src/gromacs/gmxlib/nonbonded/nb_kernel_avx_256_single/nb_kernel_ElecEwSw_VdwLJSw_GeomW3P1_avx_256_single.c

   1 /*
   2  * This file is part of the GROMACS molecular simulation package.
   3  *
   4  * Copyright (c) 2012,2013,2014, by the GROMACS development team, led by
   5  * Mark Abraham, David van der Spoel, Berk Hess, and Erik Lindahl,
   6  * and including many others, as listed in the AUTHORS file in the
   7  * top-level source directory and at http://www.gromacs.org.
   8  *
   9  * GROMACS is free software; you can redistribute it and/or
  10  * modify it under the terms of the GNU Lesser General Public License
  11  * as published by the Free Software Foundation; either version 2.1
  12  * of the License, or (at your option) any later version.
  13  *
  14  * GROMACS is distributed in the hope that it will be useful,
  15  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  16  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  17  * Lesser General Public License for more details.
  18  *
  19  * You should have received a copy of the GNU Lesser General Public
  20  * License along with GROMACS; if not, see
  21  * http://www.gnu.org/licenses, or write to the Free Software Foundation,
  22  * Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02110-1301  USA.
  23  *
  24  * If you want to redistribute modifications to GROMACS, please
  25  * consider that scientific software is very special. Version
  26  * control is crucial - bugs must be traceable. We will be happy to
  27  * consider code for inclusion in the official distribution, but
  28  * derived work must not be called official GROMACS. Details are found
  29  * in the README & COPYING files - if they are missing, get the
  30  * official version at http://www.gromacs.org.
  31  *
  32  * To help us fund GROMACS development, we humbly ask that you cite
  33  * the research papers on the package. Check out http://www.gromacs.org.
  34  */
  35 /*
  36  * Note: this file was generated by the GROMACS avx_256_single kernel generator.
  37  */
  38 #include "config.h"
  39
  40 #include <math.h>
  41
  42 #include "../nb_kernel.h"
  43 #include "types/simple.h"
  44 #include "gromacs/math/vec.h"
  45 #include "nrnb.h"
  46
  47 #include "gromacs/simd/math_x86_avx_256_single.h"
  48 #include "kernelutil_x86_avx_256_single.h"
  49
  50 /*
  51  * Gromacs nonbonded kernel:   nb_kernel_ElecEwSw_VdwLJSw_GeomW3P1_VF_avx_256_single
  52  * Electrostatics interaction: Ewald
  53  * VdW interaction:            LennardJones
  54  * Geometry:                   Water3-Particle
  55  * Calculate force/pot:        PotentialAndForce
  56  */
  57 void
  58 nb_kernel_ElecEwSw_VdwLJSw_GeomW3P1_VF_avx_256_single
  59                     (t_nblist                    * gmx_restrict       nlist,
  60                      rvec                        * gmx_restrict          xx,
  61                      rvec                        * gmx_restrict          ff,
  62                      t_forcerec                  * gmx_restrict          fr,
  63                      t_mdatoms                   * gmx_restrict     mdatoms,
  64                      nb_kernel_data_t gmx_unused * gmx_restrict kernel_data,
  65                      t_nrnb                      * gmx_restrict        nrnb)
  66 {
  67     /* Suffixes 0,1,2,3 refer to particle indices for waters in the inner or outer loop, or
  68      * just 0 for non-waters.
  69      * Suffixes A,B,C,D,E,F,G,H refer to j loop unrolling done with AVX, e.g. for the eight different
  70      * jnr indices corresponding to data put in the four positions in the SIMD register.
  71      */
  72     int              i_shift_offset,i_coord_offset,outeriter,inneriter;
  73     int              j_index_start,j_index_end,jidx,nri,inr,ggid,iidx;
  74     int              jnrA,jnrB,jnrC,jnrD;
  75     int              jnrE,jnrF,jnrG,jnrH;
  76     int              jnrlistA,jnrlistB,jnrlistC,jnrlistD;
  77     int              jnrlistE,jnrlistF,jnrlistG,jnrlistH;
  78     int              j_coord_offsetA,j_coord_offsetB,j_coord_offsetC,j_coord_offsetD;
  79     int              j_coord_offsetE,j_coord_offsetF,j_coord_offsetG,j_coord_offsetH;
  80     int              *iinr,*jindex,*jjnr,*shiftidx,*gid;
  81     real             rcutoff_scalar;
  82     real             *shiftvec,*fshift,*x,*f;
  83     real             *fjptrA,*fjptrB,*fjptrC,*fjptrD,*fjptrE,*fjptrF,*fjptrG,*fjptrH;
  84     real             scratch[4*DIM];
  85     __m256           tx,ty,tz,fscal,rcutoff,rcutoff2,jidxall;
  86     real *           vdwioffsetptr0;
  87     __m256           ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
  88     real *           vdwioffsetptr1;
  89     __m256           ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
  90     real *           vdwioffsetptr2;
  91     __m256           ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
  92     int              vdwjidx0A,vdwjidx0B,vdwjidx0C,vdwjidx0D,vdwjidx0E,vdwjidx0F,vdwjidx0G,vdwjidx0H;
  93     __m256           jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
  94     __m256           dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00;
  95     __m256           dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10;
  96     __m256           dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20;
  97     __m256           velec,felec,velecsum,facel,crf,krf,krf2;
  98     real             *charge;
  99     int              nvdwtype;
 100     __m256           rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,sh_vdw_invrcut6;
 101     int              *vdwtype;
 102     real             *vdwparam;
 103     __m256           one_sixth   = _mm256_set1_ps(1.0/6.0);
 104     __m256           one_twelfth = _mm256_set1_ps(1.0/12.0);
 105     __m256i          ewitab;
 106     __m128i          ewitab_lo,ewitab_hi;
 107     __m256           ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace,ewtabF,ewtabFn,ewtabD,ewtabV;
 108     __m256           beta,beta2,beta3,zeta2,pmecorrF,pmecorrV,rinv3;
 109     real             *ewtab;
 110     __m256           rswitch,swV3,swV4,swV5,swF2,swF3,swF4,d,d2,sw,dsw;
 111     real             rswitch_scalar,d_scalar;
 112     __m256           dummy_mask,cutoff_mask;
 113     __m256           signbit = _mm256_castsi256_ps( _mm256_set1_epi32(0x80000000) );
 114     __m256           one     = _mm256_set1_ps(1.0);
 115     __m256           two     = _mm256_set1_ps(2.0);
 116     x                = xx[0];
 117     f                = ff[0];
 118
 119     nri              = nlist->nri;
 120     iinr             = nlist->iinr;
 121     jindex           = nlist->jindex;
 122     jjnr             = nlist->jjnr;
 123     shiftidx         = nlist->shift;
 124     gid              = nlist->gid;
 125     shiftvec         = fr->shift_vec[0];
 126     fshift           = fr->fshift[0];
 127     facel            = _mm256_set1_ps(fr->epsfac);
 128     charge           = mdatoms->chargeA;
 129     nvdwtype         = fr->ntype;
 130     vdwparam         = fr->nbfp;
 131     vdwtype          = mdatoms->typeA;
 132
 133     sh_ewald         = _mm256_set1_ps(fr->ic->sh_ewald);
 134     beta             = _mm256_set1_ps(fr->ic->ewaldcoeff_q);
 135     beta2            = _mm256_mul_ps(beta,beta);
 136     beta3            = _mm256_mul_ps(beta,beta2);
 137
 138     ewtab            = fr->ic->tabq_coul_FDV0;
 139     ewtabscale       = _mm256_set1_ps(fr->ic->tabq_scale);
 140     ewtabhalfspace   = _mm256_set1_ps(0.5/fr->ic->tabq_scale);
 141
 142     /* Setup water-specific parameters */
 143     inr              = nlist->iinr[0];
 144     iq0              = _mm256_mul_ps(facel,_mm256_set1_ps(charge[inr+0]));
 145     iq1              = _mm256_mul_ps(facel,_mm256_set1_ps(charge[inr+1]));
 146     iq2              = _mm256_mul_ps(facel,_mm256_set1_ps(charge[inr+2]));
 147     vdwioffsetptr0   = vdwparam+2*nvdwtype*vdwtype[inr+0];
 148
 149     /* When we use explicit cutoffs the value must be identical for elec and VdW, so use elec as an arbitrary choice */
 150     rcutoff_scalar   = fr->rcoulomb;
 151     rcutoff          = _mm256_set1_ps(rcutoff_scalar);
 152     rcutoff2         = _mm256_mul_ps(rcutoff,rcutoff);
 153
 154     rswitch_scalar   = fr->rcoulomb_switch;
 155     rswitch          = _mm256_set1_ps(rswitch_scalar);
 156     /* Setup switch parameters */
 157     d_scalar         = rcutoff_scalar-rswitch_scalar;
 158     d                = _mm256_set1_ps(d_scalar);
 159     swV3             = _mm256_set1_ps(-10.0/(d_scalar*d_scalar*d_scalar));
 160     swV4             = _mm256_set1_ps( 15.0/(d_scalar*d_scalar*d_scalar*d_scalar));
 161     swV5             = _mm256_set1_ps( -6.0/(d_scalar*d_scalar*d_scalar*d_scalar*d_scalar));
 162     swF2             = _mm256_set1_ps(-30.0/(d_scalar*d_scalar*d_scalar));
 163     swF3             = _mm256_set1_ps( 60.0/(d_scalar*d_scalar*d_scalar*d_scalar));
 164     swF4             = _mm256_set1_ps(-30.0/(d_scalar*d_scalar*d_scalar*d_scalar*d_scalar));
 165
 166     /* Avoid stupid compiler warnings */
 167     jnrA = jnrB = jnrC = jnrD = jnrE = jnrF = jnrG = jnrH = 0;
 168     j_coord_offsetA = 0;
 169     j_coord_offsetB = 0;
 170     j_coord_offsetC = 0;
 171     j_coord_offsetD = 0;
 172     j_coord_offsetE = 0;
 173     j_coord_offsetF = 0;
 174     j_coord_offsetG = 0;
 175     j_coord_offsetH = 0;
 176
 177     outeriter        = 0;
 178     inneriter        = 0;
 179
 180     for(iidx=0;iidx<4*DIM;iidx++)
 181     {
 182         scratch[iidx] = 0.0;
 183     }
 184
 185     /* Start outer loop over neighborlists */
 186     for(iidx=0; iidx<nri; iidx++)
 187     {
 188         /* Load shift vector for this list */
 189         i_shift_offset   = DIM*shiftidx[iidx];
 190
 191         /* Load limits for loop over neighbors */
 192         j_index_start    = jindex[iidx];
 193         j_index_end      = jindex[iidx+1];
 194
 195         /* Get outer coordinate index */
 196         inr              = iinr[iidx];
 197         i_coord_offset   = DIM*inr;
 198
 199         /* Load i particle coords and add shift vector */
 200         gmx_mm256_load_shift_and_3rvec_broadcast_ps(shiftvec+i_shift_offset,x+i_coord_offset,
 201                                                     &ix0,&iy0,&iz0,&ix1,&iy1,&iz1,&ix2,&iy2,&iz2);
 202
 203         fix0             = _mm256_setzero_ps();
 204         fiy0             = _mm256_setzero_ps();
 205         fiz0             = _mm256_setzero_ps();
 206         fix1             = _mm256_setzero_ps();
 207         fiy1             = _mm256_setzero_ps();
 208         fiz1             = _mm256_setzero_ps();
 209         fix2             = _mm256_setzero_ps();
 210         fiy2             = _mm256_setzero_ps();
 211         fiz2             = _mm256_setzero_ps();
 212
 213         /* Reset potential sums */
 214         velecsum         = _mm256_setzero_ps();
 215         vvdwsum          = _mm256_setzero_ps();
 216
 217         /* Start inner kernel loop */
 218         for(jidx=j_index_start; jidx<j_index_end && jjnr[jidx+7]>=0; jidx+=8)
 219         {
 220
 221             /* Get j neighbor index, and coordinate index */
 222             jnrA             = jjnr[jidx];
 223             jnrB             = jjnr[jidx+1];
 224             jnrC             = jjnr[jidx+2];
 225             jnrD             = jjnr[jidx+3];
 226             jnrE             = jjnr[jidx+4];
 227             jnrF             = jjnr[jidx+5];
 228             jnrG             = jjnr[jidx+6];
 229             jnrH             = jjnr[jidx+7];
 230             j_coord_offsetA  = DIM*jnrA;
 231             j_coord_offsetB  = DIM*jnrB;
 232             j_coord_offsetC  = DIM*jnrC;
 233             j_coord_offsetD  = DIM*jnrD;
 234             j_coord_offsetE  = DIM*jnrE;
 235             j_coord_offsetF  = DIM*jnrF;
 236             j_coord_offsetG  = DIM*jnrG;
 237             j_coord_offsetH  = DIM*jnrH;
 238
 239             /* load j atom coordinates */
 240             gmx_mm256_load_1rvec_8ptr_swizzle_ps(x+j_coord_offsetA,x+j_coord_offsetB,
 241                                                  x+j_coord_offsetC,x+j_coord_offsetD,
 242                                                  x+j_coord_offsetE,x+j_coord_offsetF,
 243                                                  x+j_coord_offsetG,x+j_coord_offsetH,
 244                                                  &jx0,&jy0,&jz0);
 245
 246             /* Calculate displacement vector */
 247             dx00             = _mm256_sub_ps(ix0,jx0);
 248             dy00             = _mm256_sub_ps(iy0,jy0);
 249             dz00             = _mm256_sub_ps(iz0,jz0);
 250             dx10             = _mm256_sub_ps(ix1,jx0);
 251             dy10             = _mm256_sub_ps(iy1,jy0);
 252             dz10             = _mm256_sub_ps(iz1,jz0);
 253             dx20             = _mm256_sub_ps(ix2,jx0);
 254             dy20             = _mm256_sub_ps(iy2,jy0);
 255             dz20             = _mm256_sub_ps(iz2,jz0);
 256
 257             /* Calculate squared distance and things based on it */
 258             rsq00            = gmx_mm256_calc_rsq_ps(dx00,dy00,dz00);
 259             rsq10            = gmx_mm256_calc_rsq_ps(dx10,dy10,dz10);
 260             rsq20            = gmx_mm256_calc_rsq_ps(dx20,dy20,dz20);
 261
 262             rinv00           = gmx_mm256_invsqrt_ps(rsq00);
 263             rinv10           = gmx_mm256_invsqrt_ps(rsq10);
 264             rinv20           = gmx_mm256_invsqrt_ps(rsq20);
 265
 266             rinvsq00         = _mm256_mul_ps(rinv00,rinv00);
 267             rinvsq10         = _mm256_mul_ps(rinv10,rinv10);
 268             rinvsq20         = _mm256_mul_ps(rinv20,rinv20);
 269
 270             /* Load parameters for j particles */
 271             jq0              = gmx_mm256_load_8real_swizzle_ps(charge+jnrA+0,charge+jnrB+0,
 272                                                                  charge+jnrC+0,charge+jnrD+0,
 273                                                                  charge+jnrE+0,charge+jnrF+0,
 274                                                                  charge+jnrG+0,charge+jnrH+0);
 275             vdwjidx0A        = 2*vdwtype[jnrA+0];
 276             vdwjidx0B        = 2*vdwtype[jnrB+0];
 277             vdwjidx0C        = 2*vdwtype[jnrC+0];
 278             vdwjidx0D        = 2*vdwtype[jnrD+0];
 279             vdwjidx0E        = 2*vdwtype[jnrE+0];
 280             vdwjidx0F        = 2*vdwtype[jnrF+0];
 281             vdwjidx0G        = 2*vdwtype[jnrG+0];
 282             vdwjidx0H        = 2*vdwtype[jnrH+0];
 283
 284             fjx0             = _mm256_setzero_ps();
 285             fjy0             = _mm256_setzero_ps();
 286             fjz0             = _mm256_setzero_ps();
 287
 288             /**************************
 289              * CALCULATE INTERACTIONS *
 290              **************************/
 291
 292             if (gmx_mm256_any_lt(rsq00,rcutoff2))
 293             {
 294
 295             r00              = _mm256_mul_ps(rsq00,rinv00);
 296
 297             /* Compute parameters for interactions between i and j atoms */
 298             qq00             = _mm256_mul_ps(iq0,jq0);
 299             gmx_mm256_load_8pair_swizzle_ps(vdwioffsetptr0+vdwjidx0A,
 300                                             vdwioffsetptr0+vdwjidx0B,
 301                                             vdwioffsetptr0+vdwjidx0C,
 302                                             vdwioffsetptr0+vdwjidx0D,
 303                                             vdwioffsetptr0+vdwjidx0E,
 304                                             vdwioffsetptr0+vdwjidx0F,
 305                                             vdwioffsetptr0+vdwjidx0G,
 306                                             vdwioffsetptr0+vdwjidx0H,
 307                                             &c6_00,&c12_00);
 308
 309             /* EWALD ELECTROSTATICS */
 310
 311             /* Analytical PME correction */
 312             zeta2            = _mm256_mul_ps(beta2,rsq00);
 313             rinv3            = _mm256_mul_ps(rinvsq00,rinv00);
 314             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
 315             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
 316             felec            = _mm256_mul_ps(qq00,felec);
 317             pmecorrV         = gmx_mm256_pmecorrV_ps(zeta2);
 318             pmecorrV         = _mm256_mul_ps(pmecorrV,beta);
 319             velec            = _mm256_sub_ps(rinv00,pmecorrV);
 320             velec            = _mm256_mul_ps(qq00,velec);
 321
 322             /* LENNARD-JONES DISPERSION/REPULSION */
 323
 324             rinvsix          = _mm256_mul_ps(_mm256_mul_ps(rinvsq00,rinvsq00),rinvsq00);
 325             vvdw6            = _mm256_mul_ps(c6_00,rinvsix);
 326             vvdw12           = _mm256_mul_ps(c12_00,_mm256_mul_ps(rinvsix,rinvsix));
 327             vvdw             = _mm256_sub_ps( _mm256_mul_ps(vvdw12,one_twelfth) , _mm256_mul_ps(vvdw6,one_sixth) );
 328             fvdw             = _mm256_mul_ps(_mm256_sub_ps(vvdw12,vvdw6),rinvsq00);
 329
 330             d                = _mm256_sub_ps(r00,rswitch);
 331             d                = _mm256_max_ps(d,_mm256_setzero_ps());
 332             d2               = _mm256_mul_ps(d,d);
 333             sw               = _mm256_add_ps(one,_mm256_mul_ps(d2,_mm256_mul_ps(d,_mm256_add_ps(swV3,_mm256_mul_ps(d,_mm256_add_ps(swV4,_mm256_mul_ps(d,swV5)))))));
 334
 335             dsw              = _mm256_mul_ps(d2,_mm256_add_ps(swF2,_mm256_mul_ps(d,_mm256_add_ps(swF3,_mm256_mul_ps(d,swF4)))));
 336
 337             /* Evaluate switch function */
 338             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 339             felec            = _mm256_sub_ps( _mm256_mul_ps(felec,sw) , _mm256_mul_ps(rinv00,_mm256_mul_ps(velec,dsw)) );
 340             fvdw             = _mm256_sub_ps( _mm256_mul_ps(fvdw,sw) , _mm256_mul_ps(rinv00,_mm256_mul_ps(vvdw,dsw)) );
 341             velec            = _mm256_mul_ps(velec,sw);
 342             vvdw             = _mm256_mul_ps(vvdw,sw);
 343             cutoff_mask      = _mm256_cmp_ps(rsq00,rcutoff2,_CMP_LT_OQ);
 344
 345             /* Update potential sum for this i atom from the interaction with this j atom. */
 346             velec            = _mm256_and_ps(velec,cutoff_mask);
 347             velecsum         = _mm256_add_ps(velecsum,velec);
 348             vvdw             = _mm256_and_ps(vvdw,cutoff_mask);
 349             vvdwsum          = _mm256_add_ps(vvdwsum,vvdw);
 350
 351             fscal            = _mm256_add_ps(felec,fvdw);
 352
 353             fscal            = _mm256_and_ps(fscal,cutoff_mask);
 354
 355             /* Calculate temporary vectorial force */
 356             tx               = _mm256_mul_ps(fscal,dx00);
 357             ty               = _mm256_mul_ps(fscal,dy00);
 358             tz               = _mm256_mul_ps(fscal,dz00);
 359
 360             /* Update vectorial force */
 361             fix0             = _mm256_add_ps(fix0,tx);
 362             fiy0             = _mm256_add_ps(fiy0,ty);
 363             fiz0             = _mm256_add_ps(fiz0,tz);
 364
 365             fjx0             = _mm256_add_ps(fjx0,tx);
 366             fjy0             = _mm256_add_ps(fjy0,ty);
 367             fjz0             = _mm256_add_ps(fjz0,tz);
 368
 369             }
 370
 371             /**************************
 372              * CALCULATE INTERACTIONS *
 373              **************************/
 374
 375             if (gmx_mm256_any_lt(rsq10,rcutoff2))
 376             {
 377
 378             r10              = _mm256_mul_ps(rsq10,rinv10);
 379
 380             /* Compute parameters for interactions between i and j atoms */
 381             qq10             = _mm256_mul_ps(iq1,jq0);
 382
 383             /* EWALD ELECTROSTATICS */
 384
 385             /* Analytical PME correction */
 386             zeta2            = _mm256_mul_ps(beta2,rsq10);
 387             rinv3            = _mm256_mul_ps(rinvsq10,rinv10);
 388             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
 389             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
 390             felec            = _mm256_mul_ps(qq10,felec);
 391             pmecorrV         = gmx_mm256_pmecorrV_ps(zeta2);
 392             pmecorrV         = _mm256_mul_ps(pmecorrV,beta);
 393             velec            = _mm256_sub_ps(rinv10,pmecorrV);
 394             velec            = _mm256_mul_ps(qq10,velec);
 395
 396             d                = _mm256_sub_ps(r10,rswitch);
 397             d                = _mm256_max_ps(d,_mm256_setzero_ps());
 398             d2               = _mm256_mul_ps(d,d);
 399             sw               = _mm256_add_ps(one,_mm256_mul_ps(d2,_mm256_mul_ps(d,_mm256_add_ps(swV3,_mm256_mul_ps(d,_mm256_add_ps(swV4,_mm256_mul_ps(d,swV5)))))));
 400
 401             dsw              = _mm256_mul_ps(d2,_mm256_add_ps(swF2,_mm256_mul_ps(d,_mm256_add_ps(swF3,_mm256_mul_ps(d,swF4)))));
 402
 403             /* Evaluate switch function */
 404             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 405             felec            = _mm256_sub_ps( _mm256_mul_ps(felec,sw) , _mm256_mul_ps(rinv10,_mm256_mul_ps(velec,dsw)) );
 406             velec            = _mm256_mul_ps(velec,sw);
 407             cutoff_mask      = _mm256_cmp_ps(rsq10,rcutoff2,_CMP_LT_OQ);
 408
 409             /* Update potential sum for this i atom from the interaction with this j atom. */
 410             velec            = _mm256_and_ps(velec,cutoff_mask);
 411             velecsum         = _mm256_add_ps(velecsum,velec);
 412
 413             fscal            = felec;
 414
 415             fscal            = _mm256_and_ps(fscal,cutoff_mask);
 416
 417             /* Calculate temporary vectorial force */
 418             tx               = _mm256_mul_ps(fscal,dx10);
 419             ty               = _mm256_mul_ps(fscal,dy10);
 420             tz               = _mm256_mul_ps(fscal,dz10);
 421
 422             /* Update vectorial force */
 423             fix1             = _mm256_add_ps(fix1,tx);
 424             fiy1             = _mm256_add_ps(fiy1,ty);
 425             fiz1             = _mm256_add_ps(fiz1,tz);
 426
 427             fjx0             = _mm256_add_ps(fjx0,tx);
 428             fjy0             = _mm256_add_ps(fjy0,ty);
 429             fjz0             = _mm256_add_ps(fjz0,tz);
 430
 431             }
 432
 433             /**************************
 434              * CALCULATE INTERACTIONS *
 435              **************************/
 436
 437             if (gmx_mm256_any_lt(rsq20,rcutoff2))
 438             {
 439
 440             r20              = _mm256_mul_ps(rsq20,rinv20);
 441
 442             /* Compute parameters for interactions between i and j atoms */
 443             qq20             = _mm256_mul_ps(iq2,jq0);
 444
 445             /* EWALD ELECTROSTATICS */
 446
 447             /* Analytical PME correction */
 448             zeta2            = _mm256_mul_ps(beta2,rsq20);
 449             rinv3            = _mm256_mul_ps(rinvsq20,rinv20);
 450             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
 451             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
 452             felec            = _mm256_mul_ps(qq20,felec);
 453             pmecorrV         = gmx_mm256_pmecorrV_ps(zeta2);
 454             pmecorrV         = _mm256_mul_ps(pmecorrV,beta);
 455             velec            = _mm256_sub_ps(rinv20,pmecorrV);
 456             velec            = _mm256_mul_ps(qq20,velec);
 457
 458             d                = _mm256_sub_ps(r20,rswitch);
 459             d                = _mm256_max_ps(d,_mm256_setzero_ps());
 460             d2               = _mm256_mul_ps(d,d);
 461             sw               = _mm256_add_ps(one,_mm256_mul_ps(d2,_mm256_mul_ps(d,_mm256_add_ps(swV3,_mm256_mul_ps(d,_mm256_add_ps(swV4,_mm256_mul_ps(d,swV5)))))));
 462
 463             dsw              = _mm256_mul_ps(d2,_mm256_add_ps(swF2,_mm256_mul_ps(d,_mm256_add_ps(swF3,_mm256_mul_ps(d,swF4)))));
 464
 465             /* Evaluate switch function */
 466             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 467             felec            = _mm256_sub_ps( _mm256_mul_ps(felec,sw) , _mm256_mul_ps(rinv20,_mm256_mul_ps(velec,dsw)) );
 468             velec            = _mm256_mul_ps(velec,sw);
 469             cutoff_mask      = _mm256_cmp_ps(rsq20,rcutoff2,_CMP_LT_OQ);
 470
 471             /* Update potential sum for this i atom from the interaction with this j atom. */
 472             velec            = _mm256_and_ps(velec,cutoff_mask);
 473             velecsum         = _mm256_add_ps(velecsum,velec);
 474
 475             fscal            = felec;
 476
 477             fscal            = _mm256_and_ps(fscal,cutoff_mask);
 478
 479             /* Calculate temporary vectorial force */
 480             tx               = _mm256_mul_ps(fscal,dx20);
 481             ty               = _mm256_mul_ps(fscal,dy20);
 482             tz               = _mm256_mul_ps(fscal,dz20);
 483
 484             /* Update vectorial force */
 485             fix2             = _mm256_add_ps(fix2,tx);
 486             fiy2             = _mm256_add_ps(fiy2,ty);
 487             fiz2             = _mm256_add_ps(fiz2,tz);
 488
 489             fjx0             = _mm256_add_ps(fjx0,tx);
 490             fjy0             = _mm256_add_ps(fjy0,ty);
 491             fjz0             = _mm256_add_ps(fjz0,tz);
 492
 493             }
 494
 495             fjptrA             = f+j_coord_offsetA;
 496             fjptrB             = f+j_coord_offsetB;
 497             fjptrC             = f+j_coord_offsetC;
 498             fjptrD             = f+j_coord_offsetD;
 499             fjptrE             = f+j_coord_offsetE;
 500             fjptrF             = f+j_coord_offsetF;
 501             fjptrG             = f+j_coord_offsetG;
 502             fjptrH             = f+j_coord_offsetH;
 503
 504             gmx_mm256_decrement_1rvec_8ptr_swizzle_ps(fjptrA,fjptrB,fjptrC,fjptrD,fjptrE,fjptrF,fjptrG,fjptrH,fjx0,fjy0,fjz0);
 505
 506             /* Inner loop uses 345 flops */
 507         }
 508
 509         if(jidx<j_index_end)
 510         {
 511
 512             /* Get j neighbor index, and coordinate index */
 513             jnrlistA         = jjnr[jidx];
 514             jnrlistB         = jjnr[jidx+1];
 515             jnrlistC         = jjnr[jidx+2];
 516             jnrlistD         = jjnr[jidx+3];
 517             jnrlistE         = jjnr[jidx+4];
 518             jnrlistF         = jjnr[jidx+5];
 519             jnrlistG         = jjnr[jidx+6];
 520             jnrlistH         = jjnr[jidx+7];
 521             /* Sign of each element will be negative for non-real atoms.
 522              * This mask will be 0xFFFFFFFF for dummy entries and 0x0 for real ones,
 523              * so use it as val = _mm_andnot_ps(mask,val) to clear dummy entries.
 524              */
 525             dummy_mask = gmx_mm256_set_m128(gmx_mm_castsi128_ps(_mm_cmplt_epi32(_mm_loadu_si128((const __m128i *)(jjnr+jidx+4)),_mm_setzero_si128())),
 526                                             gmx_mm_castsi128_ps(_mm_cmplt_epi32(_mm_loadu_si128((const __m128i *)(jjnr+jidx)),_mm_setzero_si128())));
 527
 528             jnrA       = (jnrlistA>=0) ? jnrlistA : 0;
 529             jnrB       = (jnrlistB>=0) ? jnrlistB : 0;
 530             jnrC       = (jnrlistC>=0) ? jnrlistC : 0;
 531             jnrD       = (jnrlistD>=0) ? jnrlistD : 0;
 532             jnrE       = (jnrlistE>=0) ? jnrlistE : 0;
 533             jnrF       = (jnrlistF>=0) ? jnrlistF : 0;
 534             jnrG       = (jnrlistG>=0) ? jnrlistG : 0;
 535             jnrH       = (jnrlistH>=0) ? jnrlistH : 0;
 536             j_coord_offsetA  = DIM*jnrA;
 537             j_coord_offsetB  = DIM*jnrB;
 538             j_coord_offsetC  = DIM*jnrC;
 539             j_coord_offsetD  = DIM*jnrD;
 540             j_coord_offsetE  = DIM*jnrE;
 541             j_coord_offsetF  = DIM*jnrF;
 542             j_coord_offsetG  = DIM*jnrG;
 543             j_coord_offsetH  = DIM*jnrH;
 544
 545             /* load j atom coordinates */
 546             gmx_mm256_load_1rvec_8ptr_swizzle_ps(x+j_coord_offsetA,x+j_coord_offsetB,
 547                                                  x+j_coord_offsetC,x+j_coord_offsetD,
 548                                                  x+j_coord_offsetE,x+j_coord_offsetF,
 549                                                  x+j_coord_offsetG,x+j_coord_offsetH,
 550                                                  &jx0,&jy0,&jz0);
 551
 552             /* Calculate displacement vector */
 553             dx00             = _mm256_sub_ps(ix0,jx0);
 554             dy00             = _mm256_sub_ps(iy0,jy0);
 555             dz00             = _mm256_sub_ps(iz0,jz0);
 556             dx10             = _mm256_sub_ps(ix1,jx0);
 557             dy10             = _mm256_sub_ps(iy1,jy0);
 558             dz10             = _mm256_sub_ps(iz1,jz0);
 559             dx20             = _mm256_sub_ps(ix2,jx0);
 560             dy20             = _mm256_sub_ps(iy2,jy0);
 561             dz20             = _mm256_sub_ps(iz2,jz0);
 562
 563             /* Calculate squared distance and things based on it */
 564             rsq00            = gmx_mm256_calc_rsq_ps(dx00,dy00,dz00);
 565             rsq10            = gmx_mm256_calc_rsq_ps(dx10,dy10,dz10);
 566             rsq20            = gmx_mm256_calc_rsq_ps(dx20,dy20,dz20);
 567
 568             rinv00           = gmx_mm256_invsqrt_ps(rsq00);
 569             rinv10           = gmx_mm256_invsqrt_ps(rsq10);
 570             rinv20           = gmx_mm256_invsqrt_ps(rsq20);
 571
 572             rinvsq00         = _mm256_mul_ps(rinv00,rinv00);
 573             rinvsq10         = _mm256_mul_ps(rinv10,rinv10);
 574             rinvsq20         = _mm256_mul_ps(rinv20,rinv20);
 575
 576             /* Load parameters for j particles */
 577             jq0              = gmx_mm256_load_8real_swizzle_ps(charge+jnrA+0,charge+jnrB+0,
 578                                                                  charge+jnrC+0,charge+jnrD+0,
 579                                                                  charge+jnrE+0,charge+jnrF+0,
 580                                                                  charge+jnrG+0,charge+jnrH+0);
 581             vdwjidx0A        = 2*vdwtype[jnrA+0];
 582             vdwjidx0B        = 2*vdwtype[jnrB+0];
 583             vdwjidx0C        = 2*vdwtype[jnrC+0];
 584             vdwjidx0D        = 2*vdwtype[jnrD+0];
 585             vdwjidx0E        = 2*vdwtype[jnrE+0];
 586             vdwjidx0F        = 2*vdwtype[jnrF+0];
 587             vdwjidx0G        = 2*vdwtype[jnrG+0];
 588             vdwjidx0H        = 2*vdwtype[jnrH+0];
 589
 590             fjx0             = _mm256_setzero_ps();
 591             fjy0             = _mm256_setzero_ps();
 592             fjz0             = _mm256_setzero_ps();
 593
 594             /**************************
 595              * CALCULATE INTERACTIONS *
 596              **************************/
 597
 598             if (gmx_mm256_any_lt(rsq00,rcutoff2))
 599             {
 600
 601             r00              = _mm256_mul_ps(rsq00,rinv00);
 602             r00              = _mm256_andnot_ps(dummy_mask,r00);
 603
 604             /* Compute parameters for interactions between i and j atoms */
 605             qq00             = _mm256_mul_ps(iq0,jq0);
 606             gmx_mm256_load_8pair_swizzle_ps(vdwioffsetptr0+vdwjidx0A,
 607                                             vdwioffsetptr0+vdwjidx0B,
 608                                             vdwioffsetptr0+vdwjidx0C,
 609                                             vdwioffsetptr0+vdwjidx0D,
 610                                             vdwioffsetptr0+vdwjidx0E,
 611                                             vdwioffsetptr0+vdwjidx0F,
 612                                             vdwioffsetptr0+vdwjidx0G,
 613                                             vdwioffsetptr0+vdwjidx0H,
 614                                             &c6_00,&c12_00);
 615
 616             /* EWALD ELECTROSTATICS */
 617
 618             /* Analytical PME correction */
 619             zeta2            = _mm256_mul_ps(beta2,rsq00);
 620             rinv3            = _mm256_mul_ps(rinvsq00,rinv00);
 621             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
 622             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
 623             felec            = _mm256_mul_ps(qq00,felec);
 624             pmecorrV         = gmx_mm256_pmecorrV_ps(zeta2);
 625             pmecorrV         = _mm256_mul_ps(pmecorrV,beta);
 626             velec            = _mm256_sub_ps(rinv00,pmecorrV);
 627             velec            = _mm256_mul_ps(qq00,velec);
 628
 629             /* LENNARD-JONES DISPERSION/REPULSION */
 630
 631             rinvsix          = _mm256_mul_ps(_mm256_mul_ps(rinvsq00,rinvsq00),rinvsq00);
 632             vvdw6            = _mm256_mul_ps(c6_00,rinvsix);
 633             vvdw12           = _mm256_mul_ps(c12_00,_mm256_mul_ps(rinvsix,rinvsix));
 634             vvdw             = _mm256_sub_ps( _mm256_mul_ps(vvdw12,one_twelfth) , _mm256_mul_ps(vvdw6,one_sixth) );
 635             fvdw             = _mm256_mul_ps(_mm256_sub_ps(vvdw12,vvdw6),rinvsq00);
 636
 637             d                = _mm256_sub_ps(r00,rswitch);
 638             d                = _mm256_max_ps(d,_mm256_setzero_ps());
 639             d2               = _mm256_mul_ps(d,d);
 640             sw               = _mm256_add_ps(one,_mm256_mul_ps(d2,_mm256_mul_ps(d,_mm256_add_ps(swV3,_mm256_mul_ps(d,_mm256_add_ps(swV4,_mm256_mul_ps(d,swV5)))))));
 641
 642             dsw              = _mm256_mul_ps(d2,_mm256_add_ps(swF2,_mm256_mul_ps(d,_mm256_add_ps(swF3,_mm256_mul_ps(d,swF4)))));
 643
 644             /* Evaluate switch function */
 645             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 646             felec            = _mm256_sub_ps( _mm256_mul_ps(felec,sw) , _mm256_mul_ps(rinv00,_mm256_mul_ps(velec,dsw)) );
 647             fvdw             = _mm256_sub_ps( _mm256_mul_ps(fvdw,sw) , _mm256_mul_ps(rinv00,_mm256_mul_ps(vvdw,dsw)) );
 648             velec            = _mm256_mul_ps(velec,sw);
 649             vvdw             = _mm256_mul_ps(vvdw,sw);
 650             cutoff_mask      = _mm256_cmp_ps(rsq00,rcutoff2,_CMP_LT_OQ);
 651
 652             /* Update potential sum for this i atom from the interaction with this j atom. */
 653             velec            = _mm256_and_ps(velec,cutoff_mask);
 654             velec            = _mm256_andnot_ps(dummy_mask,velec);
 655             velecsum         = _mm256_add_ps(velecsum,velec);
 656             vvdw             = _mm256_and_ps(vvdw,cutoff_mask);
 657             vvdw             = _mm256_andnot_ps(dummy_mask,vvdw);
 658             vvdwsum          = _mm256_add_ps(vvdwsum,vvdw);
 659
 660             fscal            = _mm256_add_ps(felec,fvdw);
 661
 662             fscal            = _mm256_and_ps(fscal,cutoff_mask);
 663
 664             fscal            = _mm256_andnot_ps(dummy_mask,fscal);
 665
 666             /* Calculate temporary vectorial force */
 667             tx               = _mm256_mul_ps(fscal,dx00);
 668             ty               = _mm256_mul_ps(fscal,dy00);
 669             tz               = _mm256_mul_ps(fscal,dz00);
 670
 671             /* Update vectorial force */
 672             fix0             = _mm256_add_ps(fix0,tx);
 673             fiy0             = _mm256_add_ps(fiy0,ty);
 674             fiz0             = _mm256_add_ps(fiz0,tz);
 675
 676             fjx0             = _mm256_add_ps(fjx0,tx);
 677             fjy0             = _mm256_add_ps(fjy0,ty);
 678             fjz0             = _mm256_add_ps(fjz0,tz);
 679
 680             }
 681
 682             /**************************
 683              * CALCULATE INTERACTIONS *
 684              **************************/
 685
 686             if (gmx_mm256_any_lt(rsq10,rcutoff2))
 687             {
 688
 689             r10              = _mm256_mul_ps(rsq10,rinv10);
 690             r10              = _mm256_andnot_ps(dummy_mask,r10);
 691
 692             /* Compute parameters for interactions between i and j atoms */
 693             qq10             = _mm256_mul_ps(iq1,jq0);
 694
 695             /* EWALD ELECTROSTATICS */
 696
 697             /* Analytical PME correction */
 698             zeta2            = _mm256_mul_ps(beta2,rsq10);
 699             rinv3            = _mm256_mul_ps(rinvsq10,rinv10);
 700             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
 701             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
 702             felec            = _mm256_mul_ps(qq10,felec);
 703             pmecorrV         = gmx_mm256_pmecorrV_ps(zeta2);
 704             pmecorrV         = _mm256_mul_ps(pmecorrV,beta);
 705             velec            = _mm256_sub_ps(rinv10,pmecorrV);
 706             velec            = _mm256_mul_ps(qq10,velec);
 707
 708             d                = _mm256_sub_ps(r10,rswitch);
 709             d                = _mm256_max_ps(d,_mm256_setzero_ps());
 710             d2               = _mm256_mul_ps(d,d);
 711             sw               = _mm256_add_ps(one,_mm256_mul_ps(d2,_mm256_mul_ps(d,_mm256_add_ps(swV3,_mm256_mul_ps(d,_mm256_add_ps(swV4,_mm256_mul_ps(d,swV5)))))));
 712
 713             dsw              = _mm256_mul_ps(d2,_mm256_add_ps(swF2,_mm256_mul_ps(d,_mm256_add_ps(swF3,_mm256_mul_ps(d,swF4)))));
 714
 715             /* Evaluate switch function */
 716             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 717             felec            = _mm256_sub_ps( _mm256_mul_ps(felec,sw) , _mm256_mul_ps(rinv10,_mm256_mul_ps(velec,dsw)) );
 718             velec            = _mm256_mul_ps(velec,sw);
 719             cutoff_mask      = _mm256_cmp_ps(rsq10,rcutoff2,_CMP_LT_OQ);
 720
 721             /* Update potential sum for this i atom from the interaction with this j atom. */
 722             velec            = _mm256_and_ps(velec,cutoff_mask);
 723             velec            = _mm256_andnot_ps(dummy_mask,velec);
 724             velecsum         = _mm256_add_ps(velecsum,velec);
 725
 726             fscal            = felec;
 727
 728             fscal            = _mm256_and_ps(fscal,cutoff_mask);
 729
 730             fscal            = _mm256_andnot_ps(dummy_mask,fscal);
 731
 732             /* Calculate temporary vectorial force */
 733             tx               = _mm256_mul_ps(fscal,dx10);
 734             ty               = _mm256_mul_ps(fscal,dy10);
 735             tz               = _mm256_mul_ps(fscal,dz10);
 736
 737             /* Update vectorial force */
 738             fix1             = _mm256_add_ps(fix1,tx);
 739             fiy1             = _mm256_add_ps(fiy1,ty);
 740             fiz1             = _mm256_add_ps(fiz1,tz);
 741
 742             fjx0             = _mm256_add_ps(fjx0,tx);
 743             fjy0             = _mm256_add_ps(fjy0,ty);
 744             fjz0             = _mm256_add_ps(fjz0,tz);
 745
 746             }
 747
 748             /**************************
 749              * CALCULATE INTERACTIONS *
 750              **************************/
 751
 752             if (gmx_mm256_any_lt(rsq20,rcutoff2))
 753             {
 754
 755             r20              = _mm256_mul_ps(rsq20,rinv20);
 756             r20              = _mm256_andnot_ps(dummy_mask,r20);
 757
 758             /* Compute parameters for interactions between i and j atoms */
 759             qq20             = _mm256_mul_ps(iq2,jq0);
 760
 761             /* EWALD ELECTROSTATICS */
 762
 763             /* Analytical PME correction */
 764             zeta2            = _mm256_mul_ps(beta2,rsq20);
 765             rinv3            = _mm256_mul_ps(rinvsq20,rinv20);
 766             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
 767             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
 768             felec            = _mm256_mul_ps(qq20,felec);
 769             pmecorrV         = gmx_mm256_pmecorrV_ps(zeta2);
 770             pmecorrV         = _mm256_mul_ps(pmecorrV,beta);
 771             velec            = _mm256_sub_ps(rinv20,pmecorrV);
 772             velec            = _mm256_mul_ps(qq20,velec);
 773
 774             d                = _mm256_sub_ps(r20,rswitch);
 775             d                = _mm256_max_ps(d,_mm256_setzero_ps());
 776             d2               = _mm256_mul_ps(d,d);
 777             sw               = _mm256_add_ps(one,_mm256_mul_ps(d2,_mm256_mul_ps(d,_mm256_add_ps(swV3,_mm256_mul_ps(d,_mm256_add_ps(swV4,_mm256_mul_ps(d,swV5)))))));
 778
 779             dsw              = _mm256_mul_ps(d2,_mm256_add_ps(swF2,_mm256_mul_ps(d,_mm256_add_ps(swF3,_mm256_mul_ps(d,swF4)))));
 780
 781             /* Evaluate switch function */
 782             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 783             felec            = _mm256_sub_ps( _mm256_mul_ps(felec,sw) , _mm256_mul_ps(rinv20,_mm256_mul_ps(velec,dsw)) );
 784             velec            = _mm256_mul_ps(velec,sw);
 785             cutoff_mask      = _mm256_cmp_ps(rsq20,rcutoff2,_CMP_LT_OQ);
 786
 787             /* Update potential sum for this i atom from the interaction with this j atom. */
 788             velec            = _mm256_and_ps(velec,cutoff_mask);
 789             velec            = _mm256_andnot_ps(dummy_mask,velec);
 790             velecsum         = _mm256_add_ps(velecsum,velec);
 791
 792             fscal            = felec;
 793
 794             fscal            = _mm256_and_ps(fscal,cutoff_mask);
 795
 796             fscal            = _mm256_andnot_ps(dummy_mask,fscal);
 797
 798             /* Calculate temporary vectorial force */
 799             tx               = _mm256_mul_ps(fscal,dx20);
 800             ty               = _mm256_mul_ps(fscal,dy20);
 801             tz               = _mm256_mul_ps(fscal,dz20);
 802
 803             /* Update vectorial force */
 804             fix2             = _mm256_add_ps(fix2,tx);
 805             fiy2             = _mm256_add_ps(fiy2,ty);
 806             fiz2             = _mm256_add_ps(fiz2,tz);
 807
 808             fjx0             = _mm256_add_ps(fjx0,tx);
 809             fjy0             = _mm256_add_ps(fjy0,ty);
 810             fjz0             = _mm256_add_ps(fjz0,tz);
 811
 812             }
 813
 814             fjptrA             = (jnrlistA>=0) ? f+j_coord_offsetA : scratch;
 815             fjptrB             = (jnrlistB>=0) ? f+j_coord_offsetB : scratch;
 816             fjptrC             = (jnrlistC>=0) ? f+j_coord_offsetC : scratch;
 817             fjptrD             = (jnrlistD>=0) ? f+j_coord_offsetD : scratch;
 818             fjptrE             = (jnrlistE>=0) ? f+j_coord_offsetE : scratch;
 819             fjptrF             = (jnrlistF>=0) ? f+j_coord_offsetF : scratch;
 820             fjptrG             = (jnrlistG>=0) ? f+j_coord_offsetG : scratch;
 821             fjptrH             = (jnrlistH>=0) ? f+j_coord_offsetH : scratch;
 822
 823             gmx_mm256_decrement_1rvec_8ptr_swizzle_ps(fjptrA,fjptrB,fjptrC,fjptrD,fjptrE,fjptrF,fjptrG,fjptrH,fjx0,fjy0,fjz0);
 824
 825             /* Inner loop uses 348 flops */
 826         }
 827
 828         /* End of innermost loop */
 829
 830         gmx_mm256_update_iforce_3atom_swizzle_ps(fix0,fiy0,fiz0,fix1,fiy1,fiz1,fix2,fiy2,fiz2,
 831                                                  f+i_coord_offset,fshift+i_shift_offset);
 832
 833         ggid                        = gid[iidx];
 834         /* Update potential energies */
 835         gmx_mm256_update_1pot_ps(velecsum,kernel_data->energygrp_elec+ggid);
 836         gmx_mm256_update_1pot_ps(vvdwsum,kernel_data->energygrp_vdw+ggid);
 837
 838         /* Increment number of inner iterations */
 839         inneriter                  += j_index_end - j_index_start;
 840
 841         /* Outer loop uses 20 flops */
 842     }
 843
 844     /* Increment number of outer iterations */
 845     outeriter        += nri;
 846
 847     /* Update outer/inner flops */
 848
 849     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W3_VF,outeriter*20 + inneriter*348);
 850 }
 851 /*
 852  * Gromacs nonbonded kernel:   nb_kernel_ElecEwSw_VdwLJSw_GeomW3P1_F_avx_256_single
 853  * Electrostatics interaction: Ewald
 854  * VdW interaction:            LennardJones
 855  * Geometry:                   Water3-Particle
 856  * Calculate force/pot:        Force
 857  */
 858 void
 859 nb_kernel_ElecEwSw_VdwLJSw_GeomW3P1_F_avx_256_single
 860                     (t_nblist                    * gmx_restrict       nlist,
 861                      rvec                        * gmx_restrict          xx,
 862                      rvec                        * gmx_restrict          ff,
 863                      t_forcerec                  * gmx_restrict          fr,
 864                      t_mdatoms                   * gmx_restrict     mdatoms,
 865                      nb_kernel_data_t gmx_unused * gmx_restrict kernel_data,
 866                      t_nrnb                      * gmx_restrict        nrnb)
 867 {
 868     /* Suffixes 0,1,2,3 refer to particle indices for waters in the inner or outer loop, or
 869      * just 0 for non-waters.
 870      * Suffixes A,B,C,D,E,F,G,H refer to j loop unrolling done with AVX, e.g. for the eight different
 871      * jnr indices corresponding to data put in the four positions in the SIMD register.
 872      */
 873     int              i_shift_offset,i_coord_offset,outeriter,inneriter;
 874     int              j_index_start,j_index_end,jidx,nri,inr,ggid,iidx;
 875     int              jnrA,jnrB,jnrC,jnrD;
 876     int              jnrE,jnrF,jnrG,jnrH;
 877     int              jnrlistA,jnrlistB,jnrlistC,jnrlistD;
 878     int              jnrlistE,jnrlistF,jnrlistG,jnrlistH;
 879     int              j_coord_offsetA,j_coord_offsetB,j_coord_offsetC,j_coord_offsetD;
 880     int              j_coord_offsetE,j_coord_offsetF,j_coord_offsetG,j_coord_offsetH;
 881     int              *iinr,*jindex,*jjnr,*shiftidx,*gid;
 882     real             rcutoff_scalar;
 883     real             *shiftvec,*fshift,*x,*f;
 884     real             *fjptrA,*fjptrB,*fjptrC,*fjptrD,*fjptrE,*fjptrF,*fjptrG,*fjptrH;
 885     real             scratch[4*DIM];
 886     __m256           tx,ty,tz,fscal,rcutoff,rcutoff2,jidxall;
 887     real *           vdwioffsetptr0;
 888     __m256           ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
 889     real *           vdwioffsetptr1;
 890     __m256           ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
 891     real *           vdwioffsetptr2;
 892     __m256           ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
 893     int              vdwjidx0A,vdwjidx0B,vdwjidx0C,vdwjidx0D,vdwjidx0E,vdwjidx0F,vdwjidx0G,vdwjidx0H;
 894     __m256           jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
 895     __m256           dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00;
 896     __m256           dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10;
 897     __m256           dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20;
 898     __m256           velec,felec,velecsum,facel,crf,krf,krf2;
 899     real             *charge;
 900     int              nvdwtype;
 901     __m256           rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,sh_vdw_invrcut6;
 902     int              *vdwtype;
 903     real             *vdwparam;
 904     __m256           one_sixth   = _mm256_set1_ps(1.0/6.0);
 905     __m256           one_twelfth = _mm256_set1_ps(1.0/12.0);
 906     __m256i          ewitab;
 907     __m128i          ewitab_lo,ewitab_hi;
 908     __m256           ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace,ewtabF,ewtabFn,ewtabD,ewtabV;
 909     __m256           beta,beta2,beta3,zeta2,pmecorrF,pmecorrV,rinv3;
 910     real             *ewtab;
 911     __m256           rswitch,swV3,swV4,swV5,swF2,swF3,swF4,d,d2,sw,dsw;
 912     real             rswitch_scalar,d_scalar;
 913     __m256           dummy_mask,cutoff_mask;
 914     __m256           signbit = _mm256_castsi256_ps( _mm256_set1_epi32(0x80000000) );
 915     __m256           one     = _mm256_set1_ps(1.0);
 916     __m256           two     = _mm256_set1_ps(2.0);
 917     x                = xx[0];
 918     f                = ff[0];
 919
 920     nri              = nlist->nri;
 921     iinr             = nlist->iinr;
 922     jindex           = nlist->jindex;
 923     jjnr             = nlist->jjnr;
 924     shiftidx         = nlist->shift;
 925     gid              = nlist->gid;
 926     shiftvec         = fr->shift_vec[0];
 927     fshift           = fr->fshift[0];
 928     facel            = _mm256_set1_ps(fr->epsfac);
 929     charge           = mdatoms->chargeA;
 930     nvdwtype         = fr->ntype;
 931     vdwparam         = fr->nbfp;
 932     vdwtype          = mdatoms->typeA;
 933
 934     sh_ewald         = _mm256_set1_ps(fr->ic->sh_ewald);
 935     beta             = _mm256_set1_ps(fr->ic->ewaldcoeff_q);
 936     beta2            = _mm256_mul_ps(beta,beta);
 937     beta3            = _mm256_mul_ps(beta,beta2);
 938
 939     ewtab            = fr->ic->tabq_coul_FDV0;
 940     ewtabscale       = _mm256_set1_ps(fr->ic->tabq_scale);
 941     ewtabhalfspace   = _mm256_set1_ps(0.5/fr->ic->tabq_scale);
 942
 943     /* Setup water-specific parameters */
 944     inr              = nlist->iinr[0];
 945     iq0              = _mm256_mul_ps(facel,_mm256_set1_ps(charge[inr+0]));
 946     iq1              = _mm256_mul_ps(facel,_mm256_set1_ps(charge[inr+1]));
 947     iq2              = _mm256_mul_ps(facel,_mm256_set1_ps(charge[inr+2]));
 948     vdwioffsetptr0   = vdwparam+2*nvdwtype*vdwtype[inr+0];
 949
 950     /* When we use explicit cutoffs the value must be identical for elec and VdW, so use elec as an arbitrary choice */
 951     rcutoff_scalar   = fr->rcoulomb;
 952     rcutoff          = _mm256_set1_ps(rcutoff_scalar);
 953     rcutoff2         = _mm256_mul_ps(rcutoff,rcutoff);
 954
 955     rswitch_scalar   = fr->rcoulomb_switch;
 956     rswitch          = _mm256_set1_ps(rswitch_scalar);
 957     /* Setup switch parameters */
 958     d_scalar         = rcutoff_scalar-rswitch_scalar;
 959     d                = _mm256_set1_ps(d_scalar);
 960     swV3             = _mm256_set1_ps(-10.0/(d_scalar*d_scalar*d_scalar));
 961     swV4             = _mm256_set1_ps( 15.0/(d_scalar*d_scalar*d_scalar*d_scalar));
 962     swV5             = _mm256_set1_ps( -6.0/(d_scalar*d_scalar*d_scalar*d_scalar*d_scalar));
 963     swF2             = _mm256_set1_ps(-30.0/(d_scalar*d_scalar*d_scalar));
 964     swF3             = _mm256_set1_ps( 60.0/(d_scalar*d_scalar*d_scalar*d_scalar));
 965     swF4             = _mm256_set1_ps(-30.0/(d_scalar*d_scalar*d_scalar*d_scalar*d_scalar));
 966
 967     /* Avoid stupid compiler warnings */
 968     jnrA = jnrB = jnrC = jnrD = jnrE = jnrF = jnrG = jnrH = 0;
 969     j_coord_offsetA = 0;
 970     j_coord_offsetB = 0;
 971     j_coord_offsetC = 0;
 972     j_coord_offsetD = 0;
 973     j_coord_offsetE = 0;
 974     j_coord_offsetF = 0;
 975     j_coord_offsetG = 0;
 976     j_coord_offsetH = 0;
 977
 978     outeriter        = 0;
 979     inneriter        = 0;
 980
 981     for(iidx=0;iidx<4*DIM;iidx++)
 982     {
 983         scratch[iidx] = 0.0;
 984     }
 985
 986     /* Start outer loop over neighborlists */
 987     for(iidx=0; iidx<nri; iidx++)
 988     {
 989         /* Load shift vector for this list */
 990         i_shift_offset   = DIM*shiftidx[iidx];
 991
 992         /* Load limits for loop over neighbors */
 993         j_index_start    = jindex[iidx];
 994         j_index_end      = jindex[iidx+1];
 995
 996         /* Get outer coordinate index */
 997         inr              = iinr[iidx];
 998         i_coord_offset   = DIM*inr;
 999
1000         /* Load i particle coords and add shift vector */
1001         gmx_mm256_load_shift_and_3rvec_broadcast_ps(shiftvec+i_shift_offset,x+i_coord_offset,
1002                                                     &ix0,&iy0,&iz0,&ix1,&iy1,&iz1,&ix2,&iy2,&iz2);
1003
1004         fix0             = _mm256_setzero_ps();
1005         fiy0             = _mm256_setzero_ps();
1006         fiz0             = _mm256_setzero_ps();
1007         fix1             = _mm256_setzero_ps();
1008         fiy1             = _mm256_setzero_ps();
1009         fiz1             = _mm256_setzero_ps();
1010         fix2             = _mm256_setzero_ps();
1011         fiy2             = _mm256_setzero_ps();
1012         fiz2             = _mm256_setzero_ps();
1013
1014         /* Start inner kernel loop */
1015         for(jidx=j_index_start; jidx<j_index_end && jjnr[jidx+7]>=0; jidx+=8)
1016         {
1017
1018             /* Get j neighbor index, and coordinate index */
1019             jnrA             = jjnr[jidx];
1020             jnrB             = jjnr[jidx+1];
1021             jnrC             = jjnr[jidx+2];
1022             jnrD             = jjnr[jidx+3];
1023             jnrE             = jjnr[jidx+4];
1024             jnrF             = jjnr[jidx+5];
1025             jnrG             = jjnr[jidx+6];
1026             jnrH             = jjnr[jidx+7];
1027             j_coord_offsetA  = DIM*jnrA;
1028             j_coord_offsetB  = DIM*jnrB;
1029             j_coord_offsetC  = DIM*jnrC;
1030             j_coord_offsetD  = DIM*jnrD;
1031             j_coord_offsetE  = DIM*jnrE;
1032             j_coord_offsetF  = DIM*jnrF;
1033             j_coord_offsetG  = DIM*jnrG;
1034             j_coord_offsetH  = DIM*jnrH;
1035
1036             /* load j atom coordinates */
1037             gmx_mm256_load_1rvec_8ptr_swizzle_ps(x+j_coord_offsetA,x+j_coord_offsetB,
1038                                                  x+j_coord_offsetC,x+j_coord_offsetD,
1039                                                  x+j_coord_offsetE,x+j_coord_offsetF,
1040                                                  x+j_coord_offsetG,x+j_coord_offsetH,
1041                                                  &jx0,&jy0,&jz0);
1042
1043             /* Calculate displacement vector */
1044             dx00             = _mm256_sub_ps(ix0,jx0);
1045             dy00             = _mm256_sub_ps(iy0,jy0);
1046             dz00             = _mm256_sub_ps(iz0,jz0);
1047             dx10             = _mm256_sub_ps(ix1,jx0);
1048             dy10             = _mm256_sub_ps(iy1,jy0);
1049             dz10             = _mm256_sub_ps(iz1,jz0);
1050             dx20             = _mm256_sub_ps(ix2,jx0);
1051             dy20             = _mm256_sub_ps(iy2,jy0);
1052             dz20             = _mm256_sub_ps(iz2,jz0);
1053
1054             /* Calculate squared distance and things based on it */
1055             rsq00            = gmx_mm256_calc_rsq_ps(dx00,dy00,dz00);
1056             rsq10            = gmx_mm256_calc_rsq_ps(dx10,dy10,dz10);
1057             rsq20            = gmx_mm256_calc_rsq_ps(dx20,dy20,dz20);
1058
1059             rinv00           = gmx_mm256_invsqrt_ps(rsq00);
1060             rinv10           = gmx_mm256_invsqrt_ps(rsq10);
1061             rinv20           = gmx_mm256_invsqrt_ps(rsq20);
1062
1063             rinvsq00         = _mm256_mul_ps(rinv00,rinv00);
1064             rinvsq10         = _mm256_mul_ps(rinv10,rinv10);
1065             rinvsq20         = _mm256_mul_ps(rinv20,rinv20);
1066
1067             /* Load parameters for j particles */
1068             jq0              = gmx_mm256_load_8real_swizzle_ps(charge+jnrA+0,charge+jnrB+0,
1069                                                                  charge+jnrC+0,charge+jnrD+0,
1070                                                                  charge+jnrE+0,charge+jnrF+0,
1071                                                                  charge+jnrG+0,charge+jnrH+0);
1072             vdwjidx0A        = 2*vdwtype[jnrA+0];
1073             vdwjidx0B        = 2*vdwtype[jnrB+0];
1074             vdwjidx0C        = 2*vdwtype[jnrC+0];
1075             vdwjidx0D        = 2*vdwtype[jnrD+0];
1076             vdwjidx0E        = 2*vdwtype[jnrE+0];
1077             vdwjidx0F        = 2*vdwtype[jnrF+0];
1078             vdwjidx0G        = 2*vdwtype[jnrG+0];
1079             vdwjidx0H        = 2*vdwtype[jnrH+0];
1080
1081             fjx0             = _mm256_setzero_ps();
1082             fjy0             = _mm256_setzero_ps();
1083             fjz0             = _mm256_setzero_ps();
1084
1085             /**************************
1086              * CALCULATE INTERACTIONS *
1087              **************************/
1088
1089             if (gmx_mm256_any_lt(rsq00,rcutoff2))
1090             {
1091
1092             r00              = _mm256_mul_ps(rsq00,rinv00);
1093
1094             /* Compute parameters for interactions between i and j atoms */
1095             qq00             = _mm256_mul_ps(iq0,jq0);
1096             gmx_mm256_load_8pair_swizzle_ps(vdwioffsetptr0+vdwjidx0A,
1097                                             vdwioffsetptr0+vdwjidx0B,
1098                                             vdwioffsetptr0+vdwjidx0C,
1099                                             vdwioffsetptr0+vdwjidx0D,
1100                                             vdwioffsetptr0+vdwjidx0E,
1101                                             vdwioffsetptr0+vdwjidx0F,
1102                                             vdwioffsetptr0+vdwjidx0G,
1103                                             vdwioffsetptr0+vdwjidx0H,
1104                                             &c6_00,&c12_00);
1105
1106             /* EWALD ELECTROSTATICS */
1107
1108             /* Analytical PME correction */
1109             zeta2            = _mm256_mul_ps(beta2,rsq00);
1110             rinv3            = _mm256_mul_ps(rinvsq00,rinv00);
1111             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
1112             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
1113             felec            = _mm256_mul_ps(qq00,felec);
1114             pmecorrV         = gmx_mm256_pmecorrV_ps(zeta2);
1115             pmecorrV         = _mm256_mul_ps(pmecorrV,beta);
1116             velec            = _mm256_sub_ps(rinv00,pmecorrV);
1117             velec            = _mm256_mul_ps(qq00,velec);
1118
1119             /* LENNARD-JONES DISPERSION/REPULSION */
1120
1121             rinvsix          = _mm256_mul_ps(_mm256_mul_ps(rinvsq00,rinvsq00),rinvsq00);
1122             vvdw6            = _mm256_mul_ps(c6_00,rinvsix);
1123             vvdw12           = _mm256_mul_ps(c12_00,_mm256_mul_ps(rinvsix,rinvsix));
1124             vvdw             = _mm256_sub_ps( _mm256_mul_ps(vvdw12,one_twelfth) , _mm256_mul_ps(vvdw6,one_sixth) );
1125             fvdw             = _mm256_mul_ps(_mm256_sub_ps(vvdw12,vvdw6),rinvsq00);
1126
1127             d                = _mm256_sub_ps(r00,rswitch);
1128             d                = _mm256_max_ps(d,_mm256_setzero_ps());
1129             d2               = _mm256_mul_ps(d,d);
1130             sw               = _mm256_add_ps(one,_mm256_mul_ps(d2,_mm256_mul_ps(d,_mm256_add_ps(swV3,_mm256_mul_ps(d,_mm256_add_ps(swV4,_mm256_mul_ps(d,swV5)))))));
1131
1132             dsw              = _mm256_mul_ps(d2,_mm256_add_ps(swF2,_mm256_mul_ps(d,_mm256_add_ps(swF3,_mm256_mul_ps(d,swF4)))));
1133
1134             /* Evaluate switch function */
1135             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
1136             felec            = _mm256_sub_ps( _mm256_mul_ps(felec,sw) , _mm256_mul_ps(rinv00,_mm256_mul_ps(velec,dsw)) );
1137             fvdw             = _mm256_sub_ps( _mm256_mul_ps(fvdw,sw) , _mm256_mul_ps(rinv00,_mm256_mul_ps(vvdw,dsw)) );
1138             cutoff_mask      = _mm256_cmp_ps(rsq00,rcutoff2,_CMP_LT_OQ);
1139
1140             fscal            = _mm256_add_ps(felec,fvdw);
1141
1142             fscal            = _mm256_and_ps(fscal,cutoff_mask);
1143
1144             /* Calculate temporary vectorial force */
1145             tx               = _mm256_mul_ps(fscal,dx00);
1146             ty               = _mm256_mul_ps(fscal,dy00);
1147             tz               = _mm256_mul_ps(fscal,dz00);
1148
1149             /* Update vectorial force */
1150             fix0             = _mm256_add_ps(fix0,tx);
1151             fiy0             = _mm256_add_ps(fiy0,ty);
1152             fiz0             = _mm256_add_ps(fiz0,tz);
1153
1154             fjx0             = _mm256_add_ps(fjx0,tx);
1155             fjy0             = _mm256_add_ps(fjy0,ty);
1156             fjz0             = _mm256_add_ps(fjz0,tz);
1157
1158             }
1159
1160             /**************************
1161              * CALCULATE INTERACTIONS *
1162              **************************/
1163
1164             if (gmx_mm256_any_lt(rsq10,rcutoff2))
1165             {
1166
1167             r10              = _mm256_mul_ps(rsq10,rinv10);
1168
1169             /* Compute parameters for interactions between i and j atoms */
1170             qq10             = _mm256_mul_ps(iq1,jq0);
1171
1172             /* EWALD ELECTROSTATICS */
1173
1174             /* Analytical PME correction */
1175             zeta2            = _mm256_mul_ps(beta2,rsq10);
1176             rinv3            = _mm256_mul_ps(rinvsq10,rinv10);
1177             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
1178             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
1179             felec            = _mm256_mul_ps(qq10,felec);
1180             pmecorrV         = gmx_mm256_pmecorrV_ps(zeta2);
1181             pmecorrV         = _mm256_mul_ps(pmecorrV,beta);
1182             velec            = _mm256_sub_ps(rinv10,pmecorrV);
1183             velec            = _mm256_mul_ps(qq10,velec);
1184
1185             d                = _mm256_sub_ps(r10,rswitch);
1186             d                = _mm256_max_ps(d,_mm256_setzero_ps());
1187             d2               = _mm256_mul_ps(d,d);
1188             sw               = _mm256_add_ps(one,_mm256_mul_ps(d2,_mm256_mul_ps(d,_mm256_add_ps(swV3,_mm256_mul_ps(d,_mm256_add_ps(swV4,_mm256_mul_ps(d,swV5)))))));
1189
1190             dsw              = _mm256_mul_ps(d2,_mm256_add_ps(swF2,_mm256_mul_ps(d,_mm256_add_ps(swF3,_mm256_mul_ps(d,swF4)))));
1191
1192             /* Evaluate switch function */
1193             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
1194             felec            = _mm256_sub_ps( _mm256_mul_ps(felec,sw) , _mm256_mul_ps(rinv10,_mm256_mul_ps(velec,dsw)) );
1195             cutoff_mask      = _mm256_cmp_ps(rsq10,rcutoff2,_CMP_LT_OQ);
1196
1197             fscal            = felec;
1198
1199             fscal            = _mm256_and_ps(fscal,cutoff_mask);
1200
1201             /* Calculate temporary vectorial force */
1202             tx               = _mm256_mul_ps(fscal,dx10);
1203             ty               = _mm256_mul_ps(fscal,dy10);
1204             tz               = _mm256_mul_ps(fscal,dz10);
1205
1206             /* Update vectorial force */
1207             fix1             = _mm256_add_ps(fix1,tx);
1208             fiy1             = _mm256_add_ps(fiy1,ty);
1209             fiz1             = _mm256_add_ps(fiz1,tz);
1210
1211             fjx0             = _mm256_add_ps(fjx0,tx);
1212             fjy0             = _mm256_add_ps(fjy0,ty);
1213             fjz0             = _mm256_add_ps(fjz0,tz);
1214
1215             }
1216
1217             /**************************
1218              * CALCULATE INTERACTIONS *
1219              **************************/
1220
1221             if (gmx_mm256_any_lt(rsq20,rcutoff2))
1222             {
1223
1224             r20              = _mm256_mul_ps(rsq20,rinv20);
1225
1226             /* Compute parameters for interactions between i and j atoms */
1227             qq20             = _mm256_mul_ps(iq2,jq0);
1228
1229             /* EWALD ELECTROSTATICS */
1230
1231             /* Analytical PME correction */
1232             zeta2            = _mm256_mul_ps(beta2,rsq20);
1233             rinv3            = _mm256_mul_ps(rinvsq20,rinv20);
1234             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
1235             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
1236             felec            = _mm256_mul_ps(qq20,felec);
1237             pmecorrV         = gmx_mm256_pmecorrV_ps(zeta2);
1238             pmecorrV         = _mm256_mul_ps(pmecorrV,beta);
1239             velec            = _mm256_sub_ps(rinv20,pmecorrV);
1240             velec            = _mm256_mul_ps(qq20,velec);
1241
1242             d                = _mm256_sub_ps(r20,rswitch);
1243             d                = _mm256_max_ps(d,_mm256_setzero_ps());
1244             d2               = _mm256_mul_ps(d,d);
1245             sw               = _mm256_add_ps(one,_mm256_mul_ps(d2,_mm256_mul_ps(d,_mm256_add_ps(swV3,_mm256_mul_ps(d,_mm256_add_ps(swV4,_mm256_mul_ps(d,swV5)))))));
1246
1247             dsw              = _mm256_mul_ps(d2,_mm256_add_ps(swF2,_mm256_mul_ps(d,_mm256_add_ps(swF3,_mm256_mul_ps(d,swF4)))));
1248
1249             /* Evaluate switch function */
1250             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
1251             felec            = _mm256_sub_ps( _mm256_mul_ps(felec,sw) , _mm256_mul_ps(rinv20,_mm256_mul_ps(velec,dsw)) );
1252             cutoff_mask      = _mm256_cmp_ps(rsq20,rcutoff2,_CMP_LT_OQ);
1253
1254             fscal            = felec;
1255
1256             fscal            = _mm256_and_ps(fscal,cutoff_mask);
1257
1258             /* Calculate temporary vectorial force */
1259             tx               = _mm256_mul_ps(fscal,dx20);
1260             ty               = _mm256_mul_ps(fscal,dy20);
1261             tz               = _mm256_mul_ps(fscal,dz20);
1262
1263             /* Update vectorial force */
1264             fix2             = _mm256_add_ps(fix2,tx);
1265             fiy2             = _mm256_add_ps(fiy2,ty);
1266             fiz2             = _mm256_add_ps(fiz2,tz);
1267
1268             fjx0             = _mm256_add_ps(fjx0,tx);
1269             fjy0             = _mm256_add_ps(fjy0,ty);
1270             fjz0             = _mm256_add_ps(fjz0,tz);
1271
1272             }
1273
1274             fjptrA             = f+j_coord_offsetA;
1275             fjptrB             = f+j_coord_offsetB;
1276             fjptrC             = f+j_coord_offsetC;
1277             fjptrD             = f+j_coord_offsetD;
1278             fjptrE             = f+j_coord_offsetE;
1279             fjptrF             = f+j_coord_offsetF;
1280             fjptrG             = f+j_coord_offsetG;
1281             fjptrH             = f+j_coord_offsetH;
1282
1283             gmx_mm256_decrement_1rvec_8ptr_swizzle_ps(fjptrA,fjptrB,fjptrC,fjptrD,fjptrE,fjptrF,fjptrG,fjptrH,fjx0,fjy0,fjz0);
1284
1285             /* Inner loop uses 333 flops */
1286         }
1287
1288         if(jidx<j_index_end)
1289         {
1290
1291             /* Get j neighbor index, and coordinate index */
1292             jnrlistA         = jjnr[jidx];
1293             jnrlistB         = jjnr[jidx+1];
1294             jnrlistC         = jjnr[jidx+2];
1295             jnrlistD         = jjnr[jidx+3];
1296             jnrlistE         = jjnr[jidx+4];
1297             jnrlistF         = jjnr[jidx+5];
1298             jnrlistG         = jjnr[jidx+6];
1299             jnrlistH         = jjnr[jidx+7];
1300             /* Sign of each element will be negative for non-real atoms.
1301              * This mask will be 0xFFFFFFFF for dummy entries and 0x0 for real ones,
1302              * so use it as val = _mm_andnot_ps(mask,val) to clear dummy entries.
1303              */
1304             dummy_mask = gmx_mm256_set_m128(gmx_mm_castsi128_ps(_mm_cmplt_epi32(_mm_loadu_si128((const __m128i *)(jjnr+jidx+4)),_mm_setzero_si128())),
1305                                             gmx_mm_castsi128_ps(_mm_cmplt_epi32(_mm_loadu_si128((const __m128i *)(jjnr+jidx)),_mm_setzero_si128())));
1306
1307             jnrA       = (jnrlistA>=0) ? jnrlistA : 0;
1308             jnrB       = (jnrlistB>=0) ? jnrlistB : 0;
1309             jnrC       = (jnrlistC>=0) ? jnrlistC : 0;
1310             jnrD       = (jnrlistD>=0) ? jnrlistD : 0;
1311             jnrE       = (jnrlistE>=0) ? jnrlistE : 0;
1312             jnrF       = (jnrlistF>=0) ? jnrlistF : 0;
1313             jnrG       = (jnrlistG>=0) ? jnrlistG : 0;
1314             jnrH       = (jnrlistH>=0) ? jnrlistH : 0;
1315             j_coord_offsetA  = DIM*jnrA;
1316             j_coord_offsetB  = DIM*jnrB;
1317             j_coord_offsetC  = DIM*jnrC;
1318             j_coord_offsetD  = DIM*jnrD;
1319             j_coord_offsetE  = DIM*jnrE;
1320             j_coord_offsetF  = DIM*jnrF;
1321             j_coord_offsetG  = DIM*jnrG;
1322             j_coord_offsetH  = DIM*jnrH;
1323
1324             /* load j atom coordinates */
1325             gmx_mm256_load_1rvec_8ptr_swizzle_ps(x+j_coord_offsetA,x+j_coord_offsetB,
1326                                                  x+j_coord_offsetC,x+j_coord_offsetD,
1327                                                  x+j_coord_offsetE,x+j_coord_offsetF,
1328                                                  x+j_coord_offsetG,x+j_coord_offsetH,
1329                                                  &jx0,&jy0,&jz0);
1330
1331             /* Calculate displacement vector */
1332             dx00             = _mm256_sub_ps(ix0,jx0);
1333             dy00             = _mm256_sub_ps(iy0,jy0);
1334             dz00             = _mm256_sub_ps(iz0,jz0);
1335             dx10             = _mm256_sub_ps(ix1,jx0);
1336             dy10             = _mm256_sub_ps(iy1,jy0);
1337             dz10             = _mm256_sub_ps(iz1,jz0);
1338             dx20             = _mm256_sub_ps(ix2,jx0);
1339             dy20             = _mm256_sub_ps(iy2,jy0);
1340             dz20             = _mm256_sub_ps(iz2,jz0);
1341
1342             /* Calculate squared distance and things based on it */
1343             rsq00            = gmx_mm256_calc_rsq_ps(dx00,dy00,dz00);
1344             rsq10            = gmx_mm256_calc_rsq_ps(dx10,dy10,dz10);
1345             rsq20            = gmx_mm256_calc_rsq_ps(dx20,dy20,dz20);
1346
1347             rinv00           = gmx_mm256_invsqrt_ps(rsq00);
1348             rinv10           = gmx_mm256_invsqrt_ps(rsq10);
1349             rinv20           = gmx_mm256_invsqrt_ps(rsq20);
1350
1351             rinvsq00         = _mm256_mul_ps(rinv00,rinv00);
1352             rinvsq10         = _mm256_mul_ps(rinv10,rinv10);
1353             rinvsq20         = _mm256_mul_ps(rinv20,rinv20);
1354
1355             /* Load parameters for j particles */
1356             jq0              = gmx_mm256_load_8real_swizzle_ps(charge+jnrA+0,charge+jnrB+0,
1357                                                                  charge+jnrC+0,charge+jnrD+0,
1358                                                                  charge+jnrE+0,charge+jnrF+0,
1359                                                                  charge+jnrG+0,charge+jnrH+0);
1360             vdwjidx0A        = 2*vdwtype[jnrA+0];
1361             vdwjidx0B        = 2*vdwtype[jnrB+0];
1362             vdwjidx0C        = 2*vdwtype[jnrC+0];
1363             vdwjidx0D        = 2*vdwtype[jnrD+0];
1364             vdwjidx0E        = 2*vdwtype[jnrE+0];
1365             vdwjidx0F        = 2*vdwtype[jnrF+0];
1366             vdwjidx0G        = 2*vdwtype[jnrG+0];
1367             vdwjidx0H        = 2*vdwtype[jnrH+0];
1368
1369             fjx0             = _mm256_setzero_ps();
1370             fjy0             = _mm256_setzero_ps();
1371             fjz0             = _mm256_setzero_ps();
1372
1373             /**************************
1374              * CALCULATE INTERACTIONS *
1375              **************************/
1376
1377             if (gmx_mm256_any_lt(rsq00,rcutoff2))
1378             {
1379
1380             r00              = _mm256_mul_ps(rsq00,rinv00);
1381             r00              = _mm256_andnot_ps(dummy_mask,r00);
1382
1383             /* Compute parameters for interactions between i and j atoms */
1384             qq00             = _mm256_mul_ps(iq0,jq0);
1385             gmx_mm256_load_8pair_swizzle_ps(vdwioffsetptr0+vdwjidx0A,
1386                                             vdwioffsetptr0+vdwjidx0B,
1387                                             vdwioffsetptr0+vdwjidx0C,
1388                                             vdwioffsetptr0+vdwjidx0D,
1389                                             vdwioffsetptr0+vdwjidx0E,
1390                                             vdwioffsetptr0+vdwjidx0F,
1391                                             vdwioffsetptr0+vdwjidx0G,
1392                                             vdwioffsetptr0+vdwjidx0H,
1393                                             &c6_00,&c12_00);
1394
1395             /* EWALD ELECTROSTATICS */
1396
1397             /* Analytical PME correction */
1398             zeta2            = _mm256_mul_ps(beta2,rsq00);
1399             rinv3            = _mm256_mul_ps(rinvsq00,rinv00);
1400             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
1401             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
1402             felec            = _mm256_mul_ps(qq00,felec);
1403             pmecorrV         = gmx_mm256_pmecorrV_ps(zeta2);
1404             pmecorrV         = _mm256_mul_ps(pmecorrV,beta);
1405             velec            = _mm256_sub_ps(rinv00,pmecorrV);
1406             velec            = _mm256_mul_ps(qq00,velec);
1407
1408             /* LENNARD-JONES DISPERSION/REPULSION */
1409
1410             rinvsix          = _mm256_mul_ps(_mm256_mul_ps(rinvsq00,rinvsq00),rinvsq00);
1411             vvdw6            = _mm256_mul_ps(c6_00,rinvsix);
1412             vvdw12           = _mm256_mul_ps(c12_00,_mm256_mul_ps(rinvsix,rinvsix));
1413             vvdw             = _mm256_sub_ps( _mm256_mul_ps(vvdw12,one_twelfth) , _mm256_mul_ps(vvdw6,one_sixth) );
1414             fvdw             = _mm256_mul_ps(_mm256_sub_ps(vvdw12,vvdw6),rinvsq00);
1415
1416             d                = _mm256_sub_ps(r00,rswitch);
1417             d                = _mm256_max_ps(d,_mm256_setzero_ps());
1418             d2               = _mm256_mul_ps(d,d);
1419             sw               = _mm256_add_ps(one,_mm256_mul_ps(d2,_mm256_mul_ps(d,_mm256_add_ps(swV3,_mm256_mul_ps(d,_mm256_add_ps(swV4,_mm256_mul_ps(d,swV5)))))));
1420
1421             dsw              = _mm256_mul_ps(d2,_mm256_add_ps(swF2,_mm256_mul_ps(d,_mm256_add_ps(swF3,_mm256_mul_ps(d,swF4)))));
1422
1423             /* Evaluate switch function */
1424             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
1425             felec            = _mm256_sub_ps( _mm256_mul_ps(felec,sw) , _mm256_mul_ps(rinv00,_mm256_mul_ps(velec,dsw)) );
1426             fvdw             = _mm256_sub_ps( _mm256_mul_ps(fvdw,sw) , _mm256_mul_ps(rinv00,_mm256_mul_ps(vvdw,dsw)) );
1427             cutoff_mask      = _mm256_cmp_ps(rsq00,rcutoff2,_CMP_LT_OQ);
1428
1429             fscal            = _mm256_add_ps(felec,fvdw);
1430
1431             fscal            = _mm256_and_ps(fscal,cutoff_mask);
1432
1433             fscal            = _mm256_andnot_ps(dummy_mask,fscal);
1434
1435             /* Calculate temporary vectorial force */
1436             tx               = _mm256_mul_ps(fscal,dx00);
1437             ty               = _mm256_mul_ps(fscal,dy00);
1438             tz               = _mm256_mul_ps(fscal,dz00);
1439
1440             /* Update vectorial force */
1441             fix0             = _mm256_add_ps(fix0,tx);
1442             fiy0             = _mm256_add_ps(fiy0,ty);
1443             fiz0             = _mm256_add_ps(fiz0,tz);
1444
1445             fjx0             = _mm256_add_ps(fjx0,tx);
1446             fjy0             = _mm256_add_ps(fjy0,ty);
1447             fjz0             = _mm256_add_ps(fjz0,tz);
1448
1449             }
1450
1451             /**************************
1452              * CALCULATE INTERACTIONS *
1453              **************************/
1454
1455             if (gmx_mm256_any_lt(rsq10,rcutoff2))
1456             {
1457
1458             r10              = _mm256_mul_ps(rsq10,rinv10);
1459             r10              = _mm256_andnot_ps(dummy_mask,r10);
1460
1461             /* Compute parameters for interactions between i and j atoms */
1462             qq10             = _mm256_mul_ps(iq1,jq0);
1463
1464             /* EWALD ELECTROSTATICS */
1465
1466             /* Analytical PME correction */
1467             zeta2            = _mm256_mul_ps(beta2,rsq10);
1468             rinv3            = _mm256_mul_ps(rinvsq10,rinv10);
1469             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
1470             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
1471             felec            = _mm256_mul_ps(qq10,felec);
1472             pmecorrV         = gmx_mm256_pmecorrV_ps(zeta2);
1473             pmecorrV         = _mm256_mul_ps(pmecorrV,beta);
1474             velec            = _mm256_sub_ps(rinv10,pmecorrV);
1475             velec            = _mm256_mul_ps(qq10,velec);
1476
1477             d                = _mm256_sub_ps(r10,rswitch);
1478             d                = _mm256_max_ps(d,_mm256_setzero_ps());
1479             d2               = _mm256_mul_ps(d,d);
1480             sw               = _mm256_add_ps(one,_mm256_mul_ps(d2,_mm256_mul_ps(d,_mm256_add_ps(swV3,_mm256_mul_ps(d,_mm256_add_ps(swV4,_mm256_mul_ps(d,swV5)))))));
1481
1482             dsw              = _mm256_mul_ps(d2,_mm256_add_ps(swF2,_mm256_mul_ps(d,_mm256_add_ps(swF3,_mm256_mul_ps(d,swF4)))));
1483
1484             /* Evaluate switch function */
1485             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
1486             felec            = _mm256_sub_ps( _mm256_mul_ps(felec,sw) , _mm256_mul_ps(rinv10,_mm256_mul_ps(velec,dsw)) );
1487             cutoff_mask      = _mm256_cmp_ps(rsq10,rcutoff2,_CMP_LT_OQ);
1488
1489             fscal            = felec;
1490
1491             fscal            = _mm256_and_ps(fscal,cutoff_mask);
1492
1493             fscal            = _mm256_andnot_ps(dummy_mask,fscal);
1494
1495             /* Calculate temporary vectorial force */
1496             tx               = _mm256_mul_ps(fscal,dx10);
1497             ty               = _mm256_mul_ps(fscal,dy10);
1498             tz               = _mm256_mul_ps(fscal,dz10);
1499
1500             /* Update vectorial force */
1501             fix1             = _mm256_add_ps(fix1,tx);
1502             fiy1             = _mm256_add_ps(fiy1,ty);
1503             fiz1             = _mm256_add_ps(fiz1,tz);
1504
1505             fjx0             = _mm256_add_ps(fjx0,tx);
1506             fjy0             = _mm256_add_ps(fjy0,ty);
1507             fjz0             = _mm256_add_ps(fjz0,tz);
1508
1509             }
1510
1511             /**************************
1512              * CALCULATE INTERACTIONS *
1513              **************************/
1514
1515             if (gmx_mm256_any_lt(rsq20,rcutoff2))
1516             {
1517
1518             r20              = _mm256_mul_ps(rsq20,rinv20);
1519             r20              = _mm256_andnot_ps(dummy_mask,r20);
1520
1521             /* Compute parameters for interactions between i and j atoms */
1522             qq20             = _mm256_mul_ps(iq2,jq0);
1523
1524             /* EWALD ELECTROSTATICS */
1525
1526             /* Analytical PME correction */
1527             zeta2            = _mm256_mul_ps(beta2,rsq20);
1528             rinv3            = _mm256_mul_ps(rinvsq20,rinv20);
1529             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
1530             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
1531             felec            = _mm256_mul_ps(qq20,felec);
1532             pmecorrV         = gmx_mm256_pmecorrV_ps(zeta2);
1533             pmecorrV         = _mm256_mul_ps(pmecorrV,beta);
1534             velec            = _mm256_sub_ps(rinv20,pmecorrV);
1535             velec            = _mm256_mul_ps(qq20,velec);
1536
1537             d                = _mm256_sub_ps(r20,rswitch);
1538             d                = _mm256_max_ps(d,_mm256_setzero_ps());
1539             d2               = _mm256_mul_ps(d,d);
1540             sw               = _mm256_add_ps(one,_mm256_mul_ps(d2,_mm256_mul_ps(d,_mm256_add_ps(swV3,_mm256_mul_ps(d,_mm256_add_ps(swV4,_mm256_mul_ps(d,swV5)))))));
1541
1542             dsw              = _mm256_mul_ps(d2,_mm256_add_ps(swF2,_mm256_mul_ps(d,_mm256_add_ps(swF3,_mm256_mul_ps(d,swF4)))));
1543
1544             /* Evaluate switch function */
1545             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
1546             felec            = _mm256_sub_ps( _mm256_mul_ps(felec,sw) , _mm256_mul_ps(rinv20,_mm256_mul_ps(velec,dsw)) );
1547             cutoff_mask      = _mm256_cmp_ps(rsq20,rcutoff2,_CMP_LT_OQ);
1548
1549             fscal            = felec;
1550
1551             fscal            = _mm256_and_ps(fscal,cutoff_mask);
1552
1553             fscal            = _mm256_andnot_ps(dummy_mask,fscal);
1554
1555             /* Calculate temporary vectorial force */
1556             tx               = _mm256_mul_ps(fscal,dx20);
1557             ty               = _mm256_mul_ps(fscal,dy20);
1558             tz               = _mm256_mul_ps(fscal,dz20);
1559
1560             /* Update vectorial force */
1561             fix2             = _mm256_add_ps(fix2,tx);
1562             fiy2             = _mm256_add_ps(fiy2,ty);
1563             fiz2             = _mm256_add_ps(fiz2,tz);
1564
1565             fjx0             = _mm256_add_ps(fjx0,tx);
1566             fjy0             = _mm256_add_ps(fjy0,ty);
1567             fjz0             = _mm256_add_ps(fjz0,tz);
1568
1569             }
1570
1571             fjptrA             = (jnrlistA>=0) ? f+j_coord_offsetA : scratch;
1572             fjptrB             = (jnrlistB>=0) ? f+j_coord_offsetB : scratch;
1573             fjptrC             = (jnrlistC>=0) ? f+j_coord_offsetC : scratch;
1574             fjptrD             = (jnrlistD>=0) ? f+j_coord_offsetD : scratch;
1575             fjptrE             = (jnrlistE>=0) ? f+j_coord_offsetE : scratch;
1576             fjptrF             = (jnrlistF>=0) ? f+j_coord_offsetF : scratch;
1577             fjptrG             = (jnrlistG>=0) ? f+j_coord_offsetG : scratch;
1578             fjptrH             = (jnrlistH>=0) ? f+j_coord_offsetH : scratch;
1579
1580             gmx_mm256_decrement_1rvec_8ptr_swizzle_ps(fjptrA,fjptrB,fjptrC,fjptrD,fjptrE,fjptrF,fjptrG,fjptrH,fjx0,fjy0,fjz0);
1581
1582             /* Inner loop uses 336 flops */
1583         }
1584
1585         /* End of innermost loop */
1586
1587         gmx_mm256_update_iforce_3atom_swizzle_ps(fix0,fiy0,fiz0,fix1,fiy1,fiz1,fix2,fiy2,fiz2,
1588                                                  f+i_coord_offset,fshift+i_shift_offset);
1589
1590         /* Increment number of inner iterations */
1591         inneriter                  += j_index_end - j_index_start;
1592
1593         /* Outer loop uses 18 flops */
1594     }
1595
1596     /* Increment number of outer iterations */
1597     outeriter        += nri;
1598
1599     /* Update outer/inner flops */
1600
1601     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W3_F,outeriter*18 + inneriter*336);
1602 }