src/gromacs/gmxlib/nonbonded/nb_kernel_avx_256_double/nb_kernel_ElecEwSw_VdwLJSw_GeomW4P1_avx_256_double.c

   1 /*
   2  * This file is part of the GROMACS molecular simulation package.
   3  *
   4  * Copyright (c) 2012,2013, by the GROMACS development team, led by
   5  * Mark Abraham, David van der Spoel, Berk Hess, and Erik Lindahl,
   6  * and including many others, as listed in the AUTHORS file in the
   7  * top-level source directory and at http://www.gromacs.org.
   8  *
   9  * GROMACS is free software; you can redistribute it and/or
  10  * modify it under the terms of the GNU Lesser General Public License
  11  * as published by the Free Software Foundation; either version 2.1
  12  * of the License, or (at your option) any later version.
  13  *
  14  * GROMACS is distributed in the hope that it will be useful,
  15  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  16  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  17  * Lesser General Public License for more details.
  18  *
  19  * You should have received a copy of the GNU Lesser General Public
  20  * License along with GROMACS; if not, see
  21  * http://www.gnu.org/licenses, or write to the Free Software Foundation,
  22  * Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02110-1301  USA.
  23  *
  24  * If you want to redistribute modifications to GROMACS, please
  25  * consider that scientific software is very special. Version
  26  * control is crucial - bugs must be traceable. We will be happy to
  27  * consider code for inclusion in the official distribution, but
  28  * derived work must not be called official GROMACS. Details are found
  29  * in the README & COPYING files - if they are missing, get the
  30  * official version at http://www.gromacs.org.
  31  *
  32  * To help us fund GROMACS development, we humbly ask that you cite
  33  * the research papers on the package. Check out http://www.gromacs.org.
  34  */
  35 /*
  36  * Note: this file was generated by the GROMACS avx_256_double kernel generator.
  37  */
  38 #ifdef HAVE_CONFIG_H
  39 #include <config.h>
  40 #endif
  41
  42 #include <math.h>
  43
  44 #include "../nb_kernel.h"
  45 #include "types/simple.h"
  46 #include "vec.h"
  47 #include "nrnb.h"
  48
  49 #include "gromacs/simd/math_x86_avx_256_double.h"
  50 #include "kernelutil_x86_avx_256_double.h"
  51
  52 /*
  53  * Gromacs nonbonded kernel:   nb_kernel_ElecEwSw_VdwLJSw_GeomW4P1_VF_avx_256_double
  54  * Electrostatics interaction: Ewald
  55  * VdW interaction:            LennardJones
  56  * Geometry:                   Water4-Particle
  57  * Calculate force/pot:        PotentialAndForce
  58  */
  59 void
  60 nb_kernel_ElecEwSw_VdwLJSw_GeomW4P1_VF_avx_256_double
  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 refer to j loop unrolling done with AVX, e.g. for the four 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              jnrlistA,jnrlistB,jnrlistC,jnrlistD;
  78     int              jnrlistE,jnrlistF,jnrlistG,jnrlistH;
  79     int              j_coord_offsetA,j_coord_offsetB,j_coord_offsetC,j_coord_offsetD;
  80     int              *iinr,*jindex,*jjnr,*shiftidx,*gid;
  81     real             rcutoff_scalar;
  82     real             *shiftvec,*fshift,*x,*f;
  83     real             *fjptrA,*fjptrB,*fjptrC,*fjptrD;
  84     real             scratch[4*DIM];
  85     __m256d          tx,ty,tz,fscal,rcutoff,rcutoff2,jidxall;
  86     real *           vdwioffsetptr0;
  87     __m256d          ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
  88     real *           vdwioffsetptr1;
  89     __m256d          ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
  90     real *           vdwioffsetptr2;
  91     __m256d          ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
  92     real *           vdwioffsetptr3;
  93     __m256d          ix3,iy3,iz3,fix3,fiy3,fiz3,iq3,isai3;
  94     int              vdwjidx0A,vdwjidx0B,vdwjidx0C,vdwjidx0D;
  95     __m256d          jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
  96     __m256d          dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00;
  97     __m256d          dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10;
  98     __m256d          dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20;
  99     __m256d          dx30,dy30,dz30,rsq30,rinv30,rinvsq30,r30,qq30,c6_30,c12_30;
 100     __m256d          velec,felec,velecsum,facel,crf,krf,krf2;
 101     real             *charge;
 102     int              nvdwtype;
 103     __m256d          rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,sh_vdw_invrcut6;
 104     int              *vdwtype;
 105     real             *vdwparam;
 106     __m256d          one_sixth   = _mm256_set1_pd(1.0/6.0);
 107     __m256d          one_twelfth = _mm256_set1_pd(1.0/12.0);
 108     __m128i          ewitab;
 109     __m256d          ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace,ewtabF,ewtabFn,ewtabD,ewtabV;
 110     __m256d          beta,beta2,beta3,zeta2,pmecorrF,pmecorrV,rinv3;
 111     real             *ewtab;
 112     __m256d          rswitch,swV3,swV4,swV5,swF2,swF3,swF4,d,d2,sw,dsw;
 113     real             rswitch_scalar,d_scalar;
 114     __m256d          dummy_mask,cutoff_mask;
 115     __m128           tmpmask0,tmpmask1;
 116     __m256d          signbit = _mm256_castsi256_pd( _mm256_set1_epi32(0x80000000) );
 117     __m256d          one     = _mm256_set1_pd(1.0);
 118     __m256d          two     = _mm256_set1_pd(2.0);
 119     x                = xx[0];
 120     f                = ff[0];
 121
 122     nri              = nlist->nri;
 123     iinr             = nlist->iinr;
 124     jindex           = nlist->jindex;
 125     jjnr             = nlist->jjnr;
 126     shiftidx         = nlist->shift;
 127     gid              = nlist->gid;
 128     shiftvec         = fr->shift_vec[0];
 129     fshift           = fr->fshift[0];
 130     facel            = _mm256_set1_pd(fr->epsfac);
 131     charge           = mdatoms->chargeA;
 132     nvdwtype         = fr->ntype;
 133     vdwparam         = fr->nbfp;
 134     vdwtype          = mdatoms->typeA;
 135
 136     sh_ewald         = _mm256_set1_pd(fr->ic->sh_ewald);
 137     beta             = _mm256_set1_pd(fr->ic->ewaldcoeff_q);
 138     beta2            = _mm256_mul_pd(beta,beta);
 139     beta3            = _mm256_mul_pd(beta,beta2);
 140
 141     ewtab            = fr->ic->tabq_coul_FDV0;
 142     ewtabscale       = _mm256_set1_pd(fr->ic->tabq_scale);
 143     ewtabhalfspace   = _mm256_set1_pd(0.5/fr->ic->tabq_scale);
 144
 145     /* Setup water-specific parameters */
 146     inr              = nlist->iinr[0];
 147     iq1              = _mm256_mul_pd(facel,_mm256_set1_pd(charge[inr+1]));
 148     iq2              = _mm256_mul_pd(facel,_mm256_set1_pd(charge[inr+2]));
 149     iq3              = _mm256_mul_pd(facel,_mm256_set1_pd(charge[inr+3]));
 150     vdwioffsetptr0   = vdwparam+2*nvdwtype*vdwtype[inr+0];
 151
 152     /* When we use explicit cutoffs the value must be identical for elec and VdW, so use elec as an arbitrary choice */
 153     rcutoff_scalar   = fr->rcoulomb;
 154     rcutoff          = _mm256_set1_pd(rcutoff_scalar);
 155     rcutoff2         = _mm256_mul_pd(rcutoff,rcutoff);
 156
 157     rswitch_scalar   = fr->rcoulomb_switch;
 158     rswitch          = _mm256_set1_pd(rswitch_scalar);
 159     /* Setup switch parameters */
 160     d_scalar         = rcutoff_scalar-rswitch_scalar;
 161     d                = _mm256_set1_pd(d_scalar);
 162     swV3             = _mm256_set1_pd(-10.0/(d_scalar*d_scalar*d_scalar));
 163     swV4             = _mm256_set1_pd( 15.0/(d_scalar*d_scalar*d_scalar*d_scalar));
 164     swV5             = _mm256_set1_pd( -6.0/(d_scalar*d_scalar*d_scalar*d_scalar*d_scalar));
 165     swF2             = _mm256_set1_pd(-30.0/(d_scalar*d_scalar*d_scalar));
 166     swF3             = _mm256_set1_pd( 60.0/(d_scalar*d_scalar*d_scalar*d_scalar));
 167     swF4             = _mm256_set1_pd(-30.0/(d_scalar*d_scalar*d_scalar*d_scalar*d_scalar));
 168
 169     /* Avoid stupid compiler warnings */
 170     jnrA = jnrB = jnrC = jnrD = 0;
 171     j_coord_offsetA = 0;
 172     j_coord_offsetB = 0;
 173     j_coord_offsetC = 0;
 174     j_coord_offsetD = 0;
 175
 176     outeriter        = 0;
 177     inneriter        = 0;
 178
 179     for(iidx=0;iidx<4*DIM;iidx++)
 180     {
 181         scratch[iidx] = 0.0;
 182     }
 183
 184     /* Start outer loop over neighborlists */
 185     for(iidx=0; iidx<nri; iidx++)
 186     {
 187         /* Load shift vector for this list */
 188         i_shift_offset   = DIM*shiftidx[iidx];
 189
 190         /* Load limits for loop over neighbors */
 191         j_index_start    = jindex[iidx];
 192         j_index_end      = jindex[iidx+1];
 193
 194         /* Get outer coordinate index */
 195         inr              = iinr[iidx];
 196         i_coord_offset   = DIM*inr;
 197
 198         /* Load i particle coords and add shift vector */
 199         gmx_mm256_load_shift_and_4rvec_broadcast_pd(shiftvec+i_shift_offset,x+i_coord_offset,
 200                                                     &ix0,&iy0,&iz0,&ix1,&iy1,&iz1,&ix2,&iy2,&iz2,&ix3,&iy3,&iz3);
 201
 202         fix0             = _mm256_setzero_pd();
 203         fiy0             = _mm256_setzero_pd();
 204         fiz0             = _mm256_setzero_pd();
 205         fix1             = _mm256_setzero_pd();
 206         fiy1             = _mm256_setzero_pd();
 207         fiz1             = _mm256_setzero_pd();
 208         fix2             = _mm256_setzero_pd();
 209         fiy2             = _mm256_setzero_pd();
 210         fiz2             = _mm256_setzero_pd();
 211         fix3             = _mm256_setzero_pd();
 212         fiy3             = _mm256_setzero_pd();
 213         fiz3             = _mm256_setzero_pd();
 214
 215         /* Reset potential sums */
 216         velecsum         = _mm256_setzero_pd();
 217         vvdwsum          = _mm256_setzero_pd();
 218
 219         /* Start inner kernel loop */
 220         for(jidx=j_index_start; jidx<j_index_end && jjnr[jidx+3]>=0; jidx+=4)
 221         {
 222
 223             /* Get j neighbor index, and coordinate index */
 224             jnrA             = jjnr[jidx];
 225             jnrB             = jjnr[jidx+1];
 226             jnrC             = jjnr[jidx+2];
 227             jnrD             = jjnr[jidx+3];
 228             j_coord_offsetA  = DIM*jnrA;
 229             j_coord_offsetB  = DIM*jnrB;
 230             j_coord_offsetC  = DIM*jnrC;
 231             j_coord_offsetD  = DIM*jnrD;
 232
 233             /* load j atom coordinates */
 234             gmx_mm256_load_1rvec_4ptr_swizzle_pd(x+j_coord_offsetA,x+j_coord_offsetB,
 235                                                  x+j_coord_offsetC,x+j_coord_offsetD,
 236                                                  &jx0,&jy0,&jz0);
 237
 238             /* Calculate displacement vector */
 239             dx00             = _mm256_sub_pd(ix0,jx0);
 240             dy00             = _mm256_sub_pd(iy0,jy0);
 241             dz00             = _mm256_sub_pd(iz0,jz0);
 242             dx10             = _mm256_sub_pd(ix1,jx0);
 243             dy10             = _mm256_sub_pd(iy1,jy0);
 244             dz10             = _mm256_sub_pd(iz1,jz0);
 245             dx20             = _mm256_sub_pd(ix2,jx0);
 246             dy20             = _mm256_sub_pd(iy2,jy0);
 247             dz20             = _mm256_sub_pd(iz2,jz0);
 248             dx30             = _mm256_sub_pd(ix3,jx0);
 249             dy30             = _mm256_sub_pd(iy3,jy0);
 250             dz30             = _mm256_sub_pd(iz3,jz0);
 251
 252             /* Calculate squared distance and things based on it */
 253             rsq00            = gmx_mm256_calc_rsq_pd(dx00,dy00,dz00);
 254             rsq10            = gmx_mm256_calc_rsq_pd(dx10,dy10,dz10);
 255             rsq20            = gmx_mm256_calc_rsq_pd(dx20,dy20,dz20);
 256             rsq30            = gmx_mm256_calc_rsq_pd(dx30,dy30,dz30);
 257
 258             rinv00           = gmx_mm256_invsqrt_pd(rsq00);
 259             rinv10           = gmx_mm256_invsqrt_pd(rsq10);
 260             rinv20           = gmx_mm256_invsqrt_pd(rsq20);
 261             rinv30           = gmx_mm256_invsqrt_pd(rsq30);
 262
 263             rinvsq00         = _mm256_mul_pd(rinv00,rinv00);
 264             rinvsq10         = _mm256_mul_pd(rinv10,rinv10);
 265             rinvsq20         = _mm256_mul_pd(rinv20,rinv20);
 266             rinvsq30         = _mm256_mul_pd(rinv30,rinv30);
 267
 268             /* Load parameters for j particles */
 269             jq0              = gmx_mm256_load_4real_swizzle_pd(charge+jnrA+0,charge+jnrB+0,
 270                                                                  charge+jnrC+0,charge+jnrD+0);
 271             vdwjidx0A        = 2*vdwtype[jnrA+0];
 272             vdwjidx0B        = 2*vdwtype[jnrB+0];
 273             vdwjidx0C        = 2*vdwtype[jnrC+0];
 274             vdwjidx0D        = 2*vdwtype[jnrD+0];
 275
 276             fjx0             = _mm256_setzero_pd();
 277             fjy0             = _mm256_setzero_pd();
 278             fjz0             = _mm256_setzero_pd();
 279
 280             /**************************
 281              * CALCULATE INTERACTIONS *
 282              **************************/
 283
 284             if (gmx_mm256_any_lt(rsq00,rcutoff2))
 285             {
 286
 287             r00              = _mm256_mul_pd(rsq00,rinv00);
 288
 289             /* Compute parameters for interactions between i and j atoms */
 290             gmx_mm256_load_4pair_swizzle_pd(vdwioffsetptr0+vdwjidx0A,
 291                                             vdwioffsetptr0+vdwjidx0B,
 292                                             vdwioffsetptr0+vdwjidx0C,
 293                                             vdwioffsetptr0+vdwjidx0D,
 294                                             &c6_00,&c12_00);
 295
 296             /* LENNARD-JONES DISPERSION/REPULSION */
 297
 298             rinvsix          = _mm256_mul_pd(_mm256_mul_pd(rinvsq00,rinvsq00),rinvsq00);
 299             vvdw6            = _mm256_mul_pd(c6_00,rinvsix);
 300             vvdw12           = _mm256_mul_pd(c12_00,_mm256_mul_pd(rinvsix,rinvsix));
 301             vvdw             = _mm256_sub_pd( _mm256_mul_pd(vvdw12,one_twelfth) , _mm256_mul_pd(vvdw6,one_sixth) );
 302             fvdw             = _mm256_mul_pd(_mm256_sub_pd(vvdw12,vvdw6),rinvsq00);
 303
 304             d                = _mm256_sub_pd(r00,rswitch);
 305             d                = _mm256_max_pd(d,_mm256_setzero_pd());
 306             d2               = _mm256_mul_pd(d,d);
 307             sw               = _mm256_add_pd(one,_mm256_mul_pd(d2,_mm256_mul_pd(d,_mm256_add_pd(swV3,_mm256_mul_pd(d,_mm256_add_pd(swV4,_mm256_mul_pd(d,swV5)))))));
 308
 309             dsw              = _mm256_mul_pd(d2,_mm256_add_pd(swF2,_mm256_mul_pd(d,_mm256_add_pd(swF3,_mm256_mul_pd(d,swF4)))));
 310
 311             /* Evaluate switch function */
 312             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 313             fvdw             = _mm256_sub_pd( _mm256_mul_pd(fvdw,sw) , _mm256_mul_pd(rinv00,_mm256_mul_pd(vvdw,dsw)) );
 314             vvdw             = _mm256_mul_pd(vvdw,sw);
 315             cutoff_mask      = _mm256_cmp_pd(rsq00,rcutoff2,_CMP_LT_OQ);
 316
 317             /* Update potential sum for this i atom from the interaction with this j atom. */
 318             vvdw             = _mm256_and_pd(vvdw,cutoff_mask);
 319             vvdwsum          = _mm256_add_pd(vvdwsum,vvdw);
 320
 321             fscal            = fvdw;
 322
 323             fscal            = _mm256_and_pd(fscal,cutoff_mask);
 324
 325             /* Calculate temporary vectorial force */
 326             tx               = _mm256_mul_pd(fscal,dx00);
 327             ty               = _mm256_mul_pd(fscal,dy00);
 328             tz               = _mm256_mul_pd(fscal,dz00);
 329
 330             /* Update vectorial force */
 331             fix0             = _mm256_add_pd(fix0,tx);
 332             fiy0             = _mm256_add_pd(fiy0,ty);
 333             fiz0             = _mm256_add_pd(fiz0,tz);
 334
 335             fjx0             = _mm256_add_pd(fjx0,tx);
 336             fjy0             = _mm256_add_pd(fjy0,ty);
 337             fjz0             = _mm256_add_pd(fjz0,tz);
 338
 339             }
 340
 341             /**************************
 342              * CALCULATE INTERACTIONS *
 343              **************************/
 344
 345             if (gmx_mm256_any_lt(rsq10,rcutoff2))
 346             {
 347
 348             r10              = _mm256_mul_pd(rsq10,rinv10);
 349
 350             /* Compute parameters for interactions between i and j atoms */
 351             qq10             = _mm256_mul_pd(iq1,jq0);
 352
 353             /* EWALD ELECTROSTATICS */
 354
 355             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 356             ewrt             = _mm256_mul_pd(r10,ewtabscale);
 357             ewitab           = _mm256_cvttpd_epi32(ewrt);
 358             eweps            = _mm256_sub_pd(ewrt,_mm256_round_pd(ewrt, _MM_FROUND_FLOOR));
 359             ewitab           = _mm_slli_epi32(ewitab,2);
 360             ewtabF           = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,0) );
 361             ewtabD           = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,1) );
 362             ewtabV           = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,2) );
 363             ewtabFn          = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,3) );
 364             GMX_MM256_FULLTRANSPOSE4_PD(ewtabF,ewtabD,ewtabV,ewtabFn);
 365             felec            = _mm256_add_pd(ewtabF,_mm256_mul_pd(eweps,ewtabD));
 366             velec            = _mm256_sub_pd(ewtabV,_mm256_mul_pd(_mm256_mul_pd(ewtabhalfspace,eweps),_mm256_add_pd(ewtabF,felec)));
 367             velec            = _mm256_mul_pd(qq10,_mm256_sub_pd(rinv10,velec));
 368             felec            = _mm256_mul_pd(_mm256_mul_pd(qq10,rinv10),_mm256_sub_pd(rinvsq10,felec));
 369
 370             d                = _mm256_sub_pd(r10,rswitch);
 371             d                = _mm256_max_pd(d,_mm256_setzero_pd());
 372             d2               = _mm256_mul_pd(d,d);
 373             sw               = _mm256_add_pd(one,_mm256_mul_pd(d2,_mm256_mul_pd(d,_mm256_add_pd(swV3,_mm256_mul_pd(d,_mm256_add_pd(swV4,_mm256_mul_pd(d,swV5)))))));
 374
 375             dsw              = _mm256_mul_pd(d2,_mm256_add_pd(swF2,_mm256_mul_pd(d,_mm256_add_pd(swF3,_mm256_mul_pd(d,swF4)))));
 376
 377             /* Evaluate switch function */
 378             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 379             felec            = _mm256_sub_pd( _mm256_mul_pd(felec,sw) , _mm256_mul_pd(rinv10,_mm256_mul_pd(velec,dsw)) );
 380             velec            = _mm256_mul_pd(velec,sw);
 381             cutoff_mask      = _mm256_cmp_pd(rsq10,rcutoff2,_CMP_LT_OQ);
 382
 383             /* Update potential sum for this i atom from the interaction with this j atom. */
 384             velec            = _mm256_and_pd(velec,cutoff_mask);
 385             velecsum         = _mm256_add_pd(velecsum,velec);
 386
 387             fscal            = felec;
 388
 389             fscal            = _mm256_and_pd(fscal,cutoff_mask);
 390
 391             /* Calculate temporary vectorial force */
 392             tx               = _mm256_mul_pd(fscal,dx10);
 393             ty               = _mm256_mul_pd(fscal,dy10);
 394             tz               = _mm256_mul_pd(fscal,dz10);
 395
 396             /* Update vectorial force */
 397             fix1             = _mm256_add_pd(fix1,tx);
 398             fiy1             = _mm256_add_pd(fiy1,ty);
 399             fiz1             = _mm256_add_pd(fiz1,tz);
 400
 401             fjx0             = _mm256_add_pd(fjx0,tx);
 402             fjy0             = _mm256_add_pd(fjy0,ty);
 403             fjz0             = _mm256_add_pd(fjz0,tz);
 404
 405             }
 406
 407             /**************************
 408              * CALCULATE INTERACTIONS *
 409              **************************/
 410
 411             if (gmx_mm256_any_lt(rsq20,rcutoff2))
 412             {
 413
 414             r20              = _mm256_mul_pd(rsq20,rinv20);
 415
 416             /* Compute parameters for interactions between i and j atoms */
 417             qq20             = _mm256_mul_pd(iq2,jq0);
 418
 419             /* EWALD ELECTROSTATICS */
 420
 421             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 422             ewrt             = _mm256_mul_pd(r20,ewtabscale);
 423             ewitab           = _mm256_cvttpd_epi32(ewrt);
 424             eweps            = _mm256_sub_pd(ewrt,_mm256_round_pd(ewrt, _MM_FROUND_FLOOR));
 425             ewitab           = _mm_slli_epi32(ewitab,2);
 426             ewtabF           = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,0) );
 427             ewtabD           = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,1) );
 428             ewtabV           = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,2) );
 429             ewtabFn          = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,3) );
 430             GMX_MM256_FULLTRANSPOSE4_PD(ewtabF,ewtabD,ewtabV,ewtabFn);
 431             felec            = _mm256_add_pd(ewtabF,_mm256_mul_pd(eweps,ewtabD));
 432             velec            = _mm256_sub_pd(ewtabV,_mm256_mul_pd(_mm256_mul_pd(ewtabhalfspace,eweps),_mm256_add_pd(ewtabF,felec)));
 433             velec            = _mm256_mul_pd(qq20,_mm256_sub_pd(rinv20,velec));
 434             felec            = _mm256_mul_pd(_mm256_mul_pd(qq20,rinv20),_mm256_sub_pd(rinvsq20,felec));
 435
 436             d                = _mm256_sub_pd(r20,rswitch);
 437             d                = _mm256_max_pd(d,_mm256_setzero_pd());
 438             d2               = _mm256_mul_pd(d,d);
 439             sw               = _mm256_add_pd(one,_mm256_mul_pd(d2,_mm256_mul_pd(d,_mm256_add_pd(swV3,_mm256_mul_pd(d,_mm256_add_pd(swV4,_mm256_mul_pd(d,swV5)))))));
 440
 441             dsw              = _mm256_mul_pd(d2,_mm256_add_pd(swF2,_mm256_mul_pd(d,_mm256_add_pd(swF3,_mm256_mul_pd(d,swF4)))));
 442
 443             /* Evaluate switch function */
 444             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 445             felec            = _mm256_sub_pd( _mm256_mul_pd(felec,sw) , _mm256_mul_pd(rinv20,_mm256_mul_pd(velec,dsw)) );
 446             velec            = _mm256_mul_pd(velec,sw);
 447             cutoff_mask      = _mm256_cmp_pd(rsq20,rcutoff2,_CMP_LT_OQ);
 448
 449             /* Update potential sum for this i atom from the interaction with this j atom. */
 450             velec            = _mm256_and_pd(velec,cutoff_mask);
 451             velecsum         = _mm256_add_pd(velecsum,velec);
 452
 453             fscal            = felec;
 454
 455             fscal            = _mm256_and_pd(fscal,cutoff_mask);
 456
 457             /* Calculate temporary vectorial force */
 458             tx               = _mm256_mul_pd(fscal,dx20);
 459             ty               = _mm256_mul_pd(fscal,dy20);
 460             tz               = _mm256_mul_pd(fscal,dz20);
 461
 462             /* Update vectorial force */
 463             fix2             = _mm256_add_pd(fix2,tx);
 464             fiy2             = _mm256_add_pd(fiy2,ty);
 465             fiz2             = _mm256_add_pd(fiz2,tz);
 466
 467             fjx0             = _mm256_add_pd(fjx0,tx);
 468             fjy0             = _mm256_add_pd(fjy0,ty);
 469             fjz0             = _mm256_add_pd(fjz0,tz);
 470
 471             }
 472
 473             /**************************
 474              * CALCULATE INTERACTIONS *
 475              **************************/
 476
 477             if (gmx_mm256_any_lt(rsq30,rcutoff2))
 478             {
 479
 480             r30              = _mm256_mul_pd(rsq30,rinv30);
 481
 482             /* Compute parameters for interactions between i and j atoms */
 483             qq30             = _mm256_mul_pd(iq3,jq0);
 484
 485             /* EWALD ELECTROSTATICS */
 486
 487             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 488             ewrt             = _mm256_mul_pd(r30,ewtabscale);
 489             ewitab           = _mm256_cvttpd_epi32(ewrt);
 490             eweps            = _mm256_sub_pd(ewrt,_mm256_round_pd(ewrt, _MM_FROUND_FLOOR));
 491             ewitab           = _mm_slli_epi32(ewitab,2);
 492             ewtabF           = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,0) );
 493             ewtabD           = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,1) );
 494             ewtabV           = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,2) );
 495             ewtabFn          = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,3) );
 496             GMX_MM256_FULLTRANSPOSE4_PD(ewtabF,ewtabD,ewtabV,ewtabFn);
 497             felec            = _mm256_add_pd(ewtabF,_mm256_mul_pd(eweps,ewtabD));
 498             velec            = _mm256_sub_pd(ewtabV,_mm256_mul_pd(_mm256_mul_pd(ewtabhalfspace,eweps),_mm256_add_pd(ewtabF,felec)));
 499             velec            = _mm256_mul_pd(qq30,_mm256_sub_pd(rinv30,velec));
 500             felec            = _mm256_mul_pd(_mm256_mul_pd(qq30,rinv30),_mm256_sub_pd(rinvsq30,felec));
 501
 502             d                = _mm256_sub_pd(r30,rswitch);
 503             d                = _mm256_max_pd(d,_mm256_setzero_pd());
 504             d2               = _mm256_mul_pd(d,d);
 505             sw               = _mm256_add_pd(one,_mm256_mul_pd(d2,_mm256_mul_pd(d,_mm256_add_pd(swV3,_mm256_mul_pd(d,_mm256_add_pd(swV4,_mm256_mul_pd(d,swV5)))))));
 506
 507             dsw              = _mm256_mul_pd(d2,_mm256_add_pd(swF2,_mm256_mul_pd(d,_mm256_add_pd(swF3,_mm256_mul_pd(d,swF4)))));
 508
 509             /* Evaluate switch function */
 510             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 511             felec            = _mm256_sub_pd( _mm256_mul_pd(felec,sw) , _mm256_mul_pd(rinv30,_mm256_mul_pd(velec,dsw)) );
 512             velec            = _mm256_mul_pd(velec,sw);
 513             cutoff_mask      = _mm256_cmp_pd(rsq30,rcutoff2,_CMP_LT_OQ);
 514
 515             /* Update potential sum for this i atom from the interaction with this j atom. */
 516             velec            = _mm256_and_pd(velec,cutoff_mask);
 517             velecsum         = _mm256_add_pd(velecsum,velec);
 518
 519             fscal            = felec;
 520
 521             fscal            = _mm256_and_pd(fscal,cutoff_mask);
 522
 523             /* Calculate temporary vectorial force */
 524             tx               = _mm256_mul_pd(fscal,dx30);
 525             ty               = _mm256_mul_pd(fscal,dy30);
 526             tz               = _mm256_mul_pd(fscal,dz30);
 527
 528             /* Update vectorial force */
 529             fix3             = _mm256_add_pd(fix3,tx);
 530             fiy3             = _mm256_add_pd(fiy3,ty);
 531             fiz3             = _mm256_add_pd(fiz3,tz);
 532
 533             fjx0             = _mm256_add_pd(fjx0,tx);
 534             fjy0             = _mm256_add_pd(fjy0,ty);
 535             fjz0             = _mm256_add_pd(fjz0,tz);
 536
 537             }
 538
 539             fjptrA             = f+j_coord_offsetA;
 540             fjptrB             = f+j_coord_offsetB;
 541             fjptrC             = f+j_coord_offsetC;
 542             fjptrD             = f+j_coord_offsetD;
 543
 544             gmx_mm256_decrement_1rvec_4ptr_swizzle_pd(fjptrA,fjptrB,fjptrC,fjptrD,fjx0,fjy0,fjz0);
 545
 546             /* Inner loop uses 257 flops */
 547         }
 548
 549         if(jidx<j_index_end)
 550         {
 551
 552             /* Get j neighbor index, and coordinate index */
 553             jnrlistA         = jjnr[jidx];
 554             jnrlistB         = jjnr[jidx+1];
 555             jnrlistC         = jjnr[jidx+2];
 556             jnrlistD         = jjnr[jidx+3];
 557             /* Sign of each element will be negative for non-real atoms.
 558              * This mask will be 0xFFFFFFFF for dummy entries and 0x0 for real ones,
 559              * so use it as val = _mm_andnot_pd(mask,val) to clear dummy entries.
 560              */
 561             tmpmask0 = gmx_mm_castsi128_ps(_mm_cmplt_epi32(_mm_loadu_si128((const __m128i *)(jjnr+jidx)),_mm_setzero_si128()));
 562
 563             tmpmask1 = _mm_permute_ps(tmpmask0,_GMX_MM_PERMUTE(3,3,2,2));
 564             tmpmask0 = _mm_permute_ps(tmpmask0,_GMX_MM_PERMUTE(1,1,0,0));
 565             dummy_mask = _mm256_castps_pd(gmx_mm256_set_m128(tmpmask1,tmpmask0));
 566
 567             jnrA       = (jnrlistA>=0) ? jnrlistA : 0;
 568             jnrB       = (jnrlistB>=0) ? jnrlistB : 0;
 569             jnrC       = (jnrlistC>=0) ? jnrlistC : 0;
 570             jnrD       = (jnrlistD>=0) ? jnrlistD : 0;
 571             j_coord_offsetA  = DIM*jnrA;
 572             j_coord_offsetB  = DIM*jnrB;
 573             j_coord_offsetC  = DIM*jnrC;
 574             j_coord_offsetD  = DIM*jnrD;
 575
 576             /* load j atom coordinates */
 577             gmx_mm256_load_1rvec_4ptr_swizzle_pd(x+j_coord_offsetA,x+j_coord_offsetB,
 578                                                  x+j_coord_offsetC,x+j_coord_offsetD,
 579                                                  &jx0,&jy0,&jz0);
 580
 581             /* Calculate displacement vector */
 582             dx00             = _mm256_sub_pd(ix0,jx0);
 583             dy00             = _mm256_sub_pd(iy0,jy0);
 584             dz00             = _mm256_sub_pd(iz0,jz0);
 585             dx10             = _mm256_sub_pd(ix1,jx0);
 586             dy10             = _mm256_sub_pd(iy1,jy0);
 587             dz10             = _mm256_sub_pd(iz1,jz0);
 588             dx20             = _mm256_sub_pd(ix2,jx0);
 589             dy20             = _mm256_sub_pd(iy2,jy0);
 590             dz20             = _mm256_sub_pd(iz2,jz0);
 591             dx30             = _mm256_sub_pd(ix3,jx0);
 592             dy30             = _mm256_sub_pd(iy3,jy0);
 593             dz30             = _mm256_sub_pd(iz3,jz0);
 594
 595             /* Calculate squared distance and things based on it */
 596             rsq00            = gmx_mm256_calc_rsq_pd(dx00,dy00,dz00);
 597             rsq10            = gmx_mm256_calc_rsq_pd(dx10,dy10,dz10);
 598             rsq20            = gmx_mm256_calc_rsq_pd(dx20,dy20,dz20);
 599             rsq30            = gmx_mm256_calc_rsq_pd(dx30,dy30,dz30);
 600
 601             rinv00           = gmx_mm256_invsqrt_pd(rsq00);
 602             rinv10           = gmx_mm256_invsqrt_pd(rsq10);
 603             rinv20           = gmx_mm256_invsqrt_pd(rsq20);
 604             rinv30           = gmx_mm256_invsqrt_pd(rsq30);
 605
 606             rinvsq00         = _mm256_mul_pd(rinv00,rinv00);
 607             rinvsq10         = _mm256_mul_pd(rinv10,rinv10);
 608             rinvsq20         = _mm256_mul_pd(rinv20,rinv20);
 609             rinvsq30         = _mm256_mul_pd(rinv30,rinv30);
 610
 611             /* Load parameters for j particles */
 612             jq0              = gmx_mm256_load_4real_swizzle_pd(charge+jnrA+0,charge+jnrB+0,
 613                                                                  charge+jnrC+0,charge+jnrD+0);
 614             vdwjidx0A        = 2*vdwtype[jnrA+0];
 615             vdwjidx0B        = 2*vdwtype[jnrB+0];
 616             vdwjidx0C        = 2*vdwtype[jnrC+0];
 617             vdwjidx0D        = 2*vdwtype[jnrD+0];
 618
 619             fjx0             = _mm256_setzero_pd();
 620             fjy0             = _mm256_setzero_pd();
 621             fjz0             = _mm256_setzero_pd();
 622
 623             /**************************
 624              * CALCULATE INTERACTIONS *
 625              **************************/
 626
 627             if (gmx_mm256_any_lt(rsq00,rcutoff2))
 628             {
 629
 630             r00              = _mm256_mul_pd(rsq00,rinv00);
 631             r00              = _mm256_andnot_pd(dummy_mask,r00);
 632
 633             /* Compute parameters for interactions between i and j atoms */
 634             gmx_mm256_load_4pair_swizzle_pd(vdwioffsetptr0+vdwjidx0A,
 635                                             vdwioffsetptr0+vdwjidx0B,
 636                                             vdwioffsetptr0+vdwjidx0C,
 637                                             vdwioffsetptr0+vdwjidx0D,
 638                                             &c6_00,&c12_00);
 639
 640             /* LENNARD-JONES DISPERSION/REPULSION */
 641
 642             rinvsix          = _mm256_mul_pd(_mm256_mul_pd(rinvsq00,rinvsq00),rinvsq00);
 643             vvdw6            = _mm256_mul_pd(c6_00,rinvsix);
 644             vvdw12           = _mm256_mul_pd(c12_00,_mm256_mul_pd(rinvsix,rinvsix));
 645             vvdw             = _mm256_sub_pd( _mm256_mul_pd(vvdw12,one_twelfth) , _mm256_mul_pd(vvdw6,one_sixth) );
 646             fvdw             = _mm256_mul_pd(_mm256_sub_pd(vvdw12,vvdw6),rinvsq00);
 647
 648             d                = _mm256_sub_pd(r00,rswitch);
 649             d                = _mm256_max_pd(d,_mm256_setzero_pd());
 650             d2               = _mm256_mul_pd(d,d);
 651             sw               = _mm256_add_pd(one,_mm256_mul_pd(d2,_mm256_mul_pd(d,_mm256_add_pd(swV3,_mm256_mul_pd(d,_mm256_add_pd(swV4,_mm256_mul_pd(d,swV5)))))));
 652
 653             dsw              = _mm256_mul_pd(d2,_mm256_add_pd(swF2,_mm256_mul_pd(d,_mm256_add_pd(swF3,_mm256_mul_pd(d,swF4)))));
 654
 655             /* Evaluate switch function */
 656             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 657             fvdw             = _mm256_sub_pd( _mm256_mul_pd(fvdw,sw) , _mm256_mul_pd(rinv00,_mm256_mul_pd(vvdw,dsw)) );
 658             vvdw             = _mm256_mul_pd(vvdw,sw);
 659             cutoff_mask      = _mm256_cmp_pd(rsq00,rcutoff2,_CMP_LT_OQ);
 660
 661             /* Update potential sum for this i atom from the interaction with this j atom. */
 662             vvdw             = _mm256_and_pd(vvdw,cutoff_mask);
 663             vvdw             = _mm256_andnot_pd(dummy_mask,vvdw);
 664             vvdwsum          = _mm256_add_pd(vvdwsum,vvdw);
 665
 666             fscal            = fvdw;
 667
 668             fscal            = _mm256_and_pd(fscal,cutoff_mask);
 669
 670             fscal            = _mm256_andnot_pd(dummy_mask,fscal);
 671
 672             /* Calculate temporary vectorial force */
 673             tx               = _mm256_mul_pd(fscal,dx00);
 674             ty               = _mm256_mul_pd(fscal,dy00);
 675             tz               = _mm256_mul_pd(fscal,dz00);
 676
 677             /* Update vectorial force */
 678             fix0             = _mm256_add_pd(fix0,tx);
 679             fiy0             = _mm256_add_pd(fiy0,ty);
 680             fiz0             = _mm256_add_pd(fiz0,tz);
 681
 682             fjx0             = _mm256_add_pd(fjx0,tx);
 683             fjy0             = _mm256_add_pd(fjy0,ty);
 684             fjz0             = _mm256_add_pd(fjz0,tz);
 685
 686             }
 687
 688             /**************************
 689              * CALCULATE INTERACTIONS *
 690              **************************/
 691
 692             if (gmx_mm256_any_lt(rsq10,rcutoff2))
 693             {
 694
 695             r10              = _mm256_mul_pd(rsq10,rinv10);
 696             r10              = _mm256_andnot_pd(dummy_mask,r10);
 697
 698             /* Compute parameters for interactions between i and j atoms */
 699             qq10             = _mm256_mul_pd(iq1,jq0);
 700
 701             /* EWALD ELECTROSTATICS */
 702
 703             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 704             ewrt             = _mm256_mul_pd(r10,ewtabscale);
 705             ewitab           = _mm256_cvttpd_epi32(ewrt);
 706             eweps            = _mm256_sub_pd(ewrt,_mm256_round_pd(ewrt, _MM_FROUND_FLOOR));
 707             ewitab           = _mm_slli_epi32(ewitab,2);
 708             ewtabF           = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,0) );
 709             ewtabD           = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,1) );
 710             ewtabV           = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,2) );
 711             ewtabFn          = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,3) );
 712             GMX_MM256_FULLTRANSPOSE4_PD(ewtabF,ewtabD,ewtabV,ewtabFn);
 713             felec            = _mm256_add_pd(ewtabF,_mm256_mul_pd(eweps,ewtabD));
 714             velec            = _mm256_sub_pd(ewtabV,_mm256_mul_pd(_mm256_mul_pd(ewtabhalfspace,eweps),_mm256_add_pd(ewtabF,felec)));
 715             velec            = _mm256_mul_pd(qq10,_mm256_sub_pd(rinv10,velec));
 716             felec            = _mm256_mul_pd(_mm256_mul_pd(qq10,rinv10),_mm256_sub_pd(rinvsq10,felec));
 717
 718             d                = _mm256_sub_pd(r10,rswitch);
 719             d                = _mm256_max_pd(d,_mm256_setzero_pd());
 720             d2               = _mm256_mul_pd(d,d);
 721             sw               = _mm256_add_pd(one,_mm256_mul_pd(d2,_mm256_mul_pd(d,_mm256_add_pd(swV3,_mm256_mul_pd(d,_mm256_add_pd(swV4,_mm256_mul_pd(d,swV5)))))));
 722
 723             dsw              = _mm256_mul_pd(d2,_mm256_add_pd(swF2,_mm256_mul_pd(d,_mm256_add_pd(swF3,_mm256_mul_pd(d,swF4)))));
 724
 725             /* Evaluate switch function */
 726             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 727             felec            = _mm256_sub_pd( _mm256_mul_pd(felec,sw) , _mm256_mul_pd(rinv10,_mm256_mul_pd(velec,dsw)) );
 728             velec            = _mm256_mul_pd(velec,sw);
 729             cutoff_mask      = _mm256_cmp_pd(rsq10,rcutoff2,_CMP_LT_OQ);
 730
 731             /* Update potential sum for this i atom from the interaction with this j atom. */
 732             velec            = _mm256_and_pd(velec,cutoff_mask);
 733             velec            = _mm256_andnot_pd(dummy_mask,velec);
 734             velecsum         = _mm256_add_pd(velecsum,velec);
 735
 736             fscal            = felec;
 737
 738             fscal            = _mm256_and_pd(fscal,cutoff_mask);
 739
 740             fscal            = _mm256_andnot_pd(dummy_mask,fscal);
 741
 742             /* Calculate temporary vectorial force */
 743             tx               = _mm256_mul_pd(fscal,dx10);
 744             ty               = _mm256_mul_pd(fscal,dy10);
 745             tz               = _mm256_mul_pd(fscal,dz10);
 746
 747             /* Update vectorial force */
 748             fix1             = _mm256_add_pd(fix1,tx);
 749             fiy1             = _mm256_add_pd(fiy1,ty);
 750             fiz1             = _mm256_add_pd(fiz1,tz);
 751
 752             fjx0             = _mm256_add_pd(fjx0,tx);
 753             fjy0             = _mm256_add_pd(fjy0,ty);
 754             fjz0             = _mm256_add_pd(fjz0,tz);
 755
 756             }
 757
 758             /**************************
 759              * CALCULATE INTERACTIONS *
 760              **************************/
 761
 762             if (gmx_mm256_any_lt(rsq20,rcutoff2))
 763             {
 764
 765             r20              = _mm256_mul_pd(rsq20,rinv20);
 766             r20              = _mm256_andnot_pd(dummy_mask,r20);
 767
 768             /* Compute parameters for interactions between i and j atoms */
 769             qq20             = _mm256_mul_pd(iq2,jq0);
 770
 771             /* EWALD ELECTROSTATICS */
 772
 773             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 774             ewrt             = _mm256_mul_pd(r20,ewtabscale);
 775             ewitab           = _mm256_cvttpd_epi32(ewrt);
 776             eweps            = _mm256_sub_pd(ewrt,_mm256_round_pd(ewrt, _MM_FROUND_FLOOR));
 777             ewitab           = _mm_slli_epi32(ewitab,2);
 778             ewtabF           = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,0) );
 779             ewtabD           = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,1) );
 780             ewtabV           = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,2) );
 781             ewtabFn          = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,3) );
 782             GMX_MM256_FULLTRANSPOSE4_PD(ewtabF,ewtabD,ewtabV,ewtabFn);
 783             felec            = _mm256_add_pd(ewtabF,_mm256_mul_pd(eweps,ewtabD));
 784             velec            = _mm256_sub_pd(ewtabV,_mm256_mul_pd(_mm256_mul_pd(ewtabhalfspace,eweps),_mm256_add_pd(ewtabF,felec)));
 785             velec            = _mm256_mul_pd(qq20,_mm256_sub_pd(rinv20,velec));
 786             felec            = _mm256_mul_pd(_mm256_mul_pd(qq20,rinv20),_mm256_sub_pd(rinvsq20,felec));
 787
 788             d                = _mm256_sub_pd(r20,rswitch);
 789             d                = _mm256_max_pd(d,_mm256_setzero_pd());
 790             d2               = _mm256_mul_pd(d,d);
 791             sw               = _mm256_add_pd(one,_mm256_mul_pd(d2,_mm256_mul_pd(d,_mm256_add_pd(swV3,_mm256_mul_pd(d,_mm256_add_pd(swV4,_mm256_mul_pd(d,swV5)))))));
 792
 793             dsw              = _mm256_mul_pd(d2,_mm256_add_pd(swF2,_mm256_mul_pd(d,_mm256_add_pd(swF3,_mm256_mul_pd(d,swF4)))));
 794
 795             /* Evaluate switch function */
 796             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 797             felec            = _mm256_sub_pd( _mm256_mul_pd(felec,sw) , _mm256_mul_pd(rinv20,_mm256_mul_pd(velec,dsw)) );
 798             velec            = _mm256_mul_pd(velec,sw);
 799             cutoff_mask      = _mm256_cmp_pd(rsq20,rcutoff2,_CMP_LT_OQ);
 800
 801             /* Update potential sum for this i atom from the interaction with this j atom. */
 802             velec            = _mm256_and_pd(velec,cutoff_mask);
 803             velec            = _mm256_andnot_pd(dummy_mask,velec);
 804             velecsum         = _mm256_add_pd(velecsum,velec);
 805
 806             fscal            = felec;
 807
 808             fscal            = _mm256_and_pd(fscal,cutoff_mask);
 809
 810             fscal            = _mm256_andnot_pd(dummy_mask,fscal);
 811
 812             /* Calculate temporary vectorial force */
 813             tx               = _mm256_mul_pd(fscal,dx20);
 814             ty               = _mm256_mul_pd(fscal,dy20);
 815             tz               = _mm256_mul_pd(fscal,dz20);
 816
 817             /* Update vectorial force */
 818             fix2             = _mm256_add_pd(fix2,tx);
 819             fiy2             = _mm256_add_pd(fiy2,ty);
 820             fiz2             = _mm256_add_pd(fiz2,tz);
 821
 822             fjx0             = _mm256_add_pd(fjx0,tx);
 823             fjy0             = _mm256_add_pd(fjy0,ty);
 824             fjz0             = _mm256_add_pd(fjz0,tz);
 825
 826             }
 827
 828             /**************************
 829              * CALCULATE INTERACTIONS *
 830              **************************/
 831
 832             if (gmx_mm256_any_lt(rsq30,rcutoff2))
 833             {
 834
 835             r30              = _mm256_mul_pd(rsq30,rinv30);
 836             r30              = _mm256_andnot_pd(dummy_mask,r30);
 837
 838             /* Compute parameters for interactions between i and j atoms */
 839             qq30             = _mm256_mul_pd(iq3,jq0);
 840
 841             /* EWALD ELECTROSTATICS */
 842
 843             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 844             ewrt             = _mm256_mul_pd(r30,ewtabscale);
 845             ewitab           = _mm256_cvttpd_epi32(ewrt);
 846             eweps            = _mm256_sub_pd(ewrt,_mm256_round_pd(ewrt, _MM_FROUND_FLOOR));
 847             ewitab           = _mm_slli_epi32(ewitab,2);
 848             ewtabF           = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,0) );
 849             ewtabD           = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,1) );
 850             ewtabV           = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,2) );
 851             ewtabFn          = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,3) );
 852             GMX_MM256_FULLTRANSPOSE4_PD(ewtabF,ewtabD,ewtabV,ewtabFn);
 853             felec            = _mm256_add_pd(ewtabF,_mm256_mul_pd(eweps,ewtabD));
 854             velec            = _mm256_sub_pd(ewtabV,_mm256_mul_pd(_mm256_mul_pd(ewtabhalfspace,eweps),_mm256_add_pd(ewtabF,felec)));
 855             velec            = _mm256_mul_pd(qq30,_mm256_sub_pd(rinv30,velec));
 856             felec            = _mm256_mul_pd(_mm256_mul_pd(qq30,rinv30),_mm256_sub_pd(rinvsq30,felec));
 857
 858             d                = _mm256_sub_pd(r30,rswitch);
 859             d                = _mm256_max_pd(d,_mm256_setzero_pd());
 860             d2               = _mm256_mul_pd(d,d);
 861             sw               = _mm256_add_pd(one,_mm256_mul_pd(d2,_mm256_mul_pd(d,_mm256_add_pd(swV3,_mm256_mul_pd(d,_mm256_add_pd(swV4,_mm256_mul_pd(d,swV5)))))));
 862
 863             dsw              = _mm256_mul_pd(d2,_mm256_add_pd(swF2,_mm256_mul_pd(d,_mm256_add_pd(swF3,_mm256_mul_pd(d,swF4)))));
 864
 865             /* Evaluate switch function */
 866             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 867             felec            = _mm256_sub_pd( _mm256_mul_pd(felec,sw) , _mm256_mul_pd(rinv30,_mm256_mul_pd(velec,dsw)) );
 868             velec            = _mm256_mul_pd(velec,sw);
 869             cutoff_mask      = _mm256_cmp_pd(rsq30,rcutoff2,_CMP_LT_OQ);
 870
 871             /* Update potential sum for this i atom from the interaction with this j atom. */
 872             velec            = _mm256_and_pd(velec,cutoff_mask);
 873             velec            = _mm256_andnot_pd(dummy_mask,velec);
 874             velecsum         = _mm256_add_pd(velecsum,velec);
 875
 876             fscal            = felec;
 877
 878             fscal            = _mm256_and_pd(fscal,cutoff_mask);
 879
 880             fscal            = _mm256_andnot_pd(dummy_mask,fscal);
 881
 882             /* Calculate temporary vectorial force */
 883             tx               = _mm256_mul_pd(fscal,dx30);
 884             ty               = _mm256_mul_pd(fscal,dy30);
 885             tz               = _mm256_mul_pd(fscal,dz30);
 886
 887             /* Update vectorial force */
 888             fix3             = _mm256_add_pd(fix3,tx);
 889             fiy3             = _mm256_add_pd(fiy3,ty);
 890             fiz3             = _mm256_add_pd(fiz3,tz);
 891
 892             fjx0             = _mm256_add_pd(fjx0,tx);
 893             fjy0             = _mm256_add_pd(fjy0,ty);
 894             fjz0             = _mm256_add_pd(fjz0,tz);
 895
 896             }
 897
 898             fjptrA             = (jnrlistA>=0) ? f+j_coord_offsetA : scratch;
 899             fjptrB             = (jnrlistB>=0) ? f+j_coord_offsetB : scratch;
 900             fjptrC             = (jnrlistC>=0) ? f+j_coord_offsetC : scratch;
 901             fjptrD             = (jnrlistD>=0) ? f+j_coord_offsetD : scratch;
 902
 903             gmx_mm256_decrement_1rvec_4ptr_swizzle_pd(fjptrA,fjptrB,fjptrC,fjptrD,fjx0,fjy0,fjz0);
 904
 905             /* Inner loop uses 261 flops */
 906         }
 907
 908         /* End of innermost loop */
 909
 910         gmx_mm256_update_iforce_4atom_swizzle_pd(fix0,fiy0,fiz0,fix1,fiy1,fiz1,fix2,fiy2,fiz2,fix3,fiy3,fiz3,
 911                                                  f+i_coord_offset,fshift+i_shift_offset);
 912
 913         ggid                        = gid[iidx];
 914         /* Update potential energies */
 915         gmx_mm256_update_1pot_pd(velecsum,kernel_data->energygrp_elec+ggid);
 916         gmx_mm256_update_1pot_pd(vvdwsum,kernel_data->energygrp_vdw+ggid);
 917
 918         /* Increment number of inner iterations */
 919         inneriter                  += j_index_end - j_index_start;
 920
 921         /* Outer loop uses 26 flops */
 922     }
 923
 924     /* Increment number of outer iterations */
 925     outeriter        += nri;
 926
 927     /* Update outer/inner flops */
 928
 929     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W4_VF,outeriter*26 + inneriter*261);
 930 }
 931 /*
 932  * Gromacs nonbonded kernel:   nb_kernel_ElecEwSw_VdwLJSw_GeomW4P1_F_avx_256_double
 933  * Electrostatics interaction: Ewald
 934  * VdW interaction:            LennardJones
 935  * Geometry:                   Water4-Particle
 936  * Calculate force/pot:        Force
 937  */
 938 void
 939 nb_kernel_ElecEwSw_VdwLJSw_GeomW4P1_F_avx_256_double
 940                     (t_nblist                    * gmx_restrict       nlist,
 941                      rvec                        * gmx_restrict          xx,
 942                      rvec                        * gmx_restrict          ff,
 943                      t_forcerec                  * gmx_restrict          fr,
 944                      t_mdatoms                   * gmx_restrict     mdatoms,
 945                      nb_kernel_data_t gmx_unused * gmx_restrict kernel_data,
 946                      t_nrnb                      * gmx_restrict        nrnb)
 947 {
 948     /* Suffixes 0,1,2,3 refer to particle indices for waters in the inner or outer loop, or
 949      * just 0 for non-waters.
 950      * Suffixes A,B,C,D refer to j loop unrolling done with AVX, e.g. for the four different
 951      * jnr indices corresponding to data put in the four positions in the SIMD register.
 952      */
 953     int              i_shift_offset,i_coord_offset,outeriter,inneriter;
 954     int              j_index_start,j_index_end,jidx,nri,inr,ggid,iidx;
 955     int              jnrA,jnrB,jnrC,jnrD;
 956     int              jnrlistA,jnrlistB,jnrlistC,jnrlistD;
 957     int              jnrlistE,jnrlistF,jnrlistG,jnrlistH;
 958     int              j_coord_offsetA,j_coord_offsetB,j_coord_offsetC,j_coord_offsetD;
 959     int              *iinr,*jindex,*jjnr,*shiftidx,*gid;
 960     real             rcutoff_scalar;
 961     real             *shiftvec,*fshift,*x,*f;
 962     real             *fjptrA,*fjptrB,*fjptrC,*fjptrD;
 963     real             scratch[4*DIM];
 964     __m256d          tx,ty,tz,fscal,rcutoff,rcutoff2,jidxall;
 965     real *           vdwioffsetptr0;
 966     __m256d          ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
 967     real *           vdwioffsetptr1;
 968     __m256d          ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
 969     real *           vdwioffsetptr2;
 970     __m256d          ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
 971     real *           vdwioffsetptr3;
 972     __m256d          ix3,iy3,iz3,fix3,fiy3,fiz3,iq3,isai3;
 973     int              vdwjidx0A,vdwjidx0B,vdwjidx0C,vdwjidx0D;
 974     __m256d          jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
 975     __m256d          dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00;
 976     __m256d          dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10;
 977     __m256d          dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20;
 978     __m256d          dx30,dy30,dz30,rsq30,rinv30,rinvsq30,r30,qq30,c6_30,c12_30;
 979     __m256d          velec,felec,velecsum,facel,crf,krf,krf2;
 980     real             *charge;
 981     int              nvdwtype;
 982     __m256d          rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,sh_vdw_invrcut6;
 983     int              *vdwtype;
 984     real             *vdwparam;
 985     __m256d          one_sixth   = _mm256_set1_pd(1.0/6.0);
 986     __m256d          one_twelfth = _mm256_set1_pd(1.0/12.0);
 987     __m128i          ewitab;
 988     __m256d          ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace,ewtabF,ewtabFn,ewtabD,ewtabV;
 989     __m256d          beta,beta2,beta3,zeta2,pmecorrF,pmecorrV,rinv3;
 990     real             *ewtab;
 991     __m256d          rswitch,swV3,swV4,swV5,swF2,swF3,swF4,d,d2,sw,dsw;
 992     real             rswitch_scalar,d_scalar;
 993     __m256d          dummy_mask,cutoff_mask;
 994     __m128           tmpmask0,tmpmask1;
 995     __m256d          signbit = _mm256_castsi256_pd( _mm256_set1_epi32(0x80000000) );
 996     __m256d          one     = _mm256_set1_pd(1.0);
 997     __m256d          two     = _mm256_set1_pd(2.0);
 998     x                = xx[0];
 999     f                = ff[0];
1000
1001     nri              = nlist->nri;
1002     iinr             = nlist->iinr;
1003     jindex           = nlist->jindex;
1004     jjnr             = nlist->jjnr;
1005     shiftidx         = nlist->shift;
1006     gid              = nlist->gid;
1007     shiftvec         = fr->shift_vec[0];
1008     fshift           = fr->fshift[0];
1009     facel            = _mm256_set1_pd(fr->epsfac);
1010     charge           = mdatoms->chargeA;
1011     nvdwtype         = fr->ntype;
1012     vdwparam         = fr->nbfp;
1013     vdwtype          = mdatoms->typeA;
1014
1015     sh_ewald         = _mm256_set1_pd(fr->ic->sh_ewald);
1016     beta             = _mm256_set1_pd(fr->ic->ewaldcoeff_q);
1017     beta2            = _mm256_mul_pd(beta,beta);
1018     beta3            = _mm256_mul_pd(beta,beta2);
1019
1020     ewtab            = fr->ic->tabq_coul_FDV0;
1021     ewtabscale       = _mm256_set1_pd(fr->ic->tabq_scale);
1022     ewtabhalfspace   = _mm256_set1_pd(0.5/fr->ic->tabq_scale);
1023
1024     /* Setup water-specific parameters */
1025     inr              = nlist->iinr[0];
1026     iq1              = _mm256_mul_pd(facel,_mm256_set1_pd(charge[inr+1]));
1027     iq2              = _mm256_mul_pd(facel,_mm256_set1_pd(charge[inr+2]));
1028     iq3              = _mm256_mul_pd(facel,_mm256_set1_pd(charge[inr+3]));
1029     vdwioffsetptr0   = vdwparam+2*nvdwtype*vdwtype[inr+0];
1030
1031     /* When we use explicit cutoffs the value must be identical for elec and VdW, so use elec as an arbitrary choice */
1032     rcutoff_scalar   = fr->rcoulomb;
1033     rcutoff          = _mm256_set1_pd(rcutoff_scalar);
1034     rcutoff2         = _mm256_mul_pd(rcutoff,rcutoff);
1035
1036     rswitch_scalar   = fr->rcoulomb_switch;
1037     rswitch          = _mm256_set1_pd(rswitch_scalar);
1038     /* Setup switch parameters */
1039     d_scalar         = rcutoff_scalar-rswitch_scalar;
1040     d                = _mm256_set1_pd(d_scalar);
1041     swV3             = _mm256_set1_pd(-10.0/(d_scalar*d_scalar*d_scalar));
1042     swV4             = _mm256_set1_pd( 15.0/(d_scalar*d_scalar*d_scalar*d_scalar));
1043     swV5             = _mm256_set1_pd( -6.0/(d_scalar*d_scalar*d_scalar*d_scalar*d_scalar));
1044     swF2             = _mm256_set1_pd(-30.0/(d_scalar*d_scalar*d_scalar));
1045     swF3             = _mm256_set1_pd( 60.0/(d_scalar*d_scalar*d_scalar*d_scalar));
1046     swF4             = _mm256_set1_pd(-30.0/(d_scalar*d_scalar*d_scalar*d_scalar*d_scalar));
1047
1048     /* Avoid stupid compiler warnings */
1049     jnrA = jnrB = jnrC = jnrD = 0;
1050     j_coord_offsetA = 0;
1051     j_coord_offsetB = 0;
1052     j_coord_offsetC = 0;
1053     j_coord_offsetD = 0;
1054
1055     outeriter        = 0;
1056     inneriter        = 0;
1057
1058     for(iidx=0;iidx<4*DIM;iidx++)
1059     {
1060         scratch[iidx] = 0.0;
1061     }
1062
1063     /* Start outer loop over neighborlists */
1064     for(iidx=0; iidx<nri; iidx++)
1065     {
1066         /* Load shift vector for this list */
1067         i_shift_offset   = DIM*shiftidx[iidx];
1068
1069         /* Load limits for loop over neighbors */
1070         j_index_start    = jindex[iidx];
1071         j_index_end      = jindex[iidx+1];
1072
1073         /* Get outer coordinate index */
1074         inr              = iinr[iidx];
1075         i_coord_offset   = DIM*inr;
1076
1077         /* Load i particle coords and add shift vector */
1078         gmx_mm256_load_shift_and_4rvec_broadcast_pd(shiftvec+i_shift_offset,x+i_coord_offset,
1079                                                     &ix0,&iy0,&iz0,&ix1,&iy1,&iz1,&ix2,&iy2,&iz2,&ix3,&iy3,&iz3);
1080
1081         fix0             = _mm256_setzero_pd();
1082         fiy0             = _mm256_setzero_pd();
1083         fiz0             = _mm256_setzero_pd();
1084         fix1             = _mm256_setzero_pd();
1085         fiy1             = _mm256_setzero_pd();
1086         fiz1             = _mm256_setzero_pd();
1087         fix2             = _mm256_setzero_pd();
1088         fiy2             = _mm256_setzero_pd();
1089         fiz2             = _mm256_setzero_pd();
1090         fix3             = _mm256_setzero_pd();
1091         fiy3             = _mm256_setzero_pd();
1092         fiz3             = _mm256_setzero_pd();
1093
1094         /* Start inner kernel loop */
1095         for(jidx=j_index_start; jidx<j_index_end && jjnr[jidx+3]>=0; jidx+=4)
1096         {
1097
1098             /* Get j neighbor index, and coordinate index */
1099             jnrA             = jjnr[jidx];
1100             jnrB             = jjnr[jidx+1];
1101             jnrC             = jjnr[jidx+2];
1102             jnrD             = jjnr[jidx+3];
1103             j_coord_offsetA  = DIM*jnrA;
1104             j_coord_offsetB  = DIM*jnrB;
1105             j_coord_offsetC  = DIM*jnrC;
1106             j_coord_offsetD  = DIM*jnrD;
1107
1108             /* load j atom coordinates */
1109             gmx_mm256_load_1rvec_4ptr_swizzle_pd(x+j_coord_offsetA,x+j_coord_offsetB,
1110                                                  x+j_coord_offsetC,x+j_coord_offsetD,
1111                                                  &jx0,&jy0,&jz0);
1112
1113             /* Calculate displacement vector */
1114             dx00             = _mm256_sub_pd(ix0,jx0);
1115             dy00             = _mm256_sub_pd(iy0,jy0);
1116             dz00             = _mm256_sub_pd(iz0,jz0);
1117             dx10             = _mm256_sub_pd(ix1,jx0);
1118             dy10             = _mm256_sub_pd(iy1,jy0);
1119             dz10             = _mm256_sub_pd(iz1,jz0);
1120             dx20             = _mm256_sub_pd(ix2,jx0);
1121             dy20             = _mm256_sub_pd(iy2,jy0);
1122             dz20             = _mm256_sub_pd(iz2,jz0);
1123             dx30             = _mm256_sub_pd(ix3,jx0);
1124             dy30             = _mm256_sub_pd(iy3,jy0);
1125             dz30             = _mm256_sub_pd(iz3,jz0);
1126
1127             /* Calculate squared distance and things based on it */
1128             rsq00            = gmx_mm256_calc_rsq_pd(dx00,dy00,dz00);
1129             rsq10            = gmx_mm256_calc_rsq_pd(dx10,dy10,dz10);
1130             rsq20            = gmx_mm256_calc_rsq_pd(dx20,dy20,dz20);
1131             rsq30            = gmx_mm256_calc_rsq_pd(dx30,dy30,dz30);
1132
1133             rinv00           = gmx_mm256_invsqrt_pd(rsq00);
1134             rinv10           = gmx_mm256_invsqrt_pd(rsq10);
1135             rinv20           = gmx_mm256_invsqrt_pd(rsq20);
1136             rinv30           = gmx_mm256_invsqrt_pd(rsq30);
1137
1138             rinvsq00         = _mm256_mul_pd(rinv00,rinv00);
1139             rinvsq10         = _mm256_mul_pd(rinv10,rinv10);
1140             rinvsq20         = _mm256_mul_pd(rinv20,rinv20);
1141             rinvsq30         = _mm256_mul_pd(rinv30,rinv30);
1142
1143             /* Load parameters for j particles */
1144             jq0              = gmx_mm256_load_4real_swizzle_pd(charge+jnrA+0,charge+jnrB+0,
1145                                                                  charge+jnrC+0,charge+jnrD+0);
1146             vdwjidx0A        = 2*vdwtype[jnrA+0];
1147             vdwjidx0B        = 2*vdwtype[jnrB+0];
1148             vdwjidx0C        = 2*vdwtype[jnrC+0];
1149             vdwjidx0D        = 2*vdwtype[jnrD+0];
1150
1151             fjx0             = _mm256_setzero_pd();
1152             fjy0             = _mm256_setzero_pd();
1153             fjz0             = _mm256_setzero_pd();
1154
1155             /**************************
1156              * CALCULATE INTERACTIONS *
1157              **************************/
1158
1159             if (gmx_mm256_any_lt(rsq00,rcutoff2))
1160             {
1161
1162             r00              = _mm256_mul_pd(rsq00,rinv00);
1163
1164             /* Compute parameters for interactions between i and j atoms */
1165             gmx_mm256_load_4pair_swizzle_pd(vdwioffsetptr0+vdwjidx0A,
1166                                             vdwioffsetptr0+vdwjidx0B,
1167                                             vdwioffsetptr0+vdwjidx0C,
1168                                             vdwioffsetptr0+vdwjidx0D,
1169                                             &c6_00,&c12_00);
1170
1171             /* LENNARD-JONES DISPERSION/REPULSION */
1172
1173             rinvsix          = _mm256_mul_pd(_mm256_mul_pd(rinvsq00,rinvsq00),rinvsq00);
1174             vvdw6            = _mm256_mul_pd(c6_00,rinvsix);
1175             vvdw12           = _mm256_mul_pd(c12_00,_mm256_mul_pd(rinvsix,rinvsix));
1176             vvdw             = _mm256_sub_pd( _mm256_mul_pd(vvdw12,one_twelfth) , _mm256_mul_pd(vvdw6,one_sixth) );
1177             fvdw             = _mm256_mul_pd(_mm256_sub_pd(vvdw12,vvdw6),rinvsq00);
1178
1179             d                = _mm256_sub_pd(r00,rswitch);
1180             d                = _mm256_max_pd(d,_mm256_setzero_pd());
1181             d2               = _mm256_mul_pd(d,d);
1182             sw               = _mm256_add_pd(one,_mm256_mul_pd(d2,_mm256_mul_pd(d,_mm256_add_pd(swV3,_mm256_mul_pd(d,_mm256_add_pd(swV4,_mm256_mul_pd(d,swV5)))))));
1183
1184             dsw              = _mm256_mul_pd(d2,_mm256_add_pd(swF2,_mm256_mul_pd(d,_mm256_add_pd(swF3,_mm256_mul_pd(d,swF4)))));
1185
1186             /* Evaluate switch function */
1187             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
1188             fvdw             = _mm256_sub_pd( _mm256_mul_pd(fvdw,sw) , _mm256_mul_pd(rinv00,_mm256_mul_pd(vvdw,dsw)) );
1189             cutoff_mask      = _mm256_cmp_pd(rsq00,rcutoff2,_CMP_LT_OQ);
1190
1191             fscal            = fvdw;
1192
1193             fscal            = _mm256_and_pd(fscal,cutoff_mask);
1194
1195             /* Calculate temporary vectorial force */
1196             tx               = _mm256_mul_pd(fscal,dx00);
1197             ty               = _mm256_mul_pd(fscal,dy00);
1198             tz               = _mm256_mul_pd(fscal,dz00);
1199
1200             /* Update vectorial force */
1201             fix0             = _mm256_add_pd(fix0,tx);
1202             fiy0             = _mm256_add_pd(fiy0,ty);
1203             fiz0             = _mm256_add_pd(fiz0,tz);
1204
1205             fjx0             = _mm256_add_pd(fjx0,tx);
1206             fjy0             = _mm256_add_pd(fjy0,ty);
1207             fjz0             = _mm256_add_pd(fjz0,tz);
1208
1209             }
1210
1211             /**************************
1212              * CALCULATE INTERACTIONS *
1213              **************************/
1214
1215             if (gmx_mm256_any_lt(rsq10,rcutoff2))
1216             {
1217
1218             r10              = _mm256_mul_pd(rsq10,rinv10);
1219
1220             /* Compute parameters for interactions between i and j atoms */
1221             qq10             = _mm256_mul_pd(iq1,jq0);
1222
1223             /* EWALD ELECTROSTATICS */
1224
1225             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1226             ewrt             = _mm256_mul_pd(r10,ewtabscale);
1227             ewitab           = _mm256_cvttpd_epi32(ewrt);
1228             eweps            = _mm256_sub_pd(ewrt,_mm256_round_pd(ewrt, _MM_FROUND_FLOOR));
1229             ewitab           = _mm_slli_epi32(ewitab,2);
1230             ewtabF           = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,0) );
1231             ewtabD           = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,1) );
1232             ewtabV           = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,2) );
1233             ewtabFn          = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,3) );
1234             GMX_MM256_FULLTRANSPOSE4_PD(ewtabF,ewtabD,ewtabV,ewtabFn);
1235             felec            = _mm256_add_pd(ewtabF,_mm256_mul_pd(eweps,ewtabD));
1236             velec            = _mm256_sub_pd(ewtabV,_mm256_mul_pd(_mm256_mul_pd(ewtabhalfspace,eweps),_mm256_add_pd(ewtabF,felec)));
1237             velec            = _mm256_mul_pd(qq10,_mm256_sub_pd(rinv10,velec));
1238             felec            = _mm256_mul_pd(_mm256_mul_pd(qq10,rinv10),_mm256_sub_pd(rinvsq10,felec));
1239
1240             d                = _mm256_sub_pd(r10,rswitch);
1241             d                = _mm256_max_pd(d,_mm256_setzero_pd());
1242             d2               = _mm256_mul_pd(d,d);
1243             sw               = _mm256_add_pd(one,_mm256_mul_pd(d2,_mm256_mul_pd(d,_mm256_add_pd(swV3,_mm256_mul_pd(d,_mm256_add_pd(swV4,_mm256_mul_pd(d,swV5)))))));
1244
1245             dsw              = _mm256_mul_pd(d2,_mm256_add_pd(swF2,_mm256_mul_pd(d,_mm256_add_pd(swF3,_mm256_mul_pd(d,swF4)))));
1246
1247             /* Evaluate switch function */
1248             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
1249             felec            = _mm256_sub_pd( _mm256_mul_pd(felec,sw) , _mm256_mul_pd(rinv10,_mm256_mul_pd(velec,dsw)) );
1250             cutoff_mask      = _mm256_cmp_pd(rsq10,rcutoff2,_CMP_LT_OQ);
1251
1252             fscal            = felec;
1253
1254             fscal            = _mm256_and_pd(fscal,cutoff_mask);
1255
1256             /* Calculate temporary vectorial force */
1257             tx               = _mm256_mul_pd(fscal,dx10);
1258             ty               = _mm256_mul_pd(fscal,dy10);
1259             tz               = _mm256_mul_pd(fscal,dz10);
1260
1261             /* Update vectorial force */
1262             fix1             = _mm256_add_pd(fix1,tx);
1263             fiy1             = _mm256_add_pd(fiy1,ty);
1264             fiz1             = _mm256_add_pd(fiz1,tz);
1265
1266             fjx0             = _mm256_add_pd(fjx0,tx);
1267             fjy0             = _mm256_add_pd(fjy0,ty);
1268             fjz0             = _mm256_add_pd(fjz0,tz);
1269
1270             }
1271
1272             /**************************
1273              * CALCULATE INTERACTIONS *
1274              **************************/
1275
1276             if (gmx_mm256_any_lt(rsq20,rcutoff2))
1277             {
1278
1279             r20              = _mm256_mul_pd(rsq20,rinv20);
1280
1281             /* Compute parameters for interactions between i and j atoms */
1282             qq20             = _mm256_mul_pd(iq2,jq0);
1283
1284             /* EWALD ELECTROSTATICS */
1285
1286             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1287             ewrt             = _mm256_mul_pd(r20,ewtabscale);
1288             ewitab           = _mm256_cvttpd_epi32(ewrt);
1289             eweps            = _mm256_sub_pd(ewrt,_mm256_round_pd(ewrt, _MM_FROUND_FLOOR));
1290             ewitab           = _mm_slli_epi32(ewitab,2);
1291             ewtabF           = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,0) );
1292             ewtabD           = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,1) );
1293             ewtabV           = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,2) );
1294             ewtabFn          = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,3) );
1295             GMX_MM256_FULLTRANSPOSE4_PD(ewtabF,ewtabD,ewtabV,ewtabFn);
1296             felec            = _mm256_add_pd(ewtabF,_mm256_mul_pd(eweps,ewtabD));
1297             velec            = _mm256_sub_pd(ewtabV,_mm256_mul_pd(_mm256_mul_pd(ewtabhalfspace,eweps),_mm256_add_pd(ewtabF,felec)));
1298             velec            = _mm256_mul_pd(qq20,_mm256_sub_pd(rinv20,velec));
1299             felec            = _mm256_mul_pd(_mm256_mul_pd(qq20,rinv20),_mm256_sub_pd(rinvsq20,felec));
1300
1301             d                = _mm256_sub_pd(r20,rswitch);
1302             d                = _mm256_max_pd(d,_mm256_setzero_pd());
1303             d2               = _mm256_mul_pd(d,d);
1304             sw               = _mm256_add_pd(one,_mm256_mul_pd(d2,_mm256_mul_pd(d,_mm256_add_pd(swV3,_mm256_mul_pd(d,_mm256_add_pd(swV4,_mm256_mul_pd(d,swV5)))))));
1305
1306             dsw              = _mm256_mul_pd(d2,_mm256_add_pd(swF2,_mm256_mul_pd(d,_mm256_add_pd(swF3,_mm256_mul_pd(d,swF4)))));
1307
1308             /* Evaluate switch function */
1309             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
1310             felec            = _mm256_sub_pd( _mm256_mul_pd(felec,sw) , _mm256_mul_pd(rinv20,_mm256_mul_pd(velec,dsw)) );
1311             cutoff_mask      = _mm256_cmp_pd(rsq20,rcutoff2,_CMP_LT_OQ);
1312
1313             fscal            = felec;
1314
1315             fscal            = _mm256_and_pd(fscal,cutoff_mask);
1316
1317             /* Calculate temporary vectorial force */
1318             tx               = _mm256_mul_pd(fscal,dx20);
1319             ty               = _mm256_mul_pd(fscal,dy20);
1320             tz               = _mm256_mul_pd(fscal,dz20);
1321
1322             /* Update vectorial force */
1323             fix2             = _mm256_add_pd(fix2,tx);
1324             fiy2             = _mm256_add_pd(fiy2,ty);
1325             fiz2             = _mm256_add_pd(fiz2,tz);
1326
1327             fjx0             = _mm256_add_pd(fjx0,tx);
1328             fjy0             = _mm256_add_pd(fjy0,ty);
1329             fjz0             = _mm256_add_pd(fjz0,tz);
1330
1331             }
1332
1333             /**************************
1334              * CALCULATE INTERACTIONS *
1335              **************************/
1336
1337             if (gmx_mm256_any_lt(rsq30,rcutoff2))
1338             {
1339
1340             r30              = _mm256_mul_pd(rsq30,rinv30);
1341
1342             /* Compute parameters for interactions between i and j atoms */
1343             qq30             = _mm256_mul_pd(iq3,jq0);
1344
1345             /* EWALD ELECTROSTATICS */
1346
1347             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1348             ewrt             = _mm256_mul_pd(r30,ewtabscale);
1349             ewitab           = _mm256_cvttpd_epi32(ewrt);
1350             eweps            = _mm256_sub_pd(ewrt,_mm256_round_pd(ewrt, _MM_FROUND_FLOOR));
1351             ewitab           = _mm_slli_epi32(ewitab,2);
1352             ewtabF           = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,0) );
1353             ewtabD           = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,1) );
1354             ewtabV           = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,2) );
1355             ewtabFn          = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,3) );
1356             GMX_MM256_FULLTRANSPOSE4_PD(ewtabF,ewtabD,ewtabV,ewtabFn);
1357             felec            = _mm256_add_pd(ewtabF,_mm256_mul_pd(eweps,ewtabD));
1358             velec            = _mm256_sub_pd(ewtabV,_mm256_mul_pd(_mm256_mul_pd(ewtabhalfspace,eweps),_mm256_add_pd(ewtabF,felec)));
1359             velec            = _mm256_mul_pd(qq30,_mm256_sub_pd(rinv30,velec));
1360             felec            = _mm256_mul_pd(_mm256_mul_pd(qq30,rinv30),_mm256_sub_pd(rinvsq30,felec));
1361
1362             d                = _mm256_sub_pd(r30,rswitch);
1363             d                = _mm256_max_pd(d,_mm256_setzero_pd());
1364             d2               = _mm256_mul_pd(d,d);
1365             sw               = _mm256_add_pd(one,_mm256_mul_pd(d2,_mm256_mul_pd(d,_mm256_add_pd(swV3,_mm256_mul_pd(d,_mm256_add_pd(swV4,_mm256_mul_pd(d,swV5)))))));
1366
1367             dsw              = _mm256_mul_pd(d2,_mm256_add_pd(swF2,_mm256_mul_pd(d,_mm256_add_pd(swF3,_mm256_mul_pd(d,swF4)))));
1368
1369             /* Evaluate switch function */
1370             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
1371             felec            = _mm256_sub_pd( _mm256_mul_pd(felec,sw) , _mm256_mul_pd(rinv30,_mm256_mul_pd(velec,dsw)) );
1372             cutoff_mask      = _mm256_cmp_pd(rsq30,rcutoff2,_CMP_LT_OQ);
1373
1374             fscal            = felec;
1375
1376             fscal            = _mm256_and_pd(fscal,cutoff_mask);
1377
1378             /* Calculate temporary vectorial force */
1379             tx               = _mm256_mul_pd(fscal,dx30);
1380             ty               = _mm256_mul_pd(fscal,dy30);
1381             tz               = _mm256_mul_pd(fscal,dz30);
1382
1383             /* Update vectorial force */
1384             fix3             = _mm256_add_pd(fix3,tx);
1385             fiy3             = _mm256_add_pd(fiy3,ty);
1386             fiz3             = _mm256_add_pd(fiz3,tz);
1387
1388             fjx0             = _mm256_add_pd(fjx0,tx);
1389             fjy0             = _mm256_add_pd(fjy0,ty);
1390             fjz0             = _mm256_add_pd(fjz0,tz);
1391
1392             }
1393
1394             fjptrA             = f+j_coord_offsetA;
1395             fjptrB             = f+j_coord_offsetB;
1396             fjptrC             = f+j_coord_offsetC;
1397             fjptrD             = f+j_coord_offsetD;
1398
1399             gmx_mm256_decrement_1rvec_4ptr_swizzle_pd(fjptrA,fjptrB,fjptrC,fjptrD,fjx0,fjy0,fjz0);
1400
1401             /* Inner loop uses 245 flops */
1402         }
1403
1404         if(jidx<j_index_end)
1405         {
1406
1407             /* Get j neighbor index, and coordinate index */
1408             jnrlistA         = jjnr[jidx];
1409             jnrlistB         = jjnr[jidx+1];
1410             jnrlistC         = jjnr[jidx+2];
1411             jnrlistD         = jjnr[jidx+3];
1412             /* Sign of each element will be negative for non-real atoms.
1413              * This mask will be 0xFFFFFFFF for dummy entries and 0x0 for real ones,
1414              * so use it as val = _mm_andnot_pd(mask,val) to clear dummy entries.
1415              */
1416             tmpmask0 = gmx_mm_castsi128_ps(_mm_cmplt_epi32(_mm_loadu_si128((const __m128i *)(jjnr+jidx)),_mm_setzero_si128()));
1417
1418             tmpmask1 = _mm_permute_ps(tmpmask0,_GMX_MM_PERMUTE(3,3,2,2));
1419             tmpmask0 = _mm_permute_ps(tmpmask0,_GMX_MM_PERMUTE(1,1,0,0));
1420             dummy_mask = _mm256_castps_pd(gmx_mm256_set_m128(tmpmask1,tmpmask0));
1421
1422             jnrA       = (jnrlistA>=0) ? jnrlistA : 0;
1423             jnrB       = (jnrlistB>=0) ? jnrlistB : 0;
1424             jnrC       = (jnrlistC>=0) ? jnrlistC : 0;
1425             jnrD       = (jnrlistD>=0) ? jnrlistD : 0;
1426             j_coord_offsetA  = DIM*jnrA;
1427             j_coord_offsetB  = DIM*jnrB;
1428             j_coord_offsetC  = DIM*jnrC;
1429             j_coord_offsetD  = DIM*jnrD;
1430
1431             /* load j atom coordinates */
1432             gmx_mm256_load_1rvec_4ptr_swizzle_pd(x+j_coord_offsetA,x+j_coord_offsetB,
1433                                                  x+j_coord_offsetC,x+j_coord_offsetD,
1434                                                  &jx0,&jy0,&jz0);
1435
1436             /* Calculate displacement vector */
1437             dx00             = _mm256_sub_pd(ix0,jx0);
1438             dy00             = _mm256_sub_pd(iy0,jy0);
1439             dz00             = _mm256_sub_pd(iz0,jz0);
1440             dx10             = _mm256_sub_pd(ix1,jx0);
1441             dy10             = _mm256_sub_pd(iy1,jy0);
1442             dz10             = _mm256_sub_pd(iz1,jz0);
1443             dx20             = _mm256_sub_pd(ix2,jx0);
1444             dy20             = _mm256_sub_pd(iy2,jy0);
1445             dz20             = _mm256_sub_pd(iz2,jz0);
1446             dx30             = _mm256_sub_pd(ix3,jx0);
1447             dy30             = _mm256_sub_pd(iy3,jy0);
1448             dz30             = _mm256_sub_pd(iz3,jz0);
1449
1450             /* Calculate squared distance and things based on it */
1451             rsq00            = gmx_mm256_calc_rsq_pd(dx00,dy00,dz00);
1452             rsq10            = gmx_mm256_calc_rsq_pd(dx10,dy10,dz10);
1453             rsq20            = gmx_mm256_calc_rsq_pd(dx20,dy20,dz20);
1454             rsq30            = gmx_mm256_calc_rsq_pd(dx30,dy30,dz30);
1455
1456             rinv00           = gmx_mm256_invsqrt_pd(rsq00);
1457             rinv10           = gmx_mm256_invsqrt_pd(rsq10);
1458             rinv20           = gmx_mm256_invsqrt_pd(rsq20);
1459             rinv30           = gmx_mm256_invsqrt_pd(rsq30);
1460
1461             rinvsq00         = _mm256_mul_pd(rinv00,rinv00);
1462             rinvsq10         = _mm256_mul_pd(rinv10,rinv10);
1463             rinvsq20         = _mm256_mul_pd(rinv20,rinv20);
1464             rinvsq30         = _mm256_mul_pd(rinv30,rinv30);
1465
1466             /* Load parameters for j particles */
1467             jq0              = gmx_mm256_load_4real_swizzle_pd(charge+jnrA+0,charge+jnrB+0,
1468                                                                  charge+jnrC+0,charge+jnrD+0);
1469             vdwjidx0A        = 2*vdwtype[jnrA+0];
1470             vdwjidx0B        = 2*vdwtype[jnrB+0];
1471             vdwjidx0C        = 2*vdwtype[jnrC+0];
1472             vdwjidx0D        = 2*vdwtype[jnrD+0];
1473
1474             fjx0             = _mm256_setzero_pd();
1475             fjy0             = _mm256_setzero_pd();
1476             fjz0             = _mm256_setzero_pd();
1477
1478             /**************************
1479              * CALCULATE INTERACTIONS *
1480              **************************/
1481
1482             if (gmx_mm256_any_lt(rsq00,rcutoff2))
1483             {
1484
1485             r00              = _mm256_mul_pd(rsq00,rinv00);
1486             r00              = _mm256_andnot_pd(dummy_mask,r00);
1487
1488             /* Compute parameters for interactions between i and j atoms */
1489             gmx_mm256_load_4pair_swizzle_pd(vdwioffsetptr0+vdwjidx0A,
1490                                             vdwioffsetptr0+vdwjidx0B,
1491                                             vdwioffsetptr0+vdwjidx0C,
1492                                             vdwioffsetptr0+vdwjidx0D,
1493                                             &c6_00,&c12_00);
1494
1495             /* LENNARD-JONES DISPERSION/REPULSION */
1496
1497             rinvsix          = _mm256_mul_pd(_mm256_mul_pd(rinvsq00,rinvsq00),rinvsq00);
1498             vvdw6            = _mm256_mul_pd(c6_00,rinvsix);
1499             vvdw12           = _mm256_mul_pd(c12_00,_mm256_mul_pd(rinvsix,rinvsix));
1500             vvdw             = _mm256_sub_pd( _mm256_mul_pd(vvdw12,one_twelfth) , _mm256_mul_pd(vvdw6,one_sixth) );
1501             fvdw             = _mm256_mul_pd(_mm256_sub_pd(vvdw12,vvdw6),rinvsq00);
1502
1503             d                = _mm256_sub_pd(r00,rswitch);
1504             d                = _mm256_max_pd(d,_mm256_setzero_pd());
1505             d2               = _mm256_mul_pd(d,d);
1506             sw               = _mm256_add_pd(one,_mm256_mul_pd(d2,_mm256_mul_pd(d,_mm256_add_pd(swV3,_mm256_mul_pd(d,_mm256_add_pd(swV4,_mm256_mul_pd(d,swV5)))))));
1507
1508             dsw              = _mm256_mul_pd(d2,_mm256_add_pd(swF2,_mm256_mul_pd(d,_mm256_add_pd(swF3,_mm256_mul_pd(d,swF4)))));
1509
1510             /* Evaluate switch function */
1511             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
1512             fvdw             = _mm256_sub_pd( _mm256_mul_pd(fvdw,sw) , _mm256_mul_pd(rinv00,_mm256_mul_pd(vvdw,dsw)) );
1513             cutoff_mask      = _mm256_cmp_pd(rsq00,rcutoff2,_CMP_LT_OQ);
1514
1515             fscal            = fvdw;
1516
1517             fscal            = _mm256_and_pd(fscal,cutoff_mask);
1518
1519             fscal            = _mm256_andnot_pd(dummy_mask,fscal);
1520
1521             /* Calculate temporary vectorial force */
1522             tx               = _mm256_mul_pd(fscal,dx00);
1523             ty               = _mm256_mul_pd(fscal,dy00);
1524             tz               = _mm256_mul_pd(fscal,dz00);
1525
1526             /* Update vectorial force */
1527             fix0             = _mm256_add_pd(fix0,tx);
1528             fiy0             = _mm256_add_pd(fiy0,ty);
1529             fiz0             = _mm256_add_pd(fiz0,tz);
1530
1531             fjx0             = _mm256_add_pd(fjx0,tx);
1532             fjy0             = _mm256_add_pd(fjy0,ty);
1533             fjz0             = _mm256_add_pd(fjz0,tz);
1534
1535             }
1536
1537             /**************************
1538              * CALCULATE INTERACTIONS *
1539              **************************/
1540
1541             if (gmx_mm256_any_lt(rsq10,rcutoff2))
1542             {
1543
1544             r10              = _mm256_mul_pd(rsq10,rinv10);
1545             r10              = _mm256_andnot_pd(dummy_mask,r10);
1546
1547             /* Compute parameters for interactions between i and j atoms */
1548             qq10             = _mm256_mul_pd(iq1,jq0);
1549
1550             /* EWALD ELECTROSTATICS */
1551
1552             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1553             ewrt             = _mm256_mul_pd(r10,ewtabscale);
1554             ewitab           = _mm256_cvttpd_epi32(ewrt);
1555             eweps            = _mm256_sub_pd(ewrt,_mm256_round_pd(ewrt, _MM_FROUND_FLOOR));
1556             ewitab           = _mm_slli_epi32(ewitab,2);
1557             ewtabF           = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,0) );
1558             ewtabD           = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,1) );
1559             ewtabV           = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,2) );
1560             ewtabFn          = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,3) );
1561             GMX_MM256_FULLTRANSPOSE4_PD(ewtabF,ewtabD,ewtabV,ewtabFn);
1562             felec            = _mm256_add_pd(ewtabF,_mm256_mul_pd(eweps,ewtabD));
1563             velec            = _mm256_sub_pd(ewtabV,_mm256_mul_pd(_mm256_mul_pd(ewtabhalfspace,eweps),_mm256_add_pd(ewtabF,felec)));
1564             velec            = _mm256_mul_pd(qq10,_mm256_sub_pd(rinv10,velec));
1565             felec            = _mm256_mul_pd(_mm256_mul_pd(qq10,rinv10),_mm256_sub_pd(rinvsq10,felec));
1566
1567             d                = _mm256_sub_pd(r10,rswitch);
1568             d                = _mm256_max_pd(d,_mm256_setzero_pd());
1569             d2               = _mm256_mul_pd(d,d);
1570             sw               = _mm256_add_pd(one,_mm256_mul_pd(d2,_mm256_mul_pd(d,_mm256_add_pd(swV3,_mm256_mul_pd(d,_mm256_add_pd(swV4,_mm256_mul_pd(d,swV5)))))));
1571
1572             dsw              = _mm256_mul_pd(d2,_mm256_add_pd(swF2,_mm256_mul_pd(d,_mm256_add_pd(swF3,_mm256_mul_pd(d,swF4)))));
1573
1574             /* Evaluate switch function */
1575             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
1576             felec            = _mm256_sub_pd( _mm256_mul_pd(felec,sw) , _mm256_mul_pd(rinv10,_mm256_mul_pd(velec,dsw)) );
1577             cutoff_mask      = _mm256_cmp_pd(rsq10,rcutoff2,_CMP_LT_OQ);
1578
1579             fscal            = felec;
1580
1581             fscal            = _mm256_and_pd(fscal,cutoff_mask);
1582
1583             fscal            = _mm256_andnot_pd(dummy_mask,fscal);
1584
1585             /* Calculate temporary vectorial force */
1586             tx               = _mm256_mul_pd(fscal,dx10);
1587             ty               = _mm256_mul_pd(fscal,dy10);
1588             tz               = _mm256_mul_pd(fscal,dz10);
1589
1590             /* Update vectorial force */
1591             fix1             = _mm256_add_pd(fix1,tx);
1592             fiy1             = _mm256_add_pd(fiy1,ty);
1593             fiz1             = _mm256_add_pd(fiz1,tz);
1594
1595             fjx0             = _mm256_add_pd(fjx0,tx);
1596             fjy0             = _mm256_add_pd(fjy0,ty);
1597             fjz0             = _mm256_add_pd(fjz0,tz);
1598
1599             }
1600
1601             /**************************
1602              * CALCULATE INTERACTIONS *
1603              **************************/
1604
1605             if (gmx_mm256_any_lt(rsq20,rcutoff2))
1606             {
1607
1608             r20              = _mm256_mul_pd(rsq20,rinv20);
1609             r20              = _mm256_andnot_pd(dummy_mask,r20);
1610
1611             /* Compute parameters for interactions between i and j atoms */
1612             qq20             = _mm256_mul_pd(iq2,jq0);
1613
1614             /* EWALD ELECTROSTATICS */
1615
1616             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1617             ewrt             = _mm256_mul_pd(r20,ewtabscale);
1618             ewitab           = _mm256_cvttpd_epi32(ewrt);
1619             eweps            = _mm256_sub_pd(ewrt,_mm256_round_pd(ewrt, _MM_FROUND_FLOOR));
1620             ewitab           = _mm_slli_epi32(ewitab,2);
1621             ewtabF           = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,0) );
1622             ewtabD           = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,1) );
1623             ewtabV           = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,2) );
1624             ewtabFn          = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,3) );
1625             GMX_MM256_FULLTRANSPOSE4_PD(ewtabF,ewtabD,ewtabV,ewtabFn);
1626             felec            = _mm256_add_pd(ewtabF,_mm256_mul_pd(eweps,ewtabD));
1627             velec            = _mm256_sub_pd(ewtabV,_mm256_mul_pd(_mm256_mul_pd(ewtabhalfspace,eweps),_mm256_add_pd(ewtabF,felec)));
1628             velec            = _mm256_mul_pd(qq20,_mm256_sub_pd(rinv20,velec));
1629             felec            = _mm256_mul_pd(_mm256_mul_pd(qq20,rinv20),_mm256_sub_pd(rinvsq20,felec));
1630
1631             d                = _mm256_sub_pd(r20,rswitch);
1632             d                = _mm256_max_pd(d,_mm256_setzero_pd());
1633             d2               = _mm256_mul_pd(d,d);
1634             sw               = _mm256_add_pd(one,_mm256_mul_pd(d2,_mm256_mul_pd(d,_mm256_add_pd(swV3,_mm256_mul_pd(d,_mm256_add_pd(swV4,_mm256_mul_pd(d,swV5)))))));
1635
1636             dsw              = _mm256_mul_pd(d2,_mm256_add_pd(swF2,_mm256_mul_pd(d,_mm256_add_pd(swF3,_mm256_mul_pd(d,swF4)))));
1637
1638             /* Evaluate switch function */
1639             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
1640             felec            = _mm256_sub_pd( _mm256_mul_pd(felec,sw) , _mm256_mul_pd(rinv20,_mm256_mul_pd(velec,dsw)) );
1641             cutoff_mask      = _mm256_cmp_pd(rsq20,rcutoff2,_CMP_LT_OQ);
1642
1643             fscal            = felec;
1644
1645             fscal            = _mm256_and_pd(fscal,cutoff_mask);
1646
1647             fscal            = _mm256_andnot_pd(dummy_mask,fscal);
1648
1649             /* Calculate temporary vectorial force */
1650             tx               = _mm256_mul_pd(fscal,dx20);
1651             ty               = _mm256_mul_pd(fscal,dy20);
1652             tz               = _mm256_mul_pd(fscal,dz20);
1653
1654             /* Update vectorial force */
1655             fix2             = _mm256_add_pd(fix2,tx);
1656             fiy2             = _mm256_add_pd(fiy2,ty);
1657             fiz2             = _mm256_add_pd(fiz2,tz);
1658
1659             fjx0             = _mm256_add_pd(fjx0,tx);
1660             fjy0             = _mm256_add_pd(fjy0,ty);
1661             fjz0             = _mm256_add_pd(fjz0,tz);
1662
1663             }
1664
1665             /**************************
1666              * CALCULATE INTERACTIONS *
1667              **************************/
1668
1669             if (gmx_mm256_any_lt(rsq30,rcutoff2))
1670             {
1671
1672             r30              = _mm256_mul_pd(rsq30,rinv30);
1673             r30              = _mm256_andnot_pd(dummy_mask,r30);
1674
1675             /* Compute parameters for interactions between i and j atoms */
1676             qq30             = _mm256_mul_pd(iq3,jq0);
1677
1678             /* EWALD ELECTROSTATICS */
1679
1680             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1681             ewrt             = _mm256_mul_pd(r30,ewtabscale);
1682             ewitab           = _mm256_cvttpd_epi32(ewrt);
1683             eweps            = _mm256_sub_pd(ewrt,_mm256_round_pd(ewrt, _MM_FROUND_FLOOR));
1684             ewitab           = _mm_slli_epi32(ewitab,2);
1685             ewtabF           = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,0) );
1686             ewtabD           = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,1) );
1687             ewtabV           = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,2) );
1688             ewtabFn          = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,3) );
1689             GMX_MM256_FULLTRANSPOSE4_PD(ewtabF,ewtabD,ewtabV,ewtabFn);
1690             felec            = _mm256_add_pd(ewtabF,_mm256_mul_pd(eweps,ewtabD));
1691             velec            = _mm256_sub_pd(ewtabV,_mm256_mul_pd(_mm256_mul_pd(ewtabhalfspace,eweps),_mm256_add_pd(ewtabF,felec)));
1692             velec            = _mm256_mul_pd(qq30,_mm256_sub_pd(rinv30,velec));
1693             felec            = _mm256_mul_pd(_mm256_mul_pd(qq30,rinv30),_mm256_sub_pd(rinvsq30,felec));
1694
1695             d                = _mm256_sub_pd(r30,rswitch);
1696             d                = _mm256_max_pd(d,_mm256_setzero_pd());
1697             d2               = _mm256_mul_pd(d,d);
1698             sw               = _mm256_add_pd(one,_mm256_mul_pd(d2,_mm256_mul_pd(d,_mm256_add_pd(swV3,_mm256_mul_pd(d,_mm256_add_pd(swV4,_mm256_mul_pd(d,swV5)))))));
1699
1700             dsw              = _mm256_mul_pd(d2,_mm256_add_pd(swF2,_mm256_mul_pd(d,_mm256_add_pd(swF3,_mm256_mul_pd(d,swF4)))));
1701
1702             /* Evaluate switch function */
1703             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
1704             felec            = _mm256_sub_pd( _mm256_mul_pd(felec,sw) , _mm256_mul_pd(rinv30,_mm256_mul_pd(velec,dsw)) );
1705             cutoff_mask      = _mm256_cmp_pd(rsq30,rcutoff2,_CMP_LT_OQ);
1706
1707             fscal            = felec;
1708
1709             fscal            = _mm256_and_pd(fscal,cutoff_mask);
1710
1711             fscal            = _mm256_andnot_pd(dummy_mask,fscal);
1712
1713             /* Calculate temporary vectorial force */
1714             tx               = _mm256_mul_pd(fscal,dx30);
1715             ty               = _mm256_mul_pd(fscal,dy30);
1716             tz               = _mm256_mul_pd(fscal,dz30);
1717
1718             /* Update vectorial force */
1719             fix3             = _mm256_add_pd(fix3,tx);
1720             fiy3             = _mm256_add_pd(fiy3,ty);
1721             fiz3             = _mm256_add_pd(fiz3,tz);
1722
1723             fjx0             = _mm256_add_pd(fjx0,tx);
1724             fjy0             = _mm256_add_pd(fjy0,ty);
1725             fjz0             = _mm256_add_pd(fjz0,tz);
1726
1727             }
1728
1729             fjptrA             = (jnrlistA>=0) ? f+j_coord_offsetA : scratch;
1730             fjptrB             = (jnrlistB>=0) ? f+j_coord_offsetB : scratch;
1731             fjptrC             = (jnrlistC>=0) ? f+j_coord_offsetC : scratch;
1732             fjptrD             = (jnrlistD>=0) ? f+j_coord_offsetD : scratch;
1733
1734             gmx_mm256_decrement_1rvec_4ptr_swizzle_pd(fjptrA,fjptrB,fjptrC,fjptrD,fjx0,fjy0,fjz0);
1735
1736             /* Inner loop uses 249 flops */
1737         }
1738
1739         /* End of innermost loop */
1740
1741         gmx_mm256_update_iforce_4atom_swizzle_pd(fix0,fiy0,fiz0,fix1,fiy1,fiz1,fix2,fiy2,fiz2,fix3,fiy3,fiz3,
1742                                                  f+i_coord_offset,fshift+i_shift_offset);
1743
1744         /* Increment number of inner iterations */
1745         inneriter                  += j_index_end - j_index_start;
1746
1747         /* Outer loop uses 24 flops */
1748     }
1749
1750     /* Increment number of outer iterations */
1751     outeriter        += nri;
1752
1753     /* Update outer/inner flops */
1754
1755     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W4_F,outeriter*24 + inneriter*249);
1756 }