1              = _mm256_set1_ps(charge[inr+1]);
1695     jq2              = _mm256_set1_ps(charge[inr+2]);
1696     jq3              = _mm256_set1_ps(charge[inr+3]);
1697     qq11             = _mm256_mul_ps(iq1,jq1);
1698     qq12             = _mm256_mul_ps(iq1,jq2);
1699     qq13             = _mm256_mul_ps(iq1,jq3);
1700     qq21             = _mm256_mul_ps(iq2,jq1);
1701     qq22             = _mm256_mul_ps(iq2,jq2);
1702     qq23             = _mm256_mul_ps(iq2,jq3);
1703     qq31             = _mm256_mul_ps(iq3,jq1);
1704     qq32             = _mm256_mul_ps(iq3,jq2);
1705     qq33             = _mm256_mul_ps(iq3,jq3);
1706
1707     /* When we use explicit cutoffs the value must be identical for elec and VdW, so use elec as an arbitrary choice */
1708     rcutoff_scalar   = fr->rcoulomb;
1709     rcutoff          = _mm256_set1_ps(rcutoff_scalar);
1710     rcutoff2         = _mm256_mul_ps(rcutoff,rcutoff);
1711
1712     rswitch_scalar   = fr->rcoulomb_switch;
1713     rswitch          = _mm256_set1_ps(rswitch_scalar);
1714     /* Setup switch parameters */
1715     d_scalar         = rcutoff_scalar-rswitch_scalar;
1716     d                = _mm256_set1_ps(d_scalar);
1717     swV3             = _mm256_set1_ps(-10.0/(d_scalar*d_scalar*d_scalar));
1718     swV4             = _mm256_set1_ps( 15.0/(d_scalar*d_scalar*d_scalar*d_scalar));
1719     swV5             = _mm256_set1_ps( -6.0/(d_scalar*d_scalar*d_scalar*d_scalar*d_scalar));
1720     swF2             = _mm256_set1_ps(-30.0/(d_scalar*d_scalar*d_scalar));
1721     swF3             = _mm256_set1_ps( 60.0/(d_scalar*d_scalar*d_scalar*d_scalar));
1722     swF4             = _mm256_set1_ps(-30.0/(d_scalar*d_scalar*d_scalar*d_scalar*d_scalar));
1723
1724     /* Avoid stupid compiler warnings */
1725     jnrA = jnrB = jnrC = jnrD = jnrE = jnrF = jnrG = jnrH = 0;
1726     j_coord_offsetA = 0;
1727     j_coord_offsetB = 0;
1728     j_coord_offsetC = 0;
1729     j_coord_offsetD = 0;
1730     j_coord_offsetE = 0;
1731     j_coord_offsetF = 0;
1732     j_coord_offsetG = 0;
1733     j_coord_offsetH = 0;
1734
1735     outeriter        = 0;
1736     inneriter        = 0;
1737
1738     for(iidx=0;iidx<4*DIM;iidx++)
1739     {
1740         scratch[iidx] = 0.0;
1741     }
1742
1743     /* Start outer loop over neighborlists */
1744     for(iidx=0; iidx<nri; iidx++)
1745     {
1746         /* Load shift vector for this list */
1747         i_shift_offset   = DIM*shiftidx[iidx];
1748
1749         /* Load limits for loop over neighbors */
1750         j_index_start    = jindex[iidx];
1751         j_index_end      = jindex[iidx+1];
1752
1753         /* Get outer coordinate index */
1754         inr              = iinr[iidx];
1755         i_coord_offset   = DIM*inr;
1756
1757         /* Load i particle coords and add shift vector */
1758         gmx_mm256_load_shift_and_3rvec_broadcast_ps(shiftvec+i_shift_offset,x+i_coord_offset+DIM,
1759                                                     &ix1,&iy1,&iz1,&ix2,&iy2,&iz2,&ix3,&iy3,&iz3);
1760
1761         fix1             = _mm256_setzero_ps();
1762         fiy1             = _mm256_setzero_ps();
1763         fiz1             = _mm256_setzero_ps();
1764         fix2             = _mm256_setzero_ps();
1765         fiy2             = _mm256_setzero_ps();
1766         fiz2             = _mm256_setzero_ps();
1767         fix3             = _mm256_setzero_ps();
1768         fiy3             = _mm256_setzero_ps();
1769         fiz3             = _mm256_setzero_ps();
1770
1771         /* Start inner kernel loop */
1772         for(jidx=j_index_start; jidx<j_index_end && jjnr[jidx+7]>=0; jidx+=8)
1773         {
1774
1775             /* Get j neighbor index, and coordinate index */
1776             jnrA             = jjnr[jidx];
1777             jnrB             = jjnr[jidx+1];
1778             jnrC             = jjnr[jidx+2];
1779             jnrD             = jjnr[jidx+3];
1780             jnrE             = jjnr[jidx+4];
1781             jnrF             = jjnr[jidx+5];
1782             jnrG             = jjnr[jidx+6];
1783             jnrH             = jjnr[jidx+7];
1784             j_coord_offsetA  = DIM*jnrA;
1785             j_coord_offsetB  = DIM*jnrB;
1786             j_coord_offsetC  = DIM*jnrC;
1787             j_coord_offsetD  = DIM*jnrD;
1788             j_coord_offsetE  = DIM*jnrE;
1789             j_coord_offsetF  = DIM*jnrF;
1790             j_coord_offsetG  = DIM*jnrG;
1791             j_coord_offsetH  = DIM*jnrH;
1792
1793             /* load j atom coordinates */
1794             gmx_mm256_load_3rvec_8ptr_swizzle_ps(x+j_coord_offsetA+DIM,x+j_coord_offsetB+DIM,
1795                                                  x+j_coord_offsetC+DIM,x+j_coord_offsetD+DIM,
1796                                                  x+j_coord_offsetE+DIM,x+j_coord_offsetF+DIM,
1797                                                  x+j_coord_offsetG+DIM,x+j_coord_offsetH+DIM,
1798                                                  &jx1,&jy1,&jz1,&jx2,&jy2,&jz2,&jx3,&jy3,&jz3);
1799
1800             /* Calculate displacement vector */
1801             dx11             = _mm256_sub_ps(ix1,jx1);
1802             dy11             = _mm256_sub_ps(iy1,jy1);
1803             dz11             = _mm256_sub_ps(iz1,jz1);
1804             dx12             = _mm256_sub_ps(ix1,jx2);
1805             dy12             = _mm256_sub_ps(iy1,jy2);
1806             dz12             = _mm256_sub_ps(iz1,jz2);
1807             dx13             = _mm256_sub_ps(ix1,jx3);
1808             dy13             = _mm256_sub_ps(iy1,jy3);
1809             dz13             = _mm256_sub_ps(iz1,jz3);
1810             dx21             = _mm256_sub_ps(ix2,jx1);
1811             dy21             = _mm256_sub_ps(iy2,jy1);
1812             dz21             = _mm256_sub_ps(iz2,jz1);
1813             dx22             = _mm256_sub_ps(ix2,jx2);
1814             dy22             = _mm256_sub_ps(iy2,jy2);
1815             dz22             = _mm256_sub_ps(iz2,jz2);
1816             dx23             = _mm256_sub_ps(ix2,jx3);
1817             dy23             = _mm256_sub_ps(iy2,jy3);
1818             dz23             = _mm256_sub_ps(iz2,jz3);
1819             dx31             = _mm256_sub_ps(ix3,jx1);
1820             dy31             = _mm256_sub_ps(iy3,jy1);
1821             dz31             = _mm256_sub_ps(iz3,jz1);
1822             dx32             = _mm256_sub_ps(ix3,jx2);
1823             dy32             = _mm256_sub_ps(iy3,jy2);
1824             dz32             = _mm256_sub_ps(iz3,jz2);
1825             dx33             = _mm256_sub_ps(ix3,jx3);
1826             dy33             = _mm256_sub_ps(iy3,jy3);
1827             dz33             = _mm256_sub_ps(iz3,jz3);
1828
1829             /* Calculate squared distance and things based on it */
1830             rsq11            = gmx_mm256_calc_rsq_ps(dx11,dy11,dz11);
1831             rsq12            = gmx_mm256_calc_rsq_ps(dx12,dy12,dz12);
1832             rsq13            = gmx_mm256_calc_rsq_ps(dx13,dy13,dz13);
1833             rsq21            = gmx_mm256_calc_rsq_ps(dx21,dy21,dz21);
1834             rsq22            = gmx_mm256_calc_rsq_ps(dx22,dy22,dz22);
1835             rsq23            = gmx_mm256_calc_rsq_ps(dx23,dy23,dz23);
1836             rsq31            = gmx_mm256_calc_rsq_ps(dx31,dy31,dz31);
1837             rsq32            = gmx_mm256_calc_rsq_ps(dx32,dy32,dz32);
1838             rsq33            = gmx_mm256_calc_rsq_ps(dx33,dy33,dz33);
1839
1840             rinv11           = gmx_mm256_invsqrt_ps(rsq11);
1841             rinv12           = gmx_mm256_invsqrt_ps(rsq12);
1842             rinv13           = gmx_mm256_invsqrt_ps(rsq13);
1843             rinv21           = gmx_mm256_invsqrt_ps(rsq21);
1844             rinv22           = gmx_mm256_invsqrt_ps(rsq22);
1845             rinv23           = gmx_mm256_invsqrt_ps(rsq23);
1846             rinv31           = gmx_mm256_invsqrt_ps(rsq31);
1847             rinv32           = gmx_mm256_invsqrt_ps(rsq32);
1848             rinv33           = gmx_mm256_invsqrt_ps(rsq33);
1849
1850             rinvsq11         = _mm256_mul_ps(rinv11,rinv11);
1851             rinvsq12         = _mm256_mul_ps(rinv12,rinv12);
1852             rinvsq13         = _mm256_mul_ps(rinv13,rinv13);
1853             rinvsq21         = _mm256_mul_ps(rinv21,rinv21);
1854             rinvsq22         = _mm256_mul_ps(rinv22,rinv22);
1855             rinvsq23         = _mm256_mul_ps(rinv23,rinv23);
1856             rinvsq31         = _mm256_mul_ps(rinv31,rinv31);
1857             rinvsq32         = _mm256_mul_ps(rinv32,rinv32);
1858             rinvsq33         = _mm256_mul_ps(rinv33,rinv33);
1859
1860             fjx1             = _mm256_setzero_ps();
1861             fjy1             = _mm256_setzero_ps();
1862             fjz1             = _mm256_setzero_ps();
1863             fjx2             = _mm256_setzero_ps();
1864             fjy2             = _mm256_setzero_ps();
1865             fjz2             = _mm256_setzero_ps();
1866             fjx3             = _mm256_setzero_ps();
1867             fjy3             = _mm256_setzero_ps();
1868             fjz3             = _mm256_setzero_ps();
1869
1870             /**************************
1871              * CALCULATE INTERACTIONS *
1872              **************************/
1873
1874             if (gmx_mm256_any_lt(rsq11,rcutoff2))
1875             {
1876
1877             r11              = _mm256_mul_ps(rsq11,rinv11);
1878
1879             /* EWALD ELECTROSTATICS */
1880
1881             /* Analytical PME correction */
1882             zeta2            = _mm256_mul_ps(beta2,rsq11);
1883             rinv3            = _mm256_mul_ps(rinvsq11,rinv11);
1884             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
1885             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
1886             felec            = _mm256_mul_ps(qq11,felec);
1887             pmecorrV         = gmx_mm256_pmecorrV_ps(zeta2);
1888             pmecorrV         = _mm256_mul_ps(pmecorrV,beta);
1889             velec            = _mm256_sub_ps(rinv11,pmecorrV);
1890             velec            = _mm256_mul_ps(qq11,velec);
1891
1892             d                = _mm256_sub_ps(r11,rswitch);
1893             d                = _mm256_max_ps(d,_mm256_setzero_ps());
1894             d2               = _mm256_mul_ps(d,d);
1895             sw               = _mm256_add_ps(one,_mm256_mul_ps(d2,_mm256_mul_ps(d,_mm256_add_ps(swV3,_mm256_mul_ps(d,_mm256_add_ps(swV4,_mm256_mul_ps(d,swV5)))))));
1896
1897             dsw              = _mm256_mul_ps(d2,_mm256_add_ps(swF2,_mm256_mul_ps(d,_mm256_add_ps(swF3,_mm256_mul_ps(d,swF4)))));
1898
1899             /* Evaluate switch function */
1900             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
1901             felec            = _mm256_sub_ps( _mm256_mul_ps(felec,sw) , _mm256_mul_ps(rinv11,_mm256_mul_ps(velec,dsw)) );
1902             cutoff_mask      = _mm256_cmp_ps(rsq11,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,dx11);
1910             ty               = _mm256_mul_ps(fscal,dy11);
1911             tz               = _mm256_mul_ps(fscal,dz11);
1912
1913             /* Update vectorial force */
1914             fix1             = _mm256_add_ps(fix1,tx);
1915             fiy1             = _mm256_add_ps(fiy1,ty);
1916             fiz1             = _mm256_add_ps(fiz1,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(rsq12,rcutoff2))
1929             {
1930
1931             r12              = _mm256_mul_ps(rsq12,rinv12);
1932
1933             /* EWALD ELECTROSTATICS */
1934
1935             /* Analytical PME correction */
1936             zeta2            = _mm256_mul_ps(beta2,rsq12);
1937             rinv3            = _mm256_mul_ps(rinvsq12,rinv12);
1938             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
1939             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
1940             felec            = _mm256_mul_ps(qq12,felec);
1941             pmecorrV         = gmx_mm256_pmecorrV_ps(zeta2);
1942             pmecorrV         = _mm256_mul_ps(pmecorrV,beta);
1943             velec            = _mm256_sub_ps(rinv12,pmecorrV);
1944             velec            = _mm256_mul_ps(qq12,velec);
1945
1946             d                = _mm256_sub_ps(r12,rswitch);
1947             d                = _mm256_max_ps(d,_mm256_setzero_ps());
1948             d2               = _mm256_mul_ps(d,d);
1949             sw               = _mm256_add_ps(one,_mm256_mul_ps(d2,_mm256_mul_ps(d,_mm256_add_ps(swV3,_mm256_mul_ps(d,_mm256_add_ps(swV4,_mm256_mul_ps(d,swV5)))))));
1950
1951             dsw              = _mm256_mul_ps(d2,_mm256_add_ps(swF2,_mm256_mul_ps(d,_mm256_add_ps(swF3,_mm256_mul_ps(d,swF4)))));
1952
1953             /* Evaluate switch function */
1954             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
1955             felec            = _mm256_sub_ps( _mm256_mul_ps(felec,sw) , _mm256_mul_ps(rinv12,_mm256_mul_ps(velec,dsw)) );
1956             cutoff_mask      = _mm256_cmp_ps(rsq12,rcutoff2,_CMP_LT_OQ);
1957
1958             fscal            = felec;
1959
1960             fscal            = _mm256_and_ps(fscal,cutoff_mask);
1961
1962             /* Calculate temporary vectorial force */
1963             tx               = _mm256_mul_ps(fscal,dx12);
1964             ty               = _mm256_mul_ps(fscal,dy12);
1965             tz               = _mm256_mul_ps(fscal,dz12);
1966
1967             /* Update vectorial force */
1968             fix1             = _mm256_add_ps(fix1,tx);
1969             fiy1             = _mm256_add_ps(fiy1,ty);
1970             fiz1             = _mm256_add_ps(fiz1,tz);
1971
1972             fjx2             = _mm256_add_ps(fjx2,tx);
1973             fjy2             = _mm256_add_ps(fjy2,ty);
1974             fjz2             = _mm256_add_ps(fjz2,tz);
1975
1976             }
1977
1978             /**************************
1979              * CALCULATE INTERACTIONS *
1980              **************************/
1981
1982             if (gmx_mm256_any_lt(rsq13,rcutoff2))
1983             {
1984
1985             r13              = _mm256_mul_ps(rsq13,rinv13);
1986
1987             /* EWALD ELECTROSTATICS */
1988
1989             /* Analytical PME correction */
1990             zeta2            = _mm256_mul_ps(beta2,rsq13);
1991             rinv3            = _mm256_mul_ps(rinvsq13,rinv13);
1992             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
1993             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
1994             felec            = _mm256_mul_ps(qq13,felec);
1995             pmecorrV         = gmx_mm256_pmecorrV_ps(zeta2);
1996             pmecorrV         = _mm256_mul_ps(pmecorrV,beta);
1997             velec            = _mm256_sub_ps(rinv13,pmecorrV);
1998             velec            = _mm256_mul_ps(qq13,velec);
1999
2000             d                = _mm256_sub_ps(r13,rswitch);
2001             d                = _mm256_max_ps(d,_mm256_setzero_ps());
2002             d2               = _mm256_mul_ps(d,d);
2003             sw               = _mm256_add_ps(one,_mm256_mul_ps(d2,_mm256_mul_ps(d,_mm256_add_ps(swV3,_mm256_mul_ps(d,_mm256_add_ps(swV4,_mm256_mul_ps(d,swV5)))))));
2004
2005             dsw              = _mm256_mul_ps(d2,_mm256_add_ps(swF2,_mm256_mul_ps(d,_mm256_add_ps(swF3,_mm256_mul_ps(d,swF4)))));
2006
2007             /* Evaluate switch function */
2008             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
2009             felec            = _mm256_sub_ps( _mm256_mul_ps(felec,sw) , _mm256_mul_ps(rinv13,_mm256_mul_ps(velec,dsw)) );
2010             cutoff_mask      = _mm256_cmp_ps(rsq13,rcutoff2,_CMP_LT_OQ);
2011
2012             fscal            = felec;
2013
2014             fscal            = _mm256_and_ps(fscal,cutoff_mask);
2015
2016             /* Calculate temporary vectorial force */
2017             tx               = _mm256_mul_ps(fscal,dx13);
2018             ty               = _mm256_mul_ps(fscal,dy13);
2019             tz               = _mm256_mul_ps(fscal,dz13);
2020
2021             /* Update vectorial force */
2022             fix1             = _mm256_add_ps(fix1,tx);
2023             fiy1             = _mm256_add_ps(fiy1,ty);
2024             fiz1             = _mm256_add_ps(fiz1,tz);
2025
2026             fjx3             = _mm256_add_ps(fjx3,tx);
2027             fjy3             = _mm256_add_ps(fjy3,ty);
2028             fjz3             = _mm256_add_ps(fjz3,tz);
2029
2030             }
2031
2032             /**************************
2033              * CALCULATE INTERACTIONS *
2034              **************************/
2035
2036             if (gmx_mm256_any_lt(rsq21,rcutoff2))
2037             {
2038
2039             r21              = _mm256_mul_ps(rsq21,rinv21);
2040
2041             /* EWALD ELECTROSTATICS */
2042
2043             /* Analytical PME correction */
2044             zeta2            = _mm256_mul_ps(beta2,rsq21);
2045             rinv3            = _mm256_mul_ps(rinvsq21,rinv21);
2046             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
2047             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
2048             felec            = _mm256_mul_ps(qq21,felec);
2049             pmecorrV         = gmx_mm256_pmecorrV_ps(zeta2);
2050             pmecorrV         = _mm256_mul_ps(pmecorrV,beta);
2051             velec            = _mm256_sub_ps(rinv21,pmecorrV);
2052             velec            = _mm256_mul_ps(qq21,velec);
2053
2054             d                = _mm256_sub_ps(r21,rswitch);
2055             d                = _mm256_max_ps(d,_mm256_setzero_ps());
2056             d2               = _mm256_mul_ps(d,d);
2057             sw               = _mm256_add_ps(one,_mm256_mul_ps(d2,_mm256_mul_ps(d,_mm256_add_ps(swV3,_mm256_mul_ps(d,_mm256_add_ps(swV4,_mm256_mul_ps(d,swV5)))))));
2058
2059             dsw              = _mm256_mul_ps(d2,_mm256_add_ps(swF2,_mm256_mul_ps(d,_mm256_add_ps(swF3,_mm256_mul_ps(d,swF4)))));
2060
2061             /* Evaluate switch function */
2062             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
2063             felec            = _mm256_sub_ps( _mm256_mul_ps(felec,sw) , _mm256_mul_ps(rinv21,_mm256_mul_ps(velec,dsw)) );
2064             cutoff_mask      = _mm256_cmp_ps(rsq21,rcutoff2,_CMP_LT_OQ);
2065
2066             fscal            = felec;
2067
2068             fscal            = _mm256_and_ps(fscal,cutoff_mask);
2069
2070             /* Calculate temporary vectorial force */
2071             tx               = _mm256_mul_ps(fscal,dx21);
2072             ty               = _mm256_mul_ps(fscal,dy21);
2073             tz               = _mm256_mul_ps(fscal,dz21);
2074
2075             /* Update vectorial force */
2076             fix2             = _mm256_add_ps(fix2,tx);
2077             fiy2             = _mm256_add_ps(fiy2,ty);
2078             fiz2             = _mm256_add_ps(fiz2,tz);
2079
2080             fjx1             = _mm256_add_ps(fjx1,tx);
2081             fjy1             = _mm256_add_ps(fjy1,ty);
2082             fjz1             = _mm256_add_ps(fjz1,tz);
2083
2084             }
2085
2086             /**************************
2087              * CALCULATE INTERACTIONS *
2088              **************************/
2089
2090             if (gmx_mm256_any_lt(rsq22,rcutoff2))
2091             {
2092
2093             r22              = _mm256_mul_ps(rsq22,rinv22);
2094
2095             /* EWALD ELECTROSTATICS */
2096
2097             /* Analytical PME correction */
2098             zeta2            = _mm256_mul_ps(beta2,rsq22);
2099             rinv3            = _mm256_mul_ps(rinvsq22,rinv22);
2100             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
2101             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
2102             felec            = _mm256_mul_ps(qq22,felec);
2103             pmecorrV         = gmx_mm256_pmecorrV_ps(zeta2);
2104             pmecorrV         = _mm256_mul_ps(pmecorrV,beta);
2105             velec            = _mm256_sub_ps(rinv22,pmecorrV);
2106             velec            = _mm256_mul_ps(qq22,velec);
2107
2108             d                = _mm256_sub_ps(r22,rswitch);
2109             d                = _mm256_max_ps(d,_mm256_setzero_ps());
2110             d2               = _mm256_mul_ps(d,d);
2111             sw               = _mm256_add_ps(one,_mm256_mul_ps(d2,_mm256_mul_ps(d,_mm256_add_ps(swV3,_mm256_mul_ps(d,_mm256_add_ps(swV4,_mm256_mul_ps(d,swV5)))))));
2112
2113             dsw              = _mm256_mul_ps(d2,_mm256_add_ps(swF2,_mm256_mul_ps(d,_mm256_add_ps(swF3,_mm256_mul_ps(d,swF4)))));
2114
2115             /* Evaluate switch function */
2116             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
2117             felec            = _mm256_sub_ps( _mm256_mul_ps(felec,sw) , _mm256_mul_ps(rinv22,_mm256_mul_ps(velec,dsw)) );
2118             cutoff_mask      = _mm256_cmp_ps(rsq22,rcutoff2,_CMP_LT_OQ);
2119
2120             fscal            = felec;
2121
2122             fscal            = _mm256_and_ps(fscal,cutoff_mask);
2123
2124             /* Calculate temporary vectorial force */
2125             tx               = _mm256_mul_ps(fscal,dx22);
2126             ty               = _mm256_mul_ps(fscal,dy22);
2127             tz               = _mm256_mul_ps(fscal,dz22);
2128
2129             /* Update vectorial force */
2130             fix2             = _mm256_add_ps(fix2,tx);
2131             fiy2             = _mm256_add_ps(fiy2,ty);
2132             fiz2             = _mm256_add_ps(fiz2,tz);
2133
2134             fjx2             = _mm256_add_ps(fjx2,tx);
2135             fjy2             = _mm256_add_ps(fjy2,ty);
2136             fjz2             = _mm256_add_ps(fjz2,tz);
2137
2138             }
2139
2140             /**************************
2141              * CALCULATE INTERACTIONS *
2142              **************************/
2143
2144             if (gmx_mm256_any_lt(rsq23,rcutoff2))
2145             {
2146
2147             r23              = _mm256_mul_ps(rsq23,rinv23);
2148
2149             /* EWALD ELECTROSTATICS */
2150
2151             /* Analytical PME correction */
2152             zeta2            = _mm256_mul_ps(beta2,rsq23);
2153             rinv3            = _mm256_mul_ps(rinvsq23,rinv23);
2154             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
2155             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
2156             felec            = _mm256_mul_ps(qq23,felec);
2157             pmecorrV         = gmx_mm256_pmecorrV_ps(zeta2);
2158             pmecorrV         = _mm256_mul_ps(pmecorrV,beta);
2159             velec            = _mm256_sub_ps(rinv23,pmecorrV);
2160             velec            = _mm256_mul_ps(qq23,velec);
2161
2162             d                = _mm256_sub_ps(r23,rswitch);
2163             d                = _mm256_max_ps(d,_mm256_setzero_ps());
2164             d2               = _mm256_mul_ps(d,d);
2165             sw               = _mm256_add_ps(one,_mm256_mul_ps(d2,_mm256_mul_ps(d,_mm256_add_ps(swV3,_mm256_mul_ps(d,_mm256_add_ps(swV4,_mm256_mul_ps(d,swV5)))))));
2166
2167             dsw              = _mm256_mul_ps(d2,_mm256_add_ps(swF2,_mm256_mul_ps(d,_mm256_add_ps(swF3,_mm256_mul_ps(d,swF4)))));
2168
2169             /* Evaluate switch function */
2170             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
2171             felec            = _mm256_sub_ps( _mm256_mul_ps(felec,sw) , _mm256_mul_ps(rinv23,_mm256_mul_ps(velec,dsw)) );
2172             cutoff_mask      = _mm256_cmp_ps(rsq23,rcutoff2,_CMP_LT_OQ);
2173
2174             fscal            = felec;
2175
2176             fscal            = _mm256_and_ps(fscal,cutoff_mask);
2177
2178             /* Calculate temporary vectorial force */
2179             tx               = _mm256_mul_ps(fscal,dx23);
2180             ty               = _mm256_mul_ps(fscal,dy23);
2181             tz               = _mm256_mul_ps(fscal,dz23);
2182
2183             /* Update vectorial force */
2184             fix2             = _mm256_add_ps(fix2,tx);
2185             fiy2             = _mm256_add_ps(fiy2,ty);
2186             fiz2             = _mm256_add_ps(fiz2,tz);
2187
2188             fjx3             = _mm256_add_ps(fjx3,tx);
2189             fjy3             = _mm256_add_ps(fjy3,ty);
2190             fjz3             = _mm256_add_ps(fjz3,tz);
2191
2192             }
2193
2194             /**************************
2195              * CALCULATE INTERACTIONS *
2196              **************************/
2197
2198             if (gmx_mm256_any_lt(rsq31,rcutoff2))
2199             {
2200
2201             r31              = _mm256_mul_ps(rsq31,rinv31);
2202
2203             /* EWALD ELECTROSTATICS */
2204
2205             /* Analytical PME correction */
2206             zeta2            = _mm256_mul_ps(beta2,rsq31);
2207             rinv3            = _mm256_mul_ps(rinvsq31,rinv31);
2208             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
2209             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
2210             felec            = _mm256_mul_ps(qq31,felec);
2211             pmecorrV         = gmx_mm256_pmecorrV_ps(zeta2);
2212             pmecorrV         = _mm256_mul_ps(pmecorrV,beta);
2213             velec            = _mm256_sub_ps(rinv31,pmecorrV);
2214             velec            = _mm256_mul_ps(qq31,velec);
2215
2216             d                = _mm256_sub_ps(r31,rswitch);
2217             d                = _mm256_max_ps(d,_mm256_setzero_ps());
2218             d2               = _mm256_mul_ps(d,d);
2219             sw               = _mm256_add_ps(one,_mm256_mul_ps(d2,_mm256_mul_ps(d,_mm256_add_ps(swV3,_mm256_mul_ps(d,_mm256_add_ps(swV4,_mm256_mul_ps(d,swV5)))))));
2220
2221             dsw              = _mm256_mul_ps(d2,_mm256_add_ps(swF2,_mm256_mul_ps(d,_mm256_add_ps(swF3,_mm256_mul_ps(d,swF4)))));
2222
2223             /* Evaluate switch function */
2224             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
2225             felec            = _mm256_sub_ps( _mm256_mul_ps(felec,sw) , _mm256_mul_ps(rinv31,_mm256_mul_ps(velec,dsw)) );
2226             cutoff_mask      = _mm256_cmp_ps(rsq31,rcutoff2,_CMP_LT_OQ);
2227
2228             fscal            = felec;
2229
2230             fscal            = _mm256_and_ps(fscal,cutoff_mask);
2231
2232             /* Calculate temporary vectorial force */
2233             tx               = _mm256_mul_ps(fscal,dx31);
2234             ty               = _mm256_mul_ps(fscal,dy31);
2235             tz               = _mm256_mul_ps(fscal,dz31);
2236
2237             /* Update vectorial force */
2238             fix3             = _mm256_add_ps(fix3,tx);
2239             fiy3             = _mm256_add_ps(fiy3,ty);
2240             fiz3             = _mm256_add_ps(fiz3,tz);
2241
2242             fjx1             = _mm256_add_ps(fjx1,tx);
2243             fjy1             = _mm256_add_ps(fjy1,ty);
2244             fjz1             = _mm256_add_ps(fjz1,tz);
2245
2246             }
2247
2248             /**************************
2249              * CALCULATE INTERACTIONS *
2250              **************************/
2251
2252             if (gmx_mm256_any_lt(rsq32,rcutoff2))
2253             {
2254
2255             r32              = _mm256_mul_ps(rsq32,rinv32);
2256
2257             /* EWALD ELECTROSTATICS */
2258
2259             /* Analytical PME correction */
2260             zeta2            = _mm256_mul_ps(beta2,rsq32);
2261             rinv3            = _mm256_mul_ps(rinvsq32,rinv32);
2262             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
2263             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
2264             felec            = _mm256_mul_ps(qq32,felec);
2265             pmecorrV         = gmx_mm256_pmecorrV_ps(zeta2);
2266             pmecorrV         = _mm256_mul_ps(pmecorrV,beta);
2267             velec            = _mm256_sub_ps(rinv32,pmecorrV);
2268             velec            = _mm256_mul_ps(qq32,velec);
2269
2270             d                = _mm256_sub_ps(r32,rswitch);
2271             d                = _mm256_max_ps(d,_mm256_setzero_ps());
2272             d2               = _mm256_mul_ps(d,d);
2273             sw               = _mm256_add_ps(one,_mm256_mul_ps(d2,_mm256_mul_ps(d,_mm256_add_ps(swV3,_mm256_mul_ps(d,_mm256_add_ps(swV4,_mm256_mul_ps(d,swV5)))))));
2274
2275             dsw              = _mm256_mul_ps(d2,_mm256_add_ps(swF2,_mm256_mul_ps(d,_mm256_add_ps(swF3,_mm256_mul_ps(d,swF4)))));
2276
2277             /* Evaluate switch function */
2278             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
2279             felec            = _mm256_sub_ps( _mm256_mul_ps(felec,sw) , _mm256_mul_ps(rinv32,_mm256_mul_ps(velec,dsw)) );
2280             cutoff_mask      = _mm256_cmp_ps(rsq32,rcutoff2,_CMP_LT_OQ);
2281
2282             fscal            = felec;
2283
2284             fscal            = _mm256_and_ps(fscal,cutoff_mask);
2285
2286             /* Calculate temporary vectorial force */
2287             tx               = _mm256_mul_ps(fscal,dx32);
2288             ty               = _mm256_mul_ps(fscal,dy32);
2289             tz               = _mm256_mul_ps(fscal,dz32);
2290
2291             /* Update vectorial force */
2292             fix3             = _mm256_add_ps(fix3,tx);
2293             fiy3             = _mm256_add_ps(fiy3,ty);
2294             fiz3             = _mm256_add_ps(fiz3,tz);
2295
2296             fjx2             = _mm256_add_ps(fjx2,tx);
2297             fjy2             = _mm256_add_ps(fjy2,ty);
2298             fjz2             = _mm256_add_ps(fjz2,tz);
2299
2300             }
2301
2302             /**************************
2303              * CALCULATE INTERACTIONS *
2304              **************************/
2305
2306             if (gmx_mm256_any_lt(rsq33,rcutoff2))
2307             {
2308
2309             r33              = _mm256_mul_ps(rsq33,rinv33);
2310
2311             /* EWALD ELECTROSTATICS */
2312
2313             /* Analytical PME correction */
2314             zeta2            = _mm256_mul_ps(beta2,rsq33);
2315             rinv3            = _mm256_mul_ps(rinvsq33,rinv33);
2316             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
2317             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
2318             felec            = _mm256_mul_ps(qq33,felec);
2319             pmecorrV         = gmx_mm256_pmecorrV_ps(zeta2);
2320             pmecorrV         = _mm256_mul_ps(pmecorrV,beta);
2321             velec            = _mm256_sub_ps(rinv33,pmecorrV);
2322             velec            = _mm256_mul_ps(qq33,velec);
2323
2324             d                = _mm256_sub_ps(r33,rswitch);
2325             d                = _mm256_max_ps(d,_mm256_setzero_ps());
2326             d2               = _mm256_mul_ps(d,d);
2327             sw               = _mm256_add_ps(one,_mm256_mul_ps(d2,_mm256_mul_ps(d,_mm256_add_ps(swV3,_mm256_mul_ps(d,_mm256_add_ps(swV4,_mm256_mul_ps(d,swV5)))))));
2328
2329             dsw              = _mm256_mul_ps(d2,_mm256_add_ps(swF2,_mm256_mul_ps(d,_mm256_add_ps(swF3,_mm256_mul_ps(d,swF4)))));
2330
2331             /* Evaluate switch function */
2332             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
2333             felec            = _mm256_sub_ps( _mm256_mul_ps(felec,sw) , _mm256_mul_ps(rinv33,_mm256_mul_ps(velec,dsw)) );
2334             cutoff_mask      = _mm256_cmp_ps(rsq33,rcutoff2,_CMP_LT_OQ);
2335
2336             fscal            = felec;
2337
2338             fscal            = _mm256_and_ps(fscal,cutoff_mask);
2339
2340             /* Calculate temporary vectorial force */
2341             tx               = _mm256_mul_ps(fscal,dx33);
2342             ty               = _mm256_mul_ps(fscal,dy33);
2343             tz               = _mm256_mul_ps(fscal,dz33);
2344
2345             /* Update vectorial force */
2346             fix3             = _mm256_add_ps(fix3,tx);
2347             fiy3             = _mm256_add_ps(fiy3,ty);
2348             fiz3             = _mm256_add_ps(fiz3,tz);
2349
2350             fjx3             = _mm256_add_ps(fjx3,tx);
2351             fjy3             = _mm256_add_ps(fjy3,ty);
2352             fjz3             = _mm256_add_ps(fjz3,tz);
2353
2354             }
2355
2356             fjptrA             = f+j_coord_offsetA;
2357             fjptrB             = f+j_coord_offsetB;
2358             fjptrC             = f+j_coord_offsetC;
2359             fjptrD             = f+j_coord_offsetD;
2360             fjptrE             = f+j_coord_offsetE;
2361             fjptrF             = f+j_coord_offsetF;
2362             fjptrG             = f+j_coord_offsetG;
2363             fjptrH             = f+j_coord_offsetH;
2364
2365             gmx_mm256_decrement_3rvec_8ptr_swizzle_ps(fjptrA+DIM,fjptrB+DIM,fjptrC+DIM,fjptrD+DIM,
2366                                                       fjptrE+DIM,fjptrF+DIM,fjptrG+DIM,fjptrH+DIM,
2367                                                       fjx1,fjy1,fjz1,fjx2,fjy2,fjz2,fjx3,fjy3,fjz3);
2368
2369             /* Inner loop uses 945 flops */
2370         }
2371
2372         if(jidx<j_index_end)
2373         {
2374
2375             /* Get j neighbor index, and coordinate index */
2376             jnrlistA         = jjnr[jidx];
2377             jnrlistB         = jjnr[jidx+1];
2378             jnrlistC         = jjnr[jidx+2];
2379             jnrlistD         = jjnr[jidx+3];
2380             jnrlistE         = jjnr[jidx+4];
2381             jnrlistF         = jjnr[jidx+5];
2382             jnrlistG         = jjnr[jidx+6];
2383             jnrlistH         = jjnr[jidx+7];
2384             /* Sign of each element will be negative for non-real atoms.
2385              * This mask will be 0xFFFFFFFF for dummy entries and 0x0 for real ones,
2386              * so use it as val = _mm_andnot_ps(mask,val) to clear dummy entries.
2387              */
2388             dummy_mask = gmx_mm256_set_m128(gmx_mm_castsi128_ps(_mm_cmplt_epi32(_mm_loadu_si128((const __m128i *)(jjnr+jidx+4)),_mm_setzero_si128())),
2389                                             gmx_mm_castsi128_ps(_mm_cmplt_epi32(_mm_loadu_si128((const __m128i *)(jjnr+jidx)),_mm_setzero_si128())));
2390
2391             jnrA       = (jnrlistA>=0) ? jnrlistA : 0;
2392             jnrB       = (jnrlistB>=0) ? jnrlistB : 0;
2393             jnrC       = (jnrlistC>=0) ? jnrlistC : 0;
2394             jnrD       = (jnrlistD>=0) ? jnrlistD : 0;
2395             jnrE       = (jnrlistE>=0) ? jnrlistE : 0;
2396             jnrF       = (jnrlistF>=0) ? jnrlistF : 0;
2397             jnrG       = (jnrlistG>=0) ? jnrlistG : 0;
2398             jnrH       = (jnrlistH>=0) ? jnrlistH : 0;
2399             j_coord_offsetA  = DIM*jnrA;
2400             j_coord_offsetB  = DIM*jnrB;
2401             j_coord_offsetC  = DIM*jnrC;
2402             j_coord_offsetD  = DIM*jnrD;
2403             j_coord_offsetE  = DIM*jnrE;
2404             j_coord_offsetF  = DIM*jnrF;
2405             j_coord_offsetG  = DIM*jnrG;
2406             j_coord_offsetH  = DIM*jnrH;
2407
2408             /* load j atom coordinates */
2409             gmx_mm256_load_3rvec_8ptr_swizzle_ps(x+j_coord_offsetA+DIM,x+j_coord_offsetB+DIM,
2410                                                  x+j_coord_offsetC+DIM,x+j_coord_offsetD+DIM,
2411                                                  x+j_coord_offsetE+DIM,x+j_coord_offsetF+DIM,
2412                                                  x+j_coord_offsetG+DIM,x+j_coord_offsetH+DIM,
2413                                                  &jx1,&jy1,&jz1,&jx2,&jy2,&jz2,&jx3,&jy3,&jz3);
2414
2415             /* Calculate displacement vector */
2416             dx11             = _mm256_sub_ps(ix1,jx1);
2417             dy11             = _mm256_sub_ps(iy1,jy1);
2418             dz11             = _mm256_sub_ps(iz1,jz1);
2419             dx12             = _mm256_sub_ps(ix1,jx2);
2420             dy12             = _mm256_sub_ps(iy1,jy2);
2421             dz12             = _mm256_sub_ps(iz1,jz2);
2422             dx13             = _mm256_sub_ps(ix1,jx3);
2423             dy13             = _mm256_sub_ps(iy1,jy3);
2424             dz13             = _mm256_sub_ps(iz1,jz3);
2425             dx21             = _mm256_sub_ps(ix2,jx1);
2426             dy21             = _mm256_sub_ps(iy2,jy1);
2427             dz21             = _mm256_sub_ps(iz2,jz1);
2428             dx22             = _mm256_sub_ps(ix2,jx2);
2429             dy22             = _mm256_sub_ps(iy2,jy2);
2430             dz22             = _mm256_sub_ps(iz2,jz2);
2431             dx23             = _mm256_sub_ps(ix2,jx3);
2432             dy23             = _mm256_sub_ps(iy2,jy3);
2433             dz23             = _mm256_sub_ps(iz2,jz3);
2434             dx31             = _mm256_sub_ps(ix3,jx1);
2435             dy31             = _mm256_sub_ps(iy3,jy1);
2436             dz31             = _mm256_sub_ps(iz3,jz1);
2437             dx32             = _mm256_sub_ps(ix3,jx2);
2438             dy32             = _mm256_sub_ps(iy3,jy2);
2439             dz32             = _mm256_sub_ps(iz3,jz2);
2440             dx33             = _mm256_sub_ps(ix3,jx3);
2441             dy33             = _mm256_sub_ps(iy3,jy3);
2442             dz33             = _mm256_sub_ps(iz3,jz3);
2443
2444             /* Calculate squared distance and things based on it */
2445             rsq11            = gmx_mm256_calc_rsq_ps(dx11,dy11,dz11);
2446             rsq12            = gmx_mm256_calc_rsq_ps(dx12,dy12,dz12);
2447             rsq13            = gmx_mm256_calc_rsq_ps(dx13,dy13,dz13);
2448             rsq21            = gmx_mm256_calc_rsq_ps(dx21,dy21,dz21);
2449             rsq22            = gmx_mm256_calc_rsq_ps(dx22,dy22,dz22);
2450             rsq23            = gmx_mm256_calc_rsq_ps(dx23,dy23,dz23);
2451             rsq31            = gmx_mm256_calc_rsq_ps(dx31,dy31,dz31);
2452             rsq32            = gmx_mm256_calc_rsq_ps(dx32,dy32,dz32);
2453             rsq33            = gmx_mm256_calc_rsq_ps(dx33,dy33,dz33);
2454
2455             rinv11           = gmx_mm256_invsqrt_ps(rsq11);
2456             rinv12           = gmx_mm256_invsqrt_ps(rsq12);
2457             rinv13           = gmx_mm256_invsqrt_ps(rsq13);
2458             rinv21           = gmx_mm256_invsqrt_ps(rsq21);
2459             rinv22           = gmx_mm256_invsqrt_ps(rsq22);
2460             rinv23           = gmx_mm256_invsqrt_ps(rsq23);
2461             rinv31           = gmx_mm256_invsqrt_ps(rsq31);
2462             rinv32           = gmx_mm256_invsqrt_ps(rsq32);
2463             rinv33           = gmx_mm256_invsqrt_ps(rsq33);
2464
2465             rinvsq11         = _mm256_mul_ps(rinv11,rinv11);
2466             rinvsq12         = _mm256_mul_ps(rinv12,rinv12);
2467             rinvsq13         = _mm256_mul_ps(rinv13,rinv13);
2468             rinvsq21         = _mm256_mul_ps(rinv21,rinv21);
2469             rinvsq22         = _mm256_mul_ps(rinv22,rinv22);
2470             rinvsq23         = _mm256_mul_ps(rinv23,rinv23);
2471             rinvsq31         = _mm256_mul_ps(rinv31,rinv31);
2472             rinvsq32         = _mm256_mul_ps(rinv32,rinv32);
2473             rinvsq33         = _mm256_mul_ps(rinv33,rinv33);
2474
2475             fjx1             = _mm256_setzero_ps();
2476             fjy1             = _mm256_setzero_ps();
2477             fjz1             = _mm256_setzero_ps();
2478             fjx2             = _mm256_setzero_ps();
2479             fjy2             = _mm256_setzero_ps();
2480             fjz2             = _mm256_setzero_ps();
2481             fjx3             = _mm256_setzero_ps();
2482             fjy3             = _mm256_setzero_ps();
2483             fjz3             = _mm256_setzero_ps();
2484
2485             /**************************
2486              * CALCULATE INTERACTIONS *
2487              **************************/
2488
2489             if (gmx_mm256_any_lt(rsq11,rcutoff2))
2490             {
2491
2492             r11              = _mm256_mul_ps(rsq11,rinv11);
2493             r11              = _mm256_andnot_ps(dummy_mask,r11);
2494
2495             /* EWALD ELECTROSTATICS */
2496
2497             /* Analytical PME correction */
2498             zeta2            = _mm256_mul_ps(beta2,rsq11);
2499             rinv3            = _mm256_mul_ps(rinvsq11,rinv11);
2500             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
2501             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
2502             felec            = _mm256_mul_ps(qq11,felec);
2503             pmecorrV         = gmx_mm256_pmecorrV_ps(zeta2);
2504             pmecorrV         = _mm256_mul_ps(pmecorrV,beta);
2505             velec            = _mm256_sub_ps(rinv11,pmecorrV);
2506             velec            = _mm256_mul_ps(qq11,velec);
2507
2508             d                = _mm256_sub_ps(r11,rswitch);
2509             d                = _mm256_max_ps(d,_mm256_setzero_ps());
2510             d2               = _mm256_mul_ps(d,d);
2511             sw               = _mm256_add_ps(one,_mm256_mul_ps(d2,_mm256_mul_ps(d,_mm256_add_ps(swV3,_mm256_mul_ps(d,_mm256_add_ps(swV4,_mm256_mul_ps(d,swV5)))))));
2512
2513             dsw              = _mm256_mul_ps(d2,_mm256_add_ps(swF2,_mm256_mul_ps(d,_mm256_add_ps(swF3,_mm256_mul_ps(d,swF4)))));
2514
2515             /* Evaluate switch function */
2516             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
2517             felec            = _mm256_sub_ps( _mm256_mul_ps(felec,sw) , _mm256_mul_ps(rinv11,_mm256_mul_ps(velec,dsw)) );
2518             cutoff_mask      = _mm256_cmp_ps(rsq11,rcutoff2,_CMP_LT_OQ);
2519
2520             fscal            = felec;
2521
2522             fscal            = _mm256_and_ps(fscal,cutoff_mask);
2523
2524             fscal            = _mm256_andnot_ps(dummy_mask,fscal);
2525
2526             /* Calculate temporary vectorial force */
2527             tx               = _mm256_mul_ps(fscal,dx11);
2528             ty               = _mm256_mul_ps(fscal,dy11);
2529             tz               = _mm256_mul_ps(fscal,dz11);
2530
2531             /* Update vectorial force */
2532             fix1             = _mm256_add_ps(fix1,tx);
2533             fiy1             = _mm256_add_ps(fiy1,ty);
2534             fiz1             = _mm256_add_ps(fiz1,tz);
2535
2536             fjx1             = _mm256_add_ps(fjx1,tx);
2537             fjy1             = _mm256_add_ps(fjy1,ty);
2538             fjz1             = _mm256_add_ps(fjz1,tz);
2539
2540             }
2541
2542             /**************************
2543              * CALCULATE INTERACTIONS *
2544              **************************/
2545
2546             if (gmx_mm256_any_lt(rsq12,rcutoff2))
2547             {
2548
2549             r12              = _mm256_mul_ps(rsq12,rinv12);
2550             r12              = _mm256_andnot_ps(dummy_mask,r12);
2551
2552             /* EWALD ELECTROSTATICS */
2553
2554             /* Analytical PME correction */
2555             zeta2            = _mm256_mul_ps(beta2,rsq12);
2556             rinv3            = _mm256_mul_ps(rinvsq12,rinv12);
2557             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
2558             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
2559             felec            = _mm256_mul_ps(qq12,felec);
2560             pmecorrV         = gmx_mm256_pmecorrV_ps(zeta2);
2561             pmecorrV         = _mm256_mul_ps(pmecorrV,beta);
2562             velec            = _mm256_sub_ps(rinv12,pmecorrV);
2563             velec            = _mm256_mul_ps(qq12,velec);
2564
2565             d                = _mm256_sub_ps(r12,rswitch);
2566             d                = _mm256_max_ps(d,_mm256_setzero_ps());
2567             d2               = _mm256_mul_ps(d,d);
2568             sw               = _mm256_add_ps(one,_mm256_mul_ps(d2,_mm256_mul_ps(d,_mm256_add_ps(swV3,_mm256_mul_ps(d,_mm256_add_ps(swV4,_mm256_mul_ps(d,swV5)))))));
2569
2570             dsw              = _mm256_mul_ps(d2,_mm256_add_ps(swF2,_mm256_mul_ps(d,_mm256_add_ps(swF3,_mm256_mul_ps(d,swF4)))));
2571
2572             /* Evaluate switch function */
2573             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
2574             felec            = _mm256_sub_ps( _mm256_mul_ps(felec,sw) , _mm256_mul_ps(rinv12,_mm256_mul_ps(velec,dsw)) );
2575             cutoff_mask      = _mm256_cmp_ps(rsq12,rcutoff2,_CMP_LT_OQ);
2576
2577             fscal            = felec;
2578
2579             fscal            = _mm256_and_ps(fscal,cutoff_mask);
2580
2581             fscal            = _mm256_andnot_ps(dummy_mask,fscal);
2582
2583             /* Calculate temporary vectorial force */
2584             tx               = _mm256_mul_ps(fscal,dx12);
2585             ty               = _mm256_mul_ps(fscal,dy12);
2586             tz               = _mm256_mul_ps(fscal,dz12);
2587
2588             /* Update vectorial force */
2589             fix1             = _mm256_add_ps(fix1,tx);
2590             fiy1             = _mm256_add_ps(fiy1,ty);
2591             fiz1             = _mm256_add_ps(fiz1,tz);
2592
2593             fjx2             = _mm256_add_ps(fjx2,tx);
2594             fjy2             = _mm256_add_ps(fjy2,ty);
2595             fjz2             = _mm256_add_ps(fjz2,tz);
2596
2597             }
2598
2599             /**************************
2600              * CALCULATE INTERACTIONS *
2601              **************************/
2602
2603             if (gmx_mm256_any_lt(rsq13,rcutoff2))
2604             {
2605
2606             r13              = _mm256_mul_ps(rsq13,rinv13);
2607             r13              = _mm256_andnot_ps(dummy_mask,r13);
2608
2609             /* EWALD ELECTROSTATICS */
2610
2611             /* Analytical PME correction */
2612             zeta2            = _mm256_mul_ps(beta2,rsq13);
2613             rinv3            = _mm256_mul_ps(rinvsq13,rinv13);
2614             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
2615             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
2616             felec            = _mm256_mul_ps(qq13,felec);
2617             pmecorrV         = gmx_mm256_pmecorrV_ps(zeta2);
2618             pmecorrV         = _mm256_mul_ps(pmecorrV,beta);
2619             velec            = _mm256_sub_ps(rinv13,pmecorrV);
2620             velec            = _mm256_mul_ps(qq13,velec);
2621
2622             d                = _mm256_sub_ps(r13,rswitch);
2623             d                = _mm256_max_ps(d,_mm256_setzero_ps());
2624             d2               = _mm256_mul_ps(d,d);
2625             sw               = _mm256_add_ps(one,_mm256_mul_ps(d2,_mm256_mul_ps(d,_mm256_add_ps(swV3,_mm256_mul_ps(d,_mm256_add_ps(swV4,_mm256_mul_ps(d,swV5)))))));
2626
2627             dsw              = _mm256_mul_ps(d2,_mm256_add_ps(swF2,_mm256_mul_ps(d,_mm256_add_ps(swF3,_mm256_mul_ps(d,swF4)))));
2628
2629             /* Evaluate switch function */
2630             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
2631             felec            = _mm256_sub_ps( _mm256_mul_ps(felec,sw) , _mm256_mul_ps(rinv13,_mm256_mul_ps(velec,dsw)) );
2632             cutoff_mask      = _mm256_cmp_ps(rsq13,rcutoff2,_CMP_LT_OQ);
2633
2634             fscal            = felec;
2635
2636             fscal            = _mm256_and_ps(fscal,cutoff_mask);
2637
2638             fscal            = _mm256_andnot_ps(dummy_mask,fscal);
2639
2640             /* Calculate temporary vectorial force */
2641             tx               = _mm256_mul_ps(fscal,dx13);
2642             ty               = _mm256_mul_ps(fscal,dy13);
2643             tz               = _mm256_mul_ps(fscal,dz13);
2644
2645             /* Update vectorial force */
2646             fix1             = _mm256_add_ps(fix1,tx);
2647             fiy1             = _mm256_add_ps(fiy1,ty);
2648             fiz1             = _mm256_add_ps(fiz1,tz);
2649
2650             fjx3             = _mm256_add_ps(fjx3,tx);
2651             fjy3             = _mm256_add_ps(fjy3,ty);
2652             fjz3             = _mm256_add_ps(fjz3,tz);
2653
2654             }
2655
2656             /**************************
2657              * CALCULATE INTERACTIONS *
2658              **************************/
2659
2660             if (gmx_mm256_any_lt(rsq21,rcutoff2))
2661             {
2662
2663             r21              = _mm256_mul_ps(rsq21,rinv21);
2664             r21              = _mm256_andnot_ps(dummy_mask,r21);
2665
2666             /* EWALD ELECTROSTATICS */
2667
2668             /* Analytical PME correction */
2669             zeta2            = _mm256_mul_ps(beta2,rsq21);
2670             rinv3            = _mm256_mul_ps(rinvsq21,rinv21);
2671             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
2672             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
2673             felec            = _mm256_mul_ps(qq21,felec);
2674             pmecorrV         = gmx_mm256_pmecorrV_ps(zeta2);
2675             pmecorrV         = _mm256_mul_ps(pmecorrV,beta);
2676             velec            = _mm256_sub_ps(rinv21,pmecorrV);
2677             velec            = _mm256_mul_ps(qq21,velec);
2678
2679             d                = _mm256_sub_ps(r21,rswitch);
2680             d                = _mm256_max_ps(d,_mm256_setzero_ps());
2681             d2               = _mm256_mul_ps(d,d);
2682             sw               = _mm256_add_ps(one,_mm256_mul_ps(d2,_mm256_mul_ps(d,_mm256_add_ps(swV3,_mm256_mul_ps(d,_mm256_add_ps(swV4,_mm256_mul_ps(d,swV5)))))));
2683
2684             dsw              = _mm256_mul_ps(d2,_mm256_add_ps(swF2,_mm256_mul_ps(d,_mm256_add_ps(swF3,_mm256_mul_ps(d,swF4)))));
2685
2686             /* Evaluate switch function */
2687             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
2688             felec            = _mm256_sub_ps( _mm256_mul_ps(felec,sw) , _mm256_mul_ps(rinv21,_mm256_mul_ps(velec,dsw)) );
2689             cutoff_mask      = _mm256_cmp_ps(rsq21,rcutoff2,_CMP_LT_OQ);
2690
2691             fscal            = felec;
2692
2693             fscal            = _mm256_and_ps(fscal,cutoff_mask);
2694
2695             fscal            = _mm256_andnot_ps(dummy_mask,fscal);
2696
2697             /* Calculate temporary vectorial force */
2698             tx               = _mm256_mul_ps(fscal,dx21);
2699             ty               = _mm256_mul_ps(fscal,dy21);
2700             tz               = _mm256_mul_ps(fscal,dz21);
2701
2702             /* Update vectorial force */
2703             fix2             = _mm256_add_ps(fix2,tx);
2704             fiy2             = _mm256_add_ps(fiy2,ty);
2705             fiz2             = _mm256_add_ps(fiz2,tz);
2706
2707             fjx1             = _mm256_add_ps(fjx1,tx);
2708             fjy1             = _mm256_add_ps(fjy1,ty);
2709             fjz1             = _mm256_add_ps(fjz1,tz);
2710
2711             }
2712
2713             /**************************
2714              * CALCULATE INTERACTIONS *
2715              **************************/
2716
2717             if (gmx_mm256_any_lt(rsq22,rcutoff2))
2718             {
2719
2720             r22              = _mm256_mul_ps(rsq22,rinv22);
2721             r22              = _mm256_andnot_ps(dummy_mask,r22);
2722
2723             /* EWALD ELECTROSTATICS */
2724
2725             /* Analytical PME correction */
2726             zeta2            = _mm256_mul_ps(beta2,rsq22);
2727             rinv3            = _mm256_mul_ps(rinvsq22,rinv22);
2728             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
2729             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
2730             felec            = _mm256_mul_ps(qq22,felec);
2731             pmecorrV         = gmx_mm256_pmecorrV_ps(zeta2);
2732             pmecorrV         = _mm256_mul_ps(pmecorrV,beta);
2733             velec            = _mm256_sub_ps(rinv22,pmecorrV);
2734             velec            = _mm256_mul_ps(qq22,velec);
2735
2736             d                = _mm256_sub_ps(r22,rswitch);
2737             d                = _mm256_max_ps(d,_mm256_setzero_ps());
2738             d2               = _mm256_mul_ps(d,d);
2739             sw               = _mm256_add_ps(one,_mm256_mul_ps(d2,_mm256_mul_ps(d,_mm256_add_ps(swV3,_mm256_mul_ps(d,_mm256_add_ps(swV4,_mm256_mul_ps(d,swV5)))))));
2740
2741             dsw              = _mm256_mul_ps(d2,_mm256_add_ps(swF2,_mm256_mul_ps(d,_mm256_add_ps(swF3,_mm256_mul_ps(d,swF4)))));
2742
2743             /* Evaluate switch function */
2744             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
2745             felec            = _mm256_sub_ps( _mm256_mul_ps(felec,sw) , _mm256_mul_ps(rinv22,_mm256_mul_ps(velec,dsw)) );
2746             cutoff_mask      = _mm256_cmp_ps(rsq22,rcutoff2,_CMP_LT_OQ);
2747
2748             fscal            = felec;
2749
2750             fscal            = _mm256_and_ps(fscal,cutoff_mask);
2751
2752             fscal            = _mm256_andnot_ps(dummy_mask,fscal);
2753
2754             /* Calculate temporary vectorial force */
2755             tx               = _mm256_mul_ps(fscal,dx22);
2756             ty               = _mm256_mul_ps(fscal,dy22);
2757             tz               = _mm256_mul_ps(fscal,dz22);
2758
2759             /* Update vectorial force */
2760             fix2             = _mm256_add_ps(fix2,tx);
2761             fiy2             = _mm256_add_ps(fiy2,ty);
2762             fiz2             = _mm256_add_ps(fiz2,tz);
2763
2764             fjx2             = _mm256_add_ps(fjx2,tx);
2765             fjy2             = _mm256_add_ps(fjy2,ty);
2766             fjz2             = _mm256_add_ps(fjz2,tz);
2767
2768             }
2769
2770             /**************************
2771              * CALCULATE INTERACTIONS *
2772              **************************/
2773
2774             if (gmx_mm256_any_lt(rsq23,rcutoff2))
2775             {
2776
2777             r23              = _mm256_mul_ps(rsq23,rinv23);
2778             r23              = _mm256_andnot_ps(dummy_mask,r23);
2779
2780             /* EWALD ELECTROSTATICS */
2781
2782             /* Analytical PME correction */
2783             zeta2            = _mm256_mul_ps(beta2,rsq23);
2784             rinv3            = _mm256_mul_ps(rinvsq23,rinv23);
2785             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
2786             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
2787             felec            = _mm256_mul_ps(qq23,felec);
2788             pmecorrV         = gmx_mm256_pmecorrV_ps(zeta2);
2789             pmecorrV         = _mm256_mul_ps(pmecorrV,beta);
2790             velec            = _mm256_sub_ps(rinv23,pmecorrV);
2791             velec            = _mm256_mul_ps(qq23,velec);
2792
2793             d                = _mm256_sub_ps(r23,rswitch);
2794             d                = _mm256_max_ps(d,_mm256_setzero_ps());
2795             d2               = _mm256_mul_ps(d,d);
2796             sw               = _mm256_add_ps(one,_mm256_mul_ps(d2,_mm256_mul_ps(d,_mm256_add_ps(swV3,_mm256_mul_ps(d,_mm256_add_ps(swV4,_mm256_mul_ps(d,swV5)))))));
2797
2798             dsw              = _mm256_mul_ps(d2,_mm256_add_ps(swF2,_mm256_mul_ps(d,_mm256_add_ps(swF3,_mm256_mul_ps(d,swF4)))));
2799
2800             /* Evaluate switch function */
2801             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
2802             felec            = _mm256_sub_ps( _mm256_mul_ps(felec,sw) , _mm256_mul_ps(rinv23,_mm256_mul_ps(velec,dsw)) );
2803             cutoff_mask      = _mm256_cmp_ps(rsq23,rcutoff2,_CMP_LT_OQ);
2804
2805             fscal            = felec;
2806
2807             fscal            = _mm256_and_ps(fscal,cutoff_mask);
2808
2809             fscal            = _mm256_andnot_ps(dummy_mask,fscal);
2810
2811             /* Calculate temporary vectorial force */
2812             tx               = _mm256_mul_ps(fscal,dx23);
2813             ty               = _mm256_mul_ps(fscal,dy23);
2814             tz               = _mm256_mul_ps(fscal,dz23);
2815
2816             /* Update vectorial force */
2817             fix2             = _mm256_add_ps(fix2,tx);
2818             fiy2             = _mm256_add_ps(fiy2,ty);
2819             fiz2             = _mm256_add_ps(fiz2,tz);
2820
2821             fjx3             = _mm256_add_ps(fjx3,tx);
2822             fjy3             = _mm256_add_ps(fjy3,ty);
2823             fjz3             = _mm256_add_ps(fjz3,tz);
2824
2825             }
2826
2827             /**************************
2828              * CALCULATE INTERACTIONS *
2829              **************************/
2830
2831             if (gmx_mm256_any_lt(rsq31,rcutoff2))
2832             {
2833
2834             r31              = _mm256_mul_ps(rsq31,rinv31);
2835             r31              = _mm256_andnot_ps(dummy_mask,r31);
2836
2837             /* EWALD ELECTROSTATICS */
2838
2839             /* Analytical PME correction */
2840             zeta2            = _mm256_mul_ps(beta2,rsq31);
2841             rinv3            = _mm256_mul_ps(rinvsq31,rinv31);
2842             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
2843             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
2844             felec            = _mm256_mul_ps(qq31,felec);
2845             pmecorrV         = gmx_mm256_pmecorrV_ps(zeta2);
2846             pmecorrV         = _mm256_mul_ps(pmecorrV,beta);
2847             velec            = _mm256_sub_ps(rinv31,pmecorrV);
2848             velec            = _mm256_mul_ps(qq31,velec);
2849
2850             d                = _mm256_sub_ps(r31,rswitch);
2851             d                = _mm256_max_ps(d,_mm256_setzero_ps());
2852             d2               = _mm256_mul_ps(d,d);
2853             sw               = _mm256_add_ps(one,_mm256_mul_ps(d2,_mm256_mul_ps(d,_mm256_add_ps(swV3,_mm256_mul_ps(d,_mm256_add_ps(swV4,_mm256_mul_ps(d,swV5)))))));
2854
2855             dsw              = _mm256_mul_ps(d2,_mm256_add_ps(swF2,_mm256_mul_ps(d,_mm256_add_ps(swF3,_mm256_mul_ps(d,swF4)))));
2856
2857             /* Evaluate switch function */
2858             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
2859             felec            = _mm256_sub_ps( _mm256_mul_ps(felec,sw) , _mm256_mul_ps(rinv31,_mm256_mul_ps(velec,dsw)) );
2860             cutoff_mask      = _mm256_cmp_ps(rsq31,rcutoff2,_CMP_LT_OQ);
2861
2862             fscal            = felec;
2863
2864             fscal            = _mm256_and_ps(fscal,cutoff_mask);
2865
2866             fscal            = _mm256_andnot_ps(dummy_mask,fscal);
2867
2868             /* Calculate temporary vectorial force */
2869             tx               = _mm256_mul_ps(fscal,dx31);
2870             ty               = _mm256_mul_ps(fscal,dy31);
2871             tz               = _mm256_mul_ps(fscal,dz31);
2872
2873             /* Update vectorial force */
2874             fix3             = _mm256_add_ps(fix3,tx);
2875             fiy3             = _mm256_add_ps(fiy3,ty);
2876             fiz3             = _mm256_add_ps(fiz3,tz);
2877
2878             fjx1             = _mm256_add_ps(fjx1,tx);
2879             fjy1             = _mm256_add_ps(fjy1,ty);
2880             fjz1             = _mm256_add_ps(fjz1,tz);
2881
2882             }
2883
2884             /**************************
2885              * CALCULATE INTERACTIONS *
2886              **************************/
2887
2888             if (gmx_mm256_any_lt(rsq32,rcutoff2))
2889             {
2890
2891             r32              = _mm256_mul_ps(rsq32,rinv32);
2892             r32              = _mm256_andnot_ps(dummy_mask,r32);
2893
2894             /* EWALD ELECTROSTATICS */
2895
2896             /* Analytical PME correction */
2897             zeta2            = _mm256_mul_ps(beta2,rsq32);
2898             rinv3            = _mm256_mul_ps(rinvsq32,rinv32);
2899             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
2900             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
2901             felec            = _mm256_mul_ps(qq32,felec);
2902             pmecorrV         = gmx_mm256_pmecorrV_ps(zeta2);
2903             pmecorrV         = _mm256_mul_ps(pmecorrV,beta);
2904             velec            = _mm256_sub_ps(rinv32,pmecorrV);
2905             velec            = _mm256_mul_ps(qq32,velec);
2906
2907             d                = _mm256_sub_ps(r32,rswitch);
2908             d                = _mm256_max_ps(d,_mm256_setzero_ps());
2909             d2               = _mm256_mul_ps(d,d);
2910             sw               = _mm256_add_ps(one,_mm256_mul_ps(d2,_mm256_mul_ps(d,_mm256_add_ps(swV3,_mm256_mul_ps(d,_mm256_add_ps(swV4,_mm256_mul_ps(d,swV5)))))));
2911
2912             dsw              = _mm256_mul_ps(d2,_mm256_add_ps(swF2,_mm256_mul_ps(d,_mm256_add_ps(swF3,_mm256_mul_ps(d,swF4)))));
2913
2914             /* Evaluate switch function */
2915             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
2916             felec            = _mm256_sub_ps( _mm256_mul_ps(felec,sw) , _mm256_mul_ps(rinv32,_mm256_mul_ps(velec,dsw)) );
2917             cutoff_mask      = _mm256_cmp_ps(rsq32,rcutoff2,_CMP_LT_OQ);
2918
2919             fscal            = felec;
2920
2921             fscal            = _mm256_and_ps(fscal,cutoff_mask);
2922
2923             fscal            = _mm256_andnot_ps(dummy_mask,fscal);
2924
2925             /* Calculate temporary vectorial force */
2926             tx               = _mm256_mul_ps(fscal,dx32);
2927             ty               = _mm256_mul_ps(fscal,dy32);
2928             tz               = _mm256_mul_ps(fscal,dz32);
2929
2930             /* Update vectorial force */
2931             fix3             = _mm256_add_ps(fix3,tx);
2932             fiy3             = _mm256_add_ps(fiy3,ty);
2933             fiz3             = _mm256_add_ps(fiz3,tz);
2934
2935             fjx2             = _mm256_add_ps(fjx2,tx);
2936             fjy2             = _mm256_add_ps(fjy2,ty);
2937             fjz2             = _mm256_add_ps(fjz2,tz);
2938
2939             }
2940
2941             /**************************
2942              * CALCULATE INTERACTIONS *
2943              **************************/
2944
2945             if (gmx_mm256_any_lt(rsq33,rcutoff2))
2946             {
2947
2948             r33              = _mm256_mul_ps(rsq33,rinv33);
2949             r33              = _mm256_andnot_ps(dummy_mask,r33);
2950
2951             /* EWALD ELECTROSTATICS */
2952
2953             /* Analytical PME correction */
2954             zeta2            = _mm256_mul_ps(beta2,rsq33);
2955             rinv3            = _mm256_mul_ps(rinvsq33,rinv33);
2956             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
2957             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
2958             felec            = _mm256_mul_ps(qq33,felec);
2959             pmecorrV         = gmx_mm256_pmecorrV_ps(zeta2);
2960             pmecorrV         = _mm256_mul_ps(pmecorrV,beta);
2961             velec            = _mm256_sub_ps(rinv33,pmecorrV);
2962             velec            = _mm256_mul_ps(qq33,velec);
2963
2964             d                = _mm256_sub_ps(r33,rswitch);
2965             d                = _mm256_max_ps(d,_mm256_setzero_ps());
2966             d2               = _mm256_mul_ps(d,d);
2967             sw               = _mm256_add_ps(one,_mm256_mul_ps(d2,_mm256_mul_ps(d,_mm256_add_ps(swV3,_mm256_mul_ps(d,_mm256_add_ps(swV4,_mm256_mul_ps(d,swV5)))))));
2968
2969             dsw              = _mm256_mul_ps(d2,_mm256_add_ps(swF2,_mm256_mul_ps(d,_mm256_add_ps(swF3,_mm256_mul_ps(d,swF4)))));
2970
2971             /* Evaluate switch function */
2972             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
2973             felec            = _mm256_sub_ps( _mm256_mul_ps(felec,sw) , _mm256_mul_ps(rinv33,_mm256_mul_ps(velec,dsw)) );
2974             cutoff_mask      = _mm256_cmp_ps(rsq33,rcutoff2,_CMP_LT_OQ);
2975
2976             fscal            = felec;
2977
2978             fscal            = _mm256_and_ps(fscal,cutoff_mask);
2979
2980             fscal            = _mm256_andnot_ps(dummy_mask,fscal);
2981
2982             /* Calculate temporary vectorial force */
2983             tx               = _mm256_mul_ps(fscal,dx33);
2984             ty               = _mm256_mul_ps(fscal,dy33);
2985             tz               = _mm256_mul_ps(fscal,dz33);
2986
2987             /* Update vectorial force */
2988             fix3             = _mm256_add_ps(fix3,tx);
2989             fiy3             = _mm256_add_ps(fiy3,ty);
2990             fiz3             = _mm256_add_ps(fiz3,tz);
2991
2992             fjx3             = _mm256_add_ps(fjx3,tx);
2993             fjy3             = _mm256_add_ps(fjy3,ty);
2994             fjz3             = _mm256_add_ps(fjz3,tz);
2995
2996             }
2997
2998             fjptrA             = (jnrlistA>=0) ? f+j_coord_offsetA : scratch;
2999             fjptrB             = (jnrlistB>=0) ? f+j_coord_offsetB : scratch;
3000             fjptrC             = (jnrlistC>=0) ? f+j_coord_offsetC : scratch;
3001             fjptrD             = (jnrlistD>=0) ? f+j_coord_offsetD : scratch;
3002             fjptrE             = (jnrlistE>=0) ? f+j_coord_offsetE : scratch;
3003             fjptrF             = (jnrlistF>=0) ? f+j_coord_offsetF : scratch;
3004             fjptrG             = (jnrlistG>=0) ? f+j_coord_offsetG : scratch;
3005             fjptrH             = (jnrlistH>=0) ? f+j_coord_offsetH : scratch;
3006
3007             gmx_mm256_decrement_3rvec_8ptr_swizzle_ps(fjptrA+DIM,fjptrB+DIM,fjptrC+DIM,fjptrD+DIM,
3008                                                       fjptrE+DIM,fjptrF+DIM,fjptrG+DIM,fjptrH+DIM,
3009                                                       fjx1,fjy1,fjz1,fjx2,fjy2,fjz2,fjx3,fjy3,fjz3);
3010
3011             /* Inner loop uses 954 flops */
3012         }
3013
3014         /* End of innermost loop */
3015
3016         gmx_mm256_update_iforce_3atom_swizzle_ps(fix1,fiy1,fiz1,fix2,fiy2,fiz2,fix3,fiy3,fiz3,
3017                                                  f+i_coord_offset+DIM,fshift+i_shift_offset);
3018
3019         /* Increment number of inner iterations */
3020         inneriter                  += j_index_end - j_index_start;
3021
3022         /* Outer loop uses 18 flops */
3023     }
3024
3025     /* Increment number of outer iterations */
3026     outeriter        += nri;
3027
3028     /* Update outer/inner flops */
3029
3030     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_W4W4_F,outeriter*18 + inneriter*954);
3031 }